Systems Microbiology and Biomanufacturing (2023) 3:328–338

https://doi.org/10.1007/s43393-022-00113-8

ORIGINAL ARTICLE

Producing malonate in Saccharomyces cerevisiae via the β‑alanine

pathway
Shiyun Li1 · Wenxuan Fu2 · Ruifang Su2 · Yunying Zhao1,2 · Yu Deng1,2 

Received: 18 March 2022 / Revised: 9 May 2022 / Accepted: 13 May 2022 / Published online: 6 June 2022
© Jiangnan University 2022

Abstract
Malonate is a high-value chemical that can be used to produce value-added compounds. Due to the toxic by-products and low
product yield for malonate production through hydrolysis of cyanoacetic acid, microbial production methods have attracted
significant attention. Previously, the β-alanine pathway has been engineered in Escherichia coli for malonate production.
In this study, the β-alanine pathway was constructed in Saccharomyces cerevisiae by introducing the heterologous genes of
BcBAPAT and TcPAND to convert l-aspartate to malonic semialdehyde, combining with co-expression genes of AAT2 and
UGA2 to improve precursor supply and malonate producing. Through delta sequence-based integration of the two heterolo-
gous genes, the engineered strain produced with 7.21 mg/L malonate was screened. Further, replaced the succinic semialde-
hyde dehydrogenase gene UGA2 with yneI from E. coli which was utilized to produce malonate in previous study, increased
the malonate titer to 7.96 mg/L in flask culture. Following optimization, fermentation of the final engineered strain in shake
flasks yielded a maximum malonate titer of 12.83 mg/L, and this was increased to 91.53 mg/L during fed-batch fermentation
in a 5 L bioreactor which increased by two-fold compared with that of the engineered strain overexpressing UGA2.

Keywords Malonate · Saccharomyces cerevisiae · β-alanine · Metabolic engineering · Delta-sequence integration

Introduction and fragrance [3], specialty solvent, polymer cross-linking

[4] and pharmaceutical industries. Additionally, malonate
Malonate is one of the top 30 chemicals produced from bio- can be utilised to produce valuable chemical compounds as
mass by the U.S. Department of Energy (DOE) [1]. As a building blocks and precursors for polymers such as polyes-
highly valued chemical, malonate has been applied in a num- ter [5]. Currently, malonate is mainly produced by hydrolys-
ber of manufacturing processes in the electronics [2], flavour ing cyanoacetic acid [6], but this generates toxic by-products
and suffers from low product yield. Therefore, renewable
and environmentally friendly bio-based methods for produc-
* Yunying Zhao ing malonate are receiving greater attention.
[email protected]
In recent years, microbial metabolic engineering has
* Yu Deng expanded because it is efficient and effective for trans-
[email protected]
forming substrates into value-added chemicals. However,
Shiyun Li research on the biological production of malonate has not
[email protected]
progressed much due to a lack of knowledge on suitable
Wenxuan Fu enzymes and metabolic pathways. In the previous stud-
[email protected]
ies, three non-natural metabolic pathways have been con-
Ruifang Su structed for producing malonate by using malonyl-CoA [7],
[email protected]
oxaloacetate [8] and β-alanine [9] as the main precursor,
1
National Engineering Research Center of Cereal respectively. The malonyl-CoA pathway was constructed by
Fermentation and Food Biomanufacturing, Jiangnan modifying an acyl-CoA hydrolase to convert malonyl-CoA
University, 1800 Lihu Road, Wuxi  214122, Jiangsu, China to malonate, and achieved 82.3 mg/L malonate producing
2
Jiangsu Provincial Research Center for Bioactive Product in the engineered E. coli [7]. The oxaloacetate pathway was
Processing Technology, Jiangnan University, 1800 Lihu constructed by co-expressing α-keto decarboxylase and
Road, Wuxi  214122, Jiangsu, China

13
Vol:.(1234567890)
Systems Microbiology and Biomanufacturing (2023) 3:328–338 329

malonic semialdehyde dehydrogenase to convert oxaloac- aspartate-1-decarboxylase (PAND) from Tribolium cas-
etate to malonate, and finally produced 42.5 mg/L malonate taneum and β-alanine-pyruvate aminotransferase (BAPAT)
in cellulolytic thermophilic fungus Myceliophthora ther- from Bacillus cereus were used for this conversion, proving
mophila [8]. The β-alanine route was constructed to produce that the two exogenous enzymes could efficiently convert
malonate in Escherichia coli by co-expressing genes encod- aspartate to malonic semialdehyde in yeast. Thus, in the pre-
ing β-alanine pyruvate transaminase (pa0123) from Pseu- sent study, we constructed the β-alanine route by introducing
domonas aeruginosa, native succinic semialdehyde dehy- PAND from B. cereus and BAPAT from T. castaneum to
drogenase (yneI), native aspartate-α-decarboxylase (panD) produce malonic semialdehyde in yeast, and we also over-
and native aspartase (aspA) [9]. Additional deletion of the expressed native succinic semialdehyde dehydrogenase from
gene encoding malonic semialdehyde reductase increased S. cerevisiae and yneI from E. coli to transform malonic
malonate production to 0.450 g/L in flask culture, and fed- semialdehyde to malonate (Fig. 1). After calculation, the
batch culture achieved production of 3.60 g/L malonate from theoretical maximum yield of the β-alanine route was 1.15 g
glucose. malonate/g glucose (Supplementary Fig. 1). This is the first
However, there are few reports on malonate production by time using the β-alanine route to produce malonate in S.
Saccharomyces cerevisiae. In the β-alanine route, malonic cerevisiae. In addition, we enhanced production of malonate
semialdehyde is the main precursor for malonate. In this through further gene overexpression and optimisation of the
pathway, the malonic semialdehyde-producing pathway is fermentation conditions.
the same as the β-alanine pathway producing 3-hydroxy-
propionic acid (3-HP). Researchers have successfully con-
structed the β-alanine pathway for 3-HP and achieved a titer Materials and methods
of 13.77 ± 0.3 g/L in controlled fed-batch fermentation [10],
proving that S. cerevisiae is an excellent host for producing Strains and culture conditions
malonic semialdehyde. For the construction of the β-alanine
route in S. cerevisiae, the key step is conversion of aspar- All strains and plasmids used in this study are shown
tate to malonic semialdehyde. In this previous work [10], in Table  1, and primers used in this work are listed in

Fig. 1  Biosynthetic pathway and integration strategy for TcPAND and tate-1-decarboxylase gene; BcBAPAT, β-alanine-pyruvate aminotrans-
BcBAPAT to produce malonate in S. cerevisiae. PCR-amplified prod- ferase gene; delta1 and delta2, multi-copy DNA sequences located
ucts including TcPAND and BcBAPAT genes controlled by strong pro- in the reverse transcription transposon Ty of the chromosomal DNA
moters were integrated into the delta locus with overlapping URA3 in S. cerevisiae. ­TADH1, ADH1 terminator; P ­ TEF1, TEF1 promoter;
homology. Red line, overexpression pathways; green line, heterolo- URA3, URA3 selection marker gene; ­PGPD1, GPD1 promoter; T ­ CYC1,
gous pathways; AAT2, cytoplasmic aspartate aminotransferase gene; CYC1 terminator
UGA2, succinic semialdehyde dehydrogenase gene; TcPAND, aspar-

13

330 Systems Microbiology and Biomanufacturing (2023) 3:328–338

Table 1  Strains and plasmids Name Relevant genotype name Sources

used in this study
Strains
 E. coli JM109 ATCC 53,323
 S. cerevisiae BY4741 MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0 ATCC 201,388
 BA-1 BY4741 overexpressing AAT2 with ­PTEF1 This study
 BA-2 BA-1 overexpressing UGA2 with ­PGPD1 This study
 BA-3 BA-2 carrying pand and bapat This study
 BA-4 BA-2 carrying pand and bapat This study
 BA-5 BA-3 deleted UGA2, overexpressing yne1 with ­PGPD This study
Plasmids
 pY26 -GPD-TEF Integrating plasmid, URA3 marker, MCS derived from [28]
pBLUESCRIPT (ColE1 (derivative) ori, f1ori, ­AmpR)
pY26 -pand-bapat pY26 -GPD-TEF carrying pand and bapat This study

Supplementary Table 1. E. coli cells were cultured in LB was replaced by ­PTEF1 by integrating the HIS3-PTEF1 cas-
medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl) sette containing the selective marker HIS3, ­PTEF1, and the
at 37 ℃ with shaking at 250 rpm. Ampicillin (100 μg/mL) upstream and downstream 500 bp regions of the AAT2 gene
was added to provide selective pressure during growth when as the homologous arms. The HIS3 fragment was ampli-
necessary. Recombinant yeast cells were grown in YPD fied from the pRS313 plasmid with primers His-F/His-R,
medium (10 g/L yeast extract, 20 g/L peptone, 20 g/L glu- and ­PTEF1 was amplified from BY4741 genomic DNA with
cose) at 30 ℃ and 250 rpm. Synthetic dropout (SD) medium primers TEF1-F/TEF1-R. The upstream and downstream
(1.7 g/L yeast nitrogen base, 5 g/L ammonium sulphate, 500  bp regions of the AAT2 gene were amplified from
20 g/L glucose) with appropriate supplemental amino acids BY4741 genomic DNA with primers AAT2U-F/AAT2U-
was used to select yeast transformants. Solid medium was R and AAT2D-F/AAT2D-R, respectively. The HIS3-PTEF1
produced by adding 2% (w/v) agar. cassette was amplified with primers AAT2U-F/ AAT2D-R
For shake flask cultivation, 50 mL of YPD medium in using HIS3, ­PTEF1, and the upstream and downstream 500 bp
a 250 mL shake flask was used as the initial fermentation regions of the AAT2 gene fragments as templates. The HIS3-
medium. Cultures were first grown overnight at 30 ℃ and PTEF1 cassette was transformed into the BY4741 strain to
250 rpm in YPD medium, then inoculated to an optical den- generate the BA-1 strain, and correct integration was con-
sity at 600 nm ­(OD600) of 0.05. Cultures were incubated at firmed by PCR with primers AAT2-check-F/AAT2-check-R.
30 °C and 250 rpm for ~ 7 days. To overexpress the UGA2 gene encoding native succi-
nate semialdehyde dehydrogenase, the promoter of UGA2
Plasmid construction was replaced by ­PGPD1 by integrating the LEU2-PGPD cas-
sette containing the selective marker LEU2, ­PGPD1, and the
The pand gene encoding the PAND enzyme in T. castaneum upstream and downstream 500 bp regions of the UGA2 gene
and the bapat gene encoding BAPAT in B. cereus were as the homologous arms. The LEU2 fragment was ampli-
codon-optimised and synthesised by Genewiz (Suzhou, fied from the pHAC181 plasmid [12] with primers Leu-F/
China). The accession numbers of codon-optimized gene Leu-R, and P ­ GPD was amplified from BY4741 genomic
sequences were OL456161 (BcBAPAT) and OL456162 DNA with primers GPD-F/GPD-R. The upstream and
(TcPAND). To construct the pY26-pand plasmid, the pand downstream 500 bp regions of the UGA2 gene were ampli-
gene was first digested with BglII/NotI and ligated to the fied from BY4741 genomic DNA with primers UGA2U-
pY26-GPD-TEF plasmid [11], which was previously F/UGA2U-R and UGA2D-F/UGA2D-R, respectively.
digested with BglII/NotI, to generate the pY26-pand plas- Next, the LEU2-PGPD cassette was amplified with primers
mid. The bapat gene was digested with SalI/BamHI and UGA2U-F/UGA2D-R using LEU2, ­PGPD, and the upstream
ligated to pY26-pand previously digested with SalI/BamHI and downstream 500 bp regions of the UGA2 gene fragments
to generate pY26-pand-bapat. as templates. The LEU2-PGPD cassette was transformed into
the BA-1 strain to generate strain BA-2, and correct integra-
Genetic manipulations tion was confirmed by PCR with primers UGA2-check-F/
UGA2-check-R.
To overexpress the AAT2 gene encoding the native cyto- To integrate exogenous genes into the genome, delta1
plasmic aspartate aminotransferase, the AAT2 promoter and delta2 [13, 14] fragments were amplified from the S.

13
Systems Microbiology and Biomanufacturing (2023) 3:328–338 331

cerevisiae genome using delta1-F/delta1-R and delta2-F/ the average of the triplicate samples. Transcript levels were
delta2-R primers, respectively, and pand and bapat expres- calculated using the − ΔΔCt method [18].
sion cassettes DPU and DBU were amplified from the pY26-
pand-bapat plasmid by ura2-F/delta2-R and delta1-F/ura1-R Metabolite quantification by high‑performance
primers, respectively. In addition, two parts of the selec- liquid chromatography (HPLC) and LC mass
tion marker URA3 (URA3-1 and URA3-2) were amplified spectrometry (LC–MS)
from pY26-GPD-TEF using ura1-F/ura1-R and ura2-F/
ura2-R. The above PCR products were gel-extracted, and The fermentation mixture was centrifuged at 6000×g and the
touchdown PCR was performed to assemble two homolo- supernatant was transferred assessed by HPLC or LC–MS.
gous arms containing delta1, DBU, URA3-1 and URA3-2, For HPLC detection, the supernatant was filtered through
DPU, and delta2. Finally, the two homologous arms were a 0.22 μm membrane and analysed using a 1260 Infinity II
transformed into the BA-2 strain using the Li-Ac method instrument (Agilent Technologies, USA) equipped with an
[15], and malonic acid-producing strain BA-3 was acquired HPX-87H ion-exclusion column (300 mm × 7.8 mm; Bio-
and screened.To replace UAG2 with yneI, the open reading Rad). The mobile phase consisted of 5 mM H ­ 2SO4 at a rate
frame (ORF) of UGA2 was replaced by yneI by integrating of 0.6 mL/min, and the column temperature was maintained
kanMX-yneI cassette which contained the selective marker at 30 ℃. Compounds were detected from 30 µL injections
kanMX, ­PGPD and yneI. The kanMX-yneI cassette was con- using a refractive index detector (Agilent Technologies,
structed by a sequential PCR method [16]. First, the kanMX USA).
was amplified from pUG6 plasmid with primers kanMX-
F/R. ­PGPD was amplified from BY4741 genomic DNA with Fed‑batch Fermentation in a 5 L Bioreactor
the primers GPD-F/GPD-R. The yneI gene encoding suc-
cinate semialdehyde dehydrogenase enzyme in E. coli was For 5 L scale fermentation, the parameters and conditions
codon-optimized and synthesized by Genewiz (Suzhou, were set as follows: temperature 30  °C, rotation speed
China) and amplified with primers of yneI-F/R. The acces- 600 rpm, ventilation volume 1 vvm, medium YPD, initial
sion number of codon-optimized yneI gene sequence was glucose 20 g/L, initial pH 6.5, initial dissolved oxygen (DO)
OL473734. Next, the full-length kanMX-yneI cassette was 100%, loaded liquid 2 L. During the fermentation process,
abstained using sequential PCR method with the primers the pH was maintained at pH 5 using 5 M NaOH. In fed-
kanMX-F/yneI-R. The kanMX-yneI cassette containing the batch fermentation, glucose (80% in water) was added at a
homologous arms of the UGA2 ORF was abstained after rate of 48 g/day after glucose had run out, and regular sam-
application with yneI-IN-F/R primers and was transformed pling was carried out at the indicated timepoints.
into BA-3 strain replacing UGA2 gene. The correct integra-
tion was confirmed by PCR with the primers yneI-check-F
/R, and generated the strain BA-5. Results

Construction of the β‑alanine‑producing malonate

Relative quantification of gene expression levels pathway in S. cerevisiae

To quantify gene expression levels, yeast cells were first The E. coli β-alanine pathway has been reported previously
cultivated in YPD medium at 30 ℃ for ~ 16 h, and cells [9]. In this pathway, malonic semialdehyde is the direct
were transferred to 100 mL of fresh YPD and cultured to precursor of malonate. In this previous study, succinic
­OD600 = 0.6 ~ 0.8. Cells were collected and total RNA was semialdehyde dehydrogenase was selected as the enzyme
extracted using the hot phenol method [17]. Following transferring malonic semialdehyde to malonate because the
genomic DNA eraser treatment to remove DNA, cDNA was molecular structures of succinic semialdehyde and malonic
synthesised from 500 ng of total RNA using random primers semialdehyde are similar [9]. Thus, we selected the UGA2
with a Primer Script RT reagent kit (Cwbiotech, Beijing, gene encoding succinic semialdehyde dehydrogenase from
China). The synthesised cDNA was then amplified by PCR S. cerevisiae, which also converts succinic semialdehyde to
with primers pand-a/pand-s, bapat-a/bapat-s and act-a/act-s succinic acid [19]. To overexpress the UGA2 gene, its pro-
to verify pand and bapat mRNA levels, with the housekeep- moter was replaced by P ­ GPD1. Also, since aspartate is the
ing gene (ACT1) as an internal reference gene. Quantitative precursor for malonate production, we overexpressed genes
PCR (qPCR) was performed with SYBR Premix Ex Taq encoding native cytoplasmic aspartate aminotransferase
(Cwbiotech, Beijing, China) and a Bio-Rad CFX96 instru- AAT2 [20] together with ­PTEF1. Additionally, for synthe-
ment (Bio-Rad, USA). Each reaction was carried out in sis of malonic semialdehyde, two heterologous enzymes,
triplicate, and the reported cycle threshold (Ct) value was aspartate-1-decarboxylase (PAND) and β-alanine-pyruvate

13

332 Systems Microbiology and Biomanufacturing (2023) 3:328–338

aminotransferase (BAPAT), were added to the pathway to expression of the two genes, reverse-transcription PCR
transfer aspartate to malonic semialdehyde. According to a (RT-PCR) was carried out, and the results indicated that
previous study on the production of 3-HP in S. cerevisiae, the two exogenous genes were both successfully expressed
we selected BAPAT from B. cereus (encoded by BcBAPAT) in the BA-3 strain (Supplementary Fig. 3). These results
[21] and PAND from T. castaneum (encoded by TcPAND) confirmed that the β-alanine pathway built using exogenous
[22] as the two heterologous enzymes [10]. genes BcBAPAT and TcPAND produced malonate success-
To introduce the two exogenous genes, circumvent plas- fully in S. cerevisiae.
mid instability, and increase the number of target gene cop-
ies, a delta sequence-based constitutive expression method The production of malonate was tightly regulated
was employed to overexpress the two heterologous genes. by the expression levels of BcBAPAT and TcPAND
S. cerevisiae delta sequences are long, terminal repeats
of transposons Ty1 and Ty2, and 425 delta sequences are To determine whether the high malonate yield was due to the
dispersed through the yeast genome [23]. BcBAPAT and high expression levels of BcBAPAT and TcPAND, quantita-
TcPAND were under the control of GPD and TEF1 pro- tive real-time PCR (qRT-PCR) was performed to confirm
moters [24], respectively, were integrated into the chromo- whether the delta sequence-based integrative expression
some of BA-2 through delta sequence-based homologous obviously influenced gene expression levels of BcBAPAT
recombination (Fig. 1). Following successful integration, and TcPAND. We compared the transcription levels of
we randomly selected 27 single strains on a selective plate. BcBAPAT and TcPAND in BA-3 and the strain with low-
After shake flask fermentation with 50 mL YPD for 7 days, est titer (BA-4), the lowest titer was 4.89 mg/L, which was
malonate was successfully detected in the culture super- 32.08% less than that of BA-3 (Fig. 2). The transcription
natant of all selected strains by LC–MS (Supplementary levels of BcBAPAT and TcPAND in BA-3 were 3.46- and
Fig. 2). The strain with the highest titer was selected and 7.92-fold higher than in BA-4 (Fig. 3). Thus, the higher
named BA-3. This strain produced 7.21 mg/L malonate expression levels of BcBAPAT and TcPAND in BA-3 strain
in shake flask fermentation (Fig. 2). To verify exogenous helped to increase malonate production. Furthermore, the

Fig. 2  The malonate titer of 27 randomly selected strains. The BA-3 strain had the highest titer (7.21  mg/L) and BA-4 had the lowest titer
(4.89 mg/L)

13
Systems Microbiology and Biomanufacturing (2023) 3:328–338 333

efficiency for converting malonic semialdehyde to malonate

was unknown. For further improve malonate production, we
replaced UGA2 with another gene encoding succinic semi-
aldehyde dehydrogenase, yneI, which has been utilized to
produce malonate before [9], and generated the strain BA-5.
During fermentation, the growth has no obvious difference
between BA-3 and BA-5 (Fig. 4a). As a result, 7.96 mg/L
malonate was produced when yneI was employed (Fig. 4b).
This malonate titer increased by 11% than that of when
employing UGA2. This result indicated that YneI has higher
efficiency for converting malonic semialdehyde to malonate
compared with Uga2.
Fig. 3  Relative gene expression levels of BcBAPAT and TcPAND 
Optimising fermentation increases malonate
production
delta sequences had different copy numbers, which led to
large differences in gene expression levels among different Substrates have a major influence on batch fermentation
integrative strains. Thus, when using delta sequence-based [25]. The amount of glucose and the time of addition may
integrative methods, screening multiple integrative strains affect malonate accumulation. To optimise the glucose addi-
was an efficient approach to identify strains with high gene tion time, 5 g/L of glucose was fed once at 12, 24 or 48 h,
expression levels. respectively, when shake flask fermentation of BA-5. The
results showed that glucose fed once at 12 h was best for
Replacing succinic semialdehyde dehydrogenase increasing malonate production (Fig. 5a). Next, the glucose
to improve malonate production addition concentration was tested; 5, 10 and 20 g/L glucose
was fed at 12 h. The highest malonate titer was achieved
After successfully constructed the β-alanine producing with 5 g/L glucose, and higher glucose concentrations had a
malonate pathway, we analysed the reason for low titer of negative effect on malonate production (Fig. 5b). Excessive
malonate production which may be the low efficiency of con- glucose may have decreased the malonate titer by causing
verting malonic semialdehyde to malonate when employing flux to be diverted into the glycolytic pathway to increase
the succinic semialdehyde dehydrogenase UGA2. As UGA2 biomass. Therefore, the optimum fermentation conditions
has never been utilized for producing malonate before, its were YPD medium supplemented with 5 g/L of glucose fed

Fig. 4  The shake flask fermentation result of BA-3 and BA-5. a Growth of BA-3 and BA-5. b Malonate titer of BA-3 and BA-5

13

334 Systems Microbiology and Biomanufacturing (2023) 3:328–338

Fig. 5  Fermentation results
following addition of glucose.
a Malonate produced in YPD
medium with 5 g/L glucose
fed once at 12, 24 or 48 h. b
Malonate produced in YPD
medium with 5, 10 or 20 g/L
glucose fed at 12 h

at 12 h. Under these culture conditions, the highest malonate water). Biomass accumulated until 120 h, followed by a sta-
titer was 12.83 mg/L in shake flasks, an increase of 66% tionary phase, and the biomass reached ­OD600 > 60. Follow-
compared with no supplementation of glucose (Fig. 5b). ing glucose feeding, the malonate titer was increased, and
These results suggested that glucose as the main carbon accumulation continued to a maximum level of 91.53 mg/L
source had a significant effect on malonate production; addi- until feeding ended (Fig. 6a), and the yield was 0.28 mg/g
tion of the optimal amount of glucose at the optimal time glucose. Ethanol was produced in large quantities during
helped to increase the malonate titer. glucose feeding, and accumulated to ~ 30 g/L at 168 h, then
sharply declined when feeding ceased (Fig. 6b). The acetic
Fed‑batch fermentation acid and glycerol titer increased slowly throughout the whole
fed-batch fermentation process, and reached 6.21 g/L and
To further enhance malonate production by the BA-5 strain, 3.43 g/L, respectively (Fig. 6b).
fed-batch fermentation in a 5 L bioreactor was carried out. In In addition, we also test the malonate producing efficiency
the fed-batch fermentation process, glucose (80% in water) of BA-3 strain. We carried out the fed-batch fermentation of
was added at a rate of 48 g/day, starting at 12 h (when glu- BA-3 with the same fermentation strategy used above. The
cose had run out) and ending at 168 h, and regular sampling maximum level of malonate titer was 44.31 mg/L which
was performed out at the indicated times. The total amount was 52% less than that of BA-5 strain (Fig. 6c). The growth
of glucose fed was 318  g (397.5  mL of 80% glucose in rate of the two strains were similar, but the biomass of BA-3

13
Systems Microbiology and Biomanufacturing (2023) 3:328–338 335

Fig. 6  Fed-batch fermentation in a 5-L bioreactor using the BA-5 Growth, glucose consumption and malonate production of the BA-3
strain. a Growth, glucose consumption and malonate production of strain. d By-product accumulation in the BA-3 strain
the BA-5 strain. b By-product accumulation in the BA-5 strain. c

reached ­OD600 > 70, higher than that of BA-5 (Fig. 6c). And produced via decarboxylation of l-aspartate by PAND.
the by-products: ethanol, acetic acid and glycerol accumula- And in previous study [10], BAPAT and PAND were both
tion was much less than that of BA-3 (Fig. 6d). In general, used for efficient production of malonic semialdehyde and
BA-5 has higher efficiency for producing malonate, and achieved high-level production of the target product 3-HP in
BA-3 strain could produce less by-products and accumulate S. cerevisiae, proving that S. cerevisiae is an excellent host
more biomass. Thus, it was obvious that yneI has higher for producing malonic semialdehyde.
efficiency for converting malonic semialdehyde that UGA2. In many studies on yeast cell factories [27, 28], episo-
mal plasmids have been used to introduce pathway genes
for building engineered strains. However, this may lead to
Discussion problems such as plasmid instability [29]. To avoid this
problem and to increase the copy number of pathway genes,
In this work, we aimed to construct a malonate-producing a genomic integration method based on delta-sequences was
cell factory for the biosynthesis of this valuable chemi- applied for constitutive expression of exogenous pathway
cal using S. cerevisiae as a host. In the previous work, the genes in the present study. Besides, there are no inducer
β-alanine pathway was employed for producing malonate and antibiotics addition for the constitutive expression of the
using E. coli as host [9]. Compared with E. coli, S. cerevi- malonate synthetic pathway, which will reduce the cost of
siae has a stronger acid tolerance [26], which is important production. Accordingly, to construct the pathway for pro-
because malonate is a dicarboxylic acid. Additionally, in ducing the malonate precursor malonic semialdehyde from
the β-alanine pathway, malonic semialdehyde is the direct aspartate, the highly efficient BcBAPAT from B. cereus and
precursor of malonate. Malonic semialdehyde is converted TcPAND from T. castaneum [10] were introduced by delta-
from β-alanine by the action of BAPAT, and β-alanine is sequence-based integration. A delta sequence is a long-end

13

336 Systems Microbiology and Biomanufacturing (2023) 3:328–338

repeated DNA sequence located in the reverse transcrip- production, indicating the succinic semialdehyde dehydro-
tion transposon Ty of chromosomal DNA in S. cerevisiae. genase activity was a major bottleneck for malonate pro-
Because there are ~ 425 delta sequences, delta sequence- ducing and the activity of YneI was still too low for high-
based homologous recombination is more efficient than tra- level production of malonate. In the further studies, it may
ditional methods for exogenous DNA integration into the be necessary to improve succinic semialdehyde dehydro-
S. cerevisiae genome [23]. However, due to the different genase activity to increase the malonate titer. Additionally,
copy numbers of delta sequences, positive colonies with as a succinic semialdehyde dehydrogenase, UGA2 may
high integration copy numbers cannot be easily screened as have greater affinity for succinate than malonate, which
the integration process remains random [23]. Therefore, we may result in more carbon flux flowing toward succinate
selected 27 positive colonies to screen for high copy number when overexpressing UGA2. According to previous stud-
integration strains. Due to the variation in copy number, the ies, UGA2 is part of the gamma-aminobutyrate (GABA)
malonate titer of BA-4 was 32.08% lower than that of BA-3. shunt [30], and UGA2 is regulated by GABA, which may
The transcription levels of BcBAPAT and TcPAND in strains lead to low UGA2 activity when producing malonate [31].
BA-3 and BA-4 were determined by qRT-PCR as an indirect UGA2 has also been reported to be activated by high levels
measure of the gene copy number. The transcription levels of sugar stress [32]; hence the low glucose concentration
of BcBAPAT and TcPAND in BA-3 were 3.46- and 7.922- during fermentation may lead to low UGA2 activity.
fold higher than in BA-4 (Fig. 3). The results confirmed Besides, UGA2 and yneI are NAD(P)+-dependent suc-
that the difference in copy number of the two strains and cinic semialdehyde dehydrogenase [33]. With the over-
the increased expression levels of BcBAPAT and TcPAND expression of succinic semialdehyde dehydrogenase, it
helped to improve malonate production. required large amount of NAD(P) + (Fig.  1) and might
Because l-aspartate is the direct precursor of β-alanine, increase the ratio of NAD(P)H/ NAD(P) +, and further
we increased the flux toward aspartate by overexpressing led to redox imbalance. On the other hand, pyruvate is an
genes encoding native cytoplasmic aspartate aminotrans- important precursor of malonate producing from glucose,
ferase AAT2. However, overexpression of AAT2 did not and the synthesis of large amount of pyruvate might lead
significantly increase the malonate titer. This may be due to glycolytic NADH overproducing (Fig. 1), and reduced
to insufficient supply of oxaloacetate, the precursor of aspar- products like ethanol and glycerol will be generated [34],
tate. For this reason, we aimed to increase oxaloacetate which will lead to the reduced carbon flux to malonate. It
supply by overexpressing pyruvate carboxylases PYC1 and also explained the high production of ethanol and glycerol
PYC2 in later study, based on previous studies [29]. Under in the fed-batch fermentation (Fig. 6b). Thus, it is nec-
batch conditions, such as during growth on glucose fol- essary to maintain redox balance to reduce the reduced
lowed by ethanol, there is high pyruvate carboxylase activ- products producing and lead more carbon flux to malonate.
ity in cells [29], which could significantly improve malonate Malonate is a high value chemical and needs to be pro-
production. duced at a price competitive to the malonate by chemi-
The last step for malonate production in this pathway is cal synthesis. Considering the production parameters
oxidation of malonic semialdehyde, catalysed by malonic (91.53 mg/L titer and 0.28 mg/g glucose yield), both of
semialdehyde dehydrogenase. One previous study showed the titer and the yield need to be improved at least 100-fold
that succinic semialdehyde dehydrogenase (yneI) was to reach commercial feasibility for production of malonate
more efficient than malonic semialdehyde dehydrogenase from glucose. With the maximal theoretical yield of 1.15 g
[9]. Thus, in the present work, we first overexpressed the malonate/g glucose via β-alanine pathway (Supplementary
native succinic semialdehyde dehydrogenase (UGA2), Fig. 1), it should, however, be possible to achieve the com-
which converts succinic semialdehyde to succinic acid. mercial process requirements with further optimization.
However, overexpression of UAG2 did not significantly In this study, the β-alanine pathway was constructed in
increase the malonate titer. This may reflect low activity S. cerevisiae to produce malonate. After screening strains,
of UGA2 with malonic semialdehyde as substrate. After replacing succinic semialdehyde dehydrogenase and opti-
replacing UGA2 with yneI, the malonate titer increased by mising the glucose addition concentration and time, the
11%, which was not a significant improvement. However, maximum malonate titer reached 12.83  mg/L during
when carried out the amplification fermentation of 5-L fermentation in shake flasks. Following further fermen-
bioreactor fed-batch fermentation, the BA-5 strain over- tation optimisation, the malonate titer was increased to
expressing yneI produced 91.53 mg/L malonate, increased 91.53 mg/L during fed-batch fermentation in a 5 L biore-
by two-fold compared with that of the BA-3 strain over- actor (Fig. 5a).
expressing UGA2. The significant increase of malonate
titer indicated that the higher succinic semialdehyde dehy- Supplementary Information  The online version contains supplemen-
tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 43393-0​ 22-0​ 0113-8.
drogenase activity had a tight correlation with malonate

13
Systems Microbiology and Biomanufacturing (2023) 3:328–338 337

Acknowledgements  This work was supported by the National Key 11. Chen F, Zhou J, Liu J, Du G, Chen J. Effect of acetyl-CoA syn-
R&D Program of China (2019YFA0905502), the National Natural Sci- thase gene overexpression on physiological function of Saccha-
ence Foundation of China (21877053), Tianjin Synthetic Biotechnol- romyces cerevisiae. Acta Microbiol Sin. 2010;50:1172–9.
ogy Innovation Capacity Improvement Project (TSBICIP-KJGG-015), 12. Zhao Y, Xiong B, Xu H, Jiang L. Expression of NYV1 encoding
and the Open Foundation of Jiangsu Key Laboratory of Industrial Bio- the negative regulator of Pmc1 is repressed by two transcriptional
technology (KLIB-KF201807). repressors, Nrg1 and Mig1. FEBS Lett. 2014;588:3195–201.
https://​doi.​org/​10.​1016/j.​febsl​et.​2014.​06.​062.
Author contributions  YD and YZ: designed and conceived the idea the 13. Parekh RN, Shaw MR, Wittrup KD. An integrating vector for
whole experiments. SL, WF and RS: performed the experiments. SL: tunable, high copy, stable integration into the dispersed ty δ sites
analysed the results and wrote the first draft of the manuscript. YD and of Saccharomyces cerevisiae. Biotechnol Prog. 1996;12:16–21.
YZ: supervised the work and revised the manuscript. https://​doi.​org/​10.​1021/​bp950​0627.
14. Curran KA, Leavitt JM, Karim AS, Alper HS. Metabolic engi-
neering of muconic acid production in Saccharomyces cerevisiae.
Declarations  Metab Eng. 2013;15:55–66. https://​doi.​org/​10.​1016/j.​ymben.​
2012.​10.​003.
Conflict of interest  The authors declare that they have no conflict of 15. Gietz RD, Schiestl RH. Frozen competent yeast cells that can
interests. be transformed with high efficiency using the LiAc/SS carrier
DNA/PEG method. Nat Protoc. 2007;2:1–4. https://​doi.​org/​10.​
Ethical approval  Not applicable. 1038/​nprot.​2007.​17.
16. Zhang P, Ding Y, Liao W, Chen Q, Zhang H, Qi P, et al. A simple,
Informed consent  Not applicable. universal, efficient PCR-based gene synthesis method: sequential
OE-PCR gene synthesis. Gene. 2013;524:347–54. https://​doi.​org/​
10.​1016/j.​gene.​2013.​03.​126.
17. Köhrer K, Domdey H. Preparation of high molecular weight RNA.
Methods Enzymol. 1991;194:398–405.
18. Livak KJ, Schmittgen TD. Analysis of relative gene expression
References data using real-time quantitative PCR and the 2−ΔΔCT method.
Methods. 2001;25:402–8. https://​doi.​org/​10.​1006/​meth.​2001.​
1. Werpy T, Petersen G. Top value added chemicals from biomass. 1262.
Results of screening for potential candidates from sugars and syn- 19. Ramos F, Ei Guezzar M, Grenson M, Wiame J-M. Mutations
thesis gas, vol. 1. Golden: National Renewable Energy Lab; 2004. affecting the enzymes involved in the utilization of 4-aminobu-
2. Li B, Lin A, Gan F. Preparation and characterization of Cr–P tyric acid as nitrogen source by the yeast Saccharomyces cerevi-
coatings by electrodeposition from trivalent chromium elec- siae. Eur J Biochem. 1985;149:401–4. https://​doi.​org/​10.​1111/j.​
trolytes using malonic acid as complex. Surf Coat Technol. 1432-​1033.​1985.​tb089​39.x.
2006;201:2578–86. https://​doi.​org/​10.​1016/j.​surfc​oat.​2006.​05.​ 20. Verleur N, Elgersma Y, Van Roermund CWT, Tabak HF, Wanders
001. RJA. Cytosolic aspartate aminotransferase encoded by the AAT2
3. Aziah Serri N, Kamaruddin AH, Len KYT. A continuous esteri- gene is targeted to the peroxisomes in oleate-grown Saccharomy-
fication of malonic acid with citronellol using packed bed reac- ces Cerevisiae. Eur J Biochem. 1997;247:972–80. https://​doi.​org/​
tor: investigation of parameter and kinetics study. Food Bioprod 10.​1111/j.​1432-​1033.​1997.​00972.x.
Process. 2010;88:327–32. https://​doi.​org/​10.​1016/j.​fbp.​2008.​12.​ 21. Nakano Y, Tokunaga H, Kitaoka S. Two omega-amino
002. acid transaminases from Bacillus cereus. The J Biochem.
4. Szlosek D, Currie D. Application and mechanism of malonic 1977;81:1375–81. https://d​ oi.o​ rg/1​ 0.1​ 093/o​ xford​ journ​ als.j​ bchem.​
acid as a green alternative for protein-crosslinking. Green Sustain a1315​91.
Chem. 2016;6:6. https://​doi.​org/​10.​4236/​gsc.​2016.​62010. 22. Liu Z, Zheng W, Ye W, Wang C, Gao Y, Cui W, et al. Charac-
5. Zhang D, Dumont M-J. Advances in polymer precursors and bio- terization of cysteine sulfinic acid decarboxylase from Tribolium
based polymers synthesized from 5-hydroxymethylfurfural. J Poly castaneum and its application in the production of β-alanine.
Sci. 2017;55:1478–92. https://​doi.​org/​10.​1002/​pola.​28527. Appl Microbiol Biotechnol. 2019;103:9443–53. https://​doi.​org/​
6. Klikar M, Jelínková V, Růžičková Z, Mikysek T, Pytela O, Ludwig 10.​1007/​s00253-​019-​10139-z.
M, et al. Malonic acid derivatives on duty as electron-withdrawing 23. Lee FWF, Silva NAD. Improved efficiency and stability of mul-
units in push–pull molecules. Eur J Org Chem. 2017;2017:2764– tiple cloned gene insertions at the δ sequences of Saccharomyces
79. https://​doi.​org/​10.​1002/​ejoc.​20170​0070. cerevisiae. Appl Microbiol Biotechnol. 1997;48:339–45. https://​
7. Dietrich JA, Fortman JL, Steen EJ. Recombinant host cells for the doi.​org/​10.​1007/​s0025​30051​059.
production of malonate. 2017. US patent WO 2013/134424 A1. 24. Mumberg D, Müller R, Funk M. Yeast vectors for the controlled
8. Gu S, Zhao Z, Yao Y, Li J, Tian C. Designing and constructing a expression of heterologous proteins in different genetic back-
novel artificial pathway for malonic acid production biologically. grounds. Gene. 1995;156:119–22. https://​doi.​org/​10.​1016/​0378-​
Front Bioeng Biotechnol. 2022. https://​doi.​org/​10.​3389/​fbioe.​ 1119(95)​00037-7.
2021.​820507. 25. Dada O, SahaidKalil M, Yusoff WMW. Effects of inoculum and
9. Song CW, Kim JW, Cho IJ, Lee SY. Metabolic engineering substrate concentrations in anaerobic fermentation of treated rice
of Escherichia coli for the production of 3-hydroxypropionic bran to acetone, butanol and ethanol. Bacteriol J. 2012;2:79–89.
acid and malonic acid through β-alanine route. ACS Syn Bio. https://​doi.​org/​10.​3923/​bj.​2012.​79.​89.
2016;5:1256–63. https://​doi.​org/​10.​1021/​acssy​nbio.​6b000​07. 26. Ji R-Y, Ding Y, Shi T-Q, Lin L, Huang H, Gao Z, et al. Metabolic
10. Borodina I, Kildegaard KR, Jensen NB, Blicher TH, Maury J, engineering of yeast for the production of 3-hydroxypropionic
Sherstyk S, et al. Establishing a synthetic pathway for high-level acid. Front Microbiol. 2018. https://​doi.​org/​10.​3389/​fmicb.​2018.​
production of 3-hydroxypropionic acid in Saccharomyces cerevi- 02185.
siae via β-alanine. Metab Eng. 2015;27:57–64. https://​doi.​org/​10.​ 27. Zhang X, Liu Y, Wang J, Zhao Y, Deng Y. Biosynthesis
1016/j.​ymben.​2014.​10.​003. of adipic acid in metabolically engineered Saccharomyces

13

338 Systems Microbiology and Biomanufacturing (2023) 3:328–338

cerevisiae. J Microbiol. 2020;58:1065–75. https://​doi.​org/​10.​ coding for the inducible 4-aminobutyric-acid-specific permease
1007/​s12275-​020-​0261-7. in Saccharomyces cerevisiae. Eur J Biochem. 1989;181:357–61.
28. Chen N, Wang J, Zhao Y, Deng Y. Metabolic engineering of Sac- https://​doi.​org/​10.​1111/j.​1432-​1033.​1989.​tb147​32.x.
charomyces cerevisiae for efficient production of glucaric acid at 32. Erasmus DJ, van der Merwe GK, van Vuuren HJJ. Genome-wide
high titer. Microb Cell Fact. 2018;17:67. https://​doi.​org/​10.​1186/​ expression analyses: metabolic adaptation of Saccharomyces cer-
s12934-​018-​0914-y. evisiae to high sugar stress. FEMS Yeast Res. 2003;3:375–99.
29. Brewster NK, Val DL, Walker ME, Wallace JC. Regulation of https://​doi.​org/​10.​1016/​s1567-​1356(02)​00203-9.
pyruvate carboxylase isozyme (PYC1, PYC2) gene expression in 3 3. Fuhrer T, Chen L, Sauer U, Vitkup D. Computational prediction
Saccharomyces cerevisiae during fermentative and nonfermenta- and experimental verification of the gene encoding the N ­ AD+/
+
tive growth. Arch Biochem Biophys. 1994;311:62–71. https://d​ oi.​ NADP -dependent succinate semialdehyde dehydrogenase in
org/​10.​1006/​abbi.​1994.​1209. Escherichia coli. J Bacteriol. 2007;189:8073–8. https://​doi.​org/​
30. Bach B, Meudec E, Lepoutre J-P, Rossignol T, Blondin B, Dequin 10.​1128/​JB.​01027-​07.
S, et al. New insights into γ-aminobutyric acid catabolism: evi- 34. Vemuri GN, Eiteman MA, McEwen JE, Olsson L, Nielsen J.
dence for γ-hydroxybutyric acid and polyhydroxybutyrate syn- Increasing NADH oxidation reduces overflow metabolism in Sac-
thesis in Saccharomyces cerevisiae. Appl Environ Microbiol. charomyces cerevisiae. Proc Natl Acad Sci. 2007;104:2402–7.
2009;75:4231–9. https://​doi.​org/​10.​1128/​AEM.​00051-​09. https://​doi.​org/​10.​1073/​pnas.​06074​69104.
31. Vissers S, Andre B, Muyldermans F, Grenson M. Positive and neg-
ative regulatory elements control the expression of the UGA4 gene

13