Co-fermentation of xylose and cellobiose by an engineered Saccharomyces cerevisiae
- First Online:
- Cite this article as:
- Aeling, K.A., Salmon, K.A., Laplaza, J.M. et al. J Ind Microbiol Biotechnol (2012) 39: 1597. doi:10.1007/s10295-012-1169-y
We have integrated and coordinately expressed in Saccharomyces cerevisiae a xylose isomerase and cellobiose phosphorylase from Ruminococcus flavefaciens that enables fermentation of glucose, xylose, and cellobiose under completely anaerobic conditions. The native xylose isomerase was active in cell-free extracts from yeast transformants containing a single integrated copy of the gene. We improved the activity of the enzyme and its affinity for xylose by modifications to the 5′-end of the gene, site-directed mutagenesis, and codon optimization. The improved enzyme, designated RfCO*, demonstrated a 4.8-fold increase in activity compared to the native xylose isomerase, with a Km for xylose of 66.7 mM and a specific activity of 1.41 μmol/min/mg. In comparison, the native xylose isomerase was found to have a Km for xylose of 117.1 mM and a specific activity of 0.29 μmol/min/mg. The coordinate over-expression of RfCO* along with cellobiose phosphorylase, cellobiose transporters, the endogenous genes GAL2 and XKS1, and disruption of the native PHO13 and GRE3 genes allowed the fermentation of glucose, xylose, and cellobiose under completely anaerobic conditions. Interestingly, this strain was unable to utilize xylose or cellobiose as a sole carbon source for growth under anaerobic conditions, thus minimizing yield loss to biomass formation and maximizing ethanol yield during their fermentation.
KeywordsSaccharomyces cerevisiaeXylose isomeraseCellobiose phosphorylaseYeast fermentationCellulosic ethanol
Cellulosic ethanol is a promising alternative and renewable energy source that may significantly reduce greenhouse gas emissions and our reliance on oil. While ethanol production from the hexose sugars in sugarcane and corn starch by Saccharomyces cerevisiae are well-established processes, economical conversion of cellulosic feedstocks will also require rapid and efficient fermentation of xylose, the most abundant pentose sugar comprising up to 50 % weight of these feedstocks . In addition, cellobiose is a β(1,4)-disaccharide of glucose that can accumulate during the enzymatic hydrolysis of cellulose and its subsequent hydrolysis by β-glucosidase liberates glucose that could potentially lead to catabolite repression of xylose fermentation . Unfortunately, S. cerevisiae is unable to naturally ferment either xylose or cellobiose, and robust anaerobic fermentation of these sugars while maintaining the desirable properties of conventional yeast fermentation is essential for commercializing the next-generation cellulosic ethanol process.
There have been two main approaches to engineer xylose fermentation into yeast. The first has involved expression of a heterologous xylose assimilation pathway based on the sequential activities of NADPH-dependent xylose reductase, which catalyzes the conversion of xylose to xylitol, and a NAD+-dependent xylitol dehydrogenase, which catalyzes the conversion of xylitol to xylulose . Xylulose is then phosphorylated by the native xylulokinase to xylulose-5-phosphate, which enters glycolysis through the pentose phosphate pathway. Yeast engineered with this pathway suffer a cofactor imbalance that results in the accumulation of xylitol, reduced ethanol yield, and a micro-aerobic oxygen requirement that could be difficult to maintain at a commercial scale. A second approach has involved expression of a cofactor-neutral pathway based on the introduction of xylose isomerase to catalyze the direct conversion of xylose to xylulose. While it may seem an obvious choice to engineer into yeast a single enzyme that avoids cofactor imbalance, it took enormous effort to identify xylose isomerases that are active in yeast [4, 15, 22, 30, 41] due to improper folding and insolubility of the enzyme in the yeast cytoplasm .
There have also been two approaches to engineer cellobiose utilization into yeast. The first has involved expression of heterologous genes encoding a cellobiose transporter and β-glucosidase catalyzing its hydrolysis to two molecules of glucose, which are subsequently phosphorylated to glucose-6-phosphate by the expenditure of ATP [7, 20, 28]. The development of yeast strains capable of cellobiose utilization under aerobic conditions has been reported recently [7, 12, 20, 24]. An alternative approach has involved coordinate expression of a cellobiose transporter and cellobiose phosphorylase, instead of β-glucosidase, that utilizes inorganic phosphate to hydrolyze the disaccharide, forming one molecule each of glucose and glucose 1-phosphate . This latter pathway is hypothesized to be more energetically favorable as glucose-1-phosphate can be isomerized to glucose-6-phosphate without the expenditure of energy, thus reducing the overall ATP demand of cellobiose metabolism. This could be especially important under anaerobic fermentation conditions when ATP levels may be limiting.
We report here the metabolic engineering of an industrial strain of S. cerevisiae containing a novel xylose isomerase, two cellobiose transporters, and a cellobiose phosphorylase that is able to co-ferment both xylose and cellobiose to ethanol at high yield under completely anaerobic conditions.
Materials and methods
Strains and media
All reagents were obtained from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise indicated. E. coli DH5α was used as the recipient for construction of recombinant plasmids. Competent cells were obtained from Monserate Biotechnology Group (San Diego, CA). S. cerevisiae BY4742 (MATαhis3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) was obtained from the American Type Culture Collection (Manassas, VA, USA). The S. cerevisiae diploid strain Turbo, designated here as BF903, was acquired from HydroBrew (Oceanside, CA, USA). Yeast strains were cultured at 30 °C in either YPD (10 g/l yeast extract (BD, Franklin, NJ, USA), 20 g/l peptone (BD, Franklin, NJ), and 20 g/l glucose) or SCD-URA (6.7 g/l yeast nitrogen base (USB Swampscott, MA, USA), 20 g/l glucose and appropriate supplements). Intergenic sites for integration of expression cassettes were selected between neighboring convergent genes in transcriptionally active regions within the genome . Intergenic sites were designated by adding “.5” to the standard systematic yeast gene name, e.g., YLR113.5 indicates integration between genes YLR113 and YLR114.
Strains used in this study
BF2513 + YGL054.5::PTEF1−CBP−TCYC1 −URA3
BF2901 + YGR249.5::PTEF1−CBT1−TADH1−URA3
BF3210 + YDR223.5::PTEF1−LAC12−TADH1−URA3
BF2513 + URA3
BF3241 + YDR275.5::PTEF1−GAL2−TCYC1−URA3
BF3539 + YER150.5:: PTEF1−XKS1−TADH1−PTDH3−XKS1−TCYC1−URA3
BF3579 + Δpho13−URA3
BF3591 + Δgre3−URA3
BF3614 + YOL108.5::PTEF1−RfCO*−TCYC1−PTDH3−RfCO*−TADH1−URA3
Each expression cassette consisting of a promoter, gene of interest, and terminator was ligated along with a URA3 marker [2, 18] between the flanking regions of the targeted integration site and cloned into pUC19. The entire integration cassette was released from the vector backbone via PacI digestion and used in standard yeast transformation . Integration cassettes contained two 300-mers that flanked the site of integration. Codon optimized sequences were obtained from Integrated DNA Technologies (IDT, Coralville, Iowa) S. cerevisiae CAI score algorithm.
Phylogenetic trees were constructed using the MEGA5 program . Amino acid sequences of xylose isomerase homologs were obtained from NCBI with the following accession numbers: Ruminococcus flavefaciens 17, CAB51938; Piromyces sp. E2, CAB76571; Clostridium phytofermentans ISDg; YP_001558336; Thermus thermophilus HB8, P26997; Orpinomyces sp. ukk1, ACA65427; Bacteroides thetaiotaomicron VPI-5482, NP_809706; Escherichia coli str. K-12 substr. MG1655, NP_418022; Thermoanaerobacter brockii subsp. finnii Ako-1, YP_004186994; Streptomyces rubiginosus, P24300; Thermotoga maritima MSB8, NP_229467; Thermotoga neapolitana DSM 4359, YP_002534262; Streptomyces murinus, P37031; Lactobacillus pentosus, P21938; Geobacillus stearothermophilus, ABI49954; Bacteroides uniformis ATCC 8492, ZP_02069286; Clostridium cellulolyticum H10, YP_002507697; Ruminococcus flavefaciens FD-1, ZP_06143883; Ruminococcus sp. 18P13, CBL17278; Clostridiales genomosp. BVAB3 str. UPII9-5, YP_003474614; Bacteroides stercoris ATCC 43183, ZP_02435145. Sequences were aligned using ClustalW .
Xylose isomerase assays
Cells were inoculated to an initial OD600 of 0.1 and grown to an OD600 of 1.0 in 50 ml YPD, pelleted, and frozen at −80 °C. Cells were thawed then lysed with 2.5 ml/g cell paste of YPER-plus reagent (Thermo Fisher Scientific Inc., Rockford, IL, USA) for 30 min at room temperature on a rotating platform. Cell debris was removed by centrifugation at 3,000 rpm for 5 min at 4 °C and the supernatant was used for enzyme assays. Protein concentration was measured using a Bradford assay performed as described (Thermo Scientific Coomassie Plus Bradford Reagent; Rockford, IL, USA). Enzyme assays contained 100 mM Tris-HCl, pH7.5, 5 mM MnCl2, 0.3 mM NADH, 2 U sorbitol dehydrogenase (Roche; Basel, Switzerland), and an amount of cell extract containing 25 μg protein. Reactions were started with the addition of 500 mM xylose and absorbance measured at 340 nm for 5 min at 30 °C. A linear range of enzymatic activity was tested from 1 to 25 μg xylose isomerase. The Km was determined by varying the xylose concentration from 5 to 500 mM. Assays were performed in triplicate and normalized to the protein concentration .
Cellobiose phosphorylase assays
The cellobiose phosphorylase gene from Ruminococcus flavefaciens FD-1, cloned under the control of the TEF1 promoter and CYC1 terminator, was integrated into site YGL054.5 in Turbo strain BF2513. Cells were inoculated to an initial OD600 of 0.1 and grown to an OD600 of 1.0 in 50 ml YPD, pelleted by centrifugation and frozen at −80 °C. Cells were thawed and resuspended in 1 ml 50 mM Tris-HCl. Lysis was achieved by adding 200 μl glass beads and using a Biospec Products (Bartlesville, OK) Mini beadbeater for 1.5 min. Cell debris was removed by centrifugation at 3,000 rpm for 5 min at 4 °C and the protein concentration of cell-free lysate measured by the Bradford assay (Thermo Scientific Coomassie Plus Bradford Reagent; Rockford, IL). Activity was assayed in 50 mM PIPES pH 6.8, 33 mM sodium-potassium-phosphate buffer pH 6.8, 3 mM MgCl2, 1 mM ATP, 1 mM NADP+, 5 U hexokinase, 5 U glucose-6-phosphate dehydrogenase and 20 mM cellobiose. A linear range of activity was determined by assaying 1–50 μg cell-free lysate. Reactions were initiated by the addition of cellobiose and absorbance measured at 340 nm for 5 min at 30 °C. The Km was determined by varying the cellobiose concentration from 0 to 20 mM. Assays were performed in triplicate and normalized to the protein concentration.
Anaerobic growth on plates
Yeast nitrogen base lacking amino acids (6.7 g/l) and containing 20 g/l agar were supplemented with 2 % (w/v) glucose, 2 % (w/v) cellobiose, or 2 % (w/v) xylose, as specified. The plates were incubated under anaerobic conditions in a BD Diagnostics (Sparks, MD) Gaspak EZ incubation chamber according to manufacturer’s instructions. Inoculated plates were incubated at 30 °C under aerobic conditions for 3 days or anaerobic conditions for 4 days.
Alcohol Fermentation Monitors, AFM, (Applikon, Foster City, CA, USA) containing 400 ml of 1 % (w/v) yeast extract, 2 % (w/v) peptone, 8 % (w/v) glucose, 4 % (w/v) xylose, and 2 % (w/v) cellobiose, were inoculated to an initial OD600 of 0.6 and incubated at 30 °C with an agitation rate of 300 rpm. Six 500-ml vessels were run in parallel for 96 h. One-milliliter samples were obtained in duplicate at 0, 8, 24, 48, 72, and 96-h time points. An Agilent (Santa Clara, CA, USA) 1100 HPLC equipped with a Rezex RFQ-Fast Fruit H+ (8 %) column (Phenomenex, Torrance, CA, USA) and ultraviolet detectors was used to measure fermentation metabolites. Sulfuric acid (25 mM) in Milli-Q water (EMD Millipore Corporation, Billerica, MA, USA) was used as mobile phase. The column temperature was held constant at 60 °C with an isocratic flow rate of 0.9 ml/min.
Identification of a phylogenetically distinct xylose isomerase
Specific activities of xylose isomerases integrated in single copy in S. cerevisiae BY4742 at site YLR113.5
Specific activity (U/mg)
% increase over native
0.29 ± 0.008
0.32 ± 0.013
0.34 ± 0.013
0.37 ± 0.007
1.41 ± 0.029
Modifications to enhance xylose isomerase expression and activity
We set out to improve the xylose isomerase activity in yeast by increasing both the expression and activity of the enzyme. We hypothesized that the 5′ end of the gene and the corresponding N-terminus of the encoded protein would be important to efficient transcription and translation of XI-Rf. Thus, we replaced the first ten amino acids of the XI-Rf protein sequence with the corresponding amino acids from the Piromyces xylose isomerase, since it was previously shown to be highly expressed in S. cerevisiae . This was accomplished by aligning the XI-Rf native amino acid sequence with the Piromyces xylose isomerase. The aligned sequences indicated that the N-terminus of the Piromyces gene was two amino acids longer than the corresponding XI-Rf sequence. Replacement of the first ten amino acids of the XI-Rf protein sequence with the corresponding 12 amino acids of the Piromyces sequence resulted in an 8.9 % increase in the in vitro xylose isomerase activity over the native XI-Rf protein (Table 2).
Next, we aimed to improve XI-Rf activity by site-directed mutagenesis. The XI-Rf amino acid sequence was aligned with homologs from Pyromyces and Clostridium phytofermentans, both known to be active in S. cerevisiae, and residues near the xylose binding site were compared. XI-Rf and Clostridium phytofermentans both contained a glycine at position 179, whereas this residue was an alanine in Pyromyces. The G179A mutation led to a 15.1 % increase in activity in cell-free extracts (Table 2). Interestingly, the G179A mutation combined with the 5′-P10 modification resulted in an additive increase in xylose isomerase activity in cell-free extracts (Table 2). The XI-Rf-P10-G179A was selected as our best-engineered XI-Rf, with a 26.8 % increase in activity over the wild-type enzyme. This sequence was then codon optimized using Integrated DNA Technologies (IDT, Coralville, IA, USA) S. cerevisiae CAI score algorithm, generating RfCO* with a CAI score  of 0.83 compared to 0.72 for the native XI-Rf. Codon optimization led to a substantial increase in expression resulting in a 385.2 % (4.8-fold) increase in activity in cell-free extracts over the native enzyme (Table 2). Interestingly, RfCO* also had a lower Km for xylose, 66.7 ± 2.9 mM.
Two copies of RfCO* in divergent orientation were integrated into intergenic site YOL108.5 in BF3645. One copy was placed under the transcriptional control of the TEF1 promoter/CYC1 terminator, and the other under the TDH3 promoter/ADH1 terminator. It was found to have a specific activity of 1.2 ± 0.07 μmol/min/mg.
Cellobiose phosphorylase activity assays
A single copy of the CBP1 gene encoding cellobiose phosphorylase from Ruminococcus flavefaciens FD-1 was integrated at intergenic site YGL054.5 of BF2513 and assayed for in vitro activity. It was found to have a specific activity of 0.75 ± 0.1 μmol/min/mg protein and a Km for cellobiose of 0.55 ± 0.04 mM.
Growth on xylose and cellobiose
AFM fermentation results with standard deviation of three replicates per strain, grown in YP media with 8 % (w/v) glucose, 4 % (w/v) xylose, and 2 % (w/v) cellobiose
Xylose consumed (g/l)
Cellobiose consumed (g/l)
Glycerol produced (g/l)
Xylitol produced (g/l)
Ethanol produced (g/l)
Ethanol yield (g/l/g sugar consumed)
Ethanol production rate (g/l/h)
3.9 ± 0.71
0.4 ± 0.23
3.4 ± 0.34
0.7 ± 0.02
37.7 ± 0.25
0.45 ± 0.01
0.4 ± 0.00
19.0 ± 0.81
9.8 ± 2.19
4.3 ± 0.05
1.3 ± 0.22
48.6 ± 0.99
0.44 ± 0.01
0.5 ± 0.01
There have been several attempts to engineer yeast to produce ethanol from cellulosic feedstocks. Our approach has been to construct a single Saccharomyces cerevisiae strain capable of fermenting glucose, xylose, and cellobiose under completely anaerobic conditions. Our strategy is based on the coordinate expression of a xylose isomerase and a cellobiose phosphorylase from Ruminococcus flavefaciens, unlike previous reports that used a combination of xylose reductase, xylitol dehydrogenase [7, 20] and either an extracellularly displayed β-glucosidase  or an intracellularly expressed β-glucosidase [7, 20]. Although a xylose metabolic pathway based on the sequential activities of xylose isomerase and xylulokinase is cofactor neutral, engineering a xylose fermenting strain required several additional genetic modifications. Since the native GAL2p has been shown to function as a fortuitous xylose transporter [8, 31–33], we integrated a second copy of the gene under the control of a strong constitutive TEF1 promoter that is not regulated by glucose. We also deleted the GRE3 gene, encoding a non-specific aldose reductase, previously shown to minimize xylitol production [3, 11, 14, 16, 19, 26, 35, 37] and prevent potential competitive inhibition of xylose isomerase . Indeed, BF3645 containing the gre3 deletion produced little or no xylitol during xylose fermentations. Xylose enters the pentose phosphate pathway via the action of xylulokinase , which is known to be limiting during growth on xylose [3, 9–11, 21, 26], thus we also over-expressed this gene in our strain. We also deleted the PHO13 gene encoding a phosphatase that has been shown to increase growth on xylose while increasing expression of TAL1 and TKL1, although the mechanism is not fully understood [5, 25, 40, 41]. Additionally, we integrated the cellobiose transporters encoded by the K. lactisLAC12  and N. crassaCBT1  genes.
We increased in vitro xylose isomerase activity of XI-Rf by engineering the 5′-end of the gene, introducing a site-specific mutation, G179A, and codon optimization. Overall these modifications resulted in a 4.8-fold increase in xylose isomerase activity in cell-free extracts, and also decreased the Km for xylose. Surprisingly, this strain was incapable of growth on xylose as the sole carbon source under aerobic or anaerobic conditions. While a Km of 66.7 mM is comparable to other xylose-assimilation enzymes, including xylose reductases, it is still higher than the micromolar Km common for hexokinases involved in the first step of glucose utilization. It is rational to suspect that this low affinity for xylose may be a key hurdle for rapid and efficient utilization of xylose comparable to the utilization of glucose. We suspect that metabolism of xylose or cellobiose alone does not generate sufficient ATP to support the growth of BF3645 on either substrate as the sole carbon source  when plated on minimal media and incubated under anaerobic conditions, particularly since transport of xylose depends on expenditure of ATP. In contrast, xylose and cellobiose consumption and fermentation to ethanol took place under anaerobic conditions, but in liquid media also containing glucose and yeast extract (1 % w/v) to support the fermentation. Note that xylose and cellobiose consumption and fermentation to ethanol continued even after the complete exhaustion of glucose from the fermentation media. We suspect xylose fermentation under these fully anaerobic conditions is supported by the ATP generated via cellobiose metabolism. From a practical perspective, the inability to grow on either substrate under anaerobic fermentation conditions is perhaps fortuitous by minimizing yield loss to biomass formation and increasing the ethanol yield based on consumed sugar. While it has been hypothesized that a cellobiose fermentation pathway based on cellobiose phosphorylase may be energetically favorable over one based on β-glucosidase, it is clearly not sufficient, by itself, to allow growth on cellobiose under anaerobic conditions. It is possible that increasing cellobiose phosphorylase activity may alleviate this constraint, or that a passive cellobiose transporter may be necessary to confer anaerobic growth on cellobiose. Interestingly, both BF3264 and BF3645 had similar ethanol yields despite BF3645 consuming far more cellobiose and xylose sugars, indicating that BF3645 is capable of co-fermenting xylose and cellobiose at the same efficiency as it ferments glucose.
All authors have agreed to submit this manuscript to the “Journal of Industrial Microbiology and Biotechnology”. The authors would like to thank the following people for providing assistance to the work presented in this manuscript: Malcolm Robertson, Jessica Long, Binh Phung, Doug Decker, Aran Gregorian, June Hoang, Moe Abouzari, Jessica Greiner, Pierre-Yves De Wals, and Elaine Ito.