Xylose reductase from Pichia stipitis with altered coenzyme preference improves ethanolic xylose fermentation by recombinant Saccharomyces cerevisiae

Background Xylose reductase (XR) and xylitol dehydrogenase (XDH) from Pichia stipitis are the two enzymes most commonly used in recombinant Saccharomyces cerevisiae strains engineered for xylose utilization. The availability of NAD+ for XDH is limited during anaerobic xylose fermentation because of the preference of XR for NADPH. This in turn results in xylitol formation and reduced ethanol yield. The coenzyme preference of P. stipitis XR was changed by site-directed mutagenesis with the aim to engineer it towards NADH-preference. Results XR variants were evaluated in S. cerevisiae strains with the following genetic modifications: overexpressed native P. stipitis XDH, overexpressed xylulokinase, overexpressed non-oxidative pentose phosphate pathway and deleted GRE3 gene encoding an NADPH dependent aldose reductase. All overexpressed genes were chromosomally integrated to ensure stable expression. Crude extracts of four different strains overexpressing genes encoding native P. stipitis XR, K270M and K270R mutants, as well as Candida parapsilosis XR, were enzymatically characterized. The physiological effects of the mutations were investigated in anaerobic xylose fermentation. The strain overexpressing P. stipitis XR with the K270R mutation gave an ethanol yield of 0.39 g (g consumed sugars)-1, a xylitol yield of 0.05 g (g consumed xylose)-1 and a xylose consumption rate of 0.28 g (g biomass)-1 h-1 in continuous fermentation at a dilution rate of 0.12 h-1, with 10 g l-1 glucose and 10 g l-1 xylose as carbon sources. Conclusion The cofactor preference of P. stipitis XR was altered by site-directed mutagenesis. When the K270R XR was combined with a metabolic engineering strategy that ensures high xylose utilization capabilities, a recombinant S. cerevisiae strain was created that provides a unique combination of high xylose consumption rate, high ethanol yield and low xylitol yield during ethanolic xylose fermentation.


Background
Xylose is the second most abundant carbohydrate in nature after glucose and one of the fermentable sugars in lignocellulosic biomass. Saccharomyces cerevisiae is the industrially most commonly used ethanol producer and S. cerevisiae strains have been genetically engineered to utilize xylose (recently reviewed in [1][2][3]).
In fungi, xylose catabolism begins with its conversion by xylose reductase (XR) and xylitol dehydrogenase (XDH) to xylulose, which after phosphorylation is assimilated via the non-oxidative pentose phosphate pathway. Anaerobic xylose fermentation by recombinant S. cerevisiae strains harbouring the XR-XDH pathway generally results in ethanol yields far below the theoretical 0.51 g g -1 [4][5][6][7][8]. A significant fraction of the consumed xylose is secreted as xylitol, which has been ascribed to the difference in cofactor preference of the NAD(P)H-dependent XR and the NAD + -dependent XDH [4,9]. Xylitol formation can be limited by expressing a xylose isomerase (XI) instead of the XR-XDH pathway [10,11]. However, in a recent study, a strain carrying the Pichia stipitis XR-XDH pathway showed significantly higher xylose consumption rate and higher specific ethanol productivity compared with an isogenic strain carrying the Pyromyces XI pathway [12]. This indicates that an XR-XDH pathway engineered to be redox-neutral while maintaining the capability of high flux towards the central carbon metabolism could be the key to efficient anaerobic xylose fermentation.
In the current study, genes encoding XRs with different cofactor affinities were expressed in S. cerevisiae strains with high xylose utilization capability due to overexpression of XK and the non-oxidative PPP. Minimized background NADPH-dependent aldose reductase activity was ensured by GRE3 gene deletion. The effect of different mutations in P. stipitis XR on the kinetic properties of the enzyme and on the xylose fermentation capability of the corresponding strains was evaluated. Additionally, the C. parapsilosis XYL1 gene [25], encoding an NADH-preferring XR, was expressed and evaluated in recombinant S. cerevisiae.

Plasmids and Strains
Relevant genotype Reference Competent E. coli DH5α cells were prepared and transformed as described elsewhere [27] and transformed E. coli strains were selected on LB plates [26] containing 100 mg l -1 ampicillin (IBI Shelton Scientific, Inc., Shelton, CT). E. coli strains were grown in LB medium containing 100 mg l -1 ampicillin for plasmid amplifications. Yeast strains were transformed with the lithium acetate method [28] and transformed yeast strains were selected for prototrophy on defined mineral medium plates containing 20 g l -1 glucose.

Construction of TMB 3200
The P. stipitis XYL1 gene carrying the K270R (Lys270Arg) mutation was generated by site-directed mutagenesis using the overlap extension PCR protocol [29]. In the first step, two separate PCR amplifications were done using plasmid YIplac211 PGK XYL1(K270M) [13] as template, primers 5XYL1s and 3K270R (

Batch fermentation
Anaerobic batch fermentation was carried out in 3-litre ADI Autoclavable Bio Reactor Systems (Applikon, Schiedam, The Netherlands) with a working volume of 1 litre. Cells were pre-cultivated in shake flasks in defined mineral medium with 20 g l -1 glucose, washed with sterile water and inoculated into the bioreactor to an optical density at 620 nm (OD620) of 0.2. Defined mineral medium with doubled concentration of all salts, trace elements and vitamins, containing 20 g l -1 glucose and 50 g l -1 xylose, was used. The temperature was 30°C, stirring was set to 200 rpm and pH 5.5 was maintained with 3 M KOH. Anaerobic conditions were attained by sparging with nitrogen gas containing less than 5 ppm O 2 (AGA GAS AB, Sundbyberg, Sweden) before inoculation. During fermentation, anaerobic conditions were maintained by the produced CO 2 that diffused through a water lock. The experiments were performed at least in biological duplicates.

Continuous fermentation
Continuous fermentation was conducted anaerobically in 2-litre Biostat ® A bioreactors (B. Braun Biotech International, Melsungen, Germany) with a working volume of 1 litre. Defined mineral medium with 10 g l -1 glucose and 10 g l -1 xylose was used for pre-cultivation and continuous fermentation. Cells pre-cultivated in shake flasks and washed with sterile water were used to inoculate the bioreactor to an OD620 of 0.2. Continuous fermentation at dilution rates of 0.06 and 0.12 h -1 was started after glucose depletion. The temperature was 30°C, stirring 200 rpm and pH 5.5 was maintained with 3 M KOH. Anaerobic conditions were obtained by sparging with nitrogen gas containing less than 5 ppm O 2 (AGA GAS AB, Sundbyberg, Sweden) at a constant gas flow of 0.2 litre min -1 controlled by mass flow meters (Bronkhorst HI-TEC, Ruurlo, the Netherlands). The off-gas condensers were cooled to 4°C, and the medium reservoirs were continuously sparged with nitrogen gas. Steady-state was assumed after five residence times and verified by measurements of cell density and CO 2 production. The experiments were performed in biological duplicates.

Analyses
Growth was determined by measuring OD620 with a Hitachi U-1800 Spectrophotometer (Hitachi Ltd., Tokyo, Japan  [33] and RNA [34] in the biomass were determined at steady-state in continuous fermentation. A previously developed stoichiometric model [35] was used to estimate the intracellular carbon fluxes at steady-state in continuous fermentation.

Ethanol Evaporation
Ethanol evaporation was determined experimentally for the setup used for continuous fermentation. Ethanol was added to a fermentor sparged with a nitrogen gas flow of 0.2 litre min -1 and the ethanol concentration was measured over time. The evaporation rate followed Equation (1) with a proportionality constant of k = 0.004.
Ethanol evaporation was estimated for each continuous fermentation and constitutes together with ethanol measured by HPLC the total ethanol production.

Enzymatic activity
Strains were cultivated for enzyme activity measurements in defined mineral medium containing 20 g l -1 glucose and harvested in the exponential growth phase. Cells were washed with sterile water and treated with yeast protein extraction solution Y-PER (Pierce, Rockford, IL, USA). Coomassie Protein Assay Reagent (Pierce) was used to determine protein concentration with Albumin Standard (Pierce). NAD(P)H-dependent XR activity was determined using an Ultrospec 2100 pro spectrophotometer (Amersham Biosciences, Uppsala, Sweden) operating at 30°C and 340 nm (ε NAD(P)H = 6220 M -1 cm -1 ). Triethanolamine buffer (100 mM, pH 7.0) was used and reactions were started by addition of xylose. Crude extracts from strains Y-PsNative, Y-PsK270M, Y-PsK270R and Y-CpXR were assayed for functional XR expression using a standard assay with 200 μM NAD(P)H and 350 mM xylose as previously described [7]. XR kinetics in crude extracts from strains Y-PsNative, Y-PsK270M and Y-PsK270R were determined, with concentrations of xylose and NAD(P)H varied from less than half to more than five times the respective apparent K m value. The initial rates were fitted by unconstrained non-linear optimization in MatLab R2006a to Equation (2), which describes the initial rate for a two-substrate reaction following a compulsory-order ternary-complex mechanism [36].

Strain construction
XR encoded by the C. parapsilosis XYL1 gene is the first XR enzyme reported to prefer NADH [25]. The C. parapsilosis XR carries an arginine instead of a lysine in the Ile-Pro-Lys-Ser motif that is conserved among NADPH-dependent xylose reductases [37]. The replacement of the lysine by an arginine in the K270R mutant of P. stipitis XR was made to mimic the C. parapsilosis XR. Strain TMB 3200 expressing the K270R mutant of P. stipitis XR (Table 1) was constructed to assess the influence of the mutation on xylose fermentation by recombinant S. cerevisiae. The strain was compared in anaerobic continuous fermentation with TMB 3001 [7], which carries the native P. stipitis XR, XDH and overexpressed endogenous XK. Increased ethanol yield and decreased xylitol yield was observed but the xylose utilization rate was not improved (results not shown). It was suspected that the xylose utilization rate was limited by other factors than the cofactor imbalance caused by the NAD(P)H-dependent XR and the strictly NAD + -dependent XDH.
Overexpression of XK together with the non-oxidative PPP improved xylose utilization by recombinant S. cerevisiae [22,23]. Also, the deletion of the endogenous aldose reductase GRE3 minimized background XR activity and decreased xylitol formation [24]. Four isogenic CEN.PKbased strains (Table 1) with these features were constructed to evaluate how the kinetic properties of XR affect xylose fermentation by recombinant S. cerevisiae. Strain Y-PsNative carrying the native P. stipitis XR served as a reference strain. Y-PsK270M contained the K270M mutant of P. stipitis XR that previously has been shown to reduce xylitol yield and increase ethanol yield in xylose fermentation [13]. Y-PsK270R expressed the K270R mutant of P.  Xylose consumption, ethanol production and product yields after 117 hours (see Figure 1) anaerobic batch fermentation of 20 g l -1 glucose and 50 g l -1 xylose by strains Y-PsNative, Y-PsK270M and Y-PsK270R. a (g xylitol (g consumed xylose) -1 )    (Table 3).
A kinetic study was made on crude extracts from strains Y-PsNative, Y-PsK270M and Y-PsK270R. The data was fitted to Equation (2) and the resulting constants are summarized in Table 3. Compared with native XR from P. stipitis, the K270M mutation resulted in a significant increase in the K m values for both NADPH and NADH. In fact, the kinetic parameters for the NADH-linked reaction catalyzed by the K270M mutant could not even be determined since this mutant could not be saturated with NADH. The K270R mutation increased the K m value for NADPH 25-fold, while the K m for NADH only increased two-fold.

Batch fermentation
Strains Y-PsNative, Y-PsK270M and Y-PsK270R were compared in anaerobic batch fermentation with 20 g l -1 glucose and 50 g l -1 xylose (Figure 1). Table 4 summarizes xylose consumption, ethanol concentration and product yields after 117 h of fermentation. The reference strain Y-PsNative consumed 30.4 g l -1 xylose and produced 16.7 g l -1 ethanol while Y-PsK270R consumed 46.1 g l -1 xylose and produced 25.3 g l -1 ethanol. Y-PsK270M consumed the least xylose (16.8 g l -1 ) and produced the lowest ethanol concentration (14.1 g l -1 ) of the three strains. The reference strain Y-PsNative produced an ethanol yield of 0.33 g ethanol (g consumed sugars) -1 and a xylitol yield of 0.26 g xylitol (g consumed xylose) -1 . Both strains with mutated XR produced higher ethanol yields (0.38 g ethanol (g consumed sugars) -1 ) and significantly lower xylitol yields (0.09 g xylitol (g consumed xylose) -1 ) than the reference strain.

Continuous fermentation and flux analysis
Y-PsNative and Y-PsK270R were compared in anaerobic continuous fermentation with a feed containing 10 g l -1 glucose and 10 g l -1 xylose ( Table 5). The continuous fermentation results were generally in agreement with the batch fermentation results (Table 4). Y-PsK270R gave 4% higher ethanol yields than Y-PsNative at both dilution rates. Y-PsK270R showed 17% and 9% higher specific xylose consumption rates and gave 60% and 78% lower xylitol yields compared with the reference strain Y-PsNative at dilution rates 0.06 h -1 and 0.12 h -1 respectively. Y-PsK270R also gave 17% and 22% lower glycerol yields The metabolic fluxes through Y-PsNative and Y-PsK270R were estimated ( Figure 2) using a stoichiometric model [35]. The flux values were normalized to a total specific sugar consumption of 100 mmol g -1 biomass h -1 . The xylose fraction of the total specific sugar consumption was smaller for both strains at dilution rate 0.12 h -1 compared with 0.06 h -1 . According to the model, Y-PsK270R utilized a larger fraction of NADH in the XR reaction (90% and 100%) than Y-PsNative (59% and 74%) at dilution rates 0.06 h -1 and 0.12 h -1 respectively. The model also predicts that a smaller fraction of glucose-6-phospate enters the oxidative PPP in Y-PsK270R (11% and 7%) than in Y-PsNative (14% and 12%) at dilution rates 0.06 h -1 and 0.12 h -1 respectively.

Discussion
This is the first time that an efficient xylose fermenting S. cerevisiae strain has been generated by a targeted metabolic engineering strategy, and where the XR-XDH xylose utilization pathway was chromosomally integrated to ensure stable expression of all required components (Table 6) [38]. Expression of K270R XR proved to be superior to expression of native XR in both batch and continuous fermentation. When compared with other strains in anaerobic continuous fermentation with 10 g l -1 glucose and 10 g l -1 xylose (Table 6), strain Y-PsK270R uniquely combines high xylose consumption rate and high ethanol yield due to low xylitol production. Strains TMB3270 and TMB3271, harbouring XR with the K270M mutation, gave lower xylitol yields compared with their reference strains TMB3001 and TMB3260 (Table 6) [13]. However, the K270M mutation also decreased the xylose consumption rate. Improved xylose utilization capability has previously Metabolic flux analysis Figure 2 Metabolic flux analysis. Estimated metabolic fluxes in Y-PsNative and Y-PsK270R (bold) from continuous fermentation at dilution rate 0.06 h -1 (upper values) and 0.12 h -1 (lower values) with a feed containing 10 g l -1 glucose and 10 g l -1 xylose. All fluxes are normalized to a total specific sugar consumption of 100 mmol g -1 biomass h -1 . Substances shown inside boxes are substrates or products measured with high performance liquid chromatography. Specific consumption (negative) and production (positive) rates, product yields and carbon balances in continuous fermentation of strains Y-PsNative and Y-PsK270R under anaerobic conditions at dilution rates of 0.06 h -1 and 0.12 h -1 with 10 g l -1 glucose and 10 g l -1 xylose. a (g xylitol (g consumed xylose) -1 ) been reported for strains TMB3400 and C1, generated by random mutagenesis and evolutionary engineering, respectively (Table 6) [39,40]. However, the exact mutations that caused the improvements are not known, and it is therefore difficult to transfer these traits. In addition, strains TMB3400 and C1 produce relatively high xylitol yields (Table 6). Effective ethanolic xylose fermentation in batch has been reported for XI strains, harbouring Piromyces XI on a multicopy plasmid [23,41]. When the Piromyces XI gene was chromosomally integrated in the strain background used for constructing Y-PsK270R, aerobic xylose growth could not be achieved [12]. In contrast, strain Y-PsK270R harbours all components required for effective xylose utilization chromosomally integrated. This ensures stable expression, and facilitates the transfer of these traits into an industrial strain.
In another fermentation study, the native Candida tenuis XR has been compared with its K274R-N276D double mutant in recombinant S. cerevisiae strains also expressing XDH from Galactocandida mastotermitis and overexpressing the endogenous XK gene [19]. In contrast to the current results, the double mutant C. tenuis XR did not improve the xylose uptake rate, suggesting that the high xylose utilization background of the strains used in the present study is required to make a full examination of a xylose utilization pathway.
The metabolic flux analysis indicated that the K270R XR utilizes a larger fraction of NADH compared with the native XR in vivo. This, in turn, made more NAD + available for the XDH reaction, and resulted in higher xylose consumption rate and lower xylitol production. Both strains carrying mutated XRs also gave lower glycerol yields in anaerobic batch fermentation compared with the reference strain. This is a further indication that NAD + is more available for the XDH reaction in these strains since glycerol formation is the main NAD + generation pathway in S. cerevisiae under anaerobic conditions [42][43][44].
The estimated kinetic parameters of the native P. stipitis XR correspond well to previously published data [45]. The K270M and K270R mutations affect the kinetic properties of the enzyme similarly to the corresponding mutations in C. tenuis XR [46]. According to the metabolic flux analysis, the K270R XR appears to use more NADH than NADPH in vivo, even though the K m value for NADPH is estimated to be around half of the K m value for NADH. This suggests that that the intracellular level of NADH is much higher than the intracellular level of NADPH. The K270M mutation affected both NADPH and NADH affinity in xylose reduction, in contrast to the glyceraldehyde reduction where the apparent affinity for NADH remained unchanged [47]. The K270M mutation reduced the xylose utilization rate, which agrees with previous observations [13]. Crude extract from Y-PsK270M displayed lower XR activity in standard assay conditions which indicated lower expression of the K270M XR. However, the kinetic study revealed that the K270M XR has a lower affinity for NAD(P)H and xylose. The K270M XR was far from saturated in the standard assay conditions, which explains the lower XR activities.

Conclusion
The cofactor preference of P. stipitis XR was altered by sitedirected mutagenesis. When the K270R XR was combined with a metabolic engineering strategy that ensures high xylose utilization capabilities, a recombinant S. cerevisiae strain was created that provides a unique combination of high xylose consumption rate, high ethanol yield and low xylitol yield during ethanolic xylose fermentation. Xylose consumption rate (g (g biomass) -1 h -1 ), ethanol yield (g (g consumed sugars) -1 ) and xylitol yield (g (g consumed xylose) -1 ) in anaerobic continuous fermentation with recombinant S. cerevisiae strains, 10 g l -1 glucose and 10 g l -1 xylose. a Recalculated