ABSTRACT
Thermoanaerobacter spp. have long been considered suitable Clostridium thermocellum coculture partners for improving lignocellulosic biofuel production through consolidated bioprocessing. However, studies using “omic”-based profiling to better understand carbon utilization and biofuel producing pathways have been limited to only a few strains thus far. To better characterize carbon and electron flux pathways in the recently isolated, xylanolytic strain, Thermoanaerobacter thermohydrosulfuricus WC1, label-free quantitative proteomic analyses were combined with metabolic profiling. SWATH-MS proteomic analysis quantified 832 proteins in each of six proteomes isolated from mid-exponential-phase cells grown on xylose, cellobiose, or a mixture of both. Despite encoding genes consistent with a carbon catabolite repression network observed in other Gram-positive organisms, simultaneous consumption of both substrates was observed. Lactate was the major end product of fermentation under all conditions despite the high expression of gene products involved with ethanol and/or acetate synthesis, suggesting that carbon flux in this strain may be controlled via metabolite-based (allosteric) regulation or is constrained by metabolic bottlenecks. Cross-species “omic” comparative analyses confirmed similar expression patterns for end-product-forming gene products across diverse Thermoanaerobacter spp. It also identified differences in cofactor metabolism, which potentially contribute to differences in end-product distribution patterns between the strains analyzed. The analyses presented here improve our understanding of T. thermohydrosulfuricus WC1 metabolism and identify important physiological limitations to be addressed in its development as a biotechnologically relevant strain in ethanologenic designer cocultures through consolidated bioprocessing.
INTRODUCTION
The use of designer cocultures is a strategy that is receiving increased attention for its effectiveness at achieving improved biofuel yields and conversion efficiencies from lignocellulosic biomass through consolidated bioprocessing (CBP) (1–3). In a CBP platform, which involves concomitant enzyme production, biomass hydrolysis, and biofuel production (4), an ideal consortium would achieve (i) efficient and complete biomass hydrolysis, (ii) simultaneous, rather than sequential, utilization of cellulose, and hemicellulose constituent saccharides, and (iii) industrially relevant biofuel yields. The selection of microorganisms is therefore an important component in optimization of a successful designer consortium.
The extensive suite of lignocellulose-degrading enzymes encoded by Clostridium thermocellum has made it an attractive candidate for CBP platforms (5–7). However, its inability to grow and produce biofuels from hemicellulose constituent saccharides, most notably pentoses (8, 9), has often provided a rationale for the identification and investigation of suitable coculture partners. Previous C. thermocellum cocultures with bacteria possessing more diverse substrate utilization capabilities have resulted in improved rates of biomass degradation and biofuel yield (1, 2, 10, 11).
Hydrolysis of both the cellulose and hemicellulose fractions in a CBP platform involving C. thermocellum would generate a pool of mixed sugars available for fermentation. Although the substrate utilization capabilities of the constituent coculture members may have the potential to utilize the resulting hydrolysis products, distinct preferences for certain carbon sources, at the exclusion of others, may also exist. This preferential utilization by many bacteria, known as carbon catabolite repression (CCR), has been well documented in Firmicutes (12, 13), including strains of interest for lignocellulosic biofuel production such as Clostridium cellulolyticum (14) and Thermoanaerobacterium saccharolyticum (15). In addition, strains of the genus Thermoanaerobacter have been shown to exhibit CCR under some mixed sugar conditions (16, 17), while showing no evidence of CCR under other conditions (18–20).
Thermoanaerobacter thermohydrosulfuricus WC1, a recently characterized isolate from woodchip compost (21, 22), can hydrolyze and grow on polymeric xylan, distinguishing it from most other Thermoanaerobacter spp. (23). Since C. thermocellum downregulates expression of its own xylanases when grown on xylan containing substrates (6) in comparison to growth on cellulose alone, the xylan-hydrolyzing ability of T. thermohydrosulfuricus WC1 suggests it may be an effective C. thermocellum coculture partner. The value of this phenotype in C. thermocellum-Caldicellulosiruptor bescii cocultures has recently been reported as increased rates of biomass hydrolysis were directly attributed to the xylanolytic capabilities of Caldicellulosiruptor bescii (2). Constructing cocultures whereby both members contribute to lignocellulose hydrolysis may be particularly valuable given that biomass hydrolysis is a major limitation toward achieving industrially viable lignocellulosic biofuels (24, 25).
T. thermohydrosulfuricus WC1 is also found in a divergent lineage (clade 3) within the genus Thermoanaerobacter (22, 23). Comparative genomic analyses with better-characterized Thermoanaerobacter strains, including those for which global gene expression data (i.e., transcriptomic) are available (11, 20), has identified differences in physiological potential relating to biomass hydrolysis, substrate utilization, energy conservation, and end-product synthesis (23). However, while ethanol has been reported to be a major end product of fermentation under certain conditions using T. thermohydrosulfuricus WC1, the molar yields reported are not as high as those observed for other Thermoanaerobacter spp. As such, metabolic and quantitative proteomic analyses of T. thermohydrosulfuricus WC1 grown on single or mixed sugars was undertaken to achieve two objectives.
First, C. thermocellum-mediated hydrolysis of cellulose is reported to yield primarily higher order cellodextrins with little concomitant glucose (5). Since the T. thermohydrosulfuricus WC1 genome encodes two distinct annotated cellobiose specific phosphotransferase system (PTS) type transporters, and a connection between PTS-mediated sugar transport and CCR is well established in Firmicutes (12, 26), it was hypothesized that cellobiose consumption may repress the utilization of other sugars. Given the relative abundance of xylose in many types of hemicelluloses (27) and that xylose (as well as other higher-order xylodextrins potentially generated through combined C. thermocellum and T. thermohydrosulfuricus WC1 hydrolytic activities) utilization is an important characteristic for C. thermocellum coculture partners, the ability of T. thermohydrosulfuricus WC1 to utilize xylose in the presence of excess cellobiose was investigated.
Second, multiple redundancies for genes putatively involved with carbon utilization and end-product synthesis exist within the T. thermohydrosulfuricus WC1 genome (23). To better characterize the routes of carbon and electron flux in this divergent Thermoanaerobacter strain relative to other Thermoanaerobacter spp. and to help identify potential targets for genetic engineering as a means to improve biofuel yields in the strain, proteomic analyses was performed under single and multiple substrate conditions.
MATERIALS AND METHODS
Bacteria and culture conditions.Independent glycerol stocks of T. thermohydrosulfuricus WC1 were revived for each experimental replicate performed in this study. Cultures were grown at 60°C in ATCC 1191 medium as previously described (21) with the following exceptions: (i) the concentration of yeast extract was reduced from 2 to 0.67 g/liter to minimize the potential for growth on yeast extract components and (ii) the gassing and degassing of butyl rubber-stoppered bottles to make the medium anoxic were shortened to three cycles (1 min, gassing; 3 min, degassing) with a final gas cycle of 100% nitrogen (28). Anoxic, filter-sterilized sugar solutions were added to the medium after autoclaving so that medium contained (final concentrations) 5 g of xylose/liter (33.3 mM xylose; 27.8 mM glucose equivalents), 5 g of cellobiose/liter (14.6 mM cellobiose; 27.8 mM glucose equivalents), or 5 g of xylose/liter plus 5 g of cellobiose/liter (sugar mix). An equal volume of anoxic, filter-sterilized water was used for the no-substrate conditions. T. thermohydrosulfuricus WC1 has been deposited in the Deutsche Sammlung von Mikroorganismen und Zellkulturen culture collection (DSM 26960).
Genome analysis.Analysis of the T. thermohydrosulfuricus WC1 annotated genome was performed as previously described (23) using the Joint Genome Institute's Integrated Microbial Genome's Expert-Review (IMG-ER) online tool (29). The genome is publicly available in GenBank under accession number AMYG00000000.
An ad hoc perl script was developed to identify putative catabolite responsive elements (cre) in the T. thermohydrosulfuricus WC1 genome using the reported degenerate consensus sequences from Bacillus subtilis (30), Lactobacillus casei (31), and Clostridium difficile (32) as queries.
Growth and metabolic analyses. (i) Growth.Revived glycerol stocks were serially passaged three times at 60°C on 5 g of xylose/liter every 24 h prior to inoculation of (10% [vol/vol]) 10-ml tubes containing 5 g of xylose/liter, 5 g of cellobiose/liter, 5 g of xylose/liter plus 5 g of cellobiose/liter or no substrate. The optical density at 600 nm (OD600) of inoculated cultures was routinely measured until an increase in the OD600 was no longer observed. The reported doubling times are averages of three independent experiments, with each experiment containing three biological replicates.
(ii) Metabolic analyses.Ten-milliliter cultures, prepared as described above, were grown to target optical densities based on the growth profiles observed. For all conditions (excluding the no-substrate control), three tubes per experiment were harvested immediately after inoculation (t = 0) or upon reaching a target OD600 of 0.10, 0.20, 0.40, 0.70, 0.80, or 0.85. Cultures were then analyzed for end-product and protein/biomass synthesis, as well as residual sugar concentrations.
Gas analysis (H2 and CO2) was conducted by injecting triplicate 1-ml samples per tube into a Varian 490 Micro-GC gas chromatograph (Agilent Technologies, Santa Clara, CA) using nitrogen as a carrier gas. Gas solubility was accounted for in determining concentrations by measuring barometric pressure, tube pressure, and temperature as described previously (33). In addition, the bicarbonate fraction was determined as part of the total CO2 calculation (34). Acetate, lactate, xylose, and cellobiose concentrations were determined via high-performance liquid chromatography (Waters Corp., Milford, MA) equipped with a refractive index detector (model 2414) and an ion exclusion column (Aminex HPX-87H; Bio-Rad Laboratories, Hercules, CA) using sulfuric acid (5 mM) as the mobile phase at a flow rate of 0.6 ml/min. Ethanol and protein/biomass measurement assays were performed as described previously (21).
Proteomic analyses. (i) Growth.Cultures (10 ml) were prepared and grown on xylose, cellobiose, or xylose plus cellobiose from two independent, glycerol stock cultures of T. thermohydrosulfuricus WC1 (designated culture A or culture B). Upon reaching a target OD600 of 0.40, representing the mid-exponential phase, cultures were centrifuged at 4,700 × g for 10 min, and the resulting pellet was washed three times in 600 μl of phosphate-buffered saline (NaCl, 8 g/liter; KCl, 0.2 g/liter; Na2HPO4, 1.44 g/liter; KH2PO4, 0.24 g/liter [pH 7.4]). Cell pellets derived from the same inoculum (culture A or culture B) and grown under the same conditions were combined from ten independent 10-ml cultures and treated as a single sample. Thus, six samples (three conditions × two glycerol stock cultures) were used for proteomic analysis. Samples were frozen at −80°C until protein extractions could be performed.
(ii) Protein extraction.A modified version of the filter aided sample preparation protocol (35) was used for protein extraction. Cell pellets were resuspended in 1 ml of lysis buffer (4% [wt/vol] sodium dodecyl sulfate, 100 mM Tris-HCl, 0.1 M dithiothreitol [pH 7.6]) and heated at 95°C for 5 min. The samples were continuously sonicated at an output of 3.5 for 2 min using a Branson Sonifier 450 (Branson Ultrasonics Corp., Danbury, CT) and subsequently centrifuged at 16,000 × g for 20 min. The resulting supernatant was transferred to an Amicon Ultra-15 10K filter device (Millipore, Billerica, MA) and washed three times in 12 ml of urea solution (8 M urea in 0.1 M Tris-HCl [pH 8.5]). Each wash step included centrifugation at 4,000 × g for at least 10 min until the final volume remaining in the filter tube was <1 ml. Then, 2 ml of iodoacetamide solution (50 mM iodoacetamide in urea solution) was added to the filter device, followed by incubation at room temperature in the dark for 20 min. After centrifugation, the filter membrane was washed twice more with an additional 12 ml of urea solution. A 50-μl aliquot was taken from the filter unit and analyzed by using a BCA protein assay kit (Pierce Chemical Co., Rockford, IL) to estimate the total protein content of the sample. The filter membrane was washed twice with 12 ml of 50 mM ammonium bicarbonate in water, and the remaining protein was trypsin digested for 18 h at room temperature (trypsin/protein ratio, 1 μg:100 μg). The filter unit was transferred to a new collection tube and spun at 4,000 × g for 10 min, and the filtrate was retained for downstream analysis. The membrane was washed with 1 ml of 0.5 M NaCl, and the resulting filtrate was combined with the corresponding previous filtrate and stored at −80°C.
(iii) Peptide purification and MS analysis.Peptide concentrations in the combined filtrate were measured using a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE). Acidified aliquots of combined filtrate containing ∼150 μg of the digest were purified by loading onto a 1-by-100-mm C18 column (5-μm Luna C18[2]; Phenomenex, Torrance, CA) and eluted using 80% (vol/vol) acetonitrile. Purified aliquots were lyophilized and redissolved in buffer A (0.1% formic acid in water) for subsequent liquid chromatography-mass spectrometry (LC-MS) analysis.
A splitless nano-flow 2D LC Ultra system (Eksigent, Dublin, CA) with a 10-μl sample injection via a PepMap100 trap-column (300 μm by 5 mm; Thermo Fisher Scientific, Rockford, IL) and an analytical column (100 μm by 200 mm) packed with 5-μm Luna C18[2] were used for all LC separations. Both eluents A (water) and B (acetonitrile) contained 0.1% formic acid as the ion-pairing modifier. Tryptic digests were separated by using a 0.33% gradient (0.5 to 36% acetonitrile over 107 min), corresponding to 2 h of LC-MS instrument time. Quantities of 1.0 and 0.5 μg of the T. thermohydrosulfuricus WC1 digest were used for the injection into IDA (information-dependent acquisition) and SWATH-MS (36) analyses, respectively. Aside from the differences in the amount of injected digest, the identical chromatography settings on the same mass spectrometer were used, permitting the direct utilization of IDA-based identification data to drive the subsequent SWATH-MS analyses.
A TripleTOF5600 mass spectrometer (Applied Biosystems, Foster City, CA) was used for both IDA acquisition and SWATH-MS analyses. Each scan in the standard tandem mass spectrometry (MS/MS) IDA included a 250-ms survey of MS spectra (m/z 300 to 1,500) and up to 20 MS/MS measurements on the most intense parent ions (300-counts/s threshold, +2 to +4 charge state, m/z 100 to 1,500 mass range for MS/MS, 100 ms each). Previously targeted parent ions were excluded from repetitive MS/MS acquisition for 12 s (50-mDa mass tolerance). Raw spectrum files were converted into Mascot Generic File (MGF) format for peptide and protein identification. Each cycle in SWATH-MS acquisition mode consisted of a 250-ms survey MS spectra, followed by 34 MS/MS spectra in 25-Da wide parent blocks (100 ms each), giving a total of 3.65 s/cycle.
(iv) Data processing and analyses.Peptides were identified from the observed MS/MS spectra using an in-house GPU peptide search engine (37) from a single missed cleavage tryptic database derived from the T. thermohydrosulfuricus WC1 genome annotation with only the fixed posttranslational modification of the carbamidomethylation of cysteine residues (+57.021 Da) applied. The GPU search engine settings were 20 ppm on parent mass ions and a 0.2-Da window on fragment ions, respectively, and peptides with an expectation log(e) of ≤−1 were reported. Each identified peptide in the six IDA runs was aligned against its corresponding collision-induced dissociation fragment spectra to form a fragment ion library with each hypothetical transition (b or y ions for m/z where z = +1, z = +2) computed from the peptide sequence and integrated over a 20-ppm window of the target spectrum. A strategy was then devised to combine these six libraries into a single reference database, which could potentially allow for peptides observed in only a single IDA run to be detected in the potentially deeper SWATH-MS analysis.
For each fragment ion library, the most intense identification of every nonredundant peptide sequence was selected. Across the collection of IDA runs, the retention times of these nonredundant peptides were averaged, with the identified peptides with the greatest signal providing the fragments. This nonredundant “averaged” collection was then formatted to a tab-delimited collection of Q1/Q3 transitions for processing by PeakView software (AB Sciex, Framingham, MA). This ion library contained 111,856 transitions spanning 9,110 peptides belonging to 1,313 proteins. PeakView transition XIC extraction within detection windows of ±5 min and 20 ppm was applied to all six SWATH-MS runs, and the resulting peptide-level intensity report contained 6,264 entries spanning 1,171 proteins. Proteins identified with only a single member peptide were excluded from the analysis.
The expression measurement for each protein was total ion current (TIC), measured as the sum of the signal intensities for each protein's observed member peptides, and expressed on a log2 scale. Normalized TIC values (nTIC) of the raw data for each protein were computed by dividing the observed TIC by the log2 value of its mass in kilodaltons. An additional filtering step of removing proteins with log2 nTIC values having a difference in magnitude ≥2 across biological replicates (culture A and culture B) was applied, yielding a final data set of 832 proteins quantified across all six conditions.
The relative expression levels between growth conditions were calculated for each protein using a simple transformation, which combines difference measurements into a unified expression (W) based on protein level Z-scores and is measured in units of standard deviation. First, expression values for all populations are normalized to a mean of zero and a standard deviation of 1. For each comparison, four difference measurements were determined. These difference measurements incorporate the intrareplicate variability (R0 = BX − AX, R1 = BY − AY), as well as the intercondition variability (Z0 = AX − AY, Z1 = BX − BY), where A and B represent different glycerol stock cultures and X and Y represent different substrates. Vectors were computed as the distance from the origin to the mapped points [(R0, R1) or (Z0, Z1)]. The difference between the magnitudes was scaled by the ratio of their respective population widths, yielding W with the sign being computed via the angle subtended from the x axis to (Z0, Z1). Angles between 315° and 135° are positive; other angles are negative. Finally, W was normalized to Wnet and for the present work is represented as follows: (i) W0net = xylose-grown cells – cellobiose-grown cells; (ii) W1net = xylose-grown cells – sugar mix-grown cells, and (iii) W2net = cellobiose-grown cells – sugar mix-grown cells. Therefore, in W0net (for example), for any protein, positive values represent higher nTIC values observed for growth on xylose in comparison to cellobiose, whereas negative values represent the opposite.
Global expression trends were analyzed using an in-house analysis platform, which groups proteins and genes into membership in “higher-order variables” (HOVs) extracted from the IMG-ER “export gene information” function (29) for an individual genome. These include (i) METACYC pathways (38), (ii) enzyme class (EC) numbers (39), (iii) clusters of orthologous group (COG) classes (40), or (iv) KEGG modules (41). Coarse global expression trends associated with an HOV are defined as an asymmetry (upregulated or downregulated) in the expression of proteins in that HOV, relative to the expression profiles of the overall population. For this study, HOV analysis has been limited to COG categories and a Wnet value of ±1 standard deviation (representing the outermost 32% of the population) and has been used as an initial guide to investigate population regulation asymmetries.
For cross-species comparative “omic” analyses, the complete microarray data set for Thermoanaerobacter sp. strain X514 (20) was accessed from the NCBI GEO database (GSE24458). The retrieved data were parsed by a custom perl script, and each normalized gene expression collection (from eleven growth conditions × three biological replicates each) was reduced by summing the intensities of the three replicates per condition into a single value. The resulting 11 summed-intensity values were analyzed by using our in-house analysis platform, which mapped the values into a log2 scale and cross-referenced each gene to a biological context driven by the Thermoanaerobacter sp. strain X514 gene information file accessed in IMG (42).
RESULTS
Genomic analysis of CCR network genes.Analysis of the T. thermohydrosulfuricus WC1 genome identified genes putatively encoding a carbon catabolite control protein A (CcpA)-dependent CCR network. The identified homologous protein sequences include (i) a histidine-containing protein (HPr) (TthWC1_1711), (ii) a catabolite repression HPr-like protein (CrH) (TthWC1_2012), and (iii) a HPr kinase/phosphatase (TthWC1_1297). The His15 and Ser46 residues of HPr in Bacillus subtilis, which upon phosphorylation activate HPr for either PTS-mediated transport or CCR network signaling, respectively (43–45), were conserved in the T. thermohydrosulfuricus WC1 annotated protein sequence (see Fig. S1 in the supplemental material). In addition, sequence alignments (see Fig. S2 in the supplemental material) of the CrH protein confirms a conserved Ser46 residue suitable for HPr kinase-dependent phosphorylation, as well as the absence of the His15 residue, as described in B. subtilis (46). Nine annotated proteins homologous to the lacI family transcriptional regulator protein CcpA were also identified (based on assignment into COG1609), making it difficult to determine which homolog, if any, may function as CcpA in vivo.
Twenty-three sites homologous to the B. subtilis catabolite responsive element (cre) sequence and one site homologous to the cre sequence in C. difficile were identified (see Table S1 in the supplemental material). Only a few of these putative cre sequences were found either adjacent to, or overlapping with, genes whose products may be involved with sugar transport or catabolism, although none of these are expected to be involved with xylose catabolism. Although sequences were found both internal to coding sequence(s) (CDS), as well as in intergenic regions, 25% of the identified sequences were also found to be on the strand opposite to a CDS and are unlikely to represent cre sequences.
Growth and metabolic analyses.Based on the presence of annotated cellobiose-specific PTS transporters, as well as genes whose products may form a potential CCR network in T. thermohydrosulfuricus WC1, the substrate utilization profiles of cultures grown on xylose, cellobiose, and xylose plus cellobiose were evaluated. Cells transferred from a xylose-grown inoculum consistently lagged when transferred to medium containing cellobiose only (Fig. 1). A lag was not observed when transferred to media containing xylose or xylose plus cellobiose. All cultures, with the exception of the no-carbohydrate substrate control, grew to a similar maximum OD600 (∼0.85). The fastest doubling time (2.8 ± 0.1 h/generation) was observed for cells grown on cellobiose only, while growth on xylose (3.5 ± 0.3 h/generation) and xylose plus cellobiose (3.3 ± 0.2 h/generation) were slower.
Typical growth curve of T. thermohydrosulfuricus WC1 under single substrate, mixed substrate, or no substrate conditions. Symbols: 5 g of xylose/liter (■), 5 g of cellobiose/liter (●), 5 g of xylose/liter plus 5 g of cellobiose/liter (□), or no added sugar (○). Reported values represent the average of three biological replicates from a single experiment. Error bars represent the standard deviations between replicates.
To account for differences in growth rates, metabolic analyses across conditions were normalized by using cultures harvested at predetermined target OD600 values. The measured OD600 values from the cultures showed a mean deviation from the target OD600 values of 1.54%, with a maximum deviation of 5.5% (see Fig. S3 in the supplemental material). The differences between observed OD600 values and the target ODs were therefore considered negligible. Significant differences in biomass production between growth conditions were also not observed, a finding consistent with the cultures being harvested at similar target OD600 values (see Fig. S4 in the supplemental material).
Growth on all substrates was in carbon excess conditions for T. thermohydrosulfuricus WC1 (Fig. 2A). Residual xylose and cellobiose concentrations were higher for the mixed-substrate condition throughout growth in comparison to the single-substrate conditions. This was likely due to the simultaneous utilization of both xylose and cellobiose in the mixed-substrate condition, suggesting a lack of CCR-type repression with this sugar combination. Further analyses of the specific substrate consumption rates (Table 1) showed that xylose was consumed at a higher rate than cellobiose under single-sugar conditions. This trend was also observed in the mixed-substrate conditions, despite an overall decrease in the consumption rates of each specific sugar, providing additional support for simultaneous substrate utilization. End-product and biomass analyses revealed that most of the major products formed from sugar consumption were accounted for (see Fig. S4 and Table S2 in the supplemental material) and were detected throughout growth. Under the conditions tested, lactate was the predominant end product formed throughout growth, though by mid-exponential phase (target OD600 = 0.40) until the end of growth, the production of lactate increased disproportionately to the production of acetate or ethanol (Fig. 2C). This change correlated with a steep decline in medium pH (see Fig. S4 in the supplemental material).
Substrate consumption and metabolite production of T. thermohydrosulfuricus WC1 in single- or mixed-sugar conditions. Consumption or end-product formation on 5 g of xylose/liter (gray), 5 g of cellobiose/liter (white), or 5 g of xylose/liter plus 5 g of cellobiose/liter (black) at target ODs. Cultures harvested immediately postinoculation are assigned a target OD of 0. (A) Xylose consumption (diamonds) and cellobiose consumption (circles). (B) H2 (diamonds) and CO2 (circles) production. (C) Acetate (diamonds), ethanol (circles), and lactate (squares) production.
Specific substrate utilization rates of T. thermohydrosulfuricus WC1
Proteomic analysis and global expression trends.Six 2-h IDA runs were conducted on T. thermohydrosulfuricus WC1 yielding 205,309 MS/MS spectra (see Table S3 in the supplemental material). Over half of the MS/MS spectra yielded peptide identifications, and the ratio of nonredundant peptides to overall peptides was approximately 1:3. SWATH-MS analyses allowed for the quantification of 892 proteins under all three growth conditions, a number reduced to 832 after filtering data with large differences between biological replicates. The dynamic range of observed log2 nTIC values was >17. Good correlation was observed between biological replicates of the filtered data sets (see Fig. S5 in the supplemental material) as linear regression analysis of the replicates showed a slope of >0.96 and R2 values of >0.97 across conditions. The cross-condition correlations were also high (see Fig. S6 in the supplemental material), suggesting relatively stable protein abundance scores under all conditions tested. Proteomes derived from xylose-grown versus cellobiose-grown cells were the most different (see Fig. S6A in the supplemental material), whereas proteomes from cellobiose- versus sugar mix-grown cells were the most similar (see Fig. S6C in the supplemental material), suggesting that cellobiose may have a broader impact on global protein abundance profiles than does xylose.
To determine whether substrate specific protein profiles emerged, the observed proteins were grouped according to their respective annotated COG class. Wnet values found in the outermost 32% of each population were used to identify protein level responses of COG-designated processes as a whole. The most pronounced changes, in terms of number of proteins affected, were observed with proteins in COG class G—carbohydrate metabolism and transport (Table 2), an observation consistent with culture growth on different substrates. Under xylose-only conditions, the abundance levels of proteins involved in energy production and conversion (COG class C) were upregulated compared to the cellobiose-only or sugar mix condition. Conversely, proteins in COG class E (amino acid transport and metabolism) and COG class J (translation) were upregulated in cellobiose-containing cultures (as a single substrate or in the sugar mix) compared to growth on xylose alone. This trend is consistent with the higher growth rates observed on cellobiose or the sugar mix compared to those seen with growth on xylose alone.
Number of proteins in specific COG classes up- or downregulated under specific growth conditionsa
Proteomic analysis of CCR network genes.Proteins of a potential CcpA-dependent CCR network (HPr, CrH, and HPr kinase/phosphatase) were not observed under any conditions. In addition, only three proteins corresponding to the nine annotated ccpA homologs (TthWC1_0680, TthWC1_2179, and TthWC1_2451) were observed in the proteomes (see Table S4 in the supplemental material). Thus, given that the CCR-related network proteins were not detected, in conjunction with the simultaneous utilization of xylose and cellobiose (Fig. 2 and Table 1), it appears that if a CcpA-dependent CCR network exists in T. thermohydrosulfuricus WC1 that cellobiose consumption does not exert a repressive effect preventing xylose catabolism or vice versa.
Cellobiose and xylose transport and hydrolysis.Comparison of the average nTIC values between subunits of both annotated cellobiose-specific PTS-type transporters (Fig. 3A; see Table S4 in the supplemental material) suggest similar expression of both gene clusters when grown on cellobiose alone. However, the large negative W0net and W1net values observed for TthWC1_0920 to TthWC1_0922 proteins in comparison to TthWC1_2146 to TthWC1_2148 proteins suggests that expression of the former is more responsive to the presence of cellobiose. Both PTS-type transporters are colocalized in the genome with transcriptional antiterminator proteins containing PTS-regulatory domains (pfam00874), which may regulate their expression, similar to what has been proposed for PTS transporters in Thermoanaerobacter sp. strain X514 (20). The TthWC1_0920 to TthWC1_0922 gene cluster is also colocalized with glycoside hydrolase (GH) family 1 and family 4 proteins, which may be involved with cellobiose hydrolysis (see below), whereas the TthWC1_2146 to TthWC1_2148 gene cluster has no obvious genes involved with cellobiose hydrolysis proximally located.
Normalized relative Z-score expression ratios (Wnet) of select T. thermohydrosulfuricus WC1 sugar transport and hydrolysis mechanisms. Transport and hydrolysis mechanisms have been limited to cellobiose (A) or xylose (B) and reflect potential modes of entry into the cell and hydrolysis for entry into either the glycolytic (cellobiose) or pentose phosphate (xylose) pathways for subsequent catabolism. Locus tags in black were observed in the proteomes, while those in red were not. Solid lines represent potential routes when transport is through phosphotransferase system (PTS)-mediated mechanisms, while dashed lines represent ABC-mediated transport. W0net, W1net, and W2net are as defined in Materials and Methods. The four-color scheme represents the populations of Wnet, where red boxes are the outermost 1% (|Wnet| ≥ 2.58), green boxes are the outermost 5% (|Wnet| ≥ 1.96), blue boxes are the outermost 10% (|Wnet| ≥ 1.65), and white boxes are the innermost 90%.
The genome also encodes genes (TthWC1_1006 to TthWC1_1008) orthologous to the proposed cellobiose ABC-type importer (cglF-cglG) in Thermoanaerobacter brockii subsp. brockii HTD4 (47). The expression of this gene cluster, though, coincides with the presence of xylose rather than cellobiose (Fig. 3).
Multiple cytosolic GH enzymes potentially capable of cellobiose hydrolysis, in addition to a membrane-bound cellobiose phosphorylase, are annotated in T. thermohydrosulfuricus WC1. Three genes (TthWC1_0923, TthWC1_0924, and TthWC1_0925) are immediately adjacent to one of the cellobiose-specific PTS-type transporters. Only two of the genes (TthWC1_0923 and TthWC1_0924) follow an expression profile similar to that of the cellobiose-responsive PTS-type transporter (Fig. 3). An additional GH1 enzyme, TthWC1_0946, was also observed at a higher abundance level in the presence of cellobiose, although its average nTIC value is less than that for TthWC1_0923 or TthWC1_0924 by a log2 difference of ≥1.58 under cellobiose-only conditions (see Table S4 in the supplemental material). Of the three GH enzymes upregulated in response to cellobiose, TthWC1_0924 is the most abundant in both the cellobiose-only and the sugar mix conditions. Neither the cellobiose phosphorylase nor the annotated phosphoglucomutase needed to convert glucose-1-P to glucose-6-P was detected in the proteomes, supporting the proposal that cellobiose is transported via PTS-mediated mechanisms and not ABC-dependent mechanisms in T. thermohydrosulfuricus WC1 (Fig. 3A).
The expression of the TthWC1_1006 to TthWC1_1008 transporter correlates with growth on xylose (Fig. 3B). Its function as a presumed xylose transporter is further supported by its genomic context, which includes an annotated extracellular endoxylanase (GH10 enzyme) (TthWC1_1010), an acetyl-xylan esterase (TthWC1_1011), and an extracellular GH52 β-xylosidase (TthWC1_1012). Proteins in the region spanning TthWC1_1005 to TthWC1_1023, which also includes the xylose isomerase (TthWC1_1022) and xylulose kinase (TthWC1_1021), are more abundant when growth is on xylose or the sugar mix than when growth is on cellobiose alone (see Table S4 in the supplemental material). No ATPase subunit is found associated with the annotated ABC-type transporter proteins, and complex formation may therefore rely on an independently transcribed ATPase subunit.
A second xylose-specific ABC transporter (TthWC1_1216 to TthWC1_1219), which includes an ATPase subunit, was not observed in the proteomes. An ABC transporter (TthWC1_0997 to TthWC1_1000), annotated to be involved with ribose transport, is also encoded near the xylose-responsive gene cluster (TthWC1_1005 to TthWC1_1023). Hemme et al. (11) suggest that in strains lacking obvious xylose-specific ABC transporters, such as Thermoanaerobacter pseudethanolicus 39E, this orthologous ribose transporter may have xylose transport capabilities. However, it was also identified that this gene cluster was not upregulated in response to growth on xylose by T. pseudethanolicus 39E. In T. thermohydrosulfuricus WC1, only two of the four annotated subunits were observed in the proteomes, and only the expression of one subunit, the periplasmic component (TthWC1_0997), strongly correlated with the presence of xylose.
Central carbon metabolism.The abundance levels for individual proteins in the glycolytic pathway and the nonoxidative branch of the pentose phosphate pathway were comparable across growth conditions (see Fig. S7 in the supplemental material). However, the abundance levels of the putative xylose isomerase and xylulose kinase proteins, which feed xylose into the pentose phosphate pathway, were much higher under xylose-only and mixed-substrate conditions compared to cellobiose-only conditions. Proteomic analyses were unable to resolve which gene product(s) may form a functional transketolase enzyme due to ambiguity in the annotated sequences relative to functionally characterized enzymes (see Text S1 in the supplemental material). Catabolism of hexose sugars via the Entner-Doudoroff pathway, or through the oxidative branch of the pentose phosphate pathway, is considered unlikely based on previous genomic analyses of T. thermohydrosulfuricus WC1 (23), as well as pathway characterization in other Thermoanaerobacter strains (11, 48). Of note, a pyruvate dikinase had an average nTIC log2 of ≥1.96 versus pyruvate kinase across all growth conditions (see Table S4 in the supplemental material). This suggests a potential catabolic role for pyruvate dikinase, in addition to the pyruvate kinase, involving the conversion of phosphoenolpyruvate (PEP) to pyruvate, as has been reported in other thermophiles (49). Its comparatively high expression, along with other pyrophosphate catabolizing enzymes (see Text S1 and Table S5 in the supplemental material), also suggest a potential role for pyrophosphate as an important energy-related metabolite in both T. thermohydrosulfuricus WC1 and Thermoanaerobacter sp. strain X514, as has been described in other Clostridia (50).
End-product and cofactor metabolism.The formation of lactate is most likely catalyzed by a highly abundant lactate dehydrogenase (LDH) (TthWC1_1229) (Fig. 4A; see Table S6 in the supplemental material). As an alternative to lactate production, pyruvate may be decarboxylated to acetyl coenzyme A (acetyl-CoA), yielding CO2 and reduced ferredoxin (Fdred) via pyruvate ferredoxin oxidoreductase (POR). Two multisubunit, as well as two single-subunit PORs homologous to the characterized single subunit POR in Moorella thermoacetica (51, 52), have been identified in the genome. All subunits forming the putative multisubunit PORs were not observed, whereas both single-subunit PORs were. Expression analysis identified that TthWC1_1927 was observed at an average nTIC log2 difference ≥3.47 higher than TthWC1_1529 under each condition (see Table S6 in the supplemental material) and is likely the primary POR utilized.
Normalized relative Z-score expression ratios (Wnet) of the end-product-producing reactions in T. thermohydrosulfuricus WC1. (A) Acetate, ethanol, lactate, and CO2 producing reactions. (B) Hydrogenases. The color scheme is as identified in Fig. 3. W0net, W1net, and W2net are as defined in Materials and Methods. Acetyl-P, acetyl-phosphate.
Acetate production may proceed primarily via the colocalized phosphotransacetylase and acetate kinase genes (Fig. 4A). Gene products annotated as phosphotransbutyrylase (TthWC1_2222 and TthWC1_2230) and butyrate kinase (TthWC1_2228 and TthWC1_2231), whose specificities are not known and which may also contribute to acetate production, were also observed in the proteomes, but at much lower levels (see Table S4 in the supplemental material).
Ethanol production from acetyl-CoA in T. thermohydrosulfuricus WC1 may occur directly through a bifunctional acetaldehyde:alcohol dehydrogenase (AdhE) or via a stand-alone aldehyde dehydrogenase and independent alcohol dehydrogenase (ADH) (Fig. 4A). The stand-alone aldehyde dehydrogenase (TthWC1_1190), which exists in only a few Thermoanaerobacter spp. (23), was observed in the proteomes but was not observed in high abundance, in agreement with the expression levels observed for the orthologous gene (Teth514_1914) in Thermoanaerobacter sp. strain X514 (see Table S6 in the supplemental material). AdhE, on the other hand, was found in high abundance in all T. thermohydrosulfuricus WC1 proteomes and may represent the principal route through which acetaldehyde is formed. In addition, the protein (TthWC1_0498) encoded by the gene orthologous to adhA in Thermoanaerobacter ethanolicus JW200 (53, 54) was in the top five most abundant proteins across all growth conditions (see Tables S4 and S6 in the supplemental material). Two other ADH proteins, including the protein encoded by the adhB ortholog in T. thermohydrosulfuricus WC1 (TthWC1_0605), were also observed in the proteomes, whereas a final annotated ADH-encoding gene (TthWC1_1530) was not. Under all conditions for T. thermohydrosulfuricus WC1, and most conditions for Thermoanaerobacter sp. strain X514 (9 of 11 conditions for adhA; 6 of 11 conditions for adhE), the abundance of the adhA and adhE gene products was higher than the lactate dehydrogenase gene product (see Table S6 in the supplemental material).
Three distinct hydrogenases, including a Fd-linked energy-conserving hydrogenase (Ech), a putative bifurcating hydrogenase, and a Fd-linked Fe-hydrogenase, are present in T. thermohydrosulfuricus WC1 (Fig. 4B). The first, the Ech hydrogenase, had only three of the six annotated subunits observed in the proteomes. Although it is possible that the remaining three proteins are synthesized and were just not observed, only a single protein, TthWC1_1090, encoded within the adjacent hypA-hypF gene cluster, was also observed (see Table S4 in the supplemental material). The hypA-hypF gene products are known to be responsible for hydrogenase maturation and the assembly of the NiFe center in NiFe hydrogenases (55), and their absence from the observable proteomes suggests low, if any, expression of Ech.
Only three gene products (TthWC1_1780 to TthWC1_1782) of the five annotated genes that compose a potential bifurcating hydrogenase were identified in the proteomes (Fig. 4B; see also Table S6 in the supplemental material), although ambiguity exists in terms of what subunits are needed to form a functional complex. The gene cluster is orthologous to those found in other Thermoanaerobacter spp. (23), including those previously proposed to form a bifurcating hydrogenase (56) with the exception that TthWC1_1784 is annotated as a pseudogene. In addition, TthWC1_1783 is annotated as a histidine kinase, and its role is unknown. Average nTIC analysis shows that the TthWC1_1780 and TthWC1_1781 proteins are highly abundant (top 35 most abundant proteins) across all conditions, a finding similar to what is observed in Thermoanaerobacter sp. strain X514 (see Table S6 in the supplemental material), although the ferredoxin-encoding gene (TthWC1_1782) is not as highly expressed. A putative single subunit, Fd-linked hydrogenase (TthWC1_1787) was also not observed in the proteomes.
The adh gene products in some Thermoanaerobacter spp. have been shown to utilize both NADH and/or NADPH as cofactors (53, 54, 57). Although NADH is typically considered a by-product of glycolysis, NADPH production in Thermoanaerobacter spp. is less well understood but is an important component of metabolism and ethanol production. T. thermohydrosulfuricus WC1, like other strains of Thermoanaerobacter (48, 58), has no apparent 6-P-gluconolactonase. Its absence suggests that NADPH production through the oxidative portion of the pentose phosphate pathway may not be possible in Thermoanaerobacter spp. The genome does encode an annotated NADP-specific isocitrate dehydrogenase (TthWC1_0990), however, which is also observed in the proteomes (see Table S6 in the supplemental material).
In addition, a two-gene cluster orthologous to the nfnAB genes in Clostridium kluyveri (59) and M. thermoacetica (60), which couples the reduction of NADP+ via Fdred with the reduction of NADP+ via NADH, is annotated (TthWC1_0606-TthWC1_0607). Only one protein (TthWC1_0606) was observed in the proteomes. This contrasts with the data from Lin et al. (20), which identifies mid- to high-level expression of the orthologous genes in Thermoanaerobacter sp. strain X514 under the tested conditions (see Table S6 in the supplemental material).
Energy conservation.T. thermohydrosulfuricus WC1 has many mechanisms to conserve energy as an ion-motive force or as ATP/PPi (23). A potential ion-gradient forming reaction, catalyzed by Ech, has already been discussed. Another reaction, whereby the hydrolysis of pyrophosphate via a V-type pyrophosphatase is coupled to ion translocation, may exist as synthesis of the V-type pyrophosphatase is confirmed (see Text S1 and Table S4 in the supplemental material). No gene products in the ion-translocating mbx gene cluster (TthWC1_0717 to TthWC1_0729) were observed, whereas the gene product homologous to nhaC, encoding a potential H+/Na+ antiporter, was also not observed. Both products of the two-gene cluster (TthWC1_1955-TthWC1_1956) homologous to the natAB ATPase in B. subtilis (61) were observed (see Table S4 in the supplemental material), whereas all but subunits “a” (TthWC1_0477) and “c” (TthWC1_0476) of the FoF1-ATPase were also observed. An annotated ATP-linked (EC 4.1.1.49) (62) PEP carboxykinase (TthWC1_1173) was not observed.
Under xylose-only conditions, T. thermohydrosulfuricus WC1 upregulates synthesis of proteins belonging to COG class C (energy production and conversion) (Table 2). The upregulated gene products observed are distinct from the gene products associated with ion-motive force or ATP/PPi generation (above) (see Table S4 in the supplemental material). Included in the upregulated COG class C genes is a gene cluster (TthWC1_2220 to TthWC1_2231), which encodes gene products putatively associated with amino acid metabolism. The annotated genes include an indolepyruvate ferredoxin oxidoreductase, a multisubunit pyruvate/2-ketoisovalerate ferredoxin oxidoreductase, and a leucine dehydrogenase, which suggests a possible energy-related role for increasing amino acid metabolism when T. thermohydrosulfuricus WC1 is grown on xylose.
DISCUSSION
A lack of CCR allowing for the simultaneous utilization of lignocellulose relevant saccharides is a desirable phenotype for CBP microorganisms. Identifying possible sugar utilization preferences in select strains is particularly relevant for designer cocultures whereby substrate availability would be influenced by the differential hydrolysis mechanisms and utilization capabilities of the constituent strains. To test whether coutilization of cellobiose and xylose, two important lignocellulose constituent saccharides, was possible, T. thermohydrosulfuricus WC1 was grown on 5 g of cellobiose/liter plus 5 g of xylose/liter. For this study, it was important that each saccharide in the sugar mix condition be available in excess (Fig. 2A) to help delineate between coutilization of each saccharide due to a lack of CCR mechanisms and coutilization due to carbon limitation of a single saccharide. Despite encoding genes for a potential CCR network and that cellobiose transport and catabolism may induce a CCR effect, T. thermohydrosulfuricus WC1 was able to utilize both xylose and cellobiose simultaneously (Fig. 2 and Table 1). Although the concentrations of sugars tested (i.e., 10 g total sugar/liter) is far less than what is desirable for industrial processes of added soluble sugars (63, 64), the concentrations of each sugar likely exceeds levels that would accumulate through C. thermocellum-T. thermohydrosulfuricus WC1-mediated hydrolysis of lignocellulosic biomass (1, 65, 66) in designer cocultures for consolidated bioprocessing under these conditions.
Understanding the nature of substrate-linked gene regulation is important to improving our understanding of T. thermohydrosulfuricus WC1 physiology. The HOV analysis presented identified that growth on xylose alone lead to an upregulation in genes involved with energy production (COG class C), which is consistent with reports for Thermoanaerobacter sp. strain X514 (20). Despite this, growth of T. thermohydrosulfuricus WC1 under xylose-only conditions corresponded to the lowest growth rate and even slowed growth under the mixed-substrate conditions (above), suggesting that xylose utilization may be a bottleneck that impedes cellobiose utilization in T. thermohydrosulfuricus WC1. This is in contrast to results reported for Thermoanaerobacter sp. strain X514, which identifies cellobiose utilization and end-product formation of cellobiose-grown cultures as a metabolic bottleneck in that strain (20). The differences in growth rate for T. thermohydrosulfuricus WC1 under different conditions may be partly attributable to differences in potential energy yield from both the transport and the hydrolysis of each substrate.
Proteomic analyses suggest xylose transport is principally through ABC-type mechanisms, as has been reported in other Thermoanaerobacter spp. (17, 20, 67, 68), and would consume 1 mol of ATP per mol of xylose transported. Conversely, transport of the disaccharide cellobiose seemingly occurs via PTS-mediated mechanisms. Its transport would also consume 1 mol ATP equivalent/mol of cellobiose. In this case, the phosphate from PEP would ultimately phosphorylate the transported cellobiose molecule rather than ADP (forming ATP), as it would in the conversion of PEP to pyruvate. The energy expenditure is therefore equivalent for transporting 1 mol of either substrate.
However, assuming PTS-mediated transport of cellobiose occurs, its hydrolysis would yield 1 glucose plus 1 glucose-6-P (Fig. 3), and the theoretical ATP yield from converting 1 mol of cellobiose to pyruvate would be 5 mol of ATP. The conversion of xylose to pyruvate through the pentose phosphate pathway would only yield 1.67 mol of ATP/mol of xylose, providing less energy for growth. If true for T. thermohydrosulfuricus WC1, these differences could potentially account for the differences in the observed doubling times. This is further supported by the observed differences in substrate consumption rates (Table 1). While xylose was consumed at a higher rate than cellobiose under both the single-substrate and the mixed-sugar conditions, the calculated rate was not sufficiently fast to mitigate differences in the energetic potential yielded from each substrate. Similar findings were observed by Lacis and Lawford (69), who report lower ATP yields for Thermoanaerobacter ethanolicus grown on xylose in contrast to T. ethanolicus grown on glucose, which correlated with slower growth of T. ethanolicus on xylose under batch conditions.
To compensate for growth on a poorer energy source such as xylose, T. ethanolicus simultaneously consumed yeast extract as an additional energy source during growth but did not during growth on glucose (69). Like many other Thermoanaerobacter spp. (70), T. thermohydrosulfuricus WC1 is capable of growing on yeast extract (21), although the reduced concentrations of yeast extract and the absence of thiosulfate (70, 71) in the medium used make it unlikely to act as a major carbon or energy source here (Fig. 1). It is plausible, however, that the genes upregulated during growth on xylose (COG class C), notably the indolepyruvate ferredoxin oxidoreductase, 2-ketoisovalerate ferredoxin oxidoreductase, and leucine dehydrogenase, which are known to be involved with peptide fermentation (72, 73), may be a mechanism by T. thermohydrosulfuricus WC1 to maximize energy when grown on poorer energy sources. The specificity of these gene products is not yet known, however, and in some cases, such as with the annotated indolepyruvate ferredoxin oxidoreductase or the 2-ketoisovalerate ferredoxin oxidoreductase, the gene products may have POR activity and not be involved with amino acid fermentation.
Previous characterization of T. thermohydrosulfuricus WC1 in medium containing 2 g of cellobiose/liter plus 2 g of yeast extract/liter (21) reported a faster doubling time than was observed here with 5 g of cellobiose/liter plus 0.67 g of yeast extract/liter. Further, the maximal OD600 values reached for cultures in the present study were significantly less than those observed previously (data not shown). However, while reducing the concentrations of nutrient-rich yeast extract in the present study was important in minimizing the potential impact of amino acid metabolism on growth for evaluation of CCR, as well as our interpretation of the proteomes, its reduction may have also limited the availability of an essential nutrient required for T. thermohydrosulfuricus WC1 anabolic processes. A potential exists whereby these conditions influenced end-product distribution patterns within the strain.
Previous analyses with T. ethanolicus JW200 have found that increasing glucose or xylose concentrations to sugar excess levels led to a disproportionate increase in lactate production (69, 74)—similar to what was observed here for T. thermohydrosulfuricus WC1 in comparison to our previous results (21). In T. ethanolicus JW200-FE7, high lactate production under carbon excess conditions was correlated with reduced concentrations of yeast extract used in the fermentation medium (75). The high lactate production was proposed to result from a limitation in the strain's anabolic processes generated through nutrient (other than carbon) limitation. While the influence that anabolic reactions may have on substrate catabolism (and biofuel production) is not yet known for T. thermohydrosulfuricus WC1, the current data suggest a similar link may also exist. If high catabolic flux due to sugar excess conditions was simultaneously coupled to anabolic constraint due to nutrient limitation, this may have led to the high lactate production observed as a means of disposing excess carbon and reducing equivalents generated through sugar catabolism.
High expression of all end-product-forming enzymes was observed (see Table S6 in the supplemental material) and, in some cases, the adh gene products were even more abundant than the LDH. Thus, the relative protein abundance levels do not correlate with end-product distribution patterns in T. thermohydrosulfuricus WC1. Similar results are also identified in the transcriptomic data derived from Thermoanaerobacter sp. strain X514 (see Table S6 in the supplemental material). As such, this provides additional evidence that end-product distribution within Thermoanaerobacter spp. is controlled by factors other than gene product abundance and may therefore be related to anabolic constraint (above), catabolic bottlenecks, metabolite-based activity control (allosteric regulation), or cofactor availability.
The high lactate production observed here may occur due to a limitation in the ability of T. thermohydrosulfuricus WC1 to produce alternative end products. The decarboxylation of pyruvate through POR could be a potential bottleneck, as has been reported in other saccharolytic clostridia (76, 77), leading to the disposal of carbon, and reducing equivalents in the form of NADH, by producing lactate. Slow consumption of acetyl-CoA generated from POR mediated catalysis could also exist. This would require that the phosphotransacetylase, and/or the enzyme(s) catalyzing acetaldehyde formation, have a much lower affinity for acetyl-CoA than LDH does for pyruvate, or function at a slower catalytic rate.
Alternatively, differences in allosteric activation of specific enzymes may also influence end-product synthesis patterns. In Thermoanaerobacter brockii subsp. brockii HTD4, high lactate formation has been associated with high intracellular concentrations of fructose-1,6-bisphosphate (FBP) (78), while the LDH enzymes of T. pseudethanolicus 39E (79), T. ethanolicus JW200 (80), and Thermoanaerobacter wiegelii Rt8.B1 (81) have been shown to be allosterically activated by FBP. If also true in T. thermohydrosulfuricus WC1, intracellular FBP levels must always be sufficiently high to activate lactate formation under the conditions tested since lactate was produced throughout all stages of growth. Differences in these levels may also partly explain the different lactate production profiles observed between T. wiegelii Rt8.B1 and T. thermohydrosulfuricus WC1. T. wiegelii Rt8.B1 is reported to produce lactate in a pH-dependent manner with maximal LDH activity under slightly acidic conditions (pH 6.2), which corresponds with a declining intracellular pH throughout growth (81). Although the intracellular pH of T. thermohydrosulfuricus WC1 is unknown, a similar decline could increase LDH activity and potentially explain the increased lactate: ethanol ratio observed upon reaching a target OD600 of 0.40 until the end of growth (Fig. 2C). However, the average medium pH at mid-exponential phase was ≥6.68 (see Fig. S4 in the supplemental material) in the present study, while lactate formation in T. wiegelii Rt8.B1 was not reported until the medium pH dropped below 6.0. Therefore, differences in intracellular pH and/or LDH activity may also contribute to the different lactate formation profiles between these strains and not just differences in FBP concentrations.
The availability of specific reducing equivalents may also influence end-product formation. In T. ethanolicus JW200, the LDH enzyme has been characterized as being NADH dependent, and no activity was observed when NADPH was used as an alternative cofactor (80). Conversely, AdhA in T. ethanolicus JW200, whose ortholog in T. thermohydrosulfuricus WC1 was the most abundantly expressed adh gene product (see Table S6 in the supplemental material), is considered to be primarily NADPH dependent (53, 54). If similar specificities are also present in T. thermohydrosulfuricus WC1, and assuming that glycolysis principally generates NADH rather than NADPH, ethanol formation via AdhA may be limited by a lack of available NADPH. While the adhE gene was also highly expressed in T. thermohydrosulfuricus WC1, the study by Pei et al. (82) with T. ethanolicus JW200 suggests that AdhE may be principally involved with acetaldehyde, rather than alcohol, formation.
Flux analyses conducted in other Thermoanaerobacter spp. (11) suggest that carbon flux through an NADP-specific isocitrate dehydrogenase is unlikely to provide significant levels of NADPH for ethanol production via AdhA. Therefore, the only obvious mechanism for NADPH production in T. thermohydrosulfuricus WC1 may be the nfnAB gene products, though not all subunits were observed in the proteomes. The gene products were relatively abundant in Thermoanaerobacter sp. strain X514 (see Table S6 in the supplemental material) and may potentially contribute to differences in NADPH availability (and ethanol production) between these strains.
Differences in the consumption of POR-generated Fdred between T. thermohydrosulfuricus WC1 and other Thermoanaerobacter spp. may also influence NAD(P)H availability and ethanol production. Comparatively high expression of the putative bifurcating hydrogenase relative to other gene products was observed in the both the Thermoanaerobacter sp. strain X514 microarray data and the T. thermohydrosulfuricus WC1 proteomes (see Table S6 in the supplemental material). In both strains, this hydrogenase represents a potential Fdred sink leading to the formation of hydrogen, rather than ethanol. However, genomic analyses suggests that both strains could also putatively convert Fdred to NAD(P)H via an NADH: Fd oxidoreductase (NFO) (Thermoanaerobacter sp. strain X514) or an mbx-like enzyme (T. thermohydrosulfuricus WC1). Expression of all NFO-encoding subunits was observed for Thermoanaerobacter sp. strain X514, while the mbx-enzyme subunits of T. thermohydrosulfuricus WC1 were not (see Table S6 in the supplemental material). This difference may not only consume Fdred but also provide increased NAD(P)H in Thermoanaerobacter sp. strain X514 relative to T. thermohydrosulfuricus WC1. Depending on the nature of the reducing equivalent and enzyme specificity (NADH or NADPH), this may improve the thermodynamic conditions for ethanol formation in Thermoanaerobacter sp. strain X514 to a greater extent than possible in T. thermohydrosulfuricus WC1.
CONCLUSIONS
The present study has identified that T. thermohydrosulfuricus WC1 simultaneously ferments cellobiose and xylose with no overt repression mechanism, precluding the utilization of either sugar under mixed-substrate conditions. Coupling this phenotype with its biomass hydrolytic capacities relative to many other Thermoanaerobacter spp., as well as its diverse substrate utilization potential (23), and given the relative importance of these capabilities for achieving industrial biofuel production (24, 25), T. thermohydrosulfuricus WC1 should be considered an organism of interest for coculture development. However, the lower ethanol and higher lactate yields observed in T. thermohydrosulfuricus WC1 under the tested conditions is a significant issue in its development as a CBP-relevant, biofuel-producing organism. Proteomic analyses of the strain grown on single and multiple substrates has revealed that T. thermohydrosulfuricus WC1 end-product synthesis patterns seem to be unconnected to protein complement for end-product-forming enzymes. Cross-species “omic” analyses identified a similar trend also exists in Thermoanaerobacter sp. strain X514 and, by extension, may therefore translate to other strains within the genus as well. Through further correlation of the proteomes with the metabolic data, we have also refined the pathways of carbon and electron flux in T. thermohydrosulfuricus WC1 and identified physiological processes (e.g., xylose utilization and NADPH production) that may limit substrate utilization and biofuel production in this strain. We have not, however, been able to fully resolve pathways of carbon flux, since determination of enzyme specificity is currently limited to insights provided by genome annotation, which is not an exact science (83), and detailed knowledge of enzyme activity has not yet been conducted. Further refinement of metabolic flux and the continued investigation into the mechanisms through which end-product distribution is influenced by anabolic processes, metabolite-based control, or cofactor availability may help identify strategies to potentially alleviate the bottlenecks currently limiting ethanol production in T. thermohydrosulfuricus WC1, as well as other thermophilic Firmicutes.
ACKNOWLEDGMENTS
This study was supported by funds provided by Genome Canada for the project “Microbial Genomics for Biofuels and Co-Products from Biorefining Processes” (MGCB2), the Province of Manitoba through funds provided by the Department of Innovation, Energy, and Mines “Manitoba Research Innovation Fund” (MRIF), a Natural Sciences and Engineering Research Council strategic grant (STPGP 365076), and the University of Manitoba.
We also thank Dmitry Shamshurin and Kirill Levin at the Manitoba Centre for Proteomics and Systems Biology for technical assistance and guidance in the preparation of samples for MS/MS analysis.
FOOTNOTES
- Received 29 October 2013.
- Accepted 18 December 2013.
- Accepted manuscript posted online 20 December 2013.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.03555-13.
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.