Professional Documents
Culture Documents
Transciptome D24
Transciptome D24
Plant Genetics
Research Article
Nurul Arneida Husin1,5 , Sadequr Rahman2, Rohini Karunakaran3,4 and Subhash Janardhan Bhore1
1
AIMST University, Faculty of Applied Sciences, Department of Biotechnology, Kedah, Malaysia.
2
Monash University Malaysia, School of Science and Tropical Medicine and Biology Platform, Malaysia.
3
AIMST University, Faculty of Medicine, Unit of Biochemistry, Kedah, Malaysia.
4
Institute of Bioinformatics, Saveetha School of Engineering, Department of Computational Biology,
Chennai, India.
5
Monash University Malaysia, Jeffrey Cheah School of Medicine and Health Sciences, Malaysia.
Abstract
Durian (Durio zibethinus Murr.) fruits are famous for their unique aroma. This study analysed the Durian fruit
transcriptome to discover the expression patterns of genes and to understand their regulation. Three developmental
stages of Durian fruit, namely, early [90 days post-anthesis (DPA)], mature (120 DPA), and ripen (127 DPA), were
studied. The Illumina HiSeq platform was used for sequencing. The sequence data were analysed using four
different mapping aligners and statistical methods: CLC Genomic Workbench, HISAT2+DESeq2, Tophat+Cufflinks,
and HISAT2+edgeR. The analyses showed that over 110 million clean reads were mapped to the Durian genome,
yielding 19,976, 11,394, 17,833, and 24,351 differentially expressed genes during 90-127 days post-anthesis. Many
identified differentially expressed genes were linked to the fruit ripening processes. The data analysis suggests that
most genes with increased expression at the ripening stage were primarily involved in the metabolism of cofactors
and vitamins, nucleotide metabolism, and carbohydrate metabolism. Significantly expressed genes from the young
to mature stage were mainly associated with carbohydrate metabolism, amino acid metabolism, and cofactor and
vitamin metabolism. The transcriptome data will serve as a foundation for understanding Durian fruit development-
specific genes and could be helpful in fruit’s trait improvement.
Keywords: Durian, Durio zibethinus, ethylene, fruit ripening, transcriptome.
Received: December 23, 2021; Accepted: October 18, 2022.
family 29, SEPALLATA transcription factor family, and anticipated data output from HiSeq Illumina sequencing was
ethylene-related genes such as ACS (aminocyclopropane- 16.7 million reads or 5 GB per sample.
1-carboxylic acid synthase), a key ethylene-production
enzyme, are involved in ripening (Teh et al., 2017). On Raw read processing
the other hand, high-throughput sequencing has not been After sequencing, the reads (raw reads) contain low-
employed to analyse differential gene expression during quality reads and adapters, which will affect the analysis
development or to compare stages of Durian growth. Hence, quality. Therefore, it is crucial to filter the raw reads and get
the data reported in this paper will contribute to a better clean reads. Reads containing adapters, sequences with N less
understanding of durian fruit development. than 10 % (base cannot be determined), and sequences with
The ultimate goal of this study was to contribute (Qscore <= 5) base, which is over 50 % of the total base, were
fundamental knowledge about durian fruit development and filtered and removed. The quality of the raw data was assessed
ripening from a gene expression perspective. We have investigated using the FastQC software. The FastQC results showed per tile
the transcriptomic variations in Durian by performing differential sequence quality, base sequence content, sequence duplication
gene (DEG) discovery and gene expression characterisation level, overrepresented sequences, and adapter content. The
in edible fruit pulp in its young, mature, and ripened stages. pre-processing step was crucial to ensure the raw data was
We identified genes involved in ethylene biosynthesis, fruit cleaned before downstream analysis. In addition, the removal
softening, and other metabolic pathways essential for the of adapter contamination in sequence reads was a necessary
development and ripening of durian fruit. step. FastQC (Andrews, 2010) and Trimmomatic (Bolger et
al., 2014) tools were used to check and improve the quality
Material and Methods of the raw transcriptome.
Figure 1 − Durian fruits and pulp tissue. (a-b) Durian fruits at its young stage of growth and development (90 days); (c-d) Durian fruit at its mature stage
of growth and development (120 days); and (e-f) Durian fruit at its ripening stage of growth and development (127 days).
Functional annotation of transcripts were used for their downstream analysis. The high-throughput
and pathway analysis sequence data for nine samples were submitted to GEO (Gene
The Blast2GO 5 (Conesa and Götz, 2008) Basic software Expression Omnibus), at NCBI. Transcriptome datasets are
was used. The differential group sequence in FASTA format available through the NCBI under the accession number series
was uploaded to the software. Then, the pathway analysis GSE136290 (release date: March 31, 2023).
was conducted with the KEGG Blast2GO function. All genes
Comparison of all mapping tools
that have a q-value less than 0.05 were added to the GO
categories “biological process,” “molecular function,” and Three different mapping methods, namely CLC Mapping,
“cellular components.” TopHat, and HISAT2 were used for mapping the reads to the
Durian reference genome. The number of mapping reads using
Results CLC is 49.82-59.42 %, TopHat 84.5-90.6 %, and HISAT2
88-95 %. Figure 2 shows the graphical representation of
Sequencing, annotation, and mapping mapped and unmapped reads using three mapping tools.
to the Durian genome CLC mapping shows lower mapping rates, while TopHat and
The sequencing was carried out with the Illumina Hiseq HISAT2 offer high and competitive mapping rates. Overall,
150 PE Run by a commercial service provider. The data output HISAT2 outperformed TopHat significantly in aligning FASTQ
was around 16.7 million reads per sample, equal to 5 GB. reads to the Durian genome. TopHat and HISAT2 are a type
Total raw reads were 193,416,352 M, and filtered reads were of splice-aware aligners suitable for RNA reads. This can
166,709,698 M, with a ≥ Q30 percentage. The quality of the raw split reads at intron-exon boundaries. The high mapping rate
data was assessed with the FastQC software. Trimmomatic and could be due to the tools’ ability to handle the RNA sequences’
Fast QC were tools used to check and improve the quality of splicing sites. Also, HISAT2 is an improved TopHat version
the raw transcriptome. In total, 110,351,584 M clean sequences (Trapnell et al., 2009, Kate Shannon, 2016).
4 Husin et al.
Figure 2 − Comparison of the percentage of mapped and unmapped reads using CLC Mapping, TopHat, and HISAT2. (a) CLC Mapping, (b) TopHat
Mapping, and (c) HISAT.
Compared with other research, the selection of mapping total variation of PC1 and PC2 for YSMS was 95 %. There
tools has not given any significant indicator to determine the was a strong correlation in their particular groups at the YS
correct mapping for the aligned reads. Raplee et al. (2019) vs. ripening stage (RS). The total variation of PC1 and PC2
stated that the generally low proportion of aligned reads for YSRS was 91 %. The total variation of PC1 and PC2 for
for all input reads for HISAT2 and STAR is likely due to MSRS was 100 %. The PCA for all samples in three different
its quality. They also suggest that many input reads were groups was calculated. Principal component 1 (PC1) and
poly(A) sequences, Illumina adapter sequences, and reads principal component 2 (PC2) were identified by variance
from libraries are too uninformative for accurate mapping stabilising transformation in all three groups. The percentage of
to correspond to the very 3-end of mRNAs (Raplee et al., variance indicates how much variance was explained by PC1
2019). This study also contrasts with other previous studies, and PC2. These comparative principal component analyses
which reported that a mapping method has a limited impact demonstrate that biological replicates from all differential
on the final analysis of DE genes (Costa-Silva et al., 2017). groups are highly similar.
Table 1 shows the mapping tools’ performance in this study.
The mapped reads’ output was used to compare the number of Comparative transcriptome analysis and differential
DEGs analyzed with four selected methods. It is also known gene expression using merged DEGs
that selecting a suitable mapping tool relies on a variety of RNA-seq was used with the Illumina platform technology
other variables, such as RNA quality, cDNA libraries, genome to explore dynamic changes in gene expression in durian fruit
size, intron length, etc... pulp tissues from three developmental stages, young, mature,
and ripening. Various techniques are available to assess gene
Principal component analysis (PCA) for differential expression, each of which produces distinct results for rapid
expression groups RNA-seq growth. However, there are no clear rules as to which
Each differential studied group was subjected to PCA technique yields the best accurate gene expression estimations.
analysis to identify whether the samples clustered within In this study, four statistical techniques were used to produce a
each group or with other groups. At the young stage (YS) more accurate list of DE genes. Other researchers used similar
vs. mature stage (MS), the PCA results show that the YS techniques to study the transcriptomic data (Rajkumar et al.,
and MS samples were clustered together in their group. The 2015; Costa-Silva et al., 2017).
Table 1 − Statistics of the reads and the mapping percentage using CLC Mapping, TopHat, and HISAT2 in the present study.
YS, young stage; MS, mature stage; RS, ripening stage; DPA, days post anthesis.
Durian fruit transcriptome 5
CLC was used to investigate the utilisation of commercial Five different genes, namely, LOB domain-containing
software for Durian RNA seq experiments in this study. protein 18 (LOC111304630), LOB domain-containing protein
CuffDiff was chosen in conjunction with Top Hat, as it is 42 (LOC111282412), LOB domain-containing protein
developed explicitly from transcripts, spliced regions, and 40 (LOC111290435), LOB domain-containing protein 1
promoters for DGE analysis. In the comparison, DESEq2 and (LOC111297059), and LOB domain-containing protein 19
edgeR were combined to investigate the gene-level expression, (LOC111305909), were expressed during the early stage of
as this method works by integrating the raw reads using Ht- durian fruit pulp development. These five genes are from the
seq count. If the absolute value of the expression log2 fold LOB (lateral organ boundaries)-domain gene family. Durian’s
changes > 1.5 and < -1.5, the differentially expressed genes LOB gene family members exhibited a 9-fold increase in
and the transcript were considered statistically significant. expression. Several studies have suggested that LBD genes
Only those genes with expression identified as meaningful have a broad range of functions (Luo et al., 2016). Genes
with q value of < 0.05 (5 % of a gene for each comparative from the plant-specific Lateral Organ Boundaries Domain
group, false rate of discovery, FDR) were considered. After (LBD) family encode transcriptional regulators with roles in
mapping the reads to the reference genome using one of CLC various physiological and developmental processes in plants
mapping, TopHat or HISAT2, transcripts were assembled (Luo et al., 2016).
in separate events using the CLC assembler, Cufflink, and One of the MGL gene isoforms, namely methionine
StringTie. Table 2 highlighted the number of significant genes gamma-lyase-like (LOC111287425), responsible for the
identified in up and down-regulation among the four studied Volatile sulfur compounds (VSC) ’s production, was up-
methods. Among four methods used for RNA - seq analysis, regulated during the transition from the young to the mature
edgeR detected more DE genes in the durian transcriptome. stage. In a differential analysis of Durian, this gene is found
This finding is similar to other researchers’ findings, as they up-regulated with a fold change of 10.163. This gene is
reported that edgeR had identified more DEGs than other tools involved in the transsulfuration metabolic pathway, where
used (Fischer, 2003; Kvam et al., 2012; Seyednasrollah et al., sulfur transfer from homocysteine to cysteine occurs. The
2013; Gallego Romero et al., 2014; Rajkumar et al., 2015). pathway leads to the generation of several sulfur metabolites,
including cysteine, GSH, and the gaseous signalling molecule
Differentially expressed genes in Durian during hydrogen sulfide (H2S) (Sbodio et al., 2019).
its fruit development ATP- binding cassette (ABC) transporters are essential
Genes are differentially regulated to bring about changes for plant development in processes such as gametogenesis,
in durian fruit from a young to a mature stage. The differentially seed development, seed germination, organ formation,
expressed genes were identified in the young and mature secondary growth, protective layer formation, and
stages (Tables S1 and S2). phytohormone transport (Ha et al., 2017). In this study,
Pectinesterase 2-like (LOC111304600) had an 11- four different ABC transporter gene families, namely, ABC
fold increase in expression, and the endoglucanase-like transporter C family member 13 (LOC111274386), ABC
(LOC111278395) gene had a 10-fold rise in expression. Both transporter G family member 31 (LOC111301322), ABC
of these genes are involved in cell wall metabolism. Glucan transporter G family member 14 (LOC111286089), and ABC
endo-1,3 beta-glucosidase 11-like (LOC111287487) and transporter G family member 11 (LOC111316978) were found
cellulose synthase-like protein G2 (LOC111280213), both significantly up-regulated during the early stage. Members
implicated in the starch-sucrose breakdown, had a 9-fold and of the ABC transporter gene family in Durian exhibited a
6-fold expression, respectively. 10-fold increase in expression.
The analysis suggested that the Polygalacturonase
inhibitor-like (LOC111311552) gene was expressed highly (an Differentially expressed genes in Durian during
8-fold increase in expression) and is responsible for protein its fruit ripening
binding. Polygalacturonase inhibiting proteins (PGIPs) are cell Genes that play an essential role in the Durian fruit
wall proteins that inhibit the pectin-depolymerising activity pulps’ growth and development and are responsible for the
of polygalacturonases and play an essential role in the plant transition from a mature to a ripening stage are depicted in
defence system (Kalunke et al., 2015). Tables S3 and S4.
Table 2 − Total number of identified DEGs using four tools of CLC, CuffDiff, DESEq2, and edgeR.
Four family members of xyloglucan endotransglucosylase Venn analysis of all differentials expressed genes
(XTH) were present, namely, xyloglucan endotransglucosylase/ Venn analysis was carried out for all differential expressed
hydrolase protein 33 (LOC111317417), xyloglucan genes to observe the specific and overlapping sequences among
endotransglucosylase/ hydrolase 2- like (LOC111287953), the three stages. Figure 3 shows the intersection among YSMS
xyloglucan endotransglucosylase/ hydrolase protein 6 and YSRS and MSRS, with 67 up-regulated and 42 down-
(LOC111316166), and xyloglucan endotransglucosylase/ regulated genes overlapped in all their gene sets. The Venn
hydrolase protein 6 (LOC111292968). These XTH family diagram analysis revealed a total of 937, 504, and 1,666 of the
members were directly annotated to the term cellular glucan specific differentials upregulated expressed genes, respectively,
metabolic process. The most significant increase in the in the young, mature, and ripening stages of growth. The
expression during durian ripening was observed in this Venn diagram analysis showed a total of 1,965, 724, and 635
XTH gene family. Similar observations were also reported of the specific differentials downregulated expressed genes,
for the banana and papaya fruit softening process (Asif respectively, in the young, mature, and ripening growth stages.
et al., 2014; Yao et al., 2014). Previously reported gene
families include expansins, pectate lyases, and xyloglucan Functional analysis of differentially expressed genes
endotransglycosylases responsible for softening banana fruit Blast2GO 5 Basic was used to identify the biological
(Ha et al., 2017) Most of the genes involved in cell wall significance of each group’s transition stage. The entire DEGs
hydrolysis are also members of multigene families, and many from each group generated using CuffDiff were subjected to
have highly specialised functions in cell wall metabolism GO analysis to achieve a broader functional characterisation.
(Asif et al., 2014). The GO analysis was classified into three major categories,
Three family members, namely, glucan endo-1,3 namely, biological process (BP), molecular function (MF),
beta-glucosidase (LOC111285741), glucan endo-1,3 beta- and cellular component (CC)). The total DEGs mapped to
glucosidase 12-like (LOC111275649), and glucan endo- the three GO categories for YSMS, YSRS, and MSRS were
1,3 beta-glucosidase (LOC111295531) were among the 6,735, 3,898, and 7,310 (Table 3). All selected transcripts with
expressed transcripts. They are known for their hydrolysis q value > 0.05 were uploaded to Blast2GO for annotation to
activity (Hrmova and Fincher, 2001; Singh et al., 2016). assign the putative functions to the DEGs. After running with
These genes and galactinol-sucrose galactosyltransferase InterProscan and Annex, the transcripts were assigned to the
5 (LOC111316628), beta-amylase 3, and chloroplastic-like GO terms within three main categories (BP, MF, and CC).
(LOC111305012) are also involved in sugar metabolism, and In developmental stage YS/MS, for biological process
responsible for the degradation of starch to sugar. (BP), major sub-categories were 1,416 DEGs in the metabolic
Polygalacturonase (PG), LOC 111311521, was process (GO: 0008152), followed by 1,401 DEGs in the
significantly expressed in durian. It is responsible for fruit cellular process (GO: 0009987) and 395 DEGs in response
softening (García-Gago et al., 2009; Anand et al., 2018). Its to stimulus (GO: 0050896). As for molecular function (MF),
high expression of 12.6-fold was noticeable while doing the binding (GO: 0005488) and catalytic activity (GO: 0003824)
data analysis. This gene is also involved in starch and sucrose were the highest genes mapped to these GO terms, with a total
metabolism. Other significant up-regulated genes involved in number of 1,427 and 1,137, respectively. Cell (GO: 0005623),
the hydrolysis activities were carboxylesterase SOBER1-like cell part (GO: 0044464), and organelle (GO: 0043226) with
(LOC111275243), endoglucanase-like (LOC111278395), and total DEGs of 1,297, 1,272, and 1,049 were enriched in cellular
beta-glucosidase 11 (LOC111305070). The findings on Carica component (CC). In YS/RS, the major sub-categories for
papaya L. var solo eight (8) fruits at the post-harvest stage biological process (BP) were 862 for metabolic process, 806
reported by Yao et al. (2014) reveal that the loss of the papaya’s for cellular process, and 227 for response to the stimulus. As for
firmness was positively related to the hydrolytic enzyme molecular function (MF), binding (GO: 0005488) and catalytic
activities and the sweet taste of the presence of simple sugars activity (GO: 0003824) were the genes mapped to these GO
such as galactose liberated from the polysaccharide complexes. terms with a total number of 730 and 597, respectively. Cell
The data analysis suggests that Sulfate transporter (GO: 0005623), cell part GO: 0044464), and organelle (GO:
1.3, SULTR (LOC 111294357) is also activated during 0043226) with total DEGs of 775, 746, and 533 were enriched
durian ripening. It is involved in sulfur metabolism (Leustek in cellular component (CC). The number of DEGs for all GO
et al., 2000; Gigolashvili et al., 2014). Sulfur is required terms was lower compared to YS/MS. In late-stage MS/RS,
for the biosynthesis of proteins, co-enzymes, prosthetic 1509, 1415, and 393 DEGs represented major sub-categories
groups, vitamins, amino acids like Cys and Met, GSH, and in biological process (BP) for metabolic process, cellular
secondary metabolites such as GSL and sulfoflavonoids. process, and response to the stimulus. As for molecular
Sulfate transporters are the most prominent S-metabolite function (MF), binding (GO: 0005488) and catalytic activity
transporters in plants, as sulfate is the primary source of sulphur (GO: 0003824) were the highest genes mapped to these GO
taken from the soil. It is the most abundant S-containing terms, with a total number of 1,571 and 1,094, respectively.
metabolite in plant cells (Gigolashvili et al., 2014). The Cell (GO: 0005623), cell part (GO: 0044464), and organelle
process is assisted by several sulfate transporter members (GO: 0043226) with total DEGs of 1,354, 1,269, and 969
(SULTR) (Pinsorn et al., 2018). were enriched in cellular component (CC).
Durian fruit transcriptome 7
Figure 3 − a. Venn diagram for upregulated DEGs from three comparisons, namely YS/MS, YS/RS, and MS/RS. A total of 5,134 genes were expressed in
all three stages. The number of tissue-specific genes are 1,965, 724, and 635, respectively. b. Venn diagram for upregulated DEGs from three coparisons,
namely YSMS, YSRS, and MSRS. A total of 4,087 genes were expressed in all three stages. The number of tissue-specific genes are 937, 504, and
1,666, respectively.
Table 3 − Analysis of GO terms for differently expressed sequences. Enrichment of the KEGG Pathway
GO:0008150 GO:0003674 GO: 0005575 The Kyoto Encyclopedia of Genes and Genomes
Differential
Biological Molecular Cellular (KEGG) platform (https://www.genome.jp/kegg/) hosts various
Group
process Function Component databases and provides tools for a systematic annotation and
YS/MS 1,870 2,876 1,989 analysis of cell metabolic pathways and functions of the gene
YS/RS 1,149 1,646 1,103 product. To further investigate the functional studies, all these
MS/RS 2,038 3,160 2,112 up-regulated DEGs were enriched and annotated using the
KEGG database to obtain the link to the Enzyme Commission
YS, young stage; MS, mature stage; RS, ripening stage. number (EC). The EC was finally mapped to the relevant
KEGG pathway to obtain the annotated pathway maps. The
In the developmental stage of growth from young to mature percentage of the mapped reads to the KEGG pathway for
(YS/MS), a total of 210 DEGs mapped to the developmental each differential group was 96.50 % (YS/MS), 60.20 % (YS/
process (GO: 0032502) and 210 DEGs in anatomical structure RS), and 35.30 % (MS/RS). Table 4 shows the number of
development (GO: 0048856). Both GO terms were not present the mapped pathway, the number of mapped sequences, and
in the MS/RS. The DEGs are important for durian fruit growth the EC number for each differential group. A total of 2,705
and development. Several other GO terms of BP, MF, and DEGs were assigned to 382 EC. The EC were subsequently
CC that were not active in the developmental stage were grouped into 284 biochemical pathways.
localisation, catabolic process, and establishment of localisation, The top five KEGG class results from the young to
transport, non-membrane-bounded organelle, and intracellular mature stage data analysis indicate that most transcript-
non-membrane-bounded organelle. The highest DEGs of GO encoding enzymes were linked to carbohydrate metabolism
term associated with response to stress were in the ripening (254), metabolism of cofactors and vitamins (200), amino
stage, MS/RS with 291, followed by YS/MS with 277 DEGs. acid metabolism (176), nucleotide metabolism (142) and
As shown in Figure 4, the developmental process, biosynthesis of other secondary metabolites (82). Carbohydrates
cellular component organization, anatomical structure are highly essential due to its contribution to texture, flavour,
development, protein modification process within the color, and nutritional value in horticultural commodities (Yahia
biological process, and RNA binding within the molecular et al., 2019). The classification suggests that the highest
function were the unique groups in YSMS. Localization, number of enzyme activities in Durian growth in the young
catabolic process, the establishment of localization and stage were related to carbohydrate metabolism. This indicates
transport within the biological process, and non-membrane- that many pathways related to carbohydrate metabolism were
bounded organelle, an intracellular non-membrane-bounded activated to support the durian fruit’s growth and development.
organelle within the cellular component, were the unique Understanding the gene expression patterns in carbohydrate
groups in MSRS. This finding revealed that the DEGs metabolism will facilitate the investigation of durian fruit
classified into the groups might be playing a tissue-specific quality. Durian fruits development involves the accumulation
role during fruit development (Table S5). of starch and sucrose. According to KEGG, 254 genes were
8 Husin et al.
associated with carbohydrate metabolism, and these genes 2-fold change in expression level, indicating its role in fruit
were classified into several metabolic pathways. The highest pulp softening. Endoglucanase also showed a high level (9-
number of expressed genes during the early fruit development fold) of change in its expression.
were from starch and sucrose pathways. Starch is present in Other softening genes that showed more than 2-fold
unripe durian, mainly in its pulp, which is transformed into expression change in the ripening stage were pectinesterase,
sugars during the ripening process. These sugars contribute U-box domain, xyloglucan endotransglucosylase/hydrolase,
to the level of sweetness of the fruits. In the transition from beta-galactosidase, cellulose synthase, endonuclease 1,
the young to the ripening stage, GO analysis produced 4,154 expansin-A (also called α-expansins), beta-glucosidase, and
sequences mapped with GO terms, and 2,248 transcripts pectin. It appears that more genes were expressed during the
encoding enzymes were annotated. ripening stage than the developmental stage of durian fruits.
The top five KEGG class results showed that the
expressed genes during the transition from the young to Ethylene synthesis and signal transduction
ripening stage were mainly for the metabolism of cofactors pathway in Durian
and vitamins (109), carbohydrate metabolism (102), lipid The molecular level of ethylene-controlled ripening
metabolism (77), nucleotide metabolism (75), and amino acid in Durian has not been investigated in detail. Most of the
metabolism (65). The pathways for nitrogen and glutathione studies conducted refer to a single gene or a single family
metabolism contained the greatest number of expressed genes. of genes. The genes involved in ethylene biosynthesis and
During the transition from the mature to ripening stage, signal transduction are significantly important in the growth
GO analysis produced a total of 7,629 sequences mapped with and development of Durian fruit. Ethylene biosynthesis
GO terms, and of these, 4,096 transcripts encoding enzymes were is essential for the ripening of fruits; hence, the data were
annotated. The top five KEGG class results were the metabolism analysed to study the genes involved in this process. Durian
of cofactors and vitamins (200), nucleotide metabolism (192), shows a burst of respiration and a gaseous hormone ethylene
carbohydrate metabolism (85), lipid metabolism (61), and biosynthesis at the beginning of its ripening, which regulates
amino acid metabolism (56). During the late stage of fruit the fruit-maturing aspects.
development, the sulphur metabolism pathway had the highest The genes involved in ethylene biosynthesis, signalling
number of expressed genes, indicating the durian’s strong odour. pathway, and regulatory response process were identified from
In contrast, the classification of transcripts suggests that the transcriptome data (Table S8). S-adenosyl-L-methionine
the highest number of metabolism of cofactors and vitamins (SAM) synthase converts methionine to SAM with a 2-fold
outperformed the other pathways in the late stage of Durian expression during fruit ripening (Bouzayen et al., 2010). Several
growth. In KEGG, 180 genes were identified to be associated genes linked with ethylene biosynthesis have been identified
with thiamine metabolism to support the nutritional quality and with six ACS gene members. The ACS genes are involved
health benefit of durian fruit, mainly in the energy metabolism in converting SAM to 1-aminocyclopropane-1-carboxylic
(Figure 5 and Table S6) (Lonsdale, 2006). acid (ACC) with 2-fold expression in both stages. Most
Table 4 − No of mapped pathways and sequences. pulp samples, Durio zibethinus methionine gamma-lyase-like
No of Mapped No of Mapped (LOC111287425), transcript variant X1, mRNA, D. zibethinus
Stage No of EC methionine gamma-lyase-like (LOC111308026), transcript
Pathway Sequences
YS/MS 96 1,279 114 variant X3, misc_RNA, and D. zibethinus methionine gamma-
lyase-like (LOC111290112), partial mRNA, were found,
YS/RS 85 689 95
respectively. MGL genes abundantly expressed in the mature
MS/RS 103 737 173
and ripening stage were D. zibethinus methionine gamma-
YS, young stage; MS, mature stage; RS, ripening stage. lyase-like (LOC111308047), mRNA, and D. zibethinus
methionine gamma-lyase-like (LOC111315809), mRNA.
The correlation matrix was performed to compare the five
Role of MGL genes and VSC production in Durian` MGL gene copies to observe its expression value in three
To study the biological process related to the Durian developmental stages. There were two major clusters, major
fruit’s strong aroma, the heatmap analysis between isoforms cluster 1 (C1) comprised of 2 minor clusters (C1a and C1b).
of MGL (methionine gamma-lyase-like) gene responsible Major cluster 2 comprised of C2, the outlier. C1a showed a
for volatile sulphur compounds (VSC) production was relationship between MGLa1 and MGLa2. Both MGL genes
carried out. The VSCs have been identified as important in cluster C1a showed a varying high level of expression
factors determining the degree of aroma and taste of Durian value in the ripening stage. While C1b comprises MGLb1
fruit pulp. MGL enzymes break methionine and cysteine and MGLb2, it showed a similar expression pattern from
into methanethiol, ethanethiol, and ethionine (Fischer and the mature to ripening stage. MGL genes were up-regulated
Steinhaus, 2019). from the mature to the ripening stage. There were two
MGL genes produce VSC. Durian fruit is unique in the transcripts under MGLb2; namely XM_022906389.1 and
sense of its aroma conferred by VSCs. A study on the Durian XM_022906396.1. Major cluster C2 was an outlier, as this
genome reveals the presence of four copies of MGL, which MGL gene, MGLc, showed an uneven expression level only
contributes to making the Durian fruit pulp aroma strong (Teh in the young stage of growth. This gene was downregulated
et al., 2017). There are two main groups of volatile compounds; in the mature to ripening stage. Under this gene, there were
sulfur-containing volatiles, such as thiols, disulfides, trisulfides, three transcripts; XM_022877994.1, XM_022878003.1 and
and esters associated with a fruity aroma. MGL genes control XM_022877983.1. Overall, the highest-level expression value
the VSCs level. A distinctive sulfuric aroma can be smelled of the MGL gene was expressed in MGLb1 in the transition
in durian fruit pulp due to VSCs produced by the MGL stage from mature to ripening.
enzyme (Teh et al., 2017). Seven low molecular weight The structural analysis of MGL proteins published in
sulfurs containing compounds are reported in durian fruit the 2018 International Conference on Tropical Fruit Pests
pulp, including hydrogen sulphide, methanethiol, ethanethiol, and Diseases suggests that all four MGL genes in Durian
and propane-1-thiol (Cannon and Ho, 2018). have 89 % to 93 % similarity at the protein sequence level.
While annotating transcripts, we noticed that Durian Phylogenetic analysis demonstrated that the MGL genes in
fruit pulp contains five different MGL isoforms expressed in Durian are evolutionary-related with high genetic diversity,
three different stages of growth, YS, MS, and RS. One outlier separating MGLb_1 into different clades (Yusof, 2018). The
of the MGL gene, MGLc (LOC111287425), is found in the active sites in MGL protein sequences are highly conserved.
early stage of Durian fruit pulp. However, the regions proximal to the active sites in MGLb_1
A Heatmap showing the gene abundance of MGL showed more residue substitutions than others. Their sequence
isoforms (FPKM) in YS, MS, and RS is depicted in Figure 8. analysis also showed that LOC111315809, MGLb_1 (accession
This heatmap was generated using StringTie. The abundance number XP_022773569) contained more residue substitutions
of each reference gene was estimated with StringTie. The than other MGL genes. This gene expression was highly
output of FPKM (Fragment per kilobase) and TPM (Transcript abundant in the ripening stage, suggesting that it produces
per kilo base million) values were calculated based on the more VSCs than other isoforms.
normalisation of the sequencing depth and gene length. Furthermore, they also suggested some possible residue
The highest FPKM or TPM value was considered the most substitutions in MGLb_1 that may affect the activity of MGL
abundant genes for the particular sample. FPKM and TPM in the recognition and elimination of methionine, which
values were used to compare the relative gene expression eventually contributes to the pungent odour of ripening
levels within a sample. Durian fruit (Yusof, 2018). Our findings were similar.
We found that durian fruits possess various isoforms MGLa2 (LOC111315809) was expressed abundantly in the
of the MGL genes. The MGL gene isoforms are abundantly ripening fruit pulp tissue samples. These findings explain
expressed. The transcripts were found in young, mature, and why the strong aroma of Durian is felt when fruit reaches
ripening fruit pulp samples. In young, mature, and ripened fruit the ripening stage.
Durian fruit transcriptome 11
Figure 5 − Classification of the sub-categories of KEGG pathways in a) YSMS, b) YSRS, and c) MSRS.
12 Husin et al.
Figure 6 − Hierarchical clustering analysis, heatmap, and key regulatory genes involved in Durian fruit softening development and ripening stage. Blue
and orange bands indicate the high and low expression of genes, respectively. The color scale representing log2 fold change values is shown; young to
mature stage (YSMS), young to ripening stage (YSRS), and mature to ripening stage (MSRS).
Figure 7 − Selected members of gene families involved in ethylene biosynthesis and signal transduction in durian fruit during its developmental and
ripening stages. The colour scale is representing the log2 fold change values.
Durian fruit transcriptome 13
Figure 8 − Heatmap showing the expression of MGL (Methionine gamma lyases) gene isoforms in three different growth stages of durian fruit pulp.
YS represents for the young stage; MS represents the mature stage, and RS represents ripening stage. The FPKM value and Pearson’s distance are used
as the metric.
tools used. As a result, a comparison of the top 50 highly in ripened jackfruit (Hu et al., 2016). Some of the genes
significant genes revealed the similarities and dissimilarities identified in jackfruit are also expressed in Durian fruit pulp
between these four pipelines. Therefore, it is advantageous to tissue. The commonly expressed genes, namely 1,4-alpha-
combine the results of different tools to obtain a more detailed glucan-branching enzyme 2-1, chloroplastic/amyloplastic-like,
and reliable DEGs results output. fructokinase-like 1, chloroplastic, and hexokinase-3-like,
Previous studies published by other research groups also are expressed during the transition from young to mature
support selecting top candidate genes to explore further using stage. Similarly, during the transition from the mature to
several tools for identifying DEGs (Costa-Silva et al., 2017; ripening stage, genes, namely hexokinase-3-like, sucrose-
Raplee et al., 2019). Schurch et al. (2016), recommended to phosphatase 2-like, probable sucrose-phosphate synthase 1,
use edgeR for samples of less than 12 replicates due to the glucan endo-1,3-beta-D-glucosidase-like, and fructokinase-like
superior combination of a truly positive and false positive. 1, chloroplastic are also expressed in Jackfruit (Hu et al., 2016).
DESeq2 is recommended to use for higher replicates of As a whole, this study has successfully provided the
samples. Only DESeq2 and edgeR (TMM) can maintain a transcriptome data and gene expression data specific to
reasonable false-positive rate in the presence of top-count young, mature, and ripened developmental stages of Durian
genes without any loss of power. fruit pulp. The research findings reported in this paper will
Results suggest that the total number of identified up serve as preliminary genomics expression data, which may be
and down-regulated expressed genes in all groups is higher helpful in developing a new strategy for genetic engineering
using DESeq2 and edgeR compared with CLC and CuffDiff. to create a new Durian variety with desirable traits beneficial
The highest range of overlapping gene similarities between for consumers and Durian growers.
DESEq2 and edgeR is 72.1-85.4 %. Unknown genes and
functions dominate most of the top DEGs associated with Acknowledgements
DESeq2 and edgeR. These obscure results need further research The authors want to acknowledge that the research
work using both tools in the identification of novel genes. funding for this project was provided by AIMST University,
DESeq2 and edgeR tools identified the highest number Malaysia, under its internal research grant scheme. The
of up-regulated genes compared to other tools in the transition authors also want to thank the Agriculture Park of Universiti
of young to mature and mature to the ripening stage. EdgeR Putra Malaysia (UPM), Selangor, Malaysia, for providing the
has successfully identified the highest number of DEGs in Durian (clone D24) fruit samples for the study.
up and down-regulated genes in the transition from mature to
ripening compared with other tools. For comparative functional Conflict of Interest
analysis and validation between four tools, the top 50 DEGs
The authors declare no conflict of interest.
in up and down-regulated were selected from each tool. The
function of the top significant genes between stages shows Author Contributions
Bouzayen M, Latché A, Nath P and Pech J- C (2010) Mechanism of Kim S and Yang I (2015) Analysis of whole transcriptome sequencing
Fruit Ripening - Chapter 16. In: Pua EC and Davey MR (eds) data: Workflow and software. Genomics Inform 119, 13:119-
Plant Developmental Biology- Biotechnological Perspectives. 125.
Springer, vol. 1, pp 319-339. Kvam VM, Liu P and Yaqing S (2012) A comparison of statistical
Brownleader MD, Jackson P, Mobasheri A, Pantelides AT, Sumar methods for detecting differentially expressed genes from
S, Trevan M and Dey PM (1999) Molecular aspects of cell RNA-seq data. Am J Bot 99:248-256.
wall modifications during fruit ripening. Crit Rev Food Sci Leustek T, Melinda NM, Bick J-A and Davies JP (2000) Pathways
Nutr 39:149-164. and regulation of sulfur metabolism revealed through molecular
Cannon RJ and Ho CT (2018) Volatile sulfur compounds in tropical and genetic studies. Annu Rev Plant Physiol Plant Mol Biol
fruits. J Food Drug Anal 26:445-468. 51:141-165.
Conesa A and Götz S (2008) Blast2GO: A comprehensive suite for Lonsdale D (2006) A review of the biochemistry, metabolism and
functional analysis in plants genomics. Int J Plant Genomics clinical benefits of thiamin(e) and its derivatives. Evid Based
2008:619832. Complement Alternat Med 3:49-59
Costa-Silva J, Domingues D and Lopes FM (2017) RNA-Seq Love M, Huber W and Anders S (2014) Moderated estimation of
differential expression analysis: An extended review and a fold change and dispersion for RNA-seq data with DESeq2
software tool. PLoS One 12:e0190152. Background. Genome Biol 15:550.
Fischer W (2003) A survey of best practices for RNA-seq data Luo Y, Ma B, Zeng Q, Xiang Z and He N (2016) Identification and
analysis. Genome Biol 27:705-715. characterisation of Lateral Organ Boundaries Domain genes in
Fischer NS and Steinhaus M (2019) Identification of an important mulberry, Morus notabilis-NC-ND license. Meta Gene 8:44-50.
odorant precursor in durian: First evidence of ethionine in Marowa P, Ding A and Kong Y (2016) Expansins: Roles in plant
plants. J Agric Food Chem 68:10397-10402. growth and potential applications in crop improvement. Plant
Gallego Romero I, Pai AA, Tung J and Gilad Y (2014) RNA-seq: Cell Rep 35:949–965.
Impact of RNA degradation on transcript quantification. McCarthy DJ, Chen Y and Smyth GK (2012) Differential expression
BMC Biol 12:42. analysis of multifactor RNA-Seq experiments with respect
Gao J, Zhang Y, Li Z and Liu M (2020) Role of ethylene response to biological variation. Nucleic Acids Res 40:4288–4297.
factors (ERFs) in fruit ripening. Food Qual Saf 4:15-20. Nyffeler R and Baum DA (2000) Phylogenetic relationships of
Garcia-Gago JA, Posé S, Munõz-Blanco J, Quesada MA and Mercado the durians [Bombacaceae-Durioneae or Malvaceae]
JA (2009) The polygalacturonase FaPG1 gene plays a key role Helicteroideae/Durioneae) based on chloroplast and nuclear
in strawberry fruit softening. Plant Signal Behav 4:766-768. ribosomal DNA sequences. Plant Syst Evol 224:55-82.
Gigolashvili T, Kopriva S and Linka N (2014) Transporters in plant Padovan A, Lanfear R, Keszei A, Foley WJ and Kulheim C (2013)
sulfur metabolism. Front. Plant Sci 5:442. Differences in gene expression within a striking phenotypic
Ha T, Do T, Martinoia E and Lee Y (2017) Functions of ABC mosaic Eucalyptus tree that varies in susceptibility to herbivory.
transporters in plant growth and development. Curr Opin BMC Plant Biol 13:29.
Plant Biol 41:32-38. Palapol Y, Kunyamee S, Thongkhum M, Ketsa S, Ferguson IB
Han Y, Ban Q, Li H, Hou Y, Jin M, Han S and Rao J (2016) and Van Doorn WG (2015) Expression of expansin genes in
DkXTH8, a novel xyloglucan endotransglucosylase/hydrolase the pulp and the dehiscence zone of ripening durian (Durio
in persimmon, alters cell wall structure and promotes leaf zibethinus) fruit. J Plant Physiol 182:33-39.
senescence and fruit post-harvest softening. Sci Rep 6:39155. Pinsorn P, Hoefgen R and Sirikantaramas S (2018) Dynamic
Hariyati T, Kusnadi J and Arumingtyas EL (2013) Genetic diversity expression of sulfate transporter genes in durian flesh
of hybrid Durian resulted from cross breeding between Durio during fruit ripening. The 6th International Conference on
kutejensis and Durio zibethinus based on random amplified Biochemistry and Molecular Biology, pp 1-8.
polymorphic DNAs (RAPDs). Am J Mol Biol 3:153-157. Posoongnoen S, Ubonbal R, Thammasirirak S, Daduang J, Minami
Hrmova M and Fincher GB (2001) Structure-function relationships of H, Yamamoto K and Daduang S (2015) alpha-Amylase from
β- D-glucan endo- and exohydrolases from higher plants. Plant Mon Thong durian Durio zibethinus Murr. cv. Mon Thong-
Mol Biol 47:73-91 nucleotide sequence analysis, cloning and expression. Plant
Hu L, Wu G, Hao C, Yu H and Tan L (2016) Transcriptome and Biotechnol J 32:1-10.
selected metabolite analyses reveal points of sugar metabolism Rajkumar AP, Qvist P, Lazarus R, Lescai F, Ju J, Nyegaard M, Mors
in jackfruit (Artocarpus heterophyllus Lam.). Plant Sci 248:45- O, Børglum AD, Li Q and Christensen JH (2015) Experimental
56. validation of methods for differential gene expression analysis
Husin NA, Rahman S, Karunakaran R and Bhore SJ (2018) A review and sample pooling in RNA-seq. BMC Genomics 16:548.
on the nutritional, medicinal, molecular and genome attributes Raplee ID, Evsikov AV and De Evsikova CM (2019) Aligning the
of Durian (Durio zibethinus L.), the king of fruits in Malaysia. aligners: Comparison of rna sequencing data alignment and
Bioinformation 14:265-270. gene expression quantification tools for clinical breast cancer
Imsabai W, Ketsa S and Van DWG (2002) Effect of temperature research. J Pers Med 9: 18.
on softening and the activities of polygalacturonase and Robinson MD, McCarthy DJ and Smyth GK (2009) edgeR:
pectinesterase in durian fruit. Postharvest Biol Technol 26:347- A Bioconductor package for differential expression analysis
351. of digital gene expression data. Bioinformatics 26:139-140.
Kalunke RM, Tundo S, Benedetti M, Cervone F, De Lorenzo G and Ruwaida IP, Supriyadi S and Parjanto P (2009) Variability analysis
D’Ovidio R (2015) An update on polygalacturonase-inhibiting of Sukun durian plant (Durio zibethinus) based on RAPD
protein (PGIP), aleucine-rich repeat protein that protects crop marker. Nusant Biosci 1:84-91.
plants against pathogens. Front Plant Sci 6:146. Santoso P, Saleh G, Saleh NM and Napis S (2013) Phylogenetic
Kate SG (2016) HISAT: A fast spliced aligner with low memory relationships amongst 10 Durio species based on PCR-RFLP
requirements. Physiol Behav 176:139-148 analysis of two chloroplast genes. Indones J Agric Sci 6:20-27.
16 Husin et al.
Sbodio JI, Snyder SH and Paul BD (2019) Regulators of the (ed) Postharvest Physiology and Biochemistry of Fruits and
transsulfuration pathway. Br J Pharmacol 176:583-593. Vegetables. Elsevier, pp 760-761.
Schurch NJ, Schofiled P, Gierlinski M, Cole C, Sherstnev A, Sing V, Yao BN, Tano K, Konan HK, Bédié GK, Oulé MK, Koffi-Nevry R
Wrobel N, Charbi K, Gordon G, Owen-Highes T et al. (2016) and Arul J (2014) The role of hydrolases in the loss of firmness
How many biological replicates are needed in an RNA-seq and of the changes in sugar content during the post-harvest
experiment and which differential expression tool should you maturation of Carica papaya L. var solo 8. J Food Sci Technol
use? RNA 22:839-851. 51:3309-3316.
Sebastian A, Chang JC, Mendez ME, Murugesh D, Hatsell S, Yusof NA (2018) Turbocharged odor that attracts animals: Structural
Economides A, Christiansen B and Loots G (2018) Loots analysis of MGL proteins in Durio zibethinus (Durian).
comparative transcriptomics identifies novel genes and International Conference of Fruits Pest and Diseases.
pathways involved in post-traumatic osteoarthritis development Zhu Q, Gao P, Liu S, Zhu Z, Amanullah S, Davis A and Luan F
and progression. Int J Mol Sci 19:2657. (2017) Comparative transcriptome analysis of two contrasting
Seyednasrollah AL and Elo LL (2013) Comparison of software watermelon genotypes during fruit development and ripening.
packages for detecting differential expression in RNA-seq BMC Genomics 18:3.
studies. Brief Bioinform 16:59-70.
Sharova EI (2007) Expansins: Proteins involved in cell wall softening Supplementary material
during plant growth and morphogenesis. Russ J Plant Physiol
54:713-727.
The following online material is available for this article:
Shearman JR, Sonthirod C, Naktang C, Sangsrakru D, Yoocha Figure S1 – Experimental design of the study.
T, Chatbanyong R, Vorakuldumrongchai S, Chusri O,
Tangphatsornruang S and Pootakham W (2020) Assembly
Figure S2 – The comparative Principal component analysis
of the durian chloroplast genome using long PacBio reads. (PCA) plot of the samples at young, mature, and ripening
Sci Rep 10:15980. stages.
Singh G, Verm AK and Kumar V (2016) Catalytic properties, functional Table S1 – Differentially up-regulated expressed genes between
attributes and industrial applications of β-glucosidases. 3 the young stage and mature stage of durian fruit pulp.
Biotech 6:3.
SutthacharoenthadW, SukkoN and Sinsuebchuea J (2016) Evaluation Table S2 – Differentially down-regulated expressed genes
of reference-based Mapping Tools for RNA-Seq Pipeline. between the young stage and mature stage of durian fruit pulp.
KMUTT. DOI:10.13140/RG.2.2.22400.69120/1 Table S3 – Differentially up-regulated expressed genes between
Teh BT, Lim K, Yong CH, Ng CCY, Rao SR, Rajasegaran V, Lim the mature stage and ripening stage of durian fruit pulp. We
WK, Ong CK, Chan K, Cheng VKY et al. (2017) The draft used FDR:
genome of tropical fruit durian (Durio zibethinus). Nat Genet
49:1633-1641. Table S4 – Differentially down-regulated expressed genes
Thirugnanasambantham K, Durairaj S, Saravanan S, Karikalan between the mature stage and ripening stage of durian fruit
K, Muralidaran S and Islam VIH (2015) Role of Ethylene pulp.
Response Transcription Factor (ERF) and its regulation in Table S5 – Categorisation of expressed genes to Gene Ontology
response to stress encountered by plants. Plant Mol Biol Rep terms in YS//MS, YS/RS, and MS/RS.
33:347–357.
Trapnell C, Pachte L and Salzberg SL (2009) TopHat: Discovering Table S6 – KEGG pathway term distribution of expressed
splice junctions with RNA-Seq. Bioinformatics 25:1105-1111 genes in YS/MS, YS/RS, and MS/RS.
Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR, Pimentel Table S7 – Heatmap results for softening genes.
H, Salzberg SL, Rinn JL and Pachter L (2012) Differential gene
Table S8 – Heatmap results for signal transduction.
and transcript expression analysis of RNA-seq experiments
with TopHat and Cufflinks. Nat Protoc 7:562–578. Table S9 – Genes and metabolising enzymes involved in
Vanijajiva O (2011) Genetic variability among Durian (Durio starch and sucrose metabolism pathway.
zibethinus Murr.) cultivars in the Nonthaburi province, Thailand
detected by RAPD analysis. J Agric Technol 7:1107-1116.
Vanijajiva O (2012) The application of ISSR markers in genetic
variance detection among Durian (Durio zibethinus Murr.) Associate Editor: Roberto Hirochi Herai
cultivars in the Nonthaburi province, Thailand. Procedia
Eng 32:155-159.
License information: This is an open-access article distributed under the terms of the
Yahia EM, Bello-Perez LA, Science F, Program TP and Sciences Creative Commons Attribution License (type CC-BY), which permits unrestricted use,
C (2019) Carbohydrates 1 Carbohydrates 1. In: Yahia EM distribution and reproduction in any medium, provided the original article is properly cited.