Professional Documents
Culture Documents
US HUPO 2022 Abstract Book
US HUPO 2022 Abstract Book
Page 2 of 166
OA05.02 Quantification of Brain-Derived CSF Protein Biomarkers by Targeted Mass Spectrometry to Stage Alzheimer’s Disease
Progression .......................................................................................................................................................................................... 23
Page 3 of 166
Parallel Session 11: Proteomics and Biomimicry ............................................................. 36
IS11.01 Comparative Mammalian Proteomics: Increasing the Biomimicry Space alongside Genomes .............................................. 36
IS11.02 Insights into human pathology from proteome studies of extreme adaptations in mammals.................................................. 36
OA11.01 Leveraging large-scale comparative proteomics across the tree of life to improve human disease genetics ....................... 36
OA11.02 Are Mouse Models of Alzheimer’s Disease a Surrogate for Human Disease? – Using quantitative proteomics to monitor
time resolved formation of pathology ................................................................................................................................................... 37
Page 4 of 166
P01.07 Assessing the Precision of Detergent-based Workflows: Cartridge-Based Proteome Precipitation for Robust Sample
Preparation .......................................................................................................................................................................................... 49
P01.08 Automated S-Trap Sample Preparation on an Integrated Liquid Handling / Pneumatic Sample Processing Workstation for
MAM Applications ................................................................................................................................................................................ 50
P01.09 Multi-nanoparticle Workflow Enables Deep Plasma Proteomics at Scale, with Enhanced Precision, and Depths of Coverage.
............................................................................................................................................................................................................. 51
P01.10 The Next Generation Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (NextGen IR-MALDESI) Source.... 51
P01.11 A Universal and High-Throughput Proteomics Sample Preparation Platform.......................................................................... 52
P01.12 Real-Time Database Search Multiplex Quantitative Proteomics: Instrument Method and Data Processing ........................... 53
P01.13 Benchmark of Micro-flow Chromatograph for Robust Proteomics Analysis............................................................................. 54
P01.14 DIA Analysis in Discovery Identifications and Label-free Quantification on Orbitrap Exploris 240 Mass Spectrometer .......... 54
P01.15 Semi-automated and high-throughput homogenization technique for in-depth analysis of various tissue proteomes............. 55
P01.16 Multi-omics sample preparation workflow for DNA and proteins from a single starting sample using ProMTag ..................... 55
P01.17 Robust Identification of Low Abundant Proteins Utilizing a Precipitation Workflow on the ProTrap XG .................................. 56
P01.18 Establishing Quality Control Procedures for Sample Preparation of High-Throughput Plasma Proteomics ............................ 57
P01.19 Establishing Quality Control Procedures for Large-Scale Discovery Based Proteomics Analysis of Human Plasma ............. 57
P01.20 High throughput single-shot proteomics on the timsTOF Pro 2 with DIA-PASEF and DDA-PASEF acquisition mode ........... 58
P01.21 Analyzing protein fluorosequencing data, a new technology for single molecule proteomics.................................................. 59
P01.22 Universal Hydrophilic affinity SPE based peptide clean-up workflow for parallel sample processing ...................................... 59
P01.23 EPITOME ANALYSIS OF THE HUMAN PLASMA PROTEOME DELIVERS DIAGNOSTIC AND THERAPEUTIC
CANDIDATES. ..................................................................................................................................................................................... 60
P01.24 Giving CE-MS a New Life in Proteomics: Integrated Solid-Phase Extraction and Detergent Tolerance ................................. 60
P01.25 Quality Control of a DIA-MS High-Throughput Proteomics Workflow for CSF Samples from Parkinson’s Disease Patients .. 61
P01.26 Incorporation of glycoproteome detection into large scale unbiased proteomics studies utilizing nanoparticles ..................... 62
P01.27 "BCA-No-More": Seamless, High-Throughput Protein Quantification Directly on S-Trap Plates ............................................. 62
P01.28 Sample Preparation to Match Analytical Advances: 384-well S-Trap Plate ............................................................................. 63
P01.29 S-Trap Turbo: From Samples to Analysis in Record Time ...................................................................................................... 64
Page 5 of 166
P02.05 SIRT7 as a Novel Regulator of Cellular Senescence .............................................................................................................. 67
P02.06 Dynamics of Huntingtin Protein Interactions in the Striatum Identifies Candidate Modifiers of Huntington’s Disease ............. 68
P02.07 A proteomic network approach to define proteins associated with cognitive resilience in Alzheimer's disease ...................... 68
P02.08 Extracellular Matrix Proteins Differentially Expressed Between APOE Isoforms ..................................................................... 69
P02.10 Changes in the Proteome of Calorie Restricted Mice vs Ad Lipitum Mice ............................................................................... 70
P02.11 Multi-omics analysis of the exosome senescence-associated secretory phenotype to study aging biology............................ 70
P02.12 Deep and Unbiased Plasma Protein Profiling of Alzheimer’s and Mild Cognitive Impairment Subjects with a Novel Multi-
nanoparticle Approach ......................................................................................................................................................................... 71
P02.13 The Liver’s Role in ApoE-Mediated Alzheimer’s Disease Risk ................................................................................................ 72
P02.14 An Untargeted Metabolomic Investigation of Amyotrophic Lateral Sclerosis Decedent Brain Tissue by IR-MALDESI Mass
Spectrometry Imaging .......................................................................................................................................................................... 72
P02.15 Investigating Racial Disparities in Alzheimer’s Disease using Quantitative Brain Proteomics................................................. 73
P02.16 Mapping the 3 D Proteome using Surface Accessibility in Animal Models of Disease ............................................................ 73
P02.17 Increased Sequence Coverage of Low-abundance Biomarkers in Human Serum to Quantify Protein Folding Stability ......... 74
Page 6 of 166
P04.01 Proteomic Analysis Reveals that Musashi-2 Promotes Signaling Pathways Associated with Cell Proliferation and Metastasis
in Neuroblastoma Cells ........................................................................................................................................................................ 83
P04.02 Robust, Reliable, Reproducible Sample Preparation Enabling Confident LC-MS Analysis of Proteins in FFPE Tissues........ 83
P04.03 Deep Plasma Proteomics at Scale Enabling Proteogenomic Analyses in a Lung Cancer (NSCLC) Study ............................. 84
P04.04 Deep, Rapid and Unbiased Plasma Proteomics with Differential Analysis of Proteoforms Enabling Proteogenomic Studies in
a NSCLC Lung Cancer Study .............................................................................................................................................................. 84
P04.06 Low-Input Proteomic Phenotyping of a Metabolically Defined Putative Mammary Epithelial Stem Cell Population ................ 85
P04.07 Characterizing the Proteomic Response of Models of High-Grade Serous Tubo-Ovarian Cancer to Taxane Treatment ....... 86
P04.08 Proteogenomics reveal over-expression of HLA-I in cancer .................................................................................................... 86
Page 7 of 166
P07.11 Identifying Drivers of Human Therapeutic Antibody Production in Chinese Hamster Ovary Cells via Proteomics and
Bioinformatics .................................................................................................................................................................................... 101
P07.13 Increasing Human Leukocyte Antigen Peptide Identifications Using Fragpipe and MSBooster ............................................ 102
P07.14 One-stop DIA data analysis using MSFragger-DIA coupled with FragPipe ........................................................................... 102
P07.15 moPepGen: a fast custom database generator from multi-omics data for proteogenomics .................................................. 103
Page 8 of 166
P10: Metabolomics and Imaging Mass Spectrometry
............................................................................................................................................... 118
P10.01 Revealing the Wound Healing Process with Three-Dimensional Mass Spectrometry Imaging using Infrared Matrix-Assisted
Laser Desorption Ionization ............................................................................................................................................................... 118
P10.02 Metabolomic Characterization of Pseudomonas protegens Secretions for Identification of Biopesticides Against Bacterial
Panicle Blight of Rice ......................................................................................................................................................................... 118
P10.03 Using Covaris AFA for Fast, Reliable, and Automatable Extraction of Analytes from cells for MALDI-TOF Based Analyses119
P10.04 Leveraging Public Untargeted Metabolomics Data To Propagate Structurally Related Molecule Annotations to Millions of
MS/MS Spectra .................................................................................................................................................................................. 119
P10.05 Glycan and Collagen Imaging Mass Spectrometry Analysis of a Broad Tumor Spectrum of Prostate Cancer Tissue
Histopathologies ................................................................................................................................................................................ 120
P10.06 Improved Spatial Resolution of Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-MALDESI) Mass
Spectrometry Imaging (MSI) .............................................................................................................................................................. 121
P10.07 Novel Matrix Strategies for Improved Ionization and Spatial Resolution for IR-MALDESI Mass Spectrometry Imaging ....... 121
P10.08 A Novel Ion-beam Imaging Technique for Detection of Chromosomal Copy Number Alterations Associated with Progression
of Breast Cancer ................................................................................................................................................................................ 122
P10.09 Development of a Metabolomics System Suitability Sample for MS-based Metabolomics ................................................... 122
P10.10 Spatial Metabolomics of Xenopus laevis Tadpoles by IR-MALDESI Mass Spectrometry Imaging ........................................ 123
P10.11 Investigation of Sucrose as a Cryoprotectant for Tissue Preservation and Compatibility with IR-MALDESI Mass Spectrometry
Imaging .............................................................................................................................................................................................. 123
P10.12 Enhancing Metabolomic Coverage in Positive Ionization Mode Using Dicationic Reagents by IR-MALDESI MSI................ 124
Page 9 of 166
P12.04 Linking Dynamic Lysine Acylation with Metabolic Regulation in Bacteria .............................................................................. 131
P12.05 Compared dda-PASEF and prm-PASEF approaches for quantification of 2000 RAS induced Phosphopeptides. ............... 132
P12.06 Abscisic Acid Controlled Redox Proteome of Arabidopsis and Its Regulation by Heterotrimeric G-proteins ......................... 133
P12.07 High-throughput protein modification quantitation analysis using intact protein MRM and its application on hENGase inhibitor
screening ........................................................................................................................................................................................... 133
Page 10 of 166
P15.02 Proteomic Analysis of Dorsal Root Ganglia of Type II Diabetic Donors with Peripheral Neuropathic Pain ........................... 147
P15.03 Urinary Protein Signatures for the Detection of Antiretroviral-Induced Acute Kidney Injury: A Retrospective Study ............. 148
P15.04 Proteomics to understand effect of maternal diet during pregnancy on fetal adult life........................................................... 148
P15.05 Divergent proteostasis interactions reveal distinct pathways attenuated by correctors in CFTR misfolding mutations ......... 149
P15.06 Plasma Proteomic Changes in Response to a High Salt Diet in Female Baboons................................................................ 150
P15.07 Laser Capture Microdissection Assisted Spatial Resolved Proteomics Analysis of FFPE Human Tissues for Fibrosis-Related
Disease MOA Study........................................................................................................................................................................... 150
P15.08 Towards understanding the comprehensive extracellular matrix proteome in teeth .............................................................. 151
P15.09 A Fully Automated High-Throughput, Deep-Scale Quantitative Plasma Proteomics Workflow Enables Quantitively Profile
More Than 1000 Proteins Per Sample ............................................................................................................................................... 152
............................................................................................................................................... 153
P16.01 A Proteomics Perspective of Postnatal Swine Heart Development ....................................................................................... 153
P16.02 A complete and automated sample preparation strategy for high throughput and standardized proteomics ........................ 153
P16.03 Global multi-site assessment of LC-MS proteomics precision and accuracy with high-throughput analysis ......................... 154
P16.04 Phosphoproteomic analysis reveals kinases involved in SARS-CoV-2 induced pancreatic β-cell death............................... 155
P16.05 Quantitative proteomics of zebrafish embryos reveals wave-like expression profiles of transcription factors during maternal to
zygotic transition ................................................................................................................................................................................ 155
P16.06 Proteomic Analysis of Alcohol-associated Hepatitis Reveals Glycoprotein NMB (GPNMB) as a Novel Hepatic and Serum
Biomarker........................................................................................................................................................................................... 156
P16.07 Deep Metaproteome Analysis using a Vanquish Neo UHPLC System Coupled to an Orbitrap Eclipse Tribrid Mass
Spectrometer with FAIMS Pro Interface............................................................................................................................................. 156
P16.08 Comprehensive Comparison of Serum vs. Plasma High-Throughput Proteomics Across a Healthy Gender-Matched Cohort
........................................................................................................................................................................................................... 157
P16.09 Fast library generation using Zeno MS/MS and microflow chromatography.......................................................................... 158
P16.10 Enhanced Detection of Low Abundance Proteins in Various Mouse Tissues Using Offline HPLC as a Precursor to LC-
MS/MS. .............................................................................................................................................................................................. 158
P16.11 Deep Proteome Coverage by Offline HPLC Fractionation of Intact Proteins as a Precursor to Bottom-up Proteomic Analysis
........................................................................................................................................................................................................... 159
P16.12 Best practice of FAIMS with Single Compensation Voltage in Single-shot Proteomics using Data Independent Acquisition 160
P16.13 Identification of RNA-Binding Proteome in Mouse Macrophages .......................................................................................... 160
P16.14 Protein Folding Stabilities are a Major Determinant of Oxidation Rates for Buried Methionine Residues ............................. 161
Page 11 of 166
P17.01 Application of a library-free dia-PASEF approach for high throughput and high sensitivity proteomics ................................ 162
P17.02 Standardized and easy to deploy nanoLCMS for ultra-sensitive and high-throughput analysis of limited sample amounts and
single cells ......................................................................................................................................................................................... 162
P17.03 Multiplexed nanoLC System Produces a Label-Free Single-Cancer-Cell Proteome Every 15 Minutes ................................ 163
P17.04 Ultra-Sensitive Proteome Quantification on the timsTOF SCP Mass Spectrometer .............................................................. 164
Page 12 of 166
Parallel Session 01: Advances in Single Cell Measurements
IS01.01 Intelligent Data Acquisition for Multiplexed Single Cell Analysis
Christopher Rose, Genentech Inc.
Single cell proteomics methods often utilize a carrier proteome added at 20-500x single cell channels to enable peptide
detection, selection, and quantification. Adding excess carrier proteome may increase measurement variability and
decrease measurement accuracy. We introduce an advanced data acquisition approach that utilizes a carrier proteome
labeled with an offset mass chemical tag (i.e., TMT-SH) that separates the carrier proteome peptide cluster from the
multiplexed single cell proteomes and enables quantification of single cells in the absence of carrier proteome reporter ions.
This approach was implemented with inSeqAPI, an API program that enables facile construction of real-time data analysis
methods, such that sequencing of the TMT-SH labeled carrier proteome triggers quantitative MS2/SPS-MS3 on TMT
labeled single cell proteomes.
The recent introduction of real-time database searches (RTS) have drastically improved the number of peptides that can be
quantified during SPS-MS3 quantification of isobaric labeled samples. Here we introduce inSeqAPI - an API based
Windows application that enables the construction of custom acquisition methods on Orbitrap Eclipse mass spectrometers.
inSeqAPI is unique in that it enables modular method construction complete with a number of custom available real-time
filters as well as ‘.json’ method files that can be shared across instruments. Furthermore, inSeqAPI produces a full audit trail
of real-time filter results for each raw file, enabling optimization and troubleshooting during method development.
We apply inSeqAPI to the analysis of single cells by implementing an offset mass triggering method at proteome scale. This
approach enables a deep proteomic analysis and sensitive quantification of single cell proteomes by quantifying peptides
only after they are identified. Utilizing TMT-SH as a carrier proteome removes carrier proteome reporter ions from analysis
during quantitative scans. We demonstrate that this enables accurate quantification of proteins even when carrier proteome
levels exceed 500x.
IS01.02 Si Wu Presentation
Si Wu, University of Oklahoma
OA01.01 Improved sensitivity and depth of bottom-up and top-down proteomic and PTM profiling of ng- and sub-
ng-level samples
Alexander R. Ivanov (1)
(1) Northeastern University, Boston, Massachusetts, United States
Informative proteomic characterization of limited samples (e.g., small populations of rare cells, microneedle biopsies,
extracellular vesicles (EVs) isolated from minute volumes of physiological fluids, or even single cells) and especially,
profiling of post-translational modifications, e.g., glycosylation and phosphorylation, of such specimens have been a major
Page 13 of 166
challenge because of very low abundance and high heterogeneity in biological matrices. Alterations of proteomes,
glycomes, phosphoproteomes, and acetylomes may be associated with a number of pathologies, including immune,
cardiovascular and oncological diseases, as well as other biological phenomena. In this study, we evaluated a combination
of advanced sample preparation, ultra-low flow high-efficiency liquid phase separation coupled to mass spectrometry (MS)
via a high-field asymmetric waveform ion mobility spectrometry (FAIMS) interface to evaluate the potential applicability for
high sensitivity, robust and reproducible proteomic profiling of low ng- and sub-ng-level complex biological samples. We
explored several electric field- and pressure-driven separation approaches, including ultra-narrow bore monolithic and
porous layer open tubular (PLOT) liquid chromatography (LC) columns and capillary electrophoresis (CE) interfaced with
mass spectrometry to enhance the sensitivity and depth of proteomic, glycomic, acetylomic, and phosphoproteomic profiling
of several types of limited biological specimens. Both bottom-up and top-down MS-based proteomic strategies were
evaluated using cultured human cells, blood isolates, and commercial digest standards. The acquired results demonstrate
the potential applicability of the developed techniques in single-cell proteomic studies and analyses of minute amounts of
biological and clinical samples.
Isobaric tagging facilitates multiplexed experiments that can determine sequences and relative amounts of peptides in
biological samples using tandem mass spectrometry. Limited reporter ion generation limits quantitative accuracy and
precision. Further, unintended co-isolation of contaminating peaks in MS2 experiments distorts reporter intensities.
MS3 experiments address contamination by generating reporter ions via collisional activation (HCD) of peptide product ions
rather than the precursor ion. Because HCD performance is related to m/z, activation of synchronously isolated products
generates suboptimal reporter intensities. We supposed that a combination of IRMPD and ion parking would maximize
reporter generation from tagged peptides by eliminating the m/z dependence inherent in HCD. Peptide
ISQAVHAAHAEINEAGR was labeled with 10-plex TMT reagents. Pierce TMT11plex Yeast Digest Standard was obtained.
Experiments were performed on a quadrupole-Orbitrap-QLT MS system modified to include a 40 W CO2 continuous wave
laser which allowed for infrared photoactivation of ions within the QLT. Adjustments to the instrument control code allowed
for broadband ion parking at reporter ion m/z (126–131) during IRMPD experiments. Data-dependent MS2 scans were
followed by back-to-back HCD and IRMPD MS3 scans of the same MultiNotch isolation precursors, allowing direct
comparisons. As reporter ions are susceptible to unintended fragmentation and scattering by high-energy collisions, we
photoactivated product ions and prevented successive dissociation of generated reporter ions with ion parking, which
altogether boosted reporter ion yield by up to 55%. We show that using IRMPD to generate reporter ions from ten
synchronously isolated product ions results in a 2.4-fold increase in reporter intensities, significantly enhancing the
sensitivity and dynamic range of quantitation via isobaric tagging. We describe the inefficiency of HCD when activating
individual peptide fragments and when synchronously activating multiple precursors and demonstrate IRMPD boosts
reporter generation by up to an average of 2.4-fold, without perturbing expected quantitative ratios.
Page 14 of 166
Contact: Trenton Peters-Clarke, [email protected]
In both data dependent (DDA) and data independent acquisition (DIA) workflows, the ability to collect high quality MS/MS at
fast acquisition rates is key to maximizing peptide and protein identifications and obtaining highly reproducible
quantification. Novel Zeno trap functionality can greatly improve duty cycle in the orthogonal pulsing region of a QTOF
system, providing large gains in MS/MS sensitivity. This work investigates the impact of Zeno trap MS/MS sensitivity
increases on protein and peptide workflows, at the very fast LC-MS run times offered by microflow chromatography.
IS01.02 Automated Proteomic Sample Preparation: The Key Component for High Throughput and Quantitative
Mass Spectrometry Analysis
Qin Fu, Smidt Heart Institute
There is a great need in high-throughput workflows for reproducibility and precision during proteomic sample preparation for
mass spectrometry-based protein quantification, and large-scale LC-MS analysis. Proteomic sample preparation consists of
protein denaturation, cystine reduction and alkylation, and trypsin digestion carried out in a temperature-controlled
incubator, with salts and contaminants removed prior to MS analysis. In particular, the steps around the digestion of protein
samples are prone to inconsistency. One route for reliable sample processing is the development and optimization of a
workflow utilizing an automated liquid handling workstation. When combining robust liquid chromatography-mass
spectrometry with either discovery or targeted methods, automated sample preparation facilitates increased throughput and
reproducible quantitation of biomarker candidates.
OA02.01 ‘Think Big, Act Bigger”: Opportunity and challenges of high throughput plasma proteomics
Benoit Fatou (1), Hanno Steen (1)
(1) Boston Children's Hospital, Boston, Massachusetts, United States
Blood is widely used to diagnose disease or monitor health because blood is an ideal body fluid: it is a systemic body fluid,
easily and readily accessible using minimally invasive collection procedures. However, wide variations in measured blood
biomarkers across populations require the need of large number of subjects at different states of the immune response for
meaningful clinical proteomics studies. In addition, as multiple measurements are being made from the same samples
including metabolome and transcriptome, sample amounts can be limited particularly for very young population such as
newborns. Given these constraints, we have been developing robust sample-sparing high throughput sample processing
methods in combination with state-of-the-art instrumentation to enable large scale plasma/serum proteomic studies using
submicroliter volumes of sample. We have developed a targeted and multiplexed LC/MS method, on a Shimadzu
LCMS8060 triple quadrupole mass spectrometer to monitor the human classical plasma proteome in a high throughput
manner (<15 min per sample). We are focusing on those classical plasma proteins needed for a comprehensive description
Page 15 of 166
of the immunophenotype of humans under a wide range of immune challenges such as infection, inflammation and/or
vaccination. Our current plasma proteomics pipeline allows for the complete processing and analysis of up to 600 plasma
samples within a week. This method has been thoroughly tested by application to large-scale studies within the Human
Immunology Project Consortium (HIPC) in the context of newborn immune development, providing unique insights into the
maturation of the immune system during the first week of life (DOI: 10.3389/fped.2020.610461i) and the IMmuno
Phenotyping Assessment in a COVID-19 Cohort (IMPACC; DOI: 10.1126/sciimunol.abf3733) study, allowing us to better
understand the immune determinants associated with positive and negative outcomes from COVID-19. These studies have
resulted in the successful processing and analysis of >5,000 plasma samples (and counting).
OA02.02 Fully Automated Proteomic Sample Preparation for the iPSC Neurodegenerative Disease Initiative (iNDI):
A Large-Scale Effort at the NIH
Luke Reilly (1), Erika Lara (1), Lirong Peng (1), Daniel Ramos (1), Caroline Pantazis (1), Julia Stadler (1), Marianita
Santiana (1), Faraz Faghri (2), Mike Nalls (3), Steven Coon (4), Priyanka Narayan (1), Andrew Singleton (1), Mark Cookson
(1), Michael Ward (1), Yue A. Qi (1)
(1) NIH, Bethesda, Maryland, United States
(2) Data Tecnica International, Glen Echo, Maryland
(3) Data Tecnica International, Glen Echo, Maryland, United States
(4) NIH, Bethesda, Maryland
The induced pluripotent stem cell (iPSC) Neurodegenerative Disease Initiative (iNDI) is a large-scale effort to characterize
iPSC-derived neurons harboring one of 132 Alzheimer’s-related variants across 73 genes using multi-omics approaches,
openly providing both the lines and associated data to the scientific community. Such a large-scale and long-term project
requires consistent and efficient workflows in sample preparation and data acquisition for proteomics analyses. Here,
though combining liquid- and bead-handling systems, we report a fully automated proteomic sample preparation pipeline in
96-well format. This involves direct protein collection and reduction in 96-well tissue culture plates using high-percentage
detergent SP3 lysis buffer, automated protein and peptide concentration assays and normalization using an Agilent Bravo
liquid handler, and a customized 30-minute SP3 paramagnetic bead protocol on the Kingfisher APEX platform.
To test this pipeline, we compared three neuron differentiation protocols: overexpression of transcriptional regulator
neurogenin-2 (NGN2), co-expression of NGN2 with homeobox protein EMX1, and the NGN2-EMX1co-expression strategy
with CultureOne supplement. The iPSCs were differentiated into mature neurons using the three protocols (n = 6) and
processed via our automated pipeline. We consistently identified more than 45,000 peptides and 7,500 protein groups from
as few as 3000 cells per replicate using single-shot DIA analysis. The CVs of biological replicates were less than 15%, and
more than 90% of proteins were quantified across all replications, indicating high reproducibility and low missing values.
Our results indicated upregulated cortical neuron markers (e.g., BCL11A) and downregulated cholinergic neuron markers
(e.g., CHAT, SCL18A3) when comparing NGN2-EXM1 to the NGN2 condition. The CultureOne condition resulted in
increased synaptic markers (e.g., MAP2 and MAPT), but higher expression of cholinergic markers. Importantly, we validated
these observations using single-cell RNA sequencing datasets from the same cell lines. Together, we provide an automated
MS sample preparation protocol from sample collection to LC-MS/MS.
Page 16 of 166
Page 17 of 166
Parallel Session 03: System Suitability in Mass Spectrometry
IS03.01 Lindsay Pino Presentation
Lindsay Pino, Talus Bio
IS03.02 Presentation
Skyline introduced metabolomics and small molecules features into the software in 2017 and since then has shown to be a
powerful tool for targeted metabolomics. However, system suitability monitoring in metabolomics and small molecules
application by Skyline is relatively limited compared to proteomics applications. Therefore, we would like to share a workflow
utilizing Skyline and AutoQC for monitoring system suitability for a LC-MS/MS metabolomics platform. One challenging
aspect of system suitability in metabolomics applications is that the analytical column and the mobile phase system change
so frequently, especially in laboratory settings where multiple classes of compounds are analyzed. A dual LC column setup
is used in this scenario as one column is designated to only analyze the specific standard mixture and the other column is
for sample analysis. This setup eliminates the variables introduced from samples and makes system suitability in
metabolomics more consistent within the laboratory. The raw file of system suitability mixtures is saved in a folder that is
automatically scraped by AutoQC and uploaded to an online portal called Panorama. Using this web portal and an
accompanying Skyline document, the user can check attributes such as mass accuracy, retention time, and peak area to
evaluate if the sample analysis can be continued in this system. AutoQC can also be used as a tracking tool to monitor the
system and column performance over time, especially when multiple sets of experiments are completed using the same
column. Though mainly demonstrated previously for proteomics applications, Skyline and AutoQC together also serve as a
powerful open-source tool to perform system suitability of metabolomics LC-MS/MS applications.
OA03.02 HeLa Digest Mixed with Synthetic Standard Peptide Reference Is Used to Monitor System Suitability in a
Data-Dependent Acquisition Proteomics Workflow
Leonard Collins (1), Taufika Islam Williams (1), Jaclyn Kalmar (1), Michael Bereman (1), David Muddiman (1)
(1) North Carolina State University, Raleigh, North Carolina, United States
The goal of this study was to design a procedure to define system suitability for discovery proteomics that uses
mathematically derived thresholds to judge LC-MS system performance instead of relying on arbitrary limits. We currently
Page 18 of 166
evaluate commercially available HeLa digest using data-dependent acquisition (DDA) and bovine serum albumin digest
using parallel reaction monitoring (PRM). We set an arbitrary benchmark for non-specific, total system performance by
noting the number of protein identifications (IDs) obtained with Proteome Discoverer 2.4 (PD) for HeLa injections and track
chromatography and mass spectrometer output changes with BSA injections. There is a need for a single mechanism
specific to DDA that provides relevant system suitability information. We designed an experiment to reveal the point at
which deteriorating system performance would unacceptably impact final results generated by PD. Standard HeLa digest
was combined with Promega 6 x 5 Peptide Reference Mix and serially diluted to mimic a gradual instrument failure. Data
were collected using DDA-only experiments, and reference peptide peak intensities were correlated to the number of
peptide spectral matches (PSMs) and protein identifications found by PD. A logarithmic decay relationship of the type
y=A(1-e-kx) explained decreasing PSMs and IDs with decreasing peak intensities. Reference peptide peak intensity
thresholds were calculated from these relationships and new quality control plots were created that visually identify failure
points. The 6 x 5 reference peptides were also used to monitor spectrometer mass accuracy, signal dynamic range,
retention time stability, and chromatographic peak widths. Combining HeLa digest and Promega 6 x 5 peptides provided a
mechanism to assess chromatographic and mass spectrometric performance and predict expected PD outputs in a single
DDA experiment.
Page 19 of 166
Parallel Session 04: Functional Glycomics
IS04.01 Heparan Sulfate 6-O-endosulfatases in HNSCC and Other Malignancies
Radoslav Goldman, Georgetown University
Post synthetic editing of heparan 6-O-sulfation by the human endosulfatases SULF1 and SULF2 regulates extracellular
matrix remodeling, growth factor signaling, or chemokine distribution. These pathways modulate outcomes of cancer
diseases and patient responses to therapy. In this study, we carried out a translational study of multiple cancer diseases
which confirms the impact of the 6-O-endosulfatases on the progression of head and neck squamous cell carcinoma
(HNSCC). We optimized expression of the SULF enzymes in mammalian cell lines and began characterization of their
structure, activities, and function in tumor biology. We expect that the advances will facilitate detection and targeting of the
enzymes in HNSCC and other malignancies.
Goal: Identify a serum biomarker(s) of hepatocellular carcinoma that originate directly from cancer tissue.
Methods: Formalin-fixed human liver tissue from patients with healthy livers, cirrhotic livers, or hepatocellular carcinoma
(HCC) were evaluated using N-glycan MALDI imaging mass spectrometry. In total, 239 HCC tissue samples, many which
were genetically subtyped, and 145 control tissue samples were analyzed. Subsequently, serum glycoproteomics was
performed and proteins identified as containing the same glycans observed in HCC tissue were further examined in five
independent and cohorts consisting of 1,071 patients, 492 with liver cirrhosis and 472 with hepatocellular carcinoma,
including 270 with early-stage HCC, and 51 patients treated for the HCC, either by transplant or resection.
Results: Through tissue based glycan imaging, increased levels of fucosylation was observed in 96% of the HCC tissues
examined. Glycan alterations in tissue correlated with the underlying genetic subtype of the cancer. The glycan most often
observed in the tissue of aggressive HCC was a tetra-antennary glycan with one to three fucose residues. Using a
recombinant AAL lectin with increased affinity towards fucosylated and branched glycan, we identified low molecular weight
kininogen (LMWK) as a serum protein that contains tetra-antennary glycan with one to three fucose residues. A plate based
lectin ELISA assay was used to test the performance of fucosylated LMWK as a biomarker of HCC. In five independent
studies, a fucosylated LMWK based biomarker algorithm had a median AUC of 0.9575 in the detection of all HCC and 0.945
for early stage HCC, far exceeding the performance obtained currently used markers.
Conclusions: Using a novel tissue based glycan imaging platform, we were able to identify glycan changes that occur
directly in cancer tissue, identify glycans associated with specific subtypes of HCC and used this information to identify
serum based biomarkers of HCC that are far superior to those currently used.
Page 20 of 166
Detecting changes in the N-glycan profiles of biofluids as a reflection of disease state can potentially identify new biomarker
candidates. The goal of this study was to detect alterations in the serum and urine N-glycan profiles of lupus nephritis (LN)
patients, as well as differences between responders and non-responders of LN treatment. Forty donor matched urine and
serum samples from LN and non-autoimmune control subjects were used. Urine samples were concentrated, and buffer
exchanged with molecular weight centrifuge filters. An amine-reactive slide was used to immobilize serum and urine
glycoproteins from samples spotted on to the slide. The samples were delipidated and desalted with washes of Carnoy's
solution and water. Peptide N-glycosidase F was sprayed on the samples to cleave the N-glycans from the captured
glycoproteins, and a layer of matrix is applied. The released N-glycans were detected across the samples by MALDI-QTOF
imaging mass spectrometry (IMS). The intensities of the detected N-glycans and structural classes, such as fucosylation,
sialylation, and sulfation, were compared using SCiLS Lab software. Linear regression modelling was used for detection of
significant differences between sample groups. This recently developed workflow allowed for the detection of 70 N-glycans
in serum and 98 N-glycans in urine. Urine samples from LN patients displayed significant glycan differences from non-LN
individuals. Specifically, high-mannose glycans were lower in LN patients, while a triantennary sialylated N-glycan was more
abundant. Several of these alterations were also detected in the serum samples. While there were no differences in the
intensities of the N-glycans across the responder and non-responders to LN treatment, there were a series of hybrid N-
glycans that were present in nearly all responders but largely absent in the non-responders. Overall, serum and urine N-
glycan profiling revealed significant differences between LN patients and non-autoimmune patients, as well as between
treatment responders and non-responders.
OA04.02 Developing isotopically labeled calibrants for quantification analysis of heparan sulfate from biological
samples
Zhangjie Wang (1), Katelyn Arnold (1), Yongmei Xu (1), Jian Liu (1)
(1) University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States
Heparan sulfate (HS) is a sulfated polysaccharide that displays essential physiological functions. The amount of HS and its
structural information are critically important for investigating the functions of HS in vivo. Disaccharide compositional
analysis is the most widely used technique for the conformation of HS structure and total content determination as the
analysis is capable of handling saccharide sequence heterogeneity. In addition to eight common disaccharide compositions,
the 3-O-sulfated glucosamine in HS is also a key saccharide unit for the biological activities of HS. Here, two types of 13C-
labeled calibrant standards were prepared using chemosynthetic method: eight 13C-labeled disaccharide calibrants for the
disaccharide composition analysis, five 13C-labeled 3-O-sulfated oligosaccharides for 3-O-sulfation analysis of HS. We
performed a LC-MS/MS-based method to conduct disaccharide quantitative analysis of HS from biological sources including
fluids, tissues, and tissue slides. Using the 13C-labeled 3-O-sulfated oligosaccharides as internal standard, we determined
the levels of 3-O-sulfation in the pharmaceutical heparin, analyzed the clearance rate of enoxaparin from a mouse model,
and probed and quantified the specific 3-O-sulfated tetrasaccharides in the Alzheimer’s disease patients. The crucial
innovation in this method is to utilize a set of eight 13C-labeled disaccharide calibrants and oligosaccharides as internal
standards. The combination of both disaccharide calibrants and 3-O-sulfated oligosaccharide calibrants provides the
sensitivity and throughput to study dynamic changes of heparan sulfate in a wide array of biological samples
Page 21 of 166
Parallel Session 05: Biomarkers and Precision Medicine
IS05.01 Tony Hollingsworth Presentation
Tony Hollingsworth, University of Nebraska Medical Center
IS05.02 Glycomics and Glycoproteomics Enabling the identification of Glycan and Glycopeptide isomers as
Biomarkers for Cancer and Neurodegenerative Diseases
Yehia Mechref, Texas Tech University
Cancers and neurodegenerative diseases have an essential impact on human lives with high mortality and poor prognosis
in the early stage. Several biomarkers such as CA125 and tau have been utilized to improve the prognosis of these
diseases. However, the limited prediction accuracy is demanding to explore more reliable markers. Glycosylation is one of
the most common post-translational modifications in mammalian cells and influences a variety of essential biological
functions. Aberrant glycosylation and its isomeric distribution have been associated with many neurodegenerative diseases
such as neurological disorders, Alzheimer’s disease, Parkinson’s disease, and various cancers. However, the glycomics
and glycoproteomics methods for the isomeric glycans and glycopeptides for clinical samples remain a great challenge due
to the complex microheterogeneity of glycoproteins. Over decades, different methods have been developed to facilitate the
identification and quantitation of glycan and glycopeptide isomers. Coupling to a variety of separation techniques, mass
spectrometry has been considered the most powerful tool in isomeric glycomics and glycoproteomics due to its high
resolution, high sensitivity, and the ability to acquire sufficient structural information. Additionally, multiple sample
preparation strategies and automated software were developed to improve the identification and quantitation accuracy and
throughput for large clinical cohorts. During the past ten years, we have developed efficient methods for both glycomics and
glycoproteomics isomeric analysis using C18-, porous graphitized carbon (PGC)-, mesoporous graphitized carbon (MGC)-,
or micropillar array column (µPAC)-LC-MS/MS under different acquisition techniques, combined with several reliable
sample derivatization and multiplexing strategies. Together, these advanced technologies and approaches enhance the
pace of isomeric glycan and glycopeptide biomarker discovery. Here, we review recent glycomics and glycoproteomics
methods for better isomeric characterization and their applications in cancers and neurodegenerative diseases.
OA05.01 Urinary Extracellular Vesicle Phosphoproteome Profiling by DIA Mass Spectrometry for Renal Cell
Carcinoma Subtype Differentiation
Marco Hadisurya (1), Zheng-Chi Lee (2), Anton Iliuk (3), Ronald Boris (4), W. Andy Tao (1)
(1) Purdue University, West Lafayette, Indiana, United States
(2) West Lafayette Junior/Senior Highschool, West Lafayette, Indiana, United States
(3) Tymora Analytical Operations, West Lafayette, Indiana, United States
(4) Indiana University School of Medicine, West Lafayette, Indiana, United States
Extracellular Vesicles (EVs) are promising sources to develop disease phosphoprotein markers, based on the knowledge
that phosphorylation events are directly responsible for many signaling pathways at the cellular level. This study focused on
optimizing and applying robust data-independent acquisition using mass spectrometry for phosphoproteomic EVs profiling
that could differentiate renal cell cancer (RCC) subtypes in a non-invasive way. We compared several methods for DIA
Page 22 of 166
quantification, such as using the gas-phase fractionated library, direct DIA, forbidden zones, and several windowing
schemes. After finding the optimized DIA mass spectrometry method, we applied the strategy to identify and quantify urinary
EV phosphoproteome from 57 individuals representing four different urine samples from low-grade RCC, high-grade RCC,
chronic kidney disease (CKD), and healthy control (HC) individuals. For these samples, we utilized the EVtrap capture
method for EV enrichment and PolyMAC for phosphopeptides enrichment. In addition, we will use a bigger sample cohort
with the help of automation for EV isolation to further substantiate the results. After the optimization, we quantified around
1,400 phosphoproteins. Several RNA-related pathways, such as RNA splicing, mRNA processing, and RNA binding, were
significantly upregulated in CKD. Meanwhile, several common cancer-related pathways are upregulated in high-grade RCC
only, but not in CKD and low-grade RCC. Overall, based on our urinary EV phosphoproteins, CKD, low-grade RCC, and
high-grade RCC have unique upregulated phosphoproteins, signaling pathways, upstream kinases, and transcription
factors. These results show that EV phosphoproteome quantification utilizing our optimized EV isolation, phosphopeptides
enrichment, and DIA method could potentially be further developed for RCC subtype differentiation to benefit many newly
diagnosed RCC patients.
OA05.02 Quantification of Brain-Derived CSF Protein Biomarkers by Targeted Mass Spectrometry to Stage
Alzheimer’s Disease Progression
Caroline M. Watson (1), Lingyan Ping (1), E. Kathleen Carter (1), James J. Lah (1), Allan I. Levey (1), Blaine R. Robert
Roberts (1), Nicholas T. Seyfried (1)
(1) Emory University School of Medicine, Atlanta, Georgia, United States
Alzheimer's disease (AD) is the most common form of dementia, with cerebrospinal fluid (CSF) amyloid-beta42 (Aβ), Tau,
and phosphorylated Tau providing the most sensitive and specific biomarkers for diagnosis. However, these diagnostic
biomarkers are not dynamic throughout disease progression, and do not reflect the complex changes in AD brain beyond
plaque and tangle pathology. We recently developed an integrative brain-CSF proteomics platform and discovered CSF
proteomic changes reflective of diverse brain pathophysiologies and disease progression. We report a sensitive,
quantitative, and scalable targeted proteomics assay for novel biomarkers. We analyzed CSF from two separate Emory
Goizueta AD Research Center (ADRC) cohorts using targeted mass spectrometry. As quality controls (QCs), we pooled
CSF from controls having normal Aβ and Tau levels, and AD samples having low Aβ and high Tau levels. QCs were
processed and analyzed identically to the individual CSF samples. Collectively, 280 controls (cognitively normal), 131
asymptomatic (AsymAD) and 282 symptomatic AD, and 68 QC samples were reduced, alkylated, denatured, and
enzymatically digested with endopeptidase LysC and trypsin. Samples were analyzed as a single replicate over
approximately 11 days using a 1290 Infinity II system (Agilent) coupled with a TSQ Altis Triple Quadrupole mass
spectrometer (Thermo Fisher Scientific). Isotopically labeled peptide standards were added for relative quantification by
reporting the area ratios for each targeted peptide in Skyline. We reproducibly detected six proteins previously identified as
brain-based CSF AD biomarkers, representing a diverse range of cell types and pathophysiology. The protein targets are
consistent with previous CSF proteomic studies in their direction of change with CHI3L1, SPP1, YWHAZ, and SMOC1
increased, and NPTX2 and VGF decreased in AD. We also detected changes in protein abundance for SMOC1 and other
targets from control to AsymAD and AD states suggesting these proteins could serve as useful biomarkers for staging AD
progression.
Page 23 of 166
Parallel Session 06: Proteoform Biology
IS06.01 Toxic and Protective Proteoforms of SOD1 Inspire a Novel ALS Therapeutic
Jeff Agar, Northeastern University
Affinity capture top-down MS was used to identify prevalent SOD1 proteoforms in ALS patients and controls. The degree of
toxicity or protection and their structural correlates were assessed using neurotoxicological assays and global- and peptide-
based HDX-MS. These results indicate that oxidative modification of cysteine was toxic, that dissociation of the SOD1 dimer
initiates toxic aggregation, and that biology prevents this toxic oxidation by protective S-thiolation. Together these inspired
the development of a novel class of pharmacological chaperones, cyclic thiosulfinates, which employ S-thiolation-mediated
crosslinking. Cyclic thiosulfinates have 1) drug-like properties, 2) blood-brain-barrier penetration as measured by a novel
top-down pharmacodynamics assay, and 3) stabilize proteins by unprecedented amounts.
IS06.02 New Frontiers in Proteomics – Proteoforms, Proteoform Families, and the Human Proteoform Project
Lloyd Smith, University of Wisconsin-Madison
Proteins are the primary effectors of function in biology, and thus complete knowledge of their structure and behavior is
needed to decipher function. However the richness of protein structure and function goes far beyond the linear amino acid
sequence dictated by the genetic code. Multigene families, alternative splicing, coding polymorphisms, and post-
translational modifications, work together to create a rich variety of proteoforms, whose chemical diversity is the foundation
of the biological complexes and networks that control biology. The term “proteoforms” refers to the specific molecular
forms in which proteins are present in biological systems; only direct analysis of the proteoforms themselves can reveal their
structures, dynamics, and localizations in biological systems.
Remarkably, the current paradigm of proteomics research, “bottom-up” proteomics, is unable to identify proteoforms –
rather, proteins are enzymatically digested into peptides, which are then identified, serving as surrogates for the likely
presence of their parent proteins in the sample. This strategy destroys the information as to what form of the protein the
peptide represents, and thus the critical information needed to identify proteoforms is lost. The entire field of Biology is thus
attempting to understand life in the absence of the ability to understand the molecules that define life. This limitation of
todays technology provides a “grand challenge” to the scientific community, to devise new strategies and approaches that
are able to comprehensively and quantitatively reveal the full breadth of the proteome at the proteoform level. Developing
the technology to accomplish this, building a comprehensive atlas of proteoforms present in human systems, and eventually
deciphering the functional roles they play in normal and disease biology, comprise central elements in the quest to
understand human biology.
OA06.01 Uncovering Dehydroamino Acids within the Capsid and Matrix HIV-1 Proteins
Rachel Miller (1), Rachel Knoener (1), Bayleigh Benner (1), Brian Frey (1), Mark Scalf (1), Michael Shortreed (1), Nathan
Sherer (1), Lloyd Smith (1)
(1) University of Wisconsin-Madison, Madison, Wisconsin, United States
Human immunodeficiency virus type 1 (HIV-1), despite being heavily studied, remains a leading cause of death worldwide.
The development of new and effective therapeutics is limited by an incomplete understanding of the specific mechanisms of
Page 24 of 166
viral infectivity. Through in-depth proteomic analysis of HIV-1 virions, we discovered the novel post-translational modification
of highly conserved serine, cysteine, and threonine residues within the viral Matrix and Capsid proteins to the dehydroamino
acids, dehydroalanine (DHA) and dehydrobutyrine (DHB). Dehydroamino acids are rare, understudied, and until now have
not been observed in HIV viral proteins. We identified and confirmed, via chemical derivatization with glutathione, the
presence of nine dehydroamino acids: two DHA residues within the Matrix protein, as well as three DHA and four DHB
residues within the Capsid protein. We hypothesize these residues are important in viral particle maturation and could
provide valuable insight into HIV infectivity mechanisms. Dehydroamino acids contain an electrophilic alkene moiety which
can form covalent inter- or intramolecular crosslinks with the nucleophilic side chains of lysine, histidine, and cysteine
residues. The ability to form these crosslinks may cause dehydroamino acids to impact HIV-1 function. Existing literature
was investigated to determine the impact of mutagenesis, at the sites of confirmed dehydroamino acid modifications, on
virion production or viral infectivity. For eight of the nine confirmed dehydroamino acid sites, mutagenesis had a detrimental
effect on virion production or infectivity. Preliminary top-down proteomics of HIV-1 virions revealed 11 Capsid proteoforms
containing DHA and DHB residues. Going forward, we will investigate the origin of these dehydroamino acids within the HIV
proteome, further characterize DHA and DHB containing Matrix and Capsid proteoforms, while exploring their functional
roles.
OA06.02 Application of Top-Down Thermal Proteome Profiling to the Analysis of Staurosporine Treated HeLa
Proteome Stability
Kellye Cupp-Sutton (1), Yanting Guo (1), Trishika Chowdhury (1), Ji Kang (1), Eric Moore (1), Si Wu (1)
(1) University of Oklahoma, Norman, Oklahoma, United States
Proteoforms produced as a result of post-translational modifications (PTMs), such as phosphorylation, are critical for cellular
processes. In eukaryotic cells, phosphorylated proteins make up a substantial portion of the proteome with many proteins
demonstrating multiple functional phosphorylated sites. These proteoforms often demonstrate unique functionality in
biological systems and the high throughput study of the functionality of these proteoforms in their native or native-like
environment is valuable in the study of biological pathways and disease. We developed a high throughput top-down thermal
proteome profiling (TD-TPP) method for the direct determination of proteoform thermal stability and examination of
proteoform functionality. We have applied this platform for the thermal profiling of the HeLa proteome and, furthermore,
analyzed the difference in stability of differentially phosphorylated/dephosphorylated HeLa proteoforms produced in the
presence of the broad-spectrum kinase inhibitor Staurosporine. HeLa cell cultures were incubated with 5 uM Staurosporine
(or DMSO vehicle) for 6 hours and were subsequently lysed. Lysates were aliquoted, heated (37-67°C), and centrifuged to
remove the aggregate protein. Label-free quantitative UHPLC-MS/MS analysis was conducted on the remaining soluble
portions. We deconvoluted and examined the thermal stability of more than 300 HeLa proteoforms with a focus on the
effect of PTMs on thermal stability. For example, we identified the human Peptidyl-prolyl cis-trans isomerase protein and
observed increased thermal stability of the biologically active N-terminal acetylated proteoform when compared to the
unmodified proteoform. Furthermore, we observed 71 proteoforms that were differentially expressed between Staurosporine
treated and DMSO vehicle treated HeLa cell cultures. These proteoforms were analyzed for shifts in thermal stability using
single temperature thermal analysis. Overall, we have applied our TD-TPP for the analysis of intact HeLa proteoforms to
observe the effect of PTMs on proteoform stability. Furthermore, we observed the effect differential phosphorylation on
thermal stability of HeLa proteins via treatment with Staurosporine.
Page 25 of 166
Parallel Session 07: Glycoproteomics: New Applications
IS07.01 New Technologies for Exploiting the Human Glycoproteome for Personalized Medicine
Rebekah Gundry, University of Nebraska Medical Center
Cell surface glycoproteins and glycans play critical roles in a range of physiological functions and disease processes, are
valuable drug targets, and may be exploited as biomarkers for precision medicine. Despite their biological relevance and
utility, glycoproteins and glycans are often understudied largely due to technical challenges. This presentation will describe
CellSurfer and glyPAQ, new analytical platforms that enable rapid identification and quantification of cell surface
glycoproteins and glycans from small sample sizes. The application of these new methodologies to address outstanding
questions in cardiac physiology and disease, with an emphasis on precision medicine, will be described. CellSurfer enables
routine discovery of cell surface N-glycoproteins from samples with limited availability (100-1000 μg or 300,000 – 10 million
cells) with >80% specificity. Innovative bioinformatic tools expedite discovery, analysis, annotation, and candidate
prioritization for downstream validation. We applied CellSurfer to cells isolated from cardiac tissue to develop the first cell
type-specific maps of the cell surface N-glycoproteome of adult human cardiomyocytes from five myocardial chambers.
Primary explanted cardiac fibroblasts, cardiac microvascular endothelial and coronary artery smooth muscle cells were also
profiled. New cardiac cell-type specific markers emerged and novel insights into cardiac fibroblast surfaceome dynamics
due to extended culturing were revealed. glyPAQ enables standardized processing of biological samples for quantitative
profiling of native, reduced N- and O-glycan structures by mass spectrometry. glyPAQ is suitable for the preparation of a
broad range of sample complexities, including monoclonal antibodies, cells, tissues, serum, plasma, and urine. We applied
glyPAQ to human serum and tissue to inform new precision medicine strategies for patients at increased risk for
cardiovascular disease due to rheumatoid arthritis and generate new insights into the impact of COVID-19 infection on the
human heart.
Viral spike proteins are in general heavily glycosylated proteins that bind host receptors to facilitate host cell invasion.
These spike proteins are often the primary immunogens utilized to develop neutralizing antibodies and vaccines.
Furthermore, the host cell surface receptors themselves are also glycoproteins. Thus, a molecular level understanding of
the glycosylation of the spike protein protein of viruses and of the cell surface receptors of hosts are key to understanding
initial steps of the infection cycle as well as developing therapeutics. In this talk, we will primarily focus on the SARS-CoV-2
Spike protein and the human angiotensin converting enzyme 2 (ACE2) receptor but will also refer to other spike
glycoproteins such as that found on HIV-1.
The SARS-CoV-2 virus is responsible for the COVID-19 pandemic that has ravaged the world population for the last 2+
years. This betacoronavirus utilizes a heavy glycosylated trimer spike protein to bind to the ACE2 glycoprotein to facilitate
host cell entry. We utilized a glycomics-informed glycoproteomic approach to determine site-specific microheterogeneity at
all 22 sites of N-linked glycosylation for a stabilized recombinant trimer Spike mimetic immunogen and for a soluble version
of human ACE2. When combined with bioinformatics and molecular dynamic simulations, our results illuminate roles for
glycans in masking viral epitopes as well as directly participating in modulating the viral spike-host receptor interaction.
Page 26 of 166
OA07.01 Characterizing Aberrant IgG N-glycans During Acute and Post-treatment Lyme Disease
Ben Haslund-Gourley (1), Mary Ann Comunale (1), Anand Mehta (2), Stéphane Grauzam (2)
(1) Drexel University College of Medicine, Philadelphia, Pennsylvania, United States
(2) GlycoPath, Charleston, South Carolina, United States
There are an estimated 476,000 cases of Lyme disease (LD) in the USA each year. Acute LD is confirmed through a two-
tier serological assay using Borrelia burgdorferi (Bb) antigens. This testing system has a low sensitivity (46%) and LD
patients remain antibody reactive which negates the ability to differentiate between current and past infections serologically.
When LD is not treated within the acute phase, the spirochete can disseminate into synovial, cardiac, and neuronal tissue
resulting in long-term disability and possibly death. A subset of patients suffer from Post-Treatment Lyme Disease
Syndrome (PTLDS) which is poorly elucidated due to a lack of diagnostic tools. Glycosylation, a known regulator of
antibody response, remains uninvestigated in LD. We hypothesize that characterizing the N-glycans of IgG can improve
upon the current two-tier assay and offer insight into mechanisms of Bb immunological evasion. HPLC analysis of acute LD
IgG N-glycans reveals statistically significant decreased GlcNAcylation and increased galactosylation and sialylation. This
result was unexpected; many inflammatory diseases induce the opposite trend of IgG N-glycans during acute IgG
responses promoting pathogen clearance. The N-glycans marking the acute phase of infection are slow to return to baseline
levels over 70 days post-antibiotic treatment. To further interrogate our findings, total IgG from serum was captured via an
antibody array and analyzed using a novel high-throughput MALDI-FT-ICR method under development. This parallel
analysis of total IgG via MS aligned with the trends and significant alterations observed in HPLC in a fraction of the time
previously required and is a promising clinical diagnostic technique. This report of altered IgG N-glycans in LD is the first in
the field. Future experiments will examine the ability of IgG glycovariants to improve testing sensitivity and as a biomarker
for differentiating past infections and diagnosing PTLDS using the high-throughput MALDI-FT-ICR and Lectin ELISA.
OA07.02 Elucidating the Structural Variations and O-Glycoform Heterogeneity of SARS-CoV-2 S-RBD Variants by
Top-Down Mass Spectrometry
David Roberts (1), Morgan Mann (1), Allan Brasier (1), Song Jin (1), Ying Ge (1)
(1) University of Wisconsin-Madison, Madison, Wisconsin, United States
The novel zoonotic 2019 coronavirus disease (COVID-19) global pandemic has led to over 5 million deaths as of November
2021. SARS-CoV-2 utilizes an extensively glycosylated spike (S)-protein that protrudes from the viral surface to bind
receptor angiotensin-converting enzyme 2 (ACE2) for cell entry. The S glycoprotein initiates pathogenesis and is the main
target for therapeutic design. S-protein glycans alter viral function and infectivity, and enormous attention has been recently
focused on deciphering these complex glycan structures. However, the structures and glycan heterogeneity of the S protein
regional-binding domain (S-RBD) and ACE2 O-glycans remain cryptic because of the immense challenges in intact O-
glycoform analysis by conventional “bottom-up” glycoproteomic approaches, which cannot reveal their structural diversity.
Here we report a multipronged top-down mass spectrometry (MS) approach for the structural elucidation of intact S-
RBD/ACE2 O-glycan proteoforms for the first time. This hybrid top-down MS approach is performed by combining trapped
ion mobility spectrometry (TIMS), using a Bruker timsTOF Pro, to reveal and characterize native structural heterogeneity,
with ultrahigh-resolution Fourier transform ion cyclotron resonance (FTICR)-MS, using a 12T solariX FTICR, to elucidate the
intact O-glycoforms. We removed the N-glycans from the S-RBD/ACE2 using PNGase F treatment to minimize the
interference posed by N-glycan heterogeneity. We next used TIMS to separate and analyze the various O-glycoforms. This
Page 27 of 166
native TIMS-MS method will be used to reveal the conformational heterogeneity of S-RBD/ACE2 O-glycan structures and
probe the influence of various O-glycoforms on TIMS collision cross section (TIMSCCSN2). A 12T FTICR-MS equipped with
collisionally activated dissociation (CAD) and electron capture dissociation (ECD) fragmentation modes will be used to
isotopically resolve the intact O-glycoforms for MS/MS characterization. Finally, we will further combine this hybrid top-down
MS strategy with functional studies to reveal new insights into the structure-function roles of S-RBD/ACE2 O-glycoforms in
viral interactions.
Page 28 of 166
Parallel Session 08: Innovation in Mass Spec Imaging Omics
IS08.01 Ambient Mass Spectrometry Imaging Omics Using Nanospray Desorption Electrospray Ionization (nano-
DESI) Mass Spectrometry
Julia Laskin, Purdue University
Mass spectrometry imaging (MSI) is a powerful technique for studying the localization of lipids, metabolites, and proteins in
biological samples. We have developed nanospray desorption electrospray ionization (nano-DESI), an ambient ionization
technique that relies on localized liquid extraction of analyte molecules from the sample into a liquid bridge formed between
two glass capillaries. The extracted analytes are transferred to a mass spectrometer inlet and ionized by electrospray
ionization. Nano-DESI enables quantitative imaging of biomolecules in fully hydrated samples with minimal or no sample
pre-treatment. Recent developments in the nano-DESI MSI instrumentation have enabled quantitative imaging of lipids,
metabolites, and proteins in tissues with high sensitivity and spatial resolution down to 8-10 microns using finely pulled
capillaries. Furthermore, we have developed a microfluidic nano-DESI probe, which greatly simplifies the experimental
setup and demonstrates similar performance to the capillary probe. Correlative omics imaging of small biomolecules and
proteins provides unique insights into biochemical pathways in biological systems.
IS08.02 Highly multiplexed, multiomic and multimodal tissue imaging using novel photocleavable mass-tags
Mark Lim, AmberGen, Inc.
While the advent of mass spectrometric imaging (MSI) has extended the metabolomic/proteomic capabilities of mass
spectrometry to the spatial dimension, it is generally limited to untargeted analysis of small molecules and proteolytic protein
fragments. We have developed a novel spatialomic approach based on photocleavable mass-tags (PC-MTs) for facile
labeling of probes including antibodies, lectins and nucleic acids which enables highly multiplexed MSI of targeted
macromolecules in tissues. This approach, termed MSI-IHC, significantly exceeds the multiplexity of both fluorescence and
previous cleavable mass-tag based methods. In addition, when combined with direct label-free MSI of small biomolecules, it
provides an integrated workflow to study the spatial distribution and interaction of small molecules and larger
macromolecules, such as intact proteins, in a single tissue specimen. High-plex MSI-IHC has been achieved on brain, tonsil
and cancer tissues using a variety of probe classes, reflecting the known molecular composition, anatomy and pathology of
the targeted biomarkers. Novel dual-labeled fluorescent-PC-MT-probes extend the utility of this new approach to multimodal
imaging. Overall, MSI-IHC holds significant promise for use in the fields of tissue pathology, tissue diagnostics, therapeutics
and precision medicine as well as in research aimed at understanding the mechanisms of human disease.
OA08.01 Combined sialyl-Lewis tumor antigen immunohistochemistry, chemical labeling and N-glycan imaging
mass spectrometry for targeted tissue glycoproteomics
Richard Drake (1), Xiaowei Lu (1), Colin McDowell (1), Grace Grimsley (1), Sarah Pippen (1), Peggi Angel (1)
(1) MUSC, Charleston, South Carolina, United States
Carbohydrate tumor antigen immunohistochemistry (IHC), represented by sialyl Lewis A/CA19-9 and sialyl Lewis X, has
been evaluated by pathologists for decades for diagnostic tumor biomarkers in tissues. Our goal was to develop an
integrated workflow that combines the molecular localization features of IHC and N-glycan imaging mass spectrometry
(IMS) on the same slide with targeted chemical labeling approaches for identification of tumor associated
Page 29 of 166
glycoproteins. Using commercially available antibodies to sialyl Lewis A and sialyl Lewis X antigens, a series of formalin-
fixed paraffin embedded tumor tissues from prostate, colon, and pancreas were processed for standard IHC analysis. Each
slide was scanned by high resolution microscopy, followed by removal of the coverslip. Next, an antigen retrieval heating
step was done to dissociate bound antibodies, followed by spraying of a molecular coating of PNGase F PRIME for release
of N-glycans. N-glycan IMS was done using a Bruker MALDI timsTOFfleX instrument, and the IHC-stained slide is already
co-localized to the resulting glycan images. For both sialyl Lewis antigens, co-localized glycans in the stained tumor regions
are multi-sialylated and multi-fucosylated branched structures. This workflow can further be extended by targeting the alpha-
2,3 linked sialic acids that comprise the sialyl Lewis A/X motif. Prior to PNGase F digestion, the tissues can be treated
chemically with an amidation reaction for alpha 2,6 isomers, followed by use of a biorthogonal amine-azide that specifically
targets alpha-2,3 sialic acids. The N-glycan and IHC stained images can be used to scrape off regions of interest directly
from the tissue. The lysate can be used for chemical affinity enrichment of the alpha-2,3 sialylated glycoproteins using a
sialic acid bead enrichment reaction (SABER) by conjugation to an alkyne-bead. Alternatively, a sialic acid fluorescence
enhancement reaction (SAFER) can be done in intact tissues to obtain sialic acid distributions by fluorescence at single cell
resolution.
OA08.02 Mapping Glutathione Biosynthesis Heterogeneity via Multiple Infusion Start Times (MIST) and IR-
MALDESI-MSI
Allyson Mellinger (1), Kenneth Garrard (2), Zahid Rabbani (3), Michael Gamcsik (3), David Muddiman (2)
(1) FTMS Laboratory for Human Health Research, Department of Chemistry, North Carolina State University, Raleigh, North
Carolina, United States
(2) FTMS Laboratory for Human Health Research; Molecular Education, Technology and Research Innovation Center
(METRIC); Department of Chemistry, North Carolina State University, Raleigh, North Carolina, United States
(3) University of North Carolina/North Carolina State University Joint Department of Biomedical Engineering, North Carolina
State University, Raleigh, North Carolina, United States
Elevated levels of glutathione are characteristic to drug resistant tumor tissues. These tissues are highly heterogeneous and
contain many different cellular microenvironments that may affect the observed disruption in glutathione abundances. These
effects and the mechanisms involved require further elucidation. To map this heterogeneity two-dimensional sections of
tissue, we developed a technique combining a stable isotope labeling tracer method and mass spectrometry imaging. While
tracer studies are not uncommon to study biosynthesis kinetics, we uniquely employ a multiple infusion start time (MIST)
method to monitor the incorporation of three glycine isotopologues into glutathione over the course of infusion time. We also
use infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) coupled with a high resolution/accurate
mass Orbitrap mass spectrometer to measure the degree of incorporation of the isotopologues across tissue sections. This
platform allows us to ionize and fully resolve signals from the glutathione isotopologues without chemical derivatization. We
then quantify the degree of incorporation in each analyzed tissue voxel using the newly developed percent isotope
enrichment (PIE) tool within MSiReader. This tool first computes the percent isotope enrichment after correcting
abundances for overlapping isotopic distributions within each voxel. It then generates heatmaps for visualization of flux
heterogeneity across the tissues. This method has been used successfully to observe approximately 2-6% total isotopic
enrichment within mouse liver tissue. Current efforts are focused towards adapting this protocol for studying tumor tissues
and towards further developing the PIE tool for the computation of kinetic curves. Preliminary results in studying tumor
Page 30 of 166
tissues reveal visible differences in isotopic enrichment across 4T1 mammary tumors in mice, which provide more dynamic
information than heat maps of single time point measurements of glutathione molecules.
Page 31 of 166
Parallel Session 09: Microbiome Multi-omics
IS09.01 Ran Blekhman Presentation
Ran Blekhman, University of Minnesota
Researchers have established substantial evidence regarding the impact of the gut microbiome on human health and the
potential for its use as a diagnostic tool and target for therapeutic interventions. Many gut microbiome studies depend on the
analysis of stool samples, which present a complex matrix composed of microbes, protein, undigested plant matter and fat
content that is strongly influenced by diet. Given the complexity of the gut microbiome (microbes and metabolites) and its
importance for human health, the gut microbiome community has recognized a need to develop standards (e.g. reference
material) to validate analytical workflows, improve reproducibility, and allow comparability among different efforts. To this
end, the National Institute of Standards and Technology (NIST) has undertaken the effort to develop a human stool
reference material to support this community. The development of a reference material, particularly a complex material
such as human stool, occurs in several stages. We will present a detailed overview of the development process including
key considerations for production and characterization of the material. Notably, this material will consist of pooled material,
from two distinct cohorts, and the preparation will be suitable for multi-omic analysis. In addition to the detailed overview of
the material development, we will present preliminary data on the metabolomic and metagenomic characterization of pilot
materials and the results from an interlaboratory study assessing the current state of the field in metabolomic analysis of a
standard fecal sample.
OA09.02 Next-gen Sequencing and Targeted Proteomics Reveal Global Binding and Suppression of the
Herpesvirus Genome by the DNA Sensor IFI16
Timothy Howard (1), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
Human cells express a wide array of pattern recognition receptors (PRRs) to identify invading pathogens and activate
immune signaling pathways. The interferon-inducible protein 16 (IFI16) is a nuclear PRR that recognizes double-stranded
DNA (dsDNA) from a number of viral sources, including genomes of nuclear-replicating viruses (e.g., herpesviruses and
Page 32 of 166
papillomaviruses) and dsDNA intermediates formed during retrovirus (e.g., HIV) replication. Our lab and others have found
that upon binding to the dsDNA genome of herpes simplex virus 1 (HSV-1), IFI16 induces antiviral cytokine expression and
suppresses virus gene expression. Though structural studies have determined IFI16 binds dsDNA in a sequence-
independent manner, the interaction between IFI16 and the 152 kilobase-pair HSV-1 genome remains uncharacterized.
Further, it has been hypothesized that IFI16 promotes the heterochromatinization of the HSV-1 genome to achieve
suppression of viral genes, but the direct effect of IFI16 binding to the virus DNA has not been investigated. Here, we
utilized a multi-omics approach of DNA sequencing techniques paired with targeted mass spectrometry proteomics to obtain
a complete view of the interaction between IFI16 and the HSV-1 genome, and how this binding affects virus DNA structure
and protein expression. Specifically, we used chromatin immunoaffinity purification (ChIP)-sequencing and identified regions
of IFI16 enrichment on the virus genome at loci that are crucial for virus replication. We next examined infected wild-type
and IFI16-knockout cells for differences in virus genome structure and protein expression, using the assay for transposase
accessible chromatin with sequencing (ATAC-seq) to probe IFI16-mediated changes in virus DNA conformation, along with
parallel reaction monitoring-mass spectrometry to measure concomitant changes in the expression of more than 60 HSV-1
proteins. Altogether, we present the most complete view of a nuclear DNA sensor binding viral dsDNA, and we provide
insights into how IFI16 represses HSV-1 replication by modifying the 3-dimensional structure of the virus genome.
Page 33 of 166
Parallel Session 10: Computational Proteomics: From Machine
Learning to Human Insight
IS10.01 Olga Vitek Presentation
Olga Vitek, Northeastern University
OA10.01 Presentation
OA10.02 Creating a global map of cell structure from protein interactions and images
Leah V. Schaffer (1), Yue Qin (1), Edward L. Huttlin (2), Casper F. Winsnes (3), Gege Qian (1), Laura Pontano Vaites (2), J.
Wade Harper (2), Steven P. Gygi (4), Emma Lundberg (3), Trey Ideker (5)
(1) University of California, San Diego, La Jolla, California, United States
(2) Harvard Medical School, Boston, Massachusetts, United States
(3) KTH-Royal Institute of Technology, Stockholm, Sweden
(4) Harvard Medical School, Boston, Massachusetts
(5) University of California, San Diego, La Jolla, California
Introduction Determining the structure of biological systems and its relation to function is one of the ultimate goals of the
biological sciences. There has been a recent push to produce large-sized data sets from two of the different major
approaches for mapping cell structure including the Human Protein Atlas (HPA), a major imaging effort consisting of
>10,000 immunofluorescence (IF) images, and the BioPlex networks, derived from systematic affinity-purification mass
spectrometry (AP-MS) experiments. Integrating the resulting datasets from these two different approaches provides an
opportunity to generate a more complete map of cell structure across scales, from individual protein interactions to
subcellular location of whole complexes. Methods Towards this goal, a recent collaborative effort developed an approach
for creating Multi-Scale Integrated Cell (MuSIC) maps of cellular structure by integrating HPA IF images and the BioPlex
protein interaction network. Pairs of protein embeddings from the two measurement types were integrated using a machine
learning algorithm trained with Gene Ontology similarities to determine pairwise distances between proteins, and
subsequently protein communities were detected from the network of these integrated protein-protein distances. Results
This approach resulted in a MuSIC map for the 661 proteins in HEK-293 present in both the HPA and BioPlex data sets
consisting of 69 subcellular systems, approximately half of which were novel. This map of the cell determined roles for
poorly characterized proteins and identified new protein assemblies in processes such as ribosomal biogenesis and RNA
Page 34 of 166
splicing. We will describe the initial developments to build MuSIC map v2.0 in the U2OS cell line consisting of the ~5,000
proteins that have HPA IF images and are present in the new U2OS BioPlex AP-MS network, representing a ten-fold
expansion of the first MuSIC map in the HEK-293 cell line and thus providing a substantially more global view of cell
structure.
Page 35 of 166
Parallel Session 11: Proteomics and Biomimicry
IS11.01 Comparative Mammalian Proteomics: Increasing the Biomimicry Space alongside Genomes
Michael Janech, College of Charleston/Grice Marine Lab
With over 240 mammalian genome assemblies available through open-source repositories like DNAzoo, Zoonomia, and the
Earth Biogenome Project, scientists are now able to interrogate chromosomal rearrangements, positive gene selection,
copy number, and deletions across mammalian taxa like never before. Differences in genomic architecture due to
evolutionary pressure result in much of the phenotypic diversity witnessed today and give rise to unique physiological
adaptations that have become the topic of study for comparative physiologists. Some of these unique physiological
adaptations hold promise for medical researchers interested in understanding ischemia/reperfusion injury (e.g. dive
response in marine mammals) and uremia tolerance (hibernation in bears) amongst others. Although genome information
is useful for understanding evolutionary relationships and physiological adaptations, differences in gene copy number or
positive gene selection is difficult to translate into protein abundance, especially in biofluids. For this reason, parallel
reaction monitoring was utilized to quantify the serum protein, pantetheinase, in 44 different mammals to determine whether
a high abundance of serum pantetheinase was unique to diving marine mammals. Although levels could not be predicted
from comparative genomics, diving marine mammals in general had relatively high levels of pantetheinase. In addition, the
high pantetheinase phenotype also defined the ungulates (odd and even-toed), a previously unreported observation. To
expedite hypothesis testing across taxa the Comparative Protein Aggregator Resource (CoMPARe) was initiated in 2018 to
examine differences between mammalian taxa. This ongoing venture is collecting proteomic data from serum across 52
mammalian species. Ranked comparisons are being made following “humanization” of the proteomes to capture information
that is not predicted from genomes. Exploration of the proteome adds another layer of information to describe
interrelationships amongst mammals and will help to define the landscape of comparative mammalian physiology from
which biomimicry can be employed to extract novel solutions for humans and advance industry.
IS11.02 Insights into human pathology from proteome studies of extreme adaptations in mammals
Jane Khudyakov, University of the Pacific
Many wild mammals are adapted to tolerate "extreme" environmental and physiological challenges, such as prolonged
fasting, hypoxia, and rapid skin regeneration, among others. Understanding the molecular basis of natural tolerance in wild
mammals may provide insights into the treatment of human pathologies such as obesity and skin disorders. The northern
elephant seal is an excellent non-model "model system" for studies of natural adaptation to prolonged fasting, catastrophic
molting, and hypoxia in mammals. We examined changes in skeletal muscle, adipose tissue, skin, and plasma proteomes of
elephant seals over 5 weeks of fasting associated with rapid molting and identified enzymes, signaling molecules, and
transcriptional regulators of these processes, which are potential targets for further studies of human disease. Our findings
include changes in apolipoprotein composition that may underlie maintenance of naturally high HDL levels and resistance to
oxidative stress during fasting and alterations in extracellular matrix composition and abundance of metabolic enzymes that
may support rapid skin regeneration during catastrophic molting.
OA11.01 Leveraging large-scale comparative proteomics across the tree of life to improve human disease genetics
Rachael Cox (1), Ophelia Papoulas (1), Edward Marcotte (1)
(1) University of Texas at Austin, Austin, Texas, United States
Page 36 of 166
Proteins interacting in the same biochemical complex are often linked to similar genetic traits. Moreover, previous studies
have shown that evolutionarily conserved (ancient) proteins are enriched for disease traits and are abundant across human
cell types and tissues. A significant portion of these deeply conserved genes are known to be responsible for a large and
diverse subset of major human diseases, spanning developmental disorders (e.g., Leigh syndrome, microcephaly, neural
tube defects), cancers (e.g., leukemia, breast cancer, colorectal cancer), chronic respiratory diseases (e.g., ciliary
dyskinesia, asthma), neurological disorders (e.g., encephalopathy, schizophrenia, autism) and motor dysfunction (e.g.,
dystonia, spastic paraplegia). Thus, we sought to better characterize the ancient eukaryotic protein complexes that likely
contribute to human disease by computationally integrating thousands of of mass spectrometry proteomics experiments
spanning an evolutionarily diverse set of 31 eukaryotes. We combine this data with phylostratigraphy and machine learning
to reconstruct a likely set of ancestral eukaryotic protein assemblies. We will present our progress on the analysis and
application of these data to the discovery of disease associations and new functions for poorly characterized human genes.
OA11.02 Are Mouse Models of Alzheimer’s Disease a Surrogate for Human Disease? – Using quantitative
proteomics to monitor time resolved formation of pathology
Kathrin Wenger (1), Arthur Viode (1), Kinga Smolen (1), Christoph Schlaffner (1), Patrick van Zalm (1), Tammy Delovade
(2), Xavier Langlois (2), Anthony Bannon (3), Juri Rappsilber (4), Hanno Steen (5), Judith Steen (1)
(1) Boston Children's Hospital, Boston, Massachusetts, United States
(2) Abbvie, Cambridge, Massachusetts, United States
(3) Abbvie, Cambridge, Massachusetts
(4) Technische Universität Berlin, Berlin, Germany
(5) Boston Children's Hospital, Boston, Massachusetts
While Alzheimer’s Disease (AD) was described and characterized in 1901 by Alois Alzheimer, a cure has remained elusive
121 years later. Most if not all therapeutics are tested preclinically in mice and as such understanding if these mouse
models recapitulate human disease is important to finding therapeutics. Transgene mouse models that overexpress human
Tau with known risk mutations such as P301S or P301L are a widely used tool for Tau drug development in AD as these
models exhibit robust Tau aggregation during disease progression. It is, however, not clear to which extent these models
reflect human Tau pathology and thus AD at the molecular level. An in-depth, mass spectrometry-based proteomic analysis
was applied to study pathological and functional Tau, derived from affected brain tissue of two commonly used mouse
models of AD: the P301S (hTau.P301S)) and P301L (rTg(tauP301L)4510) model expressing mutant Tau forms.
Identification of post translational modifications (PTMs) on Tau and quantification of the extent and changes thereof during
Tau aggregation provides molecular insights into the drivers during the appearance of the pathology. Comparison to human
AD (Wesseling, Mair et al, Cell., 2020) shows similarities, but also highlights the dramatic differences between the human
disease and the preclinical animal models. Our data show that the two mouse models reflect a subset of modifications of
pathological Tau which are observed in human AD. However, many other disease and disease stage associated PTMs that
are important in symptomatic human AD, are not represented in these mouse models. This analysis provides guidance
regarding the extent to which mouse models can be used in preclinical studies of therapeutics. Thus, such in depth
molecular characterization and comparison of mouse models and human disease are crucial in designing mechanistic
studies and preclinical testing of therapeutics for application in human disease.
Page 37 of 166
Parallel Session 12: Proteomics to Advance Equity
IS12.01 Differences in Stromal Patterns from Breast Cancer Metastatic Lymph Between Black Women and White
Women in the South Carolina Sea Islander Population
Peggi Angel, Medical University of South Carolina
In South Carolina (SC), there is a disproportionate increase in female breast cancer death rates in black women (BW)
compared to white women (WW) (30.1 and 21.2/100,000 respectively, years 2000-2019). Historical accounts of the SC
slave trade report that the majority of the enslaved African originated from the sub-Saharan West African Coastal regions.
After slavery, these Africans remained in relative isolation within communities along the south eastern coastal islands, called
Sea Islands (SIs). The West African origins of the Sea Islanders predispose this diasporic population to higher breast
cancer risk and development of more aggressive breast cancers. Contemporary literature reports that stromal collagen
differences are predictive of breast cancer progression and survival. We hypothesized that collagen stroma variations
between SI BW and WW may be involved in disproportionate SC breast cancer outcomes. The study investigated collagen
stromal variations by targeted collagen tissue imaging proteomics in breast cancer tumor, normal adjacent tissue, normal
adjacent lymph, and metastatic lymph tissue. Newly diagnosed patients were from documented SI geographic regions (BW
n=10; WW n=21). Collagen peptide peak intensities were analyzed using Area Under the Receiving Operating Curve, p-
value <0.01 to determine differentiating signatures between BW and WW. Intriguingly, the largest variation when comparing
by race occurred in the lymph nodes. Normal lymph tissue showed 83/1377 significantly different collagen peptides while
the metastatic lymph tissue showed significant changes in 74 collagen peptides, t-test p-value ≤0.001. Certain
peptides were significant both metastatic and normal lymph tissue, whereas others were uniquely significant in either
metastatic or normal lymph. This study suggests that there may be an ancestry-dependent immune involvement to
metastatic breast cancer that could contribute to higher breast cancer mortality rates in BW from SC SI regions. Additional
studies are in progress investigating collagen stroma and immune system involvement in West African-origin breast cancer.
IS12.02 In-Cell Protein Footprinting Coupled with Mass Spectrometry for Structural Biology Across the Proteome
Lisa Jones, University of Maryland
In recent years, protein footprinting coupled with mass spectrometry has been extensively used to analyze the higher order
structure (HOS) of proteins. These methods have been successfully used to identify protein-protein interaction sites and
regions of conformational change through modification of solvent accessible sites in proteins. The footprinting method, fast
photochemical oxidation of proteins (FPOP), utilizes hydroxyl radicals to modify these solvent accessible sites. To date,
FPOP has been used in vitro on relatively pure protein systems. We have further extended the FPOP method for in-cell and
in vivo analysis of proteins. This will allow for study of proteins in their native cellular environment and be especially useful
for the study of membrane proteins which can be difficult to purify for in vitro studies. A major application of these methods
is for proteome-wide structural biology. In one such application, we used in-cell FPOP (IC-FPOP) to identify on and off
targets of the cancer drug methotrexate in leukemia cells. By obtaining structural information on proteins across the
proteome, we were able to distinguish structural changes that occur in response to drug treatment. We have further
extended the FPOP method for analysis in C. elegans, a member of the nematode family. This allows us to study protein
structure directly in animal model for human disease. These methods have the potential to become a powerful tool in the
structural biology toolbox.
Page 38 of 166
OA12.01 Proteomic Changes Associated with Survival Outcomes of Sepsis Patients with Primary Intra-Abdominal
Infection
Kathryn Kapp (1), Albert Arul (1), Kevin Zhang (2), Liping Du (2), Sachin Yende (3), John Kellum (3), Derek Angus (3),
Octavia Peck-Palmer (4), Renã Robinson (1)
(1) Vanderbilt University, Nashville, Tennessee, United States
(2) Vanderbilt University Medical Center, Nashville, Tennessee, United States
(3) University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania, United States
(4) University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania
The plasma proteome can distinguish patient survival outcomes as early as hospital admission. Patient response to sepsis
is highly heterogeneous and depends on primary source of infection and potentially the patient’s racial background. For
example, African American/Black patients have higher mortality rates of sepsis than non-Hispanic Whites. To better
understand acute changes in sepsis that impact survival outcomes, we analyzed plasma samples from a racially diverse
cohort of sepsis patients with primary intra-abdominal (IA) infection. Blood plasma samples (N = 107) from non-Hispanic
White and African American/Black survivors and non-survivors of sepsis secondary to IA infection were obtained from the
Protocolized Care for Early Septic Shock cohort at emergency department admission. Survivorship was defined as a
patient’s status at 60 days post-admission. Plasma samples were immunodepleted, digested using trypsin/Lys-C, tagged
with tandem mass tag reagents, fractionated, and analyzed with high-resolution liquid chromatography-tandem mass
spectrometry (LC-MS/MS). Linear regression models assessed the effects of survival on protein outcome with and without
stratification based on the patients’ self-identified race. In total, 2,540 plasma proteins were identified, and from regression
analysis, 224 of these proteins were differentially-expressed between non-survivors (N = 45) and survivors (N = 62).
Differentially-expressed proteins are primarily involved in the hepatic fibrosis/stellate cell activation pathway, acute phase
response signaling, and the coagulation and complement systems. In race-stratified comparisons, 151 proteins were
differentially-expressed between non-survivors and survivors, 103 of which were unique to one racial group. We identified
eight proteins with a significant race x survival outcome interaction (FDR adjusted p < 0.05) that are involved in the acute
phase response, complement system, and liver X receptor/retinoid X receptor and farnesoid X receptor/retinoid X receptor
activation pathways. This presentation will discuss study findings and implications of the plasma proteome toward
identifying acute changes related to patient survival and racial background.
OA12.02 Resolving racial differences in the brain proteome of African Americans with Alzheimer’s disease
Erica Modeste (1), Lingyan Ping (1), Duc M. Duong (1), Marla Gearing (1), James J. Lah (1), Allan I. Levey (1), Nicholas T.
Seyfried (1)
(1) Emory University, Atlanta, Georgia, United States
Chronic health conditions including cardiovascular disease and diabetes, lower quality and level of education, higher rates
of poverty, and greater exposure to discrimination disproportionally affect African Americans, putting this population at a
heightened risk for Alzheimer’s disease (AD) and related dementias. Neuropathological hallmarks of AD include
accumulations of amyloid-beta (Aβ) plaques and tangles of hyperphosphorylated tau in the brain. Levels of Aβ and tau in
cerebrospinal fluid (CSF) can act as proxies for AD pathology in the brain. Emerging biomarker studies in CSF show,
Page 39 of 166
however, that African Americans have lower levels of tau, compared to Caucasians. This suggests that race may influence
tau and related AD pathophysiology in African Americans. Whether race impacts tau levels or other pathophysiologies in the
brain is not well established. Here, we sought to characterize the brain proteome of Caucasians and African Americans with
AD. Using tandem mass tag mass spectrometry-based proteomics coupled with off-line fractionation, we quantified ~8,000
proteins across 129 matched African-American and Caucasian AD and control postmortem brain tissues. Notably,
characteristic hallmarks of AD, including increases in Aβ peptide and total tau levels in brain, did not differ by race.
Following phosphopeptide enrichment using iron-chelated magnetic beads, approximately 80 tau phosphopetides were
identified and quantified. Four of those tau phosphopeptides were statistically different between Caucasians and Africans
Americans. None, however, fell within the microtubule-binding region most associated with disease and tau tangles.
Furthermore, co-expression network analysis revealed pathways of proteins specific to neuronal, microtubule binding,
vascular, and RNA splicing / binding biology that differed between African Americans and Caucasians, particularly in the
controls. Together, this data suggests that race plays a role in AD, particularly at earlier stages and these differences exist
outside of the pathological hallmark proteins, Aβ and tau.
Page 40 of 166
Parallel Session 13: Proteomics in Drug Discovery
IS13.01 Profiling the Diversity of Agonist Selective Effects on the Proximal Proteome Environment of GPCRs
Ruth Huttenhain, University of California, San Francisco
G protein-coupled receptors (GPCRs) represent the largest family of signaling receptors and drug targets. Activation of
GPCRs can initiate complex signaling networks from different cellular compartments. Depending on the cellular location and
the signaling pathways activated, GPCRs can elicit different cellular effects. This capacity may be exploited in drug
discovery by designing ligands selectively activating specific receptor-mediated signaling pathways or target receptors at
selected cellular locations. However, the efforts to create pathway or location specific ligands is limited by our understanding
of GPCR-based signaling and trafficking mechanisms. One reason for this is the lack of technologies that can capture the
full complexity of GPCR activity.
Recently, we developed a proximity biotin labeling method for GPCRs combining the engineered peroxidase APEX with
quantitative mass spectrometry (GPCR-APEX). GPCR-APEX simultaneously captures proximal protein interaction networks
and cellular location of the activated receptor with high temporal resolution. While this represents a powerful unbiased
method to profile agonist-selective effects on receptor interaction networks and cellular localization, a major limitation of its
application has been to deconvolve this information from the proximity labeling datasets.
Here, we present a novel computational framework which predicts time and agonist dependent subcellular location of the
receptor and quantitatively deconvolutes the effect of receptor location and proximal interactors in the proximity labeling
data. As our model system for developing the framework, we selected the mu-opioid receptor (MOR). Activating MOR with
chemically distinct agonists, we show pronounced differences in the proximal proteome of MOR elicited by their distinct
capacity to initiate receptor trafficking. Moreover, we discovered two novel G protein associated proteins–KCTD12 and
EYA4. We show that recruitment of KCTD12 and EYA4 is dependent on Gɑi activity and suggest that KCTD12 and EYA4
form a buffer system for G protein signaling by interacting directly with the Gβ/γ and Gɑ subunits, respectively.
Mass spectrometry based proteomic technologies are indispensable for small molecule drug discovery programs. For
example, proteomics and chemical proteomics are used, along with other technologies, to identify protein targets from well-
defined phenotypic readouts. Intact protein mass spectrometry (MS) is performed to ensure the appropriate protein
constructs are made, define PTM status, and can be used in covalent inhibitor programs as a screening method. Chemical
proteomic experiments are conducted to assess both on-target and off-target engagement. This presentation will present
practical examples and insights into how proteomic technologies are used throughout the drug discovery pipeline. Also,
featured will be the development and incorporation of a new technology for intact protein and protein characterization
studies, infrared matrix assisted laser desorption ionization mass spectrometry IR-MALDESI-MS.
OA13.01 Unveiling of RF-amide family GPCRs interaction networks and signaling pathways with proteomics for
analgesics development
Qiongyu Li (1), Benjamin Polacco (1), Trupti Patil (1), Aaron Marley (1), Aliza Ehrlich (1), Jiewei Xu (1), Xiaofang Zhong (1),
Page 41 of 166
Nevan Krogan (2), Mark Von Zastrow (1), Ruth Huttenhain (2)
(1) UCSF, San Francisco, California, United States
(2) UCSF, San Francisco, California
Studies on a group of GPCRs, the RF-amide peptide receptors (RFPR), demonstrated their involvement in the regulation of
nociception, which further revealed the potential of the development of analgesics targeting RFPRs. In this study, two
receptors of RFPR family, including NPFFR1 and QRFPR have been studied for their interaction networks and signaling
pathways with mass spectrometry (MS)-based proteomics. To delineate interaction networks of NPFFR1 and QRFPR, an
APEX-based proximity labeling strategy was employed on HEK293T cells stably expressing either receptor. Receptors were
activated with their corresponding agonists for different time courses ranging from 1 to 30 min. To understand further the
signaling pathways downstream of activated QRFPR, the global quantitative phosphoproteomics was conducted with
QRFPR activated by its agonist. For NPFFR1, further study of the regulation of its interaction network by its agonist was
achieved by conducting APEX-based proximity labeling on wild and phosphosite-mutant types of NPFFR1 in parallel. The
APEX-labeling data revealed different interaction networks of NPFFR1 and QRFPR. The data demonstrated that the
activation of QRFPR lead to the labeling of proteins known as subunits of complexes involved in actin filament
polymerization. For NPFFR1, several components of the CUL1-E3 ligase were labeled with the APEX-proximity labeling.
The phosphoproteomics for QRFPR activation generated 4000 significantly regulated phosphorylation sites, with more than
200 sites being regulated in a Gq-dependent manner. Furthermore, the comparative APEX-labeling data of wild and
phosphosite-mutant NPFFR1 revealed the dependence of the recruitment of CUL1-E3 ligase on the phosphorylation of
phosphodegron on NPFFR1. With novel proteomics approaches combining with molecular biology tools, we were able to
reveal the molecular changes downstream of two RFPR family receptors. This leads to the further investigation of molecular
mechanisms of their activation, which will guide the design of novel molecules targeting them for the development of
analgesics.
OA13.02 Identification of Host Cell Proteins Critical to the Hepatitis B Viral Life Cycle
Isabella Whitworth (1), Sofia Romero (1), Abena Kissi-Twum (1), Rachel Knoener (1), Mark Scalf (1), Daniel Loeb (1),
Nathan Sherer (1), Lloyd Smith (2)
(1) University of Wisconsin-Madison, Madison, Wisconsin, United States
(2) University of Wisconsin-Madison, Madison, Wisconsin
Hepatitis B Virus (HBV), a leading cause of liver cancer, chronically infects over 200 million people worldwide, has no
reliable cure, and lacks targetable host factors for development of therapeutics. Identification of new drug targets will require
moving beyond just protein characterization. Physical interactions between biomolecules, including protein and RNA,
provide essential information about viral life cycles. Hybridization-Purification of RNA-protein complexes followed by Mass
Spectrometry (HyPR-MS) is an RNA-centered approach to characterize RNA-protein complexes by purifying target RNA
and performing mass spectrometric analysis to identify bound proteins. Previous research focused on the packaging of HBV
pregenomic RNA (pgRNA) has suggested unknown factors are required for the specific packaging of the genome into
capsids. By looking at the differentially expressed proteins bound to pgRNA in cells transfected with a packaging competent
HBV construct (ε+) and with a packaging incompetent HBV construct (ε-), host proteins specifically associated with genome
packaging can be identified. Hybridization and capture of pgRNA as well as scrambled and mRNA controls was performed
on 4 bioreplicates of cell transfected with ε+ plasmids and 4 bioreplicates of cells transfected with ε- plasmids. Pairwise T-
Page 42 of 166
tests identified 267 proteins with a p-value less than 0.05 that were at least 2X more abundant in cells transfected with the
ε+ construct. This includes proteins that are known to promote HBV capsid formation, catalyze HBV viral protein
dephosphorylation, post-transcriptionally modify the ε loop of pgRNA, as well as proteins known to participate more broadly
in viral regulation. 30 of these proteins have been selected for further evaluation via siRNA knockdowns based on their fold-
change differences and previous identifications in the literature. After host protein knockdowns, the impact on packaging will
be evaluated by comparing the pgRNA packaging efficiency of knockdown and wildtype cells.
Page 43 of 166
Parallel Session 14: Enabling Pandemic Proteomics
IS14.01 Douglas Fraser Presentation
Douglas Fraser, London Health Sciences Centre
The ongoing COVID-19 pandemic caused by SARS-CoV-2 has brought fresh attention to the enormous global health risks
posed by RNA viruses and the need for broad antiviral strategies. Our focus is on host-targeted therapeutics by expanding
our understanding of common host processes critical for virus infections. Comparative mass spectrometry-based
interactomics is a powerful approach to identify and sensitively compare the enrichment of shared and unique host cell
dependencies exploited by virus proteins. We coupled affinity purification to TMTpro-multiplexed quantitative proteomics to
profile the host cell interactions of several nonstructural proteins from SARS, non-pathogenic coronavirus strains, as well as
newly emerging SARS-CoV-2 variants. We identified pan-strain interactions with mitochondria-associated endoplasmic
reticulum (ER) membrane factors and ER proteostasis pathways. As RNA virus replication and assembly frequently occurs
at the ER membrane, viruses extensively remodel this organelle and the ER proteostasis network requiring precise
modulation of the unfolded protein response (UPR). We found divergent roles for two nonstructural proteins, nsp4 and nsp3,
in activating and suppressing the UPR to finetune its activity. To disrupt these host ER proteostasis dependencies during
viral propagation, we can take advantage of recently developed small-molecule ER proteostasis regulator compounds. We
discovered that such compounds are broadly effective at inhibiting infections with flaviruses (dengue and Zika) and we are
currently exploring their use against human and mouse coronavirus strains. These results open up a broadly applicable
therapeutic strategy to target essential host ER protein folding processes to impair viral infections.
Human coronaviruses (CoVs) are a threat to global health and society, as evident from the SARS outbreak in 2002, the
MERS outbreaks in 2012 and 2014, and the most recent COVID-19 pandemic. Despite the sequence similarity between
severe disease-causing CoVs, each strain has distinctive virulence. A better understanding of the basic molecular
mechanisms mediating changes in virulence is needed. Here, we profile the virus-host protein-protein interactions of three
CoV nonstructural proteins (nsps) that are critical for virus replication. We use tandem mass tag-multiplexed quantitative
proteomics to sensitively compare and contrast the interactomes of nsp2, nsp3, and nsp4 from three severe disease-
causing CoV strains (SARS-CoV-2, SARS-CoV, MERS-CoV) and two common-cold CoVs (hCoV-OC43, hCoV-229E). This
approach enabled us to identify both unique and shared host cell protein interactors and quantitatively compare the
enrichment of interactions between homologs. Both nsp2 and nsp4 common interactors are strongly enriched for ER-
mitochondria contact site proteins, suggesting a new functional role for modulating host processes such as calcium
homeostasis and mitochondrial function. Furthermore, using functional genetic screens we show many of these interactors
Page 44 of 166
are pro-viral for CoV infection. Nsp3 homologs showed more strain-specific interactor profiles, including interactions with
nuclear import machinery for hCoV-229E and ribosomal RNA processing for MERS-CoV. Lastly, we find that SARS-CoV-2
nsp3 interacts with ATF6, a regulator of the Unfolded Protein Response, and can suppress the ATF6 stress response. Our
results shed light on the role these CoV proteins play in the infection cycle, as well as host factors that may mediate the
divergent pathogenesis of common cold CoVs from SARS/MERS strains. Our mass spectrometry workflow enables rapid,
robust comparisons of multiple bait proteins and identification of common CoV-host dependencies to be targeted by host-
directed anti-viral therapeutics.
OA14.02 Plasma proteomics of human Ebola virus infection distinguishes and defines outcomes
Arthur Viode (1), Kinga K Smolen (1), Benoit Fatou (1), Zainab Wurie (1), Patrick van Zalm (1), Mandy Kader Konde (2),
Balla Moussa Keita (3), Richard Amento Ablam (4), Eleanor N. Fish (5), Hanno Steen (6)
(1) Boston Children's Hospital, Boston, Massachusetts, United States
(2) Sustainable Health Foundation (FOSAD), Center of Excellence for Training on Research & Priority Diseases
(CEFORPAG), Conakry, Guinea
(3) Sustainable Health Foundation (FOSAD), Conakry, Guinea
(4) Sustainable Health Foundation (FOSAD), Conakry
(5) Toronto General Hospital Research Institute, University Health Network, Toronto, Canada
(6) Boston Children's Hospital, Boston, Massachusetts
The 2013-2016 Ebola Zaire virus (EBV) outbreak in West Africa resulted in over 28,000 cases and 11,000 deaths. Ebola
virus disease (EVD) is a highly virulent systemic disease with a high case fatality rate of ~50%. EVD results in hemorrhagic
fever marked by an exaggerated systemic inflammatory response, and impaired vascular and coagulation systems. The
immune response of patients who either survived or died is characterized by pronounced differences. Notably, fatalities
showed a diminished capacity to mount an appropriate immune response, resulting in high viremia and increased pro-
inflammatory cytokine production. In this study, we analyzed 38 sequential samples collected from 12 patients: 8 survivors
and 4 fatalities. Our analytical strategy combined three analytical platforms covering three different fractions of the plasma
proteome: the undepleted classical plasma proteome, the depleted plasma proteome, and cytokines/chemokines, using
LC/MS- and antibody-based assays. This analysis resulted in over 1000 quantified host and pathogen proteins. For
depletion of the most abundant plasma proteins, we advanced a perchloric acid-based precipitation method. This method is
low cost, high-throughput and has proven its robustness for thousands of plasma samples in the Immunophenotyping
Assessment in a COVID-19 Cohort study (IMPACC; Sci Immunol. 2021), which bring significant advancement in the field of
high-throughput proteomics. We first characterized the early host response to EBV infection and identified distinct protein
signatures that could differentiate between fatalities and survivors. Next, we focused on proteins whose abundance levels
changed based on viral load and disease course. Our data revealed distinct protein signatures and identification of a
prognostic 4-protein biomarker panel that performs better than the widely used viral load, comprising a histone, a ribosomal
protein, moesin, and kininogen. Given the easy implementation of protein assays, this 4-protein panel may help direct care
in emergency settings in resource poor regions.
Page 45 of 166
P01: Advances in Technology
P01.01 Evaluating Impacts of Nanoflow Column Length and Core-shell Particle Size for Proteomics DDA and
SWATH Workflows
Alexandra Antonoplis (1), Jason Anspach (2), Christie Hunter (1)
(1) SCIEX, Redwood City, California, United States
(2) Phenomenex, Torrance, California, United States
Introduction: Nanoflow chromatography is often used in proteomics DDA and SWATH acquisition workflows in order to
obtain the highest sensitivity on digested samples. Peak shape and resolution are critical to allow the MS system to sample
as many peptides as possible. Here, new long nanoflow columns packed with core-shell chromatographic phase were
evaluated in DDA and SWATH workflows with digested K562 protein extracts. Methods: In the current study, three different
nanoflow column lengths (15, 25, and 50 cm) were packed with core-shell Kinetex C18 phase. For the 50 cm column, both
2.6 µm and 5.0 µm particle sizes were tested. Gradient durations ranging from 60-180 minutes were used across the
column types for DDA workflows. K562 cell digest was evaluated at several loads and a trap-elute workflow was
implemented with an analytical flow rate of 300 nL/min. The TripleTOF 6600+ system and the ZenoTOF 7600 system were
used for all data acquisition, and data was processed using the ProteinPilot App in OneOmics suite. Results: Here, the
impacts of column length, column particle size, and gradient duration on protein and peptide identifications were assessed.
As expected, protein and peptide identifications increased as the gradient length increased and column length increased,
with optimal identifications observed with a 180 min gradient and 50 cm column length. The impacts of higher sensitivity
MS/MS (Zeno MS/MS) on protein ID gains were evaluated using the final nanoflow chromatographic conditions. Optimized
DDA and SWATH acquisition methods were also determined for nanoflow system suitability analysis. Conclusions:
Using 50 cm nanoflow columns packed with core-shell chromatographic phase improved protein and peptide identifications
in DDA and SWATH workflows.
P01.02 Infrared Matrix-Assisted Laser Desorption Electrospray Ionization and Quadrupole Time-of-Flight Mass
Spectrometer: Design, Coupling and Challenges.
Cristina Arciniega (1), Ken Garrard (1), Jacob Guymon (1), Alex Apffel (2), John Fjeldsted (2), David Muddiman (1)
(1) North Carolina State University, Raleigh, North Carolina, United States
(2) Agilent Technologies, Santa Clara, California, United States
Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-MALDESI) is an ambient ionization source that
combines matrix-assisted laser desorption with electrospray ionization. This source has been widely employed for mass
spectrometry imaging (MSI) and high-throughput screening (HTS) applications. It has been previously coupled to FT-ICR
and Orbitrap mass spectrometers due to their high resolving power and mass accuracy. These are pulsed instruments that
require tens to hundreds of milliseconds between scan, consequently limiting throughput. This work focuses on the
integration and characterization of IR-MALDESI source to a commercial Q-TOF mass spectrometer for the analysis of
biological samples. For this purpose, an IR-MALDESI source has been designed as a proof of concept to perform analyses
of samples directly from microliter well plates. Also, an extended metal capillary has been engineered for the front end of the
mass spectrometer. The ionization source is equipped with a 2970nm IR laser which operates at 100Hz using 1 pulse-per-
Page 46 of 166
burst (PPB) producing a quasi-continuous ion generation. Since the Q-TOF is a continuous mass analyzer, IR-MALDESI
was adapted to properly couple the source with the instrument. In this study, various samples were analyzed to prove the
functionality of the IR-MALDESI source paired with a Q-TOF mass spectrometer.
P01.03 Highly multiplexed, spatially resolved protein and gene expression profiling in FFPE tumor tissues
Jennifer Chew (1), Cedric Uytingco (1), Naishitha Anaparthy (1), Jun Chiang (1), Christina Galonska (2), Karthik Ganapathy
(1), Ryo Hatori (1), Alexander Hermes (3), Layla Katiraee (1), Anna-Maria Katsori (4), William Nitsch (1), Patrick Roelli (2),
Joe Shuga (1), Rapolas Spalinkas (2), Benton Veire (1), Dan Walker (1), Neil Weisenfeld (1), Stephen Williams (1), Zachary
Bent (1), Marlon Stoeckius (2)
(1) 10X Genomics, Pleasanton, California, United States
(2) 10X Genomics, Stockholm, Sweden
(3) 10X Genomics, Pleasanton, California
(4) 10X Genomics, Stockholm
The tumor microenvironment (TME) is composed of highly heterogeneous structures and cell types that dynamically
influence and communicate with each other. Protein interactions within these cell types often play a critical role in how the
cancer develops, progresses, and responds to therapies. Traditionally, immunofluorescence or immunohistochemistry has
been used to annotate and detect proteins and other antigens in tissue sections. More recent higher throughput approaches
capture phenotypes using a limited number of protein markers while spatially interrogating targeted or transcriptome-wide
expression of RNA in tissue sections. However, for a more comprehensive understanding of the unique characteristics of
cells within the TME, multiple layers of information must be studied together. Here we demonstrate a novel, streamlined
multiomic spatial assay that integrates histological staining and imaging with simultaneous transcriptome-wide gene
expression and highly multiplexed protein expression profiling from the same formalin-fixed paraffin embedded (FFPE)
tissue section. In short, tissue sections from archived FFPE samples were placed on slides containing arrayed capture
oligos with unique positional barcodes. The tissues were stained with hematoxylin and eosin, imaged and then incubated
with transcriptome-wide probes and a high-plex DNA-barcoded antibody panel containing intra- and extracellular markers.
Transcriptome probes and antibody-barcodes were then spatially captured on the slide and converted into sequencing-
ready libraries. We apply this method to simultaneously measure gene and protein expression within the TME of human
breast cancer FFPE samples using whole transcriptome probes and an immune-oncology antibody panel. This data
enables comparison and correlation of multiple analytes and their patterns within the same sample section and
demonstrates that a spatially resolved, multiomic approach provides a more comprehensive understanding of cellular
behavior in and around tumors, yielding new insights into disease progression, predictive biomarkers, drug response and
resistance, and therapeutic development.
P01.04 Deep Plasma Proteomics at Scale with Proteograph Product Suite: A Performance Evaluation with Label-
free and TMT Multiplexing Methods
Ryan Benz (1), Xiaoju Zhang (1), Tianyu Wang (1), Mathew Ellenberger (1), Gabriel Castro (1), Ray Schmidt (1), Veder
Garcia (1), Qiu Yang (2), Chenyi Zhao (1), Purvi Tandel (2), Michelle Dubuke (1), Ramón Díaz Peña (3), Alex Campos (3),
Page 47 of 166
Khatereh Motamedchaboki (1)
(1) Seer, Redwood City, California, United States
(2) Seer, Redwood City, California
(3) Sanford Burnham Prebys Medical Discovery Institute, La Jolla, California, United States
Human blood plasma is a widely accessible sample for assessing individual health status. However, the large dynamic
range of circulating proteins combined with the diversities of proteoforms present in plasma have limited the comprehensive
characterization of the plasma proteome in a high throughput manner. To address such challenges, current plasma
proteomics workflows combine immunodepletion of high abundance proteins, peptide fractionation and sample multiplexing
approaches such as tandem mass tags (TMT). Recent advancement in sample preparation (Seer’s Proteograph Product
Suite), coupled with improved mass spectrometry instrument sensitivity and speed, enable the quantification of thousands of
proteins from plasma without compromising throughput or reproducibility, creating a unique opportunity to detect robust
protein biomarkers for complex diseases. Here we evaluate the performance of label-free and TMT multiplexing methods
with a set of control plasma samples processed with Proteograph Product Suite for deep plasma proteomic analysis. Pooled
control human plasma samples were processed with Proteograph in 4 batches prepared on 4 different days. Tryptic
peptides were either directly analyzed by LC-MS/MS or labeled with one of the TMTpro reagents followed by peptide
fractionation (high pH RP) and LC-MS/MS analysis. LC-MS analysis were performed with 40-48-hours workflow (LFQ and
TMTpro 16plex). TMT with peptide fractionation, resulted in ~3,000 protein groups. Approximately 86% of the peptide
features are detected across all 4 batches, with a median CV < 20%, (PSM level). Label-free performance across four
plates run on two different instrument and two different days were evaluated with protein group intensity for most NPs with
CVs < 20%, enabling large scale plasma proteomics without compromising depth or precision. We detected plasma proteins
spanning 9 orders of magnitude including 40 cytokine activity proteins and several members of TNF superfamily. This study
evaluates performance of the Proteograph platform combined with label-free or TMT technology mass spectrometry
analysis.
P01.05 Optimization of an Automated 96-well Plate Protein Depletion Protocol for Plasma
Min Choi (1), Albert Arul (1), Rena Robinson (1)
(1) Vanderbilt University, Nashville, Tennessee, United States
Depletion of high-abundant proteins in plasma increases the coverage of the plasma proteome detectable by mass
spectrometry (MS). Traditional methods of protein depletion can require processing one sample at a time using an
immunoaffinity-based depletion column, which presents a bottleneck in a high-throughput proteomics workflow. Single-use
spin columns can shorten the processing time through sample parallelization but then becomes limited to centrifugal rotor
capacities. The goal of this study was to optimize an automated platform for depleting multiple plasma samples
simultaneously on a 96-well plate packed with High Select Top 14 resin (Thermo Fisher). In order to benchmark our results,
we compared depletions of human plasma standard on the 96-well plate method with multiple injections on an Agilent
Human 14 Multiple Affinity Removal column (MARS14). Plasma standards were depleted by the 96-well plate method using
an automated liquid handling device. Different plasma-to-resin ratios and incubation methods were tested to optimize
depletion efficiency calculated using bicinchoninic acid (BCA) protein assays of the automated and column methods. The
binding efficiencies of the top 14 abundant proteins captured by the High Select Top 14 resin and MARS14 column were
evaluated by bottom-up proteomics. Compared to the MARS14 column, the automated 96-well method using the High
Page 48 of 166
Select Top14 resin resulted in decreased depletion and binding efficiencies. However, the depletion and binding efficiencies
of the automated platform were reproducible across the technical replicates. Overall, the automated platform can reduce
processing time by depleting up to 96 samples simultaneously in less than an hour.
Purpose Urine has tremendous value for non-invasive biomarker identification and routine detection, but intra- and inter-
individual variability present a challenge with interpretation. Here, we aimed to standardize urine collection and disease-
specific processing and analysis. Our end-to-end workflow enables rapid detection of proteins with minimum sample
requirements to support large cohort studies. Methods Eight pooled urine samples, 10 donors per pool, were digested and
fractionated by HPLC into 12 fractions. Each fraction was analyzed with an optimized data-dependent acquisition (DDA)
method and data-independent acquisition (DIA) method on an Orbitrap Exploris 480. DDA and DIA data were used to
generate a spectral library. Subsequently, individual urine specimens were analyzed without fractionation and searched
against a library or analyzed using directDIA for large cohorts. High abundance proteins in chronic kidney disease (CKD)
urine, such as albumin, presented a challenge and thus, strategies to reduce albumin interference were introduced for
disease-specific sample processing to improve proteome coverage. Results The impact of processing variables including
centrifugation and storage temperature on the true urine proteome were evaluated. We assessed a range of LC gradient
lengths, chromatography columns and ionization sources and determined that a 30 minute gradient with an ES906 column
and Easy-Spraysource on the Exploris 480 was optimal for rapid and comprehensive analysis. For DIA analysis, multiple
spectral libraries were generated and evaluated based on size and proteome coverage, with current libraries containing
>3400 protein groups. Using this rapid DIA workflow, single shot analysis of individual healthy urine samples detected
>2000 protein groups without fractionation. Currently, the workflow is further being optimized and applied to the analysis of
CKD urine, to address numerous analytic challenges due to extensive proteinuria and high levels of albumin. Conclusion
We have optimized and standardized an end-to-end high-throughput workflow that enables rapid urine analysis.
P01.07 Assessing the Precision of Detergent-based Workflows: Cartridge-Based Proteome Precipitation for Robust
Sample Preparation
Alan Doucette (1), Jessica Nickerson (1), Victoria Miller (2), Sara Little (2), Ziheng Dang (1)
(1) Dalhousie University, Halifax, Nova Scotia, Canada
(2) Proteoform Scientific, Halifax, Nova Scotia, Canada
Introduction: Multiple advances in MS instrumentation have improved the rate, depth and accuracy of proteome
profiling. Particularly for comparative and quantitative proteome characterization, a reliable analytical platform is essential
for accurate and precise measurements. Inconsistencies during proteome sample preparation can contribute even greater
inconsistencies than the MS platform, and thereby presents a bottleneck to delivering repeatable results. Here we evaluate
Page 49 of 166
the analytical precision and robustness of an SDS-based workflow for bottom-up proteome analysis. The ProTrap XG is a
disposable cartridge designed to automate protein precipitation for SDS depletion, together with in-cartridge digestion and
subsequent peptide enrichment. Each component of the workflow (proteome extraction, purification & recovery through
precipitation, digestion efficiency and peptide isolation) will be evaluated relative to the resulting MS precision.
Methods: Samples of bovine liver were extracted with varying levels of SDS (0, 1, 5%), and independently processed in
triplicate using the ProTrap XG cartridge. Protein recovery, residual SDS, and peptide yield following digestion are
reported. Reproducibility was evaluated following a bottom-up workflow across technical replicates vs. replicate MS
injections, comparing ID counts and intensity. Preliminary Results: Inclusion of SDS is favorable for complete proteome
extraction of bovine liver. However, acetone precipitation of samples with 1% SDS increases recovery, with 45% more
proteins (avg 8103) and 23% more peptides (29478) vs extraction at 5%. More importantly in 1%, RSD drops from >50% to
9% at the peptide level, and from 28 to 4% at the protein level. MS replicates of an identical sample provided 7% absolute
difference in PSMs between runs, with 78 ± 0.5% overlap in identified proteins. Adding the complete ProTrap-based
workflow, the agreement in individual peptide PSMs across runs remained within 9%, with 76 ± 0.5% overlapping
identification. Conclusions: The ProTrap XG provides a highly reproducible workflow for automated protein sample
preparation for bottom-up analysis.
P01.08 Automated S-Trap Sample Preparation on an Integrated Liquid Handling / Pneumatic Sample Processing
Workstation for MAM Applications
Heather Eastwood (1), John Laycock (1), Claudia Martelli (2), Manuel Bauer (2), John Wilson (3), Xin Zhang (4), James
Teuscher (1), Shang Tsai (5)
(1) Tecan, Baldwin Park, California, United States
(2) Tecan, Männedorf, Switzerland
(3) ProtiFi, LLC, Farmingdale, New York, United States
(4) Sciex, Redwood City, California, United States
(5) Tecan, Baldwin Park, California
Purpose Multiple Attribute Monitoring (MAM) is an LC-MS technique designed to simultaneously and directly monitor critical
quality attributes (CQAs) of biologics including impurities, post-translational modifications (PTMs) and primary sequence.
The highly varied nature of biotherapeutic manufacturing from bioreactors containing surfactants to final formulations
presents analytical challenges. Here we present a universal, automatable MAM kit capable of removing contaminants from
salts to surfactants to excipients to dyes in a streamlined workflow usable without change at all stages from bioreactor to
final product. Methods A MAM-specific kit and automated workflow based on the S-Trap 96-well plate was optimized to the
specific needs and requirements of MAM at all stages of biotherapeutic manufacturing. The kit was stress tested using the
NISTmAb RM 8671 monoclonal antibody spiked into various PEG or surfactant containing solutions. MAM sample
preparation was performed with full automation on the Tecan Freedom EVO® liquid handler with an integrated A200.
Methods and scripts for (semi-)automated analyses are included in the ProtiFi/Tecan MAM workstation configuration.
Samples were analyzed on a Sciex ZenoTOF 7600 in CID and EAD modes for comparative analysis. Results The MAM-
sample processing kit was effective without protocol modification at removing all contaminant challenges including salts,
surfactants, emulsifiers, dyes and other small molecules; this performance matches previous studies in the analysis of
surfactant-containing bioreactor supernatants. Excellent reproducibility was observed for both unmodified and PTM-modified
peptides with detection of the expected glycosylations of NISTmAb RM 8671. Residual host cell proteins (HCPs) were also
Page 50 of 166
detected. Conclusion Quantification variability from sample preparation and especially automated sample preparation was
similar to levels of variability observed in technical replicate injections. The MAM automation workflow and the
corresponding kit has utility at all levels of therapeutics manufacturing, from initial R&D and characterization through
process control, formulations, and QC and release testing.
P01.09 Multi-nanoparticle Workflow Enables Deep Plasma Proteomics at Scale, with Enhanced Precision, and
Depths of Coverage.
Shadi Ferdosi (1), Tianyu Wang (1), Renata Blatnik (2), Moaraj Hasan (3), Jessica Chu (3), Eltaher M. Elgierari (3), Xiaoyan
Zhao (3), Martin Goldberg (4), Asim Siddiqui (3), Serafim Batzoglou (4), Michael Krawitzky (2), Nagarjuna Nagaraj (2), Omid
C. Farokhzad (3), Daniel Hornburg (3)
(1) Seer, Redwood City, California, United States
(2) Bruker Daltonics, Bremen, Germany
(3) Seer, Redwood City, United States
(4) Seer, Redwood City
To overcome limitation of deep plasma proteomics in large cohorts, we have developed a fast and scalable technology that
employs intricate protein-nano interactions. Introducing a nanoparticle (NP) into a biofluid such as blood plasma leads to the
formation of a selective, specific, and reproducible protein corona at the nano-bio interface driven by the relationship
between protein-NP affinity, protein abundance and protein-protein interactions. We previously demonstrated that this
process, incorporated within the Seer Proteograph™ Product Suite, offers superior performance in terms of depth, breadth,
precision, and throughput compared to conventional deep workflows. The ratio of plasma-to-nanoparticles determines the
competition between proteins for binding surface, which plays an important role in protein corona composition and can
be optimized to enhance and differentiate protein selectivity. Here we investigate effects of different conditions on
protein corona composition enabling enhanced performance of Proteograph. We have investigated compositional
changes of protein coronas from 5 NPs with blood plasma at different ratios. Samples were analyzed with timsTOF Pro
mass spectrometry and UltiMate3000 Dionex LC system using 30min DIA runs. We evaluated depth, dynamic range,
coverage, and precision of quantification at a wide range of concentrations for each NP. By limiting the available binding
surface of NPs and increasing the binding competition, we are able to identify 20 – 60% more proteins on the surface of
each NP. Moreover, by increasing the competition the proteins are more reproducibly identified and quantified across the
replicates of the same NP. In addition, protein selectivity was enhanced, leading to improved coverage of plasma proteome
when using multiple physicochemically distinct NPs. In summary, NP panels with optimized workflow, capture a
large and diverse set of proteins and biological pathways based on their specific physicochemical makeup.
P01.10 The Next Generation Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (NextGen IR-
MALDESI) Source
Kevan Knizner (1), Jacob Guymon (2), Kenneth Garrard (3), Guy Bouvree (4), Jan-Peter Hauschild (5), Kerstin Strupat (5),
Kyle Fort (5), Lee Earley (6), Eloy Wouters (7), Fan Pu (8), Nathaniel Elsen (9), Jon Williams (9), Mark Pankow (2), David
Muddiman (10)
Page 51 of 166
(1) FTMS Laboratory for Human Health Research, Department of Chemistry, North Carolina State University, Raleigh, North
Carolina, United States
(2) Precision Engineering Consortium, Department of Mechanical and Aerospace Engineering, North Carolina State
University, Raleigh, North Carolina, United States
(3) FTMS Laboratory for Human Health Research, Department of Chemistry, North Carolina State University; Precision
Engineering Consortium, Department of Mechanical and Aerospace Engineering, North Carolina State University, Raleigh,
North Carolina, United States
(4) GB Conseil & Services, Brie-Comte-Robert, France
(5) Thermo Fisher Scientific, Bremen, Germany
(6) Thermo Fisher Scientific, San Jose, California
(7) Thermo Fisher Scientific, San Jose, California, United States
(8) Drug Discovery Science and Technology, Abbvie Inc., North Chicago, Illinois
(9) Drug Discovery Science and Technology, AbbVie Inc., North Chicago, Illinois, United States
(10) FTMS Laboratory for Human Health Research, Department of Chemistry, North Carolina State University; Molecular
Education, Technology, and Research Innovation Center (METRIC), North Carolina State University, Raleigh, North
Carolina, United States
Mass spectrometry imaging (MSI) is an effective analysis method for determining the spatial distribution of molecules within
biological tissues. Infrared matrix-assisted laser desorption ionization mass spectrometry (IR-MALDESI-MS) is a versatile
MSI platform that has been utilized to image diverse biological samples without tedious sample preparation. To improve IR-
MALDESI-MSI analyses, user-friendliness of the source, and overall versatility of the platform, the key components of the
source have been upgraded. Here, we present the next generation (NextGen) IR-MALDESI source which features a
vertically mounted IR laser, planar translation stage with computer-controlled z-axis sample stage, external source
component adjusters, redesigned source-to-mass spectrometer interface plate, and reduced plastic use. Also, the NextGen
IR-MALDESI source can be easily interfaced with numerous Orbitrap mass spectrometers to accommodate a greater
number of research labs. In this work, we present the characterization of the NextGen IR-MALDESI source to establish it as
an improved platform for MSI and direct analysis applications.
Proteomics has become an essential tool to address critical biological questions through the identification and
characterization of total protein content within a biological system. The discovery of distinct protein biomarkers for specific
disease targets can aid in the development of therapeutic treatments. Major advances have been made to improve the
sensitivity of mass analyzers, spectral quality, and the speed of data processing enabling more comprehensive proteome
discovery and quantitation. While focus has recently begun shifting toward robust proteomic sample preparation efforts,
high-throughput proteomics sample preparation capabilities has been lacking. This led us to develop a highly automated
Page 52 of 166
universal 384-well plate sample preparation platform with high reproducibility and adaptability for the extraction of proteins
from cells within a culture plate. Digestion efficiency was excellent in comparison to a commercial digest peptide standard
with minimal sample loss while sample preparation throughput was improved by 20- to 40-fold (the entire process from
plated cells to clean peptides is complete in ~300 min). Analysis of six human cell types including two primary cell samples,
identified and quantified ~4,000 proteins for each sample in a single HPLC-MS/MS injection with only 100 – 10,000 cells,
thus demonstrating universality of the platform. A selected protein was further quantified with a developed HPLC-MRM
method for HeLa digests utilizing a spiked in heavy-labelled internal standard peptide. Excellent linearity was achieved
across different cell numbers indicating a potential for target protein quantitation in clinical research. These reported
advancements in high-throughput proteomics sample preparation brings us a step closer to realizing a fully automated
proteomics screening platform thus enabling an efficient large-scale characterization of diverse biological signatures across
a wide array of conditions and parameters.
P01.12 Real-Time Database Search Multiplex Quantitative Proteomics: Instrument Method and Data Processing
Yang Liu (1), Frank Berg (2), William D. Barshop (1), Jesse D. Canterbury (1), David M. Horn (1), David Bergen (1), Romain
Huguet (1), Rosa Viner (3)
(1) Thermo Fisher Scientific, San Jose, California, United States
(2) Thermo Fisher Scientific, Bremen, Germany
(3) Thermo Fisher Scientific, San Jose, California
Introduction Real-time search (RTS) on the Thermo Scientific™ Orbitrap Eclipse™ Tribrid™ mass spectrometer has
enabled the capability to triggered SPS-MS3 scans only upon confident peptide spectrum matches (PSMs) identified from
MS2 spectra via online database search using Comet. Here we introduce RTS TMT quantitation workflow using the latest
version of instrument control software (ICSW) and the implementation of Comet in Thermo ScientificTM Proteome
DiscovererTM Software (PD) 3.0 to provide the best alignment between online and post-acquisition data analysis. Method For
RTS experiments and its extended features, 500ng Thermo ScientificTM PierceTM TMT11plex Yeast Digest Standard or TMT
pro 18 plex Yeast Digest Standard (prototype) was analyzed by an Orbitrap Eclipse Tribrid mass spectrometer (ICSW 3.5).
MS2 spectra were searched against a yeast proteome database during acquisition using the Comet search algorithm
(2019.01 rev.1). The data were analyzed with PD 3.0, using both Comet and SequestHT search engine. Result RTS TMT
SPS MS3 strategy was able to quantify 40-50% more protein groups and 30-40% more peptide groups, compared the SPS-
MS3 method. The close-out feature can further boost the number of quantified protein groups without sacrificing
quantification accuracy. The online FDR estimation can improve the percentage of quantified protein groups. In both
SequestHT and Comet search in PD, parameters were matched to the RTS Comet settings. 98% of PSMs that were
confirmed by RTS Comet search were confidently identified in the post-acquisition analysis using PD Comet algorithm. By
combining both SequestHT and Comet in PD processing workflow, the number of identification and quantified IDs can be
improved by 10-15%. Conclusion The RTS-MS3 workflow using the latest ICSW, together with post-acquisition data
analysis optimization using Comet node in PD 3.0, largely improved the performance of the multiplex quantification
proteomics.
Page 53 of 166
P01.13 Benchmark of Micro-flow Chromatograph for Robust Proteomics Analysis
Yang Liu (1), Amirmansoor Hakimi (1), David M Horn (1), Runsheng Zheng (2), Oleksand Boychenko (2), Daniel Lopez-
Ferrer (1)
(1) Thermo Fisher Scientific, San Jose, California, United States
(2) Thermo Fisher Scientific, Germering, Germany
Introduction Liquid chromatography-mass spectrometry (LC-MS) has been a powerful analytical tool in protein identification
and quantification. In the past few decades, nano-flow LC-MS has been the primary approach due to its high sensitivity.
However, challenges always come from the needs of high throughput, reproducibility and robustness. Here we present a
micro-flow LC-MS workflow using a robust setup with Thermo ScientificTM NG micro-flow UHPLC System coupled to Thermo
ScientificTM Orbitrap ExplorisTM platforms. Methods Thermo ScientificTM PierceTM HeLa protein digest standard and Thermo
ScientificTM PierceTM TMT-11plex Yeast digest standard were analyzed on the Orbitrap Exploris 240 MS for LFQ analysis.
Digested peripheral blood mononuclear cells (PBMCs) from a variety of animal species (human, mouse, etc) were analyzed
to demonstrate the robustness over 100 injections. Thermo ScientificTM High-Field Asymmetric Waveform Ion Mobility
Spectrometry (FAIMS) Pro Duo interface was installed to provide gas phase fractionation. Data was analyzed on Thermo
ScientificTM Proteome DiscovererTM 3.0 software using MSPepSearch and CHIMERYS in parallel. Result We were able to
identify ~3400 protein groups and ~27500 peptide groups from 1ug of HeLa digest, ~3800 protein groups and ~34000
peptide groups from 5ug of HeLa digest within 30min gradient. The micro-flow LC-MS system showed excellent
reproducibility of protein group IDs (<3% coefficient of variation, CV) and protein group abundance (median CV ><11%). In
multiplex quantitation 90% of identified proteins and peptides were successfully quantified. Digested PBMC peptides were
fractioned by 6 compensation voltages using FAIMS Pro Duo interface and were acquired in triplicate on the Orbitrap
Exploris 480MS. GPF provides deep proteome profiling without the need for off-line RPLC fractionation, which reduced the
overall experiment time. Conclusion This micro-flow LC-MS setup has been demonstrated to be highly reproducible and
robust without sacrificing performance for both discovery and quantitation.>
P01.14 DIA Analysis in Discovery Identifications and Label-free Quantification on Orbitrap Exploris 240 Mass
Spectrometer
Yang Liu (1), Khatereh Motamedchaboki (1), Sega Ndiaye (2), Aman Makaju (1), Amarjeet K. Flora (3), Ryan Bomgarden
(3), Daniel Lopez-Ferrer (1)
(1) Thermo Fisher Scientific, San Jose, California, United States
(2) Thermo Fisher Scientific, Courtaboeuf, France
(3) Thermo Fisher Scientific, Rockford, Illinois, United States
Introduction Label-free quantitation (LFQ) is the simplest way to explore the proteome. Data-independent acquisition (DIA)
mass spectrometry (MS) can rapidly identify and reproducibly quantify all ions without the bias from peak intensity, providing
deep proteome analysis for large-scale profiling and quantitative studies. In this work, we present a MS-only workflow that
combines gas-phase fractionation (GPF) and DIA acquisition, saving significant experiment time while maintaining high
reproducibility Methods Different amount of digested E.coli peptides (in ratio 0:1:2:4:8) were added to a fixed amount of
HeLa digest peptides. Mixed peptides were loaded on 25cm Aurora Series emitter column (25 cm*75 μm ID, 1.6 μm, C18),
separated by a 90 min LC gradientat in direct injection mode on an EASY-nLC 1200 system before being injected onto the
Thermo ScientificTM Orbitrap ExplorisTM 240 mass spectrometer. Empirically corrected spectral library was built by matching
Page 54 of 166
the gas-phase fractionations (GPF) DIA chromatogram spectral library to the predicted spectral library of HeLa and E. coli
tryptic peptides1. Acquired DIA data were analyzed by SpectronautTM 14.0. Results Over 9,000 proteins were identified from
~1 μg of the mixed proteome sample when matching to the empirical spectral library. Over 75% of the proteins were
quantified in each of 15 LC-MS runs. In addition, the Orbitrap Exploris 240 mass spectrometer achieved over 6 orders of
magnitude in protein dynamic range. Furthermore, we also proved that matching with the library built at a higher sample
load would improve the identification coverage and confidence of identifications. Therefore, once built at ultimate setup, the
spectral library can be used for universal DIA discovery of the same sample. Conclusions Orbitrap Exploris 240 MS had
excellent performance in DIA analysis in discovery and quantitation. References 1. Searle, B.C., Swearingen, K.E.,
Barnes, C.A. et al. Nat Commun 11, 1548 (2020).
P01.15 Semi-automated and high-throughput homogenization technique for in-depth analysis of various tissue
proteomes
Kristine Bissell (1), Berit Mang (1), Katrin Hartinger (1), Sebastian H. Johansson (1), Jasmin Johansson (1), Nils A. Kulak
(1)
(1) PreOmics GmbH, Planegg, Germany
Purpose: Efficient protein extraction is a crucial and challenging step in tissue sample preparation for LC-MS analysis. While
conventional methods are faced with various challenges - such as low-throughput and time-consuming workflows - our
novel tissue lysis workflow on the semi-automated BeatBox platform enables efficient protein extraction for 96 samples in
parallel in as little as 10 minutes. Coupled to the robust iST technology that is fully automatable on common liquid handling
platforms, we combined in-depth sample preparation for large-scale applications with minimal hands-on time. Methods The
described workflow is based on 96 well sample processing on the BeatBox platform and applicable for a plethora of tissue
types ranging from soft brain to rigid heart muscle samples. Utilizing innovative bead-based technologies, a defined energy
input is applied to each sample facilitating highly efficient protein extraction. For complete nanoLC-MS sample preparation,
we combined the BeatBox-based tissue lysis with the iST workflow. Improved proteomic depth could be achieved by a 3-
step peptide fractionation. Results The BeatBox platform speeds up and simplifies the tissue preparation for proteomic
experiments. From intact tissue sample to finished data acquisition (DDA on timsTOF Pro, Bruker) in less than 4 hours, we
identified ~2300 proteins from mouse heart muscle, ~3000 proteins from mouse liver tissue and >4000 proteins from mouse
lung tissue. In comparison to the traditional iST workflow, we gained up to 13 % more protein identifications. At the same
time, excellent technical variability (e.g. for mouse liver tissue: median CV of 9.2 %) was obtained. Furthermore, a
combination of the BeatBox-iST workflow with tip-based peptide fractionation increased the protein identifications by over 40
% compared to unfractionated samples. Conclusion: The innovative BeatBox-based workflow will set a new standard in
large-scale tissue sample preparation by enabling ultra-fast and highly efficient protein extraction in a high-throughput
manner.
P01.16 Multi-omics sample preparation workflow for DNA and proteins from a single starting sample using
ProMTag
Page 55 of 166
Amber Lucas (1), Stephanie Biedka (1), Jonathan Minden (1)
(1) Impact Proteomics, Pittsburgh, Pennsylvania, United States
Extracellular vesicles (EVs) are complex, cell-derived nanoparticles generated by all cell types. EVs are composed of lipid
bilayer membranes and their associated membrane proteins, nucleic acids, and luminal proteins. The mechanism by which
Gram-positive bacteria shed EVs is still unknown. EVs from the Gram-positive human pathogen S. pneumoniae, which is a
major cause of otitis medi and pneumonia, are of particular interest because of how they EVs modulate the host immune
response. To uncover possible mechanisms for EV production and shedding in S. pneumoniae, we have performed a
comparative proteomics analysis of EV membrane proteins versus whole-cell membrane proteins.
Membrane proteins were enriched from intact S. pneumoniae cells or their EVs using a ProMTag labeling and capture
workflow. ProMTag is a bifunctional protein tag where one moiety of the tag is able to form a reversible, covalent link to
primary amines on proteins. The other moiety is methyltetrazine, which can form an irreversible, covalent bond with trans-
Cyclooctene (TCO) on the surface of beads to capture ProMTagged proteins for cleanup and elution. Using this workflow
plasma membrane proteins can be tagged, captured, washed to remove non-plasma membrane proteins, and then eluted in
their original, unmodified state.
In this study, intact cells and EVs from S. pneumoniae cultures were separated and the extracellular domains of membrane
proteins in these two fractions were labeled with ProMTag. The membrane proteins were then enriched, washed, and eluted
using the ProMTag workflow. These membrane protein populations were then analyzed using mass spectrometry to identify
what proteins were present in EV membranes and how that compared to the composition of whole cell membranes and
whole EVs. Analysis revealed membrane proteins that are concentrated or absent in EV membranes relative to bulk plasma
membrane from whole cells, indicating a selective process for EV formation in S. pneumoniae. With this information, we
present a new model for EV formation and shedding in S. pneumoniae.
P01.17 Robust Identification of Low Abundant Proteins Utilizing a Precipitation Workflow on the ProTrap XG
Victoria A Miller (1), Sara Lahsaee Little (1), Jessica L Nickerson (2), Alan A Doucette (2)
(1) Proteoform Scientific Inc, Halifax, Nova Scotia, Canada
(2) Dalhousie University, Halifax, Nova Scotia, Canada
In biomarker discovery, conclusions are dependent on the quality of the data, which is contingent on the integrity of sample
preparation. Efficient recovery of all proteins, regardless of abundance, is necessary to accurately represent the proteome.
The ProTrap XG is a filter-based sample preparation cartridge that simplifies the traditional precipitation workflow. Sample
preparation conditions for the ProTrap XG were optimized to deliver highly efficient and robust identification of low
abundance proteins from a complex system. The effect of 2% SDS on the precipitation and recovery of 3 standard proteins,
spanning 4 orders of magnitude: Beta-galactosidase (50 µg), cytochrome c (50 ng) and enolase-1 (5ng) were combined in
the presence of 2% SDS. 50 ug of the mixture was subjected to acetone precipitation followed by resolubilization and
trypsin digestion in the ProTrap XG. A control (no SDS) was also digested in solution. Bottom-up analysis on a Thermo
Orbitrap Fusion Lumos with easy nLC identified all 3 proteins with equivalent sequence coverage in the SDS-based
workflow relative to the control. Moreover, the SDS workflow yielded higher peptide abundance for Beta-galactosidase (4x)
and cytochrome c (2x) vs. the control. Bovine liver was extracted in 1% SDS and 50 µg protein was processed in the
ProTrap XG. Bottom-up MS yielded on average ~8000 proteins and ~30,000 peptides. Several low abundance proteins,
Page 56 of 166
such as transcription factors (e.g. NFκB) were identified. Further data analysis is ongoing. A detergent-based workflow for
proteome processing in the ProTrap XG is therefore shown to improve the MS identification of low abundance proteins in
biological samples.
P01.18 Establishing Quality Control Procedures for Sample Preparation of High-Throughput Plasma Proteomics
Nekesa Oliver (1), Min Ji Choi (1), Albert Arul (1), Renã Robinson (1)
(1) Vanderbilt University, Nashville, Tennessee, United States
Clinical biomarker discovery and verification that uses mass spectrometry (MS) –based proteomics must have workflows
that are dependable and reliable. However, bottom-up plasma proteomic workflows can suffer from reproducibility issues in
large cohorts (N>100) due to losses that occur with numerous sample preparation steps such as protein depletion, tryptic
digestion, chemical labeling, sample cleanup and peptide separation. Automation of sample preparation workflows with a
robotic liquid handler can alleviate these losses, manage the workflow, and increase reproducibility. Designing quality
control (QC) measures is key to ensuring reliable MS data is obtained. This study of 800 plasma samples assesses robust
quality control decision checkpoints for an automated proteomic preparation workflow using a robotic liquid handler. TMTpro
isotopic labeling is used to increase multiplexing capabilities while an equimolar sample pool labeled with TMTzero is
prepared as an instrument QC for liquid chromatography (LC)-MS/MS analysis. Across 1600 sample MARS-14 depletions
(800 in duplicate), QC samples gave 2.91 and 0.68 CV% for unbound and bound peak retention times, respectively. Plasma
sample concentrations, determined from bicinchoninic acid assay (BCA), varied less than 10% across this set of samples
while albumin standards had less than 5% variation. Plasma digests randomly selected along with QCs for shotgun
proteomic analysis resulted in protein identifications with less than 5% CV. Here, we will present preliminary data of QC
performance checks of sample preparation aspects of the automated workflow and provide recommendations for large-
scale studies. Overall, defining quality control decision points for a high-throughput and automated workflow is highly
valuable for robust plasma proteomics in large-scale proteomics.
P01.19 Establishing Quality Control Procedures for Large-Scale Discovery Based Proteomics Analysis of Human
Plasma
Khiry L. Patterson (1), Albert B. Arul (1), Min Choi (1), Nekesa C. Oliver (1), Shania Hansen (2), Angela L. Jefferson (2),
Timothy J. Hohman (2), Logan C. Dumitrescu (3), Renã A. S. Robinson (1)
(1) Vanderbilt University, Nashville, Tennessee, United States
(2) Vanderbilt University Medical Center, Nashville, Tennessee, United States
(3) Vanderbilt University Medical Center, Nashville, Tennessee
Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS) platforms for large-scale clinical proteomics enable
analysis of complex proteomes, where thousands of proteins can be quantified in biological samples. Datasets generated
from such analyses are then used to evaluate significant biological findings relevant for disease, assuming that changes
associated with disease are real and not a reflection of experimental variability. Experimental variability can occur at any
step in the proteomics workflow (e.g., sample preparation, fractionation, and LC-MS/MS analysis); therefore, stringent
Page 57 of 166
quality control (QC) protocols are needed to ensure optimal instrument performance, especially when the number of
samples exceeds 100. While use of robotic platforms and sophisticated instruments can reduce variability during sample
preparation and fractionation, daily consideration of LC-MS/MS metrics can inform optimal, “real-time” chromatographic and
instrument performance needs. The goal of this study is to evaluate and determine what appropriate LC-MS/MS metrics
should be considered on a daily basis when performing large-scale plasma proteomics analyses. Using a proteomic dataset
of ~330 human plasma samples, we performed two independent QC protocols: 1) internal QC checks within tandem mass
tag batches and 2) a daily QC check before sample injections. The observed coefficient of variation (CV %) for protein and
peptide identifications was <5 and ~8 CV %, respectively, and similar for peptide spectral matches. Other metrics such as
peptide retention times, signal intensities of precursors, and intensities of TMT reporter ions were also used to evaluate LC
and instrument daily performance. This poster will present performance analysis of QCs in this plasma proteomics dataset,
considerations for improving LC-MS/MS instrument performance, and recommendations for daily QC monitoring in large-
scale plasma proteomics experiments.>
P01.20 High throughput single-shot proteomics on the timsTOF Pro 2 with DIA-PASEF and DDA-PASEF acquisition
mode
Francesco Pingitore (1), Verena Tellstroem (2), Kristina Marx (2), Dirk Wunderlich (2), Nagarjuna Nagaraj (2), Gary Kruppa
(3)
(1) Bruker, San Jose, California, United States
(2) Bruker Daltonics GmbH & Co KG, Bremen, Germany
(3) Bruker s.r.o., Brno, Czech Republic
Introduction Successful proteomics research requires the combination of rapid and in depth coverage along with quantitative
accuracy with high sample throughput in applications like clinical and personalized medicine proteomics. TimsTOF Pro 2
enables sensitive and accurate proteomic analysis from a variety of samples including cell lysates, tissues, pulldowns and
body fluids in a robust fashion. Here we show the performance of the instrument to quantify cell line proteomes in different
gradient lengths and discuss the best usage of the instrument for a versatile use in biological experiments. Methods
Digested HeLa peptides (Pierce), digested K562 peptides (Promega) and inhouse prepared digests from HEK cell lines
were used for benchmark experiments. Peptides were directly loaded on a Aurora-25 cm column using nanoElute coupled
to a timsTOF Pro 2 instrument via a Captive Spray ionization source. Data were acquired using DDA- and DIA-PASEF
methods with different parameters tested for best performance. Data were directly streamed to PaSER box for all DDA data
unless otherwise specified and processed offline in MaxQuant. For DIA, data were processed in Spectronaut (v15). Data
were filtered at a FDR of 1% for peptide and protein groups. Preliminary Data Preliminary analyses were performed with
200 ng of HeLa or K562 peptides and measured on 60-minute gradient. These runs typically resulted in identifications in the
range of 6000 protein groups (PG). Using inhouse digested HEK peptides with an improved sensitive sample preparation
protocol, we could quantify in the range of 7000 PG using DDA methods. With DIA analysis, HEK peptides resulted in about
8000 protein groups and more than 70000 unique peptide sequences. Measuring 20 ng of peptides in relatively shorter
gradients resulted in more than 3500 PG. Novel Aspect Rapid and sensitive quantification of about 7000 protein groups in
single-shot injections.
Page 58 of 166
P01.21 Analyzing protein fluorosequencing data, a new technology for single molecule proteomics
Matthew Smith (1), Edward Marcotte (2)
(1) Oden Institute, The University of Texas at Austin, Austin, Texas, United States
(2) Department of Molecular Biosciences, The University of Texas at Austin (COI: E.M.M. is a co-founder, shareholder, and
SAB member of single molecule protein sequencing company, Erisyon Inc.), Austin, Texas, United States
Purpose Tools for protein identification and quantification lag DNA and RNA sequencing techniques in sensitivity and
throughput, pushing proteomics researchers to pursue new high-performance approaches. To address these issues, our
group invented fluorosequencing, a single molecule protein sequencing technology that incorporates features of nucleic acid
sequencing for proteomics. In fluorosequencing, proteins are proteolytically digested into peptides, and specific amino acids
are labeled with fluorescent dyes. Labeled peptides are immobilized in a flow-cell where, using Edman degradation
chemistry, they are sequenced in parallel while being imaged by single molecule microscopy. Fluorosequencing produces
sequencing reads from many individual molecules simultaneously, with the expected elevation in noise and errors that must
be addressed in subsequent computational analysis. Methods By modeling sequencing errors (failed chemistry, labeling,
dye bleaching, etc) on large synthetic datasets, we considered four machine learning approaches to accurately assign
fluorosequencing reads to parent peptides: (1) the k-Nearest-Neighbor method, (2) a Random Forest classifier, (3) Bayesian
classification based on Hidden Markov Models of the chemical processes used in sequencing, and (4), a hybrid approach
combining the k-Nearest-Neighbor method with the Hidden Markov Models. Results For small sets of proteins, RF, HMM,
and kNN+HMM classifiers significantly outperformed kNN for classification precision and recall, with the HMM based
classifier giving the best results. However, at human proteome scale, involving hundreds of thousands of possible peptides,
RF and HMM models become intractable. We found that kNN+HMM offers a good compromise, scaling to the full proteome
and significantly outperforming kNN. Conclusions A hybrid kNN+HMM machine learning strategy successfully assigns
fluorosequences to their parent peptides at good precision and recall, while also scaling to the full human proteome and
modeling known sources and rates of fluorosequencing errors in a human interpretable form.
P01.22 Universal Hydrophilic affinity SPE based peptide clean-up workflow for parallel sample processing
Stoyan Stoychev (1), Ireshyn Govender (1), Previn Naicker (1), Isak Gerber (1), Justin Jordaan (1)
(1) ReSynBio, Pretoria, South Africa
Several methods for proteome sample preparation exist that deal with upfront protein capture, clean-up and digestion.
These include classical affinity-based precipitation, in-solution digestion, FASP and more recent on-membrane and on-bead
precipitation workflows. However, on peptide level C18-based desalting is still the go-to-method for peptide clean-up. While
this approach is efficient, it suffers from well-documented drawbacks such as the inability to deal with/eliminate
contaminants including detergents and polymers as well as the requirement to remove (or at the very least reduce) the high
concentration of the organic solvent utilised for peptide elution. Here we present a workflow that can be coupled for
universal clean-up prior to LC-MS analysis, using semi-porous hydrophilic affinity material (ReSyn HILIC) in a 96 well plate
format for parallel sample processing in a high-throughput manner. The performance was compared to traditional C18 SPE
using colorimetric peptide quantification to estimate the recoveries. LCMS-analysis was further applied to assess potential
bias in peptide capture and release, as well as the overall proteome coverage. The material allows for efficient capture of
Page 59 of 166
peptides under high organic conditions and the efficiency to remove salts, polymers and detergents from peptide samples
will be showcased.
P01.23 EPITOME ANALYSIS OF THE HUMAN PLASMA PROTEOME DELIVERS DIAGNOSTIC AND THERAPEUTIC
CANDIDATES.
Laszlo Takacs (1), Jozsef Lazar (1), Peter Antal-Szalmas (2), Istvan Kurucz (1), Barbara Uzonyi (3), John Lamont (4),
Janos Fekete (5), Ilona Tornyi (6), Alexandra Pap (7), Alexandra Matola (8), Mihaly Jozsi (3)
(1) Biosystems Immunolab Ltd, Debrecen, Hungary
(2) Univesity of Debrecen, Debrecen, Hungary
(3) Department of Immunology, Eötvös Loránd University, Budapest, Hungary
(4) 4Randox Laboratories Ltd, Crumlin, United Kingdom
(5) Adware Research Kft, Balatonfured, Hungary
(6) Biosystems Immunolab Ltd, Debrecen
(7) Department of Immunology Eotvos university, Budapest, Hungary
(8) Department of Immunology, Eötvös Loránd University, Budapest
Current proteomics technologies focus solely on the quantitative determination of individual proteins, practically no efforts
have been dedicated to the development of proteomics systems tools to simultaneously monitor proteome variability,
accessibility and abundance. Here, we present a robust and analytically validated mAb proteomics technology for profiling
epitopes of the plasma proteome. To this end we developed mAb libraries, QuantiplasmaTM against natural, accessible and
immunogenic plasma protein epitopes. We show that biomarker epitope panels represent a rich and thus far unexplored
source of protein biomarkers with diagnostic potential. We developed, validated and registered PulmolisaTM as a tool to aid
differential diagnostic of lung cancer. Additionally, we tested 156 complement component (C1q, C3, C4, C5, C6, C7, C8,
C9, C4b-binding protein (C4BP) or factor H (FH)) epitope specific mAbs in functional assays and found 42 with biological
activity (inhibition or potentiation). In summary, we validate a novel and expandable proteomics technology that goes
beyond the current limitations of MS tools, apparently to the level of epitopes. The number of protein coding genes is far
exceeded by the number of protein variants that become accessible via epitome analysis. The high ratio of in-vitro functional
activity of the epitope specific mAbs support strongly that epitome profiling, done with BSI’s natural protein immunization
approach, monitors biologically relevant differences, with direct applicability as diagnostic tools and as therapeutic
candidates.
P01.24 Giving CE-MS a New Life in Proteomics: Integrated Solid-Phase Extraction and Detergent Tolerance
J. Will Thompson (1), J. Scott Mellors (2), Joshua P. Gurrette (3), Erin A. Redman (1), Christopher D. Brown (4)
(1) 908 Devices Inc., Boston, Massachusetts, United States
(2) 908 Devices Inc., Boston, North Carolina, United States
(3) 908 Devices Inc, Boston, Massachusetts, United States
(4) 908 Devices, Boston, Massachusetts, United States
Page 60 of 166
Microchip CE-MS is an inherently low-volume separation which achieves high separation efficiency. It is capable of
producing small ‘peak volumes’, meaning peptides may typically elute in about 1 to 10 nanoliters, resulting in the potential
for a very sensitive separation. Additionally, peptide CE-MS separation can be performed from samples containing
detergents and other carriers, which could be a distinct advantage for low-concentration samples. However, CE-MS
traditionally suffers from poor concentration-sensitivity because of the small sample injections, which are on the order of
100-1000 times smaller than typical volumes for sample loading in capillary liquid chromatography. To address this
microchip CE limitation, we have integrated a small solid phase extraction bed in the sample channel of the microfluidic
device, allowing a large sample volume (up to 2 uL) to be loaded directly into the separation channel prior to CE-MS
separation, a 400-fold volumetric loading improvement. The SPE-CE-MS method including a 0.5 mm bed packed with 5 um
C18 particles demonstrates a capacity in excess of 100 ng peptide directly into the CE-MS channel. Using HeLa lysates
and serum as model systems, we perform proteome separations in 11 minutes with sample concentrations below 0.2 ng/uL
on a modified ZipChip CE coupled to an Exploris 240 MS. Additionally, CE-MS separations are detergent tolerant, so the
SPE-CE-MS procedure can be performed with detergents and other carriers present in the sample. The ability to utilize
carriers could be a major advantage for proteomics sample preparation of single cells, but has been generally avoided
because of the inherent problems with reversed-phase LC intolerance to detergents. We observe up to 100-fold recovery
improvements when samples are prepared and stored with nonionic surfactants present in the sample. Data will be
presented on the relative merits and of various detergents, chaotropes, and small molecule carriers in the SPE-CE-MS
separation.
P01.25 Quality Control of a DIA-MS High-Throughput Proteomics Workflow for CSF Samples from Parkinson’s
Disease Patients
Kirstin E. Washington (1), Rakhi Pandey (1), Danica-Mae Manalo (1), Alejandro Rivas (1), Connor Phebus (1), Matthew
Ayres (1), Koen Raedschelders (1), Niveda Sundararaman (2), Vidya Venkatraman (1), Jennifer E. Van Eyk (2)
(1) Cedars Sinai Medical Center, Los Angeles, California, United States
(2) Cedars Sinai Medical Center, Los Angeles, California
The deployment of automated sample preparation using 96-well plate format, efficient LC-MS workflows and quality control
measure is helpful for consistent large cohort proteomics analysis. As a part of the Accelerating Medicines Partnership in
Parkinson’s Disease (AMP-PD), 2284 human cerebrospinal fluid (CSF) samples from Parkinson’s Disease Patients along
with 624 Quality Control samples have been analyzed over 39 weeks. 4 digestion control reference (DCR) samples were
included along with 75 patient samples per 96-well plate which were processed on the i7 automated workstation for protein
denaturation (TFE), reduction (DTT), alkylation (IAA), and digestion (1:10 protein:trypsin; 4hrs; 42°C). A total of 39 x 96-well
plates were processed and analyzed on a EVOSEP One system coupled to a Orbitrap Exploris 480 MS (Thermo) using
DIA-MS (21 Da windows). Furthermore, a weekly MS-system suitability was carried out using a commercial Hela peptide
digest reference standard and MS-technical control reference (TCR) samples consisting of a pooled pre-digested
Parkinson’s Disease CSF sample run 9 times per 96 plate. In total, 156 DCR, 117 HeLa digests (3 times per week) and
351 TCR were run. For efficiency, we automated QC reports that provided total # peptides and proteins with <20% and
><40% CV% for each 96-well plate. On average, the DCR and TCR samples quantified >13000 peptides corresponding to a
total of 450 456 proteins with >70% of proteins having intraday CV% <40 across all 39 plates, while TCR samples quantified
>600 peptides and 392 proteins, with 70% of proteins having CV% <40%. Variance analysis (PCA) of the DCRs and TCRs
Page 61 of 166
across the 39 plates of PC1 contributing 12.45 and 11.31% and PC2 contributing 6.01 and 5.93%, respectively. Batch
correction was undertaken using Random Forest further reducing this technical variance.>
P01.26 Incorporation of glycoproteome detection into large scale unbiased proteomics studies utilizing
nanoparticles
Bruce Wilcox (1), Kavya Swaminathan (1), John Blume (1), Jared Deyarmin (1), Preston Williams (1), Chinmay Belthangady
(1), Manway Liu (1), Mi Yang (2), Philip Ma (1)
(1) Prognomiq, San Mateo, California, United States
(2) Prognomiq, San Mateo, California
Robust identification, characterization, and quantitation of glycosylated post-translational modifications (PTMs) is of key
interest in disease pathogenesis as glycoproteins play vital roles in maintaining cellular health and glycoprotein
dysregulation is directly associated with numerous diseases, including cancer. Historically, identification, characterization,
and quantitation of the circulating glycoproteome from biofluids has posed challenges due to wide range of abundances,
stability, heterogeneity, and data analysis limitations. While enrichment strategies yield deeper investigation of glycoproteins
and glycoforms, experimental reproducibility and biases still pose considerable challenges. By leveraging Seer’s
ProteographTM nanoparticle technology that enables robust and reproducible measurements across the plasma proteome,
unique proteoforms are specifically captured in corona formation and allow for comprehensive assessment of the circulating
glycoproteome. Furthermore, using Proteograph technology unlocks the possibility to monitor ratio abundances of
glycosylated vs. native forms of proteins, which unveils a new dimension in glycoproteomics for the development of novel
biomarkers. Using the five-nanoparticle panel workflow, we report the detection of distibuitions of glycoproteins across all
five nanoparticles (up to 1% per nanoparticle) in a high-throughput cancer biomarker study of 212 subjects (96 non-cancer,
116 cancer). In this study, we identified a median 1,592 protein groups per subject, a total of 5,099 protein groups across all
samples (process samples included) while maintaining median CV’s ranging 20-25% across all 5 nanoparticles. With this
technology, we detect a higher composition of glycoproteins relative to unenriched sample preparation strategies for
circulating proteins in literature. We also describe a new approach to investigate glycoprotein abundance,
glycoprotein/glycoform identification and characterization across a large cancer cohort study while overcoming the need for
separate upfront sample enrichment. Our results highlight the application of Proteograph nanoparticle technology for deep,
unbiased proteomics biomarker studies including the simultaneous enrichment of glycoproteins to further advance the
detection of cancer biomarkers.
Page 62 of 166
Purpose Recent advances in the throughput of analytical proteomics, now often requiring only minutes per sample for
identification and quantification, necessitate concomitant advances in bottom-up sample preparation workflows. Specifically,
the simplification and ideally elimination of any extraneous steps in a sample preparation workflow increases both
throughput and robustness. Post lysis, one of the most standard steps preceding sample preparation is a protein
concentration assay, often via BCA. While compatible with detergents such as SDS, all assays, including BCA, are not
instant, consume some of your sample, and are subject to interference and, if performed in a 96-well plate, edge effects.
Here, we demonstrate the new concept of direct quantification of cleaned, surface-bound protein on S-Trap 96-well plates
using intrinsic tryptophan fluorescence. Methods S-Trap 96-well plates were made and used as per standard protocols.
Samples of varied hydrophobicity were bound onto plates and washed as per standard protocols; standard BSA curves
were also loaded. Tryptophan fluorescence in both a wet and dry state was measured with an excitation between 269 and
280 nm and an emission of 325 to 475 nm using a Tecan Spark plate reader in top-read emission mode. Protein
concentrations determined via tryptophan fluorescence were compared to BCA for limit of detection, reproducibility, and
dynamic range. Results Fluorescence was measured using a Tecan Spark plate reader for proteins already bound and
washed on an S-Trap 96-well plate and provided with digestion buffer. Fluorescence tracked with protein load at 277 nm
excitation and 350 nm emission in a roughly linear fashion up to 100ug load. The direct-determination method afforded
protein quantification from 1ug-100ug in a significantly reduced time compared to BCA assays, without any loss of
sample. Conclusion Direct determination of protein concentration with intrinsic clean-up removing the need for protein
assays.
Purpose Recent advances in analytical proteomics throughput, necessitate concomitant progress in bottom-up sample
preparation workflows. With the ability to handle extremely diverse samples without the need for protocol modification, the
S-Trap sample preparation system has found widespread adoption. To date, S-Traps have been available as both spin
columns of varying capacities and in 96-well plate format. To keep pace with ever-increasing throughput and single-cell
analyses, we developed the new S-Trap 384-well plate, suited for protein loads from single cells and sub µg quantities to
100 µg. Methods 384-well S-Traps were manufactured to match the performance of S-Trap micros. Replicate sample
preparations were performed on 384-well plates; S-Trap micro columns, with 96-well plates used as a baseline. The
standard steps of lysis, reduction, alkylation, denaturation, binding, washing and digestion were performed. Well-to-well and
plate-to-plate variation were compared via analysis by LC-MS. Results 384-well S-Trap plates performed consistently with
the results of spin columns and 96-well plates. Well-to-well and plate-to-plate variation was equivalent to the reproducibility
of replicate technical injections. With 1 hr digestion at 47 °C, an average sample processing speed of less than 10
sec/sample could be attained, a speed compatible with current rates of detection and quantification. Initial experiments in
single-cell analyses indicate that the 384-well plate affords more accurate representation of the underlying biological states
Page 63 of 166
by quenching biochemical reactions through direct dispensing of cells into 5% SDS. Conclusion Robust sample preparation
suited for scales from single cell to 100 µg at an average of < 10 sec/sample.
Purpose Recent advances in proteomics data acquisition throughput necessitate concomitant advances in bottom-up
sample preparation. The elimination of any and all extraneous elements in a workflow increases both throughput and
robustness. One of the more tedious steps in proteomics sample processing is post-elution sample dry-down. We present
the new S-Trap turbo: yielding minimal elution volumes of highly concentrated peptides suited for immediate analysis.
Methods New snap-cap S-Trap turbo micro columns were constructed via plastic injection molding. The columns
incorporated newly developed, compressed polymeric traps derivatized with novel surface modifications. The standard S-
Trap protocol steps of lysis, reduction and alkylation, denaturation, binding, washing and tryptic digestion were performed
both for standard and turbo S-Traps. Samples were analyzed by LC-MS. Sample yield was compared and quantified using
BCA and/or fluorescent assays. Sample quality was compared by peptide and protein identification rate and reproducibility
of quantifications. Results Turbo traps were compared to traditional S-Traps using three samples of highly varied
hydrophobicity between < 1 µg to 100 µg of protein was processed as described above. S-Trap turbo elutions as low as 5
µL were found to be reproducible and similar or better than standard S-Trap digestions as judged by completeness of
digestion, peptide yield and identifications. S-Trap turbo elutions could be immediately loaded onto an autosampler. No
significant loss of hydrophobic peptides between the standard S-Trap protocol and S-Trap turbo was observed. S-Trap
turbo allows proteomics researchers to go from complex sample to ready-to-inject peptides in record time, with minimal
steps and equipment. Conclusion A highly robust sample preparation system, which negates speedvacing and increases
throughput.
Page 64 of 166
P02: Aging and Neurodegenerative Diseases
P02.01 Deep and quantitative succinylation profiling from dietary treatment in liver
Conor Mullens (3), Joanna Bons (1), Jacob Rose (1), Kevin Peasley (2), Takuto Chiba (2), Francesco Pingitore (4),
Christopher Adams (4), Eric Goetzman (5), Birgit Shilling (1)
(1) Buck Institute for Research on Aging, Novato, California, United States
(2) University of Pittsburgh, Pittsburgh, Pennsylvania, United States
(3) Bruker Datonic, San Jose, California, United States
(4) Bruker Daltonics, San Jose, California, United States
(5) University of Pittsburgh, Pittsburgh, Pennsylvania
Abstract: Protein succinylation is an unexplored PTM and thought to have dramatic consequences on protein structure and
thereby function given the large size of the succinyl moiety (100 Da) and amino acid site selectivity (K) changing charge
from +1 to -1. In this study, we combine succinylated peptide-level enrichment with timsTOF technology (DDA and DIA) to
both identify sites of lysine succinylation and measure changes in succinylation upon dietary treatment in mouse liver
tissues. Methods: Control and dietary-treated wild-type mouse livers (N=8) were homogenized, digested and enriched for
succinyl modifications using the anti-succinyl (Succ-K) affinity motif from Cell Signaling Technologies. Peptides were run in
both a DDA and dia-PASEF schema(s) on the timsTOF Pro using variable CE’s and mobility windows in a 90 min. gradient
and short 45 min. gradients. Data was analyzed by PaSER (DDA) and Spectronaut v15 in directDIA mode (dia-
PASEF). Preliminary Data: The DDA search results produced in real time by PaSER identified more than 6,300 peptides of
which greater than 3,400 were succinylated as identified in a single shot analysis. Within biological replicates dia-PASEF
identified on average 8,200 peptide groups of which on average 4,600 were succinylated. Comparing control versus treated
groups, more than 4200 succinylated peptides were shown to be statistically significant (q<0.05, FC>0.58) in either up or
down regulation, where approximately 3100 were upregulated upon dietary treatment. In previous studies it has been shown
that the proteome upon dietary supplement remains largely unchanged. These data suggest the complete succinylome re-
wiring upon dietary feed and point toward Sirtuins as a novel drug target. Novel Aspect: High sensitivity profiling of succinyl
PTMs using trapped ion mobility
P02.02 Utilizing an Iodine Unfolding Assay to Analyze the Effects of Aging on Protein Stability
Tyler Bateman (1), Chad Hyer (1), Lavender Lin (1), Michael Zackrison (1), John Price (1)
(1) Brigham Young University, Provo, Utah, United States
As humans age, diseases like Alzheimer’s, cancer, and type II diabetes become more prevalent in the population. This is
due in part to the loss of proteostasis, or the ability to regulate the concentration and quality of protein in the body. The
proteostasis of an organism is dictated by the quality of its proteins. One metric of protein quality is the thermodynamic
stability of the folded protein structure. Overly stable proteins can form aggregates or linger in a biological system for too
long, while unstable proteins might not be retained long enough to perform their specialized tasks. Understanding protein
stability can help researchers to combat age-related diseases by revealing novel mechanism of these ailments. We have
developed a proteome-wide folding stability assay, wherein denaturation curves and an iodine reporter are used to quantify
protein stability at multiple residues within the protein sequence. In this study, we use the unfolding assay to examine the
Page 65 of 166
protein stability of mouse liver proteins during dietary interventions which modify their aging rate and lifespan. For example,
caloric restriction of mice has been shown to extend their life span in previous studies. This extension is thought to result
from improved proteostasis in mice with restricted diets. We can see how changes in proteostasis reflect on protein stability
by comparing the stability of proteins from mice with different diets. This will provide unique insights into the effects of aging
on an organism’s proteome and overall human health.
P02.03 Monitoring regionally-regulated control of protein turnover and concentration across the brain proteome
using targeted Microsampling and AutoPOTS
Rebecca S. Burlett (1), Leena M. Patil (1), Holden Kelly (1), John C. Price (1)
(1) Brigham Young University, Provo, Utah, United States
Imbalances within the brain can lead to neurodegenerative diseases like Alzheimer’s Disease (AD). The risk of developing
AD has been linked to the isoforms of Apolipoprotein E (ApoE) that the body produces. However, it is unknown how ApoE
(a lipid transport protein) affects brain proteostasis. Thus, it is important to understand the connection to AD. To understand
this model mice were given deuterated water, over a timeline, prior to harvesting tissue. Utilizing a cryostat, we cut 20
micron slices of brain tissue that have been labeled metabolically with deuterium to allow analysis of the Hippocampus,
Entorhinal, and Visual regions. Slices were then treated with a Nissl Stain and identified using Allen Brain Atlas. Next, Laser
Capture Microdissection (LCM) was used to collect target specific 200x200x20 micron samples. AutoPOTS methods were
used to prepare samples for mass spectrometry. Trypsinized samples were analyzed using LC-MS to measure protein
abundance and turnover. Readings for each of the regions were compared against each other to identify patterns of
regulation within protein expression. We have reproducibly identified ~2500 proteins with this targeted microsampling
technique, almost as much as a normal LC-MS/MS run. Within the ~2500 proteins we have been able to identify, further
analysis has provided us with turnover rates and concentrations that were put through analysis software, i.e. DeuteRater or
Metaboanalyst. The outputs given by DeuteRater and Metaboanalyst were then run through python scripts that grouped the
proteins into specific pathways provided by DAVID and original data. Those identified pathways have allowed us to pinpoint
changes seen between the regions being compared. Partnering protein turnover rates with protein concentration we are
able to investigate which pathways are being regulated by synthesis or by degradation. We are working towards a better
understanding of how the brain’s regions differ and are affected by AD overall.
P02.04 The Role of Epigenetics-Mediated Neuroplasticity in Alzheimer’s Disease as a Means of Understanding the
Impacts of Apolipoprotein E Variants.
Noah Earls (1), Benjamin Jones (2), Katherine Brown (1), Kimberly Wagstaff (1), Ethan Smith (1), Noah Moran (1), Nathan
Zuñiga (1), John C. Price (3)
(1) Brigham Young University, Provo, Utah, United States
(2) Brigham Young University, Utah, Utah, United States
(3) Brigham Young University, Provo, Utah
Page 66 of 166
One of the primary modes of thought behind the causes of Alzheimer’s Disease (AD), is the role of Apolipoprotein E (ApoE)
variants. Several studies have highlighted the protective qualities of the ApoE 2 variant against several neurodegenerative
disorders such as AD, Huntington’s Disease, and Parkinson’s Disease. Conversely, the ApoE 4 isoform contributes to the
risk for these same neurological disorders, while ApoE 3 serves as a “neutral” predictor for AD. Additionally, AD
has long been associated with the effects of neuroplastic changes found in patients with AD and how they relate to memory
loss and dementia. Research has provided evidence that epigenetic factors (i.e., DNA Methylation, Histone Modifications,
etc.) have a significant contribution to the neuroplasticity of AD. While ApoE has long been associated with these
epigenetic variations, it remains to be seen what effect ApoE has on neuroplasticity-related proteins and how these changes
demonstrate the predisposition for Alzheimer’s Disease associated with ApoE 4. Our research aims to discover the
connections between these ideas by analyzing the many factors that result in the neuroplastic changes common with
AD. Using a transgenic mouse model on an LCMS system to understand the human proteome, we have analyzed both
brain membrane and brain cytosolic proteins of these mice in search of significant changes between ApoE 2, ApoE 3, and
ApoE 4. Our preliminary data shows several proteins strongly associated with neuroplasticity, such as Synaptopodin, that
are significantly changed in concentration in ApoE 2 as compared to ApoE 3, and in ApoE4 as compared to ApoE 3. We
expect this trend to also be expressed when we analyze the turnover rates. Here we show that we have been able to use
LCMS data to detect significant changes in Neuroplasticity related proteins in both ApoE 2 and ApoE 4.
Aging is a highly complex process with distinct cellular hallmarks that include epigenetic alterations, mitochondrial
dysfunction, nutrient sensing, and cellular senescence. While these processes are distinct, they are also highly connected,
as the perturbation of one of these hallmarks can influence another. As such, elucidating the areas of intersection between
hallmarks could help define the mechanistic underpinnings of aging. Thus, the overarching goal of this proposal is to identify
novel crosstalk mechanisms between metabolism, epigenetics, senescence, and other aging hallmarks to uncover novel
targets that can promote healthy aging. With regards to cellular senescence, the process where a cell limits its ability to
proliferate, it is a critical component of aging as senescent cells are accumulated in aged tissue and the selective removal of
these cells can increase lifespan and ameliorate age-related diseases. Senescent cells undergo multiple cellular and
molecular changes which include altered chromatin landscape, metabolic reprogramming, and the secretion of inflammatory
cytokines and growth factors (SASP). We discovered that the least studied sirtuin, SIRT7, which has histone and non-
histone substates, is targeted for degradation upon oncogene-induced senescence (OIS) and this precedes the
upregulation of senescence associated proteins like p16 and Lamin B1. Rescue of SIRT7 loss during OIS by exogenous
SIRT7 expression delayed senescence and importantly, diminished expression of SASP. These data suggest SIRT7 is a
key regulator during senescence. Through immunoprecipitation-mass spectrometry (IP-MS), we determined E3 ubiquitin
ligases that interact with SIRT7 upon senescence induction. These studies reveal a novel function of an epigenetic factor
that is critical in aging and senescence and will allow deeper investigation into the crosstalk of epigenetics and metabolism.
Page 67 of 166
P02.06 Dynamics of Huntingtin Protein Interactions in the Striatum Identifies Candidate Modifiers of Huntington’s
Disease
Todd Greco (1), Christopher Secker (2), Eduardo Silva Ramos (2), Joel Federspiel (1), Jeh-Ping Liu (3), Alma Perez (4),
Ismael Al-Ramahi (4), Jeffrey Cantle (5), Jeffrey Carroll (6), Juan Botas (7), Scott Zeitlin (3), Erich Wanker (8), Ileana
Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
(2) Max Delbrück Centre for Molecular Medicine, Berlin, Germany
(3) University of Virginia School of Medicine, Charlottesville, Virginia, United States
(4) Baylor College of Medicine, Houston, Texas, United States
(5) Western Washington University, Bellingham, Washington
(6) Western Washington University, Bellingham, Washington, United States
(7) Baylor College of Medicine, Houston, Texas
(8) Max Delbrück Centre for Molecular Medicine, Berlin, United States
Introduction: Huntington’s disease (HD) is an autosomal dominant fatal neurodegenerative disorder with one causative
gene, huntingtin (HTT). Yet, the pathobiology is complex, suggesting that cellular factors influence disease progression.
Here, we define HTT protein-protein interactions (PPIs) that are altered in abundance and stability due to the mutant protein
with expanded polyglutamine in the mouse striatum, a brain region with selective HD vulnerability. By combining protein
interactions investigations in mouse brain with validation in human cells and functional assays in Drosophila, we discover
mutant HTT-dependent interactions whose genes modify disease pathogenesis. Method: The striatum of 2- and 10-month-
old mice expressing 3XFLAG-Htt with a knock of human exon 1 containing either normal (20Q) or expanded (140Q)
polyglutamine tracts were used for label-free and isotope-labeled affinity purification-mass spectrometry experiments to
determine the specificity and relative stability of the interacting protein, except here adapted for tissues. Polyglutamine-
regulated PPIs were validated in a human HD cell model using bioluminescence-based two-hybrid assays and Drosophila
gene homologs were tested by a motor performance assay in an HD fly model expressing fragment (HTTNT231[Q128]) or full
length (HTTQ200) mHTT. Results: Using metabolically labeled tissues and immunoaffinity purification-mass spectrometry, we
found that polyglutamine-dependent modulation of HTT PPI abundances and relative stability starts at an early stage of
pathogenesis in a 140Q HD mouse model. We identified direct and indirect PPIs that are also genetic disease modifiers
using in-cell two-hybrid and behavioral assays in HD human cell and Drosophila models, respectively. Validated, disease-
relevant mHTT-dependent interactions encompassed mediators of synaptic neurotransmission (SNAREs and glutamate
receptors) and lysosomal acidification (V-ATPase). Conclusion: Using complementary AP-MS approaches, in-cell two
hybrid assays, and fly genetic disease modifier assays, our results provide a resource for understanding mutant HTT-
dependent dysfunction in cortico-striatal cellular networks, partly through impairment in synaptic communication and the
endosomal-lysosomal system.
P02.07 A proteomic network approach to define proteins associated with cognitive resilience in Alzheimer's
disease
Cheyenne Hurst (1), Derian Pugh (2), Duc Duong (3), David Bennett (4), Jeremy Herskowitz (2), Nicholas Seyfried (1)
(1) Emory University, Atlanta, Georgia, United States
(2) University of Alabama Birmingham, Birmingham, Alabama, United States
Page 68 of 166
(3) Emory Universoty, Atlanta, Georgia, United States
(4) Rush University Medical Center, Chicago, Illinois, United States
The core pathologies of Alzheimer’s disease (AD) include extracellular plaques of aggregated amyloid-β (Aβ) and
intracellular accumulation of tau neurofibrillary tangles. There is increasing recognition of a subpopulation of “cognitive-
resilient” individuals that live to advanced age with intact cognitive function despite the presence of plaque and tangle
pathology. Thus, identifying molecular mechanisms in brain associated with cognitive resilience may reveal novel
therapeutic pathways and targets for AD. Here, we performed isobaric tandem mass tag (TMT) labeling coupled with off-line
fractionation and liquid chromatography with tandem mass spectrometry (LC-MS/MS) on post-mortem brain tissue samples
from the Religious Order Study and Memory and Aging Project longitudinal cohorts across two brain regions (Broadman
area 6 and 37). This included pathology-free controls (n=24), cognitively normal “resilient” individuals with AD pathology
(n=53) and symptomatic AD (n=34) cases. Following database search, ~8000 high abundance proteins were quantified.
Consensus Weighted Gene Correlation Network Analysis (cWGCNA) of both brain regions generated 38 modules, groups
of highly co-expressed proteins, which were correlated to clinical and pathological phenotypes and assessed for cell-type
specificity and gene ontologies. Modules positively correlated to core AD pathology were significantly increased in
both resilient and AD cases compared to controls. Conversely, modules enriched for synaptic and other vulnerable proteins,
including VGF, NPTX2, and NRN1, were significantly decreased in AD, but not resilient cases. To further validate our
systems-level data and explore potential neuroprotective mechanisms, we treated rat primary neuronal cultures with
recombinant NRN1 protein with or without toxic Aβ 1-42 oligomer insult followed by dendritic spine morphometric analysis
and in-depth proteomics. Morphometrics revealed NRN1 treatment in the presence of Aβ oligomers rescued overt dendritic
spine loss observed with Aβ insult alone. Together, our findings support network nomination of resilience-associated
proteins, including NRN1 which protects against amyloid-induced neurotoxicity that may in-turn promote cognitive resilience
in AD.
Apolipoprotein E (APOE) isoform expression is a significant risk factor for Alzheimer’s Disease (AD). The three most
common APOE alleles include: APOE3, the baseline variant, APOE 2, which bestows increased resistance against AD
development, and APOE4, which increases the risk of the development of AD. The biochemical mechanism behind this
disparity in AD risk remains unknown. To analyze the differences between polymorphisms, we homogenized brain samples
from transgenic mice with each genotype; these homogenized samples were centrifuged to separate cytosolic proteins from
membrane-bound proteins. Additional washes to the membrane pellet remove cytosolic contaminants that would otherwise
limit our ability to see low abundance proteins on the membrane. Membrane samples were processed via state-of-the-art
liquid chromatography and mass spectrometry (LC-MS), and this data was used to identify and quantify proteins in the
membrane fractions.This data was further processed by comparing the concentrations of the proteins between
polymorphisms to discover which proteins were upregulated or downregulated between polymorphisms. The regulation of
Page 69 of 166
membrane-bound proteins in brain tissues may offer a deeper understanding of the signaling and structural differences
between polymorphisms. Proteins in these tissues include receptors involved in cell signaling. Preliminary data shows a
large number of proteins with significantly changed metabolism between APOE2 and APOE3. Collagen proteins and
components of the extracellular matrix were some of the most significantly changed. Comparing both concentration and
turnover rate changes gives us insight into the changes in protein metabolism. We will discuss our collagen and
extracellular matrix protein observations in the context of the development and spread of amyloid beta aggregates, which
are a leading biochemical marker for AD.
Human health is determined by the proper balance and function of proteins, which is called proteostasis. When proteostasis
is interrupted, disease and death occur. Therefore, minimizing disease and maximizing life are top priorities in the scientific
community. Some literature has shown that calorie restriction (CR) in mice prolongs lifespan relative to mice on an ad
lipitum (AL) diet. This difference suggests that there are changes in the proteome. Those changes have not been fully
revealed, so we did a follow up experiment. We put mice on two main diets (AL and CR) and five sub diets of protein to
carbohydrate (PC) ratios (0.1, 0.2, 0.4, 0.8, 1.6). Liver tissue samples were homogenized and proteins were extracted,
cleaned, and digested by trypsin. After analysis on LC/MS, data was put though PEAKS software. Our preliminary data
suggest that mice on CR respond more quickly to dietary change than AL mice. This suggest changes in the proteome that
lead to those fluctuations. We used LC/MS to quantify the differences in protein concentrations between AL and CR mice.
This knowledge will identify changes in metabolic pathways which support longer lifespan due to nutrient type and total
calories. We will present this data and compare against efforts to increase human lifespan.
P02.11 Multi-omics analysis of the exosome senescence-associated secretory phenotype to study aging biology
Sandip Kumar Patel (1), Jacob Rose (1), Rebecca Beres (2), Joanna Bons (1), Nathan Basisty (3), Roland Bruderer (4),
Lukas Reiter (5), Erin Baker (6), Judith Campisi (1), Birgit Schilling (1)
(1) Buck Institute for Research on Aging, Novato, California, United States
(2) North Carolina State University, Raleigh, North Carolina, United States
(3) National Institute on Aging, Baltimore, Maryland, United States
(4) Biognosys AG, Schlieren, California, United States
(5) Biognosys AG, Schlieren, Switzerland
(6) North Carolina State University, Raleigh, North Carolina
Cellular senescence induced by endogenous and exogenous stresses is a common aging hallmark. Senescent cells can
drive aging phenotypes by paracrine signaling of the senescence-associated secretory phenotype (SASP), which can impair
tissue regeneration and drive chronic age-associated diseases. Exosomes, a component of the SASP, are small, lipid-
bilayer enclosed, cell-derived nanoparticles that play a vital role in cell-to-cell signaling by delivering bioactive
Page 70 of 166
cargo. Senescence-associated exosomes, largely unexplored, could serve as aging biomarkers. We developed robust and
efficient plasma exosome isolation and analytical workflows using genotoxic- and mitochondrial dysfunction-induced
senescent IMR-90 human fibroblasts. For model development, 30 plasma samples from young and old humans and 20
conditioned media samples from senescent IMR-90 cells were analyzed using proteomics, lipidomics, and miRNA
sequencing. We generated a deep plasma exosome-specific spectral library comprising ~2,300 exosome proteins by
combining data-dependent acquisitions (DDA) and direct Data-Independent Acquisitions (directDIA) spectral libraries. Our
data demonstrate that the directDIA spectral library enables efficient, time-saving proteomic workflows. Partial least
squares-discriminant analysis (PLS-DA) of DIA-data could differentiate the plasma donors into young and old groups with
no overlap. We identified 1,310 plasma exosome proteins with two unique peptides, of which 144 show differential
abundance comparing old and young donors. In IMR-90 cells, 1852 exosome proteins with 2 unique peptides were
identified, of which ~1000 changed with senescence. Interestingly, 42 significantly altered proteins are common in human
plasma and the exosome SASP. 331 miRNAs were detected in human plasma exosomes, of which 88 and 17 miRNAs
were unique to old and young individuals, respectively. Exosome lipidomics on a timsTOF ion mobility MS platform
identified >300 lipid species and 38 were differentially present in young and plasma. We will validate our model with larger
cohorts and integrate proteomics lipidomics, and miRNA data to discover molecular aging markers.
P02.12 Deep and Unbiased Plasma Protein Profiling of Alzheimer’s and Mild Cognitive Impairment Subjects with a
Novel Multi-nanoparticle Approach
A Siddiqui (1), RB Benz (1), M Goldberg (1), JC Cuevas (1), W Manning (1), X Zhao (1), TL Platt (1), M Ko (2), EM Elgierari
(1), M Figa (2), H Xia (1), H Guturu (1), M Zamanighomi (1), J Zhang (1), S Batzoglou (1), OC Farokhzad (1)
(1) Seer, Redwood City, California, United States
(2) Seer, Redwood City, California
Blood plasma is a rich source of protein biomarkers for early detection of diseases, but its large dynamic range of protein
concentrations necessitates complex workflow and trade-offs between throughput, scalability, coverage, and precision. Here
we use a deep and quantitative proteome profiling platform, ProteographTM Product Suite, which leverages multiple
nanoparticles, engineered with distinct physicochemical properties to provide broad coverage of the plasma proteome at
scale. In this study, we aim to identify protein biomarkers for Alzheimer’s disease (AD) from blood with this untargeted
plasma protein profiling approach, for AD and Mild Cognitive Impairment (MCI) condition. Previously, plasma samples from
200 subjects comprising 50 AD, 50 MCI, and 100 controls were profiled using Proteograpgh plasma protein profiling
platform1. Using 5 injections per sample, proteins were quantified by data-independent acquisition (DIA) liquid-
chromatography mass-spectrometry (LC-MS). Normalized peptide intensities were used to develop models for class
discrimination. Across the samples, 2,391 plasma proteins were detected, with 2,085 in at least 25%. 25,593 peptides were
detected, with 15,661 in at least 25% of the samples. Univariate analysis identified 441 and 526 proteins that were
significantly different in AD or MCI versus control, respectively. Random-forest classification with ten rounds of 10-fold
cross-validation yielded ROC AUCs that were at least 0.90 for AD and MCI. These analyses identified a combination of
known and potential new candidate plasma protein markers, confirming the Proteograph platform’s ability to generate
profiling data in a deep, broad, and rapid fashion, enabling large-scale studies to detect novel insights with clinically relevant
potential. We are in process of conducting a second study of 200 subjects comprising 100 AD and 100 controls and will
present the results from this second cohort at the meeting. References. 1. Blume JE et al. Nature Communications.
2020;11(1):3662–1
Page 71 of 166
Contact: Asim Siddiqui, [email protected]
Alzheimer’s Disease (AD) carries finality to the declining quality of life in many aging members of the population. While
several genetic risk factors have been characterized, the mechanisms by which they are associated with AD are not well
understood. Pathological patterns have been observed in AD progression, but we do not understand them well enough to
intervene medically. Using proteomics methods to better characterize the biochemistry of AD risk across different tissues
and organs is a very promising way to identify the mechanisms of risk and potential interventions. The genetic risk factor
that we study involves three variants of the lipid transport protein ApoE. This is among the strongest genetic correlates of
AD risk, and it does much of its work between the liver and the brain. Both organs produce ApoE, and the liver produces
many lipids for energy and structure in the brain. As AD risk increases, we and others find this relationship dysregulated,
with changes in both organs. The clinical manifestations are far more obvious in the brain, but MS proteomic analysis also
gives insight into the mechanistic underpinnings that are less obvious in the liver. Here I will present our results investigating
proteomic perturbations in mice with transgenic insertions of ApoE variant 2, 3, or 4 as our AD risk models. We analyzed
both membrane and cytosolic protein populations from homogenized liver and brain tissue, and we compared differential
impacts of the ApoE variants on the two organs. As we compare these organs, we can better understand the full-body
contributions to the risk for AD pathology.
P02.14 An Untargeted Metabolomic Investigation of Amyotrophic Lateral Sclerosis Decedent Brain Tissue by IR-
MALDESI Mass Spectrometry Imaging
Alexandria L. Sohn (1), Lingyan Ping (2), Jonathan D. Glass (2), Nicholas T. Seyfried (2), David C. Muddiman (1)
(1) North Carolina State University, Raleigh, North Carolina, United States
(2) Emory University School of Medicine, Atlanta, Georgia, United States
Amyotrophic lateral sclerosis (ALS) is a fatal, idiopathic neurodegenerative disease characterized by progressive motor
neuron degradation with limited treatment options. This disease state is classified based on presence or absence of family
history, respectively differentiating familial ALS (fALS) and sporadic ALS (sALS), and further specified with regard to genetic
mutations if applicable (e.g. C9orf72). Because metabolites provide insight into dysregulation of upstream biological
pathways, an untargeted metabolomic study of brain tissue could identify biomarkers and exogenous compounds related to
ALS in hopes to optimize diagnosis protocol and extend the life expectancy of ALS patients. Infrared matrix-assisted laser
desorption electrospray ionization (IR-MALDESI) mass spectrometry imaging (MSI) is a powerful approach leveraged in this
study to analyze the spatial distribution of analytes across a tissue with respect to their abundance without the necessity of
an organic matrix. Post-mortem human brain tissue was imaged at 10-micron thickness from three subgroup classifications:
non-neurological controls, sALS decedents, and sALS decedents positive for the C9orf72 repeat expansion mutation.
Selection of sALS and C9orf72 positive cases for analysis respectively was due to the significantly higher rate of incidence
of sALS (~90%) relative to fALS, and the C9orf72 mutation relative to other genetic mutations associated with ALS (e.g.
SOD1, TARDBP, FUS). Data collected was annotated by METASPACE and further analyzed utilizing MSiReader and
Page 72 of 166
Excel. Additionally, a normalization approach was developed and employed for the comparison of tissues within and across
diagnosis subgroups to reduce analytical variability between biological samples for a more accurate representation of
biological variability in a MSI workflow. Ultimately, over 100 annotated metabolites were common between all samples and
their respective differences in abundance between groups are under further investigation for biological differences between
disease classes. Furthermore, MS/MS follow-on studies will be pursued for metabolite structure validation to enhance our
understanding of ALS-related metabolites.
P02.15 Investigating Racial Disparities in Alzheimer’s Disease using Quantitative Brain Proteomics
Jasmin B. Tindal (1), Kaitlyn E. Stepler (2), Shania Hansen (3), Timothy J. Hohman (3), Lisa L. Barnes (4), David A. Bennett
(4), Renã A.S. Robinson (1)
(1) Vanderbilt University, Nashville, Tennessee, United States
(2) Vanderbilt Univerisity, Nashville, Tennessee, United States
(3) Vanderbilt Memory and Alzheimer’s Center, Nashville, Tennessee, United States
(4) Rush Alzheimer’s Disease Center, Chicago, Illinois, United States
Compared to the non-Hispanic White population, the African American/Black population is disproportionately affected by
Alzheimer’s disease (AD) with a 2-3 times higher incidence rate. Understanding the molecular basis of this disparity has
been challenged by a lack of samples from African American/Black participants in basic science studies. Previously, our
laboratory determined many proteomics changes in AD that are universal across racial/ethnic backgrounds. However, we
also observed proteins that had significant differences between cognitively unimpaired and AD participants that were
associated with racial and ethnic background of the participants. Here we applied a proteomic workflow to postmortem
tissue from the inferior parietal lobule (IPL) and prefrontal cortex (PFC) brain regions across cognitively unimpaired and AD
adults from the Religious Orders Study and Rush Memory and Aging Project. The individuals in the cohort self-identified as
African American/Black or non-Hispanic White. We employed multiplexed proteomics with Tandem Mass Tags (TMT),
reversed-phase fractionation, and nanoflow LC-MS/MS analysis using Q-Exactive HF and Orbitrap Lumos mass
spectrometers. Overall, >9000 proteins were identified across the two regions and were subjected to regression analysis.
Preliminary data of the IPL suggest >75 proteins had changes in AD that were significantly impacted by race. This
presentation will highlight differentially expressed brain proteins in AD, including those associated with racial backgrounds,
and implications for disease understanding.
P02.16 Mapping the 3 D Proteome using Surface Accessibility in Animal Models of Disease
John Yates (1), Ahrum Son (1), Casimir Bamberger (1), Sandra Pankow (1)
(1) The Scripps Research Institute, Lajolla, California, United States
Methods to determine the surface accessibility of amino acids on proteins can provide information about conformational
changes to proteins because of mutation, drug treatment, or other perturbations. These methods such as FPOP, SPROX or
LiP are collectively described as protein footprinting. We describe an approach using formaldehyde to label accessible
lysine residues in animals using perfusion of formaldehyde into the animal which is then converted to dimethyl tags using
Page 73 of 166
cyanoborohydride. Proteins are labeled both in situ and in vivo. As has been used in quantitative proteomics, different stable
isotope elements can be used in formaldehyde and cyanoborohydride to create different weight tags and thus a secondary
labeling after proteolytic digestion of proteins using different weight reagents labels lysine residues that were not. After
removal of tissue it is homogenized, solubilized and digested and then labeled with a different weight combination of
formaldehyde and cyanoborohydride. Quantitative differences in surface accessibility can be observed to identify changes
from a control or healthy state. This method is being used to look at protein misfolding in connection with Alzheimer's
disease in a mouse model.
P02.17 Increased Sequence Coverage of Low-abundance Biomarkers in Human Serum to Quantify Protein Folding
Stability
Michael Zackrison (1), Lavender Lin (1), JC Price (1), Tyler Bateman (1), Chad Hyer (1)
(1) Brigham Young University, Provo, Utah, United States
The structural integrity of proteins is foundational for their ability to carry out their biological function. The protein folding
stability (PFS) can be quantified by the amount of energy needed to unfold the structure. The alteration of PFS can result in
aggregation and a decrease in proteostasis. With the novel iodinated protein folding stability assay developed by our lab, we
can take a snapshot of localized PFS of the proteins across the proteome and compare the stability between various
experimental conditions. Cardiac amyloidosis is a disease caused by amyloid aggregation of misfolded serum protein
transthyretin (TTR) in the heart. Unfortunately, it usually can’t be diagnosed until it is too late. TTR functions as a tetramer.
When the tetramer can’t be stabilized, it dissociates into beta sheet-stabilized monomers that then aggregate as amyloids.
Our goal is to measure the PFS of TTR under different states of disease and intervention and develop a metric to aid early
diagnosis. However, TTR is a low-abundance protein in human serum, and most of its signals are covered by the most
abundant serum protein: serum albumin. Consequently, it is challenging for us to repeatedly measure the PFS of the same
site on a TTR protein across different treatment conditions, affecting our ability to have a reliable measurement. Here we
present our efforts on increasing the sequence/PFS site coverage of TTR while also retaining the effect from its interactions
using offline HPLC fractionation and serum albumin depletion after iodinating our samples. The methods used here not only
promote the PFS measurement of TTR, but also can be applied to other low-abundance biomarkers for further
understanding the mechanism of disease.
Page 74 of 166
P03: Biomarkers and Targeted MS Assays
P03.01 Urine Proteomics for Noninvasive Monitoring Health and Disease in Preterm Neonates: Urinary Biomarkers
for Bronchopulmonary Dysplasia
Saima Ahmed (1), Oludare A. Odumade (1), Patrick van Zalm (1), Kinga K. Smolen (1), Kimino Fujimura (1), Jan Muntel (1),
Melissa S. Rotunno (1), Abigail B. Winston (2), Judith Steen (1), Richard B. Parad (2), Linda J. Van Marter (3), Stella
Kourembanas (1), Hanno Steen (1)
(1) Boston Children’s Hospital and Harvard Medical School, Boston, Massachusetts, United States
(2) Brigham & Women’s Hospital and Harvard Medical School, Boston, Massachusetts
(3) Brigham & Women’s Hospital and Harvard Medical School, Boston, Massachusetts, United States
Introduction: Current techniques to diagnose and/or monitor critically ill neonates with bronchopulmonary dysplasia (BPD)
require repeated invasive sampling of body fluids, which by itself interferes with the wellbeing of these extremely frail
patients. We tested our hypothesis that it is feasible to use non-invasively collected urine samples for a urine proteomics
study with extremely low gestational age newborns (ELGANS) at risk for bronchopulmonary dysplasia to identify proteins,
which have previously been associated with BPD albeit never in urine. Methods: Our urine proteomics pipeline requires as
little as 50 microliters of urine and is thus fully compatible with sample amounts collected from ELGANs in Neonatal
Intensive Care Units (NICUs). We performed a proof-of-concept urine proteomics study validating proteins, which had
previously been identified in invasively collected samples such as blood, broncho-alveolar lavage and/or tracheal aspirates.
We used urine collected within 72 hours of birth from ELGANS (gestational age (26+1.2) weeks) who had been admitted to
a NICU; half of whom eventually developed BPD (n=21), while the other half remained BPD free, i.e., serving as controls
(n=21). Results: Our sample-sparing urine proteomics pipeline clearly identified several BPD-associated changes in the
urine proteome recapitulating well-described changes in the blood, broncho-alveolar lavage and/or tracheal aspirate
proteomes. In addition, several novel urinary biomarker candidates were discovered that allowed to identify the ELGANs at
risk for BPD. Interestingly, sixteen of the identified urinary proteins are known targets of FDA-approved drugs pointing
towards potential novel therapeutic strategies. Conclusion: In addition to validating numerous proteins, previously found in
invasively collected blood, tracheal aspirate, and broncho-alveolar lavage, that have been implicated in BPD
pathophysiology, urine proteomics also suggested novel potential therapeutic targets. Ease of access to urine could allow
for repeated non-invasive monitoring of disease progression and impact of therapeutic intervention in future studies.
P03.02 Reducing Sample Input Requirements for Parallel Reaction Monitoring of Affinity-Enriched Trace
Biomarkers
Nathaniel Axtell (1), Jacob J. Kennedy (2), Rachel Lundeen (2), Andikan Nwosu (1), Thy Truong (1), Jeffrey R. Whiteaker
(2), Xiaofeng Xie (1), Lei Zhao (3), Amanda G. Paulovich (2), Ryan T. Kelly (4)
(1) BYU Department of Chemistry and Biochemistry, Provo, Utah, United States
(2) Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, Washington, United States
(3) Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, Washington
(4) BYU Department of Chemistry and Biochemistry, Provo, Utah
Page 75 of 166
Studies of protein pathways often reveal key proteins whose expression levels can be excellent indicators of normal,
pathogenic, and response states of biological systems. Such proteins make ideal biomarkers for medical diagnostics, as
well as targets for therapeutics. Potential protein biomarkers are routinely quantified in laboratory settings using mass
spectrometry (MS) methods such as multiple reaction monitoring (MRM) and parallel reaction monitoring (PRM). A
significant challenge in translating these measurements to clinical application is overcoming sensitivity hurdles in analyzing
limited clinical sample material. Immunoaffinity enrichment paired with MRM/PRM analysis has been demonstrated to
increase measurement sensitivity for some biomarker candidates; however, this can still require hundreds of micrograms of
protein input to yield quantitative measurements. Here we report on improvements to the standard immuno-PRM workflow
that reduce required sample input requirements by more than an order of magnitude. By employing a MS-compatible
surfactant during sample preparation, we simplified clean-up steps, preventing protein losses due to nonspecific surface
adsorption, and maximizing target proteotypic peptide recovery. Also, our ultra-low flow nanoLC setup increased ionization
efficiency significantly, further reducing sample loss. These changes resulted in a cumulative increase in immuno-PRM
assay sensitivity over current methods. We demonstrated the applicability of our immuno-PRM workflow through detection
and quantitation of DNA damage response proteins FANCD2 and RAD50 expressed in lymphoblast cells. We validated our
results through comparison with a previous study utilizing a conventional immunoaffinity-enriched MRM method.
P03.03 Zeno MS/MS powers quantitative and qualitative workflows for proteomics using the ZenoTOF 7600 system
Alexandra Antonoplis (1)
(1) SCIEX, Redwood City, California, United States
Proteomic workflows cover a wide range depending on project goals, from fully untargeted data dependent acquisition
(DDA) approaches for protein identification, comprehensive data independent acquisition (DIA) strategies for large scale
quantification, and fully targeted quantitative assays for the highest specificity/sensitivity (MRM). Novel Zeno trap
functionality can greatly improve duty cycle in orthogonal pulsing region of a QTOF system, providing large gains in MS/MS
sensitivity. This work investigates the impact of increased MS/MS sensitivity on protein/peptide identification and targeted
peptide quantitation. Microflow chromatography was performed on a Phenomenex Kinetex column at 5 µL/min using a
range of gradients (5, 10, 20, and 45 min). Digested K562 cell lysate was used for DDA testing and the PQ500 kit
(Biognosys, 804 heavy peptides) was used for peptide quantification. Using the ZenoTOF 7600 system, all experiments
were performed with and without the Zeno trap activated. Data was processed using OneOmics suite, SCIEX OS software
and Skyline. To test the impact on peptide MS/MS sensitivity, a targeted peptide quant assay was built using Skyline for the
PQ500 kit (804 peptides). Using the final MRMHR assay with Zeno trap off vs. Zeno trap on, the average gain in MS/MS
fragment peak area was ~5.6 fold. Median peak area CV was 6.1% for 10 replicate injections and median LLOQ was found
to be 193 amol on column. Next, Zeno IDA acquisition parameters were optimized for 4 different microflow gradient lengths
using a design of experiments (DOE) approach. Using optimized settings, comparison data for Zeno trap on and off was
generated for all gradients. Specific gains with Zeno on increase with gradient length, with gains in protein identifications of
more than 35% for longer gradients. Sample loading curves were also generated at all gradient lengths to fully explore
acquisition space, and results will be discussed.
Page 76 of 166
P03.04 Developing and Validating Targeted Mass Spectrometry Assays for Pan-Herpesvirus Viral Protein Detection
and Monitoring of Infection Progression
Michelle Kennedy (1), Joel Federspiel (1), Cora Betsinger (1), Matthew Tyl (1), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
INTRODUCTION: The presence and abundance of viral proteins within host cells signify the cellular stages of viral
infections. Viral proteins are either brought into cells by infectious particles or expressed at specific steps of the replication
cycle. However, methods that can comprehensively detect and quantify these proteins are limited, particularly for viruses
like herpesviruses that boast large protein coding capacities. Importantly, the development of such methods would provide a
comprehensive portrait of viral replication and allow for the screening of small molecules and other cellular perturbations
with potential therapeutic and clinical applications. METHODS: By integrating mass spectrometry and molecular virology,
we have designed and experimentally validated a set of targeted proteomics assays for monitoring human viruses
representing the three Herpesviridae subfamilies—herpes simplex virus type 1, human cytomegalovirus, and Kaposi’s
sarcoma-associated herpesvirus. Our assays target 200+ unique peptides covering 50-80% of the predicted proteomes for
each of these viruses and span proteins that are representative of different temporal protein expression classes as well as
virion components. RESULTS: During wild-type virus infections, we first demonstrated that the breadth of proteins
monitored by our methods captures the temporal cascades of the replication cycles of these viruses. Additionally, we
illustrated that our assays can detect viral proteins at clinically relevant levels of infection. We next showed that these
assays can be used to quantify the effects of long-established and recently-discovered antiviral agents, and further captures
their precise temporal regulation of specific viral proteins. Finally, we demonstrated their broad utility for monitoring different
viral strains, including laboratory and clinical isolates. CONCLUSIONS: Altogether, our assays provide a reproducible
framework for monitoring the progression of herpesvirus infections and are broadly applicable across a variety of model
systems and contexts, including drug screening, detecting infections in clinical settings, and genetic manipulations of virus
or host factors.
P03.05 Beyond Concentration: Quantifying changes in protein fold stability of human serum proteins due to drug
binding using a novel iodination assay
Hsien-Jung L. Lin (1), Isabella James (1), Chad Hyer (1), Connor Haderlie (1), Michael Zackrison (1), Tyler Bateman (1),
Anisha Daley (1), Monica Berg (1), Stella Park (1), Nate Zuniga (1), John Price (1)
(1) Brigham Young University, Provo, Utah, United States
Currently, most biomarker discoveries focus on concentration differences. However, concentration is not sufficient to assess
the impact of the observed changes. There is a need to quantify the quality of the proteins in the proteome to assist
diseases diagnosis. Our lab has developed a novel assay that uses protein folding stability (PFS) as a metric to describe
protein quality in the proteome. Our assay measures the classical idea of denature midpoint (C1/2 ). A tissue homogenate is
aliquoted into a GdmCl gradient to denature each protein of the proteome in a controlled manner. Iodine is then added to
covalently label surface-exposed amino acids (Y, H, C, M, W). The modified peptides served as reporters of solvent
accessibility. The different amino acids we targeted compared to other PFS methods allow us to track the stability of a
protein across its sequence, which provides another perspective to understand protein structure and protein-protein
interaction. Here we present the application of the assay on human serum albumin (HSA) from undepleted human serum.
HSA is the most abundant protein in serum. It accounts for 60% of the serum proteome, which often presents a challenge
Page 77 of 166
for serum proteomics. In most of the applications, HSA, and other abundant proteins, are removed for meaningful
biomarkers analysis. However, studies have shown that the stability of HSA is altered in disease states. The results show
how the PFS at different regions of HSA, and the PFS of other serum proteins, respond to drug interventions. It also shines
a light on the potential of using HSA as a diagnosis tool. Further, our folding stability assay in combination with other
proteome scale metrics like concentration, turnover, or PTM can promote in vivo assessments of proteome regulation in
humans, and aid in biomarker discoveries.
Autoimmune diseases affect >20 million people in the US today. Currently, disease-specific autoantibodies are thought to
be the best biomarkers for diagnosis. Conventional immunoprecipitation methods have been used to identify autoantigens
from the most common autoimmune diseases. However, these diseases account for only 6.5 million of the 20 million
patients suffering from autoimmune diseases, leaving many without diagnoses until irreversible damage occurs. The
remaining 13.5 million patients have >70 autoimmune disorders without well characterized autoantibodies. The state-of-the-
art diagnostic test of these remaining diseases relies on gel electrophoresis of immunoprecipitated radiolabeled proteins,
which cannot be identified by MS due to safety issues and the overwhelming presence of immunoglobulins. We have
created an immunoprecipitation method that uses serum from patients with any autoimmune disorder to identify patient-
specific autoantigen proteins. This method uses a reversible click chemistry tag, called ProMTag. One end of the ProMTag
forms a reversible, covalent bond with protein by coupling to lysines and amino termini. The other end of the ProMTag can
form an irreversible, covalent bond with a solid bead support via a click chemistry, methyltetrazine-TCO, pairing. In this
study, the proteins of cell lysates that contain potential autoantigens were labeled with ProMTag. The ProMTagged-proteins
were exposed to patient antibodies bound to Protein A beads, thus capturing the ProMTagged autoantigens. All proteins
were released from the Protein A beads, including ProMTagged-autoantigens and untagged-antibodies. The ProMTagged-
autoantigens were subsequently coupled to TCO beads, and the untagged-antibodies were washed away. The linkage
between the ProMTag and autoantigens was then reversed, yielding autoantigen proteins with greatly reduced antibody
contamination ready for MS analysis. MS analysis successfully identified autoantigens from serums of patients with
rheumatoid arthritis. This autoimmune biomarker discovery method can accelerate sample testing for known autoantigens
and facilitating rapid discovery of novel autoantigens for both diagnostic and predictive biomarkers.
P03.07 Quantitative Proteomic Analysis of the Senescence-Associated Secretory Phenotype in Primary Human
Endothelial Cells to Identify Senescence Biomarkers
Francesco Neri (1), Pei-Hsun Wu (2), Akos Gerencser (1), Sandip Patel (3), Jacob Rose (1), Shuyuan Zheng (2), Jude
Phillip (2), Pierre-Yves Desprez (4), Denis Wirtz (2), Judith Campisi (5), Birgit Schilling (1)
(1) Buck Institute for Research on Aging, Novato, California, United States
Page 78 of 166
(2) Department of Chemical and Biomolecular Engineering, Johns Hopkins Physical Sciences Oncology Center, Johns
Hopkins Institute for Nanobiotechnology (INBT), Johns Hopkins University, Baltimore, Maryland, United States
(3) Buck Institute for Research on Aging, Novato, Novato, California, United States
(4) Buck Institute for Research on Aging, Novato, CA, USA, Novato, Maryland
(5) Buck Institute for Research on Aging, Novato, California
Age is the primary risk factor for major human pathologies, such as cardiovascular disorders, cancer, and
neurodegenerative diseases. Senescent cells and their senescence-associated secretory phenotype (SASP) drive aging
and promote such age-related pathologies. Identifying robust and easily-accessible biomarkers for the quantification of
senescence burden in humans is essential for translating therapeutic interventions that target senescent cells. Endothelial
cells line blood vessels and can secrete biomolecules directly into circulation. Thus, we hypothesized that studying the
SASP of senescent endothelial cells holds great promise for the identification of plasma biomarkers of senescence burden.
To do this, we optimized a cell culture model of primary human senescent endothelial cells by treating lung microvascular
endothelial cells – a cell type that comprises 50% of the capillary surface of our entire body – with the genotoxic
chemotherapeutic drug doxorubicin. Our protocol induces senescence robustly, as shown by high senescence-associated
β-galactosidase staining, halted proliferation, and a senescent gene signature. We are now optimizing a workflow to study
the SASP of senescent endothelial cells by data-independent acquisition (DIA) mass spectrometry. This approach will
enable us to study the senescent secretome in a comprehensive and unbiased fashion and identify potential plasma
biomarker candidates for human senescence burden. We are also investigating the senescent cell morphology in order to
eventually identify senescent cells in tissues based on their morphological traits. Thus, we started quantifying
morphological features of senescent endothelial cells by combining immunofluorescence staining of known senescence
markers and cytoskeleton dyes using an artificial intelligence-based algorithm that enables high content image
analysis. Our preliminary data shows that senescent cell populations show a strikingly altered morphology and are highly
heterogeneous compared to control populations. Thus, our translational approach utilizes multi-omics technologies for
potential senescence-derived biomarker identification.
P03.08 SureQuant targeted mass spectrometry standards and assay panel for quantitative analysis of
phosphorylated proteins from multiple signaling pathways
Bhavin Patel (1), Penny Jensen (2), Amirmansoor Hakimi (3), Sebastien Gallien (4), Aaron Gajadhar (3), Ana Martinez Del
Val (5), Jesper Olsen (5), Andreas Huhmer (6), Daniel Lopez-Ferrer (3), Ryan Bomgarden (7), Kay Opperman (1), John
Rogers (1)
(1) Thermo Fisher Scientific, Rockford, Illinois, United States
(2) Thermo Fisher Scientific, Rockford, United States
(3) Thermo Fisher Scientific, San Jose, California, United States
(4) Thermo Fisher Scientific, Paris, France
(5) University of Copenhagen, Copenhagen, Denmark
(6) Thermo Fisher Scientific, San Jose, California
(7) Thermo Fisher Scientific, Rockford, Illinois
Introduction There is broad interest in quantifying dynamic protein phosphorylation states in cellular signaling pathways
under different conditions. We have combined EasyPepTM technology, phosphopeptide enrichment, validated multipathway
Page 79 of 166
AQUATM heavy-labeled phosphopeptide standards, and SureQuantTM targeted mass spectrometry to quantitate changes in
phosphorylated protein abundance across multiple stimulated cell lines. Specific phosphopeptide standards were chosen
representing phosphosites from several different pathways. This novel workflow enables targeted quantitation of biologically
relevant phosphorylation sites with high accuracy, precision, and specificity. Methods Multiple cell lines were grown with
different stimulation conditions before in-solution digestion using EasyPep Maxi MS sample prep kit. One milligram of each
digest spiked with phosphopeptides standard was subjected to analysis using the Thermo ScientificTM PierceTM Fe-NTA
phosphopeptide enrichment kit. Discovery and targeted LC-MS/MS analysis were performed using Thermo Scientific Dionex
nanoLC™ system or Thermo Scientific Vanquish Neo UHPLC system coupled to Thermo Scientific™ Q Exactive™ HF
Hybrid Quadrupole-Orbitrap or Thermo Scientific™ Orbitrap Exploris™ 480 or Orbitrap Eclipse™ Tribrid™ Mass
Spectrometers. To ensure optimal measurement for targets, a novel SureQuant targeted acquisition method was performed
where real-time heavy peptides detection triggered high-sensitivity measurement of endogenous targets. Data analysis was
performed with Proteome Discoverer and Skyline software. Results Our optimized workflow includes EasyPep MS sample
prep kits and Fe-NTA enrichment with 131 AQUA heavy-isotope phosphopeptide standards to monitor multi-pathway
signaling. Two targeted MS methods (PRM and SureQuant) were compared to assess the relative performance for
quantitation of the desired endogenous peptides. More than 100 endogenous phosphopeptides from multiple stimulated
cancer cell lines and all 131 heavy phosphopeptides were quantitated with high sensitivity and reproducibility. SureQuant
method allowed quantitation of endogenous phosphopeptides at 10X lower levels than PRM. Conclusion SureQuant
multipathway phosphopeptide standard with novel SureQuant MS analysis allows reproducible, routine, and simultaneous
quantitation of functionally relevant phosphorylation sites.
P03.09 Compared dia-PASEF and prm-PASEF approaches for the absolute quantification of 500 human plasma
proteins in colon cancer plasma samples.
Gunnar Dittmar (1), Antoine Lesur (1), Francois Bernardin (1), Elisabeth Letellier (2), Gary Kruppa (3), Pierre-Olivier Schmit
(4)
(1) LIH-Luxembourg Institute of Health, Strassen, Luxembourg
(2) UNiversity of Luxembourg, Luxembourg, Luxembourg
(3) Bruker Daltonik, Bremen, Germany
(4) Bruker France S.A., Wissembourg, France
Introduction Targeted proteomics methods are a traditional choice for absolute protein quantitation of proteins in biofluids.
We recently evaluated the prm-PASEF acquisition strategy which, compared to traditional prm approaches, allows to further
increase the number of addressable targets and the method’s selectivity without compromising the sensitivity. In parallel,
the even higher multiplexing potential of dia-PASEF approaches (now regularly used in discovery studies) also triggered our
interest. We are now applying both approaches to the absolute quantitation of 500 blood proteins in colon cancer plasma
samples. Methods The plasma sample cohort consisted in 10 patients affected by a colon cancer (adeno carcinoma) and 10
controls. Plasma samples were depleted with a Mars 14 depletion column (Agilent), digested with a trypsin protease and
spiked the PQ500 (Biognosys) synthetic peptides mixture. Samples and controls were separated by nano-HPLC
(nanoElute, Bruker Daltonics) on a 25cm pulled emitter column (IonOpticks) using a 100 min gradient. Peptides were
analyzed on a timsTOF Pro2 instrument (Bruker Daltonics) operated in prm-PASEF and dia-PASEF modes. Data
processing has been done with Spectronaut (Biognosys), MaxQuant and Skyline-daily. Results We evaluated the
quantification performance of the prm-PASEF in depleted plasma samples by monitoring 1564 precursor-ions
Page 80 of 166
corresponding to 782 peptides from 565 proteins while using a 2 min retention time window. Some peptides could be
quantified down to 7 amol with prm-PASEF and 20 amol with dia-PASEF . The median relative standard deviation of the
signal of the peptides was of 3%. 98% of the 574 quantified peptide pairs could be quantified from the prm-PASEF
experiment, while 96% could be quantified using dia-PASEF. The results obtained with both approaches were highly
correlated. Conclusions Both approaches have been successfully applied to the analysis of colon cancer plasma samples
and allowed to spot regulated proteins that had formerly been spotted from cancer tissue analysis.
P03.10 Discovery of Senescence Biomarker Candidates from Monocyte Proteomes and Secretomes
DIMITRIOS TSITSIPATIS (1), ANJANA RAM (1), MARTINA ROSSI (1), MYRIAM GOROSPE (1), NATHAN BASISTY (1)
(1) National Institute on Aging, Baltimore, Maryland, United States
Introduction: Cellular senescence is a complex stress response characterized by permanent cell-cycle arrest and the
activation of a pro-inflammatory senescence-associated secretory phenotype (SASP). Accumulation of senescent cells and
SASP are drivers of aging and many pathological processes including chronic inflammation and cancer. Biomarkers of
senescent cell burden may be clinically useful for treatment of age-related pathologies in humans. Here we perform
comprehensive and quantitative proteomic profiling of intracellular and secreted proteomes of senescent monocytes to
identify biomarker candidates of monocyte senescence burden in human peripheral blood. Methods: To model the full
range of physiological oxygen exposure experienced by monocytes in circulation, we induced senescence using ionizing
radiation in two monocyte cell lines, U937 and THP-1, under hypoxic or hyperoxic oxygen conditions. After 6 days,
proliferating and senescent cells were placed in serum- and phenol-red-free RPMI media for a 24-hour period. Secreted
media and cell pellets were collected for LC-MS/MS analysis and assessment of a panel of canonical senescence and
viability markers. Analysis of all samples was performed on the Q-Exactive HF Orbitrap mass spectrometer using Data-
Independent Analysis and analyzed in Spectronaut. Results: We established a model of senescence in two independent
monocyte cell lines that is characterized by senescence markers of reduced proliferation (assessed by EdU incorporation),
increased expression of p21, p16, DPP4, and IL6, and increased senescence-associated β-galactosidase activity. We
report the first comprehensive and unbiased assessment of the senescent monocyte intracellular proteome and
secretomes, with many unique features no shared with other senescent cell types. We compared the proteomes of
senescent monocytes with existing biomarkers of aging in humans to identify promising peripheral biomarkers of
senescence in humans. Conclusion: These results may provide clinically useful information for establishing therapeutic
targets or biomarkers to aid in the translation of senescence-targeted therapies to treat chronic inflammation and age-
related decline.
P03.11 Monitoring sirtuin dynamics during viral infection using a targeted mass spectrometry assay
Matthew D. Tyl (1), Yana D. Miteva (1), Ileana M. Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
Sirtuins are NAD+-dependent enzymes that function as “sensors” of the cellular environment by transmitting information
through the addition or removal of posttranslational modifications (PTMs). Although initially designated as Class III histone
Page 81 of 166
deacetylases, recent work has demonstrated that sirtuins have a range of enzymatic activities and substrates through which
they maintain cellular homeostasis. Our lab has previously uncovered that sirtuins exert both proviral and antiviral roles
during infection, but the underlying mechanisms remain unknown. Sirtuins may have compensatory effect, with one sirtuin
modulating the levels and functions of other sirtuins. Here, we developed a targeted mass spectrometry assay based on
parallel reaction monitoring that allows simultaneous detection and quantification of all human sirtuins. We demonstrate the
sensitivity of this assay for detecting these low abundance proteins and its generalizability to different cell types. We profiled
changes after treatment with the antiviral compound, trans-resveratrol, which activates sirtuins. Further, we monitored sirtuin
dynamics during cellular remodeling events: differentiation of monocyte-like cells into macrophages, as well as during
infection with two herpesviruses, herpes simplex virus 1 (HSV1) and human cytomegalovirus (HCMV). Finally, we used this
assay in our investigation of the nuclear sirtuin 6 (SIRT6) during infection. Using SIRT6 knockdown, CRISPR-mediated
knockout, and overexpression, we assessed the effect of SIRT6 on HCMV replication in primary human fibroblasts (MRC5).
We further investigated whether perturbing SIRT6 levels drives compensatory effects by other sirtuins. Lastly, our
quantitative sirtuin measurements were integrated with investigations of SIRT6 protein interactions and localization during
the progression of HCMV infection.
Page 82 of 166
P04: Cancer Proteomics
P04.01 Proteomic Analysis Reveals that Musashi-2 Promotes Signaling Pathways Associated with Cell Proliferation
and Metastasis in Neuroblastoma Cells
Pritha Bagchi (1), Adeiye Pilgrim (1), Selma Cuya-Smith (2), Dongdong Chen (1), Nicholas Seyfried (1), Robert Schnepp (1)
(1) Emory University, Atlanta, Georgia, United States
(2) Kennesaw State University, Kennesaw, Georgia, United States
Neuroblastoma (NBL) is the most common extracranial solid tumor of childhood. Many patients present with widespread
metastasis and may have secondary malignancies and other toxicities in adulthood even after extensive multimodal
treatment. Thus, there is a dire need for druggable targets, which minimize toxicity and maximize efficacy in the treatment of
NBL. RNA-binding protein Musashi-2 (MSI2) has been shown to be a mediator of numerous cancer-related hallmarks, but
its role in NBL remains unexplored. Based on our preliminary data from cell-based assays, MSI2 appears to be an
oncogenic driver in NBL as well. Hence, we explored the MSI2-influenced proteome to understand the mechanism and to
identify potential NBL-specific therapeutic targets. We genetically reduced MSI2 expression in two NBL cell lines, SK-N-
BE(2)C and Kelly, using short hairpin RNA (shRNA) clones with two unique targets within the MSI2 coding sequence.
Subsequently, we performed unbiased proteomics analysis using LC-MS/MS. The label-free quantification algorithm in
MaxQuant was employed for protein identification and quantification; statistical analysis was performed by Perseus and
DAVID. The proteomics data verify successful knockdown of MSI2 protein in both cell lines. Gene Ontology (GO) analysis
reveals that MSI2 knockdown affected different pathways in a cell-line specific manner. In SK-N-BE(2)C cells, processes
related to cell cycle progression and cell growth are downregulated (e.g., DNA replication and telomere maintenance via
recombination), whereas mechanisms that induce apoptosis are upregulated (e.g., glycosphingolipid metabolic process and
lysosome). In Kelly cells, downregulated proteins show overrepresentation for “nervous system development” and
“postsynaptic density”, indicating attenuation of metastasis, while upregulated proteins are related to “mitochondrial matrix”
and “metabolic pathways”. In summary, NBL cells with genetically reduced MSI2 demonstrate cell cycle arrest, increased
apoptosis, and altered metabolism. Thus, MSI2 and its downstream proteins could potentially be targeted to develop cancer
therapeutics, especially to inhibit tumor growth and metastasis in neuroblastoma.
P04.02 Robust, Reliable, Reproducible Sample Preparation Enabling Confident LC-MS Analysis of Proteins in FFPE
Tissues
DEBADEEP BHATTACHARYYA (1), Eugenio Daviso (1)
(1) Covaris, Woburn, Massachusetts, United States
Formalin-fixation and paraffin-embedding (FFPE) of tissue samples are routinely used in clinical histopathology for
preserving tissue architecture. However, paraffin removal is needed for high quality LC-MS data. Typically Xylol is used
but has health risks and compromises reproducibility. Better sample preparation removes unwanted matrix components
while ensuring reliability, reproducibility and robustness. Covaris' Adaptive Focused Acoustics® (AFA®) Technology can
enable a modified protein aggregation capture (PAC) protocol with remarkable robustness, reliability, and
reproducibility. FFPE tissue scrolls can be incubated in a Covaris plate post-treatment with “Covaris Tissue Lysis Buffer”.
The plates were processed with the Covaris LE220-Plus Focused-ultrasonicator for deparaffinization and DNA
Page 83 of 166
shearing. PAC protocol enabled extraction from FFPE lysate and Sera Mag Speed beads induced on-bead protein
aggregation followed by digestion Trypsin/LysC before conducting LC-MS analysis. This work demonstrates a non-toxic,
scalable, and robust method for LC-MS based analysis of FFPE scrolls with human adenoma samples that were processed
in AFA compatible 96-well plate. Heating and AFA based sonication ensured reversal of crosslinks and paraffin removal
from tissue samples. Protein purification was achieved using the PAC protocol and a tryptic ‘on-bead’ digestion resulted in a
clean peptide solution for LC-MS analysis. A comparison of the xylol-free workflow with that of the common xylol
deparaffinization resulted in highly similar peptide, protein identification rates, and protein abundance ranges. On average,
3,900 proteins and 20,842 peptides per single run measurement were quantified. A comparison of this method with the
xylol-based protocol showed that crosslink reversal was equally efficient in both protocols. The developed protocol not only
offers ability to characterize and quantify a large number of proteins from complex tissue matrices, but also can support low-
medium-high throughput capabilities as needed by the organization.
P04.03 Deep Plasma Proteomics at Scale Enabling Proteogenomic Analyses in a Lung Cancer (NSCLC) Study
Harendra Guturu (1), Mahdi Zamanighomi (1), Jian Wang (1), Amir Alavi (1), Tristan Brown (1), Daniel Hornburg (1), Moaraj
Hasan (1), Shadi Ferdosi (1), Khatereh Motamedchaboki (1), Margaret Donovan (1), Theodore Platt (1), Ryan Benz (1),
Asim Siddiqui (1), Serafim Batzoglou (1)
(1) Seer, Inc., Redwood City, California, United States
Our ~20,000 genes encode over one million protein variants, given alternative splice forms, allelic variation, and protein
modification. Though large-scale genomics studies have expanded our understanding of biology, similarly, scaled deep and
untargeted proteomics studies of biofluids have remained impractical due to complex workflows. To address this need, we
have previously described Proteograph(TM), a novel platform that leverages the protein-corona interactions of nanoparticles
for deep and untargeted proteomic sampling at scale. Using Proteograph(TM) in a non-small cell lung cancer (NSCLC) cohort,
we previously conducted a deep interrogation of plasma from 141 subjects: 61 early-stage NSCLC subjects and 80 non-
cancer controls. We identified 2,499 plasma proteins, with 1,992 present in ≥ 25% of the samples. Leveraging this data, we
created a biomarker classifier distinguishing NSCLC from controls with area under the receiver operating characteristic
curve of 0.91 [1]. In this study, we now re-analyze the data with the more sensitive DIA-NN software to enhance protein
depth while preserving the accuracy of the classifier. In addition, to show the added value of our platform in combination
with genomic data, we integrate previously sequenced exome data with our proteomic data to build a multi-modal
proteogenomic deep learning classifier. Our results outline proteogenomic workflows for robust biomarker discovery and
cohort subtyping. The Proteograph(TM) platform interrogates the plasma proteome at previously impractical combinations of
scale, depth and coverage, and enables the development of improved classification models and the study of
proteogenomics. [1]Blume et al. Nat. Comm. (2020)
P04.04 Deep, Rapid and Unbiased Plasma Proteomics with Differential Analysis of Proteoforms Enabling
Proteogenomic Studies in a NSCLC Lung Cancer Study
Yingxiang Huang (1), Margaret Donovan (1), John Blume (1), Jian Wang (1), Marwin Ko (1), Ryan Benz (1), Theodore Platt
Page 84 of 166
(1), Juan Cuevas (1), Serafim Batzoglou (1), Asim Siddiqui (1), Omid Farokhzad (1)
(1) Seer Inc., Redwood City, California, United States
Comprehensive assessment of the human proteome remains elusive due to multiple forms of a protein, each of which can
serve distinct functions, arising from alternative splicing, allelic variation, and protein modifications. Characterization of the
variable protein forms, or proteoforms, will expand our understanding of the molecular mechanisms
underlying disease, however identification of these variable forms requires unbiased protein coverage at sufficient
scale. Scalable, deep, and unbiased proteomics studies have been impractical due to cumbersome and lengthy
workflows required for complex samples, like blood plasma. Here, we demonstrate the power of Proteograph in a proof-of-
concept proteogenomic analysis of 80 healthy controls and 61 early-stage non-small-cell lung cancer (NSCLC) samples to
dissect differences between protein isoforms arising from alternative gene splicing, as well as the identification of novel
peptides arising from allelic variation. Processing 141 plasma samples
with Proteograph yielded 22,993 peptides corresponding to 2,569 protein groups. Using peptides
with significant abundance differences (p < 0.05; Benjamini-Hochberg corrected), we extracted proteins comprised of
peptides where at least one peptide had significantly higher plasma abundance, and another significantly lower plasma
abundance in controls vs. cancer, resulting in a set of four proteins. For three of these proteins (BMP1, C4A, C1R), the
abundance variation is possibly explained by underlying protein isoforms. To identify protein variants, we performed exome
sequencing on 29 individuals from the NSCLC study, created personalized mass spectrometry search libraries for each
individual, and identified 464 protein variants. In conclusions, Proteograph can generate unbiased and deep plasma
proteome profiles that enable identification of protein variants and peptides present in plasma, at a scale sufficient to enable
population-scale proteomic studies.
P04.06 Low-Input Proteomic Phenotyping of a Metabolically Defined Putative Mammary Epithelial Stem Cell
Population
Matthew Waas (1), Pirashaanthy Tharmapalan (1), Rama Khokha (1), Thomas Kislinger (1)
(1) Princess Margaret Cancer Centre, Toronto, Ontario, Canada
Breast cancer, a heterogenous disease of multiple origins, is the most common cancer in women. Studies of the various
subtypes of breast cancer have revealed similarities between the molecular profile of these subtypes to distinct populations
of mammary gland epithelial cells. This evidence supports the hypothesis that the different epithelial cell populations serve
as the cell-of-origin for the corresponding subtypes of cancer. Our hypothesis is that defining functional and molecular
identities of distinct stem cell populations within the mammary gland epithelium will reveal potential markers and molecular
vulnerabilities. Recently, we identified a metabolically defined subpopulation of mammary epithelial basal cells which harbor
an increased capacity for colony formation representing a novel putative progenitor/stem population. We have adapted an
innovative droplet-digestion sample preparation strategy relying on commercially available consumables and, in
combination with systematic optimization of sample preparation, instrument acquisition, and data analysis methods, we
reproducibly obtain >3,000 proteins on a Thermo Fusion mass spectrometer from 500 sorted cells. We have performed a
repeatability study to characterize the observed variability inherent to sample preparation and data acquisition. The median
correlation between cell type replicates was 0.96 and 0.91 within and between days, respectively. Finally, we have applied
this platform to profile the proteomic phenotype of the metabolically defined putative epithelial stem/progenitor cells from six
individual mice (~2000 cells/mouse). We identify over 3500 proteins per sample with high repeatability. Comparisons to
Page 85 of 166
control populations reveal 9 potential cell surface markers and 5 potential druggable targets. Future studies will compare the
proteomic phenotype to stem cell populations sorted with other established markers and will investigate the functional
capacity of this population. We anticipate the results of this research will harmonize our understanding of mammary stem
cell biology and support breast cancer prevention, prognosis, and treatment efforts.
P04.07 Characterizing the Proteomic Response of Models of High-Grade Serous Tubo-Ovarian Cancer to Taxane
Treatment
John Muroski (1), Oscar Pundel (2), Krystal Lum (1), Benjamin Neel (2), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
(2) New York University, New York, New York, United States
Ovarian cancer is the deadliest form of gynecological cancer, with high-grade serous tubo-ovarian cancer (HGSC)
presenting itself as the most common form. Advancements in treating HGSC have shown less progress in overall survival,
due in part to the high heterogeneity of cancer cells. Recent advances in organoid generation have allowed the
development of genetic models that represent different subclasses of HGSCs. Major subtypes include a homologous-
recombination (HR)-deficient model (Trp53-/-;Brca1-/-;MycOE) and a model of a poorly characterized subtype (Trp53-/-;Pten-/-;
Nf1-/-). Previous work to elucidate features of this model unexpectedly revealed a significantly greater susceptibility to
treatment with a commonly used anti-cancer drug, Paclitaxel, than compared to other subclass models. Here, we used
proteomics to analyze 2D and 3D cell culture model systems of HGSC in order to better understand drug hypersensitivity
and cell autonomous/non-autonomous effects on HGSC evolution. Alterations in the proteomes of cell and organoid
cultures, with and without Paclitaxel treatment, were analyzed using quantitative mass-spectrometry. Over 700 proteins
significantly altered in abundance and major alterations in the HR-deficient model indicate an increase in proteins related to
cytoskeletal organization, extracellular components, and immune signaling. Protein network analysis highlights enriched
processes such as mitosis and immune signaling. In contrast, proteins enriched in the Trp53-/-;Pten-/-;Nf1-/- model are
involved in oxidation reduction and lipid metabolism. These findings indicate that alterations in cytoskeletal compositions do
correlate with differences in Paclitaxel sensitivity. To further define the system-wide changes occurring in drug-treated
HGSC models, thermal proximity coaggregation analysis (TPCA) is used to investigate activated or inhibited cellular
pathways in these model systems.
Accurate quantification of HLA class I gene expression is important to understand the interplay of tumor microenvironment
of antitumor cytotoxic T cell activity. Because HLA-I sequences are highly variable, common RNAseq and mass
spectrometry based quantification workflows using standard genome and protein sequence references do not provide HLA-I
allele specific quantifications. This study improved the quantification processes by using personalized HLA-I nucleotide and
protein reference sequences based on the subjects’ HLA-I genotypes, and surveyed tumor and adjacent normal samples
Page 86 of 166
from patients from nine cancer types. DDA-MS data was validated to be sufficient to estimate HLA-A protein expression at
allele level. B15 allele group was over-represented in LUAD patient with allele frequency of 76% more than the expected
value. HLA-I proteins were observed to be at significantly higher levels in tumors compared to adjacent normal tissues by
41% - 63% of head and neck, endometrial, and kidney tumors. These higher protein levels were mostly driven by increased
level of HLA-I transcripts from both alleles. Most immune cell types are universally enriched in HLA-I high tumors while
endothelial and neuron cells showed divergent relationships with HLA-I. Pathway analysis revealed that tumor senescence
and autophagy activity was likely to influence the level of HLA-I proteins in glioblastoma. Most proteins correlated to HLA-I
protein levels are directly involved in HLA-I function in immune response and cell death, while co-expression of
glycosylation genes with HLA-I are exclusively observed at protein level.
Page 87 of 166
P05: Cardiovascular Proteomics
P05.01 CellSurfer platform reveals the first experimentally derived surfaceome map of primary human cardiac cells
Linda Berg Luecke (1), Matthew Waas (2), Jack Littrell (3), Rebekah Gundry (3)
(1) Medical College of Wisconsin, Milwaukee, Wisconsin, United States
(2) Princess Margaret Cancer Centre, Toronto, Canada
(3) University of Nebraska Medical Center, Omaha, Nebraska, United States
In the heart, cell surface glycoproteins in cardiomyocytes and cardiac fibroblasts are essential for sustaining normal organ
function and play critical roles in cardiac development, disease, and drug uptake. However, the lack of a detailed cell type-
or chamber-resolved view of the cell surface proteome of the adult human heart currently limits discovery of new targets for
precision drug delivery and the development of practical approaches for studying how different cell types contribute to the
development of cardiac disease. CellSurfer, a new analytical platform, was applied to cardiac cells isolated from human
hearts. Briefly, cell surface N-glycoproteins on ~1 million cells were labeled, digested, selectively enriched using streptavidin
magnetic beads, cleaned using SP2, and analyzed by MS. Sample preparation was automated using liquid handling
robotics. MS data were analyzed using Proteome Discoverer, Spectronaut, MSstats, and R. Results were curated and
annotated using Veneer. Integrating CellSurfer with an optimized strategy for isolating intact primary cardiomyocytes and
fibroblasts from human donor heart tissue resulted in the generation of the first chamber-, cell-type-, and patient-specific
map of the cell surface N-glycoproteome in the adult heart. Overall, >1100 cell surface N-glycoproteins were detected,
including 48 and 21 putative selective markers for cardiomyocytes and cardiac fibroblast, respectively. Novel monoclonal
antibodies generated for one cardiomyocyte marker uniquely localize to cardiomyocytes within human heart tissue sections
and stem cell derivatives, suggesting its value for cell-type specific targeting and immunophenotyping. Comparisons of
explanted cardiac fibroblasts within the first three passages reveals previously undescribed remodeling of the surfaceome,
justifying caution when using cultured cells. These data represent the first major step towards a comprehensive, donor, cell-
type, subtype, and chamber-resolved reference map of cell surface phenotypes in the adult human heart and reveal new
targets for immunophenotyping, drug delivery, and benchmarking explanted cells and stem cell derivatives.
Purpose: Matrix metalloproteinases (MMPs) are an endogenous family of enzymes that contribute to matrix remodeling in
several diseases, including cardiomyopathies. After cardiac ischemia, MMP induction and activation may generate a
predominantly proteolytic environment, leading to degradation of extracellular matrix (ECM). Matricryptins are biologically
active peptides, generated from ECM proteolysis, able to regulate cell function and survival. We recently identified a
collagen-derived matricryptin (p1159), gradually forms post-myocardial infarction (MI) and is a substrate of MMP-9. Mice
treated with p1159 post-MI show both improved function and reduced dilation of the left ventricle (LV). To evaluate whether
p1159 presence modulates MMP-9 substrate preference in the myocardium, we used N-terminal amine isotopic labeling of
substrates (TAILS) proteomics to compare and quantify the LV peptidome. Methods: iTRAQ-TAILS multiplex quantitative
proteomics was applied to naïve LV tissues (male, C57Bl/6 mice, 6-8 months old, n=6/group) to assess effects of p1159 in
Page 88 of 166
MMP-9 substrate preference. One mg of protein was used for each sample ± p1159 ± MMP-9; samples were processed
according to the previously published protocol (1). The samples were run on a Thermo Scientific Q-Exactive Orbitrap, then
analyzed using Proteome Discoverer 2.5™. Results: We identified 837 MMP-9 substrates in naïve mouse LV. Of these, 49
were ECM molecules and 47 were plasma membranes. For the first time, cofilin-1 and nischarin were identified as potential
MMP-9 cardiac substrates. The presence of p1159 changed the relative abundance of MMP-9 cleavage products.
Conclusion: This study both identified novel MMP-9 cardiac substates and determined p1159 presence changes MMP-9
substrate preference. Future studies will identify which substrates may be responsible for the peptide beneficial effects on
LV remodeling. Literature cited: (1) Kleifeld, Oded et al. “Identifying and quantifying proteolytic events and the natural N
terminome by terminal amine isotopic labeling of substrates.” Nature protocols vol. 6,10 1578-611. 22 Sep. 2011,
doi:10.1038/nprot.2011.382
P05.03 Proteomics to understand cholesterol efflux capacity of High density Lipoprotein (HDL) particles from
human plasma
Zeeshan Hamid (1), Anamika Gangwar (2), Kip Zimmerman (1), Sneha Deodhar (2), Kit Wallis (1), Anand Rohatgi (2),
Michael Olivier (1)
(1) Wake Forest University School of Medicine, Winston Salem, North Carolina, United States
(2) University of Texas Southwestern Medical Center, Dallas, Texas, United States
Cardiovascular disease (CVD) remains the leading cause of death worldwide and atherosclerosis has been shown to be a
major underlying factor. HDL is known to exert various anti-atherosclerotic functions with its strong ability to efflux
cholesterol from periphery and transport it to the liver for excretion. Cholesterol efflux capacity (CEC) of HDL has been
extensively studied in various animal models, but has not been studied in detail in humans. Here, we identified HDL plasma
proteins with high throughput mass spectrometry and associated protein abundance with measured CEC. A bottom up
proteomics workflow was followed involving protein precipitation from HDL particles, followed by reduction, alkylation and
trypsin digestion. Resulting peptides were cleaned up using C18 desalting tips, and data was acquired on a Q-Exactive
mass spectrometer with an easyspray nano-LC. Raw data analysis was done with MaxQuant using H. Sapiens uniprot
protein database. We obtained quantitative data for 540 proteins across all samples. A series of linear mixed models were
computed on the log-transformed and normalized abundance of each distinct protein. Protein abundance was the outcome
in our model and CEC value was the predictor. The model included age as a covariate as well as the individual identifier as
a random effect to account for repeated measures. A Benjamini-Hochberg FDR was computed on the raw p-values to adjust
for multiple comparisons, and proteins meeting an FDR < 0.05 were retained for further downstream analysis. Of 540
quantified proteins, 47 proteins were either positively or negatively associated with the CEC values. Using DAVID
enrichment analysis of these 47 proteins, we observed strong enrichment of proteins involved in lipoprotein metabolic
process, high density lipoprotein particle remodeling and reverse cholesterol transport (APOA2, APOC2, APOC3, APOE
and LCAT). Overall, this study has highlighted some significant associations of some proteins involved in CEC of HDL
particles.
Page 89 of 166
P05.04 Cardiac proteomics reveals sex chromosome-dependent differences between males and females that arise
prior to gonad formation
Josiah Hutton (1), Wei Shi (2), Xinlei Sheng (1), Kerry Dorr (2), James Emerson (2), Haley Davies (2), Tia Andrade (2),
Lauren Wasson (3), Todd Greco (1), Yutaka Hashimoto (4), Joel Federspiel (1), Zachary Robbe (2), Xuqi Chen (5), Arthur
Arnold (5), Frank Conlon (2), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
(2) UNC-Chapel Hill, Chapel Hill, North Carolina, United States
(3) UNC-Chapel Hill, Chapel Hill, North Carolina
(4) Princeton University, Princeton, New Jersey
(5) University of California, Los Angeles, Los Angeles, California, United States
Sex disparities in cardiac homeostasis and heart disease are well documented, with differences attributed to actions of sex
hormones. However, studies have indicated sex chromosomes act outside of the gonads to function without mediation by
gonadal hormones. Here, we performed transcriptional and proteomics profiling to define the differences between male and
female mouse hearts. We first applied this workflow to the C57BL/6J mouse strain, where we observed that there were
fewer protein expression differences between male and female hearts compared to the number of differentially expressed
mRNA transcripts. We next performed quantitative proteomic analysis of eight different mouse strains and determined that
proteins involved in metabolic processes were significantly enriched in male mouse hearts, while proteins involved in
nuclear membrane structure were significantly enriched in female mouse hearts. We then utilized the Four Core Genotype
(FCG) mouse strain to demonstrate, contrary to current dogma, cardiac sex disparities are controlled not only by sex
hormones but also through a sex-chromosome mechanism. Using Turner syndrome (XO) and Klinefelter (XXY) mouse
models, we determined that the sex-chromosome pathway is established by X-linked gene dosage. Additionally, we
demonstrated that cardiac sex disparities occur at the earliest stages of heart formation, a period before gonad formation.
Throughout these analyses, alpha-1B-glycoprotein (A1BG) was consistently observed to be significantly increased in female
hearts compared to male hearts. We utilized CRISPR/Cas9 to generate cardiac conditional loss-of-function alleles of A1BG,
and we demonstrated that that the loss of A1BG in the heart leads to cardiac defects in females, where we specifically
observed defects in the outer left ventricle wall function and thickness, but these cardiac defects were not observed in
A1BG-KO males. These results provide a powerful resource for studying sex based cardiac disease states, and we
demonstrated the critical role of increased A1BG expression in female heart function.
P05.05 Integrating a high-throughput proteomic workflow for in vitro screening of cell-based assays
Saeed Seyedmohammad (1), Alejandro Rivas (1), Simion Kreimer (1), Qin Fu (1), Jennifer Van Eyk (1)
(1) Cedars Sinai Medical Institute, Los Angeles, California, United States
Introduction: We have developed a robust and reproducible high-throughput (HTP) workflow for large-scale protein assay.
This bottom-up proteomics workflow directs automated processing of cells grown in 96-well plates for LC-MS analysis. The
HTP system leverages the COVARIS LE220 Plus sonication system for cell lysis and solubilization in tandem with the
Beckman i7 liquid handling workstation for sample preparation. Methods: AC16 human cardiomyocytes were grown to
different cell densities in 96-well plates. Sonication parameters were optimized for maximal performance, and suitability of
different solubilization buffers were determined on COVARIS 96 AFA-Tube TPX 150 Plates. The i7 automation workstation
were employed in conjunction with S-trap digestion for protein denaturation, Cys reduction, alkylation, and trypsin digestion
Page 90 of 166
(1:20 µg ratio of trypsin/protein) in a humidity regulated on-deck incubator (2 hours, 47°C). Data analysis involved BCA
measurements and LC-MS acquisition on Thermo Orbitrap Lumos Micro system coupled to capLC column at 60-minute
gradient. Results: Average protein concentration of 1 µg/µl (n=4, 1.7 STD) were obtained from approximately 100,000 cells
sonicated in each well of a 96-well COVARIS plate. Optimal cell lysis was observed at the representative AFA settings
condition (200 PIP, 50 DF and 1,000 CPB) in 5 % SDS using the V-shaped AFA-Tube TPX 150 plate (PN-520291,
COVARIS). A specifically designed cell-based prototype plate (PN-520310, COVARIS) was also optimized to achieve DNA
sheering and full cell lysis (half a fold increase with higher reproducibility and less miss-cleavages), by incorporating an
initial high power-burst cycle prior to standard AFA sonication. Conclusions: By employing the COVARIS sonication
system, simultaneous cell lysis of multiple samples was achieved. The developed HTP pipeline can be applied to both label
free and stable isotope-labeling as well as investigating important key regulators in cell-based assays, including post-
translational modifications. Furthermore, drug targets and cellular perturbations could be rapidly investigated and studied.
Page 91 of 166
P06: Cellular Signaling and Systems Biology
P06.01 Heterogeneity of PTEN and PTEN-alpha expression in prostate cancer
Alakesh Bera (1), Meera Srivastava (2), Madhan Subramanian (1), Eric Russ (1), Surya Radhakrishnan (1), Adam Landa
(1), Nahbuma Gana (1), Lukas Bubendorf (2)
(1) Uniformed Services University of the Health Sciences, Bethesda, Maryland, United States
(2) University Hospital Basel, Basel, Swaziland
Prostate cancer (PrCa) is among the most common cancers and remains the second leading cause of cancer-associated
death in men worldwide. Several genetic alterations are known to be responsible for the progression, aggressiveness,
chemo-resistance, and heterogeneity of prostate cancer. For instance, phosphatase and tensin homolog (PTEN) is an
essential tumor suppressor gene which regulates the PI3K-AKT pathway and whose mutation and deletion are among the
most common genomic aberrations in prostate cancer. Moreover, recent studies have indicated that the longer isoform,
termed PTEN-Long (PTEN-Ln) or PTEN-alpha, also plays a critical role in normal cellular functions such as proliferation and
apoptotic pathways. This is significant as our findings from tissue micro-array and cellular assays, which compared the
protein expression of the standard PTEN (now termed PTEN-st) and PTEN-Ln, have indicated a robust correlation between
loss of PTEN expression and the balance between PTEN-st and PTEN-Ln protein expression; showing adverse
pathological features with aggressive Gleason grades of prostate cancers. In other words, in patients with prostate cancer,
the ratio of PTEN-st to PTEN-Ln expression drastically decreases. We also found a novel Alu element insertion in the PTEN
gene of a prostate cancer cell line, similar to the reported Alu element insertions at a PTEN hotspot in Cowden syndrome. In
this current study, we have determined the molecular basis of PTEN-standard and PTEN-Long expression variability and
their genetic aberrations, including the Alu insertion mechanism, which leads to the loss of functional PTEN activity.
Furthermore, efforts have also applied to determine the signaling cascade which is involved in regulating this genetic
aberration and how to inhibit that signaling pathway. Finally, our current findings could help in the development of both
clinical diagnostic capabilities and in predicting patient response to targeted therapy in prostate cancer.
The National Institutes of Health Human BioMolecular Atlas Program was established to drive the large-scale mapping of
healthy human tissue from the level of single cells to whole organs. As part of this effort, we assessed an initial set of tumor
resections comprising histologically normal tissue from 35 patients. The body mass index (BMI) of these patients ranged
from normal (20-25) to severe obesity (>40), and additional conditions, such as hypertension and diabetes, existed within
the cohort. While it is well established that obesity can negatively impact kidney health, the molecular mechanisms
responsible have yet to be comprehensively elucidated. Additionally, the effect of confounding comorbidities can further
Page 92 of 166
obscure these pathomechanisms. We applied bulk proteomics and systems biology workflows to interrogate molecular
changes associated with BMI, comorbidities, and sample composition (i.e., percent cortex versus medulla). The initial data
reveal major trends in changes to oxidative phosphorylation and immune response associated with BMI. Oxidative stress
can trigger inflammation, leading to a negative feedback loop and furthering tissue injury. These data will be interrogated
further to understand molecular changes with increasing levels obesity and other comorbidities. Furthermore, we will
implement spatial proteomics and imaging mass spectrometry to resolve a spatially specific understanding of the molecular
effects of these conditions on kidney tissue. Spatial omics will enable the association of specific molecular mechanisms with
tissue functional regions, such as glomeruli. These approaches will be applied to additional patient samples in order to map
the proteomic landscape of healthy kidney in relation to BMI, comorbidities, and tissue functional regions.
P06.03 Proteomic characterization of the virus microenvironment and proximal immune signaling
Bokai Song (1), Xinlei Sheng (1), Peter Metzger (1), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
Efficient virus dissemination relies on the release of infectious viral particles from the infected cells and attachment to the
neighboring cells. Viruses have acquired mechanisms to alter the proteome of infected cells to promote infection
progression. Upon infection, the infected cells secrete various chemokines and cytokines to signal the immune system to
prevent the spread of infection, while viral proteins are also secreted to dampen the host response. However, it remains
largely unknown how cells either in close or distant proximity to the infected cell respond to the infection. No study has yet
attempted to investigate cellular proteomes of the virus microenvironment. Here, we integrated fluorescence-activated cell
sorting (FACS), quantitative mass spectrometry, microscopy, and molecular virology assays to provide the first view of
cellular proteome alterations within the virus microenvironment. We generated fibroblasts secreting a lipid-permeable
mCherry tag, which we infected with GFP-tagged human cytomegalovirus (HCMV). Using a strategy that involved the co-
culture of these cells with wild type cells, we distinguished and enriched infected, neighboring, and distal cells. By defining
the differential proteomes of these cell populations, we discovered that multiple interferon-inducible proteins are down-
regulated in infected and neighboring cells compared to the distal cells, suggesting a viral strategy to promote the spread of
infection. We corroborated these results by confocal microscopy analyses. Next, we performed superinfection assays with
both HCMV and herpes simplex virus type 1 (HSV-1), observing increased susceptibility of neighboring cells to
superinfection compared to distal cells. Furthermore, neighboring cells also display alterations in cell cycle progression
when compared to distal cells. Overall, our study is the first proteomic characterization of the virus microenvironment,
revealing global alterations in immunity-related pathways. These findings provide insights into how virus infection reshapes
the surrounding environment to facilitate progression of infection.
P06.04 Defining Toll-like Receptor Signaling Networks in Cellular Space and Time Using Quantitative Mass
Spectrometry
Trisha Tucholski (1), Aleksandra Nita-Lazar (1)
(1) Functional Cellular Networks Section, Laboratory of Immune System Biology, National Institute of Allergy and Infectious
Diseases, Bethesda, Maryland, United States
Page 93 of 166
Pattern recognition receptor (PRR) signaling initiates host innate immune response upon the recognition of pathogen-
associated molecular patterns (PAMPs). Toll-like receptors (TLRs) are an important class of evolutionarily-conserved PRRs
localized to the surface and endosomes of professional innate immune cells. Protein expression data alone may not
sufficiently describe regulation of early innate immune response. Quantitative experimental data detailing the subcellular
location and movement of signaling network components in the cell as innate immune response progresses are important
for understanding how hosts initially respond to infection. Herein, we use subcellular fractionation combined with
quantitative mass spectrometry (MS)-based proteomics to define the spatial proteome of human monocyte-derived
macrophages, an important professional innate immune cell type. Preliminarily, we have demonstrated reproducible
enrichment of organelle protein markers using the spatial proteomics workflow. We have also identified important proteins
involved in TLR signaling, including components of the myddosome (e.g., MyD88, IRAK1, IRAK4). We show that the
proteins TAB1, TAB2, TAK1, which associate during TLR signaling, are co-enriched in subcellular fractions. Furthermore,
we aim to characterize early TLR signaling by incorporating spatial proteomic analysis of macrophages following time-
course stimulation of TLR4, and important sensor of lipopolysaccharides derived from Gram negative pathogens. We will
monitor protein subcellular location at early timepoints post-stimulation (15, 30, 45, and 60 min) to capture early protein
translocation events. Global spatiotemporal proteomic profiles as well as targeted absolute quantitative data will be
collected. The quantitative data will eventually be used as input for computational modeling of innate immune signaling
networks in the spatial and temporal dimensions. The resulting quantitative spatiotemporal proteomic data and
computational models of innate immune response can ultimately contribute to improved understanding of how proteins
cooperate in space and time to enact cellular processes.
Page 94 of 166
P07: Computational Proteomics and Data Science
P07.01 Characterizing and Correcting Batch Effects in Large Scale Proteomics Data.
Amir Alavi (1), Harendra Guturu (1), Mahdi Zamanighomi (1), Tristan Brown (1), Jian Wang (1), Sam Cutler (1), Biao Li (1),
Khatereh Motamedchaboki (2), Asim Sidiqqui (1), Serafim Batzoglou (1)
(1) Seer, Inc., Redwood City, California, United States
Recent advances in liquid chromatography mass spectrometry (LCMS)-based proteomics have enabled
the efficient profiling of thousands of proteins from single LCMS runs. The ability to run untargeted, high throughput LCMS
experiments has opened the door to large-scale cohort studies for biomarker and drug target discovery. When conducting
large-scale cohort studies, technical confounding can be introduced as samples are run across different MS instruments, LC
columns, dates, and geographic locations. In order to integrate these samples across datasets for joint analyses, one needs
to both diagnose this batch effect and apply methods to correct for it. In order to benchmark batch effect characterization
and batch effect correction methods for protein abundance data, we collect a batch-diverse dataset using Seer’s
ProteographTM Product Suite. Our dataset includes over 800 LCMS runs across two biosamples, multiple nanoparticles,
three MS instruments, and eight LC columns. We evaluate the presence of a batch effect using multiple batch effect
diagnosis methods, including Principal Components Analysis-based approaches, local-neighborhood diversity measures,
machine learning classifier-based methods, and a statistical hypothesis testing framework. We show that technical
covariates such as mass spectrometry machine and liquid chromatography column have a significant contribution to the
variability observed in protein abundance in this data. To address the observed batch effect, we survey the space of
batch effect correction methods for protein abundance data. These include traditional methods often used in proteomics and
genomics such as ComBat, dimensionality reduction methods, and nearest neighbors-based approaches. In addition, we
apply a novel deep learning-based batch correction method along with published methods, including metric learning-based
methods and adversarial approaches. We find that our proposed deep learning-based approach can remove technical
variation while preserving biological variation better than prior methods and conclude with a discussion on the advantages
and disadvantages of these different techniques.
P07.02 Deep Learning enables a Joint Embedding of Peptides and MS/MS Spectra allowing for Exploratory
Analyses via UMAP
Tom Altenburg (1), Maroua Filali (1), Jennifer I. Daniel Onwuchekwa (1), Thilo Muth (2), Bernhard Y. Renard (1)
(1) Hasso Plattner Institute, Potsdam, Germany
(2) Federal Institute for Materials Research and Testing (BAM), Berlin, Germany
MS-based proteomics, in contrast to other omics types, typically requires additional efforts in terms of peptide identification
before the actual data exploration, or even biological interpretation can take place. Here, we present a deep learning
approach that jointly embeds MS/MS spectra and peptides into the same vector space such that embeddings can be
compared easily via euclidean distances. As a result, manifold learning or clustering approaches can be directly applied to
the embeddings of MS/MS spectra. Note, our approach differs from existing spectral embedding approaches because our
joint embedding yields meaningful spectrum embeddings. In particular, the manifold of spectrum embeddings is tied to its
Page 95 of 166
peptide counterpart. We share a collection of UMAP-visualizations of our embeddings that render useful for sample
exploration, quality control, and annotation of peptide properties.
P07.03 Subtyping the Aging Brain Based on an Unbiased Brain Proteome Network Clustering Approach
E. Kathleen Carter (1), Eric B. Dammer (2), Duc M. Duong (2), David A. Bennett (3), James J. Lah (2), Allan I. Levey (4),
Nicholas T. Seyfried (1)
(1) Emory University School of Medicine, Department of Biochemistry, Atlanta, Georgia, United States
(2) Emory University, Atlanta, Georgia, United States
(3) Rush Alzheimer’s Disease Center, Rush University Medical Center, Chicago, Illinois, United States
(4) Emory University School of Medicine, Department of Neurology, Atlanta, Georgia, United States
Aging is the largest risk factor for Alzheimer’s disease (AD)--a progressive neurological disease characterized by memory
loss and cognitive dysfunction. Recent studies have shown that it may be possible to define subtypes of AD based on
clustering of brain transcriptomic data. Here, we aimed to develop an approach to classify the brain proteome in a
community-based cohort of normal aging that included AD cases using unbiased network clustering to uncover distinct
proteomic subtypes. Four hundred dorsolateral prefrontal cortex tissues were analyzed from the Religious Orders Study
and Memory and Aging Project (ROSMAP), a longitudinal study of aging, by tandem mass tag mass spectrometry (TMT-
MS)-based quantitative proteomics. After data processing, MONET M1 was used to cluster the samples using network
modularity optimization while minimizing the percentage of samples not assigned to the network. The resultant classes were
then related to case metadata. Three distinct classes of brain proteome profiles were resolved. The first class was enriched
with cognitively normal cases with modest cortical amyloid and tau pathological burden, like cases typically defined as
controls. The second class was enriched with cases having both high amyloid and tau burden and significant global
cognitive dysfunction prior to death, consistent with an AD classification. Interestingly, cases within the third class showed
changes in episodic memory performance without a significant difference in amyloid and tau burden compared to the
“control” class, suggesting that this group may represent an intermediate stage in disease progression, or be a phenotype
distinct from AD. By integrating the results of the unbiased classification with a recent consensus AD brain proteomic
network, we found that this third class had distinct patterns of protein expression compared to the AD class. These findings
highlight the utility of unbiased network analysis to uncover classes and subtypes of the aging brain in postmortem tissue.
Introduction Privacy protection is a core principle of genomic research, but has not been addressed in high-throughput
proteomic platforms. Methods Using 1,184 subjects from COPDGene and 1,500 subjects from JHS who had 1,305 proteins
Page 96 of 166
measured by SomaScan, we identified 100 proteins which had independent single nucleotide polymorphism (SNP)
quantitative trait loci (pQTL) and then calculated genotype probability functions for each protein-genotype combination
(training dataset). To test privacy protection (i.e., assess the ability to identify subjects based on their proteome profile), we
applied a naïve Bayes framework to match the protein profile to the genotype profile. Accuracy of the algorithm was tested
using 547 independent protein profiles from 9,970 genotyped subjects in COPDGene, 258 independent protein profiles from
2,638 genotyped subjects in SPIROMICS, 1,000 independent profiles from 5,000 genotyped subjects in JHS, and 980
subjects from MESA. Results For SPIROMICS, the probabilities of a correct exact subject match (1 in 2,638) was 86%, a
near subject match (top 3 in 2,638) was 93% and a close subject match (top 1%) was 96%. In COPDGene, the probabilities
for a correct exact subject match (1 in 9,970) was 83%, a near subject match (top 3 in 9,970) was 87%, and a close subject
match (top 1%) was 94%. Accuracy was lower in Blacks compared to other ancestries. When newer data sets with 4,776
proteins were used, exact subject matching increased to 98%. Reidentification accuracy plummeted and approached that
of a random guess when we adjusted out the mean genotype effect of proteins. Conclusion This study demonstrates that
proteomic data cannot be considered deidentified and suggests that large scale proteomic data used in proteomic research
should be given the same privacy protections as genomic data. Alternatively, bioinformatic transformations could be applied
to protect privacy (such as adjustment for genotype effect).
P07.05 CHalf: A User-friendly, Publicly Available Computational Tool for Protein Folding Structure Quantification
Chad D. Hyer (1), Connor Haderlie (1), Monica Berg (1), Hsien-Jung Lavender Lin (1), Isabella James (1), Tyler Bateman
(1), Michael Zackrison (1), S Anisha Daley (2), John C Price (1)
(1) Brigham Young University, Provo, Utah, United States
(2) Brigham Young University, Provo, Utah
Traditionally, protein structure has been described using methods such as X-ray crystallography and NMR spectroscopy.
While these methods have greatly expanded our understanding of protein structure, they are inherently flawed as they do
not accurately describe proteins in biological conditions and are only able to describe a small fraction of the proteome. We
have developed an assay that examines in vivo protein stability through mass spectrometry and a computational tool known
as CHalf that can calculate the chemical denature midpoints (a term we have dubbed CHalf values) of residues throughout
proteins across the proteome and is the first user-friendly, publicly available program of its kind. This assay and CHalf can
help remediate flaws in existing protein structure calculation methods by providing a quantitative measure of protein folding
strength in biological conditions across the entire proteome. Using findings from this assay and CHalf, we plan to use
agnostic machine learning methods to examine possible correlations between CHalf values and current literature solvent-
accessible surface area, Rosetta energy, and b-factor values as well as qualitative observations in protein structure. Using
these methods, we hope to be able to accurately describe protein folding quantitatively and provide insight into implications
of structural motifs on folding strength and vice versa. These aims, if realized, have potential to greatly expand the
framework of computational protein structure determination and extrapolation and to provide insight into the effects of
protein structure motifs on folding strength and function. Applying these findings in biological conditions, we expect to be
able to detect changes in folding strength across conditions to identify changes in protein folding as a result of proteomic
diseases. This application shows promise in increasing our understanding of folding patterns of diseased protein biomarkers
and is useful in understanding the biochemical pathways of disease and the effects of drug treatments.
Page 97 of 166
P07.06 Progress Identifying and Analyzing the Human Proteome: 2021 Metrics from the HUPO Human Proteome
Project
Gilbert S. Omenn (1), Lydie Lane (2), Christopher M. Overall (3), Young-Ki Paik (4), Ileana M. Cristea (5), Fernando J.
Corrales (6), Cecilia Lindskog (7), Susan Weintraub (8), Michael H.A. Roehrl (9), Siqi Liu (10), Nuno Bandeira (11), Sudhir
Srivastava (12), Yu-Ju Chen (13), Ruedi Aebersold (14), Robert L. Moritz (15), Eric W. Deutsch (15)
(1) University of Michigan, Institute for Systems Biology, Ann Arbor, Michigan, United States
(2) CALIPHO Group, SIB Swiss Institute of Bioinformatics, Geneva, Switzerland
(3) University of British Columbia, Vancouver, British Columbia, Canada
(4) Yonsei Proteome Research Center and Yonsei University, Seoul, Korea, Republic of
(5) Princeton University, Princeton, New Jersey, United States
(6) Centro Nacional de Biotecnologia, Madrid, Spain
(7) Uppsala University, Uppsala, Sweden
(8) University of Texas Health San Antonio, San Antonio, Texas
(9) Memorial Sloan Kettering Cancer Center, New York, New York, United States
(10) BGI Group, Shenzhen
(11) University of California, San Diego, San Diego, California, United States
(12) National Cancer Institute, Bethesda, Maryland, United States
(13) Institute of Chemistry, Academia Sinica, Taipei, Taiwan (Greater China)
(14) ETH-Zurich and University of Zurich, Zurich, Switzerland
(15) Institute for Systems Biology, Seattle, Washington, United States
The 2021 Metrics of the HUPO Human Proteome Project (HPP) show that protein expression has now been credibly
detected (neXtProt PE1 level) for 18,357 (92.8%) of the 19,778 predicted proteins coded in the human genome, a gain of
483 since 2020 from reports throughout the world reanalyzed by the HPP. Conversely, the number of neXtProt PE2, PE3,
and PE4 missing proteins has been reduced by 478 to 1421. This represents remarkable progress on the proteome parts
list. The utilization of proteomics in a broad array of biological and clinical studies likewise continues to expand with many
important findings and effective integration with other omics platforms. We present highlights from the Immunopeptidomics,
Glycoproteomics, Infectious Disease, Cardiovascular, Musculo-Skeletal, Liver, and Cancers B/D-HPP teams and from the
Knowledgebase, Mass Spectrometry, Antibody Profiling, and Pathology resource pillars, as well as ethical considerations
important to the clinical utilization of proteomics and protein biomarkers.
P07.07 Deep Learning Algorithm Using Collision Cross-Section (CCS) for DIA data analysis
Robin Park (1), Patrick Garrett (2), Tharan Srikumar (3), Sven Brehmer (4), Christopher Adams (5), Ignacio Jauregui (6),
Jens Decker (4), Dennis Trede (7), John Yates (2), Rohan Thakur (8)
(1) Bruker, San Diego, California, United States
(2) Scripps Research, San Diego, California, United States
(3) Bruker, Toronto, Canada
(4) Bruker, Bremen, Germany
(5) Bruker, San Jose, California, United States
(6) Bruker, Madrid, Spain
Page 98 of 166
(7) Bruker, Bremen
(8) Bruker, San Jose, California
Since Venable et al. first introduced data-independent acquisition (DIA) in 2004, DIA acquisition and data analysis tools
have been continuously improved, making DIA a vital technology to identify and quantify thousands of proteins with high
reproducibility and deep proteomics coverage. The advantage of DIA strategy is to cover all ions in a sample by a pre-
defined set of precursor isolation windows over the entire m/z range. Meanwhile, a recent breakthrough of 4D proteomics
by the timsTOF Pro adds an extra dimension of separation, providing a solution to accurately clarifying a significant number
of spectra with CCS (collisional cross-section) values. We developed algorithms to build features using CCS values and
incorporate them into deep learning to improve peptide identification. The features include an ion mobility-based search
score called TIMScore, p-values derived from the CCS distribution based on peptide candidates. To generate predicted
CCS values, we used the training data from the PRIDE repository (PXD019086) and developed a model using transformer,
neural network architecture. To evaluate the algorithms, we acquired the spectra on a timsTOF PRO from 400ng of
digested Hela proteins using a 30-minute gradient. We ran CCS-aware deep learning algorithm together with the CCS-
prediction model on the DIA data on the triplicates of the samples. We plugged the algorithm into DIA-NN open-source
version and compared it to the original version. We identified 7941 protein groups, which is 20 percent more than the
original version. We also identified 43,783 stripped peptides, which is 40 percent more. CCS-aware deep learning
algorithm showed better classification for identifying both proteins and peptides.
Matching peptide sequences to tandem mass spectra is integral to bottom-up proteomics, where chimeric spectra are
estimated to constitute >50% of data-dependent acquisition data. Some search engines allow multi-pass searches or
account for multiple possible precursors, however such approaches result in an over or underutilization of measured
fragment ions. This introduces errors or leaves valuable information unused, resulting in far fewer peptide identifications
than could be obtained in the data. Here, we describe Proteome Discoverer 3.0 software with CHIMERYS, an intelligent
search algorithm, that rethinks the analysis of tandem mass spectra. This innovative approach routinely doubles the number
of peptide identifications and reaches identification rates of >80% for typical proteomic data sets. CHIMERYS uses accurate
predictions of peptide fragment ion intensities and retention times provided by the deep learning framework INFERYS.
CHIMERYS aims to explain as much measured intensity with as few candidate peptides as possible, resulting in the
deconvolution of chimeric spectra. We validated this approach in multiple ways including entrapment searches and an in-
silico chimeric spectra system. Here, analyzing a HeLa cell lysate digest with CHIMERYS identified 114k PSMs, 61k unique
Page 99 of 166
peptides and 7,300 unique protein groups at 1% FDR. This is a 3.5-, 2- and 1.5-fold increase when compared to Sequest
HT, respectively, resulting on average in 2.5-fold more identified peptides per protein. CHIMERYS is compatible with all
Orbitrap mass spectrometers but provides more additional identifications from the increased sensitivity of recent
instruments. CHIMERYS provides exceptional performance with short chromatographic gradients and high protein loads,
enabling higher throughput and a deeper mining of data. CHIMERYS is the first search algorithm that embraces chimeric
spectra in a highly scalable, cloud-native, AI-powered implementation.
P07.09 MetaNetwork: Systems Biology Insights through Quantitative Proteomics and Weighted Correlation
Network Analysis
Austin Carr (1), Brian Frey (1), Mark Scalf (1), Anthony Cesnik (1), Zach Rolfs (1), Kyndal Pike (1), Bing Yang (2), David
Jarrard (3), Mark Keller (4), Lloyd Smith (5)
(1) Department of Chemistry, University of Wisconsin-Madison, Madison, Wisconsin, United States
(2) Department of Urology, University of Wisconsin-Madison, Madison, Wisconsin, United States
(3) Department of Urology, University of Wisconsin-Madison, Madison, Wisconsin
(4) Department of Biochemistry, University of Wisconsin-Madison, Madson, Wisconsin, United States
(5) Department of Chemistry, University of Wisconsin-Madison, Madison, Wisconsin
Proteomics experiments can enable deep insight into biological differences between sample groups. Differential expression
analysis is commonly used to identify the subset of proteins with the most statistically significant and interesting fold-
changes. By contrast, weighted correlation network analysis (WCNA), uses all measured gene product abundances in an
unsupervised clustering and network analysis with a proven capacity for revealing important networks and pathways.
However, the use of WCNA in proteomics has been limited due to difficulties in implementing WCNA workflows.
Consequently, we developed MetaNetwork, which includes a customizable workflow, visualization of outputs, and a built-in
functional annotation enrichment tool. We use MetaNetwork to identify changes in the biological pathways that may enable
the differentiation of indolent from aggressive prostate tumors. Six aggressive and six indolent radical prostatectomy
specimens sectioned into tumor or histologically “normal” tissue. Two-dimensional LC/MS-MS was performed on trypsinized
peptides derived from each sample. The data was searched with MetaMorpheus version 0.0.311. WCNA was performed on
the protein intensity values using MetaNetwork v1.0 with default settings. From the 2441 proteins uploaded to MetaNetwork,
the WCNA workflow identified ten modules. We found that the networks identified by MetaNetwork’s unsupervised
clustering are highly enriched in protein-protein interactions. Next, we used module eigenprotein analysis to identify two
statistically significant up-regulated and one down-regulated modules. These significant changes only occurred between the
aggressive tumor samples and their paired histologically normal tissue. Using MetaNetwork’s built-in functional annotation
enrichment analysis tool, we identified potential biological roles for modules with statistically significant changes: the down-
regulated module is enriched in terms related to the cytoskeleton and the extra-cellular matrix, and the two up-regulated
modules are enriched in terms related to mRNA translation, RNA metabolism, and protein trafficking. MetaNetwork is also
capable of generating many additional layers of information to facilitate the interpretation of differential expression analysis.
Purpose: Clinical Proteomic Tumor Analysis Consortium (CPTAC) represents the largest proteogenomic data and is a
valuable resource for enhancing our understanding of cancer biology. Proteomics data from CPTAC was primarily
generated using TMT multiplex approach, which was designed to provide relative protein quantification between samples.
We aim to develop a computational approach to estimate absolute protein abundance from CPTAC TMT global proteomics
data. Methods: Label-free proteomics raw intensity data (100 tumors) of colon cancer from CPTAC were processed using
MaxQuant. TMT-10 plex mzML files (97 tumor & 100 NAT) were downloaded from CPTAC Data Portal and processed
through FragPipe to generate ratio and abundance (ion abundance) based protein quantification. Label-free raw intensity
and TMT abundance data were analyzed by total protein approach (TPA, two methods) and intensity based absolute
quantification (iBAQ) to derive estimated absolute protein abundance. Results: Using CPTAC colon cancer dataset, we
found that estimated absolute protein abundance derived from label-free proteomics data using TPA (two methods) and
iBAQ are highly correlated (TPA1 vs TPA2: r = 0.90; TPA1 vs iBAQ: r = 0.85; and TPA2 vs iBAQ: r = 0.95). Further, our
analysis revealed high concordance between TMT ratio data generated by FragPipe and published value (Vasaikar et al.
2019) (r = 0.86). Lastly, we showed that estimated absolute protein abundance derived from TMT and label-free using TPA
displayed relatively good correlation (TPA1: r = 0.72; TPA2: r = 0.76). Conclusions: We observed relatively good correlation
of estimated absolute protein abundance derived from both label-free and TMT global proteomics data of colon cancer in
CPTAC, suggesting TPA and iBAQ could be applied to TMT global proteomics data to derive estimated absolute protein
abundance. Reference: Vasaikar et al. 2019, Cell, 177, 1035-1049
P07.11 Identifying Drivers of Human Therapeutic Antibody Production in Chinese Hamster Ovary Cells via
Proteomics and Bioinformatics
Junmin Wang (1), Raghothama Chaerkady (1), Lina Chakrabarti (1), Shao Huan Samuel Weng (1), Jie Zhu (1), Diane
Hatton (1), Lisa Cazares (1), Sonja Hess (2)
(1) AstraZeneca, Gaithersburg, Maryland, United States
(2) AstraZeneca, Gaithersburg, Maryland
The Chinese hamster ovary (CHO) cell is the primary workhorse for human therapeutic antibody production. Chakrabarti et
al. (2019) showed that mitochondrial membrane potential (MMP) identifies CHO cells with high recombinant protein
productivity. Nevertheless, our ability to engineer high-yielding cell lines has been limited by insufficient knowledge of
proteins driving antibody production in CHO cells. Poor characterization of protein functions and limited sample sizes
constitute major challenges for mass spectrometry-based studies of the Chinese hamster proteome. To identify drivers of
antibody production, we performed an LC-MS/MS proteomic analysis on a Q-Exactive HF-X mass spectrometer in CHO
cells with normal and high MMP. The MaxQuant protein and peptide output was used for subsequent analysis. To improve
protein annotations, we mapped CHO proteins to murine homologs via pairwise global sequence alignment and a logistic
regression model with sequence percentage identity as its feature. Subsequently, we repurposed CAMERA, a gene set test
developed by Wu and Smyth (2012), for differential protein expression analysis. Our approach permits analysis to be carried
out on an ensemble of peptides originating from the same protein, facilitating the validation of subtle protein changes. Our
P07.13 Increasing Human Leukocyte Antigen Peptide Identifications Using Fragpipe and MSBooster
Kevin Yang (1), Fengchao Yu (1), Guo Ci Teo (1), Vadim Demichev (2), Markus Ralser (2), Alexey Nesvizhskii (1)
(1) University of Michigan, Ann Arbor, Michigan, United States
(2) Charite Universitatsmedizin, Berlin, Germany
Peptide antigen presentation via the major histocompatibility complex (MHC) is a crucial process by which the immune
system can detect and target bacterial and viral infections as well as cancers. The human leukocyte antigen (HLA)
peptidome presents a unique challenge for database searching of LC-MS/MS data, as the number of possible sequence
identities far exceeds those possible in standard enzymatic digests such as tryptic peptide databases. Previous software
solutions have used deep learning (DL) models to rescore PSMs using features describing the consistency of predicted
MS2 spectra and retention time with experimentally observed values. Here, we present a workflow in Fragpipe with
Percolator for PSM rescoring of HLA-I peptides. Incorporation of DL features is available through MSBooster, a new
addition to Fragpipe. Using reversed-phase fractionated samples from an A∗02:01 monoallelic cell line (Klaeger et al,
2021), rescoring with DL features improved peptide identifications by more than 25% compared to a baseline model without
DL rescoring. Another difficulty of searching HLA data is that they fragment differently from tryptic peptides due to their non-
enzymatic nature. Some HLA peptides display strong internal fragment ion series, which are not typically handled by DL
models predicting b- and y-ions. We perform a comparison of spectral similarity features using either 1) a baseline model
trained on tryptic peptides, 2) the Prosit HLA model, or 3) the PredFull full spectrum prediction model. This is the first
application of testing full spectrum prediction for HLA peptide rescoring, rather than using just b- and y-ions.
P07.14 One-stop DIA data analysis using MSFragger-DIA coupled with FragPipe
Fengchao Yu (1), Kevin Yang (1), Vadim Demichev (2), Sarah Haynes (1), Markus Ralser (2), Alexey Nesvizhskii (1)
(1) University of Michigan, Ann Arbor, Michigan, United States
(2) Charité - Universitätsmedizin Berlin, Berlin, Germany
Data-independent acquisition (DIA) is a widely used data acquisition approach in bottom-up proteomics. Most DIA peptide
identification methods require extraction of precursor and fragment ions, either by detecting features or using spectral
libraries, prior to database searching. Here, we propose an approach to directly search DIA spectra against a database
using MSFragger-DIA, a new module in MSFragger. We can identify peptides directly from DIA data and then build a
spectral library for quantification. To increase the sensitivity of peptide identification, we have also developed MSBooster, a
P07.15 moPepGen: a fast custom database generator from multi-omics data for proteogenomics
Chenghao Zhu (1), Lydia Y. Liu (2), Paul C. Boutros (1)
(1) Department of Human Genetics, University of California, Los Angeles, Los Angeles, California, United States
(2) Department of Medical Biophysics, University of Toronto, Toronto, Canada
Purpose: Cancers are driven by genomic variants such as SNV (single nucleotide variants) and INDEL, often accompanied
by many transcriptional variants. Modern mass spectrometry based proteomics is able to identify and quantify peptides and
proteins comprehensively, however the variant-harboring peptides that are absent in canonical databases are largely under-
studied. A common problem in custom database construction is the large number of combinations of variants to consider.
Thus major proteogenomic studies often choose a strategy that only captures peptides harboring a single variant. Existing
algorithms also suffer from limited sources of variation. Methods: Here we present moPepGen (multi-omics peptide
database generator) that aims at accelerating proteogenomic researching by generating custom peptide database from
variety of genomic and transcriptional variants. MoPepGen uses a graph-based algorithm that achieves a linear time
complexity in integrating variants in contrast to the exponential complexity of exhaust searching. MoPepGen integrates
variants from a variety of sources including SNV, indel, transcription fusion, alternative splicing, RNA editing, and circRNA. It
is able to generate peptides harboring any combinations of variants on the single peptide. MoPepGen is also highly
extensible for other types of genomic and transcriptional variants. Results: With moPepGen we created sample-specific
database noncanonical peptide harboring SNV, transcriptional fusion and RNA editing events from ~1000 cancer cell lines.
The run time for generating custom database is around 40 minutes per sample. We then applied a tiered library searching
strategy using three independent search engines and identified tens to thousands of variant-harboring peptides across 36
cancer types. Conclusion: This study demonstrated the capacity of moPepGen in identifying noncanonical peptides in
proteogenomics studies and extended the potentiality of proteomics in precision medicine and biomarker discovery.
Abstract: Data independent acquisition has become the go to method for deep and quantitative proteomic analysis
given the ability to sample large m/z windows in a reproducible and non-stochastic manner. Using a method termed dia-
PASEF on a TIMS enabled Q-TOF lends additional advantages in both duty cycle and selectivity using the ion mobility
space. Dia-PASEF allows for deep proteomes in short gradient times (<20 min.) and therefore 100’s of LCMS runs can be
generated in short times. Data analysis in a streamlined automated fashion expedites the time from experiments to
discovery. DIA-NN is a novel software package that uses neural networks providing best-in-class DIA output. Here we
integrate DIA-NN onto the PaSER platform for a streamlined workflow for the analysis of many samples in a short analysis
time with no file transfer or data migration. Methods: A timsTOF Pro using variable CE’s and mobility windows in gradients
ranging from 5 min. to 90 min. were used. DIA-NN was modified to become CCS-enabled and process data in the
most expedient fashion. The PaSER GUI was designed such that first-pass analysis is predefined automatically triggering
quantitative analysis. Retrospectively, match-between-runs (MBR) analysis can be triggered on the whole project or subset
of user defined experiments. Preliminary Data: Human, Yeast and E. coli (HYE) digested mixtures at different but known
ratios with injection loads from 50ng to 600ng were run at different gradient lengths in replicate resulting in >2500 proteins
at short gradients to >9000 proteins identified and quantified at longer gradients. Quantitative accuracy was shown to
be <20% CV. Using the DIA-NN as integrated into PaSER creates a seamless approach to dia-PASEF analysis. Novel
Aspect: Automated workflows for dia-PASEF using DIA-NN on PaSER streamlines experiments to results >
P08.03 A Cloud-scalable Software Suite for Large-Scale Proteogenomics Data Analysis and Visualization
Margaret Donovan (1), Harsharn Auluck (1), Arjun Vadapalli (1), Yan Berk (1), Aaron Gajadhar (1), Khatereh
Motamedchaboki (1), Yuandan Lou (1), Theo Platt (1), Asim Siddiqui (1)
(1) Seer, Redwood City, California, United States
Assessment of the flow of genetic information through multi-’omic data integration can reveal the molecular consequences
of genetic variation underlying human disease. Next-generation sequencing (NGS) can be used to identify
genetic variants, while mass spectrometry can be used to assess the
proteome. The ProteographTM Product Suite1, leverages multiple nanoparticles with distinct physiochemical
properties to enable large-scale, deep plasma proteome analyses. Integration of proteomics and genomics
data requires many tools of which require complex workflows that can act as a barrier for researchers to adapt new analysis
tools. Here, we present a cloud-based, data analysis software platform called Proteograph Analysis Suite (PAS) for
proteogenomic data analyses through the integration of Proteograph proteomics data
with NGS variant information. PAS includes an experiment data management system, analysis protocols, analysis
setup wizard, and result visualizations. PAS can support both Data Independent Analysis (DIA) and Data Dependent
Analysis (DDA) workflows and is compatible with variant call format (.vcf) files, enabling personalized database
FragPipe is a complete proteomics pipeline for a variety of workflows and experiment types. It can be used on both Linux
and Windows systems, where users can choose from an array of complete analysis workflows to easily load and run without
the need to set options manually. Typical analyses with FragPipe involve only a few steps: load a workflow, add LC-MS/MS
spectral files, and click "Run". FragPipe is a graphical interface that brings together a number of tools which can be used in
complete workflows or as individual data processing steps. Fast database searching can be performed with MSFragger.
Direct analysis of data-independent acquisition (DIA) data can be done with MSFragger-DIA or DIAUmpire combined with
DIA-NN quantification. Crystal-C can be used for post-search of mass shift artifacts from open search. Peptide validation
can be performed with either PeptideProphet or Percolator, with protein inference by ProteinProphet. Philosopher can
perform isobaric labeling-based quantification and generate multi-level reports. PTM-Shepherd provides summarization and
characterization of possible post-translational modifications (PTMs) found from open searches. Label-free quantification
(with optional FDR-controlled match-between-runs) can be accomplished with IonQuant, and TMT or iTRAQ data can be
analyzed with TMT-Integrator. Available workflows include traditional (closed) search with label-free quantification; open
(mass tolerant) or mass offset searches for PTM discovery; non-specific (HLA or peptidome) searches; isobaric
quantification (TMT or iTRAQ) at the gene, protein, peptide, and PTM site levels; isotopic labeling (SILAC, dimethyl, custom
MS1-based labels); glycopeptide (N- and O-linked; and DIA search and quantification. FragPipe can be used with a variety
of LC-MS/MS data types, including raw files directly from Orbitrap and timsTOF instruments as well as spectral files
converted to the mzML format.
P08.05 TIMS Viz for Mobility Offset Mass Aligned interrogation of complex samples
Tharan Srikumar (3), Philipp Strohmidel (1), Sebastian Wehner (1), Jens Decker (1), Ignacio Auregui (1), Christopher
Adams (2), Sven Brehmer (1)
(1) Bruker Daltonics GmbH & Co. KG, Bremen, Germany
The PASEF® acquisition mode of the timsTOF Pro has the power to isolate co-eluting, quasi-isobaric peptides separately
for fragmentation, based on differences in the peptide’s ion mobility. Such an event is called Mobility Offset Mass Aligned
(MOMA) and results in non-chimeric spectra, despite a quadrupoles fidelity. This is especially valuable for PTM analysis, for
example to resolve positional isomers of phosphopeptides. TIMS Viz was introduced in PaSER to visualize MOMA events in
complex samples. During acquisition, MS/MS spectra are streamed to a GPU-powered processing computer performing a
real-time database search, called PaSER. The database search utilizes all four dimensions – retention time, CCS value, m/z
and fragment spectra – to increase confidence in the identification results. TIMS Viz, a novel data visualization tool to
display an interactive heatmap in the m/z ion mobility space, maps MOMA features. Herein, we show the number of MOMA
groups, which are sets of at least two MOMA features, that could be identified by TIMS Viz in two different data sets with
different m/z tolerance settings. Setting tolerances to 500 mDa (typical lower limits of quadrupole isolation) and a retention
time window of 10 s resulted in more than 40,000 MOMA groups containing more than 90,000 spectra for both, the cell
lysate sample and the phosphopeptide enriched sample. Without the power of ion mobility separation these spectra would
likely be chimeric in nature. Lowering the m/z tolerance to 25 mDa (well below the tolerance of any quadrupole) still leads to
more than 18,000 MOMA groups (> 40,000 spectra) for the cell lysate and more than 23,000 MOMA groups (> 52,000
spectra) for the phosphopeptide enriched sample. TIMS Viz helps user to explore their data for MOMA features and is a
powerful demonstration how the TIMS dimension can improve the spectral quality for co-eluting, quasi-isobaric peptides.
P08.06 ASSESSMENT OF DISCORDANCE BETWEEN PEPTIDE AND PROTEIN LEVEL DIFFERENTIAL EXPRESSION
ANALYSIS RESULTS IN CANCER BIOMARKER DISCOVERY
Ginny Xiaohe LI (1), Alexey Nesvizhskii (1), Marcin Cieslik (1), Michael Mumphrey (1)
(1) University of Michigan, Ann Arbor, Michigan, United States
Nowadays, molecular biomarker discovery relies highly on mass spectrometry proteomics. In bottom-up proteome inference
and quantitation, a single abundance measurement on proteins are routinely used in comparing between subgroups of
patients. Such measurement on protein is obtained by aggregating intensities of peptides mapped. There are two major
sources of inaccuracy that might weaken the reliability of protein level quantification. First, a given tryptic peptide can be
derived from multiple different proteoforms. Assigning it to any of the proteoforms is a simplification of reality without full
confidence. Second, the assumption underlying the aggregation that multiple peptides mapping to a gene behave in the
same direction is often violated. Hence, protein level aggregation of multiple peptides may average out true biological
effect. In this manuscript, we study discordant discoveries made when contrasting abundance measurements of tumor with
normal adjacent tissue (NAT) samples in protein-level as opposed to peptide-level based on data in major cancer cohorts
from the Clinical Proteomics Tumor Analysis Consortium (CPTAC). To start with, we investigate the discordances seen in
the highly polymorphic HLA proteins. As an example, in 212 Lung Squamous Cell Carcinoma (LSCC) samples, 31 peptides
are assigned to HLA-A protein (ENSP00000366002.5), among these peptides, 4 and 2 peptides are found to be significantly
enriched and depleted in tumor samples compared to NAT samples respectively. We then calculate protein level abundance
measurement only from peptides that are uniquely mapped to the protein, and compare with conventional protein level
quantification. In addition, we provide visualization of the peptide-proteomform mapping topology together with projection of
conserved protein domains and known PTMs. In sum, this work explores the complexity in bottom-up proteomics presented
P08.07 Absolute Quantification (AQUA) of Vitellogenin Proteins in Striped Bass Ovary: Considering the Traditional
PRM and the Nontraditional DDA Workflows
Taufika Islam Williams (1), Cara Kowalchyk (2), Leonard Collins (1), Benjamin Reading (1)
(1) NCSU, Raleigh, North Carolina, United States
(2) NC Division of Marine Fisheries, Raleigh, North Carolina, United States
Mass spectrometry has steadily but surely moved into the forefront of quantification-centered protein research. The AQUA
method of Protein Cleavage Isotope Dilution Mass Spectrometry (PC-IDMS) is a proven way for quantifying proteins by
using an isotope-labeled analogue of a peptide fragment of the parent protein as an internal standard. Such targeted
analyses are widely performed using triple-quadrupole-MS-based Multiple Reaction Monitoring (MRM) experiments and
orbitrap-based Parallel Reaction Monitoring (PRM) workflows. PRM has lately been the go-to approach for such
quantification, as it is believed to offer highest instrument sensitivity in an orbitrap-based instrument, just as MRM offers
lowest limits of detection (LOD) in a triple-quadrupole-MS. Enhanced sensitivity is considered to be elemental in the
accurate and precise delivery of quantitative data. We performed a comparative study of DDA- and PRM-based workflows
using the quantification of vitellogenins in striped bass (Morone saxatils) eggs. Striped bass are an important recreational
and commercial fish species native to North Carolina, however, natural recruitment is limited in key water systems such as
the Tar, Neuse, and Cape Fear Rivers. The study of vitellogenins, which are glycolipophosphoproteins critically important to
the striped bass reproductive cycle, can shed some light on adaptations associated with wild type stock diminishing.
Vitellogenins are precursors to egg yolk proteins (YPs), and the ratios of key vitellogenins plays an important role in egg
quality throughout development and different environmental conditions. We employed both DDA and PRM experiments to
quantify VTG Aa, Ab and C in striped bass ovaries. Our results clearly indicate that it may not be necessary to perform a
targeted analysis such as PRM, as DDA is able to provide both absolute quantification data and discovery data in a single
nanoLC-MS/MS run.
Purpose Technological advances mean we are becoming increasingly efficient at producing large amounts of omics data;
however, our ability to translate this into tangible meaning is falling behind. Large numbers of analytical tools tends to
hinder, rather than help; the complexity only getting worse with multi-omics studies. To enable everyone to understand and
use omics, we created SimpliFi, the world’s first cloud-based, GPU-driven, streamlined and browser-accessible data-to-
meaning engine for integrated omics analyses. SimpliFi obligates QC including batch and run order effects and results.
Results are easily shared, explored or published simply by sending a URL. Methods SimpliFi models biology, using
Intact glycopeptide analysis using mass spectrometry has become increasingly effective due to hybrid fragmentation
protocols such as EThcD and stepped HCD. These allow the identification of thousands of glycopeptide spectra from
complex mixtures. However, within these datasets there are typically spectra of other glycoforms of the same glycopeptides
that are not identified due to either poorer quality fragmentation data or unanticipated modifications. Bioinformatic strategies
to uncover these extra glycoforms would provide more comprehensive sample characterization. Different glycoforms of the
same glycopeptides produce the same Y0/Y1/Y2 ions. We have previously adapted the software MS-Filter to automatically
calculate the expected masses of these ions from list of input peptides and report all spectra in peak list files that contain
these ions along with tentative glycan assignments. We have now added peptide and glycan scoring to the output, which
allows reliability filtering of these results and viewing of annotated spectra for all assignments. We have evaluated the
performance of this strategy for analysis of both EThcD and HCD data. Comparing to results from database searching in
each case we were able to identify more glycoforms. For many of these spectra the same identification was reported by
database searching but was below the 1% FDR threshold applied. A particular advantage of the strategy is the ability to
report masses of modifications that do not correspond to anticipated glycan structures. This has allowed detection of
unusual glycan modifications such as multiply phosphorylated glycans and the presence of salt and metal adduct spectra.
Analytical workflows relying on higher energy collisional dissociation (HCD) tandem mass spectrometry have evolved into
powerful tools for glycopeptide analysis, but topological characterization of multiply-fucosylated glycopeptides continues to
challenge the field. Common glycopeptide fragmentation strategies prioritize peptide backbone fragmentation, and rarely
consider the value of glycan topological information in glycopeptide tandem mass spectra. We sought to apply stepped
collision energy HCD (sceHCD) and electron transfer dissociation with HCD supplemental activation (EThcD) to the analysis
of multiply-fucosylated glycopeptides from head and neck squamous cell carcinoma lysates, to produce information-rich
glycopeptide spectra with both peptide and glycan topological information and aid in the assignment of fucosylated
carbohydrate epitope topology. Fucosylated glycopeptides were enriched via (a) AAL lectin, and (b) anti-LeY antibody from
CAL27 and HSC-3 tongue squamous cell carcinoma lysate digests, followed by nLC-MS/MS on an Orbitrap Eclipse Tribrid
Mass Spectrometer with an EASY-nLC 1200 system. An Acclaim PepMap 100 trapping column and a PepMap RSLC C18
analytical column were employed for chromatographic separation, and HCD (35% collision energy) followed by oxonium ion
triggered sceHCD and EThcD dependent scans were performed. Analyses were performed in triplicate. Spectra were
analyzed using Byonic, and assigned spectra were manually verified. The nLC-MS/MS analyses led to the successful
assignment of 212 unique high confidence glycopeptides (score ≥ 200) representing 73 glycoproteins, and 52% of the
P09.03 Correlation of Quantitative Analysis of Glycoproteomic Biomarkers from Dried Blood Spots and Serum.
Hector Huang (1), Liam Magee (1), Apoorva Srinivasan (1), Prasanna Ramachandran (1), Darragh Buckley (1), Kaity Moser
(1), Ley Hian Low (1), Carrie Smith (2), Xin Cong (1)
(1) InterVenn Biosciences, South San Francisco, California, United States
(2) InterVenn Biosciences, South San Francisco, California
Dried blood spot (DBS) cards have been used as a biosampling method for over a century and were introduced into the
clinic fifty years ago. These cards, which collect finger or heel prick blood in infants, provide an effective and affordable
alternative to traditional phlebotomy in many different applications including genomics, metabolomics and proteomics. We
have recently demonstrated that glycoproteomic analysis of peripheral blood proteins provides a powerful method for
ovarian cancer screening. Here we present a method for extracting glycoproteins from DBSs followed by Multiple Reaction
Monitoring (MRM) liquid chromatography-mass spectrometry (LC-MS) analysis to quantify glycopeptide abundances. The
work presented here provides a quantitative comparison of glycopeptide abundances extracted from DBS and serum. In a
pilot study we found that glycopeptides can be extracted from DBS samples in a reproducible and consistent manner, and
that relative abundances of individual glycopeptides correlate well with relative abundances determined in serum with a
Pearson product-moment correlation coefficient > 0.8. To validate these results, we applied our extraction protocol to DBS
samples collected from 29 individuals enrolled in the Clinical Validation of the InterVenn Ovarian CAncer Liquid biopsy
clinical trial (ClinicalTrials.gov NCT03837327), which examines the ability to discriminate malignant from benign pelvic
tumors, based on their blood glycoproteomic signature. Roughly half of the 29 assessed individuals had benign histologies,
while the other half were confirmed to have an advanced form of ovarian cancer. Principal component analysis of
glycoproteomic signatures of DBS samples confirmed this distribution. Our results indicate that samples collected on DBS
can be used as a source for MRM-based analysis of glycoproteomic biomarkers in peripheral blood, with sufficiently robust
performance to be applied for distinguishing between benign and malignant ovarian tumors.
Aberrant glycosylation has been observed in solid tumors for decades, but little is known about the significance of altered
glycosylation and its impact on immunomodulation and tumor infiltration. We hypothesize that glycosylation patterns at the
maternal-fetal interface are highly reminiscent of that in tumors, and the goal of this study is to determine changes in
P09.05 Selective enrichment of cell surface proteins using ProMTag to detect O-glycosylation-dependent changes
Amber Lucas (1), Jonathan Minden (1), Collin Bachert (2), Adam Linstedt (2)
(1) Impact Proteomics, Pittsburgh, Pennsylvania, United States
(2) Carnegie Mellon University, Pittsburgh, Pennsylvania, United States
O-linked glycosylation, which is regulated by the enzyme GalNAc transferase, plays a vital role in cellular function by
regulating membrane protein composition and sorting. Changes in O-linked glycosylation are very common in diseases
such as cancers, diabetes, and Alzheimer’s. It is important to study the effects of changes in O-link glycosylation on cell
membrane composition and function to understand how these changes affect the onset, phenotype, and progression of
these diseases. However, membrane proteins are notoriously challenging to analyze at the proteomic level due to their
hydrophobicity and the inability to separate the membrane proteins from other proteins in the cell.
Here we introduce a new sample preparation workflow that enriches membrane proteins for mass spectrometry analysis to
understand changes in protein composition of the plasma membrane as the result of changes in O-linked glycosylation. To
do this, we used the bifunctional tag ProMTag to label and capture plasma membrane proteins via their extracellular
domains in cells treated with GalNAc Transferase inhibitors. One end of the ProMTag forms a reversible, covalent bond with
protein the other end of the ProMTag is capable of forming an irreversible, covalent bond with a solid bead support via the
click chemistry pair of methyltetrazine and trans-Cyclooctene (TCO). The extracellular domains of proteins from intact wild
type HEK cells, HEK cells treated with GalNAc Transferase inhibitors, and HEK cells with GalNAc Transferase mutations
were ProMTagged, captured, cleaned up, and eluted. The samples were then TMT labeled and analyzed using mass
spectrometry (MS). MS confirmed enrichment of plasma membrane proteins after sample preparation using this new
ProMTag workflow. Protein changes in the plasma membrane as a result of GalNAc Transferase inhibition or mutation were
characterized. This work establishes a method for enrichment of plasma membrane proteins and reveals how changes in O-
linked glycosylation via GalNAc Transferase inhibition leads to changes in plasma membrane protein composition due to
altered membrane protein stability and sorting.
P09.06 Sialic Acid Derivatization and Bioorthogonal Labeling for Multiplexed Mass Spectrometry and Fluorescence
Imaging
Colin McDowell (1), Xiaowei Lu (1), Rachel Stubler (1), Luke Wisniewski (2), ChongFeng Gao (2), Peggi Angel (1), Brian
Haab (2), Richard Drake (3)
(1) Medical University of South Carolina, Charleston, South Carolina, United States
(2) Van Andel Institute, Grand Rapids, Michigan, United States
(3) Medical University of South Carolina, Charleston, South Carolina
Sialic acid-modified glycans are critical regulators of cell-cell communication through their direct interaction with a host of
carbohydrate binding proteins. α2,3- and α2,6-sialic acid linkage isomers have disparate biological functions and are
differentially modulated by cancer and other nonmalignant diseases. Imaging mass spectrometry (IMS) is a well-established
tool for the analysis of glycosylation, however the inability to resolve isomers by mass makes characterization of sialic acids
challenging. Here we present a novel sialic acid derivatization approach for isomeric characterization by mass spectrometry
and fluorescence imaging. In preparation for N-glycan imaging mass spectrometry, sialic acid residues on fixed cells and
FFPE tissues were differentially modified with alkyne (α2,3) and amide (α2,6) moieties by a novel propargylamine-based
amidation-amidation, termed AAXL. AAXL-labeled cell slides were then subsequently integrated into a previously described
N-glycan IMS sample preparation workflow1. For fluorescence imaging, sialic acids were derivatized with azide (α2,3) and
amide (α2,6) groups via a similar amine-azide-based amidation, termed AAN3. Azide-functionalized α2,3 isomers were
further reacted with a biotin-alkyne through click-chemistry, which then captured a streptavidin-Cy5 probe for fluorescent
detection. The AAXL synthesis successfully differentiates α2,3- and α2,6-sialylated N-glycan isomers by imaging mass
spectrometry analysis. AAXL-derivatization introduces +37.0316 m.u. (α2,3) and +27.0473 m.u. (α2,6) mass shifts which
resolve the isomers by mass spectrometry. Via AAXL we detect over 100 sialylated N-glycan species from FFPE tissues
and fixed cells, where AAXL revealed distinct sialic acid isomer localization. Using AAN3 and subsequent biotin labeling of
α2,3-linked isomers, we have successfully detected α2,3-sialylated glycans by fluorescence imaging. The AAN3 fluorescent
labeling strategy demonstrated disparate α2,3-sialic acid expression across FFPE tissue regions and between multiple cell
lines. This dual bioorthogonal labeling system is a novel tool for multiplexed sialic acid imaging, which is further adaptable to
glycomic and glycoproteomic enrichment strategies.
P09.07 Cross-Ring Fragmentation of N-linked Glycans via Lithium-Doped ESI Using Matrix-Assisted Laser
Desorption Electrospray Ionization Mass Spectrometry
Tana Palomino (1), Crystal Pace (1), David Muddiman (1)
(1) North Carolina State University, Raleigh, North Carolina, United States
N-linked glycans play an important role in essential biological processes. They consist of highly complex structures due to
their immense diversity in possible isomers and linkages. To characterize glycan structures with high specificity, tandem
mass spectrometry is required to induce glycosidic cleavages and cross-ring fragments. While glycosidic cleavages provide
sequencing and branching information, cross-ring fragmentation elucidate to specific linkage information between
monosaccharides. However, cross-ring fragmentation is not as abundant as glycosidic cleavages and thus, linkage
P09.08 MS-Based Glycomics and Glycoproteomics facilitating the Identification of Glycan and Glycopeptide
Isomers as Reliable Alzheimer's Disease Biomarkers
Yehia Mechref (1)
(1) Texas Tech University, Lubbock, Texas, United States
Alzheimer’s disease (AD) results in a progressive loss of cognitive function and dementia, affecting in the United States
more than one in nine people by the time they reach 65 years of age. Strikingly, the neuropathological changes in the brain
can precede the onset of symptoms by decades, providing an important diagnostic and therapeutic window. Reliable,
sensitive, and specific biomarkers for early diagnosis and disease monitoring would be of paramount importance in clinical
practice and for the development of specific therapeutic interventions. Glycosylation is emerging as a new frontier in
biomarker development, but it is still a largely unexplored field in AD. The relevance of using glycome as a biomarker source
is underlined by the key role of glycan structures in disease pathogenesis and pathology modulation, owing to their
influence on a variety of essential pathobiological functions, including cell growth and differentiation, cell communication,
and neurodegeneration, among others. Also, glycans represent organ-specific markers. We have developed a powerful,
highly sensitive, and reproducible LC-MS/MS-based platforms to analyze glycoproteins and glycans. Due to the fact that all
well-known AD-related proteins are glycosylated, AD patients brain tissue, CSF and serum would present a unique
glycome/glycoproteome biosignature indicative of AD pathophysiology;which will have distinctive representation that show
common and unique representation in tissue and biofluids. Specific glycans, glycan isomers and/or glycoproteins detectable
in biofluids are markers associated with neurodegenerative processes in patients with AD. Appropriate markers, alone
and/or in combination, can provide information critical for the diagnosis of neurodegenerative diseases such as AD. LC-
MS/MS will be utilized to define the glycome and glycoproteome changes associated with the development and progression
of AD. The LC-MS/MS to be used will facilitate the characterization of glycan and glycopeptide isomers, providing the first
assessment of the changes in the isomeric distribution of glycan/glycopeptide in AD.
P09.09 Full Glycoproteomics Analysis Using FragPipe: From Peptide Identification to Glycan FDR Assessment and
Quantitation
Daniel Polasky (1), Daniel Geiszler (1), Fengchao Yu (1), Yi Hsiao (1), Hui-Yin Chang (1), Guo Ci Teo (1), Alexey
Nesvizhskii (1)
(1) University of Michigan, Ann Arbor, Michigan, United States
P09.10 Ancestry-Dependent Metabolic Links to Socioeconomic Stresses Revealed in Normal Breast Tissue at Risk
for Breast Cancer
Denys Rujchanarong (1), Yeonhee Park (2), Danielle Scott (1), Sean Brown (3), Richard Drak (1), Marvella Ford (1),
Michael Ostrowski (1), George Sandusky (4), Harikrishna Nakshatri (5), Peggi Angel (6)
(1) Medical University of South Carolina, charleston, South Carolina, United States
(2) University of Wisconsin-Madison, Madison, Wisconsin, United States
(3) Medical University of South Carolina, Charleston, United States
(4) Indiana University School of Medicine, Indianapolis, Indiana
(5) Indiana University School of Medicine, Indianapolis, Indiana, United States
(6) Medical University of South Carolina, Charleston, South Carolina
A primary difference between black women (BW) and white women (WW) diagnosed with breast cancer is not incidence,
but the aggressiveness of the tumor. BW have higher mortalities with similar incidence of breast cancer compared to other
race/ethnicities. Black women are diagnosed at younger ages with more advanced tumors that have double the rate of
lethal, triple negative breast cancers. There is ongoing debate regarding whether the underlying cause of higher mortality
relates more to healthcare inequalities or to ancestry-dependent molecular features found in normal breast tissue that
facilitate the aggressive phenotype found in BW. Our hypothesis is that chronic social and economic stressors that differ by
race/ethnicity result in molecular responses that create a tumor permissive tissue microenvironment in normal breast tissue.
In this study, we investigate molecular pattern differences in a cohort of ancestry defined normal breast tissue from the
Susan G. Komen tissue bank with a significant 5-year risk of breast cancer by Gail score. The Gail model is a well-
established tool for assessing breast cancer risk based on demographic and clinical data (bcrisktool.cancer.gov). We test N-
glycosylation alterations, a glucose metabolism-linked post-translational modification, against stressors marital status,
P09.11 A confidence score for distinguishing structural isomers on N-linked and O-linked glycopeptides with ion-
mobility mass spectrometry
Zia Rahman (3), Wendy Sun (1), Xiyue Zhang (1), Hieu Tran (2), Baozhen Shan (1)
(1) Bioinformatics Solutions Inc, Waterloo, Ontario, Canada
(2) University of Waterloo, Waterloo, Ontario, Canada
(3) Bioinformatics Solutions Inc., Waterloo, Ontario, Canada
Introduction: Understanding the impact of glycosylation on protein function requires detailed knowledge of individual glycan
structures and their site-specific distribution. LC-MS is a very sensitive technique that is widely used for studying site-
specific protein glycosylation. Ion-mobility mass spectrometry provides additional structural information, CCS values, for
distinguishing structural isomers. However, very few reports focus on the algorithm to evaluate the glycan structural
confidence for glycopeptide identification with the differentiation of minor changes in glycan structure. In this work, we
proposed an S-Score for a glycopeptide identification from a MS2, indicating structural confidence of a glycan among
different isomers. This S-Score applies both N-linked and O-linked glycopeptide analysis with tandem mass
spectra.
Methods: There are three steps to calculate S-Score. 1. Given a MS2 from a glycopeptides, a protein database and
glycan database, find the candidates of glycans which match the glycan-associates fragment ions. 2. For each glycan
candidate, find the peptide sequence which best fits the peptide fragment ions. A glycan score is associated with each
glycan candidate, including glycan fragment ions, retention time and CCS values. 3. Evaluate all glycopeptide
candidates, and associated a probability score for each candidate. The S-Score is the glycan score gap of the top two
candidates. Output the glycopeptides identification with best probability score, associating the S-score as the structural
confidence. Preliminary data: The algorithm was implemented into PEAKS GlycanFinder, and tested with two large-scale
mass spectrometry datasets from timsTOF instrument. 19315 distinct site-specific intact N-glycopeptides from 2453
glycoproteins in mouse brain tissue were identified. The glycoprotein site-specific microheterogeneity for 6932 sites was
characterized. 2685 O-glycopeptides were identified on a human serum. Among all the identification, the S-Score showed
good accuracy to distinguishing structural isomers. Conclusion: A structural confidence score was proposed for the
glycopeptide identification.
P09.12 Evaluation of Serum/Plasma Sample Preparation Methods for Mass Spectrometry-Based Glycoproteomic
Analysis
Purpose: Efficient and reproducible sample preparation is critical to proteomic and glycoproteomic analysis of body fluids. A
variety of sample preparation protocols and kits are available for proteomic analysis, all based on enzymatic digestion.
However, glycopeptides are more susceptible to missed cleavages and degradation, and little is known about the
performance of these methods for glycoprotein digestion. Here we evaluated different serum/plasma digestion protocols to
show their applicability for targeted glycopeptide quantification. Methods: Pooled human serum and plasma were used to
evaluate the digestion efficiency of glycoproteins. Commercially available products including RapiGest SF surfactant
(Waters), S-Trap (Protifi), and rapid trypsin/LysC (Promega) were evaluated. Samples were digested in-solution or on-
column following protocols in the respective manuals and compared with samples prepared using overnight water bath
incubation and microwave-assisted digestion at optimized temperatures and durations. All experiments were performed on
both serum and plasma with four replicates per condition and the digested samples were analyzed by SDS-PAGE, UHPLC-
QTOF and UHPLC-QQQ to assess the digestion completeness, precision, and undesired modifications. Results: RapiGest
SF surfactant, S-Trap, Rapid trypsin/LysC, and microwave-assisted digestion protocols have the advantage of completing
digestion within 30 to 120 min while achieving sufficient digestion efficiency and precision for most proteomics applications.
However, we found that for targeted glycopeptide analysis each of these approaches introduced certain biases towards
individual classes of glycopeptides. For example, Rapid trypsin/LysC resulted in incomplete digestion of glycopeptides from
certain proteins, e.g. antichymotrypsin. The RapiGest SF surfactant protocol caused degradation of multiply sialylated
glycopeptides due to the high acid concentration required in the quenching step. S-Trap was helpful for reducing
glycopeptide missed cleavages in plasma, but less effective for serum. Conclusion: The results of this study can serve as a
reference for selecting the optimal sample preparation method for glycoproteomic analysis based on digestion efficiency,
time, and cost.
P09.13 A high-throughput C18 clean-up workflow for glycoproteomic analysis of peripheral blood samples
Bo Zhou (1), Diane Tu (1), Khushbu Desai (1), Daniel Serie (1), Klaus Lindpaintner (1), Xin Cong (1)
(1) InterVenn Biosciences, San Francisco, California, United States
Purpose: Clean-up of serum/plasma samples prior to mass spectrometry is essential for robust performance of
(glyco)proteomic analyses. Reverse-phase (RP) resins, particularly C18 matrix, are often used to accomplish this. However,
the hydrophilicity of glycopeptides makes them bind to C18 resins unfavorably. Moreover, manual methods are time-
consuming and labor-intensive. To address these limitations, we developed a high-throughput clean-up workflow suitable for
population-scale glycoproteomic analysis. Methods: For workflow optimization, human serum samples were digested with
trypsin. The resulting peptides were loaded onto AssayMAP C18 cartridges on an Agilent Bravo Platform for sample
cleanup, followed by analysis on a triple quadrupole mass spectrometer in dynamic multiple reaction monitoring (dMRM)
mode. 1,158 MRM transitions were used to quantify 525 glycopeptides and 151 peptides. Raw data was processed using
an in-house developed data processing software. Parameters such as sample loading, sample buffer, equilibration buffer,
loading volume/flow rate, and elution buffer were optimized. Subsequently, the optimized workflow was applied to analyze
plasma samples from 132 melanoma patients. Results: Pooled human serum samples (n=8) were used for workflow
optimization. The median coefficients of variation (CVs) were 12.7% and 8.5% for glycopeptides and peptides, respectively
after C18 clean-up, and 13.1% and 10.2% prior to C18 clean-up. The median recovery of glycopeptides and peptides after
Chronic wound is caused by a wound failing to proceed through a timely reparative order. Millions of people live with chronic
wounds presenting an economic burden to the healthcare system in the US. Understanding lipid and metabolite migration
during the wound healing process will offer insights into how skin responds to a wound and may provide guidance for wound
healing solutions. 3D mass spectrometry imaging (MSI) can determine volumetric molecular distribution, with the mass
spectrum and localization information collected in both XY- and Z-dimensions throughout different skin layers giving a
comprehensive molecular image of the wound-healing process. Most 3D MSI approaches use MALDI or DESI and
reconstruct 3D images from 2D mappings of serial sections, resulting in loss of 3D information and challenges of image
registration. To overcome these limitations, we applied infrared matrix-assisted laser desorption electrospray ionization (IR-
MALDESI), which is a hybrid ambient ionization technique, to sequentially image a sample by consecutive ablation events
for a 3D image generation. In this study, 3D IR-MALDESI MSI is applied to a wounded skin created by biopsy punch, the
molecular profile of which is then compared to the healthy skin in the same nude mouse. Depth resolution was explored with
a confocal laser scanning microscope to establish the imaging parameters for the 3D experiment. A total of 810 and 1269
lipids were putatively detected in the wildtype and wounded skin respectively in a 70-layer 3D experiment. Detailed lipid
profiles across different skin layers were revealed and correlated with histological images to characterize the skin structure
with a lateral resolution of 50 μm and a depth resolution of 10 μm. Current work involves further statistical analysis and
integration of proteomics data, with the aim for multi-omics interpretation of the wound healing process and providing
potential therapeutic targets for chronic wound.
Bacterial Panicle Blight (BPB) is a rice disease caused by Burkholderia glumae that has become a global concern in
reducing rice yields. Currently, there are no effective chemical treatments or resistant cultivars to control BPB. We have
observed that bacterial secreted fractions from the antagonistic microorganism Pseudomonas protegens PBL3 inhibits B.
glumae infections in rice plants. However, the responsible antimicrobial molecules are unknown. Genome comparison
between P. protegens PBL3 and the reference strain P. protegens CHA0, a well-characterized microbial biological control
agent, revealed the presence of 14 known biosynthetic gene clusters (BGCs) in P. protegens PBL3. From this genomic
data, it is likely that P. protegens PBL3 secretes metabolites with activity against BPB that are not produced by P. protegens
CHA0. Cultures of P. protegens PBL3 and the reference strain, P. protegens CHA0, were grown in M9 minimal media
P10.03 Using Covaris AFA for Fast, Reliable, and Automatable Extraction of Analytes from cells for MALDI-TOF
Based Analyses
Tom O'Hare (1), Jay Abramo (1), DEBADEEP BHATTACHARYYA (1)
(1) Covaris, Woburn, Massachusetts, United States
Comprehensive proteomic and metabolomic profile can help in understanding the nature of a disease to developing
therapeutics to monitoring patients and determining their therapeutic regime. Regardless of the detection technology,
extraction of relevant biomolecules from cells can be laborious and complex. Typical approaches such as, bead beating,
tissue homogenizers, etc. require high temperature or solvents requiring long time to stabilize. Adaptive Focused
Acoustics(R) technology (AFA(R)) enables reliable extraction of biomolecules under isothermal conditions and can isolate
metabolites and proteins from up to 96 different samples in ~ 2 mins. This study showcases AFA based sample
preparation for rapid extraction of proteins and metabolites from a variety of microbes. Covaris truXTRAC Extraction Solvent
was used for suspension of cells in a 96 AFA-TUBE TPX Plate, where lysis was performed using AFA followed by elution
and MALDI-TOF analysis in positive ion linear mode with a nitrogen laser in a randomized pattern. The number of cells per
well of the plate ranged from 1 x 107 cells for yeast to 1 x 105 cells for bacteria. Manual analysis of MS data using E. Coli
Metabolome Database and UniProt accurately identified several proteins (50S ribosomal protein, subunit of DNA
polymerase, surface membrane lipopolysaccharides) and metabolites critical for cell growth. Comparison of MS data of
samples extracted from E. Coli grown in LB and M9 minimal media shows several similar signals. However, several signals
in the lower mass range (~1-5 kDa) was observed only in LB media highlighting production of bacteria under stressful
conditions. This study demonstrates reproducible and robust MALDI-TOF MS data after AFA enabled cell lysis of cultured
cells with minimal variability, and can support high-throughput requirements. This robust method can extract sufficient
sample from bacteria with a 2-second pulse or from yeast in 1 minute.
P10.04 Leveraging Public Untargeted Metabolomics Data To Propagate Structurally Related Molecule Annotations
to Millions of MS/MS Spectra
Wout Bittremieux (1), Sydney P. Thomas (1), Nicole Avalon (1), Mingxun Wang (1), Pieter C. Dorrestein (1)
(1) UC San Diego, La Jolla, California, United States
One of the key goals of untargeted tandem mass spectrometry (MS/MS) metabolomics is discovering biologically relevant
molecules. Currently there is still a lot of discovery potential: using spectral library searching, on average only ~5% of the
P10.05 Glycan and Collagen Imaging Mass Spectrometry Analysis of a Broad Tumor Spectrum of Prostate Cancer
Tissue Histopathologies
Jordan Hartig (1), Peggi Angel (1), Lydia Liu (2), Amanda Khoo (2), Stan Liu (3), Michelle Downs (3), Paul Boutros (4),
Thomas Kislinger (5), Richard Drake (1)
(1) Medical University of South Carolina, Charleston, South Carolina, United States
(2) University of Toronto Princess Margaret Cancer Centre, Toronto, Ontario, Canada
(3) Sunnybrook Health Science Centre, Toronto, Ontario, Canada
(4) University of California, Los Angeles, Los Angeles, California, United States
(5) University of Toronto Princess Margaret Cancer Centre, Toronto, Ontario
Prostate cancer progression can be characterized at the tissue level by changes in glycosylation and the emergence of
reactive stroma, which represent potential target areas for cancer biomarker discovery. Using established imaging mass
spectrometry (IMS) methods designed to target N-glycans and collagen/extracellular matrix (ECM), a cohort of pathology-
annotated FFPE prostate tissues linked to a multi-omic and mult-investigator consortia were evaluated. Tissues represented
a broad spectrum of prostate cancer histopathologies, with the goal of identifying specific glycan and collagen biomarker
candidates for integration with other omic data sets from the same cohort. analyses. Each slide was processed for N-glycan
IMS by peptide N-glycosidase F (PNGase F PRIME) digestion using established protocols. Released N-glycans were
detected by MALDI-QTOF IMS on a timsTOF flex mass spectrometer. Following enzyme digestion and imaging, slides were
stained using established H&E staining protocols. High resolution images of the stained tissues were saved. Collagenase
digestion and further MALDI-QTOF detection of ECM peptides was performed following H&E staining. Each tissue sample
contained different combinations of benign tumors, fused tumors, cribriform tumors, glomeruloid bodies, poorly formed
glands (PFG), and intraductal carcinomas, all with a Gleason score of at least 3, 4, or 5 When comparing H&E images to
glycan profiles, it was found that the major tumor-associated glycans were in the high mannose and paucimannose
P10.06 Improved Spatial Resolution of Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-
MALDESI) Mass Spectrometry Imaging (MSI)
Alena Joignant (1), Hellena Bai (1), David Muddiman (1)
(1) North Carolina State University, Raleigh, North Carolina, United States
Mass spectrometry imaging (MSI) is a powerful method for mapping chemical information to tissue structures. Infrared
matrix-assisted laser desorption electrospray ionization (IR-MALDESI) is a hybrid ionization method that regularly achieves
120 microns in its current configuration, yet definition of minute structures is limited to the spatial resolution capability of the
optical train. Spatial resolution of this method was previously increased with oversampling, where ablation spots are
overlapped to achieve cellular and sub-cellular ablation areas1. A different laser and optic configuration was previously
shown to decrease the available spatial resolution to 50 microns, which was nearly diffraction-limited2. We propose a new
optical train involving a reflective, Schwarzschild-like optic that gets us closer to a diffraction-limited spot size and improves
IR-MALDESI spatial resolution without the need for oversampling. By avoiding oversampling, the sample does not need to
be completely ablated in order to achieve high spatial resolution, which is advantageous for the use of thicker samples.
Additionally, decreasing the spot size lowers the IR laser energy output required for sufficient sample ablation, which allows
for increased speed of analysis. A decrease in spot size scales the spatial resolution to that of optical imaging, allowing for
more direct multi-modal imaging. The optical path and laser profile are optimized as examined on thermal paper by
microscopy. Although a decrease in spot area is logistically challenging due to increased analysis time and is limited by the
sensitivity of the instrument, the information received by such fine combing of tissue structure has positive implications for
cellular analysis, such as imaging islet cells in pancreas tissue and similar structures. References: 1. Anal. Bioanal. Chem.
2015, 407, 2265-2271. 2. J. Am. Soc. Mass. Spectrom. 2017, 28, 2099–2107.
P10.07 Novel Matrix Strategies for Improved Ionization and Spatial Resolution for IR-MALDESI Mass Spectrometry
Imaging
Russell Kibbe (1), Allyson Mellinger (1), David Muddiman (1)
(1) North Carolina State University, Raleigh, North Carolina, United States
Infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) utilizes energy-absorbing matrices that
surround the analyte and desorb molecules from the surface through infrared laser ablation for post-ionization and
detection. A matrix with high infrared absorbance is important to maximize laser efficiency and analyte desorption. The
deposited ice layer matrix is fundamental to IR-MALDESI and mass spectrometry imaging (MSI) because of the prevalence
of water in biological systems and the absorbance of IR wavelengths in the O-H stretch. In this effort, we designed new
P10.08 A Novel Ion-beam Imaging Technique for Detection of Chromosomal Copy Number Alterations Associated
with Progression of Breast Cancer
Rashmi Kumar (1), Sushama Varma (2), Mike Angelo (1), Robert West (1)
(1) Stanford University School of Medicine, Stanford, California, United States
(2) Stanford University, Stanford, California, United States
Purpose:Ductal Carcinoma in situ (DCIS) comprises about 25% of breast cancers detected annually. It is hard to predict
which DCIS cases will progress to invasive breast cancer. Recent studies have demonstrated that certain genomic copy-
number-alterations (CNAs) are associated with higher risk of DCIS progression. In this project, we aim to develop a mass-
spectrometric imaging method to detect and spatially map CNAs in FFPE tissues. Using this novel technique, we plan to
investigate the role of tumor genomics in DCIS progression. Methods:FFPE sections were deparaffinized and prepared for
in situ hybridization (ISH) assay. Haptens were incorporated into the genomic targets. Next the tissues were incubated with
metal-tagged antibodies, specific to haptens. Multiplexed Ion Beam Imaging (MIBI) technique was used for analysis. MIBI is
a novel imaging technique, combining immunohistochemistry with time-of-flight (TOF) mass-spectrometry. An ion-beam gun
rastered on the tissue-sections containing slides under vacuum. This freed the metal ions from specimen to be measured by
TOF. The abundance of each metal was read out in pixel-by-pixel fashion which corresponded to the abundance of that
antibody Results:HER2, an important chromosomal ROIs in breast cancer was chosen as the target for all the
experiments. Bacterial artificial chromosomes containing chromosomal region of interest were used as DNA-probes and
nick translation was carried out to introduce haptens. The probes were then allowed to hybridize overnight. After
hybridization, the tissue sections were incubated with metal-tagged anti-hapten antibodies followed by imaging using
MIBI. We observed significant correlation in spatial map of HER2 between immunofluorescence and MIBI images. A pair of
HER2 +signal in each cell of lymph-node and placenta control tissues was observed. We observed amplified HER2 signal in
HER2-positive breast tissues whereas HER2 signal was depleted in HER2-negative breast tissues. Conclusions:A new
technique for spatially mapping genomic-ROIs using MIBI has been developed.
Prior to the analysis of biological samples, one must first assess the suitability of the analytical instrumentation. In this
regard, the metabolomics community requires an everyday system suitability standard by which to benchmark instrument
performance for untargeted MS based approaches. A complex solution that mimics biological samples is needed to
P10.10 Spatial Metabolomics of Xenopus laevis Tadpoles by IR-MALDESI Mass Spectrometry Imaging
Whitney Stutts (1), Julia Grzymkowsi (2), Nanette Nascone-Yoder (2)
(1) Molecular Education, Technology and Research Innovation Center (METRIC) and The Department of Molecular and
Structural Biochemistry, North Carolina State University, Raleigh, North Carolina, United States
(2) Department of Molecular Biomedical Sciences, College of Veterinary Medicine, North Carolina State University, Raleigh,
North Carolina, United States
Xenopus laevis are an important model organism for studying vertebrate development and disease. At the tadpole stage,
functional organs have formed and metabolic perturbations resulting from experimental interventions can be visualized. In
this work, infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) mass spectrometry imaging (MSI)
was employed to spatially resolve tissue-specific molecular markers in frontal, whole organism sections from Xenopus
laevis tadpoles. High-resolution accurate mass measurements were acquired with a resolving power of 240,000 (FWHM
m/z 200) in positive and negative ionization modes to enhance metabolome coverage. Resulting mass spectra were
converted to imzML files and uploaded to METASPACE, a web-based platform for metabolite annotation of MSI data. Over
1000 metabolites and lipids were annotated at an FDR of ≤10% after filtering for on-sample only ions. Highly localized
metabolites including amino acids, fatty acyls, glycerolipids, glycerophospholipids, sphingolipids and sterol lipids were
detected in distinct tissue regions such as the eyes, facial cartilage, including the branchial arches and Meckel’s cartilage,
intestines and stomach. In addition to metabolic profiling, this work presents the application of an optimized embedding and
cryosectioning method for delicate tissues such as the Xenopus gut. The results described demonstrate the utility of IR-
MALDESI MSI for studying the Xenopus metabolome and illustrate the potential for novel insights in developmental biology
and toxicology studies.
P10.11 Investigation of Sucrose as a Cryoprotectant for Tissue Preservation and Compatibility with IR-MALDESI
Mass Spectrometry Imaging
Formalin-fixed paraffin embedded (FFPE) and flash-frozen tissues are methods previously used to preserve biological
tissue samples prior to analysis with infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). FFPE
tissues have been used primarily for glycan research as the tissues are more conducive to IR-MALDESI MSI when
compared to fresh or flash-frozen tissues. FFPE tissues require sample preparation due to formalin fixation and paraffin wax
embedding. Other methods of tissue preservation such as sucrose embedding are being investigated as an alternative to
FFPE and flash frozen tissues. Sucrose embedding is a specialized version of fresh or flash-frozen tissue preparation using
sucrose as a cryoprotectant for samples in order to protect the cell structure morphology during the freezing process.
Sucrose embedding tissues for metabolite, lipid, and/or glycan imaging could yield an alternative method – results will
compare molecular classes and their ion abundance relative to fresh frozen tissues. In this study, two female mice brains
underwent separate preservation methods where one was flash frozen sample without sucrose embedding and the other
embedded in sucrose embedding prior to freezing. Tissues were sectioned and evaluated using IR-MALDESI mass
spectrometry imaging to determine the optimal tissue fixation method using ion abundance as a function of molecular class
(polar metabolites, non-polar metabolites and N-linked glycans).
P10.12 Enhancing Metabolomic Coverage in Positive Ionization Mode Using Dicationic Reagents by IR-MALDESI
MSI
Ying Xi (1), David Muddiman (1)
(1) NCSU, Raleigh, North Carolina, United States
Mass spectrometry imaging is a powerful tool to analyze a large number of metabolites with their spatial coordinates
collected throughout the sample. However, the significant differences in ionization efficiency pose a big challenge to
metabolomic mass spectrometry imaging. To solve the challenge and obtain a complete data profile, researchers typically
perform experiments in both positive and negative ionization modes, which is time-consuming. In this work, we evaluated
the use of the dicaionic reagent, 1,5-pentanediyl-bis(1-butylpyrrolidinium) difluoride (abbreviated to [C5(bpyr)2]F2) to detect a
broad range of metabolites in the positive ionization mode by infrared matrix-assisted laser desorption electrospray
ionization mass spectrometry imaging (IR-MALDESI MSI). [C5(bpyr)2]F2 at 10 µM was doped in 50% MeOH/H2O (v/v)
electrospray solvent to form +1 charged adducted ions with anionic species (-1 charged) through post-electrospray
ionization. This method was demonstrated on sectioned rat liver and hen ovary. 73 deprotonated metabolites from rat liver
tissue sections were successfully detected and putatively identified in the positive ionization polarity through adduct
formation with the dicationic reagent, along with 164 positively charged metabolite ions commonly seen in positive ionization
mode, which resulted in 44% increased molecular coverage. In addition, we were able to generate images of hen ovary
sections showing their morphological features. Following-up tandem mass spectrometry (MS/MS) indicated that this
dicationic reagent [C5(bpyr)2]2+ could bond with the headgroup of lipid ions, implying a possible mechanism behind the
adduct formation. The application of the dicationic reagent [C5(bpyr)2]2+ on tissue sections provides a rapid and effective way
to enhance the detection of metabolites in positive ionization mode.
Extracellular vesicles (EVs) are cell-derived nanoparticles that can serve as carriers of DNA, RNA, and protein. For the
pathogen S. pneumoniae, which causes diseases such as pneumonia and meningitis, EVs can be used to transport these
macromolecules to recipient cells, facilitating communication and signaling. However, the overall composition of these EVs
and how the macromolecules packaged within these EVs contribute to the pathogenicity of S. pneumoniaeis is still
unknown. In this work, we have developed a novel multi-omic workflow that isolates DNA, RNA and protein from a single S.
pneumoniae EV starting sample, enabling analysis of these macromolecules without the need for multiple sample
preparation steps and allowing us to gain insights into the composition and functionality of S. pneumoniae EVs. EVs from a
S. pneumoniae culture were isolated and the proteins in the EV homogenate were labeled with the bifunctional ProMTag.
One end of the ProMTag forms a reversible, covalent bond with proteins. ProMTag’s other functional group is
methyltetrazine, which forms an irreversible bond with trans-Cyclooctene (TCO), allowing capture of EV proteins on TCO-
agarose beads. Organic solvent was then added to precipitate the nucleic acids in the EV homogenate. Nucleic acids were
released by a series of washes and subsequently separated into DNA and RNA fractions by RNase or DNase treatment,
respectively. Proteins were then released from the TCO-agarose beads by reversing the ProMTag-protein linkage, yielding
proteins in their original, unmodified state ready for analysis. DNA and RNA were sequenced, and proteins were identified
using mass spectrometry. Using this workflow, we were able to describe the composition of multiple macromolecules within
S. pneumoniae EVs and gain insights into how these macromolecules facilitate communication and infection within the S.
pneumoniae population.
P11.02 Microbiome multi-omics analysis of human whole stool samples utilizing a single Bligh-Dyer sample
preparation method
Stephanie Servetas (1), Tracey Schock (2)
(1) NIST, Gaithersburg, Maryland, United States
(2) NIST, Charleston, South Carolina, United States
The microbiome is a critical component of an individual’s health. The structure (bacterial communities) and function
(metabolic activity) of one’s microbiome can elicit positive and negative physiological mechanisms either leading to a robust
health state or incite a foundation for the development of disease. Microbiome composition can be readily determined using
16S rRNA gene sequencing or whole genome sequencing from stool. Metagenomic and metatranscriptomic sequencing
technologies can illuminate microbes’ functional capacities and states while additional measurements are needed to
elucidate the microbiome-host interactions. Stool proteomic, metabolomic, and lipidomic, measurements offer the ability to
simultaneously assess the complex interplay between the microbiota, host, and environment. In this study, genomic,
proteomic, metabolomic, and lipidomic analyses were conducted on reference stool material as part of the National Institute
of Standards and Technology (NIST) Whole Stool Interlaboratory Study (ILS). The primary goals for the ILS were to: 1)
P11.03 Proteogenomic approach to studying the functional effect of cellular aneuploidy in cancer
Lizabeth Katsnelson (1), Xin Zhao (1), Pan Cheng (1), Teresa Davoli (1), David Fenyö (1)
(1) NYU School of Medicine, New York, New York, United States
Aneuploidy, a hallmark of cancer, is the presence of an abnormal number of chromosomes. While rare and typically
detrimental to cellular fitness in normal cells, aneuploidy occurs in ~90% of solid tumors in the form of somatic copy number
alterations (SCNAs). It is believed that aneuploidy arises from defective regulation during mitosis, which leads to
chromosomal instability. However, its role in tumorigenesis remains unknown. Tumors with high levels of aneuploidy (> 8-10
gains or losses) correlate with poor prognosis, increased immune evasion (decreased immune infiltration), and lowered
response to immunotherapy. This suggests that aneuploidy may be both a predictive and prognostic biomarker, as well as a
therapeutic target for cancer. Since proteins perform the majority of essential cellular tasks and are the targets of most
drugs, it is crucial to study the downstream effects of aneuploidy at the proteomic level. Using publicly available data from
the Clinical Proteomic Tumor Analysis Consortium (CPTAC) and the Cancer Cell Line Encyclopedia (CCLE), we reveal
differences in the cancer proteome among low and high aneuploid tumors and cell lines across several cancer types.
Upregulated pathways in highly aneuploid tumors include translation and protein regulation, mitochondrial regulation, and
cellular respiration and metabolism, while downregulated pathways are enriched for immune responses. Utilizing multi-
omics data, we also compare immune profiles of low and high aneuploid tumors, which we may then compare to proteomics
data to find possible mechanisms of immune evasion in aneuploid tumors. Together, these findings may uncover
therapeutic vulnerabilities in highly aneuploid, aggressive cancers.
P11.04 Spatially Targeted Proteomics and Multimodal Imaging Distinguish Molecular Signatures of Amyloid
Plaque-Containing Islets from Type 2 Diabetic Donors
Angela R. S. Kruse (1), Josiah C. McMillen (1), John T. Walker (2), Audra M. Judd (1), Nathan Heath Patterson (1), Danielle
B. Gutierrez (1), Jeremy L. Norris (1), Jeffrey M. Spraggins (3), Alvin C. Powers (4), Richard M. Caprioli (3)
(1) Vanderbilt University, Nashville, Tennessee, United States
(2) Vanderbilt University School of Medicine, Nashville, Tennessee, United States
The pancreas is a complex organ with a crucial role in digestion and blood glucose regulation. Approximately 2% of the
pancreas consists of islets, which are groups of hormone-producing cells. The islets of individuals with type 2 diabetes
(T2D) are dysfunctional and may undergo cell death. Amyloid plaques contribute to islet cell death but the molecular
environment supporting plaque formation remains poorly understood. Since amyloid plaques vary in morphology and
location, comparing molecular signatures of islets with and without plaques is essential. To accomplish this, we developed a
multimodal platform combining microscopy, micro-Liquid Extraction Surface Analysis (microLESA), and imaging mass
spectrometry (IMS) to study individual islets in pancreas tissue from non-diabetic and T2D human donors. Autofluorescence
and polarized light microscopy were used to localize islets and amyloid plaques. Microscopy approaches were integrated
with microLESA, in which trypsin droplets were targeted to individual islets for on-tissue digestion. Liquid chromatography-
tandem mass spectrometry was performed on digested islet proteins to generate chemically rich data from non-diabetic and
T2D islets. Additionally, matrix-assisted laser desorption/ionization-IMS was performed to spatially analyze peptides from
non-diabetic and T2D pancreas tissue sections. Differential expression analysis of amyloid-containing versus non-amyloid-
containing islets in T2D pancreata identified 15 proteins associated with amyloid plaques, including collagen IV and
myristolated alanine-rich C-kinase substrate protein. Further comparison of all islets from non-diabetic and T2D donors
identified 85 differentially abundant proteins including peroxiredoxin, carboxypeptidase, and calmodulin. These results
suggest that specific enzymes and chaperones are associated with amyloid plaques, potentially as a stress response or an
aberrant signal involved in plaque formation. Additionally, peptide IMS showed changes in protein abundance consistent
with microLESA data, but also identified peptides differing in localization but not abundance. Together, these findings
underscore the importance of a multimodal approach to begin unraveling the molecular basis for amyloid plaque formation
in T2D.
Purpose: Multi-omics approaches have emerged to understand disease pathology and drug resistance, probing both the
proteome and metabolome of the sample. We herein present a rapid precipitation-based approach to selectively fractionate
cellular components, isolating intact proteins, the low-mass peptidome and with opportunity for metabolome profiling by MS.
Methods: A test sample of S. cerevisiae lysate was subject to sequential precipitation from 20, 40, 60, 80, and 97% acetone
with zinc sulfate. Recovered protein fractions were quantified and visualized by SDS PAGE, and subject to bottom-up LC-
MS/MS. A low-molecular weight sample was prepared by trypsin digestion of yeast. Peptides were precipitated with 97%
acetone and zinc sulfate in the ProTrap XG. Pellets were re-solubilized quantified and characterized by bottom-up LC-
MS/MS. Results: Intact protein pellets recovered from 2 min precipitation in 80% acetone revealed > 95% yield. Bottom-up
LC-MS/MS characterization of the pellet and supernatant demonstrated a bias towards the rapid recovery of high molecular
weight species (p < 0.001). Recovery of low molecular weight peptides revealed 90 ± 10% precipitation yield in 97%
acetone and zinc sulfate (LC-MS/MS confirmed a median molecular weight of 2.25 kDa). SDS PAGE analysis of the
sequentially precipitated sample revealed a molecular weight trend in the fractions collected from increasing acetone
content. Future experiments will apply a sequential precipitation approach to isolate the proteome, and peptidome from a
Introduction: In data dependent acquisition workflows, collision induced dissociation (CID) is commonly used to fragment
peptide ions for analysis. While CID enables identification of many high-confidence peptides and proteins, preserving labile
post-translational modifications (PTMs) can be challenging. Electron activated dissociation (EAD) is a fragmentation mode
that involves capture of electrons by molecular ions to form radicals that dissociate into ions. EAD often preserves PTMs
and provides complimentary sequence information to CID. This work focuses on information dependent acquisition (IDA)
method development and analysis of HeLa digest using EAD fragmentation to localize peptide phosphorylation sites.
Methods: 100 µg of digested HeLa cell lysate was fractionated using high pH RP-HPLC (43 fractions). Fractions were
separated using a 20 min microflow gradient with a flow rate of 6 µL/min. Zeno EAD IDA was performed using the ZenoTOF
7600 system, with EAD reaction time, electron beam current and electron kinetic energy optimized for analysis. The Zeno
trap was activated for increased MS/MS sensitivity. MASCOT was used for all database searching and search results were
imported into Scaffold for analysis. Results: For Zeno EAD IDA, several method parameters were optimized for data
acquisition including EAD reaction time, electron beam current and electron kinetic energy. Analysis of 43 high pH fractions
of HeLa digest resulted in 5825 protein identifications and 36,588 peptide identifications, including 135 phosphopeptides.
EAD spectra from phosphopeptides identified the sequence and site of phosphorylation using complete z+1 ion series,
including localization in peptides with multiple serines. Conclusions: Zeno EAD IDA analysis of 43 HeLa digest fractions
using microflow chromatography enabled large scale protein identification, with over 5000 proteins identified and confident
identification and site assignment of peptide phosphorylation sites.
The dynamic, stress and nutrient-sensitive modification of protein Ser/Thr residues with B-N-acetylglucosamine (O-GlcNAc)
impacts multiple regulatory processes enhancing proliferation, survival, and metastasis of cancer cells. While global O-
GlcNAc modification and O-GlcNAc transferase (OGT) expression is elevated in a number of cancers including breast
cancer, characterization of the sites of O-GlcNAc modification has been challenging due to the difficulty in efficiently
enriching O-GlcNAcylated peptides and the labile nature of the O-linkage by collisional dissociation during tandem mass
spectrometry. To uncover mechanisms regulated by O-GlcNAc cycling enzymes, immunoaffinity enrichment was performed
using a newly available anti-O-GlcNAc antibody (Cell Signaling) prior to LC-MS/MS. O-GlcNAcylated peptides were
immunoprecipitated from trypsin digested MDA-MB-231 cell lysates following high pH RP fractionation and
P12.03 Discovery of the Post-Translational Modification Mechanisms of the Antibacterial Peptide TCLE, to Lead to
the Creation of Novel Antibiotics
S. Anisha Daley (1), Diana Calvopina (1), Kimberly Wagstaff (1), Katherine Brown (1), Joel Griffitts (1), John C. Price (1)
(1) Brigham Young University, Provo, Utah, United States
Bacteria are evolving, and in order to keep ourselves from being left behind, we need to improve our antibiotics as well. The
TCLE peptide is a bacteria-produced antibiotic that is heavily modified after translation. By learning about the modification
processes, we will be able to purposefully alter it to create novel anti-bacterial compounds effective against many diverse
bacteria. The TCLE peptide is made up of two parts: the core portion and the leader portion. The core portion is modified by
several protein complexes, and the leader portion is where enzyme complexes attach while modifying the core. The core
has six cysteines, and the first post-translational modification changes those cysteines into thiazoline intermediates then
thiazoles. This is done by a protein complex called TCLIJN. TCLIJN grabs on to the leader portion of the peptide and uses it
to find the core and modify the cysteines. The objective of this study is to learn how much of the leader peptide TCLIJN
needs to modify the core cysteines. To discover this, we truncated amino acids off the leader portion, starting at 6 and going
up to 18 by intervals of three. After co-expressing the peptide with TCLIJN, we then measured how many cysteine to
thiazole modifications occurred by using mass spectrometry. The data suggested that at truncations of 15 and beyond,
complete cysteine modification is significantly decreased, causing the number of thiazolines and thiazoles to increase. This
data explains how much of the leader peptide is essential for TCLIJN’s modifications. As we continue to discover the
characteristics of the TCLE peptide, we will learn how to more purposefully modify this peptide to create the desired
products. Understanding the TCLE peptide and all its facets will pave the way to greater and more purposeful creation of
antibiotics.
The post-translational modification (PTM) N-ε-lysine acylation has gained attention due to its impact in regulating diverse
biological mechanisms. Lysine acylation is tightly linked to metabolism as the production of intrinsically reactive metabolites,
P12.05 Compared dda-PASEF and prm-PASEF approaches for quantification of 2000 RAS induced
Phosphopeptides.
Elizabeth GORDON (1), Matthew Willetts (1), Diego Assis (1), Rodrigo M Ferreira (2), Uma K. Aryal (2)
(1) Bruker Daltonics Inc, BIllerica, Massachusetts, United States
(2) Department of Comparative Pathobiology, Purdue University, West Lafayette, Illinois, United States
Introduction Understanding molecular factors and biological processes of cellular senescence provides important insights
into the intrinsic cellular mechanisms for cancer prevention, and organismal aging. Herein, we have used dda-PASEF and
prm-PASEF for quantification of RAS induced phospho-peptides, our analysis is added by the ability of the trapped ion
mobility to give greater separation to phospho position specific isomers. Methods Human diploid fibroblast strain IMR90
(CCL-186; ATCC, USA) cells were transduced with ER:RAS lentivirus and treated with 100 nM of (Z)-4-Hydroxytamoxifen
(4-OHT) for ER:RAS activation and induction of oncogene induced senescence (OISCells were harvested after 6 days of 4-
OHT activation for nuclear extraction, trypsinization, and enrichment of phosphopeptides using Polymer-based Metal-ion
Affinity Capture (PolyMAC) spin tips. Samples were run on a nanoElute LC (Bruker Daltonics) using a 25 cm Aurora nano
column (IonOpticks, Australia) at 400 nl/min with a 70 min gradient, and a longer 120min gradient for global samples. LC-
TIMS MS/MS data were obtained from a timsTOF Pro instrument. dda-PASEF data were analyzed using PEAKS OnLine
(Bioinformatics Solutions). PRM PASEF data was analyzed with Skyline. Results Global proteome analysis identified 55,000
peptides and 4500 proteins, but less than 1% total identifications came from phosphopeptides. Analysis of 400ng of
PolyMAC enriched IMR90 cell digest identified around 20,000 unique phosphopeptides and over 2800
phosphoproteins. This study shows over 1,200 protein groups added quantifiable changes in phosphorylation state. Unlike
dda-PASEF, targeted prm-PASEF of selected phosphopeptides, showed that phosphorylation could be seen with RAS
induction in both enriched and unenriched samples. Conclusions In IMR90 cells induction of the Estrogen receptor RAS
fusion shows that 3,984 peptides were observed to be phosphorylated using dda-PASEF and that they then could be
targeted with PRM-PASEF and quantified in a label free approach.
P12.06 Abscisic Acid Controlled Redox Proteome of Arabidopsis and Its Regulation by Heterotrimeric G-proteins
Amanda Smythers (1), Nikita Bhatnagar (2), Chien Van Ha (2), Evan McConnell (1), Boominathan Mohanasundaram (2),
Sona Pandey (2), Leslie Hicks (1)
(1) University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, United States
(2) Donald Danforth Plant Science Center, St. Louis, Missouri, United States
Reactive oxygen species (ROS) are highly efficient, membrane permeable cell signals that are essential for cellular
regulation. These signals can mark macromolecules for degradation, act as a retrograde signal for transcriptional regulation,
as well as directly modulate enzymatic activity via the reversible oxidation of cysteine thiols. Although this reversible
cysteine oxidation is an essential regulatory mechanism, the resulting modifications are labile and infrequent compared to
the global proteome, making them challenging to maintain ex vivo and detect reproducibly. In our lab, we have optimized an
approach for oxidized resin assisted capture (OxRAC) that enables thorough, reproducible analysis of the full reversible
cysteine redoxome. By employing iodoacetamide as a blocking reagent during cell lysis, we ensure that cysteines reduced
in vivo are excluded before non-specifically reducing reversibly oxidized cysteines for enrichment and analysis via label free
quantitative LC-MS/MS. In this work, our optimized OxRAC platform was implemented to probe the dependence of
heterotrimeric G-proteins on reversible cysteine oxidation for the mediation of abscisic acid (ABA) signaling pathways in
Arabidopsis thaliana. By cross comparing the wildtype Col0 plants with the G-beta protein (AGB1) null mutant, agb1, we
were able to quantify 6,891 unique oxidized cysteine-containing peptides and reveal 923 significant changes in oxidation
following ABA treatment. Divergent pathways, including primary metabolism, ROS response, translation, and
photosynthesis, exhibited both ABA and G-protein dependent redox changes, many of which occurred on proteins not
previously linked to either ABA or G-proteins. Together, these data uncover a complex network of reversible oxidations that
allow ABA and G-proteins to rapidly adjust cellular signaling to adapt to changing environments and suggest that a
functional G-protein complex is required to maintain intracellular redox homeostasis and fully execute plant stress
responses.
P12.07 High-throughput protein modification quantitation analysis using intact protein MRM and its application on
hENGase inhibitor screening
Dingyin Tao (1), Miao Xu (1), Atena Farkhondeh (1), Andrew Burns (1), Steven Rodems (2), Matthew Might (3), Wei Zheng
(1), Christopher LeClair (4)
(1) National Center for Advancing Translational Sciences, National Institutes of Health, Rockville, Maryland, United States
(2) Travere Therapeutics, San Diego, California, United States
(3) University of Alabama at Birmingham, Birmingham, Alabama, United States
(4) National Center for Advancing Translational Sciences, National Institutes of Health, Rockville, Maryland
The emerging demand for protein quantitation has been increasing in multiple fields, which includes protein biomarker
screening and validation, protein therapeutics (antibody and antibody-drug conjugation), and drug screening and
development. Currently, there remains a large gap for performing high-throughput protein quantitation at the intact protein
level using label-free methods. In our effort to develop such a high-throughput protocol, we selected ribonuclease B (RNB),
Understanding the complexity of the mammalian brain has been a long-standing challenge, not least because of the many
specialized proteins and assemblies required for proper neuron function and connectivity. In particular, protein-protein
interactions (PPIs) are critical to the proper functioning of neural proteins, and neurons from different mammals share many
stable macromolecular protein complexes that perform conserved cellular functions. However, such data are only partially
known due to the challenge of defining PPIs at high throughput directly from mammalian brains. In this project, we
determined neuronal PPIs and “protein neighborhoods” at large scale by applying a tag-less proteomics technique called
co-fractionation/mass spectrometry (CF-MS) to primary brain tissues of several mammals, including humans. CF-MS
involves non-denaturing extraction of endogenous proteins in their native assembly states, followed by chromatographic
separation into biochemical fractions, then mass spectrometry to define co-eluting proteins. Given sufficient independent
separations and the appropriate statistical framework, stable protein complexes can be reliably identified based on protein
co-elution patterns. We will describe our progress in combining this integrative proteomic approach with available brain PPI
data and information from sequence orthology across mammals in order to define a large-scale map of conserved protein
complexes in the mammalian brain.
P13.02 System Wide Profiling of Protein Interaction Dynamics Links Host Innate Immunity and DNA Damage
Responses
Joshua L. Justice (1), Josiah E. Hutton (1), Michelle A. Kennedy (1), Brett Phelan (1), Bokai Song (1), Dawei Liu (1), Ileana
M. Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
Protein-protein interactions (PPIs) underly the cellular capacity to sense viral infection and subsequently mount an immune
response. Viruses have acquired mechanisms to counteract cellular immunity by suppressing, subverting, or hijacking the
very interactions that drive the antiviral program. This interplay between infection and host defense occurs at multiple sub-
cellular arenas through at the interface of dynamically shifting and interchanging virus-virus, virus-host, and host-host
PPIs—the scope of which has not yet been fully captured. Using Thermal Proteome Profiling coupled to Mass Spectrometry
(TPP-MS), we constructed a cell-wide portrait of the PPI dynamics that coordinate the innate immune response to Herpes
Simplex Virus 1 (HSV-1) infection in primary fibroblasts. We found that IFI16, a nuclear DNA sensor that serves as a central
platform for HSV-1 immune responses, coordinates with the DNA damage response (DDR) by activating the master DDR
kinase, DNA-PK. Microscopy and molecular virology assays revealed that IFI16 recruits DNA-PK to the genome at the
nuclear periphery. DNA-PK was necessary to suppress viral spread and mount an inflammatory antiviral response. The
discovery of DNA-PK as a major contributor to the nuclear immune response further underscored a need to characterize its
kinase-substrate network. Coupling phospho-peptide enrichment with DNA-PK chemical inhibition revealed hundreds of
DNA-PK dependent modifications after both viral infection and DNA damage. Integrating the phospho-peptide and TPP-MS
P13.03 Thermal proximity coaggregation mass spectrometry reveals shared and distinct rewiring of interactomes
in alpha, beta, and gamma herpesvirus infection
Tavis Reed (1), Olga Troyanskaya (1), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
Determining protein-protein interactions (PPIs) and their dynamics is key to uncovering the function and regulation of
proteins in both homeostatic and perturbed states. Viruses rely on temporal remodeling of PPIs, including virus-virus, virus-
host, and host-host PPIs, in order to successfully replicate within host cells. Defining dynamic interactomes at a system
level is key to understanding the biology of viral infections and identifying possible targets for therapeutic intervention. An
unbiased method for global profiling of dynamic changes in interactions is Thermal Proximity Coaggregation (TPCA) mass
spectrometry. TPCA relies on the principle that interacting proteins co-aggregate under temperature-induced stress,
resulting in overlapping aggregation curves. Analysis of the distance between curves, their melting temperatures, and their
clustering are used to predict PPIs. Previously, our lab has leveraged TPCA to study the nuclear-replicating human
cytomegalovirus (HCMV; a beta-herpesvirus) and herpes simplex virus type I (HSV-1; an alpha-herpesvirus). Here, we 1)
optimize the TPCA sample preparation and 2) apply this to a model gamma-herpesvirus, Kaposi’s Sarcoma-associated
herpesvirus (KSHV). First, we performed comparative experimental and computational analysis of a multitude of different
cell lysis conditions and set of denaturation temperatures, identifying a workflow that improves the coverage of proteins from
different functional and subcellular localization classes and the prediction of PPIs. Next, we used this workflow to perform
TPCA during the reactivation of KSHV in a latency model. Integration of the KSHV dataset with our previous HCMV and
HSV-1 TPCA datasets revealed conserved PPI rewiring events involving both host-host and viral-host PPIs. We show that
known homologs of viral proteins share similar interactions and temporal interaction dependence, highlighting the promise
of these comparative analyses to reveal functions and properties of poorly characterized viral proteins. Altogether, this work
helps to guide TPCA experimental designs and reveals temporal virus-host interactions critical for the replication of these
ancient viruses.
Erythrocytes (red blood cells; RBCs) are the simplest primary human cells, lacking nuclei and all major organelles. Despite
their simplicity, RBCs dynamically change cellular morphology and physiology in response to pathophysiological cues,
mediated in the absence of gene expression. While hemoglobin accounts for 98% of expressed RBC proteins, the full
proteome is predicted to include >1,000 distinct proteins of which the roles and those of complexes they form remain
elusive. In this study, we first defined a canonical RBC proteome of 1,202 proteins (1% FDR) using machine learning and
data from quantitative mass spectrometry and RNA-seq on RBCs and other blood cell types. We then determined the
soluble stable protein complexes in mature RBCs, based on mass spectrometry of 1,944 native biochemical fractions of
hemoglobin-depleted hemolysate and detergent solubilized membrane protein complexes. Our data reveal an RBC
interactome dominated by protein homeostasis, redox biology, cytoskeletal dynamics, and carbon metabolism, and give
insights into the biophysical organization of integral membrane proteins, channels, cytoskeletal proteins, and metabolic
enzymes at the molecular level. We validated protein complexes through electron microscopy and chemical crosslinking,
and with these data, built 3D structural models of the ankyrin/Band 3/Band 4.2 complex that bridges the spectrin
cytoskeleton to the RBC membrane to confer the characteristic RBC cell shape and morphological flexibility. The model
provides detailed interactions at the amino acid level for the first time and suggests spring-link behavior of ankyrin in linking
the cytoskeleton to the membrane. Taken together, our study provides a comprehensive view of the protein organization of
the simplest primary human cell and serves as a comprehensive resource for future research.
P13.05 Temporally resolved protein-protein interactions to guide therapeutic intervention in protein misfolding
diseases
Madison Taylor Wright (1), Bibek Timalsina (1), Valeria Garcia Lopez (1), Lars Plate (1)
(1) Vanderbilt University, Nashville, Tennessee, United States
Cells are under pressure to properly fold, assemble, and maintain the integrity of thousands of proteins throughout changes
in cellular conditions and stress. The proteostasis network (PN), consisting of folding, trafficking, and degradation
components, helps facilitate the proper processing, trafficking, localization, and secretion of these proteins in a highly
coordinated process referred to as protein quality control (PQC). Protein misfolding diseases such as thyroglobulin-related
congenital hypothyroidism (CH) are a result of improper partitioning of proteins between PQC pathways. Coordination of
PN-client interactions are critical for mediating proper PQC decisions. Recently, the PN has become an attractive target for
therapeutic intervention to combat such disease states. Yet, one of the bottlenecks in the implementation of such strategies
is a lack of understanding in the timing and coordination of PN-client interactions as there are no methodologies available to
identify and quantify transient protein-protein interactions with time-resolution at an organelle-wide scale. We are therefore
developing a quantitative mass spectrometry method, time-resolved interactome profiling (TRIP), to characterize the
temporal interactions between client proteins and the PN to better understand PQC processes. TRIP is achieved with a two-
stage enrichment strategy which utilizes unnatural amino acid incorporation through biorthogonal noncanonical amino acid
tagging (BONCAT) and streptavidin-biotin enrichment to label and subsequently purify nascent client proteins into time-
resolved fractions for affinity purification – mass spectrometry (AP-MS) analysis. We have coupled this methodology with
functional genomic screening to identify key PQC regulators that control folding and secretion of WT thyroglobulin and CH-
The COVID-19 pandemic has highlighted the importance of rapidly responding to emerging pathogens. The use of
Remdesivir, initially investigated as a broad-acting RNA virus therapeutic, as a treatment for SARS-CoV-2 shows the
advantage of having existing antiviral drugs which may potentially be used against future outbreaks. However, many viruses
that pose a large global health burden have no current therapeutic options available. Among these are the flaviviruses,
which include dengue, Zika, and yellow fever virus. Direct-acting antiviral compounds developed against these pathogens
have been examined, but not successfully translated into therapeutics due to rapid emergence of resistance in the viral
genome. An alternative option for broad-spectrum antiviral agents is to target host processes required for the viral life cycle
and disrupt macromolecular interactions to block viral replication. To this end, we showed that two proteostasis regulator
compounds (147 and 263) can act as antivirals against multiple strains of dengue and Zika virus. These compounds were
initially developed to enhance host protein folding, but we discovered that they also potently inhibit flavivirus infection.
Mechanistic investigations show 147 works in a thiol-dependent manner, and we use chemoproteomic approaches to
identify endoplasmic-reticulum (ER) resident chaperone as the predominant protein targets of this molecule. Despite
compound 147 and 263 possessing similar chemical structure, mechanistic studies and target ID approaches show the
modes of action of the two molecules likely diverge. Both molecules are effective against a wide panel of viruses, providing
a potential new avenue for host-centered therapeutics by targeting conserved ER proteostasis processes to act as broad-
spectrum antivirals.
P14.02 Applying proteomics to address a conundrum in virology: how mitochondrial fragmentation can be linked
to increased bioenergetics
Cora Betsinger (1), Connor Jankowski (1), William Hofstadter (1), Joel Federspiel (1), Clayton Otter (1), Pierre Jean Beltran
(1), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
Alterations to mitochondrial functions and cellular metabolism are hallmarks of nearly all viral infections. As obligate
intracellular parasites, viruses rely on mitochondria for the production of biosynthetic precursors and energy necessary for
generating new viral particles. However, how the widely-spread human cytomegalovirus (HCMV) remodels mitochondrial
structure and function had presented a conundrum. It remained unclear how HCMV increases mitochondrial bioenergetics
despite triggering mitochondrial fragmentation. Employing a multidisciplinary approach integrating proteome and
interactome datasets with super-resolution confocal microscopy, LC-MS based metabolite profiling and metabolic assays,
we identify a previously uncharacterized viral protein, pUL13, which targets the mitochondria and increases oxidative
phosphorylation during infection. We use targeted mass spectrometry analysis of the HCMV proteome during infection,
coupled with molecular virology techniques to establish that pUL13 is required for productive HCMV replication. We then
quantify temporal cellular proteome changes during infection and demonstrate that pUL13 alters electron transport chain
(ETC) protein abundances. Using LC-MS based metabolite profiling and live-cell Seahorse metabolic assays to monitor
P14.03 Membrane protein composition of S. pneumoniae extracellular vesicles provide insights into extracellular
vesicle formation and shedding
Stephanie Biedka (1), Rory Eutsey (2), Luisa Hiller (2), Amber Lucas (1)
(1) Impact Proteomics, Pittsburgh, Pennsylvania, United States
(2) Carnegie Mellon University, Pittsburgh, Pennsylvania, United States
Extracellular vesicles (EVs) are complex, cell-derived nanoparticles generated by all cell types. EVs are composed of lipid
bilayer membranes and their associated membrane proteins, nucleic acids, and luminal proteins. The mechanism by which
Gram-positive bacteria shed EVs is still unknown. EVs from the Gram-positive human pathogen S. pneumoniae, which is a
major cause of otitis medi and pneumonia, are of particular interest because of how they EVs modulate the host immune
response. To uncover possible mechanisms for EV production and shedding in S. pneumoniae, we have performed a
comparative proteomics analysis of EV membrane proteins versus whole-cell membrane proteins. Membrane proteins were
enriched from intact S. pneumoniae cells or their EVs using a ProMTag labeling and capture workflow. ProMTag is a
bifunctional protein tag where one moiety of the tag is able to form a reversible, covalent link to primary amines on proteins.
The other moiety is methyltetrazine, which can form an irreversible, covalent bond with trans-Cyclooctene (TCO) on the
surface of beads to capture ProMTagged proteins for cleanup and elution. Using this workflow plasma membrane proteins
can be tagged, captured, washed to remove non-plasma membrane proteins, and then eluted in their original, unmodified
state. In this study, intact cells and EVs from S. pneumoniae cultures were separated and the extracellular domains of
membrane proteins in these two fractions were labeled with ProMTag. The membrane proteins were then enriched,
washed, and eluted using the ProMTag workflow. These membrane protein populations were then TMT labeled and
analyzed using mass spectrometry. Comparative analysis revealed membrane proteins that are concentrated or absent in
EV membranes relative to bulk plasma membrane from whole cells, indicating a selective process for EV formation in S.
pneumoniae. With this information, we gained new insights into extracellular vesicle formation and shedding.
P14.04 Human Viruses Rewire Membrane Contact Sites for Pro-Viral Organelle Remodeling
Katelyn Cook (1), Elene Tsopurashvili (1), Jason Needham (2), Sunnie Thompson (2), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
(2) University of Alabama Birmingham, Birmingham, Alabama, United States
P14.05 Viral Remodeling of the Host Proteome and Host Factors Associated with Viral Genomes
Joseph Dybas (1), Krystal Lum (1), Katarzyna Kulej (1), Emigdio Reyes (1), Richard Lauman (2), Matthew Charman (1),
Caitlin Purman (1), Robert Steinbock (3), Nicholas Grams (2), Alexander Price (4), Lydia Mendoza (2), Benjamin Garcia (2),
Matthew Weitzman (2)
(1) Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, United States
(2) University of Pennsylvania, Philadelphia, Pennsylvania, United States
(3) University of Pennsylvania, Philadelphia, Pennsylvania
(4) Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
Interactions between viral and host proteins determine the ultimate outcome of virus infection. Cellular antiviral proteins
present the first line of host defense, while viral proteins act to establish an environment conducive to infection. Viruses
remodel the cellular proteome by altering protein abundance, rewiring protein-protein or protein-nucleic acid interactions,
and redirecting protein post-translational modifications. Global proteomics provides an ideal approach to investigate virus-
host interactions within the infected host cell. Adenovirus (AdV) is a ubiquitous human pathogen. AdV is known to induce
degradation of host antiviral factors during infection and is an important model system to study viral manipulation of the host
proteome. However, the full repertoire of virus-host interactions, host antiviral factors, and viral countermeasures are not
fully understood. We describe an integrated proteomics strategy designed to define host proteins that target replicating AdV
viral genomes, and quantify AdV-induced manipulations to the host proteome. We used Isolation of Proteins on Nascent
DNA (iPOND) combined with mass spectrometry, to quantify the proteome associated with AdV genomes. We compared
wildtype AdV to a mutant lacking viral proteins known to counteract host antiviral factors. Our rationale was that cellular
proteins enriched on viral genomes, in the absence of viral countermeasures, may include host antiviral factors. We used
P14.06 Direct antigen epitopes mapping in human serum samples against AVA vaccine-elicited polyclonal
antibodies
Mulin Fang (1), Kenneth Smith (2), Zhe Wang (3), Kellye Cupp-Sutton (3), Kathleen Haley (4), Judith James (4), Si Wu (1)
(1) University of Oklahoma, Norman, Oklahoma, United States
(2) Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma, United States
(3) University of Oklahoma, Norman, United States
(4) Oklahoma Medical Research Foundation, Oklahoma City, United States
Anthrax vaccine adsorbed (AVA) containing protective antigen (PA) is the only FDA-licensed anthrax vaccine in the US.
Characterization of the binding of AVA-induced anti-PA polyclonal human antibodies to the PA antigen after vaccination is
crucial to understanding the mechanisms of AVA-elicited humoral immune response as well as to help refine the vaccine for
protection against anthrax. However, techniques that measure conformational epitope binding in complex mixtures of
antibodies (e.g., serum) remain elusive. We have recently developed and optimized a long-gradient (e.g., 90 minutes) sub-
zero temperature ultra-high performance liquid chromatography hydrogen deuterium exchange mass spectrometry (UPLC-
HDX-MS) platform for the analysis of protein interactions in complex biological samples including human plasma or sera.
Here, we applied this novel platform to characterize PA epitopes against 4 fully human monoclonal antibodies. We found
that p1C03 demonstrates partial neutralizing activity in vivo and binds to a flexible loop region on domain 1A of PA. p6C04
and p1A06 bind to the same epitope in domain 3 in a manner that prevents oligomerization but is not neutralizing. p6C01
binds to an epitope in domain 3 in adjacent helices and is highly neutralizing. The epitopes we identified to be associated
with neutralizing performance will contribute to the successful development of novel therapeutics and subunit vaccines
against anthrax. The performance of the UPLC-HDX-MS platform was further evaluated for direct epitope mapping in total
IgG pooled from healthy controls without AVA vaccination (spiked with 2% p1C03 and p6C01) and total IgG pooled from
AVA-vaccinated donors. The epitopes against p1C03 and p6C01 were identified in the control sample and two protected
regions were identified in the vaccinated sample. Our results demonstrated the feasibility of the application of our novel
platform for direct mapping of antigen epitopes in vaccine-elicited serum samples.
P14.07 Peroxisomes are remodeled into dual structural populations to promote pro-viral cellular reprogramming
during herpesvirus infection
William Hofstadter (1), Katelyn Cook (1), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
P14.08 Dynamic nuclear organization and protein translocation underlie nuclear viral DNA sensing and intrinsic
immune signaling
Dawei Liu (1), Krystal Lum (1), Joshua Justice (1), Peter Metzger (1), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
In mammalian cells, DNA sensors are critical players in innate immunity, serving to detect the presence of pathogenic DNA
and activate innate immune responses. The interferon inducible protein 16 (IFI16) has emerged as a key sensor of foreign
DNA in the nucleus. Its function is especially critical in the defense against nuclear-replicating viruses, such as herpes
simplex virus 1 (HSV-1). During viral infection, IFI16 quickly localizes to the nuclear periphery, where it binds to incoming
viral DNA and oligomerizes. Upon viral DNA binding, IFI16 inhibits virus replication by inducing cytokine expression and
inhibiting viral gene expressions. However, little is known about how IFI16 transmit its downstream signals to activate
intrinsic and innate immune responses upon binding to viral DNA. Here, we first investigated the reorganization of nuclear
composition early in infection. Using biochemical fractionation and quantitative mass spectrometry, we defined the dynamic
proteomes of the nuclear periphery compared to the nuclear core in uninfected and infected primary fibroblasts. This
provided insights into which host factors are specifically recruited to the nuclear periphery in response to the deposition of
viral genome. Next, to identify IFI16-dependent protein complex formation changes during infection, we performed thermal
proximity coaggregation (TPCA) mass spectrometry analyses in control and IFI16 CRISPR-knockout cells. Finally, we
investigated whether, upon binding to viral DNA, IFI16 induces nuclear-cytoplasmic shuttling of yet unknown host factors to
communicate with the ER-localized STING adaptor protein. We infected cells with wild type HSV-1 or a strain that lacks the
ability to inhibit IFI16, and performed nuclear-cytoplasmic fractionation and tandem mass tagging (TMT)-MS to predict
P14.09 Architecture of Mammalian Innate Immunity: How Interferon Classes Tune the Protein Interactions and
Abundances of Interferon-Stimulated Genes
Krystal Lum (1), Ileana Cristea (1)
(1) Princeton University, Princeton, New Jersey, United States
The mammalian innate immune system has been shaped by the coevolution of host-pathogen interactions over millions of
years. To control the replication and spread of infection by a vast diversity of pathogens including viruses, the innate
immune response has evolved to converge on a core cellular activity: the expression of several interferon (IFN) classes,
which induce the transcription of hundreds of interferon-stimulated genes (ISG). While the significance of ISGs in stimulating
adaptive immunity and promoting antiviral functions is well accepted, and mechanisms of action have been characterized
for some ISGs, several fundamental aspects of this sophisticated host response remain unknown. Our knowledge of the
cellular co-factors associating with ISGs in immune cells and how canonical IFN classes (Type I, II, and III) organize these
ISG protein interaction networks are unclear. We also lack methods that reliably and specifically monitor ISG protein
abundances en masse. Here, we adapt thermal proximity co-aggregation coupled to mass spectrometry to capture the
system-wide dynamics of ISG protein complexes upon stimulation of immune cells with different IFN types. We identify de
novo ISG-ISG interactions and protein assemblies that are commonly engaged across several IFN treatments, or specific to
an IFN class. We further develop a targeted mass spectrometry assay based on parallel reaction monitoring to quantify
protein abundances of ISGs representative of shared and distinct IFN classes throughout treatments with different IFN
types. We observe that ISGs exhibit several distinct protein expression profiles, suggesting ISG subsets are temporally and
differentially regulated during an innate immune response. Altogether, our assays and findings establish a cartographic
protein map of mammalian innate immunity as it is shaped by canonical IFN classes. Our framework can be applied to any
biological question involving ISGs, including assessing clinical samples and applications in viral infections and immune-
related disease states such as autoimmunity.
P14.10 Proteomic characterization of human small airway epithelial cells and the effect of the BRD4-specific
inhibitor ZL0454 on RSV infection by diaPASEF-MS
Morgan Mann (1), David Roberts (1), Ying Ge (1), Allan Brasier (1)
(1) UW - Madison, Madison, Wisconsin, United States
Respiratory Syncytial Virus (RSV) is a ubiquitous airway virus that causes severe inflammation and airway pathology in
children and the elderly. RSV infects the epithelial cells of the upper and lower respiratory tract, resulting in toll-like receptor
and NF-kB mediated innate inflammation. In the process, the NF-kB subunit RelA/p65 translocates into the nucleus and
interacts with the epigenetic scaffold BRD4. The resulting complex traffics to NF-kB early-intermediate genes to trigger the
coordinated expression of pro-inflammatory chemokines and interferons to signal nearby immune cells and form an early
The thymus is an important site to the development of an immune-competent and self-tolerant T cell repertoire. Studies
suggest that stromal cells in the thymus express a wide range of the human proteome. Cleaved peptide/epitopes of this
proteome are displayed on these cells via MHC1 receptors. The thymus selects against T cell receptors that either bind too
tightly or do not bind to these displayed epitopes. The thymus encounters profound anatomical changes throughout its
lifespan, where organ atrophy begins at 1 year old and continues into adolescence. The decline of naïve T cell output was
correlated with thymus involution. While Age-related morphological changes were broadly characterized, molecular and
cellular mechanisms driving this regression are still elusive. Despite this key role in promoting immune system homeostasis,
the thymus has not been studied using deep proteomics in humans. Here, we present a global proteomics investigation of
fresh-frozen human thymi (n=73), obtained from patients undergoing corrective cardiac surgery. The ages ranged from 1
day to 15 years of life. Using the SP3 method coupled with the tims-TOF Pro we identified and quantified >8500 proteins.
Our analysis revealed significant enrichment of proteins associated with positive regulation of T cell migration and
proliferation in the thymi from newborns and infants until 4 months. In thymi from older children, we measure significant
enrichment of proteins associated with tissue remodeling as well as progesterone and testosterone metabolic process.
Additionally, we develop a workflow for proteomics analysis of 14 FACS sorted thymic cells populations derived from pooled
human tissues. In summary, using the largest collection of fresh-frozen thymi, we are the first to characterize the human
P14.12 Identifying prognostic signatures in the serum proteome of recuperated COVID-19 patients.
Smruti Pushalkar (1), Shaohuan Wu (1), Matthew Pressler (1), Justin Rendleman (1), Burcu Vitrinel (1), Elodie Ghedin (2),
Ted Ross (3), Hyungwon Choi (4), Christine Vogel (1)
(1) New York University, New York, New York, United States
(2) National Institute of Health, Bethesda, Maryland, United States
(3) University of Georgia, Athens, Georgia, United States
(4) National University of Singapore, Singapore
The COVID-19 pandemic affected >260 million people worldwide with >5 million deaths. Presented by severe acute
respiratory syndrome coronavirus 2 (SARS-CoV-2), the disease displays extensive clinical heterogeneity, ranging from
asymptomatic to severe symptoms. While specific examples are known, e.g., elevated blood cytokine levels correlating with
severe symptoms, the more general changes in the patient serum proteome are still under investigation. We used
quantitative shotgun proteomics to characterize serum proteins of 29 recuperated COVID-19 patients and age-, race-, and
gender-matched healthy controls. COVID-19 patients were 9 to 70 days post diagnosis and symptom-less at the time of
sampling. We quantified ~330 proteins involved in inflammation, activation of complement pathways, blood coagulation and
antimicrobial response. Of these, 22 and 15 serum proteins were significantly up- and down-regulated, respectively in
COVID-19 patients compared to healthy controls (q-value<0.05). Recuperated COVID-19 patients had significantly elevated
immunoglobulins for antigen recognition, Peroxiredoxin-2, hemoglobin subunits, Carbonic Anhydrase1, and Apolipoprotein
A1. Interestingly, they also showed significant depletion of actin cytoskeleton signaling proteins, such as filamin, profilin and
cofilin, suggesting possible links to hampered T cell signaling as well as manipulation of host actin cytoskeleton by the virus.
Moreover, the COVID-19 infected individuals showed significant downregulation of thymosin ß, transgelin-2 and plasma
serine protease inhibitor critical for lung functioning and enzymes that activate inflammatory cascades consistent with the
skewed immune and organ function in COVID-19 patients. Comparing our results with other analyses, we identified
intriguing similarities with acute COVID infections such as elevated hemolysis markers providing potential etiological insight
into hemolytic complications, e.g., anemia or irreversible organ damage.>
Alzheimer’s disease (AD) biofluid biomarkers reflecting brain pathophysiology beyond amyloid-β and tau (A/T) are
needed. We analyzed the proteomes of paired blood plasma and cerebrospinal fluid (CSF) of the same AD and A/T
biomarker-negative control subjects using three highly specific and sensitive quantitative assays to compare protein
measurements across platforms and provide a consensus atlas of biofluid protein changes in AD that may serve as
mechanistic, therapeutic, or diagnostic biomarkers. Blood plasma and cerebrospinal fluid from patients positive for A/T
biomarker AD status with cognitive impairment (n=18) were assayed for quantitative protein differences relative to age-,
sex-, and race-matched A/T biomarker-negative cognitively normal control subjects (n=18). Assays were discovery-mode
mass spectrometry (MS), Olink antibody proximity extension assay (PEA) (>1100 proteins), and SomaScan aptamer-based
fluorescent array quantitation (>6600 proteins). Biofluids were assayed without depletion on all three platforms, and also
with immunodepletion of the 14 most abundant proteins prior to additional mass spectrometry. All samples were analyzed
from equal biofluid volume. Proteomic changes were summarized using integrative analysis incorporating knowledge from
brain tissue proteomic systems biology. Measurements of protein changes across the platforms were consistent overall and
robust to depletion of the most abundant proteins in either CSF or plasma. Overall, more proteins were decreased in AD
CSF and plasma. In AD CSF, increased proteins included neuron/synapse-derived proteins of the 14-3-3 family,
neurogranin (NRGN), and neurofilament light polypeptide (NEFL). Metabolic enzymes were found increased or decreased.
Neuronal marker protein neuronal pentraxin-2 (NPTX2), and vascular/blood signature proteins such as alpha-1-
antichymotrypsin (SERPINA3), were found decreased. In AD plasma, many of the same neuronal marker proteins
decreased that were increased in CSF, including 14-3-3 proteins and NRGN, with strong anticorrelation. Multi-platform CSF
and plasma AD biomarkers reflect consistent measurements, biased towards decreased levels, and reflecting negative
correlation of some markers across the two biofluids.
P15.02 Proteomic Analysis of Dorsal Root Ganglia of Type II Diabetic Donors with Peripheral Neuropathic Pain
Megan Doty (1), Yan Wang (1), Bradford Hall (1), Margaret Cassidy (1), Sijung Yun (2), Ashok Kulkarni (1)
(1) NIH/NIDCR, Bethesda, Maryland, United States
(2) Yotta Biomed, LLC., Potomac, Maryland, United States
Diabetic peripheral neuropathy (DPN) is characterized by spontaneous pain often in the extremities. Incidence of DPN is
anticipated to rise with the global increase in diabetes. However, there remains a lack of safe, effective analgesics to
control this painful condition. The dorsal root ganglia (DRG) contain soma of sensory neurons and modulate sensory signal
transduction into the central nervous system. The aim of this study was to gain a deeper understanding of changes in
molecular pathways in the DRG of DPN patients with chronic pain. Towards this goal, we used bottom-up proteomics
approach to compare the proteome and phosphoproteome profiles of DRGs from donors with DPN and non-diabetic
P15.03 Urinary Protein Signatures for the Detection of Antiretroviral-Induced Acute Kidney Injury: A Retrospective
Study
Ireshyn Govender (1), Previn Naicker (1), Neil Martinson (2), Stoyan Stoychev (1)
(1) Council for Scientific and Industrial Research, Pretoria, South Africa
(2) Perinatal HIV Research Unit, Johannesburg, South Africa
Introduction: Approximately one in five South Africans are HIV positive, which accounts for ~10% of the global HIV/AIDS
burden. South Africa administers the largest antiretroviral therapy (ART) program worldwide. However, 8-10% of first-line
ART patients experience acute kidney injury (AKI). Current AKI tests are unreliable and reflect following significant kidney
damage. Hence there is a need to identify sensitive and accurate markers for early detection of AKI. This study employed
mass spectrometry-based urinary proteomics for biomarker discovery in ART-AKI in two distinct South African cohorts.
Methods: Two, case-control (age-race-gender) matched, cohorts (n=74, n=80) were analysed. A simple and robust sample
collection procedure was coupled with reproducible, high-throughput, automated magnetic microparticle-based protein
capture, clean-up, and digestion. Proteome profiles were generated via data independent acquisition using an UltiMate
3000 UHPLC coupled to a TripleTOF 6600. Raw data was processed using a study-specific spectral library in Spectronaut
14. Proteins common to both cohorts were filtered as reporters of ART-AKI and processed using Perseus (v1.6.12.0) and
Cytoscape (v3.8.2). Machine learning (ML) was applied to protein abundance data using OmicLearn (v1.1.0) Results: Two-
hundred and two common proteins (inclusion criteria: q-value ≤ 0.01, unique peptides ≥ 2) showed the same directional
change in abundance. These indicated that damaged kidneys may undergo renal hepatisation showing increased levels of
liver-associated proteins. Immune system, complement cascade and serine protease inhibitors, were the dominant protein
families showing increased abundance in AKI patients. The XGBoost algorithm showed ≥ 95% sensitivity and specificity in
classifying AKI and severity. Conclusions: The workflow applied here showed clinical robustness and can find utility as a
routine workflow for urinary proteomics studies. ML offers promise for use with complex urinary proteomics datasets for
clinically applicable outcomes. Validation of the identified protein biosignatures is currently on-going in a large cohort (n =
2000).
P15.04 Proteomics to understand effect of maternal diet during pregnancy on fetal adult life
Zeeshan Hamid (1), Cun Li (2), Peter Nathanielsz (2), Laura Cox (1), Michael Olivier (1)
(1) Wake Forest University School of Medicine, Winston Salem, North Carolina, United States
(2) Southwest National Primate Research Center, San Antonio, Texas, United States
Introduction: Maternal diet during pregnancy has an important impact on growth and development of the fetus.
Epidemiological and animal studies have shown that quality of diet during pregnancy affects health of the child, with
P15.05 Divergent proteostasis interactions reveal distinct pathways attenuated by correctors in CFTR misfolding
mutations
Minsoo Kim (1), Eli McDonald (1), Carleen Sabusap (1), Lars Plate (1)
(1) Vanderbilt University, Nashville, Tennessee, United States
Cystic fibrosis (CF) is one of the most prevalent lethal genetic diseases that has over 2000 identified genetic variants in the
cystic fibrosis transmembrane conductance regulator (CFTR) gene. Corrector drugs such as lumacaftor (VX-809),
tezacaftor (VX-661) and elexacaftor (VX-445) combat mutation-induced defects by stabilizing the structure of CFTR. The
stabilization enables proper folding and thus facilitates processing and trafficking to increase the amount of functional CFTR
on the cell surface. Here, we identified differential responses to correctors by CFTR variants and investigated the underlying
cellular mechanisms of how CFTR biogenesis is altered for these variants. We employed affinity purification-mass
spectrometry (AP-MS) multiplexed with isobaric tandem mass tags to define the remodeling of CFTR proteostasis
interactions for multiple CFTR mutant variants in response to correctors. We identified several dysregulated pathways in the
CFTR interactome in misfolding variants. We determined that correctors restore early proteostasis pathways such as
translation and folding as well as degradation machinery. In particular, degradation fates diverged for mutants in which
inhibiting the proteasome attenuated VX-809 response in responsive mutations but inhibiting the lysosome attenuated
F508del mutant VX-809 response. Our data suggest a previously unidentified role for proteasomal degradation in VX-809
correction of CFTR. Moreover, VX-445 treated CFTR exhibited unique interactomics profiles compared to VX-809
treatment, indicating how critical pathways and proteostasis factors differ for individual correctors. We narrowed down
several proteins as novel targets of interest for validation via specific pathway modulators and knockdowns. Our approach is
unique in quantitatively identifying proteostasis characteristics associated with mutant-specific therapeutic response to
correctors. Our results could provide a better understanding of the corrector mechanism of action and reveal novel cellular
targets for therapeutic approaches.
P15.06 Plasma Proteomic Changes in Response to a High Salt Diet in Female Baboons
Sumaiya Nazli (1), Angelica Riojas (1), Zeeshan Hamid (1), Kimberly D. Reeves (1), Laura A. Cox (1), Michael Olivier (1)
(1) Wake Forest School of Medicine, Winston-Salem, United States
Hypertension is a complex medical condition that affects 45% of U.S. adults, and the prevalence increases with age. The
kidneys play a key role in regulating blood pressure, predominantly by regulating salt homeostasis. Non-human primates
are an ideal animal model to study physiological responses to diet due to their similarity to humans. In our study, female
baboons (n=8) were maintained on a low-sodium chow diet and fed on a high-sodium diet for six weeks. Blood samples
were collected before and after the high-salt challenge. Blood pressure was monitored throughout the study using telemetry.
We analyzed plasma by using mass spectrometry. For initial comparisons, we included proteins that were present in at least
50% of the samples at each time point (e.g., high vs low-sodium intake). We identified 180 proteins that overlapped
between groups, while 43 proteins were unique to the samples collected after the high-sodium diet and only 3 proteins were
unique to the samples collected on the low-sodium diet. We computed logistic regression with each protein to test for
differential detection rates between groups. False Discovery Rate (FDR) was computed to adjust for multiple comparisons.
No results met an FDR < 0.05, but 3 proteins met an unadjusted p-value < 0.05 (F13A1, KRT2, LAMP1). For proteins that
had less than 30% missingness across groups, we computed linear regressions to test for differential abundance between
groups. No proteins met an FDR < 0.05, but 17 proteins met an unadjusted p-value < 0.05. Many of these proteins were
related to heme signaling pathways. The plasma proteomics differences between the two groups could play a key role in
functional differences in the hypertensive group. This represents a promising avenue of investigation that may lead to more
development of these proteins in relation to possible impairments in kidney function with regulation of hypertension.
P15.07 Laser Capture Microdissection Assisted Spatial Resolved Proteomics Analysis of FFPE Human Tissues for
Fibrosis-Related Disease MOA Study
Yu Tian (1), Xue Wang (1), Jan Schejbal (1), Liang Jin (1), Baoliang Cui (1), Yupeng (David) He (2), Yonghao Cao (3),
Annette Schwartz Sterman (4), Robert Dunstan (5), Chenqi Hu (6)
(1) DMPK-BA, Abbvie Bioresearch Center, Worcester, Massachusetts, United States
(2) Discovery Dermatology and Fibrosis, Abbvie, North Chicago, Illinois, United States
(3) Transformational and Translational Immunology Discovery, Abbvie Bioresearch Center, Worcester, Massachusetts,
United States
(4) Immunology Pathology, Abbvie Bioresearch Center, Worcester, Massachusetts
(5) Immunology Pathology, Abbvie Bioresearch Center, Worcester, Massachusetts, United States
(6) DMPK-BA, Abbvie Bioresearch Center, Worcester, Massachusetts
Traditional omics with bulk sampling poorly defines disease heterogeneity and lack the sensitivity to identify true disease
targets. Therefore, for diseases with a high degree of heterogeneity, like idiopathic pulmonary fibrosis (IPF), and hidradenitis
suppurativa (HS), spatially resolved omics analysis is critical. laser capture microdissection (LCM) is a powerful technology
for spatial omics analysis, which can achieve spatial resolution as well as morphology-directed area selection based on
histo/path classification with strong biology information association. However, it’s very challenging to execute downstream
Background: The human tooth is a unique organ composed of mineralized tissues enamel and dentin, that have a
prominent extracellular matrix (ECM). Several rare diseases affect teeth, but the proteomic outcome is unknown. Our goal
was to establish new protocols for analysis of teeth proteome to provide a foundation to investigate diseased teeth.
Methods: Wisdom teeth were extracted at the NIH-DIR Dental Clinic and processed for protein extraction using protocols
established previously. Participants consented to clinical protocol (NCT01805869) before collecting teeth. In this work, we
attempt to optimize conditions for storage as well as protein extraction and analysis. For this purpose, teeth from healthy
volunteers were processed in separate conditions, including storage in RNALater vs Trizol, protein extraction using
Guanidine vs urea, and benefit of multiple steps with demineralization and limited trypsin digestion. Tryptic peptides were
analyzed using standard nanoLC-MS/MS workflow with data dependent analysis using orbitrap Fusion Lumos mass
spectrometer. Proteome Discoverer 2.5 software was used for protein identification and quantification. Results: More than
2000 proteins were detected from the mass spec analysis. RNALater and Trizol both preserved similar number of proteins.
Guanidine had higher proteins dissolved than urea, however, some proteins, especially the non-collagenous proteins
dissolved more in Urea. While most of the proteins identified were from the initial Guanidine or urea extraction, the follow up
extracts after demineralization released collagens and non-collagenous proteins which make up the bulk of the ECM
composition of dentin in tooth. Conclusion: We tested different conditions to test the best protocol for extracting proteins
from human teeth. RNALater could be an optimal solution for preserving proteins in teeth and other mineralized tissues.
Guanidine and Urea each extracted higher number of specific proteins, so they should be used depending on the affected
target proteins for downstream analysis.
Introduction: LCMS-based plasma proteomics is advancing our understanding of human molecular pathophysiology and
empowering the discovery of therapeutic targets and biomarkers. However, managing throughput, proteome depth and
reliability altogether represent a major gap to fully enable meaningful large cohort proteomics studies. Methods: We
developed an automated and high-throughput solution to enable large-scale proteomics analysis of proteins from 1 μL of
plasma. We automated the sample preparation protocol using a commercially available 96 MS Sample Prep Kit and a liquid
handling robotic platform. Digested peptides were separated using a Thermo ScientificTM Easy-Spray™ PepMap™ Neo
column on a Thermo ScientificTM VanquishTM Neo UHPLC system coupled to a Thermo ScientificTM Orbitrap ExplorisTM 480
mass spectrometer with a High-Field Asymmetric Waveform Ion Mobility Spectrometry (FAIMSTM). Results: We standardized
a plasma profiling workflow solution using the following building blocks: 1) A liquid handler for automated sample
preparation; 2) A next-generation LC that enables higher robustness and peak capacity; 3) The unified integration with
FAIMS Pro Interface coupled to a Thermo Scientific Orbitrap Exploris 480 mass spectrometer. 4) Furthermore, we fully
benefited from a new search engine that uses deep neural networks to maximize the retrieval of protein identifications
yielding significant improvement compared to conventional search engines. The integration of these technologies and
devices allowed us the scalability to profile >1000 proteins from 1µL plasma samples without compromising reliability,
robustness, and data quality. 85% of 1014 identified protein group has %CV of less than 20% in triplicates injections. 82
FDA-approved biomarkers can be identified, and 66 FDA-approved biomarkers can be reproducibly quantified (%CV<20%,
n=3) with our high throughput label-free LCMS method. Conclusions: We developed a highly reproducible quantitative
plasma proteomics workflow with data confidence to increase the statistical power of human large cohort plasma proteomics
study.>
Purpose: The neonatal swine heart possesses an ability to regenerate injured myocardium through the proliferation of pre-
existing cardiomyocytes (CMs), but this regenerative capacity is lost shortly after birth. The mechanisms governing the
proliferative capacity of CMs during this early postnatal stage are unknown; therefore, there is great need to define the
proteomic landscape during postnatal development to identify regulators of this regenerative process. Methods: Left
ventricular (LV) tissue was isolated from swine at postnatal days (P) 1, 7, 28 and 56 (n=3). For top-down analysis,
myofilament proteins were enriched using a differential pH extraction and analyzed by reversed phase liquid
chromatography (RPLC) interfaced with a Q-TOF mass spectrometer. For bottom-up analysis, LV tissue was homogenized
in Azo, and proteins were digested with trypsin prior to RPLC interfaced with the timsTOF Pro. Results: Top-down
proteomics identified over 75 sarcomeric proteoforms and revealed a concerted transition of fetal sarcomeric proteoforms
into their more mature counterparts. More specifically, we observed a marked decrease in ssTnI expression, cTnT2
expression, and α-Tpm phosphorylation throughout postnatal cardiac development. Global bottom-up proteomics quantified
over 4,000 protein groups throughout the developmental window. We identified >900 differentially expressed proteins
(adjusted p<0.05) and mapped these proteins to different biological processes, revealing a steep decrease in translational
processes and a sharp increase in metabolic proteins in P56 hearts compared to P1 hearts. Conclusion: Using these
methods, we comprehensively defined how the proteomic landscape of swine hearts changes throughout postnatal
development. We identified significant alterations in sarcomere proteoform expression patterns, which may influence
sarcomere disassembly, a process necessary for CM proliferation. Additionally, we mapped global changes in protein
abundance to determine there was a significant increase in metabolic activity and a substantial decrease in processes
related to translation. Overall, we will use these results to guide future investigations into swine heart regeneration.>
P16.02 A complete and automated sample preparation strategy for high throughput and standardized proteomics
Nicolai Bache (3), Dorte B. Bekker-Jensen (1), Florian S. Harking (2), Angela McArdle (1), Lasse Falkenby (1), Jakob
Bunkenborg (1), Ole B. Hørning (1), Jesper V. Olsen (2)
(1) Evosep Biosystems, Odense, Denmark
(2) University of Copenhagen, Copenhagen, Denmark
(3) Evosep Biosystems, Odense
Purpose Mass spectrometry-based proteomics is fast growing and provides a powerful set of technologies, with the
potential to revolutionize health care and enable precision medicine. We recently developed the Evosep One, which is
specifically designed for high throughput applications, with a focus on clinical analysis of large sample cohorts. For a
widespread adoption of proteomics in the clinic, the entire workflow from sample preparation through LC-MS and data
P16.03 Global multi-site assessment of LC-MS proteomics precision and accuracy with high-throughput analysis
Oleksandr Boychenko (1), Runsheng Zheng (1), Tabiwang Arrey (2), Amirmansoor Hakimi (3), Christopher Pynn (1), Alec
Valenta (4)
(1) Thermo Fisher Scientific, Germering, Germany
(2) Thermo Fisher Scientific, Bremen, Germany
(3) Thermo Fisher Scientific, San Jose, California, United States
(4) Thermo Fisher Scientific, Michigan, Michigan, United States
Throughput limitations and often preclude the adoption of nanoLC-MS methods for translational proteomics applications
such as biomarker validation because the large sample cohorts need to be analyzed to reveal changes that stand out of
biological, analytical, and sample preparation variation. The throughput (the number of samples per day) and MS utilization
(the ratio of peptide elution window to total cycle time) can be boosted by utilizing LC instrumentation with the extended flow
and pressure capabilities and minimal delay volumes. Optimized LC-MS methods operated at flow rates in the range from
1.3 to 0.3 µL/min were created to maximize MS utilization (calculated as the ratio of peptide elution window to cycle time).
The MS utilization gradually increased from 68% for the shortest method to 95% for the longest 60 min method. The length
of the methods is also linked to the increased FWHM from ca. 3 sec for 8 min method to ca. 10 sec for 60 min method. To
comprehensively estimate the analytical variability were compared the quantification and identification results collect (i) on
the same low-flow LCMS instrument using the same separation column; (ii) on the same instrument using three separation
columns of the same dimensions; (iii) on different instruments, with different separation and trap columns, located at three
different laboratories in the US and Europe. The label-free quantification showed that 72% of identified proteins have less
than 25% variation of abundance for 100 continuous injections. The obtained results allow setting realistic estimates for
abundance variation that are required to confidently detect targets above the multi-site analytical variation.
While many people infected with COVID-19 were fortunate enough to not succumb to the disease, hundreds of millions of
people were still afflicted by COVID-19 in the past two years and it is imperative to examine possible long-term
complications that might arise due to the infection. While the airway epithelium is the initial site of infection, SARS-CoV-2
can disseminate to infect other organs. The pancreas secretes hormones that control blood glucose levels and
dysfunctional pancreas is often the cause of diabetes. Pancreatic islets express ACE2, the receptor for SARS-CoV-2 Spike
protein, and is believed to be a target of SARS-CoV-2. Here we examine effects of SARS-CoV-2 infection of the pancreas
and how it relates to diabetes through phosphoproteomic analysis. We were able to detect on average 20,000 unique
peptides with above 85% phosphoenrichment. From this dataset, our KSEA was able to predict the kinase activities of 67
different kinases. Doing a gene ontology (GO) biological process over-representation analysis of kinases with predicted
upregulated activities revealed cell death pathways as the most significantly enriched pathways 30 minutes post RBD
treatment. This trend was also observed in pancreatic islet cells 24 hours post SARS-CoV-2 infection, indicating COVID-19
induces cell death. TUNEL staining, an indicator of cell death, confirmed an increase in cell death in β-cells infected with
SARS-CoV-2. This process is similar to the cause of Type 1 diabetes where insulin deficiency due to autoimmune response
related loss of β-cells, indicating COVID-19 patients are more prone to developing diabetes. Induction of the JNK-MAPK
pathway in β-cells post SARS-CoV-2 infection was predicted using KSEA, suggesting that the JNK-MAPK pathway could be
involved in inducing apoptosis in infected β-cells and a candidate for potential target for therapeutic intervention.
P16.05 Quantitative proteomics of zebrafish embryos reveals wave-like expression profiles of transcription factors
during maternal to zygotic transition
Fei Fang (1), Daoyang Chen (1), Liangliang Sun (1)
(1) Michigan State University, East Lansing, United States
The transition from fertilized egg to embryo is accompanied by a multitude of changes in gene expression. Zebrafish is a
widely used vertebrate model organism in embryological developmental biology due to its high genomic conservation with
humans and rapid development. While the transcriptional events during the development were extensively studied on
zebrafish, the data on the proteome profile are scarce, especially during embryogenesis. In the present study, we coupled
iTRAQ labeling with both CZE-MS/MS and LC-MS/MS to profile protein expression levels of zebrafish embryos across four
stages during the maternal-to-zygotic transition between fertilization and 6 hours post fertilization (hpf). Nearly 5,000
proteins were quantified with biological replicates. The clusters of protein expression profiles clearly indicated the
expression level of nucleus proteins dramatically increased in pre-midblastula transition (MBT) and MBT stages, involving in
the assembly of RNA Polymerase II Complex and protein transport. Among the 110 quantified transcription factors (TFs), 41
TFs have significant abundances changes across the four stages and generate 4 waves of zygotic transcription. The wave-
like expression patterns showed that the TFs containing homeobox and zf-H2C2-2 domains playing dominant roles during
MBT. To further investigate how nanog, an important homeobox TF, instruct cell fate, we investigated proteomics,
morphological and functional changes of zebrafish embryos upon morpholino-mediated knockdown of Nanog expression.
P16.06 Proteomic Analysis of Alcohol-associated Hepatitis Reveals Glycoprotein NMB (GPNMB) as a Novel Hepatic
and Serum Biomarker
Kristofer Fritz (1), Peter Harris (1), Cole Michel (1), Youngho Yun (1), Courtney McGinnis (1), Mohammed Assiri (2), Ali
Ahmadi (3), Zhaoli Sun (3), James Roede (4), Matthew Burchill (1), David Orlicky (4), Rebecca McCullough (1)
(1) University of Colorado Anschutz Medical Campus, Aurora, Colorado, United States
(2) King Saud University, Riyadh, Saudi Arabia
(3) Johns Hopkins School of Medicine, Baltimore, Maryland, United States
(4) University of Colorado Anschutz Medical Campus, Aurora, Colorado
Alcohol consumption remains a leading cause of liver disease worldwide, resulting in a complex array of hepatic
pathologies. Individuals who progress to a rarer form of alcohol-associated liver disease (ALD), alcohol-associated hepatitis
(AH), require immediate life-saving intervention in the form of liver transplantation. While multiple mechanisms have been
identified that contribute to ALD, no cures exist and mortality from AH remains high. The rapid onset of AH is poorly
understood, and an early-stage serum biomarker could prove to be life-saving. To identify novel pathways associated with
AH, our group utilized proteomics to investigate AH-specific biomarkers in liver explant tissues. Protein abundance and
acetylomic analyses were performed utilizing nHPLC-MS/MS, revealing significant changes to proteins associated with
metabolic and inflammatory fibrosis pathways. Here, we describe a novel hepatic and serum biomarker of AH, glycoprotein
NMB (GPNMB). The anti-inflammatory protein GPNMB was significantly increased in AH explant liver (HPLC-MS/MS) and
serum (ELISA) compared to healthy donors by 50-fold and 6.5-fold, respectively. Further, bioinformatics analyses identified
an AH-dependent decrease in protein abundance across fatty acid degradation, biosynthesis of amino acids, and carbon
metabolism. The greatest increases in protein abundance were observed in pathways for focal adhesion, lysosome,
phagosome, and actin cytoskeleton. In contrast with the hyperacetylation observed in murine models of ALD, protein
acetylation was decreased in AH compared to normal liver across fatty acid degradation, biosynthesis of amino acids, and
carbon metabolism. Interestingly, immunoblot analysis found epigenetic marks were significantly increased in AH explants,
including Histone H3K9 and H2BK5 acetylation. The increased acetylation of histones likely plays a role in the altered
proteomic profile observed, including increases in GPNMB. Understanding the origin and consequences of these changes
will yield new mechanistic insight for ALD as well as identify novel hepatic and serum biomarkers, such as GPNMB.
P16.07 Deep Metaproteome Analysis using a Vanquish Neo UHPLC System Coupled to an Orbitrap Eclipse Tribrid
Mass Spectrometer with FAIMS Pro Interface
Amirmansoor Hakimi (1), Kevin L. Schauer (1), Runsheng Zheng (2), Alexander Boychenko (2), David Horn (1), Steven L.
Reeber (1), Daniel Lopez Ferrer (1)
Purpose: LC-MS-based proteomics is a powerful tool for deep profiling of peptides and proteins in complex biological
samples. Analyzing the proteome of microbial communities represents a challenge for current proteomics workflows due to
the wide dynamic range of metaproteomes. Extensive fractionation is required to address this challenge. Here we use a
newly developed high performance nLC column that provides extensive peak capacity and is hyphenated to a High-Field
Asymmetric Waveform Ion Mobility Spectrometry (FAIMS) interface coupled to a Thermo ScientificTM Orbitrap
EclipseTM TribridTM mass spectrometer to maximize the proteome coverage of metaproteome samples with very high protein
diversity. Methods: 1µg of peptides were separated using a Thermofisher Scientific™ EASY-Spray™ PepMap™ Neo
column on a Vanquish Neo UHPLC system coupled to an Orbitrap Eclipse Tribrid MS with and without FAIMS. Three CVs
with the least overlap in the peptides were used. The final FAIMS MS method was set to switch between different CVs with
a top-speed method in a 3 second total cycle time over a 140 minutes gradient. Results: The effect of gas-phase
fractionation using the FAIMS was evaluated. A Vanquish Neo UHPLC system utilizing a 75µm x 75cm EASY-Spray
PepMap-Neo column was used to separate 1µg load of peptides from different microbiome standards. The raw files were
searched against concatenated databases downloaded from UniPort using Thermofisher Scientific™ Proteome
Discoverer™ 3.0 software utilizing, SEQUEST-HT, and INFERYS re-scoring algorithm. Preliminary results showed over
10,000 proteins and 70,000 peptides in the ZymoBIOMICS Microbial Community standard for the No-FAIMS dataset. The
addition of FAIMS improved protein identifications by about 19%. Similar improvement was observed in the ZymoBIOMIC
Gut Microbiome standard dataset with identification of over 11,000 proteins and 80,000 peptides in the No-FAIMS
experiment with a 17.5% improvement in protein identification when FAIMS is used.
P16.08 Comprehensive Comparison of Serum vs. Plasma High-Throughput Proteomics Across a Healthy Gender-
Matched Cohort
Ventzi Hristova (1), H. Sophia Chung1 (1), Chiung-Yun Chang (1), Junmin Wang (1), Sonja Hess (1)
(1) Dynamic Omics, Centre for Genomics Research (CGR), Discovery Sciences, BioPharmaceuticals R&D, AstraZeneca,
Gaithersburg, Maryland, United States
Purpose: Clinical proteomics focuses on the discovery of novel, early biomarkers in liquid biopsy samples that can be
translated into clinical and precision medicine applications. Here we aim to comprehensively compare advantages and
challenges associated with serum and plasma, to guide specimen selection. This is achieved using our high-throughput
end-to-end proteomics workflow of a healthy gender-matched cohort. Methods Fasting EDTA-treated plasma and
matched serum were simultaneously collected from 40 healthy donors, 20 male and 20 female. To assess clot formation
variability, multiple serum samples were collected from a subset of donors. Specimens were centrifuged at 1500xg for 15
minutes to remove contaminants including platelets and clots. Samples were either directly digested with EasyPep 96 MS
Sample Prep Kit (Thermo Fisher Scientific) or subjected to depletion (High Select Top14 Resin Pierce) prior to digestion.
Peptide analysis was performed using 45- and 20-minute gradients with data-independent acquisition (DIA) methods on an
Orbitrap Exploris 480 mass spectrometer. DIA bioinformatic analysis was performed using library-based and non-library
approaches. Results Proteomics analysis of plasma and serum without depletion consistently demonstrated improved
proteome coverage in serum with up to 19% more protein groups identified compared to matched plasma. Replicate
analysis show at least 70% of protein groups identified in each sample have <20% CV. Following depletion, the proteome
P16.09 Fast library generation using Zeno MS/MS and microflow chromatography
Alexandra Antonoplis (1), Nick Morrice (2)
(1) SCIEX, Redwood City, California, United States
(2) SCIEX, Alderley Park, United Kingdom
Generation of spectral libraries is often thought of as a time consuming procedure, but in reality they can be generated very
quickly. Using microflow chromatography with very fast MS/MS acquisition on a QTOF system, then searching the data in
the cloud can enable library generation in <1 day. Here this workflow was explored using the ZenoTOF 7600 system, and
the impact of larger libraries on DIA protein quantification was tested. Forty-three fractions from high pH reversed phase
separations of complex digested samples (K562, Hela) were collected and then analyzed using microflow chromatography
(20 min gradients). Data dependent analysis and SWATH acquisition was performed on the ZenoTOF 7600 system using
Zeno MS/MS for fast, high sensitivity MS/MS analysis. Protein identification and SWATH acquisition data processing was
performed in the cloud using OneOmics suite. The libraries generated from each fractionated cell line contained >8000
proteins and >175000 peptides. Comparing the proteins found from each cell line, >500 new proteins were added to the
library by analyzing a second cell line. Processing SWATH acquisition data with the large, combined library enabled
extraction of many more proteins from a standard cell line acquisition, resulting in a 30-40% improvement from the
previously generated libraries on older platforms. SWATH acquisition experiments were performed with and without Zeno
MS/MS activated across a range of sample loadings and the impact on proteins quantified was assessed. For loads in the
100-400ng on column range, a 2-3 fold increase in proteins quantified <20% CV was observed. For lower loads, even
higher gains were observed. Using fast microflow LC and Zeno MS/MS combined with cloud-based data processing, very
large scale libraries can be generated in under 24 hours. Higher quality MS/MS libraries can improve information content
extracted for SWATH acquisition data. Zeno SWATH can provide significant improvements in protein quantification>
P16.10 Enhanced Detection of Low Abundance Proteins in Various Mouse Tissues Using Offline HPLC as a
Precursor to LC-MS/MS.
David Parkinson (1), Leena Patil (1), John Price (1)
(1) Brigham Young University, Provo, Utah, United States
The human proteome is vast, and most tissues are dominated by a few distinct high abundance proteins (HAPs). In bottom-
up proteomics, peptides from HAPs significantly suppress LC-MS detection of peptides from low abundance proteins
(LAPs). This is problematic because detecting LAPs is essential to understanding cellular regulation of the proteome.
P16.11 Deep Proteome Coverage by Offline HPLC Fractionation of Intact Proteins as a Precursor to Bottom-up
Proteomic Analysis
Leena Patil (1), David Parkinson (1), John Price (1)
(1) Brigham Young University, Provo, Utah, United States
The human proteome includes more than 20,000 proteins ranging from <10 kDa to >1 MDa and thus the separation and
identification of proteins is challenging. Bottom-up proteomics have been extensively used for protein analysis, wherein the
proteins are enzymatically digested into smaller peptides prior to mass spectrometry (MS) analysis. The most significant
disadvantage of this technique lies in the difficulty to detect low abundance proteins (LAPs) in the presence of high
abundance proteins (HAPs). Many disease associated proteins and drug targets are LAPs. In order to visualize these
proteins, we have incorporated offline-HPLC fractionation and concatenation as a preparatory step, allowing us to separate
and enrich LAPs from HAPs prior to bottom-up analysis. Protein level fractionation dramatically decreases sample
complexity as proteins are diverse in their chemical nature unlike than peptides which show a uniform behavior. We have
compared weak anion exchange (WAX) and reversed phase liquid chromatography (RPLC) as an offline HPLC techniques
in the first dimension and LC-MS in the second dimension. Using intact protein mixtures from mouse liver tissue, we
optimized separation by selecting the stationary phase and modifying mobile phase conditions, as well as flow rate and
temperature. Each fraction was concatenated into one of four different groups, based on relative protein abundance. Each
group was then analyzed separately with LC-MS. By applying this method as a precursor to LC-MS analysis, we have
detected LAPs that would otherwise be missed due to HAPs interference. Offline RPLC fractionation resulted in a better
separation at protein level and subsequently better identification in Bottom-up LC-MS analysis than WAX. This could be
attributed to the better selectivity and higher interaction in RPLC. We have applied the developed method to serum, brain
and muscle tissue and witnessed a substantial increase in the proteome coverage.
Using high field asymmetric waveform ion mobility spectrometer (FAIMS) pro interface for proteomic analyses has gained
significant interests because FAIMS can reduce the background contamination peaks and boost the low intensity peaks.
Although researchers have demonstrated that multiple compensation voltages (CVs) in FAIMS benefit the total proteome
coverage for data dependent acquisition (DDA), this approach may increase cycle time due to switching among multiple
CVs, which could affect data independent acquisition (DIA) even more because higher quality MS2 spectra are commonly
required for DIA. Herein, we report the best practice of using single CV FAIMS-DIA approach for large-scale proteomics
project. We have systematically evaluated multiple factors in this workflow, including (1) segmented (acquisition windows)
and FAIMS-based gas phase fractionation, and liquid phase fractionation for spectral library generation; (2) a range of
single CV FAIMS for both DDA and DIA; (3) comparison of spectral library and sequence FASTA library-based database
search using 3 most common search engines, which are Spectronaut, EncyclopeDIA and PEAKS studio. We used standard
HeLa digests, two induced pluripotent stem cell (iPSC) lines for the method development and validate the optimized method
in a temporal proteomic profiling of iPSC-derived neurons during a course over 28 days differentiation. Importantly, we
generated companion single-cell transcriptomic datasets in parallel to verify our findings in proteomics. We found single CV
at -35V outperformed other tested CVs at DIA mode. The FAIMS-DIA (CV = -35V) approach provides the deepest
proteome, less inter-sample variation and better reproducibility. Interestingly, we found FASTA-based directDIA outperforms
any spectral library-based searches, of which Spectronaut provides the highest peptide identification. This extensively
optimized and validated proteomic pipeline can readily identify ~8,000 proteins from any single shot run. Additionally, we
innovatively used single-cell transcriptomic to validate neuron markers identified by proteomics.
RNA-protein interactions play key roles in cell metabolism and gene expression. Proteins bind to certain structural motifs in
RNA using RNA binding domains (RBDs), however, unconventional binding can also occur that does not require
conventional RBDs. This binding results in a ribonucleoprotein complex (RNP) which are dynamic in nature. Any
perturbations in RNA-protein interactions can leads to cellular dysfunction and disease. Multiple approaches exist to
catalogue RNA binding proteins (RBPs), but none comprehensively capture those due to inherent limitations. We developed
a mass spectrometry-based approach to identify RNA associated proteome at a global level in mouse macrophages, which
entails 254nm ultraviolet crosslinking of RBPs to RNA in live cells, followed by their capture on silica and selective elution of
peptides that represent those RBPs. The RBPs are then enzymatically digested, cleaned up of any non-peptide impurities
and analyzed with liquid chromatography online (LC) with orbitrap tandem mass spectrometry (MS/MS). The UV dosages
P16.14 Protein Folding Stabilities are a Major Determinant of Oxidation Rates for Buried Methionine Residues
ETHAN WALKER (1), John Bettinger (1), Kevin Welle (1), Jennifer Hryhorenko (1), Sina Ghaemmaghami (1)
(1) University of Rochester, Rochester, New York, United States
The oxidation of protein-bound methionines to form methionine sulfoxides has a broad range of biological ramifications and
it is therefore important to delineate factors that influence methionine oxidation rates within a protein. This is especially
important for biopharmaceuticals, such as antibody-based therapeutics, where slow oxidation over long term storage can
lead to their eventual deactivation, or degradation. Understanding the factors that drive methionine oxidation can allow one
to predict whether a given Met residue will be labile, and thus streamline the development of drug candidates. Previously,
neighboring residue effects and solvent accessibility (SA) had been shown to impact the susceptibility of methionine
residues to oxidation. In this study, we provide proteome-wide evidence that oxidation rates of buried methionine residues
are also strongly influenced by the thermodynamic folding stability of the domains where they reside. We surveyed the E.
coli proteome using several proteomic methodologies and globally measured oxidation rates of methionines in the presence
and absence of tertiary structure as well as folding stabilities of methionione containing domains. The data indicate that
buried methionines have a wide range of protection factors (PFs) against oxidation that correlate strongly with folding
stabilities. Concordantly, we show that in comparison to E. coli, the proteome of the thermophile T. thermophilus is
significantly more stable and thus more resistant to methionine oxidation. These results indicate that oxidation rates of
buried methionines from the native state of proteins can be used as a metric of folding stability. Additionally, to demonstrate
the utility of this correlation, we used native methionine oxidation rates to survey the folding stabilities of E. coli and T.
thermophilus proteomes at various temperatures and suggest a model that relates the temperature dependence of the
folding stabilities of these two species to their optimal growth temperatures.
dia-PASEF (Meier et.al., 2019) takes advantage of the additional dimension of separation provided by trapped ion mobility
(TIMS). The dia-PASEF cycle time can be reduced to make it compatible with short gradient separation while
preserving high selectivity. This combination of DIA and PASEF compensates for the traditional DIA pitfalls: by using a
pattern of m/z isolation windows within consecutive tims events, the percentage of ions used in the dia-PASEF can be
greatly increased. Here, we evaluate benefits of dia-PASEF including library-free data processing for very short gradients
enabling ultra-high sample throughput proteomics. Moreover, we demonstrate the performance of dia-PASEF on low
sample amounts. K562 tryptic digests (Promega) were analyzed by coupling an EVOSEP One (EVOSEP) to a trapped ion
mobility spectrometry – quadrupole time of flight mass spectrometer (timsTOF Pro 2 for short gradient and timsTOF SCP for
low sample loads) using optimized dia-PASEF schemes. Data was processed using a library-free approach with DIA-NN
(Demichev et.al, 2021) incorporated into the PASER software (Bruker) and Spectronaut (Biognosys) using default
settings. Using a 300 samples per day method (300 SPD) with 200 ng on column, 3,600 protein groups and 23,500
peptides were identified with a run time of only 4.8 min. To demonstrate the potential of dia-PASEF for high sensitivity
proteomics we used the timsTOF SCP mass spectrometer in combination with low flow rate delivery from the Evosep
system (Whisper100, 40 samples per day with gradient flow of 100nl/min). With this setup and loading only 125 picograms,
1,249 (±123, n=6) protein groups could be reproducibly identified using dia-PASEF using the library-free based approach.
To further increase proteome coverage for both, high throughput, and high sensitivity proteomics, we will investigate
additional new developments. Our results demonstrate the benefits of using dia-PASEF acquisition for high throughput and
high sensitivity proteomics studies using library-free approaches.
P17.02 Standardized and easy to deploy nanoLCMS for ultra-sensitive and high-throughput analysis of limited
sample amounts and single cells
Oleksandr Boychenko (1), Runsheng Zheng (1), Christopher Pynn (1), Xuefei Sun (2), Alec Valenta (3)
(1) Thermo Fisher Scientific, Germering, Germany
(2) Thermo Fisher Scientific, Sunnyvale, California, United States
(3) Thermo Fisher Scientific, Michigan, Michigan, United States
Proteomics profiling of limited sample amounts like single-cells or cells populations, laser capture microdissected tissue, etc.
requires the maximum sensitivity from ESI interface and MS. In addition, the analysis of individual targets like single cells
needs sufficient throughput to draw meaningful statistically relevant conclusions. The existing approaches for nanoLCMS
analysis of limited samples are not widely adopted due to special setups, non-standard hardware configurations, advanced
technical skills required to make connections and operate nanoLC systems at ultra-low nano-flow rates below 100 nL/min.
P17.03 Multiplexed nanoLC System Produces a Label-Free Single-Cancer-Cell Proteome Every 15 Minutes
Kei Webber (1), Sebastian Zapata (1), Thy Truong (1), Jacob Davis (1), Alex Buttars (1), Yiran Liang (1), Madisyn Johnston
(1), Fletcher Smith (1), Arianna Mahoney (1), Hailey Jones (1), Jacob Heninger (1), Andikan Nwosu (1), Richard Carson (1),
Greg Nordin (1), Ying Zhu (2), Ryan Kelly (1)
(1) Brigham Young University, Provo, Utah, United States
(2) Pacific Northwest National Laboratory, Richland, Washington, United States
Cellular phenotypes can be measured directly using single-cell proteomics (SCP) by mass spectrometry (MS). Such
measurements are more sensitive with extremely low-flow separations (<100 nL/min) because of improved ionization
efficiency. However, the time required for the sample and mobile phase to clear extra-column volumes in these low-flow
systems can lead to inefficient duty cycles wherein the mass spectrometer sits idle for significant amounts of time between
analyses. The resulting low-throughput has hindered the adoption of label-free SCP compared to approaches using isobaric
multiplexing. As label-free SCP has some distinct advantages over isobaric-labeling methods, we sought to address the
low-throughput problem. We here present a multicolumn liquid chromatography (LC) system that more than doubles the
throughput of label-free SCP by using two parallel systems to multiplex sample loading, online desalting, analysis, and
column regeneration steps. Reliance on feature mapping of high resolution MS1 spectra to generated libraries for peptide
identification allowed us to compensate for the Orbitrap’s slower acquisition speed and achieve faster nanoLC separations.
This high-throughput column system showed good reproducibility between the parallel columns with an 8% difference in
median abundances across 15-minute (n = 100) replicate analyses of single-cell-sized aliquots of commercial human cell
lysate tryptic digest, with an average of 937 protein groups identified and a CV of 18.5%. Altering the total analysis times to
7, 10, 15, and 30 minutes resulted in an average of 621, 774, 952, and 1622 protein groups identified, respectively. When
applied to single HeLa cells, we identified an average of 986 protein groups per cell (n=53) using 30-minute cycles. Upon
reducing the cycle time to 15 minutes, we identified an average of 660 protein groups per cancer cell (n=150). This
throughput is easily double that of any previous label-free SCP approach, and begins to approach parity with isobarically-
labeled approaches.
Introduction Single-cell proteomics is a relatively young niche of proteomics compared to single-cell genomics. In recent
years, significant progress has been made in sample handling and boosting the sample signal by multiplexing with mass
spectrometers. timsTOF SCP is the first commercially introduced mass spectrometer for single-cell proteomics. The
modified front-end (orthogonal ion-guiding) of the instrument increases the ion transfer up to five times keeping ultra-high
robustness – the default attribute of the timsTOF platform. Here we demonstrate the performance of the instrument for low
sample loads in the range of 250 pg to 1 ng combined with robust low flow rate delivery from the Evosep system. Methods
Commercially available K562 tryptic peptide digests (Promega) were loaded on the Evotips according to the recommended
protocol provided by the vendor. Peptide amounts from 125 pg to 25.6 ng were used to evaluate the performance of the
instrument. The Whisper low flow method – 40 samples per day (SPD) with a gradient time of about 28 minutes was used
for separating the eluting peptides on a 15 cm column with 75 µm ID coupled to captive spray ionization source using a 10-
micron ID zero-dead volume emitter. Samples were analyzed using a dia-PASEF method tuned for high sensitivity
measurements. Data were processed in Spectronaut 15. Preliminary Data diaPASEF data processing was performed with
Spectronaut 15 using a spectral library of 5,200 protein groups and 54,000 peptides and a further improved library with
9,300 protein groups and 116,000 peptides, respectively. Using the combination of Whisper 40 SPD and the high sensitivity
methods on the timsTOF SCP, we could quantify around 1,500 protein groups from 250 pg sample loads and close to 2,000
protein groups from 500 pg loads. Performing a dilution series experiment, around 4,500 protein groups could be quantified
in 6.4 ng peptide loads.
Systematic Lupus Erythematosus (SLE) is an autoimmune disease that leads to widespread inflammation. SLE biomarker
discovery at an early stage is critical for effective treatment, and avoidance of irreversible tissue/organ damage. Plasma is a
key clinical sample type for biomarker discovery because its collection is non-invasive, and it circulates among all organs to
reflect downstream damage. High-throughput, quantitative MS-based proteomics would be advantageous for plasma
biomarker discovery, but this analysis is challenging due to the wide dynamic range of plasma protein abundance. Top-
down proteomics techniques can discern unique proteoforms that may also be relevant biomarkers but could become
indistinguishable using typical bottom-up proteomics. Here, we present a novel quantitative top-down proteomics platform
that integrates simple, precipitation-based protein depletion, protein-level TMT labeling, and top-down MS for intact plasma
proteome profiling for SLE biomarker discovery. Our results indicated that a 1:2 (v/v) plasma-to-isopropanol depletion
precipitated most of the albumin and enriched low abundance proteins between 10-50 kDa. Depleted SLE and healthy
control samples were TMT-labeled using our recently optimized protein-level TMT labeling conditions and analyzed by
nano-LC-MS/MS. All proteoforms were deconvoluted by BioPharma Finder and identified by TopPIC Suite. In total, 2727
unique proteoforms were detected. Reporter ion intensities were also extracted from the raw files using an in-house python
code and a volcano plot was constructed that showed 47 proteoforms were upregulated in SLE samples compared to
controls (cutoff: P-value ≤ 0.05, fold-change ≥ 2). Some PTMs were enriched in SLE samples, such as glycation that has
been reported to be related to chronic inflammation in SLE and the disturbance of the oxidative/antioxidative balance. This
oxidative stress leads to intensified advanced glycation end products (AGEs) formation. Overall, our results demonstrate
that optimized organic precipitation and top-down TMT protocols can be applied to plasma samples for potential biomarker
discovery in SLE.
Post-translational modifications (PTMs) contribute to the vast diversity of proteoforms that drive biological processes.
Currently, very little is known about the molecular heterogeneity of proteoforms. This gap can be dominantly attributed to a
lack of methods to analyze PTMs on intact single protein molecules on a proteome-scale. Here, we introduce an antibody-
based single-molecule analysis platform that includes novel biochemistry for single-protein molecule immobilization,
instrumentation for highly sensitive detection, and computation for data interpretation. We use this platform to study Tau
proteoforms. Aberrant PTMs on Tau generate a wide diversity of proteoforms that lead to Tau misfolding and aggregation
and are associated with neurodegenerative disorders collectively referred to as Tauopathies. First, we validate the platform
by measuring defined mixtures of recombinant tau proteins. Next, we use the platform to examine tau protein enriched from
human induced pluripotent stem cell (iPSC)-derived neurons. As PTMs can be assigned to individual protein molecules, the
platform is able to reveal the molecular heterogeneity of tau proteoforms that is missed by both bulk measurements and by
peptide-centric proteomics approaches. Ultimately, quantitative analysis of proteoform molecular heterogeneity will lead to a
more detailed view of the tau proteoform landscape towards advancing precision medicine of tauopathies, such as
Alzheimer’s disease.
P18.04 A survey of truncated proteins across humans and plants reveals novel proteoforms and proteoform-
dependent protein complexes
Claire McWhite (1), Wisath Sae-Lee (2), Nicholas Gort Frietas (3), Anna Mallam (2), Edward Marcotte (2)
(1) Princeton University, Princeton, New Jersey, United States
(2) The University of Texas at Austin, Austin, Texas, United States
(3) Harvard Medical School, Cambridge, Massachusetts, United States
Variability of proteins at the sequence level creates an enormous potential for proteome complexity. Here, we systematically
survey human and plant high-throughput native proteomics data for abundant truncated variants of protein, where
substantial regions of the full-length protein are missing from an observed protein product. In all surveyed species,
approximately one percent of proteins across show a short form, which we can assign by comparison to RNA isoforms as
either deriving from transcriptional processes or limited proteolysis. Some truncated proteins are observed in both humans
and plants, suggesting that they date to the last eukaryotic common ancestor. While many of our detected protein fragments
align with known splice forms and protein cleavage events, multiple examples are previously undescribed. We find that
truncations occur almost entirely between structured protein domains, even when short forms are derived from transcript
variants. Intriguingly, multiple endogenous protein truncations of phase-separating translation-related proteins are similar to
fragments produced by enteroviruses during infection. We describe common features of proteins with clear short forms,
finding new classes of proteins with potential bi-functionality or switch-like behavior. Finally, we describe novel proteoform-
specific protein complexes.