Metabolic pathway engineering for fatty acid ethyl ester production in Saccharomyces cerevisiae using stable chromosomal integration
Fatty acid ethyl esters are fatty acid derived molecules similar to first generation biodiesel (fatty acid methyl esters; FAMEs) which can be produced in a microbial cell factory. Saccharomyces cerevisiae is a suitable candidate for microbial large scale and long term cultivations, which is the typical industrial production setting for biofuels. It is crucial to conserve the metabolic design of the cell factory during industrial cultivation conditions that require extensive propagation. Genetic modifications therefore have to be introduced in a stable manner. Here, several metabolic engineering strategies for improved production of fatty acid ethyl esters in S. cerevisiae were combined and the genes were stably expressed from the organisms’ chromosomes. A wax ester synthase (ws2) was expressed in different yeast strains with an engineered acetyl-CoA and fatty acid metabolism. Thus, we compared expression of ws2 with and without overexpression of alcohol dehydrogenase (ADH2), acetaldehyde dehydrogenase (ALD6) and acetyl-CoA synthetase (acsSEL641P) and further evaluated additional overexpression of a mutant version of acetyl-CoA decarboxylase (ACC1S1157A,S659A) and the acyl-CoA binding protein (ACB1). The combined engineering efforts of the implementation of ws2,ADH2, ALD6 and acsSEL641P, ACC1S1157A,S659A and ACB1 in a S. cerevisiae strain lacking storage lipid formation (are1Δ, are2Δ, dga1Δ and lro1Δ) and β-oxidation (pox1Δ) resulted in a 4.1-fold improvement compared with sole expression of ws2 in S. cerevisiae.
KeywordsYeast Industrial strain Chromosomal integration Homologous recombination Fatty acid ethyl ester (FAEE) Metabolic pathway engineering
Fatty acid ethyl esters (FAEEs) are compounds with molecular properties similar to crude oil derived diesel and first generation biodiesel (fatty acid methyl esters; FAMEs) and can be produced in a microbial cell factory [4, 13, 26]. Microbial conversion of sugars derived from plant biomass into FAEEs could pave the road towards a production method with a more consistent, scalable and renewable commodity supply which is more cost efficient for biodiesel than the current chemical transesterification of vegetable oils and methanol . During microbial production, the precursors acyl-coenzyme A (acyl-CoA) and ethanol are converted to FAEEs by a transesterification reaction catalyzed by a wax ester synthase . This process could replace the current biodiesel production method for a more sustainable future.
The first metabolic engineering projects for the creation of a microbial cell factory producing FAEEs were performed in Escherichia coli [12, 13, 30]. Combined efforts resulted in a highly engineered E. coli strain with cytosolic expression of thioesterase ’TesA, overexpression of acyl-CoA ligase, inhibition of β-oxidation, introduction of the alcohol synthesis pathway from Zymomonas mobilis, and high expression of a heterologous wax ester synthase (atfA), which led to an FAEE production of 674 mg/l during cultivation on glucose .
FAEEs are meant to become bulk products for the transport industry which implies large scale and long term microbial cultivations in media with low quality. The natural ethanol producer Saccharomyces cerevisiae has a long history of being used in industrial large scale fermentations and therefore represents an ideal candidate for this purpose. Due to the requirements for production in large scale, including extensive propagation of the yeast strain, it is important to create microbial strains with a stable genotype. However, most previous modifications for FAEE production in yeast were based on expression of pathway enzymes from self-replicating plasmids, which might cause a metabolic burden on the host cell. In addition, this leads to genetic instability, especially for heterologous genes, mainly caused by segregational and structural instability, which makes these strains non-suitable for industrial fermentations [11, 14, 32].
In yeast, the formation of triacylglycerols (TAGs), steryl esters (both serving as storage lipids) and the β-oxidation pathway compete for acyl-CoA with the production of FAEEs. Therefore, these pathways have previously been inactivated by disruption of DGA1, LRO1 (both encoding diacylglycerol acyltransferases), ARE1, ARE2 (both encoding sterol acyltransferases) and POX1 (encoding acyl-CoA oxidase) [12, 33]. The knock-out strain was used to express a wax ester synthase from Marinobacter hydrocarbonoclasticus (ws2)  and resulted in a yeast strain with the inability to synthesize storage lipids and an FAEE production of 17.2 mg/l . Furthermore, the combined expression of Acc1S1157A,S659A, an acetyl-CoA carboxylase variant, in which posttranslational regulation by phosphorylation is abolished, and ws2 resulted in an FAEE production of 15.8 mg/l . In another recent attempt to increase the production of FAEEs in yeast, overexpression of native ACC1 and FAS1 and FAS2 (both subunits of fatty acid synthase), in combination with deletion of the β-oxidation pathway and expression of a wax ester synthase from Acinetobacter baylyi (AtfA) resulted in an FAEE production of 5.44 mg/l . Following up on the previous work, a FAEE titer of 13.9 mg/l was achieved with an abolished β-oxidation pathway, disruption of acyl-CoA synthetase gene FAA2 and acyl-CoA transporter gene PXA2 as well as elimination of acyl-CoA binding protein Acb1 in combination with expression of codon optimized atfA under nitrogen limitation cultivation conditions . Finally, one study focused on increasing the expression level of the wax ester synthase, when six copies of ws2 were randomly integrated into the genome, which resulted in a titer of 34 mg/l .
Materials and methods
During this research, yeast strain S. cerevisiae CEN.PK 113-11C (MATa MAL2-8cSUC2ura3-52 his3-Δ1) was used which was kindly provided by P. Kötter, University of Frankfurt, Germany. Additionally, strain JV04, based on strain CEN.PK113-11C, was used which contains five different gene deletions (MATa MAL2-8cSUC2ura3-52 his3-Δ1 are1Δ dga1Δ are2Δ lro1Δ pox1Δ) to prevent the conversion of free fatty acids to sterol esters, triacylglycerides and their degradation to acetyl-CoA via the β-oxidation pathway. With the exception that JV04 contains an additional his3-Δ1 mutation, the strain is the same as JV03 (MATa MAL2-8cSUC2ura3-52 are1Δ dga1Δ are2Δ lro1Δ pox1Δ) and was constructed following the same gene deletion and strain crossing procedure . Escherichia coli strain DH5α  was used for amplification of the plasmids used for pathway engineering.
Yeast pathway construction
List of strains used in this study
Genotype or relevant characteristicsa
MATa MAL2-8cSUC2 ura3-52 his3-Δ1
CEN.PK113-11C ws2 ADH2↑ ALD6↑ acsseL641P
WS + EDP
CEN.PK113-11C ws2 ADH2↑ ALD6↑ acsseL641PACC1S659A,S1157AACB1↑
WS + EDP + FASP
MATa MAL2-8cSUC2ura3-52 his3-Δ1 are1Δ dga1Δ are2Δ lro1Δ pox1Δ
JV04 ws2 ADH2↑ ALD6↑ acsseL641P
WS + EDP
JV04 ws2 ADH2↑ ALD6↑ acsseL641PACC1S659A,S1157AACB1↑
WS + EDP + FASP
List of plasmids used in this study
specific recognition sites XI-5 up/down; URA3 selection marker flanked by directed repeats
specific recognition sites XI-3 up/down; URA3 selection marker flanked by directed repeats
2 µm expression plasmid containing ACC1S659A,S1157A and ACB1
2 µm expression plasmid containing ADH2, ALD6, acsseL641P and ws2
2 µm expression plasmid containing HIS3 selection marker
List of oligonucleotide primers
Oligonucleotide primers used for construction of the linear integration cassettes
Additional oligonucleotide primers used for verification and sequencing of correct integration insert
genome XI-5 up-F
genome XI-5 down-R
genome XI-3 up-F
genome XI-3 down-R
Two-step fusion PCR
A two-step fusion PCR  was performed to create the linear chromosomal integration modules if necessary. The first PCR reaction was performed by mixing 100–300 ng of DNA-bricks in equimolar amounts with Phusion GC buffer, dNTPs, and Phusion polymerase in a total volume of 25 μl. The PCR conditions for amplification were 98 °C for 3 min, 15 cycles of 98 °C for 10 s, 68 °C for 30 s, and 72 °C for 30 s/kb, followed by a 10 min incubation at 72 °C. 2 μl of unpurified PCR product were used as a template for the second PCR-step. Additionally, a forward and reverse primer, dNTPs, Phusion GC buffer and Phusion polymerase were added to the PCR mix in a final volume of 50 µl. The reaction was performed identically to the PCR reaction of the DNA bricks. These fusion PCR-products (modules) were separated on a 1 % agarose gel and purified with a gel purification kit as above. The concentration was determined by a Nanodrop 2,000 spectrophotometer.
Homologous recombination and transformation
The different DNA modules formed the chromosomal integration cassettes and were transformed into S. cerevisiae by electroporation at 1.5 kV, 10 μF and 200 Ω using a 0.2 cm gap electroporation cuvette and an Eppendorf Eporator (Eppendorf AG, Hamburg, Germany). 400–800 ng of each of the DNA modules were mixed with the 50 μl of competent cells. Competent cells were prepared following instructions . After electro-transformation the cells were incubated for 3 h in 1 M sorbitol at 30 °C and plated on appropriate dropout plates. Colonies were screened by PCR for correctly integrated cassettes and the PCR fragments spanning the entire integrated constructs were sequenced (Eurofins MWG Operon, Ebersberg, Germany). The additional primers used for the verification are listed in Table 3.
The first chromosomal integration cassette contained only the wax ester synthase gene (ws2) controlled by the TEF1 promotor and the ADH1 terminator and the resulting strains functioned as reference strains. The second linear integration cassette contained over-expressed genes ADH2, ALD6 and acsSEL641P, which improve the provision of the precursor acetyl-CoA, together with ws2 [2, 5, 17]. The genes were regulated by constitutively active promoters HXT7 (truncated; ), PGK1 and TEF1 and terminators ADH2, CYC1, and ADH1, respectively. Both cassettes (1 and 2) contained the chromosomal recognition sites to be integrated into chromosome XI at integration position 5 characterized in a previous study  and contained a URA3 selection marker. The third linear integration cassette contained genes ACC1S1157A,S659A and ACB1 controlled by, respectively, PGK1 and TEF1 promoters and ADH1 and CYC1 terminators. This cassette was integrated into chromosome XI at integration position 3  and contained a HIS3 selection marker. Figure 1 shows the different linear integration fragments, including promoters, terminators, recognition sites and selection marker. It has been demonstrated that both chromosomal integration sites, XI-5 and XI-3, resulted in the highest beta-galactosidase activity during a comparison study for strains with the lacZ cassette integrated at 14 integration sites .
Media and growth conditions
All described strains of S. cerevisiae were cultivated in shake flasks at 30 °C and continuous shaking at 200 rpm in synthetic dextrose (SD) medium containing 20 g l−1 glucose, 6.7 g l−1 yeast nitrogen base without amino acids (YNB-AA) (Formedium, Hunstanton, UK), and complete supplement mixture (0.750 g l −1; CSM, Formedium) lacking, if necessary, uracil and histidine, respectively.
Samples were taken from the cultures after 100 h and handled as previously described . The quantification of fatty acid ethyl esters was performed with a Focus ICQ single quadrupole GC-MS from Thermo Fisher Scientific with a Zebron (ZB-WAX) GC column with 30 m × 25 mm internal diameter and 0.25 µm film thickness (Phenomenex, Macclesfield, UK). The precise GC-MS conditions, compound identification and quantification conditions were described previously .
Metabolic pathway integration
In the background strain JV04, the conversion of free fatty acids towards sterol esters, triacylglycerides and acetyl-CoA via the β-oxidation pathway were eliminated and combined with the chromosomal overexpression of ws2 to create the reference strain, which resulted in 506 ± 58 μg total FAEEs gCDW−1 (BdJ13). The additional expression of genes ADH2, ALD6, acsSEL641P, yielded 797 ± 69 μg total FAEEs gCDW−1, which was a 1.6-fold improvement compared to BdJ13. Strain BdJ15, in which the strategies of expressing ws2,ADH2, ALD6, acsSEL641P, ACC1S1157A,S659A, ACB1 and eliminating the fatty acid competing pathways were combined, resulted in a yield of 1,072 ± 160 μg total FAEEs gCDW−1 which was the highest FAEE yield detected here (Fig. 3b).
The most extensively engineered strain (BdJ15) resulted in a 4.1-fold improvement of the FAEE yield compared with the reference strain, in which S. cerevisiae only carried the ws2 gene (BdJ10). Overall, expression of genes in the JV04 background resulted in an 1.85 ± 0.11-fold improvement in FAEE production compared with the respective counterpart strains based on CEN.PK113-11C (Fig. 3a, b). However, the mutant strains based on JV04 resulted in a lower final biomass concentration after substrate depletion (data not shown). Therefore, while comparing the titers of the different strains with CEN.PK113-11C and JV04 background, the BdJ10, BdJ11 and BdJ12 strains showed titers of, respectively, 1.3 ± 0.3, 2.3 ± 0.4 and 3.5 ± 0.8 mg l−1 versus FAEE titers of strains BdJ13, BdJ14 and BdJ15 of 2.0 ± 0.2, 2.2 ± 0.2 and 4.4 ± 0.7 mg l−1, respectively (Fig. 3c, d).
Chain length distribution
All constructed strains contain a wax ester synthase, which catalyzes the transesterification reaction that converts ethanol and acyl-CoA into FAEEs. In S. cerevisiae, the predominant fatty acids are sixteen and eighteen carbon atoms long and exist in their saturated form as well as in their mono-unsaturated form . The most obvious change in FAEE chain length distribution occurred for unsaturated 16:1 FAEEs after elimination of the fatty acid competing pathways (Fig. 3). BdJ10-BdJ12 produced 27 ± 3.5 % of unsaturated 16:1 FAEE, whereas BdJ13-BdJ15 produced 56 ± 6.5 % of this FAEE. This increase is almost solely responsible for the improved production of total FAEEs. The fraction of saturated 16:0 FAEE and saturated 18:0 FAEE of strains BdJ10-BdJ12 compared to strain BdJ13-BdJ15 on the other hand were reduced from, respectively, 24 ± 1 to 14 ± 3 % and from 20 ± 1 to 8 ± 2 %. No significant difference for the 18:1 unsaturated FAEE content was detected. Also the implementation of genes ADH2, ALD6 and acsSEL641P or ACC1S1157A,S659A and ACB1 did not show any significant difference for the distribution of the carbon-chain length.
Expression of genes ws2,ADH2, ALD6 and acsSEL641P in S. cerevisiae for improving the production of FAEEs was previously evaluated using multicopy 2 μg plasmids, but resulted in a slightly lower and a more unstable yield of FAEEs than reported here (408 μg total FAEEs gCDW−1 compared to 463 μg total FAEEs gCDW−1) . The expression of ACC1S1157A,S659A and ACB1 in S. cerevisiae was also earlier reported to have a positive effect on the FAEE production [25, 27]. The combined expression of the above listed genes on two 2 μg plasmids with a total size of 15 and 18 kb however, decreased the yield of FAEE in S. cerevisiae (unpublished results). It was thought that the expression of these multiple genes from high-copy plasmids resulted in a metabolic burden, which caused a (partial) loss of expression resulting in a negative effect on FAEE production. Although it was not explicitly shown here, chromosomal integration has been demonstrated to be a more stable way of expressing genes compared to episomal plasmid expression [11, 18, 34]. We also noticed in this study a much reduced clonal variation of the integrated strains in comparison to the plasmid-based strains used in previous studies (data not shown). Additionally, the most favorable (with regard to growth impairment and gene expression) integration sites investigated by Mikkelsen et al.  separated by essential genetic elements, were used in this work and therefore reduced the probability of loss of the integration cassettes. Also, the use of repetitive sequences within the integration cassettes and the introduction of sequence homology in multiple tandem inserts were avoided, because of the risk of homologous recombination and chromosomal rearrangements [11, 18, 34].
In contrast to the observed FAEE decrease in a plasmid-based strain, here we show a clear improvement of the FAEE yield in BdJ15, which contains all the above discussed modifications and this clearly shows the importance of stable expression through genome integration when several metabolic engineering strategies have to be combined and when large genes have to be expressed.
Earlier, several metabolic engineering strategies have been implemented for improving FAEE production by S. cerevisiae. These strategies mostly focused on the production of acyl-CoA, its precursors acetyl-CoA [1, 5, 28] and malonyl-CoA , the elimination of the production of fatty acid related non-vital byproducts  and improved supply of NADPH . Here, the combined implementation of ws2,ADH2, ALD6, acsSEL641P, ACC1S1157A,S659A and ACB1 into a are1Δ dga1Δ are2Δ lro1Δ pox1Δ S. cerevisiae strain resulted in a FAEE yield of 1.1 mg total FAEEs gCDW−1 and a titer of 4.4 mg l−1 during growth on glucose. Despite the clear improvement obtained through the combined metabolic engineering strategies, the chromosomal integration leads to lower FAEE titers than reported in earlier studies. We believe this is primarily due to a limitation of the wax ester synthase, which in all previous studies was expressed in multi-copies, either from multi-copy plasmids or through multi-copy integration. This is consistent with our earlier study which showed that six randomly integrated copies of the gene ws2 into the chromosomes of S. cerevisiae improved the FAEE production compared with a single copy expression of ws2 . The integration of the above described pathway into the chromosome of the strain with six integrated copies of ws2, which is based on an ura3-52 selection marker, could possibly improve the yield of FAEEs and take away the limiting esterification step of the pathway. However, for large and complex pathway integrations, as presented here, there is a demand of recycling the auxotrophic selection marker of the strain which comes along with several rounds of strain transformation and selection. As reported earlier, Jensen et al.  demonstrated that selection marker loop-out for a new round of metabolic engineering resulted in low efficiency for correct integrations, which was thought to be due to the proximity of the integration sites and possible homologous recombination resulting in a loss of pathway genes. We attempted to integrate the above described pathway in two rounds of transformation and marker recycling into the strain with six integrated copies of ws2, but a drastic loss of product yield was observed after URA3 selection marker loop-out, which indicated a loss of pathway genes (data not shown).
Acyl-CoA binding protein (Acb1) plays an important role for facilitating intracellular transport of acyl-CoA esters to the membrane of the endoplasmic reticulum and lipid bodies for phospholipid and TAG biosynthesis . Additionally, it was reported that Acb1 had an attenuating effect on the negative feedback mechanism of acyl-CoAs on fatty acid synthase, acetyl-CoA carboxylase and long chain acyl-CoA synthetase [10, 15, 22]. However, it remains unclear which engineering strategy concerning ACB1 results in the highest FAEE production. Both the overexpression of the ACB1 gene  to improve the attenuation of the negative feedback loop as well as its deletion to prevent transport of acyl-CoA towards unwanted cellular locations  resulted in a positive effect on FAEE production. Here, FAEE production improved with up-regulation of ACB1 (in combination with ACC1S1157A,S659A), but understanding the exact mechanisms will require further investigations.
Distribution acyl moiety chain length
The mechanism for distribution and regulation of the carbon chain length of fatty acids derived products in S. cerevisiae is only partly known [6, 9]. Here, we observed a large increase in 16:1 unsaturated FAEE in mutant strains with disrupted DGA1, LRO1 (both encoding diacylglycerol acyltransferases), ARE1, ARE2 (both encoding sterol acyltransferases) and POX1 (encoding acyl-CoA oxidase) [12, 33]. This increase was surprising, because earlier reports with a yeast strain containing the same gene deletions did not observe any difference in the relative amount of 16:1 free fatty acids . However, in this earlier study a high-copy expression of ws2 was used, which may affect the fatty acid composition in the produced FAEEs. Another study, in which storage lipid were eliminated (are1Δ dga1Δ are2Δ lro1Δ) also showed a much larger ratio of unsaturated fatty acids compared to saturated fatty acids during different growth conditions , whereas a third study only detected a smaller change for the same factor . The disrupted capacity to store fatty acids as TAGs was mentioned to be critical for buffering the regulation of fatty acid composition .
The synthesis of the acyl-chain is a major energy- and carbon consuming metabolic step and acyl-CoA also functions as a substrate for vital membrane phospholipids. However, the mechanism regulating the total amount and the specific composition of acyl-CoA and its downstream products, which is under several levels of metabolic and transcriptional control, remains largely unclear . In S. cerevisiae, the acyl chain length is partly determined by the fatty acid synthase complex  and recently it was found that acetyl-CoA decarboxylation also has a fundamental impact on the relative distribution and quantity of acyl-CoA chains in cellular lipids . In S. cerevisiae, Snf1-dependent phosphorylation attenuates the activity of acetyl-CoA carboxylase (Acc1), for maintaining the appropriate acyl-chain length distribution and the ratio of C16 versus C18 fatty acids . It is also thought that the specific molecular composition of acyl-CoA and its downstream products directly regulates genes of the fatty acid synthesis pathway via differential interaction with the repressor Opi1 . In S. cerevisiae, a higher content of free fatty acids (and related molecules) was yielded by the elimination of two post-translational sites in acetyl-CoA decarboxylase (ACC1) [3, 9, 25].
However, here it was shown that specifically the production of 16:1 unsaturated FAEEs was increased due to the elimination of genes DGA1, LRO1, ARE1, ARE2 and POX1. Most likely eliminating production of TAGs and sterol esters influence the specific regulation of acyl-CoA formation, which subsequently changes the composition of the produced FAEEs. The sensitive balance between acyl-CoA, its downstream products, mainly TAGs in wild type yeast cells, which was interrupted, might cause a change in the carbon chain distribution and influence the energy dependent fatty acid elongation. In addition, it is known that ws2 prefers longer alcohols and has a low affinity towards short alcohols, like the ethanol used for FAEEs produced by S. cerevisiae . Matching the demands of the enzymes which catalyze the downstream reaction with the substrate specificity could likely improve the productivity substantially. Understanding the regulatory mechanisms determining the composition and quantity of downstream acyl-CoA derived intermediates is crucial for future production of designer fatty acid derived products, like FAEEs.
Fatty acid ethyl esters are suitable molecules for substitution of crude oil and current FAME biodiesel. Here, it was demonstrated that chromosomal integration of the metabolic engineering strategy into the chromosomes of S. cerevisiae resulted in stable expression and it was therefore possible to quantitatively evaluate complex engineering strategies involving several different genetic modifications. Furthermore, the here presented strains represent a very good starting point for further engineering, e.g. evaluation of the effect of increased expression of the wax ester synthase, which is likely to be limiting production the engineered strains.
BWDJ has performed all experimental work, analyzed the data and drafted the manuscript. BWDJ, SS, VS and JN designed and coordinated the work and edited the manuscript. JOV-R provided strain JV04.
This research has been funded by the Knut and Alice Wallenberg Foundation, Vetenskapsrådet, FORMAS, and Ångpanneföreningens Forskningsstiftelse.
- 5.de Jong BW, Shi S, Siewers V, Nielsen J (2014) Improved production of fatty acid ethyl esters in Saccharomyces cerevisiae through up-regulation of the ethanol degradation pathway and expression of the heterologous phosphoketolase pathway. Microb Cell Fact 13(1):39PubMedCentralPubMedCrossRefGoogle Scholar
- 9.Hofbauer HF, Schopf FH, Schleifer H, Knittelfelder OL, Pieber B, Rechberger GN, Wolinski H, Gaspar ML, Kappe CO, Stadlmann J, Mechtler K, Zenz A, Lohner K, Tehlivets O, Henry SA, Kohlwein SD (2014) Regulation of gene expression through a transcriptional repressor that senses acyl-chain length in membrane phospholipids. Dev Cell 29(6):729–739PubMedCentralPubMedCrossRefGoogle Scholar
- 10.Hong S-P, Xue Z (2008) Manipulation of acyl-CoA binding protein expression for altered lipid production in microbial hosts. Patent: US20090291479 A1Google Scholar
- 20.Okuyama H, Saito M, Joshi VC, Gunsberg S, Wakil SJ (1979) Regulation by temperature of the chain-lenght of fatty-acids in yeast. J Biol Chem 254(24):2281–2284Google Scholar
- 24.Sambrook J, Russell DW (2001) Molecular cloning: a laboratory manual. In: Argentine J (ed), 3rd ed. Cold Spring Harbor Laboratory Press, New York, ISBN 0-87969-87579-87965Google Scholar