Systematic optimization of gene expression of pentose phosphate pathway enhances ethanol production from a glucose/xylose mixed medium in a recombinant Saccharomyces cerevisiae

Second-generation technology has been developed for bioethanol production from non-edible biomass composed of cellulose and hemicellulose, which are sugar polymers of d-glucose, d-xylose and l-arabinose (Naik et al. 2010) that can be hydrolyzed by cellulase and hemicellulase (Bhattacharya et al. 2015). The yeast Saccharomyces cerevisiae readily produces ethanol from d-glucose, but not from pentose sources such as d-xylose and l-arabinose (Young et al. 2010). In 1990, introduction of two Scheffersomyces stipites (formerly known as Pichia stipites) genes, XYL1 and XYL2, encoding NADPH-dependent xylose reductase (XR) and NAD⁺-dependent xylitol dehydrogenase (XDH), respectively, into S. cerevisiae was shown to allow successful xylose fermentation (Kötter et al. 1990). Moreover, various genetic engineering trials with endogenous or exogenous genes have been applied to yeast transformants carrying xylose-metabolizing enzyme genes to improve ethanol production in xylose-containing media. For instance, xylose metabolism was markedly improved by deletion of the putative alkaline phosphate gene PHO13 in XLY1/XYL2 transformants (Van Vleet et al. 2008; Kim et al. 2015; Xu et al. 2016; Kobayashi et al. 2017). Overexpression of HXT7, a known xylose-permeable hexose transporter in S. cerevisiae, also improved xylose consumption (Gonçalves et al. 2014). However, critical issues remained for efficient xylose fermentation by XYL1/XYL2 transformants such as accumulation of xylitol, sedoheptulose-7-phosphate and pyruvate, and depletion of fructose-1,6-bisphosphate because of insufficient capacity of the glycolytic pathway (Kötter and Ciriacy 1993). An imbalance of cofactors can also slow xylose metabolism (Van Vleet and Jeffries 2009). Previous studies reported that overexpression of non-oxidative pentose phosphate pathway (PPP) genes did not improve the xylose fermentation rate (Walfridsson et al. 1995; Johansson and Hahn-Hägerdal 2002; Matsushika et al. 2009).

Another strategy to enhance xylose metabolism in S. cerevisiae depended on the xylose isomerase (XI) gene derived from Thermus thermophiles (Walfridsson et al. 1996). The S. cerevisiae XI transformants had a better ethanol yield than did XYL1/XYL2 since production inhibition by xylitol and cofactor imbalances was avoided (Brat et al. 2009). Meanwhile, deletion of either PHO13 (Lee et al. 2014; Bamba et al. 2016) or the aldo–keto reductase gene GRE3 (Träff et al. 2001), as well as introduction of an HXT7 mutant (Reider Apel et al. 2016) improved xylose metabolism in the XI transformants.

In addition to introduction of the xylose metabolism gene, balanced expression of PPP genes is important for bioethanol production (Kuyper et al. 2005). Many studies revealed that overexpression of four non-oxidative PPP genes, transaldolase (TAL1), transketolase (TKL1), ribose-5-phosphate ketol-isomerase (RKI1) and d-ribulose-5-phosphate 3-epimerase (RPE1) increased ethanol production in medium in which d-xylose was the sole carbon source (Kuyper et al. 2005; Karhumaa et al. 2007; Lee et al. 2014; Qi et al. 2015). However, the effect of exogeneous expression of PPP genes on the rates of xylose consumption and ethanol production, particularly in a glucose/xylose-mixed medium, remained unclear, and no studies had addressed suitable activities for PPP enzymes that would enhance xylose metabolism. For ethanol production from multiple sugar sources, the metabolic flow of each sugar through the glycolytic pathway (GP) and PPP may be intricately involved in xylose metabolism. Furthermore, fermentation at high temperatures is preferable in the simultaneous saccharification and co-fermentation (SSCF) process because the optimal temperature for the cellulase reaction is much higher than that for fermentation by most ethanologenic microorganisms (Paulova et al. 2014; Kawaguchi et al. 2016).

To explore the optimal expression levels of PPP genes needed to maximize ethanol production from cellulosic sources, here we systematically investigated the effect of overexpression of single and multiple PPP genes derived from S. cerevisiae and the thermotolerant yeast Kluyveromyces marxianus in a S. cerevisiae strain harboring the XI gene that allows glucose/xylose co-fermentation at high temperatures. We explored differences in the favorable combination of PPP genes in S. cerevisiae between our previous study using XR-XDH transformants in medium having d-xylose as the sole sugar (Kobayashi et al. 2017) and this study using the XI transformants in the glucose/xylose-mixed medium.

For the commercial production of bioethanol from cellulosic materials, development of microorganisms capable of highly efficient xylose fermentation is critical. To achieve this objective, many studies have demonstrated that expression of PPP genes can improve xylose metabolism (Kuyper et al. 2005; Karhumaa et al. 2007; Lee et al. 2014; Qi et al. 2015). We recently showed that the best combination of PPP gene overexpression in the S. cerevisiae haploid strain IR-2 that carries XR-XDH genes depended on the fermentation temperature in 60 g/L xylose medium (Kobayashi et al. 2017). For the SSCF process, one of the most economical fermentation processes to produce ethanol from cellulosic material at an industrial scale, the preferable temperature is higher than that for ethanologenic microorganisms. In addition, simultaneous consumption of d-glucose and d-xylose is desirable for economical ethanol production from cellulosic biomass. The concentrations of sugars in the raw material and the utilization of these sugars during the course of fermentation are also important factors for economical ethanol production, because changes in the concentrations of these sugars during glucose/xylose co-fermentation can alter gene expression levels in the microorganisms. For these reasons, it is critical to optimize expression levels of non-oxidative PPP genes under conditions that mimic those in the industrial process rather than experimental conditions such as low temperature growth on d-xylose-containing medium that was used for our previous study. However, the effects of overexpression of PPP genes on glucose/xylose co-fermentation, particularly at higher temperature, of S. cerevisiae IR-2 expressing XI were unknown.

In the present study, we systematically investigated how overexpression of PPP genes derived from S. cerevisiae and thermotolerant yeast Kluyveromyces marxianus expressed either singly or in combination by a S. cerevisiae strain harboring the XI gene affected glucose/xylose co-fermentation at 36 °C. The concentration of sugars, 85 g/L d-glucose and 35 g/L d-xylose, used in the fermentation experiments are based on typical sugarcane bagasse saccharified solutions. Among strains overexpressing a single PPP gene, TKL1 (YK185) or RKI1 (YK183) showed a dramatically increased d-xylose consumption rate at 24 h on glucose/xylose co-fermentation, whereas TAL1 (YK184) and RPE1 (YK186) did not (Fig. 1 and Table 2, also summarized in Fig. 4). Further experiments with yeast expressing multiple PPP genes clearly showed an optimal combination of PPP genes that could be used to improve xylose fermentation at high temperature: overexpression of TKL1 from S. cerevisiae with RKI1 and TAL1 from K. marxianus produced the highest rates of xylose consumption and ethanol productivity among the strains examined (YK246, Figs. 2, 3 and Table 2, also summarized in Fig. 4). This result may appear to be inconsistent with a previous report suggesting that the efficiency of xylose fermentation should be attributed, at least in part, to increased TAL1 expression in a S. cerevisiae strain expressing XI obtained by an adaptive evolution experiment (Vilela Lde et al. 2015). This study suggested that the amount of intermediary metabolites would differ between glucose/xylose co-fermentation at 36 °C and xylose fermentation alone at 30 °C. Other studies also reported that overexpression of four PPP genes could improve xylose metabolism (Kuyper et al. 2005; Qi et al. 2015). However, comparisons of our results with these previous studies is difficult due to differences in the experimental conditions, including genes for xylose metabolism, genetic background of the parental strain, fermentation temperature, inoculum dose and sugar concentrations. Thus, several points raised by our results should be addressed by future studies, including: (i) the suggestion that sedoheptulose 7-phosphate levels will be deficient in strains that do not overexpress RKI1 and TKL1 during glucose/xylose co-fermentation at high temperature because TKL1 synthesizes sedoheptulose 7-phosphate and glyceraldehyde 3-phosphate from xylose 5-phosphate and ribose 5-phosphate; and (ii) whether the synthesis of ribose 5-phosphate from ribulose 5-phosphate by RKI1 is insufficient. Overexpression of RPE1 may not be needed to increase the d-xylose consumption rate on glucose/xylose co-fermentation at a high temperature as its gene product could induce competition of enzyme reactions between xylulose 5-phosphate from xylose to ribulose 5-phosphate from glucose. In fact, only slight differences in d-xylose consumption rates at 24 h and ethanol production at 72 h were seen between YK246 and YK193 (Fig. 1 and Table 2, also summarized in Fig. 4). Thus, the present study clearly showed that overexpression of the PPP genes KmRKI1, KmTAL1 and ScTKL1 provided the best performance for xylose consumption and ethanol production on glucose/xylose co-fermentation at high temperature, indicating that optimizing xylose metabolism according to the individual and realistic experimental conditions, especially fermentation temperature, type and concentration of sugars and cell growth during fermentation, will be important for generating strains that have the best performance under these conditions.

To improve xylose metabolism, factors other than PPP may also need to be considered. Elimination of NAD-dependent glycerol 3-phosphate dehydrogenase (GPD) that has two homologous genes, GPD1 and GPD2, which are involved in glycerol metabolism by S. cerevisiae (Eriksson et al. 1995), increased ethanol yield while decreasing glycerol production (Björkqvist et al. 1997; Henningsen et al. 2015). In a preliminary examination, we verified that SS118 (XI gpd2Δ) exhibited decreased glycerol production and improved ethanol production (Seike et al. unpublished observations). As such, in the present study, each PPP gene was inserted in the GPD2 locus. As expected, glycerol production in all the strains was reduced compared with the parental strain SS82 (Additional file 2: Table S2).

On the other hand, whether deletion of PHO13, a putative alkaline phosphate gene, improves xylose fermentation is controversial. Several studies showed that deletion of PHO13 improves xylose metabolism in S. cerevisiae expressing XR-XDH. The improvement yielded by PHO13 deletion could be due to altered redox levels on d-xylose (Van Vleet et al. 2008) and xylulose-5-phosphate phosphatase (Kim et al. 2013) that induce a global transcriptional response (Kim et al. 2015). The deletion of PHO13 had a beneficial effect on fermentation with improved activity of the XI pathway, and high ethanol yield (0.45 g/g) in 40 g/L d-xylose medium at 30 °C (Lee et al. 2014). Another study demonstrated that disruption of PHO13 enhanced the d-xylose consumption rate (0.31 g/g-cell/h), ethanol productivity (0.11 g/g-cell/h) and ethanol yield (0.45 g/g) in an XI-expressing, xylose-fermenting yeast strain under oxygen-limited conditions in 50 g/L d-xylose at 30 °C (Bamba et al. 2016). In contrast to these reports, deletion of PHO13 in S. cerevisiae expressing XI sharply weakened growth on xylose plates (Shen et al. 2012) and promoted accumulation of 4-phospho-erythronate and 6-phospho-gluconate (Collard et al. 2016). Based on these earlier results, we examined and evaluated the effect of deletion of PHO13 and overexpression of all PPP genes on a glucose/xylose mixed medium or only xylose medium at 36 °C in the recombinant S. cerevisiae strains YK001 (XR-XDH), YK002 (XR-XDH pho13Δ), YK115 (XR-XDH pho13Δ:ScRKI1-ScRPE1-ScTAL1-ScTKL1), YK149 (XI pho13Δ) and YK150 (XI pho13Δ:ScRKI1-ScRPE1-ScTAL1-ScTKL1). S. cerevisiae strains expressing XR-XDH consumed 85 g/L d-glucose within 24 h but had incomplete consumption of 35 g/L d-xylose even at 72 h on glucose/xylose co-fermentation at 36 °C when cells were initially inoculated at OD₆₀₀ = 3 (Additional file 3: Figure S1). Notably, no difference was observed between the parental (YK001, SS82) and PHO13Δ (YK002, YK149) strains, but xylose consumption was decreased in the strain that overexpressed all four PPP genes (YK115, YK150) on glucose/xylose co-fermentation (Additional file 3: Figure S1). Incidentally, the concentrations of ethanol production of YK001 and SS82 were about 45.1 g/L and 48.6 g/L at 72 h, respectively. Ethanol production by strains carrying XI could have been higher than those with XR-XDH due to a decrease in the by-product xylitol compared with xylose consumption. Although S. cerevisiae strains expressing XI can directly convert d-xylose to d-xylulose without a redox reaction, GRE3 encoding an intrinsic aldo-keto reductase and producing xylitol should be deleted to reduce xylitol formation (Yamanaka. 1969; Träff et al. 2001).

In xylose fermentation, however, markedly different results were obtained for strains with XR-XDH and XI. For strains harboring XR-XDH, xylose metabolism was clearly improved by deletion of PHO13 (YK002) and additional overexpression of the four PPP genes (YK115) as expected. For strains harboring XI, however, xylose consumption was decreased by the deletion of PHO13 and recovered to control levels by overexpression of the four PPP genes (Additional file 4: Figure S2, Kobayashi et al. 2017). Ismail et al. (2013) showed that the expression of several genes was up- or downregulated by more than twofold at 38 °C compared to at 30 °C in S. cerevisiae carrying XR-XDH. These results suggested that PPP could be a limiting step during ethanol production from d-xylose at high temperature when severely downregulated PPP gene expression might result in accumulation of certain intermediates that in turn affect the metabolic flux in PPP. Whatever the case, our results indicate that PHO13 deletion should not be adopted due to its negative effect on strains carrying XI.

Developing an ideal strain for cellulosic ethanol production is challenging due to the complex nature of glucose/xylose co-fermentation. The entire fermentation process can be divided into three phases: (i) glucose assimilation (on catabolite repression), (ii) simultaneous assimilation of glucose and xylose (off catabolite repression), and (iii) subsequent assimilation of residual xylose. The interphase transitions can dramatically affect the expression of various metabolic genes such as hexose transporters since activities of gene promoters can be altered by extracellular concentration and uptake of sugars (Sedlak and Ho 2004). Here, we used the well-characterized strong promoter TDH3 and CYC1 terminator to control expression of PPP genes. Future studies should examine suitable promoters for the regulation of the expression of individual PPP genes to provide the necessary levels of PPP enzymes that promote harmonized xylose metabolic flux. Taken together, the activities of PPP enzymes required for ideal xylose metabolism differ according to glucose/xylose co-fermentation and for xylose fermentation at high temperature (Fig. 5).