Engineering Oleaginous Yeast Rhodosporidium Toruloides for Overproduction of Fatty Acid Ethyl Esters


 BackgroundProduction of biofuels and green chemicals by microbes is currently of great interest due to the increasingly limited reserves of fossil fuels. Biodiesel, especially fatty acid ethyl esters (FAEEs), is considered as an attractive alternative because of its similarity with petrodiesel and compatibility with existing infrastructures. Cost-efficient bio-production of FAEEs requires a highly lipogenic production host that is suitable for large-scale fermentation. As a non-model oleaginous yeast that can be cultured to an extremely high cell density and accumulate over 70 % biomass as lipids, Rhodosporidium toruloides represents an attractive host for FAEEs production. ResultsWe first constructed the FAEE biosynthetic pathways in R. toruloides by introducing various wax ester synthase genes from different sources, and the bifunctional wax ester synthase /acyl-CoA-diacyglycerol acyltransferase (WS/DGAT) gene from Acinetobacter baylyi was successfully expressed, leading to a production of 826 mg/L FAEEs in shake-flask fermentation. We then mutated this bifunctional enzyme to abolish the DGAT activity, and further improved the titer to 1.02 g/L. Finally, by fed-batch fermentation in a 1-L fermenter, the titer of FAEEs reached 9.2 g/L. It is worth mentioning that most of the produced FAEEs were secreted out of the cell, which should greatly reduce the cost of downstream processing.ConclusionWe achieved the highest FAEEs production in yeast with a final titer of 9.2 g/L and demonstrated that the engineered R. toruloides has the potential to serve as a platform strain for efficient production of fatty acid derived molecules.


Results
We rst constructed the FAEE biosynthetic pathways in R. toruloides by introducing various wax ester synthase genes from different sources, and the bifunctional wax ester synthase /acyl-CoA-diacyglycerol acyltransferase (WS/DGAT) gene from Acinetobacter baylyi was successfully expressed, leading to a production of 826 mg/L FAEEs in shake-ask fermentation. We then mutated this bifunctional enzyme to abolish the DGAT activity, and further improved the titer to 1.02 g/L. Finally, by fed-batch fermentation in a 1-L fermenter, the titer of FAEEs reached 9.2 g/L. It is worth mentioning that most of the produced FAEEs were secreted out of the cell, which should greatly reduce the cost of downstream processing.

Conclusion
We achieved the highest FAEEs production in yeast with a nal titer of 9.2 g/L and demonstrated that the engineered R. toruloides has the potential to serve as a platform strain for e cient production of fatty acid derived molecules.

Background
The global energy demand and environmental concerns have attracted worldwide attention to green and sustainable energy sources [1]. As one of the most promising alternative energy sources, biodiesel has high energy density, and is compatible with current infrastructure [2]. Moreover, it showed several advantages over petrodiesel, such as higher lubricity, and lower tailpipe emissions [3,4]. Biodiesel consists of long-chain alkyl esters, mainly fatty acid methyl esters (FAMEs) and fatty acid ethyl esters (FAEEs) [5]. Currently, commercial strategies for biodiesel production mainly use plant oils or animal fats as feedstock to produce FAMEs through chemical transesteri cation [6,7]. However, these strategies compete for lands and materials required for food production and also cause environmental problems [8,9]. Therefore, it is desirable to use renewable plant biomass as a feedstock to produce FAEEs by microbial fermentation [10].
Over the past decades, various microbial hosts such as Escherichia coli, Saccharomyces cerevisiae, and Yarrowia lipolytica have been engineered to produce FAEEs. In E. coli, the production of FAEEs was achieved by heterologously expressing genes encoding the pyruvate decarboxylase (PDC), alcohol dehydrogenase (ADH), and wax ester synthase/acyl-CoA-diacyglycerol acyltransferase (WS/DGAT) gene from A. baylyi [5]. The maximum FAEEs titer reached 19 g/L by adding exogenous oleic acids in fed-batch pilot scale fermentation [11]. Röttig and coworkers showed that use of a mutant AbWS from A. baylyi (Ile355Gly) led to a higher FAEEs titer than the wild-type AbWS in E. coli [12]. In S. cerevisiae, Shi and coworkers constructed a similar FAEE biosynthetic pathway and demonstrated the WS from Marinobacter hydrocarbonsticus led to the highest FAEEs titer of 6.3 mg/L [13]. Later, Yu and coworkers improved the titer to 0.52 g/L FAEEs by adding exogenous fatty acids [14]. In both E. coli and S. cerevisiae, the production of FAEEs was regarded to be limited by the supply of lipid precursors, and the addition of exogenous oleic acids can improve the nal titer signi cantly [11,14], which suggests that use of host strains with robust lipid producing capacity such as oleaginous yeasts may be advantageous. Indeed, Xu and coworkers engineered a representative oleaginous yeast Y. lipolytica to produce FAEEs and found that the subcellular localization of WSs was essential for the production of FAEEs in the strain [15].
Cytosolic expression of WS from A. baylyi ADP1 only resulted in marginal production at 7.1 mg/L, and when WS was targeted to the endoplasmic reticulum (ER), the FAEEs titer was improved signi cantly to 136.5 mg/L. In a further study, Gao and coworkers revealed that the expression of WS from M. hydrocarbonsticus led to the highest FAEEs titer of 0.4 g/L with additional exogenous ethanol, and through pathway optimization for supplying cytosolic acyl-CoA, the production titer reached 1.18 g/L [16].
In addition to Y. lipolytica, R. toruloides is another promising oleaginous yeast. Compared with Y. lipolytica, R. toruloides produces a higher titer of lipids [17], and can naturally utilize a variety of carbon sources derived from plants, including xylose and cellobiose [18], thereby exhibiting a broader industrial application prospect in the production of lipid related products. However, metabolic engineering of R. toruloides was hindered by the lack of e cient genetic manipulation tools [17,19], and only limited metabolic engineering efforts were reported [17,18]. To date, there is no report using this strain to produce biodiesels using a metabolic engineering approach.
In the present study, we explored the potential of R. toruloides to produce FAEEs. Five WSs previously reported in S. cerevisiae were codon optimized and introduced to R. toruloides [13]. However, only the AbWS from A. baylyi was successfully expressed, and the recombinant strain successfully produced FAEEs by adding exogenous ethanol to the culture. Then, the AbWS was engineered to improve the production of FAEEs by abolishing its side activity. Finally, by carrying out fed-batch fermentation in a 1-L fermenter, the production reached a maximum of 9.2 g/L FAEEs (including both extracellular and intracellular titers). To the best of our knowledge, this engineered cell factory possessed the highest FAEEs production levels in a yeast host. Notably, the FAEEs produced by R. toruloides were mainly secreted outside the cell, which should greatly reduce the cost in the subsequent downstream processes.

Results
Overexpression of WSs for FAEEs production in R. toruloides WS catalyzes esteri cation of fatty acyl coenzyme A (fatty acyl-CoA) with alcohols of various chain lengths for synthesizing fatty esters. WSs derived from different sources have a wide range of substrate preferences, leading to different catalytic e ciencies. When it is expressed in a heterologous host, the source of the enzyme is a key factor to affect the catalytic e ciency.
To compare the e ciency of various WSs in R. toruloides, ve previously reported wax ester synthase genes [13] were selected and codon optimized. They were MhWS from M. hydrocarbonoclasticus DSM 8798 (GenBank ID: EF219377.1), AbWS from A. baylyi ADP1 (GenBank ID: AF529086), RoWS from Rhodococcus opacus PD630 (GenBank ID:GQ923886), MmWS from Mus musculus C57BL/6 (GenBank ID: AY611032), and PaWS from Psychrobacter arcticus 273-4 (GenBank ID: YP_263530). A Flag tag was added at the start of each gene and subcloned into plasmid pKOCAR2. The resulting plasmids were transformed to R. toruloides Δku70, yielding the recombinant strains named Δku70-MhWS, Δku70-AbWS, Δku70-RoWS, Δku70-MmWS and Δku70-PaWS ( Table 1). The expression of WSs was examined by Western blot method. It was found that only AbWS with a molecular weight of 53.78 kD was successfully expressed (Fig. 1a), with a band around 53 kDa obtained. Therefore, the recombinant Δku70-AbWS was used as the candidate for further research.

Exploration of the optimal exogenous ethanol concentration for FAEEs production
Although ethanol is required for FAEEs production, its toxicity may affect cell growth and lead to a decreased FAEEs yield. Therefore, we sought to determine an optimal ethanol concentration for both minimizing the substrate cost and maximizing the production titer. Ethanol was added to the medium at a nal concentration of 1%, 3%, 5%, and 7%, respectively. Figure 2a showed the effect of different ethanol concentrations on cell growth of the recombinant Δku70-AbWS strain in shake-ask fermentation. As expected, ethanol had an inhibitory effect on cell growth. When the ethanol concentration was 1%, the effect on cell growth was inconspicuous. But when the ethanol concentration was above 3%, the effect on cell growth was signi cant. The addition of ethanol also had a similar effect on the nal biomass. Figure 2b showed the effect of different ethanol concentrations on biomass formation and glucose consumption. When the ethanol concentration was above 3%, the nal biomass almost decreased by half, and the glucose consumption was signi cantly reduced.
The extracellular and intracellular FAEEs were extracted by different methods, and the production titers were shown in Fig. 3. When the ethanol concentration was 1% and 3%, the extracellular FAEEs titers were signi cantly lower than that of 5% and 7%. A similar trend was also noted for the production of intracellular lipids analyzed by thin layer chromatography (TLC) (Fig. S2). When the ethanol concentration was 1% and 3%, the fatty acid precursor was mainly stored in the triglycerides (TAG) form, and when the ethanol concentration rose to 5% and 7%, most of the fatty acid precursor owed to the synthesis of FAEEs. Taken together, the maximum FAEEs titer was 826 mg/L, including 625 mg/L extracellular and 201 mg/L intracellular titers, under the 5% exogenous ethanol condition (Fig. 3). When the ethanol concentration was further increased to 7%, the FAEEs titer did not show a further increase. Thus, 5% is considered as the optimal ethanol concentration for FAEEs production. It is worth mentioning that the background strain Δku70 also appeared to produce a small amount of extracellular FAEEs after adding exogenous ethanol (Fig. S3). This is possibly due to the existence of native lipases in the strain, which can catalyze the esteri cation process [20].
Modi cation of the AbWS enzyme to improve the FAEEs production As reported previously, a point mutation of AbWS at residue 355 from glycine to isoleucine led to a shifted substrate selectivity toward shorter chain alcohols and an impaired DGAT activity [12]. Guided by this nding, we created the same mutant AbWS* (Fig. 4a) and introduced it to the Δku70 strain, yielding a recombinant Δku70-AbWS* strain. The extracellular and intracellular FAEEs were measured respectively, and the production titers were shown in Fig. 4b. Compared with the wild-type AbWS, the mutant showed higher e ciency in producing FAEEs. The maximum FAEEs titer reached 1024 mg/L under the addition of 5% exogenous ethanol, including 810 mg/L extracellular and 214 mg/L intracellular titers. When ethanol concentration varied from 1% to 7%, compared to Δku70-AbWS, the recombinant Δku70-AbWS* strain tended to synthesize more palmitic acid ethyl ester and less linoleic acid ethyl ester ( Fig. 3 and Fig. 4b). To evaluate its performance in a bioreactor, we cultivated the Δku70-AbWS* strain in a 1-L bioreactor under aerobic fed-batch fermentation. The glucose concentration, biomass, and FAEEs concentration were all monitored over time. As shown in Fig. 5, after 48 hours of fermentation, 50 mL of glucose with a concentration of 500 g/L was added to the bioreactor to avoid carbon limitation. The FAEEs concentration continuously increased with fermentation and reached a maximum of 9.2 g/L, including 7.9 g/L extracellular and 1.3 g/L intracellular titers. During the whole fermentation process, a total of 47.5 g glucose was consumed, and 15.9 g biomass was formed.

Discussion
Biodiesel can contribute signi cantly to the development of sustainable transportation fuels in the future [1], and the development of "third-generation" biodiesels calls for alternative biosources, such as oilaccumulating microalgae, bacteria and yeasts. R. toruloides, a non-model oleaginous yeast, is an ideal organism to produce biodiesel because of its robust ability to accumulate single cell oils (SCOs) and capability to utilize various carbon sources [17]. Previously, this organism was used to produce biodiesel esters consisting of FAMEs through a "microbial oil process", which involves the production of bio-oil by microbial cells and an additional transesteri cation reaction [21]. Moreover, a whole-cell catalyst technology was reported to convert 73% total lipids into FAEEs in the strain [20]. However, the catalytic mechanism was unclear and it was only hypothesized that the lipases on the membrane of lipid droplets might play an important role in the synthesis of FAEEs. All these reported processes had to lyse the cells for lipids extraction, which increased the production costs.
Recently, the emerging use of WSs highlighted the rational construction of microbial cell factories for directly producing FAEEs by fermentation in various hosts [5,[11][12][13][14][15][16]22]. Inspired by these studies, we engineered a recombinant R. toruloides strain Δku70-AbWS* capable of directly producing FAEEs by fermentation. Compared with the previous studies [20,21], which needed to separate the oil-accumulation and the esteri cation processes, the strategy reported here synchronized such two courses, thus avoiding additional esteri cation procedure. Moreover, it will be accessible to control the synthesis of FAEEs through metabolic engineering, which is not available in previous reports [20,21].
Through fed-batch fermentation in a 1-L fermenter, we obtained a maximal FAEEs production of 9.2 g/L. This is to date the highest production titer in any yeast including the commonly used oleaginous yeast Y. lipolytica [16]. It is worth mentioning that the FAEEs produced by the engineered strain were mainly secreted outside the cell (7.9 g/L extracellular FAEEs were obtained). This is possibly due to its ability to assimilate hydrophobic carbon sources as substrates using an unknown transport system [17], which can work bidirectionally to secret the product of FAEEs. For example, the transporter FATP1 can not only assimilate fatty acids, but also work as a fatty acid exporter [23]. The secretion may greatly reduce the production cost by avoiding product extraction as reported in S. cerevisiae [13][14]24]. Therefore, the engineered cell factory serves as a potential platform for the industrial production of FAEEs. In the future, genetic manipulation strategies such as overexpressing genes related to acetyl-CoA accumulation and eliminating competing pathways (peroxisome β-oxidation and TAG biosynthesis) are expected to further improve the production level.
While we have demonstrated that R. toruloides has great potential for large-scale production of biodiesels during this study, a major bottleneck to implement an economically feasible process is the additional cost associated with providing exogenous ethanol. Although it was reported that R. toruloides contained ethanol synthesis genes encoding PDC and ADH [25], it failed to produce ethanol under current experimental conditions (data not shown). Also there has been no report that R. toruloides has the ability to produce ethanol. To avoid the need to feed ethanol, two heterologous genes that encode PDC1 [26] and ADH4 [27] from S. cerevisiae were introduced to R. toruloides Δku70, with an aim to create an aerobic ethanol biosynthetic pathway. Unfortunately, the two corresponding enzymes could not be solubly expressed in the engineered strain based on Western blot (data not shown). In the future, more ethanol biosynthetic pathway genes from different ethanologenic organisms need to be evaluated and introduced to our engineered R. toruloides strain to construct an endogenously ethanol-producing pathway. Recently, this proposed strategy was demonstrated to be feasible in the oleaginous yeast Y. lipolytica for the synthesis of FAEEs without the addition of ethanol, but the titer of produced FAEEs was quite low (0.3 mg/L) due to the limited supply of ethanol [28]. Thus, further efforts to increase the production of ethanol is required. If successful, it will greatly promote the industrial development of biodiesel synthesis in the oleaginous yeasts.

Conclusions
In this study, R. toruloides was engineered to produce FAEEs by fermentation for the rst time by heterologously expressing the WS from A. baylyi ADP1. The optimal ethanol concentration for FAEEs production was proved to be 5%, which led to a maximum FAEEs titer of 0.82 g/L. To further improve the FAEEs production, the AbWS enzyme was modi ed by site-directed mutagenesis to change its substrate preferences, which signi cantly increased the production titer to 1.02 g/L. Finally, by carrying out fedbatch fermentation in a 1-L fermenter, the engineered strain was able to produce FAEEs up to a titer of 9.2 g/L. This is to date the highest FAEEs production level in a yeast host. Furthermore, the FAEEs produced by the strain were mainly secreted outside the cell, which can greatly save the costs for FAEE extraction. Overall, R. toruloides has the potential to become an excellent platform organism for industrial production of biodiesels and other fatty acid derived green fuels or chemicals.

Strains and media
The parent strain in this study was R. toruloides Δku70, a derivative of R. toruloides IFO0880 with a deletion in the non-homologous end joining (NHEJ) gene, ku70 [29]. Agrobacterium tumefaciens AGL-1 strain was used for transformation experiments [18,30]. The E. coli DH5α was used for plasmid construction.
R. toruloides strains were routinely grown in liquid or solid Yeast Extract-Peptone-Dextrose (YPD) medium (10 g/L of yeast extract, 20 g/L of peptone, 20 g/L glucose) at 30 °C for strain construction and activation, and in fermentation medium (glucose 60.0 g/L, peptone 15.7 g/L, yeast extract 15.7 g/L) for both shake-ask batch fermentation and bioreactor fed-batch fermentation. E. coli strains were routine grown in Luria-Bertani broth (LB) medium at 37 °C. A. tumefaciens was grown at 28 °C in 2YT medium (16 g/L tryptone, 10 g/L yeast extract, 5 g/L NaCl). Kanamycin, ampicillin, hygromycin, cefotaxime and rifampicin were supplemented to the medium to a nal concentration of 50 µg/mL, 100 µg/mL, 100 µg/mL, 300 µg/mL, and 25 µg/mL, respectively. Strain information used in this study are listed in Table 1.

Plasmid Construction And Yeast Transformation
Standard genetic manipulations were performed as previously described [32]. pKOCAR2 plasmid was used as the gene expression vector for yeast transformation [29]. The WS genes were codon optimized and synthesized by Nanjing GeneScript Biotech Co., Ltd (Nanjing, China). They were ampli ed and ligated to the pKOCAR2 plasmid using the NEBuiler HiFi Assembly kit (New England Biolabs, USA), yielding plasmids p101, p102, p103, p104, and p105, respectively. The optimized gene sequences and primers used in this study were listed in Tables S1 and S2, respectively.
For enzyme engineering of AbWS, overlap-extension PCR was used for site-directed mutagenesis [33]. Two pairs of primer were designed to amplify the upstream and downstream DNA fragments, respectively. The modi ed nucleic base was contained in the primers. The mutated AbWS gene was ligated to the pKOCAR2 plasmid, yielding recombinant plasmid p1022. All plasmids used in this study were listed in Table 1.
The recombinant plasmids were transformed into R. toruloides by ATMT as described previously [34]. The pKOCAR2-derived plasmids were rstly electroporated into A. tumefaciens AGL-1. Then the A. tumefaciens strain harboring the corresponding plasmid was co-cultured with the R. toruloides Δku70, and the mixture was spread on YPD solid plate supplemented with hygromycin and cefotaxime for colony screening. Colonies appeared after 2 days.

Fermentation Conditions
Shake-ask batch fermentation was conducted in a 50-ml Erlenmeyer ask containing 10 ml of fermentation medium. The culture was inoculated with 10% seed culture, and performed in an orbital shaker at a rotary rate of 250 rpm at 30 °C. After culturing for 24 h, exogenous ethanol was added to the culture to a nal concentration of 1%, 3%, 5%, 7%, respectively. To facilitate the extraction process, 10% dodecane was added to the culture as reported previously [16]. Heptadecanoic acid ethyl ester was used as the internal standard.
Fed-batch fermentation was conducted in a 1-liter stirred bioreactor (Eppendorf, Germany). Culture pH was monitored in real-time by a pH meter (Mettler-Toledo, Switzerland), and dissolved oxygen was monitored in real-time by an oxygen probe (Mettler-Toledo, Switzerland). The initial volume was 500 mL. The cultivation conditions were the same with what was previously described [35]. The glucose and ethanol concentrations were determined by High-Performance Liquid Chromatography (HPLC). The glucose concentration was kept above 10 g/L by feeding 50 mL glucose with a concentration of 500 g/L. 50 mL 50% ethanol was pumped to the fermenter at a ow rate of 1 mL/min to a nal concentration of 5% after inoculation of 24 hours. 50 mL dodecane together with ethyl heptadecanoate (C17:0) internal standard were injected to the fermenter through a syringe.

Analytical Methods
The growth curves of the R. toruloides strains were determined by measuring the cell density at 600 nm using a Genesys 20 spectrophotometer (Thermo Fischer Scienti c Inc., USA). The dry cell weight (DCW) was determined by the previously described method [5]. The concentrations of glucose and ethanol were monitored by a LC-20A HPLC instrument (Shimadzu Inc., Japan) equipped with a RID detector as previously reported [16].

Total Lipids And Faees Extraction And Analysis
For extracellular FAEEs, the concentration was determined by directly injecting the dodecane layer to a SHIMADZU Japan GCMS-QP2010 PLUS mass spectrometer coupled with a SHIMADZU gas chromatograph (Shimadzu Inc., Japan). The heptadecanoic acid ethyl ester was added to the medium as the internal standard. The FAEEs dissolved in the dodecane layer were separated using a SHIMADZU 5MS capillary column (30 m × 0.25 mm I. D., 0.25 µm lm thickness) as previously described [36].
For intracellular FAEEs, the concentration was measured by the following procedure. First, the total lipids were extracted. Cells were harvested and washed twice with double-distilled water (ddH 2 O), and then freeze-dried until the samples were dry. Lipids were extracted from the lyophilized cells as previously reported [37]. Heptadecanoic acid ethyl ester was added as an internal standard. Second, lipids were separated and fractionated. FAEE components in the total lipid extracts were separated by thin layer chromatography using TLC Silica gel 60 F254 plates as previously reported [13] (Merck, Darmstadt, Germany). The FAEE components were then scraped from the TLC plate and extracted by an organic solvent containing 3 mL hexane, 2 mL methanol and 2 mL ddH 2 O and vigorously vortexed for 1 h at room temperature. After centrifuging at 3000 rpm for 5 min, the organic phase was transferred to a new glass tube and dried under nitrogen. The residues were dissolved in 200 µL methanol: chloroform (95: 5) for GC-MS analysis.
To determine the fatty acid composition, total lipids were transmethylated and analyzed according to a previously reported method [38].

Western Blot
Western blot The cells were sampled in the exponential growth period, and homogenized using a FastPrep instrument (MP biomedicals, Solon, OH, USA) for 6 cycles at a speed of 6.0 m/s for 30 s each, with 5 minutes of interval. The suspensions were centrifuged at 15, 000 g for 30 min, and the supernatants were collected [31]. Then the total proteins were detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) with 12% polyacrylamide gels, which were then immediately transferred to a polyvinylidene di uoride (PVDF) membrane (BIO-RAD, USA). A Flag tag was added in the 5' end of the AbWS gene. A monoclonal anti-FLAG M2 antibody was used for Western blot analysis as previously described [39], and the membrane was covered with electrochemiluminescence (ECL) plus solution for color reaction. Finally, the membrane was exposed to the Gel Imager System (Azure, USA) by the chemiluminescence mode.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
All authors agree to submit the work to the journal.

Availability of data and materials
All data generated or analyzed during this study are included in this published article and its additional les.