Comparison of the xylose reductase-xylitol dehydrogenase and the xylose isomerase pathways for xylose fermentation by recombinant Saccharomyces cerevisiae

Background Two heterologous pathways have been used to construct recombinant xylose-fermenting Saccharomyces cerevisiae strains: i) the xylose reductase (XR) and xylitol dehydrogenase (XDH) pathway and ii) the xylose isomerase (XI) pathway. In the present study, the Pichia stipitis XR-XDH pathway and the Piromyces XI pathway were compared in an isogenic strain background, using a laboratory host strain with genetic modifications known to improve xylose fermentation (overexpressed xylulokinase, overexpressed non-oxidative pentose phosphate pathway and deletion of the aldose reductase gene GRE3). The two isogenic strains and the industrial xylose-fermenting strain TMB 3400 were studied regarding their xylose fermentation capacity in defined mineral medium and in undetoxified lignocellulosic hydrolysate. Results In defined mineral medium, the xylose consumption rate, the specific ethanol productivity, and the final ethanol concentration were significantly higher in the XR- and XDH-carrying strain, whereas the highest ethanol yield was achieved with the strain carrying XI. While the laboratory strains only fermented a minor fraction of glucose in the undetoxified lignocellulose hydrolysate, the industrial strain TMB 3400 fermented nearly all the sugar available. Xylitol was formed by the XR-XDH-carrying strains only in mineral medium, whereas in lignocellulose hydrolysate no xylitol formation was detected. Conclusion Despite by-product formation, the XR-XDH xylose utilization pathway resulted in faster ethanol production than using the best presently reported XI pathway in the strain background investigated. The need for robust industrial yeast strains for fermentation of undetoxified spruce hydrolysates was also confirmed.


Background
Saccharomyces cerevisiae is a promising candidate for industrial bioethanol production due to its robustness, inhibitor tolerance and high ethanol productivity [1,2].
[26] resulted in the highest activity. However, the sole expression of Piromyces XI did not result in appreciable growth on xylose [26]. Only when adaptation [27] or extensive genetic engineering [28] was also applied, were growth and fermentation on xylose achieved.
Genetic modifications other than the sole introduction of initial xylose utilization pathway are needed for efficient xylose metabolism. The combination of overexpressed XK, overexpressed non-oxidative pentose phosphate path-way (PPP) and deletion of the endogenous aldose reductase gene GRE3 have been shown to enhance both aerobic and anaerobic xylose utilization in XR-XDH-as well as XIcarrying strains [15,28,29]. The overexpression of XK is necessary to overcome the naturally low expression level of this enzyme [6,8,30,31], and the overexpression of the PPP enzymes enables efficient incorporation of xylulose-5-phosphate into the central metabolism [32]. Furthermore, the gene GRE3 codes for an unspecific reductase that functions as an NADPH-dependent xylose reductase [33], and contributes to xylitol formation [34,35] with concomitant inhibition of XI activity [36].
To the best of our knowledge, no comparative assessment has been reported for a pair of isogenic strains with the two functional xylose utilization pathways, XR and XDH from P. stipitis, or XI from Piromyces sp. In the present study, an engineered strain with overexpressed XK, overexpressed non-oxidative PPP and GRE3 deletion in CEN.PK background was used [29] to generate two isogenic strains that carried either the XR-XDH or the XI pathway. The xylose fermentation performance of the two strains was compared.
Recombinant xylose-fermenting strains are aimed to be used in industrial ethanol production from lignocellulose hydrolysates. The hydrolysates contain large amounts of inhibitory compounds such as furfural, hydroxymethyl furfural, acetate and phenolics, the concentrations of which depend on the raw material and hydrolysis method [37-39]. As industrial yeast strains are generally more tolerant to such inhibitors than laboratory strains [40,41], the two isogenic strains were also compared with the industrial xylose-fermenting strain TMB 3400 [42] with respect to hexose and pentose fermentation in an undetoxified lignocellulose hydrolysate.

Strain construction
To generate an isogenic pair of strains, a strain with highlevel expression of XI was constructed similar to the previously generated strain TMB 3057 with high-level expression of XR and XDH [29] (  Figure 2). Anaerobic growth was not measured. XI activity in strain TMB 3066 was 0.82 ± 0.01 U/mg protein.

Anaerobic xylose fermentation in mineral medium
Xylose fermentation by strains TMB 3057 and TMB 3066 was compared in anaerobic batch culture. The industrial strain TMB 3400 with chromosomally integrated XR-XDH genes [42] was also included in the comparison. Buffered mineral medium containing xylose (50 g/l) and an initial biomass concentration of 5 g dry weight/l were used. Representative fermentation profiles for the three strains are shown in Figure 3.
The XR-XDH-carrying strain TMB 3057 consumed xylose at a rate of 0.13 g xylose g cells -1 h -1 (Table 2), with ethanol, xylitol and glycerol as fermentation products ( Figure  3A). The isogenic strain TMB 3066 carrying XI consumed xylose at a 2.5 fold lower rate of 0.05 g xylose g cells -1 h -1 ( Table 2). Ethanol was the major fermentation product of TMB 3066 ( Figure 3B), with a yield of 0.43 g g consumed xylose -1 , which is close to the theoretical yield of 0.51 g g consumed xylose -1 ( Table 2). Despite the formation of the by-products, TMB 3057 produced twice as much ethanol (13.3 g/l) within 50 hours as a result of the higher xylose consumption rate. The specific ethanol productivity was 2-fold higher in the XR-XDH-carrying strain TMB 3057 than in the XI-carrying strain TMB 3066 ( Table 2). The industrial strain TMB 3400 consumed xylose more slowly than the laboratory strain TMB 3057 (Table 2) ( Figure  3C).

Fermentation of lignocellulose hydrolysate
The three strains studied were also compared in batch fermentation of an undetoxified spruce hydrolysate [44]. The initial concentrations of the hydrolysate components were as follows: glucose, 16 g/l; mannose, 10 g/l; galactose, 4 g/l; xylose, 7 g/l; arabinose, 3 g/l; and acetic acid, 2.5 g/l. Fermentation was started with an initial biomass concentration of 5-10 g dry weight/l and pH 5.5. Representative fermentation profiles are shown in Figure 4, and the fermentation parameters are summarized in Table 3.
Only a small amount of glucose was consumed by the two laboratory strains ( Figures 4A, 4B), whereas nearly all glucose, mannose and xylose were consumed by the industrial strain TMB 3400 ( Figure 4C). The final ethanol concentration was about 3 g/l for TMB 3057 (XR-XDH), 2 g/l for TMB 3066 (XI), and 16 g/l for the strain TMB 3400 (the XR-XDH-carrying industrial strain) ( Figure 4). No xylitol was produced by TMB 3400, although nearly all the xylose was consumed within 100 h ( Figure 4C).

Discussion
Isogenic strains were used to compare the efficiency of xylose fermentation via the P. stipitis XR and XDH route and the Piromyces XI route. Comparing the two pathways in isogenic strains eliminates the influence of strain background on xylose uptake rate and product formation. In anaerobic batch fermentation of xylose in mineral medium, the xylose consumption rate was higher in the XR-XDH-carrying strain TMB 3057 than in the XI-carrying TMB 3066 in the strain background used. However, the XI-carrying strain TMB 3066 produced ethanol at almost theoretical yield. Nevertheless, TMB 3057 produced twice as much ethanol due to faster and almost complete xylose consumption. This is in accordance with the 8-fold higher aerobic growth rate that was observed on xylose for the XR-XDH-carrying strain ( Figure 2).
The XI-carrying strain TMB 3066 reported in the present study displayed slower aerobic growth on xylose and slower anaerobic xylose fermentation than the previously designed strain RWB217 [28] ( Because the design of xylose-fermenting strains for bioethanol production aims at fermentation of industrial substrates, the three strains TMB 3057 (XR-XDH), TMB 3066 (XI) and TMB 3400 (integrated XR-XDH) were also compared regarding the fermentation of an undetoxified spruce hydrolysate. Laboratory strains are considered to    from most of the sugar present. A slightly higher amount of glucose was consumed by the XR-XDH-carrying laboratory strain TMB 3057 than by the XI-carrying laboratory strain TMB 3066, which may reflect the possible function of the heterologous XR in stress response, similarly to the native aldose reductases in S. cerevisiae [35,58].
Nearly all the xylose was consumed by TMB 3400 in the hydrolysate, with ethanol as the main product. No xylitol was produced. The absence of xylitol production in undetoxified lignocellulose hydrolysate is in agreement with previous studies, in which the presence of external electron acceptors such as furfural reduced xylitol formation [59-61]. Absence of xylitol formation has frequently been observed in lignocellulose fermentation with XR-and XDH-carrying strains [48,49,[62][63][64]. Indeed, in a comparison of different hydrolysates for fermentability by an XR-XDH-carrying strain, xylitol formation was only detected in a corn fiber hydrolysate [49], which is known to contain less furfural than softwood hydrolysates [65].

Conclusion
When comparing two isogenic strains with high-level expression of the XR-XDH pathway or the XI pathway, higher specific ethanol productivity was observed with the XR-XDH pathway as a result of significantly higher xylose consumption rate in the strain background investigated. On the other hand, the ethanol yield was higher in the XI carrying strain. The CEN.PK-based laboratory strains did not ferment the lignocellulose hydrolysate used in this study, demonstrating that robust industrial strains are essential for the fermentation of industrial substrates. Xylitol formation, which has been often described as being the major drawback of the XR-XDH strategy, was not observed in the fermentation of lignocellulosic hydrolysate by TMB 3400.

Strains and media
Plasmids and yeast strains used in this study are summarized in Table 1. Escherichia coli strain DH5α (Life Technologies, Rockville, MD, USA) was used for cloning. E. coli was grown in LB medium [66] supplemented with ampicillin (100 mg/l). Liquid cultures of S. cerevisiae were grown in yeast nitrogen base medium (YNB) (Difco YNB without amino acids (6.7 g/l), Becton, Dickinson and Company, Sparks, MD, USA) supplemented with glucose (20 g/l) or xylose (50 g/l) (Acros Organics, NJ, USA) and buffered at pH 5.5 with potassium hydrogen pththalate (10.2 g/l) [37]. YPD or YNB plates were used for plate cultures. When required, the medium was supplemented with uracil (40 μg/ml).
Lignocellulose hydrolysate generated by two-step dilute acid hydrolysis of spruce [44] was used. The hydrolysate was stored at 4°C at pH 1.5. Before usage, the pH was adjusted to 5.5 and the hydrolysate was sterile filtered.

Molecular biology methods and cloning
Standard molecular biology techniques were used [66]. The calcium chloride method was used for transformation of E. coli [67] and the lithium acetate method was used for transformation of S. cerevisiae [68]. Taq polymerase, restriction enzymes and T4-nucleotide ligase were purchased from Fermentas (Vilnius, Lithuania). DNA sequencing was performed with an Abi-Prism BigDye cycle sequencing kit (Applied Biosystems, Weiterstadt, Germany).

Growth experiments
Pre-cultures were cultivated until the late exponential phase in YNB medium with glucose. Cells were inoculated at an initial OD 620nm of 0.2 in YNB medium with xylose (50 g/l), and grown in shake flasks at 30°C with 200 rpm agitation (INR-200 shake incubator, Gallenkamp, Leicester, UK). Experiments were repeated at least twice.

Batch fermentation
Pre-cultures were grown until the late exponential growth phase (OD 620nm 20) in YNB medium with xylose (50 g/l). Cells were harvested by centrifugation for 3 minutes at 4°C, after which the cell pellets were mixed with the fermentation medium. Anaerobic batch fermentation was performed at 30°C in 25 ml serum flasks containing 23 ml medium. The flasks were sealed with rubber stoppers into which cannulas were inserted for sampling and the release of exhaust gas, and were stirred with small magnetic stirrers at 150 rpm. A layer of mineral oil was added to ensure anaerobiosis. Samples were withdrawn during 100 hours of fermentation. Experiments were performed at least in duplicate.
For fermentation in mineral medium, YNB medium containing xylose (50 g/l) was used, complemented with potassium hydrogen phthalate buffer (pH 5.5) [37], Tween 80 (0.4 g/l) and ergosterol (0.01 g/l). Lignocellulose hydrolysate was complemented with concentrated YNB, Tween 80 (0.4 g/l) and ergosterol (0.01 g/l). The total volume of the added medium components did not exceed 10% of the culture volume. Tween 80 and ergos-terol were added dissolved in ethanol, which resulted in an initial ethanol concentration of about 1 g/l, which was taken into account in all calculations. The initial cell concentration was 5 g dry weight/l for the xylose fermentation and 5-10 g dry weight/l for lignocellulose hydrolysate fermentation.

Analyses
Concentrations of xylose, xylitol, glycerol, acetic acid and ethanol from fermentation in mineral medium were analysed by HPLC (Waters, Milford, Massachusetts, USA) with an Aminex HPX-87H ion-exchange column (Bio-Rad, Hercules, USA). A mobile phase of 5 mM H 2 SO 4 was used at a flow rate 0.6 ml/min and the column temperature was 45°C. For xylose fermentation experiments, the ethanol concentration was corrected for evaporation with the help of degree of reduction balance calculation. In the case of hydrolysate fermentation, the concentrations of glucose, xylose, galactose, arabinose, glycerol, xylitol and arabitol were analysed with HPLC (Waters, Milford, Massachusetts, USA) using an HPX-87P (Bio-Rad, Hercules, USA) ion exchange column at 85°C with water as the mobile phase at a flow rate of 0.5 ml/min. Glucose, galactose, mannose, glycerol, acetic acid and ethanol were analysed by HPLC (Beckman Instruments, Fullerton, CA, USA) using three Aminex HPX-87H ion exchange columns (Bio-Rad, Hercules, USA) connected in series, to allow separation of the peaks. A temperature of 45°C and a mobile phase of 5 mM H 2 SO 4 at a flow rate 0.6 ml/min were used.
Cell dry weight was determined in triplicate by filtering 1 ml of the culture with a pre-weighed nitrocellulose filter with 0.45 μm pores. Filters were dried in a microwave oven and weighed.