High‐yield lipid production from lignocellulosic biomass using engineered xylose‐utilizing Yarrowia lipolytica

Abstract

Lignocellulosic biomass shows high potential as a renewable feedstock for use in biodiesel production via microbial fermentation. Yarrowia lipolytica, an emerging oleaginous yeast, has been engineered to efficiently convert xylose, the second most abundant sugar in lignocellulosic biomass, into lipids for lignocellulosic biodiesel production. Yet, the lipid yield from xylose or lignocellulosic biomass remains far lower than that from glucose. Here we developed an efficient xylose‐utilizing Y. lipolytica strain, expressing an isomerase‐based pathway, to achieve high‐yield lipid production from lignocellulosic biomass. The newly developed xylose‐utilizing Y. lipolytica, YSXID, produced 12.01 g/L lipids with a maximum yield of 0.16 g/g, the highest ever reported, from lignocellulosic hydrolysates. Consequently, this study shows the potential of isomerase‐based xylose‐utilizing Y. lipolytica for economical and sustainable production of biodiesel and oleochemicals from lignocellulosic biomass.

INTRODUCTION

Lignocellulosic biomass is the most abundant resource for biofuel and biochemical production, with the advantage of being renewable. With recent advances in the cellulosic bioethanol industry, the lignocellulosic biorefinery concept has expanded to biodiesel production, wherein vegetable oils are replaced with microbial oils for increased sustainability and reduced negative environmental impact (Ko & Lee, 2018). Of the numerous hosts for microbial oil production, the non‐model yeast Yarrowia lipolytica has emerged as one of the most promising strains because of its superior lipid accumulation capacity and robustness in harsh environments such as low pH and high salt conditions (Cui et al., 2017; Gao et al., 2016). Moreover, the relative ease of metabolic engineering with a fully sequenced genome and available tools, such as finely tuned promoters and a CRISPR‐Cas9 system (Blazeck et al., 2013; Schwartz, Hussain, Blenner, & Wheeldon, 2016), makes this strain more attractive as a cell factory for the production of microbial oils used for biodiesel or oleochemicals. Reported engineering efforts to maximize the lipid accumulation capacity of Y. lipolytica have resulted in high titer (98.9 g/L), yield (0.27 g lipids/g glucose), and productivity (1.3 g L⁻¹ hr⁻¹) during lipid production from glucose (Qiao, Wasylenko, Zhou, Xu, & Stephanopoulos, 2017).

Unlike the superior performance on glucose, the conversion of lignocellulosic biomass to lipids by Y. lipolytica is poor. Though the utilization of xylose, the second most abundant sugar in lignocellulosic biomass, has been reported in the presence of glucose, the lipid yield from lignocellulosic hydrolysates was limited to approximately 30% of the typical values obtained with glucose (0.06 g/g from lignocellulosic biomass vs. 0.2 g/g from glucose; Qiao et al., 2017; Quarterman, Slininger, Kurtzman, Thompson, & Dien, 2017; Slininger et al., 2016). Owing to its cryptic and conditional xylose catabolism, Y. lipolytica cannot utilize xylose as the sole carbon source (Zhao et al., 2015); the inefficiency in xylose catabolism hinders the use of Y. lipolytica in lipid production.

Therefore, to improve lignocellulosic lipid production, the xylose catabolic pathway was engineered in Y. lipolytica. To supplement its cryptic native xylose metabolism (Rodriguez et al., 2016), a heterologous xylose‐utilizing pathway based on oxidoreductase, xylose reductase (XR), and xylitol dehydrogenase (XDH), has been overexpressed in Y. lipolytica enabling it to produce lipids from xylose as the sole carbon source (Ledesma‐Amaro et al., 2016; Li & Alper, 2016). However, the lipid yields from xylose are much lower than those obtained from glucose owing to an innate cofactor imbalance in the oxidoreductase‐based xylose‐utilizing pathway introduced into Y. lipolytica. Cofactor imbalance in oxidoreductase‐based xylose‐utilizing strains has been previously observed in bioethanol‐producing Saccharomyces cerevisiae, which severely lowered ethanol yield during lignocellulosic fermentation. Given that lipid production pathways involve more cofactors, such as NADPH, which provide the necessary reducing power, cofactor imbalance would cause a more severe problem in xylose‐utilizing Y. lipolytica resulting in limited lipid yields from xylose. An alternative xylose‐utilizing pathway based on a redox‐neutral isomerase would have an advantage compared with the oxidoreductase‐based pathway (Figure 1). However, to the best of our knowledge, an isomerase‐based xylose‐utilizing Y. lipolytica strain has yet to be successfully developed.

1 FIGURE. A schematic illustration of metabolic pathways for the bioconversion of glucose and xylose into lipids in Yarrowia lipolytica: XDH, xylitol dehydrogenase; XK, xylulose kinase; XR, xylose reductase. Dashed lines represent multiple metabolic steps; red line represents a heterologous pathway

In this study, we developed an isomerase‐based xylose‐utilizing strain of Y. lipolytica for high‐yield lipid production from lignocellulosic biomass. To this end, we constructed an isomerase‐based xylose utilization pathway by introducing a mutated xylose isomerase gene, xylA3*, expressed in one of the best xylose‐utilizing S. cerevisiae strains with improved activity (Lee, Jellison, & Alper, 2012, 2014), and overexpressing an endogenous xylulokinase (XK), a proven overexpression target for improved xylose metabolism (Ledesma‐Amaro et al., 2016; Li & Alper, 2016). To better support the newly introduced xylose utilization pathway, adaptive evolution was then conducted to generate changes in cellular metabolism favorable for xylose metabolism even in the presence of xylose as the single carbon source. Xylose utilization was then reinforced by pulling metabolic flux directly toward lipid production by overexpressing the diacylglycerol acyltransferase gene (DGA1) and blocking lipid degradation by deleting peroxisome biogenesis factor 10 (PEX10). These engineering efforts led, for the first time, to high‐yield lipid production from xylose using an isomerase‐based xylose‐utilizing Y. lipolytica strain. Finally, high‐yield lipid production was achieved, with the highest yield reported to date from lignocellulosic lipid production. This study demonstrates the potential of xylose‐utilizing Y. lipolytica harboring an isomerase‐based pathway, which could serve as a promising host for economically feasible microbial oil production from renewable lignocellulosic biomass.

MATERIALS AND METHODS

Strains and culture conditions

The Y. lipolytica PO1f strains (ATCC MYA‐2613) used in this study are summarized in Table 1. Yeast strains were routinely cultured in yeast synthetic complete (YSC) medium containing 6.7 g/L yeast nitrogen base (YNB), 20 g/L carbon source (glucose or xylose), and complete supplement mixture (CSM) or CSM‐Leu‐Ura (MP Biomedicals). To evaluate xylose utilization, Y. lipolytica grown for 1.5 days was inoculated into 250 ml flasks containing 50 ml of the respective medium at an initial OD₆₀₀ of 0.1 or 0.2 and grown at 28°C with constant shaking at 200 rpm. Escherichia coli DH10β (New England BioLabs) was used for cloning and plasmid propagation. E. coli DH10β cells were grown at 37°C in lysogeny broth (Sitepu, Selby, Lin, Zhu, & Boundy‐Mills) supplemented with 100 µg/ml of ampicillin (Sigma Aldrich) with constant shaking at 225 rpm.

1 TABLEComparison of xylose fermentation performance of YSXI and YSXID with those of previously reported recombinant xylose‐utilizing strains

Strain	Description	Initial fermentation condition	Total sugar consumption (g/L)	Lipid yield (g lipid/g sugar)	Lipid titer (g/L)	Lipid content (%)	Reference
YSXI	YSX Δpex10 XylA, XK	Batch, C/N: 100, X: 160	X: 53.67	0.074	3.96	30.16	This study
PO1f_XDH_XR	PO1f XDH XK	Batch, C/N: 8.8, X: 20	X: 20	0.015	0.3	10	Rodriguez et al. (2016)
XYL+	PO1d ssXR ssXDH ylXK	Fed‐Batch, C/N: 45, X: 150	X: 250	0.02	5.9	12.7	Ledesma‐Amaro et al. (2016)
XYL + Obese	PO1d Δpox1‐6 Δtgl4 GPD1 DGA2 ssXR ssXDH ylXK	Fed‐Batch, C/N: 45, X: 150	X: 250	0.08	20.1	35	Ledesma‐Amaro et al. (2016)
E26 XUS	E26 XR XDH (Evolved strain)	Batch, C/N: 35, X: 160	X: 160	0.094	15.06	N/A	Li and Alper (2016)
YSXI	YSX Δpex10 XylA, XK	Batch, C/N: 100, G: 80, X: 80	G: 63.8, X: 27.8	0.07	7.3	51.6	This study
YSXID	YSX Δpex10::DGA1 XylA, XK	Batch, C/N: 100, G: 80, X: 80	G: 80, X: 34.28	0.12	13.5	56.7	This study
XYL+	PO1d ssXR ssXDH ylXK	Fed‐Batch, C/N: 45, X: 150	G: 125, X: 250	0.02	7.8	13.62	Ledesma‐Amaro et al. (2016)
XYL + Obese	PO1d Δpox1‐6 Δtgl4 GPD1 DGA2 ssXR ssXDH ylXK	Fed‐Batch, C/N: 45, X: 150	G: 125, X: 250	0.06	22.5	30.57	Ledesma‐Amaro et al. (2016)

Abbreviations: C/N, carbon‐to‐nitrogen molar ratio; G, glucose; N/A, not applicable; X, xylose.

Construction of xylose‐utilizing strain

To construct the xylose‐utilizing strain, a codon‐optimized xylose isomerase mutant gene (xylA3*; Lee, Jellison, & Alper, 2014) and an endogenous XK gene (ylXK, YALI0F10923) were expressed in Y. lipolytica. Specifically, a codon‐optimized xylA3* for Homo sapiens was synthesized by GeneScript (NJ, USA) and cloned into pMCS‐UAS1B16‐TEF plasmid to form pMCS‐UAS1B16‐TEF‐xylA‐CYCt. For expression of the XK gene, ylXK was PCR amplified with primers 001‐f and 002‐r (Table S1) and cloned into the pMCS‐UAS12‐TEF plasmid, with URA2 as the selection marker, generating the pMCS‐UAS12‐TEF‐ylXK‐CYCt plasmid. To support metabolic flux through the lipid production pathway, a copy of an endogenous diacylglycerol acyltransferase gene (DGA1, YALI0E32769g) was integrated into the peroxisomal biogenesis factor 10 (PEX10) locus using the CRISPR‐Cas9 system reported by Schwartz et al. (2016). Briefly, the expression cassette of the endogenous DGA1 under control of the TEFint promoter (Tai & Stephanopoulos, 2013) cloned into the pBHA‐TEFint plasmid was co‐transformed into Y. lipolytica with a plasmid expressing Cas9 and an sgRNA targeting PEX10 to generate the DGA1‐overexpressing and PEX10 deleted strain (YSX‐D).

Evolutionary engineering of xylose‐utilizing strain

In order to obtain efficient xylose‐utilizing Y. lipolytica, Y. lipolytica PO1f expressing the xylose isomerase gene was subjected to evolutionary engineering by subculturing the cells in 20 ml glass tubes containing 10 ml of YSC medium supplemented with 20 g/L xylose as the sole carbon source. The cells were transferred to fresh medium at the exponential growth phase using 1% inoculum. After five rounds of subculturing, the cells were plated on YSC medium with 20 g/L xylose and size‐based colony selection was conducted. The cell growth of 60 colonies isolated on xylose was initially analyzed using a TECAN instrument (Infinite 200 Pro; Tecan Group Ltd.) and then further confirmed by flask culture test using 50 ml of YSC medium with 20 g/L xylose in 100 ml culture flasks. Finally, the best xylose‐utilizing strain, YSX, was selected based on its cell growth and xylose utilization. To identify the beneficial mutations generated in strain YSX during adaptive evolution, whole‐genome sequencing was conducted by Macrogen Inc. using an Illumina HiSeq 2500 platform.

Bioreactor fermentation

To evaluate the performance of the xylose‐utilizing Y. lipolytica strains, batch fermentation was conducted in a 3 L bioreactor with a working volume of 1 L. For the inoculum, precultured cells grown in 3 ml of YSC medium containing 0.69 g/L CSM without leucine or uracil were transferred into 100 ml of the same medium and incubated for 1.5 days until the cells reached the exponential phase. The cells were then washed and inoculated into a bioreactor at an initial OD₆₀₀ of 0.2–0.5. The detailed composition of medium used for batch reactor fermentation was as follows: In the pure sugar fermentation, YSC medium containing 160 g/L xylose or 80 g/L glucose and xylose as the carbon source, 0.69 g/L CSM‐Leu‐Ura, 1.76 g/L YNB w/o amino acids and ammonium sulfate, and 3.52 g/L ammonium sulfate to achieve a C/N ratio of 100. In the lignocellulosic biomass fermentation, YSC medium containing lignocellulosic hydrolysate composed of 35 g/L glucose and xylose, 0.69 g/L CSM‐Leu‐Ura, 1.76 g/L YNB w/o amino acids and ammonium sulfate, and 3.3 g/L ammonium sulfate to achieve a C/N ratio of 46. The lignocellulosic hydrolysate derived from Miscanthus sacchariflorus Goedae Uksae 1 (Korea), pretreated by a H₂SO₄‐catalyzed hydrothermal process, was purchased from Sugaren (Yongin‐si, South Korea). The glucose and xylose‐rich hydrolysates from solid and liquid fractions of pretreated biomass, respectively, were prepared and concentrated separately, so lignocellulosic fermentation could be conducted with sufficient concentration of xylose. During lignocellulosic fermentation, the pH was maintained above 3.5 with 2.5 M NaOH and the temperature was strictly controlled at 28°C. The dissolved oxygen (DO) was maintained above 50% saturation by varying the agitation from 250 to 800 rpm.

Lipid quantification

Total lipid quantification was carried out as previously described (Bligh & Dyer, 1959; Bourque & Titorenko, 2009). Briefly, 3 ml of culture broth containing yeast cells were centrifuged and the supernatant was discarded. The remaining cells were vigorously mixed with an organic solvent mixture of chloroform and methanol (2:1 v/v) and heptadecanoic acid (C17:0, Sigma Aldrich) was added as an internal standard for further FAME (Fatty acid methyl ester) quantification. The solvent mixture containing the lipid extracts was evaporated using a nitrogen evaporation system (Caliper Life Sciences, USA) and the weight of the remaining lipids was gravimetrically measured to estimate lipid content (g lipid/g cell biomass). Solvent‐evaporated lipids were converted into their methyl esters using 2.5% sulfuric acid in methanol reagent and used for gas chromatography analysis (GC, 6890N, Agilent; equipped with a flame ionization detector and an INNOWAX capillary column [30 m × 0.32 mm × 0.5 µm, Agilent]). FAMEs were identified by comparing the peak areas and retention times of the final samples with those of the standard FAME mixture containing C₈‐C₂₄ methyl esters and heptadecanoic acid (C17:0) was used as an internal standard (Sigma‐Aldrich).

Analytical method

Yeast cell growth was estimated by measuring the optical density of the culture broth using a spectrophotometer (Shimadzu UV‐1240) and dry cell weight (DCW). For DCW, 1 ml of culture was washed three times with distilled water and dried in a 50°C oven until no difference in weight was observed. The DCW of cells with high lipid content was measured by filtering 5 ml of the cell culture through a 0.45 µm nitrocellulose filter paper (Sartorius). The xylose concentration was quantified using a high‐performance liquid chromatography system (HPLC, Agilent Technology 1100 series) equipped with UV and refractive index detectors with an Aminex HPX‐87H column (Bio‐Rad Inc.); 5 mM H₂SO₄ was used as the mobile phase with a flow rate of 0.6 ml/min and the column temperature was maintained at 50°C. Each sample was passed through a 0.22 µm syringe filter (Whatman) prior to analysis.

RESULTS

Introduction of the xylose isomerase pathway into Y. lipolytica strain PO1f

To develop an efficient xylose‐utilizing strain based on the isomerase pathway, a codon‐optimized xylose isomerase gene mutant, previously introduced to support efficient xylose metabolism in the model‐yeast S. cerevisiae, was introduced into wild‐type Y. lipolytica PO1f. Given the positive effect of XK on xylose metabolism (Niehus, Crutz‐Le Coq, Sandoval, Nicaud, & Ledesma‐Amaro, 2018), an endogenous XK gene, ylXK, was also overexpressed for synergetic improvement of cell growth on xylose. Although the xylose isomerase expressing strain showed slightly higher cell growth on xylose (Figure 2), the introduction of a heterologous xylose isomerase pathway did not fully support xylose metabolism in Y. lipolytica, as previously reported (Li & Alper, 2016).

2 FIGURE. Cell growth of Yarrowia lipolytica PO1f expressing xylose isomerase pathways on xylose as a sole carbon source: Flask culture test was conducted in complete supplement mixture medium with 20 g/L of xylose as a sole carbon source. The cell growth of Y. lipolytica PO1f expressing xylA3* (●), ylXK (□), and xylA3* and ylXK (■), were compared to that of the strain expressing empty plasmids (○). Error bars represents standard deviation of biological triplicates

Adaptive laboratory evolution of Y. lipolytica to support xylose metabolism based on the isomerase pathway

As substantial cell growth was not observed in strains expressing the isomerase‐based pathway using xylose as the sole carbon source, adaptive laboratory evolution was applied to generate changes in the genomic DNA to better support xylose metabolism by xylose isomerase. To this end, the wild‐type strain expressing the xylose isomerase gene (xylA3*) was serially transferred into fresh xylose medium every 2 days with an inoculum size of 1%. After five rounds of subculturing, the 60 largest colonies were selected and their cell growth on xylose was evaluated using the TECAN Infinite Pro 200 system. The selected isolates were then further confirmed by culturing in 20 ml culture tubes and the fastest growing strain, YSX, was selected and used for further engineering. As shown in Figure 3, the evolved strain, YSX, exhibited sufficient cell growth on xylose as the sole carbon source compared to the limited cell growth observed for the wild‐type strain.

3 FIGURE. Cell growth of evolved strain of YSX on xylose as a sole carbon source: The cell growth of YSX (●) on xylose was compared to wild‐type strain of Yarrowia lipolytica PO1f (○) during flask culture in complete supplement mixture medium supplemented with 20 g/L of xylose as a sole carbon source. Error bars represents standard deviation of biological triplicates

Though unable to grow on xylose as the sole carbon source, Y. lipolytica possess putative xylose catabolic pathway genes, xylulose reductase (ylXR), xylitol dehydrogenase (ylXDH), and XK (ylXK; Ledesma‐Amaro et al., 2016). To confirm the possible upregulation of endogenous ylXR, ylXDH, and/or ylXK through the evolutionary process, quantitative real‐time PCR was conducted. However, no significant difference in the expression levels of the ylXR, ylXDH, and ylXK genes was observed between the wild‐type and YSX strains (data not shown). This suggests that the improved xylose growth phenotype was possibly caused by genomic changes indirectly associated with the endogenous initial xylose catabolic pathway. Thus, to investigate the changes in the genomic DNA generated during adaptive laboratory evolution, whole genome sequencing was performed and 59 mutations throughout the whole genome were identified (Table S1). However, further validation of the beneficial mutation was not successfully conducted because of the limited genetic information available for the non‐model yeast Y. lipolytica.

Introduction of the xylose isomerase pathway in evolved Y. lipolytica strain YSX

Encouraged by the improved cell growth of strain YSX, the xylose isomerase (xylA3*) and XK (ylXK) genes were co‐expressed in YSX with a rewired cellular network, which would facilitate the successful expression of the isomerase‐based xylose catabolic pathway in Y. lipolytica. When xylA3* and ylXK were expressed, the YSX strain (YSX‐AK) showed significantly increased cell growth on xylose as the sole carbon source (Figure 4). Without overexpression of ylXK (YSX and YSX‐A), only marginal cell growth was observed until 300 hr regardless of the presence of xylA3*. In contrast, when XK was overexpressed (YSX‐K and YSX‐AK) a drastic improvement in cell growth on xylose was observed resulting in 10 and 13.7‐fold increase in the final OD and 3.8 and 5.1‐fold increase in xylose consumption when compared to YSX and YSX‐A, respectively. It should be noted that the highest cell growth and xylose consumption was observed in YSX‐AK, co‐expressing xylA3* and ylXK, clearly demonstrating that the isomerase‐based xylose catabolic pathway is functionally working in Y. lipolytica. The final OD and xylose consumption of YSX‐AK was 15.4 and 6.8‐fold higher than those of YSX.

4 FIGURE. Cell growth and xylose utilization of YSX expressing xylose isomerase and/or xylulokinase: (a) cell growth (OD600) and (b) xylose consumption of YSX strains expressing xylA3* (■), ylXK (□), xylA3*, and ylXK (●) were compared to that of YSX expressing empty plasmids (○) during aerobic fermentation with xylose as a sole carbon source. Error bars represents standard deviation of biological triplicates

Lipid production of YSXI during batch fermentation with xylose as the sole carbon source

Xylose‐utilizing Y. lipolytica with the isomerase‐based pathway is expected to produce higher yield of lipids than a strain with the oxidoreductase‐based pathway because of its cofactor neutral nature. To evaluate the lipid production capacity of YSX‐AK, batch fermentation with a high xylose concentration was conducted in a bioreactor, which provides controlled pH and dissolved oxygen. To prevent peroxisomal lipid degradation via β‐oxidation in YSX‐AK, PEX10 was deleted generating the YSXI strain, which was then used for subsequent experiments. During batch fermentation, YSXI utilized 54 g/L of xylose and produced 3.97 g/L of lipids with a yield of 0.074 g lipids/g xylose (Figure 5; Table 1). The measured lipid content of YSXI was 30%. In contrast, no cell growth was observed in the wild‐type strain for nearly 200 hr.

5 FIGURE. Lipid production of YSXI during batch fermentation of xylose in a 3 L bioreactor: (a) cell growth measured by OD600 (○ and ●) and xylose consumption (□ and ■) of wild‐type strain (open symbols) and YSXI (closed symbols), (b) lipid titer, and (c) lipid contents of YSXI during batch fermentation of xylose were compared to that of a wild‐type strain (WT). The fermentation was conducted in a 3 L bioreactor containing 1 L of complete supplement mixture media with 160 g/L of xylose as a sole carbon source. Initial OD600 was 0.2 and the reactor was operated at 28°C with controlled pH above 3.5

To evaluate the co‐fermentation performance of YSXI, batch fermentation was conducted in a 3 L bioreactor using glucose and xylose as co‐substrates. During 173 hr of co‐fermentation, YSXI utilized 63.8 g/L of glucose and 27.8 g/L of xylose and produced 7.3 g/L of lipids. The lipid content of YSXI at the end of the fermentation was 51.6% (Figure 6). The addition of glucose increased the lipid content of YSXI by 1.71‐fold and promoted rapid cell growth, reducing the lag phase compared to xylose fermentation. Although simultaneous utilization of glucose and xylose was observed, the glucose consumption rate was higher than that of xylose, requiring further improvement of xylose utilization efficiency in YSXI.

6 FIGURE. Lipid production of YSXI and YSXID during glucose and xylose co‐fermentation in a 3 L bioreactor: (a) glucose (▲) and xylose (●) consumptions and cell growth (■) of YSXI during co‐fermentation. (b) Lipid titer of YSXI and YSXID during co‐fermentation. (c) Glucose (▲) and xylose (●) consumptions and cell growth (■) of YSXID during co‐fermentation. (d) Lipid contents of YSXI and YSXID during co‐fermentation. The fermentation was conducted in a 3 L bioreactor containing 1 L of complete supplement mixture media with 80 g/L of glucose and 80 g/L of xylose. Initial OD600 was 0.5 and the reactor was operated at 28°C with controlled pH above 3.5

To improve xylose utilization by pulling carbon flux toward lipid production, DGA1, encoding diacylglycerol acyltransferase (YALI0E32769g), was overexpressed in YSXI generating strain YSXID. When the downstream metabolic flux was fortified by DGA1 overexpression, the xylose consumption was accelerated when glucose was depleted resulting in a 57% increase in maximum xylose consumption rate as compared to YSXI (0.53 g L⁻¹ hr⁻¹ for YSXID vs. 0.34 g L⁻¹ hr⁻¹ for YSXI). With increased carbon utilization and an enhanced lipid production pathway, YSXID accumulated lipids with a high titer and content compared to YSXI (13.5 g/L and 56.7% for YSXID vs. 7.3 g/L and 51.6% for YSXI, respectively; Figure 6b).

Lignocellulosic lipid production using xylose‐utilizing Y. lipolytica YSXID

Finally, the lipid production performance of YSXID was evaluated using miscanthus hydrolysates as substrates for lignocellulosic biodiesel production. To maximize lipid production, fed‐batch fermentation was conducted, separating the growth and production phases of YSXID during co‐fermentation.

During the first stage of fed‐batch fermentation, YSXID yielded 14.9 g/L cell biomass in 123 hr, while consuming 35.2 g/L glucose and 23.5 g/L xylose derived from miscanthus hydrolysates. After supplementing glucose and xylose to a final concentration of 35 g/L, YSXID more efficiently accumulated lipids, reaching a final lipid titer of 12.01 g/L, with an overall yield of 0.11 g/g (Figure 7; Table 2). During the second stage of fed‐batch fermentation, YSXID converted 32.7 g/L glucose and 25.7 g/L xylose into 9.35 g/L lipids, with a yield of 0.16 g/g, demonstrating successful high‐yield lipid production from lignocellulosic biomass.

7 FIGURE. Lipid production of YSXID during fed‐batch fermentation using lignocellulosic hydrolysates in a 3 L bioreactor: (a) glucose (▲) and xylose (●) consumptions and cell growth (■) of YSXID during co‐fermentation in lignocellulose hydrolysate. (b) Fatty acid profile of lipids produced by YSXID using lignocellulosic hydrolysates. The fermentation was conducted in a 3 L bioreactor containing 1 L of complete supplement mixture media with lignocellulose hydrolysate composed of 35 g/L of glucose and 35 g/L of xylose initially. Initial OD600 was 0.5 and the reactor was operated at 28°C with controlled pH above 3.5

2 TABLEComparison of glucose and xylose co‐fermentation performance of YSXID with previous reports during lignocellulosic lipid production

Strain	Biomass	Medium composition (g/L)	Inhibitor concentration (g/L)	Total sugar consumption (g/L)	Lipid yield (g lipid/g sugar)	Lipid titer (g/L)	Lipid contents (%)	Reference
YSXID	Miscanthus	G: 35.2 + 32.7 X: 32.8 + 22.8	HMF: 0.178 Furfural: 0.212	G: 67.92 X: 49.22	0.11	12.01	42.4	This study
W29	Switchgrass	G: 36.4 X: 28.3	HMF: 0.32a Furfural: 0.85a	G: 36.4 X: 24	0.06	2.2	15.0	Quarterman et al. (2017)
YB‐392	Corn stover	G: 59.3 X: 36.3	HMF: 0.6 Furfural: 0.09	G: 59.3 X: 35.9	0.06a	5.8	N/A	Slininger et al. (2016)
YB‐437	Corn stover	G: 59.3 X: 36.3	HMF: 0.6 Furfural: 0.09	G: 59.3 X: 35.7	0.06	5.8	N/A	Slininger et al. (2016)

Abbreviations: G, glucose; H, hydroxymethylfurfural; N/A, not applicable; X, xylose.

^aCalculated based on paper.

The composition of the lipids accumulated in YSXID during lignocellulosic fermentation was similar to that seen in previous reports (Ledesma‐Amaro et al., 2016; Quarterman et al., 2017). The fatty acids, analyzed as FAMEs, consisted predominantly of oleic acid (C18:1) and palmitic acid (C16:0), followed by stearic acid (C18:0), palmitoleic acid (C16:1), and linoleic acid (C18:2; Figure 7b). The lipid composition determines the fuel quality of a biodiesel (Meng et al., 2009). The unaltered lipid composition obtained with the additional sugar xylose therefore guarantees the quality of the biodiesel produced from lignocellulosic biomass.

DISCUSSION

Efficient xylose utilization offers complete conversion of all available sugars in lignocellulosic biomass. Here we developed an efficient xylose‐utilizing Y. lipolytica strain harboring isomerase‐based xylose utilization pathway for successful conversion of both glucose and xylose in lignocellulosic hydrolysates. Through pathway engineering and adaptive laboratory evolution, the newly developed strain converted lignocellulosic hydrolysates into lipids with the highest yield ever reported.

To improve the limited lipid yields during lignocellulosic biodiesel production, we developed cofactor neutral isomerase‐based xylose‐utilizing Y. lipolytica, so NADPH can be used mainly for lipid production. To accumulate a high content of lipids in the form of triacylglycerol (TAG), the major lipid storage compound, Y. lipolytica requires high NADPH redox power. The oxidation of two moles of NADPH to NADP⁺ is required to elongate two carbons in fatty acids. Previously developed xylose‐utilizing Y. lipolytica utilize xylose through oxidoreductase‐based pathway (Niehus et al., 2018), which required one mole of NADPH for the conversion of xylose into xylitol causing a drain on NADPH pool inside the cell (Ko & Lee, 2018). In this study, we clearly showed the potential of isomerase‐based Y. lipolytica for high‐yield lipid production from xylose. As shown in Table 1, the lipid yield in strain YSXI containing the isomerase‐based pathway, in which the lipid production pathway was not engineered, was comparable to that of previously reported high‐lipid producing strains with an oxidoreductase‐based pathway (Ledesma‐Amaro et al., 2016; Li & Alper, 2016). When the oxidoreductase‐based xylose catabolic pathway was overexpressed in the wild‐type Y. lipolytica PO1f strain, the lipid yield was limited to 0.015 g lipid/g xylose (Rodriguez et al., 2016), approximately 20% of that of YSXI. Even with the potential of high‐yield lipid production, isomerase‐based xylose‐utilizing Y. lipolytica has never been reported. To the best of our knowledge, we report the functional expression of the xylose isomerase pathway in Y. lipolytica supported by a proper cellular network of YSX strain generated through adaptive laboratory evolution.

Owing to its cofactor‐neutral nature of the isomerase‐based xylose‐utilizing pathway, YSXID produced lipids with a high titer of 12.01 g/L and the highest yield of 0.11 g/g (g lipids/g sugars) reported to date for lignocellulosic lipid fermentation using hydrolysates (Table 1). Previously, lipid production by Y. lipolytica from lignocellulosic biomass has been demonstrated using switchgrass (Quarterman et al., 2017) and corn stover (Slininger et al., 2016). Though both studies observed significant xylose utilization during lignocellulosic fermentation, the lipid yields were limited to 0.06 g/g (Quarterman et al., 2017; Slininger et al., 2016), indicating inefficient bioconversion of lignocellulosic biomass‐derived sugars into lipids. Moreover, it should be noted that the lipid yield of YSXID (0.16 g/g) at the second stage of lignocellulosic fermentation was close to the values (0.2 g/g) typically observed during glucose fermentation by engineered Y. lipolytica strains (Blazeck et al., 2014; Qiao et al., 2017). In fed‐batch bioreactor, cells tend to accumulate lipids in the second stage of fed‐batch fermentation, where the C/N ratio is increased due to nitrogen sources being depleted in the first stage and extra carbon sources being added in the second stage (Beopoulos et al., 2009). Nitrogen limitation is known to increase lipid accumulation by redirecting the metabolic flux through ATP citrate lyase, thus supplying more acetyl‐CoA for lipid production (Zhang, Wu, Wu, Dai, & Song, 2016).

Overexpression of DGA genes in Y. lipolytica has been previously shown to improve lipid production, carbon utilization, and cell growth during glucose fermentation (Lazar, Dulermo, Neuvéglise, Crutz‐Le Coq, & Nicaud, 2014; Qiao et al., 2015; Sagnak, Cochot, Molina‐Jouve, Nicaud, & Guillouet, 2018; Tai & Stephanopoulos, 2013; Zhang, Damude, & Yadav, 2012). Unexpectedly, DGA1 overexpression inhibited xylose consumption when glucose was presented in the growth medium possibly due to extensive requirement of NADPH for lipid production. According to Wasylenko, Ahn, and Stephanopoulos (2015), the oxidative pentose phosphate pathway is the primary source of NADPH during lipogenesis in Y. lipolytica. The theoretical yield of NADPH generated from glucose through the oxidative pentose phosphate pathway is 20% higher than that generated from xylose (Qiao et al., 2017). Because of the high demand for NADPH due to DGA overexpression, YSXID seemed to preferably utilize glucose over xylose, possibly because more NADPH can be supplied from glucose than from xylose. The sequential utilization of glucose and xylose is a well‐known bottle‐neck in oxidoreductase‐based xylose‐utilizing yeast strains, including Y. lipolytica, with cofactor imbalance (Ko & Lee, 2018; Ryu, Hipp, & Trinh, 2016). Therefore, supplying sufficient NADPH during co‐fermentation would be required to achieve simultaneous conversion of glucose and xylose into lipids with a high titer and yield.The overall lipid yield and content obtained during lignocellulosic lipid production (0.11 g/g and 42.4%) were a little lower than those from co‐fermentation of glucose and xylose (0.12 g/g and 56.7%; Tables 1 and 2). This slight decrease in lipid production could possibly be due to the presence of inhibitory compounds derived from pretreated lignocellulosic hydrolysates, such as furfural and 5‐hydroxymethylfurfural (HMF). Furan aldehydes such as furfural and HMF are well‐known inhibitory compounds in lignocellulosic fermentation (Allen et al., 2010; Ask, Bettiga, Mapelli, & Olsson, 2013). Though the toxic levels of individual inhibitory compounds have not yet been evaluated for the growth and lipid production of Y. lipolytica, a toxic effect of 0.5 g/L furfural has been reported previously (Sitepu, Selby, Lin, Zhu, & Boundy‐Mills, 2014). The miscanthus hydrolysates used in this study contained 0.05 g/L furfural, which could increase to 0.212 g/L during fed‐batch fermentation, thus possibly affecting the lignocellulosic lipid production of YSXID. YSXID showed reduced cell growth in the presence of 0.1 g/L furfural, and almost no cell growth with 0.25 g/L furfural (Figure S2). Therefore, improving furfural tolerance could further enhance the lignocellulosic lipid production of YSXID.

CONCLUSIONS

In this study, we developed a cofactor neutral isomerase‐based xylose‐utilizing Y. lipolytica strain enabling high‐yield lipid production from xylose as the sole carbon source. Once the lipid biosynthesis pathway was improved, YSXID efficiently converted glucose and xylose derived from lignocellulosic hydrolysates into lipids, resulting in the highest yield ever reported. Although repressed xylose utilization was observed during co‐fermentation, further engineering to increase the NADPH supply and tolerance toward inhibitory compounds could significantly improve lipid production. Consequently, this study shows the potential of isomerase‐based xylose‐utilizing Y. lipolytica as an important host for microbial oil production from lignocellulosic biomass.

ACKNOWLEDGEMENTS

This research was supported by the Renewable Energy Core Technology Program of the Korea Institute of Energy Technology Evaluation and Planning (KETEP), a granted financial resource from the Ministry of Trade, Industry & Energy, Republic of Korea (No. 20153030091360). The authors also appreciate additional support by Korea Institute of Science and Technology (KIST) Institutional Program (2E30170).