Growth and fermentation of D-xylose by Saccharomyces cerevisiae expressing a novel D-xylose isomerase originating from the bacterium Prevotella ruminicola TC2-24

Cloning and expression of bacterial D-xylose isomerases in Saccharomyces cerevisiae

Prevotella spp. and Bacteroides spp. are among the most common xylan-degrading microorganisms isolated from the bovine rumen and human colon, representing a dominant phylum (i.e. Bacteroidetes) in these ecosystems. Bacterial D-xylose isomerases have been reported with much higher affinity for D-xylose when compared to the fungal Piromyces sp. E2 or Orpinomyces D-xylose isomerases. With the goal of identifying D-xylose isomerases with increased affinity for D-xylose, D-xylose isomerase genes from Bacteroides and Prevotella spp. were isolated and expressed in S. cerevisiae. D-xylose isomerase genes were isolated from three Bacteroides spp. (B. uniformis, B. distasonis, and B. ovatus) and P. ruminicola strain TC2-24. Attempts to isolate D-xylose isomerase genes from other P. ruminicola strains failed (see methods). Several Prevotella spp., including the sequenced type strain P. ruminicola 23, have been reported to be missing the gene for D-xylose isomerase. When grown on D-xylose, these strains have detectable but low D-xylose isomerase activity and high D-xylulokinase activity [18, 19]. Yet, they lack an obvious D-xylose reductase and xylitol dehydrogenase pathway. It is unclear whether the absence of a D-xylose isomerase gene is due to an incomplete genome sequence or if they possess an alternate mechanism to convert D-xylose to D-xylulose. P. ruminicola strain TC2-24 shows many similarities with P. ruminicola 23 but does exhibit differences when compared to this type strain [20, 21]. Although we have not investigated further, the presence of a D-xylose isomerase in TC2-24 may be another characteristic of this group of “P. ruminicola 23-like” strains. The D-xylose isomerase identified from P. ruminicola TC2-24 was 439 amino acids long and 79% to 85% identical to D-xylose isomerases from other Prevotella spp. The TC2-24 XI was 79% identical to the Piromyces sp. E2 and Orpinomyces XIs and only 53% identical to the C. phytofermentans XI. The Bacterodies spp. XIs analyzed in this study were 81% to 83% identical to the fungal isomerases.

Aerobic growth in D-xylose medium

To determine if the D-xylose isomerase genes were functional, each gene was expressed in S. cerevisiae along with the S. cerevisiae D-xylulokinase gene, XKS1, and screened for growth on D-xylose in aerobic liquid culture (Figure 1A). The control strain YRH561, which did not express D-xylose metabolism genes, did not grow. Strain YRH562, expressing the Piromyces sp. E2 XI (and S. cerevisiae XKS1), was included for comparison. YRH562 had a specific growth rate of 0.07 h^-1 on D-xylose medium under aerobic growth conditions (Table 1). Several of the bacterial isomerases conveyed the ability for growth on D-xylose at specific growth rates similar to YRH562. As expected, specific activity of a D-xylose isomerase in the S. cerevisiae strain was correlated with the growth of that strain on D-xylose (data not shown). For example, the Bacteroides ovatus XI expressed in strain YRH565 had very low activity and this strain also grew poorly in D-xylose medium (Figure 1A). Equilibrium kinetics of the D-xylose isomerase reaction do not favor production of D-xylulose and co-expression with a D-xylulokinase has been recommend by prior studies to help pull D-xylose into the pathway. As evidence, strains not over expressing D-xylulokinase (YRH628 in Figure 1A) grew slower compared to strains expressing elevated levels of D-xylulokinase. Endogenous XK expression in YRH628 allowed for some growth on D-xylose.

Strain YRH587, expressing the XI from Prevotella ruminicola (TC2-24) and ScXKS1 grew as well as strain YRH562 expressing the Piromyces XI and ScXKS1. Specific growth rates for the two strains were also comparable (Table 1) indicating that the P. ruminicola XI was functional when expressed in S. cerevisiae. Codon optimization of the P. ruminicola XI (YRH629) did not enhance growth relative to YRH587. Additionally, the P. ruminicola D-xylulokinase was substituted for the ScXKS1 gene. YRH631 cells expressing the bacterial XK gene had a similar growth rate as the YRH629 strain expressing the S. cerevisiae XK, indicating that the P. ruminicola XK was also functionally expressed. Strain YRH630, expressing the Piromyces XI gene and the P. ruminicola XK (Figure 1A), grew as well as strain YRH562 expressing the Piromyces XI gene with the S. cerevisiae XK. Cells expressing the B. uniformis XK with the B. uniformis XK reached a higher cell density with the B. uniformis XK compared to the strain expressing the S. cerevisiae XK (YRH592 vs. YRH563). These results demonstrate that the B. uniformis and P. ruminicola XKs are active in S. cerevisiae.

Next, the strains were evaluated for growth under oxygen-limited conditions using the Bioscreen C™, a microtiter based system for measuring microbial growth rates. Aerobic growth studies were performed using a 6-edged flower shaped microtiter plate in the BioLector® under conditions that were optimized for high oxygen transfer rate [22]. Round microtiter plates such as those used in the Bioscreen C™ have been shown to provide lower oxygen transfer rates [22], and decreased oxygen levels are a likely cause of the poor cell growth seen when the Bioscreen C™ is used to culture cells on respiratory carbon sources like ethanol [23]. All of the strains grew poorly when cultured using the Bioscreen C™. The best strains reached a cell density of less than 0.8 OD₆₆₀ in 96 hours (Figure 1B). Such poor growth under these conditions suggested that adaptation to an oxygen-limited environment would be beneficial.

Adaptation for improved D-xylose fermentation

Adaptation, either by serial batch or continuous cell culture, has been successful for improving growth and fermentation of S. cerevisiae strains expressing different D-xylose isomerase genes [6, 14] as well as for cells expressing the reductase/dehydrogenase genes for D-xylose metabolism [7]. We used serial passage of cells to select for spontaneous changes that resulted in increased growth under fermentative conditions (microaerobic). YRH631 was grown in YPX and transferred weekly for a total of six transfers. Residual D-xylose and the fermentation products ethanol and xylitol were measured for each culture (Figure 2). Steady strain improvement was observed as greater D-xylose utilization and increased production of ethanol (Figure 2). Acetate (not shown) remained constant at 2 g/L.

The adapted strain YRH1114 was isolated from the last serial culture and subsequently evaluated for growth in D-xylose medium under aerobic and microaerobic conditions (Figure 3A, B and Table 1). Strains expressing the Piromyces XI (YRH562) and the unadapted strain expressing the P. ruminicola XI (YRH631) were included for comparison. The adapted strain, YRH1114, showed a significant increase in growth under both culture conditions compare to the unadapted strain. The specific growth rate for YRH1114 was 0.23 h^-1, an increase of 3.8-fold compared to the unadapted strain, is among the highest reported growth rates which range from 0.01 h^-1 to 0.22 h^-1 for a S. cerevisiae strain expressing D-xylose isomerase [6, 13, 14, 16, 17, 24–26].

D-xylose fermentation

The adapted XI yeast strain (YRH1114) was also evaluated for its ability to ferment D-xylose to ethanol using microaerobic conditions. For comparison, the unadapted yeast strain expressing the P. ruminicola XI and XK was also included in the experiment. Finally, YRH400, a D-xylose fermenting yeast that expresses the alternate D-xylose utilization genes (S. stipitis XR and XD, with elevated ScXKS1) in a commercial yeast background was also included; this strain is currently our best performing strain for D-xylose fermentation [27]. Progress of the fermentations was monitored by measuring production of CO₂ (Figure 4): one mole of CO₂ is produced for each mole of ethanol. YRH1114 outperformed these strains producing 13.6 g/L ethanol with a metabolic yield of 82.9% of theoretical (Table 2). The unadapated XI yeast strain YRH631 had a much lower ethanol titer (4.1 g/L) even though the metabolic yield was closer to that of the adapted strain YRH1114. The difference in final ethanol concentration between the adapted and unadapted strain arose primarily from differences in D-xylose consumption rate and specific ethanol productivity. The adapted strain had a significantly higher ethanol productivity compared to the unadapted strain also expressing the P. ruminicola XI.

The most notable comparison is between the adapted YRH1114 strain and YRH400. Even though YRH400 had likewise been adapted and possessed an industrial yeast background, YRH1114 produced nearly twice as much ethanol (Table 2). This difference is also apparent in lower specific ethanol productivity for YRH400 compared to YRH1114. While YRH400 consumed nearly as much D-xylose, it funneled approximately 50% of the D-xylose to the production of xylitol, which is approximately three times as much as the XI expressing strains. This difference in xylitol production is reflected in the much lower metabolic ethanol yield for YRH400 compared to the other strains. Most strains using the XR/XD pathway suffer from low ethanol yield, although an adapted strain using this pathway was recently reported to have a 79% of theoretical ethanol yield from D-xylose [28].

Selective ethanol production is a hallmark of XI expressing strains because the isomerase step avoids the redox imbalance observed for expression of the genes from S. stipitis. This redox imbalance arises because each enzyme (see introduction) favors an alternative electron carrier. This imbalance has been partially relieved by redox engineering of the XR and XD enzymes [29–32]. Increases in ethanol yield during D-xylose fermentation are seen for some redox engineered strains, but ethanol yields from D-xylose for these strains are still 10% to 19% lower than the yield obtained with the adapted strain described in this study. Thus, the adapted strain expressing the P. ruminicola XI and XK compares favorably to results reported for other strains when cultured on D-xylose. It is also likely the performance of the P. ruminicola XI can be further improved by expressing the gene in a more robust yeast strain such as D5A, which was used for constructing YRH400 [27].

Kinetic parameters

As discussed previously, one concern with XI-based systems is poor affinity of the enzyme for D-xylose. S. cerevisiae does not possess a native D-xylose transporter and intracellular D-xylose concentrations might be rate limiting. Therefore, it was of interest to measure the Michaelis-Menten kinetic constants for the P. ruminicola XI in comparison to other D-xylose isomerases. Enzyme kinetic parameters for the P. ruminicola XI were measured for the pre and post adapted strains.

The measured V _max for the unadapted strain was 0.28 μmole/min/mg protein (Table 3) and was not significantly different from Piromyces sp. E2 XI, which was measured at 0.25 μmole/min/mg protein when expressed in our strain. Prior studies have reported V _max for the Piromyces XI at 0.05 μmole/min/mg protein (Table 3). Higher enzyme activity than 0.05 μmole/min/mg for the Piromyces sp E2 XI has been observed, but V _max was not reported in these studies [12, 25, 26]. Following strain adaptation, V _max for the P. ruminicola XI increased 2.9-fold from 0.28 to 0.81 μmole/min/mg protein. This increase probably reflects a higher XI protein concentration in the cell lysate, either from increased expression or stability.

The increase in XI activity for the adapted strain suggested that the copy number of the XI gene may have increased leading to higher protein expression. Gene amplification is a common mechanism for adaption in S. cerevisiae[33, 34] and has been shown to occur under a wide range of circumstances including evolutionary engineering for increased D-xylose metabolism. For instance, one study reported an increase in XI copy number due to integration of up to 32 copies of the Piromyces sp. E2 XI gene [6]. To determine if the P. ruminicola XI gene had integrated into the genome in our adapted strain, we allowed the strain to lose the plasmid containing the XI gene. The strain lacking the XI plasmid was no longer able to grow on D-xylose medium (data not shown), suggesting that integration of the XI gene into the genome did not occur, at least not at levels high enough to support growth on D-xylose in the absence of the XI expression vector. The plasmid containing the XI gene was also rescued from the adapted strain and sequenced. No mutation was found in the XI open reading frame or the promoter or terminator for the gene, indicating that the causative mutation for increased ability to ferment D-xylose resides in the genome. Consistent with this hypothesis, replacing the XI vector in the adapted strain with the original parent XI vector restored the increased growth rate on D-xylose, similar to the original adapted strain (data not shown) while putting the same vector into the unadapted CEN.PK2-1C parent did not. Further analysis of the genome sequence of the adapted strain compared to the parent should reveal the genomic changes responsible for increased D-xylose fermentation.

Strains, media, and general methods

Cloning of D-xylose isomerase genes for expression in S. cerevisiae

Since a gene for D-xylose isomerase had not been found in P. ruminicola, degenerate primers #243 and #244 were used to amplify an isomerase gene from P. ruminicola strains. Primer #243 was 32-fold degenerate and #244 was 64-fold degenerate. These primers were designed based on Bacteroides spp., and Piromyces sp. E2. Using these primers and a touch-down PCR cycle, we were unable to successfully amplify a D-xylose isomerase gene from any of the P. ruminicola strains tested. A DNA fragment containing a D-xylose isomerase gene was isolated when primers based on the XI from sequence from Prevotella bergensis (accession # ACKS01000045.1) were used (#310 and #311). The only P. ruminicola strain to yield a D-xylose isomerase gene by PCR amplification was strain TC2-24. The P. ruminicola gene from TC2-24 (accession # KC847096) was cloned into pRH167 to create pRH367 for expression of the unoptimized gene in S. cerevisiae.

Codon-optimized genes were obtained from DNA2.0 (Menlo Park, CA,USA) and cloned into vectors for expression in S. cerevisiae. The optimized P. ruminicola (TC2-24) D-xylose isomerase gene was cloned into the high-copy vector pRH167 using the restriction endonucleases Spe I and Sal I, creating pRH384. To create pRH385 the optimized P. ruminicola (T23) D-xylulokinase was cloned into the low-copy vector pRH164 using the same restriction endonucleases. Each vector contained the truncated HXT7 promoter for constitutive expression of the heterologous gene.

Aerobic growth kinetics

Yeast pre-cultures were grown to mid-log phase in synthetic complete medium with 20 g/L D-glucose and washed with sterile water prior to inoculation. YP medium supplemented with 50 g/L D-xylose (YP5X) was used to determine each yeast strain’s ability to assimilate D-xylose aerobically. Cultures were started using 6-edged flower shaped microtiter plates in 800 μl of YP5X at an OD₆₆₀ of 0.05 and incubated at 30°C, mixing at 1000 rpm using a BioLector® (m2p-labs; Baesweiler, Germany). The BioLector® measures cell density by scattered light, which was measured every 30 minutes. A gain of 20 was used for the experiments to avoid saturation at high optical densities. At this gain, scattered light units were linearly proportional to OD₆₆₀. Scattered light units were converted to optical density using the conversion factor (OD₆₆₀ = background subtracted scattered light unit/22.93). Data shown represent the mean values from experiments that were repeated in triplicate. Standard deviation values were less than 10% of the mean. Specific growth rates for each replicate were determined by linear regression of the natural log transformed data (i.e. background subtracted scattered light units) vs. time. The slope of the fitted line during exponential growth was used as the specific growth rate μ (h^-1).

Cell culture using the Bioscreen C™ automated microbiology growth curve analysis system (Growth Curves USA; Piscataway, NJ, USA) was performed using 100 well honeycomb plates. Microtiter plates for the Bioscreen C™ have a round geometry unlike the flower-shaped plates used in the BioLector®. Plates were shaken with a high amplitude setting and normal speed setting for 30 seconds with an interval of 60 seconds. OD₆₆₀ was determined by converting the wideband OD values from the Bioscreen C™ using the equation (OD₆₆₀ = 1.74OD + 0.535(OD)² - 0.749(OD)³ + 0.675(OD)⁴, R² = 0.999). Data shown represent the mean values from triplicate experiments.

Strain adaptation

Strain adaptation to aerobic conditions was achieved by serial passage of cells in YP medium supplemented with 50 g/L of D-xylose as the only energy source. Serial passage under aerobic conditions was continued for eight weeks. Strain adaptation for D-xylose fermentation was then performed in duplicate by weekly transfer of cells grown under fermentative conditions (microaerobic) for six weeks. During adaptation to fermentative conditions cultures were sampled at each passage and analyzed for optical density, residual D-xylose, and fermentation products (ethanol, glycerol, xylitol, and acetate).

Enzyme assays

Total protein from clarified cell lysates was prepared from mid-log phase cells grown in synthetic complete medium using D-glucose as the carbon source. Cells were collected by centrifugation, washed once with sterile ice-cold water, and centrifuged again. Cell pellets were stored at −80°C for later use. Cells were resuspended in an appropriate amount of Y-PER reagent (Pierce; Rockford, IL, USA) plus protease inhibitors (Complete, mini, EDTA-free protease inhibitor cocktail, Roche; Indianapolis, IN, USA) and processed according to the manufacturer’s instructions. Protein concentrations were determined with the Quick Start Bradford Protein Assay (Bio-Rad; Hercules, CA, USA) against a bovine serum albumin standard.

D-xylose isomerase activity was assayed essentially as described in [12]. XI assays were performed using buffer containing 100 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 0.15 mM NADH, 2 U sorbitol dehydrogenase (Roche, Mannheim, Germany), and an appropriate amount of cell lysate. Reactions were started by the addition of D-xylose (Sigma; St. Louis, MO, USA) to a final concentration of 500 mM and reactions monitored at 340 nm using a Cary 50 Bio UV-Visible spectrophotometer (Varian; Palo Alto, CA, USA). Specific activity (μmole/min/mg lysate) was determined using the molar absorption coefficient, ϵ₃₄₀, of 6.22 mM^-1 cm^-1 for NADH. To ensure that the sorbitol dehydrogenase coupling enzyme was not saturated, the ratio of sorbitol dehydrogenase activity to D-xylose isomerase activity was maintained at > 20 for the assay. Additionally, specific activities reported were proportional to the amount of lysate added in a dilution series. Kinetic parameters were determined by varying the D-xylose concentration from 5 to 500 mM. Each assay was done using lysates prepared from three independent cultures.

Batch fermentation

D-xylose fermentation was investigated by inoculating 100-ml YP cultures with 50 g/L D-xylose at a starting OD₆₆₀ of 2.0. CO₂ production was monitored continuously over the course of the fermentation using a wireless gas production measurement system (Ankom Technologies; Macedon, NY, USA). The wireless system monitors gas production indirectly by measuring cumulative gas pressure; CO₂ production is calculated using the ideal gas law. The system was set to vent when the overhead pressure achieved 1 psi and to monitor the accumulated pressure every 15 min. Exponentially growing cells were used for the inoculum. Post fermentation samples were used to determine cell biomass (by OD₆₆₀), residual sugars, and fermentation products (by high-performance liquid chromatography, HPLC). All fermentation experiments were performed three to four times. Accumulative pressure values were used to calculate the amount of CO₂ produced during the fermentation. Carbon recoveries were determined using HPLC data. Carbon recovery calculations were based on total input carbon from D-xylose and measured (HPLC) remaining D-xylose, along with fermentation products. CO₂ amounts used for carbon recovery calculations assumed 1 mole of CO₂ is produced for every 1 mole of ethanol. Specific ethanol productivity (g ethanol/g cell dry weight/h) was determined using the final OD₆₆₀ measurement (91 hours) for each batch fermentation. Cell dry weight (CDW) was calculated using an OD-to-CDW conversion factor for the yeast strain CEN.PK2-1C (CDW = 0.58 g/L/OD). The conversion factor was determined by drying cells at differing OD to constant weight at 100°C. Cells were washed three times with distilled water prior to drying to remove trace amounts of medium. OD was measured using a GeneSys 10 vis spectrophotomerer (Thermo Fisher Scientific Inc.; Waltham, MA, USA).

Analytical methods

Extracellular metabolites were measured using HPLC as previously described [27]. Samples were analyzed using a SpectraSYSTEM liquid chromatography system (Thermo Electron Corporation, CA, USA) equipped with an automatic sampler, column heater, isocratic pump, refractive index detector, and computer based integrator running Chromquest ver. 2.5 software (Thermo Electron Corporation). Samples were injected (20 μl) onto a sugar column (Aminex HPX-87H Column, 300 x 7.8 mm, Bio Rad Laboratories, Inc.) and eluted with 5 mM sulfuric acid at 0.6 ml/min and 65°C.

Statistical analyses

For experiments with three or greater biological replicates, probability analyses were performed using the Student’s t-test with a two-tailed distribution and compared to the appropriate control strain. Values with p < 0.05 were considered significant for this study. Statistical analysis was performed using Microsoft Excel.