The Cytosolic Oligosaccharide-Degrading Proteome of Butyrivibrio Proteoclasticus

The growth and productivity of ruminants depends on a complex microbial community found in their fore-stomach (rumen), which is able to breakdown plant polysaccharides and ferment the released sugars. Butyrivibrio proteoclasticus B316T is a Gram-positive polysaccharide-degrading, butyrate-producing bacterium that is present at high numbers in the rumen of animals consuming pasture or grass silage based diets. B316T is one of a small number of rumen fibrolytic microbes capable of efficiently degrading and utilizing xylan, as well as being capable of utilizing arabinose, xylose, pectin and starch. We have therefore carried out a proteomic analysis of B316T to identify intracellular enzymes that are implicated in the metabolism of internalized xylan. Three hundred and ninety four proteins were identified including enzymes that have potential to metabolize assimilated products of extracellular xylan digestion. Identified enzymes included arabinosidases, esterases, an endoxylanase, and β-xylosidase. The presence of intracellular debranching enzymes indicated that some hemicellulosic side-chains may not be removed until oligosaccharides liberated by extracellular digestion have been assimilated by the cells. The results support a model of extracellular digestion of hemicellulose to oligosaccharides that are then transported to the cytoplasm for further digestion by intracellular enzymes.

xylosidase/arabinofuranosidase Xsa43J is expected to hydrolyze xylobiose, arabinoxylans and arabinogalactans. Several ATP-binding cassette (ABC) transporter substrate-binding proteins were also found to be significantly more abundant in the xylan-grown cell culture medium [10], and in the membrane proteome [13], which implies they are important for uptake of oligosaccharides derived from xylan disassembly.
Based on these results the proposed model of hemicellulose degradation and assimilation by B. proteoclasticus suggests extracellular hydrolysis of the xylan backbone of hemicellulose followed by transport of substituted or un-substituted xylooligosaccharides into the cell for further metabolism [10]. This model necessitates that B. proteoclasticus possesses the intracellular catalytic machinery that allows cells to rapidly degrade and metabolize the internalized oligosaccharides. The B. proteoclasticus genome encodes 90 predicted cytosolic oligosaccharide-degrading enzymes that collectively represent 38 distinct GH, CE, and glycosyl transferase (GT) families [9]. Members of the GH2, GH3, GH13, GH31, and GH43 families are the most prevalent. Notably, all of the GH2 and GH31 enzymes and most of the GH3, GH13, and GH43 enzymes are predicted to be localized to the cytosol.
We have therefore now carried out a proteomic analysis of cytosolic fractions from B. proteoclasticus grown on xylan or xylose with the goal of identification of intracellular carbohydrate active enzymes associated with the further metabolism of the products of extracellular digestion of polysaccharides. Two previously described proteomic workflows [10,13] were used: (1) separation of proteins by 2DE followed by use of matrix-assisted laser desorption ionization time-offlight (MALDI-TOF and TOF/TOF) mass spectrometry for identification of protein spots excised from the gels; and (2) liquid chromatography-tandem mass spectrometry (LC-MS/MS) of tryptic digests of the cytosolic samples.

Sample Preparation
B. proteoclasticus cells were grown at 37 °C under anaerobic conditions in modified DSM Medium 704 that included 0.2% yeast extract and 0.2% trypticase peptone, with either 0.1% (w/v) oat-spelt xylan or 0.5% (w/v) xylose [10,13]. Mid-log phase and stationary phase cultures were harvested at approximately OD600 0.5 and OD600 0.7, respectively. Cells were separated from the culture medium by centrifugation at 3000× g for 30 min at 4 °C. The cell pellet was then resuspended in 10 mL of ice-cold 50 mM Tris-HCl (pH 7.0) lysis buffer containing 2% (w/v) CHAPS, cOmplete ® Protease Inhibitor (Roche Applied Science, Mannheim, Germany), and 100 mg of Zirconia/Silica Beads (Biospec Products, Bartlesville, OK, USA) and the cells were lysed by bead-beating for 5 min using a FastPrep FP120-230 bead beater (MP Biomedicals, Santa Ana, CA, USA). The disrupted cell suspension was centrifuged at 16,000× g for 30 min at 4 °C to pellet the beads and cell debris. The supernatant containing the soluble cytosolic fraction was removed and stored immediately at −80 °C.
Protein extracts were purified using the phenol/methanol/ammonium acetate procedure described by Carpentier et al. [14]. Briefly, the cell supernatant was mixed with an equal volume of ice-cold Tris-HCl buffered phenol (pH 8.0) and vortexed vigorously for 30 min at 4 °C. The mixture was centrifuged at 8000× g for 5 min and the phenolic phase was removed and placed in a fresh 1.5 mL microcentrifuge tube with an equal volume of ice-cold 50 mM Tris-HCl (pH 7.0). The mixture was vigorously vortexed for 30 min at 4 °C and then centrifuged at 8000× g for 5 min and the phenol phase removed and placed in a fresh 1.5 mL microcentrifuge tube. Proteins were precipitated by the addition of 4 vols of −20 °C methanol containing 100 mM ammonium acetate. The mixture was vigorously vortexed for 60 s and incubated overnight at −20 °C. The precipitated protein was pelleted by centrifugation at 13,600× g for 60 min at 4 °C, and the protein pellet was air-dried in a fume cupboard overnight.

Two-Dimensional Electrophoresis and Image Analysis
Protein samples were resuspended in 7 M urea, 2 M thiourea, 2% (w/v) CHAPS and 50 mM dithiothreitol and incubated overnight at 4 °C with shaking. Protein concentration was assayed using a 2-D Quant Kit (GE Healthcare, Uppsala, Sweden). Each sample was supplemented with 0.5% (v/v) Immobilized pH Gradient (IPG) Buffer (GE Healthcare, Uppsala, Sweden) and 125 µL was applied to 7 cm IPG strips that were passively rehydrated at room temperature for 16 h. Isoelectricfocusing (IEF), followed by second-dimension SDS-PAGE, and staining with colloidal Coomassie Brilliant Blue G250 was as previously described [10]. Briefly, IEF was for 9-11 kVh in an IPGphor (GE Healthcare, Uppsala, Sweden) followed by incubation of focused IPG strips for 15 min in 1 × LDS Buffer (Invitrogen, Carlsbad, CA, USA) with 1 × Reducing Agent (Invitrogen, Carlsbad, CA, USA), then for 15 min in LDS Sample Buffer containing 100 mM iodoacetamide. Second dimension electrophoresis was at 200 V using NuPAGE ® Novex 4%-12% Bis-Tris ZOOM ® gels in a NuPAGE ® MOPS Buffer (Invitrogen, Carlsbad, CA, USA). Three biological replicates were analyzed for each growth condition, and each biological replicate was analyzed in triplicate (nine gels per treatment). Stained gels were scanned using a Molecular Dynamics Personal Densitometer SI (Sunnyvale, CA, USA). Image Master™ 2D Platinum software (Version 5.0, GE Healthcare, Uppsala, Sweden) was used for gel analysis. Protein spot volumes were calculated for each spot that could be matched in at least seven of the nine analyzed replicate gels for each treatment. All detected spots were given C (cytosol fraction) numbers. A larger number of spots was detected from mid-log growth phase cells (1264 spots) compared to stationary phase cells (1180 spots). All spot volumes were within the linear range previously defined for colloidal Coomassie staining [10]. Missing data points were imputed using the lowest spot volume value for that protein detected on other gels. SPSS (Statistical Package for the Social Sciences) (Version 14.0, IBM Corporation, New York, NY, USA) software was used for statistical analysis with two-tailed Student t-tests (p < 0.01) to assess statistical significance of protein abundance changes.

Identification of Proteins Excised from 2DE Gels
Protein spots excised from stained 2DE gels were prepared for peptide mass fingerprinting using an Ettan Digester (GE Healthcare, Uppsala, Sweden) with 25 ng trypsin per gel plug [15]. Extracted tryptic peptides were resuspended in 1.5 µL of 10 mg.mL −1 α-cyano-4-hydroxycinnamic acid in 50% acetonitrile: 0.1% trifluoroacetic acid (v/v) and analyzed using a Voyager DE Pro MALDI-TOF mass spectrometer (Applied Biosystems, Foster City, CA, USA) as previously described [10,13,16]. MALDI MS/MS using an ABSCIEX 5800 TOF/TOF mass spectrometer (SCIEX, Framingham, MA, USA) was used for repeat analysis of protein spots that contained fibrolytic enzymes. TOF/TOF analysis parameters were: parent ion spectra by accumulating four sub-spectra each of 200 laser shots and spectra; then the 15 most intense ions per MS spectrum (S/N > 10) were selected for MS/MS analysis and fragment ion spectra were acquired by accumulating four sub-spectra each of 250 laser shots using air as the collision gas.

LC-MS/MS
Precipitated cytosolic protein extracts were resuspended in 8 M urea, 100 mM Tris-HCl (pH 8.5) at a final concentration of 0.4 µg.µL −1 . Protein disulfide bonds were reduced with 10 mM dithiothreitol at 56 °C for 30 min, followed by alkylation with 55 mM iodoacetamide at room temperature for 40 min in the dark in the urea-buffer solution. The reduced and alkylated samples were diluted four-fold with 100 mM Tris-HCl (pH 8.5) and digested with trypsin (Roche, modified sequencing grade) at an enzyme-to-substrate ratio of 1:50 w/w with 1 mM CaCl2 overnight at 37 °C. Digestion was stopped by addition of formic acid to 4% (v/v) final concentration. LC-MS/MS was carried out as previously described [16] using a Finnigan LTQ™ Linear Ion-trap (Thermo Scientific, Waltham, MA, USA) with a Dionex UltiMate ® 3000 Nano HPLC system (Thermo Scientific, Waltham, MA, USA) containing an Acclaim ® PepMapTM C18 nano-column (75 µm i.d. × 15 cm, 100 Å pore size). The mobile phase gradient was constructed from 0.1% formic acid (v/v) (Solvent A) and 80% acetonitrile: 0.1% formic acid (v/v) (Solvent B). The gradient consisted of 0 to 15% linear gradient Solvent B over 25 min; 15% to 35% linear gradient Solvent B from 25 to 67 min; 35% to 100% linear gradient Buffer B from 67 to 77 min. Each biological replicate was analyzed five times to give 15 replicate raw data files per growth substrate. LC-MS/MS data files were processed using BioWorks 3.3.1 (Thermo Scientific, Waltham, MA, USA).

Bioinformatics
MALDI-TOF peak mass lists and MS/MS data files were searched in-house using MASCOT (Version 2.2.03) [17] against the functionally annotated B. proteoclasticus genome sequence database (Genbank Accession Numbers CP001810-CP001813). MASCOT search parameters were: one missed tryptic cleavage site; cysteine carbamidomethylation and methionine oxidation as fixed and variable modifications respectively; 50 ppm and 0.5 Da maximum mass tolerance for parent and fragment ions. The peptide false discovery rate (FDR) was determined by calculating the ratio of the number of unique peptides matched to the reverse protein sequences over the number of peptides matched to the reverse and forward protein sequences.
For analysis of LC-MS/MS spectra BioWorks search parameters were optimized to obtain a peptide FDR of less than 1%. Following peptide to protein assignment, the protein expectation value (less than 1.0 × 10 −4 ), peptide cross correlation scores (greater than 1.5 for single charge, 2.5 for double charge and 3.3 for triple charge), and minimum number of matched peptides per protein (minimum 2) were used as filters to obtain an average protein FDR across all technical replicate search results of below 1%. Protein FDR was calculated as the ratio of the number of proteins matched to the reverse protein sequences over the number of proteins matched to the reverse and forward protein sequences, respectively. Proteins were included in the BioWorks identification dataset only if they were identified with statistical significance in at least two of the three biological replicate samples.

Theoretical Proteome
The theoretical 2DE map of the B. proteoclasticus intracellular cytoplasmic proteome ( Figure 1) was dominated by acidic proteins: 79% (2242) of the predicted cytoplasmic proteins possessed a theoretical pI < 7, and 89 of the 90 predicted cytoplasmic oligosaccharide-degrading enzymes had a pI < 6.8 with a cluster of 81 enzymes between pI 4 and 5.6. Ninety percent of the predicted cytoplasmic proteome (2575 proteins) had predicted polypeptide masses between 10 and 150 kDa, and all cytoplasmic oligosaccharidases had a predicted size between 26 and 134 kDa. The group of 802 proteins with a predicted pI value > 6.5 was dominated by hypothetical proteins and mobile elements (transposases), as well as proteins involved in nucleic acid metabolism, protein synthesis, transcriptional regulation, and cell envelope biogenesis. Following initial 2DE using pH 3-10 IEF strips, all further analysis was carried out using narrow range focusing to enhance separation of the acidic proteins.

Identification of Proteins Separated by 2DE
Cytosolic subcellular fractions from cells harvested at mid-log or stationary growth phase were analyzed using 2DE of multiple gels (pI 3-5.6, 5.3-6.5, 6-11 IEF), followed by MALDI-TOF and MALDI-TOF/TOF analysis of tryptic digests of protein spots excised from the gels. Proteins were identified from 481 of 625 excised spots, and were the products of 219 genes. Table 1 summarizes the 17 2DE spots ( Figure 2) from which 15 different enzymes that are predicted to be associated with catabolism of oligosaccharides were identified. All enzymes were identified from the pI 3-5.6 2DE separations. Two protein spots, C1120 and C1147, contained more than one glycosidase. C1147 contained β-galactosidase Bga2B, β-glucosidase Bgl3C and β-mannosidase Man2A. The predicted pI and size of the three enzymes were similar. Spot C1120 contained β-galactosidase Bga35B and β-xylosidase Xyl3A, both of which were also of similar predicted pI and size. Other identified proteins are listed in Supplementary Table S1 including 28 enzymes from intracellular pathways of carbohydrate metabolism and 11 oligosaccharide ABC-transporter associated proteins.   Table 1.

Protein Identification by LC-MS/MS
Tryptic digests of cytosolic extracts from xylan-grown and xylose-grown, mid-log phase cells were digested and analyzed by LC-MS/MS, using three separate cultures (biological replicates) for each condition and each biological replicate analyzed by LC MS/MS five times. A total of 329 non-redundant proteins were identified including 50 proteins involved in carbohydrate metabolism, 20 of which were oligosaccharidases (Table 2). Notably, nine of these had not been identified by the 2DE-based analyses, bringing the total number of oligosaccharidases identified in the cytosol of B. proteoclasticus cells to 24 (15 enzymes identified from 2D gels plus nine new identifications by LC-MS/MS). Similar to the pattern observed in the gel-based analyses, all LC-MS/MS identified enzymes had a predicted pI value within a narrow range between 4.1 and 6. Other identified proteins are listed in Supplementary Table S2. Figure 3 summarizes the functional analysis of all proteins identified either from the 2DE gels or by LC-MS/MS. Carbohydrate metabolism was the major class, comprising 16.2% of the total 394 protein identifications. In accordance with the predicted metabolic pathways of B. proteoclasticus [9], enzymes from the pentose phosphate pathway and enzymes involved in production of butyrate and formate were detected (Supplementary Tables S1 and S2). Other major classes included protein synthesis, transporters, amino acid biosynthesis and energy metabolism.

Functional Domains of the Oligosaccharidases
Pfam, BLASTp and Tigrfam analyses were used to predict carbohydrate active domains in each oligosaccharidase amino acid sequence. Four proteins, β-glucosidase Bgl3A, glycoside hydrolase GH30A, pullulanase Pul13A, and xylosidase/arabinofuranosidase Xsa43A, were each predicted to contain an N-terminal secretory signal peptide (Figure 4). Bgl3A and Pul13A also contained a predicted transmembrane helix within the C-terminal region, while GH30A and Xsa43A contained lipobox motifs at the N-terminus. It is therefore possible that Bgl3A and Pul13A are membrane-anchored proteins and that GH30A and Xsa43A are cell wall lipoproteins. Although cross-contamination of the cytosolic subcellular fraction by proteins released from the membrane or cell wall is possible, it should be noted that none of the four proteins were detected in the previously reported membrane [13] or secreted [10] fractions. Like some of the extracellular polysaccharidases [10], the potentially membrane anchored GH13 pullulanase and secreted GH43 xylosidase/arabinofuranosidase Xsa43A contained carbohydrate-binding modules (CBMs). CBM48 modules mediate enzyme attachment to α-1,4-linked glucose monomers [27], while CBM6 domains have been shown to target the parent enzyme to cellulose, xylooligosaccharides, and mixed linkage glucans [28,29]. The possible functional association of these four proteins with the membrane and cell wall remains to be assessed. Homology to the functionally characterized Bacillus subtilis subsp. subtilis ATCC 6051 XynD [30] suggests that Xsa43A may target intact or partially degraded arabinose containing xylans, hydrolysing (1→2)-αlinked and (1→3)-α-linked L-arabinofuranosyl groups from xylooligosaccharides of varying lengths. It is possible that Xsa43A augments the activity of the secreted Xsa43J [10] to hydrolyse α-linked L-arabinofuranosyl groups from the xylan backbone. GH13 pullulanases (EC 3.2.1.41) (also called debranching enzymes) hydrolyze α-1,6 glycosidic linkages in polymers such as starch, pullulan, and other branched oligosaccharides, generating α-1,4 linked linear oligosaccharides. Although the catalytic domain of GH30A is similar to the GH30 domain found within the Xsa43J [10], without functional analysis it is difficult to predict the catalytic potential of the GH30A enzyme. Functional analysis is also required to predict the catalytic activity of Bgl3A. The remaining 20 identified oligosaccharidases are predicted to be localized to the cytoplasm. This group of enzymes included three members each of the GH2 and GH3 families and single representatives of 11 other GH families ( Figure 5). Two esterase families and a glycosyl transferase were also represented. Several enzymes contained more than one predicted catalytic domain, including all the GH3 family enzymes that contained GH3 and GH3C catalytic domains, and the β-galactosidase Bga2A that contained three catalytic domains. A distinguishing feature of both the GH43 xylosidase/arabinofuranosidase/esterase Xsa43H and the GH10 endo-1,4-β-xylanase/esterase Xyn10D was that each of those proteins contained two putative catalytic domains that were predicted to be active upon different substrates, namely GH and CE activities. Interestingly, among the predicted cytosolic enzymes a CBMX domain was found within the N-terminal region of the GH94 cellobiose phosphorylase, and a CelD domain is predicted in the GH9 cellodextrinase. Both non-catalytic domains are often associated with these types of enzymes, and both may play a role in enzyme/substrate attachment.

Effects of Growth on Xylan or Xylose
To further investigate the potential roles of carbohydrate metabolizing enzymes in the processing of xylan, 2DE was used to quantify the effects of growth substrate and growth phase on the relative abundance of fibrolytic enzymes and transport proteins. Table 3 and Supplementary Figure S1 summarize mid-log phase proteins that differed in abundance between cells grown on xylose or xylan.
Three biological replicates (separate cultures) were analyzed for each substrate and growth condition, and each biological replicate was analyzed on triplicate gels. Four of the identified enzymes were only detected in cells grown on xylan. Two ABC transporter substrate-binding proteins were unique to growth on xylan. Pullulanase (Pul13A) was detected as a pair of adjacent spots at approximately pI 4.2 and 100 kDa, which corresponded closely with the theoretical values for this protein. The more acidic form was also only detected in cells grown on xylan, while the more basic form was 2.8 times more abundant in cells grown on xylan compared to xylose. A further eight protein spots showed greater than two-fold differences (p < 0.05) between growth on xylan compared to xylose, although two of these spots, C733 and C1010, each contained two identified proteins. The most strongly up-and down-regulated proteins in the mid-log phase proteome of xylan grown cells were a subtilisin family serine protease and glutamate synthase GltA, respectively. Positive values denote proteins with increased mean normalized spot volume in xylan grown cells. "Xylan" denotes the protein was uniquely detected in culture medium harvested from xylan grown cells. In these cases it was not possible to calculate a p Value and "n/a" (non-applicable) is entered.
For cells harvested at stationary growth phase, two identified proteins were detected only in cells grown on xylan. They were a sugar ABC transporter substrate-binding protein (abundant spot C1074) and an amino acid ABC transporter substrate-binding protein (minor spot C1135) ( Table 4, Supplementary Figure S2) that were also unique to growth on xylan in mid-log phase cells. The high abundance of the sugar transport protein Bpr_I0182 is in good agreement with its increased abundance in the membrane fraction of B. proteoclasticus grown with xylan compare to fructose [13]. In addition, the more basic pullulanase identified in mid-log phase xylan-grown cells was replaced by a spot of slightly lower mass that was also identified as this protein. Other identified proteins differed quantitatively between cells grown on xylan compared to xylose. Analysis of the differences was complicated by identification of two proteins in each of the spots C704, C785, and C1120. C1120 contained β-galactosidase Bga35A and β-xylosidase Xyl3A and was more abundant after growth in xylan. In contrast, the xylose ABC transporter substrate-binding protein product of Bpr_I1173 (spot C1057) was less abundant in xylan-grown cells. A U62 family peptidase identified in spot C784 (Bpr_I2456) was also more abundant in xylan grown cells. This peptidase was up to 51% identical to many clostridial zinc-dependent proteases and U62 family peptidases that modulate DNA gyrase activity [31]. T, Transporters; ii Expectation value is the statistical probability of the top ranked protein match being a false positive identification, p < 0.05; iii Fold-change was calculated as the ratio of the mean normalized protein spot volumes in the xylan versus xylose growth conditions as stated. Positive values denote proteins with increased mean normalized spot volume in xylan grown cells. "Xylan" denotes the protein was uniquely detected in culture medium harvested from xylan grown cells. In these cases it was not possible to calculate a p Value and "n/a" (non-applicable) is entered.

Predicted Activities and Pathways
The results of this study are consistent with the model of B. proteoclasticus polysaccharide metabolism that has emerged from the genomic [9], and proteomic analyses of the extracellular [10] and membrane compartments [13]. This model proposes that a core set of secreted enzymes target primarily the xylan backbone and are therefore likely to liberate a range of substituted or unsubstituted xylooligosaccharides. One caution is that gene identifications were based on homology with other bacterial sequences, where possible experimental evidence for function is described below. The GH10 endoxylanase Xyn10B that was detected at high abundance in the growth medium of cells grown in the presence of xylan [10] is expected to produce xylooligomers with arabinofuranosyl, O-acetyl, glucuronyl, and ferulic acid substituents [32][33][34][35]. B. proteoclasticus also appears to synthesize several sugar binding protein dependent, ABC-transporter systems that presumably allow it to actively transport these substituted xylooligomers across the cell membrane for further metabolism [10,13]. Such a model requires B. proteoclasticus to possess a range of cytosolic enzymes that are capable of processing substituted xylooligomers to their constituent monomers.
The oligosaccharidases identified in the B. proteoclasticus cytoplasm (Table 5) include enzymes with potential to hydrolyze ferulated glucuroarabinoxylans, acetyl and glucuronic acid substituted sugars, as well as xylobiose and arabinogalactans (Figure 6), all of which are the expected products of extracellular xylan degradation [10]. Other identified oligosaccharide degrading enzymes including a glycogen phosphorylase (consistent with the demonstration of glycogen-like storage polysaccharide in B316 T [9]), α-amylase, pullulanase, and cellobiose phosphorylase that are implicated in the breakdown of storage polysaccharides.     Based upon either direct functional characterization or homology to other bacterial enzymes with known catalytic activities, the diverse range of internalized substituted xylooligosaccharide are expected to be targeted by enzymes including two acetyl-xylan esterases (Est1E and Est2A), two enzymes with arabinofuranosidase catalytic potential (Arf51A and Xsa43H), at least one glucuronidase (Agu67A), and a multi-domain endoxylanase (Xyn10D). Both acetyl-xylan esterases have been functionally characterized [36,37]. Est1E is the only enzyme of its type identified in the B. proteoclasticus proteome and may be an important mediator of the degradation and assimilation of feruloyl xylooligosaccharides. When incubated with esterified synthetic substrates or purified rye-grass hemicellulose, Est1E released ferulic and p-coumaric acids [37]. In addition, a variety of ester linkages were hydrolyzed and phenolic compounds released when Est1E was incubated with birch-wood xylan, suggesting that the enzyme has wide substrate specificity. Est2A was confirmed as having true acetyl-xylan esterase activity [36], and may therefore target only acetylated xylooligosaccharides.
Homology to characterized enzymes from other bacterial species suggests that Arf51A, Agu67A, and Xsa43H are likely to mediate almost complete hydrolysis of xylooligosaccharide substituent groups. Arf51A is homologous to several Clostridial α-L-arabinofuranosidases [38][39][40] that preferentially hydrolyzed arabinose containing polysaccharides such as wheat arabinoxylan over other hemicellulosic substrates. Agu67A is structurally and functionally similar to several well-characterized bacterial GH67 enzymes [41][42][43] that are induced by, and are active upon, methylglucuronoxylan, which suggests that Agu67A is likely to remove 4-O-methylglucuronic acid residues from the backbone of assimilated methyglucuronoxylooligosaccharides. α-D-Glucuronidases cleave the α-1,2-glycosidic linkage between 4-O-methyl-D-glucuronic acid and the xylan backbone, and are vital for the complete degradation of hemicelluloses such as GAX. The C-terminal, GH43 domain of Xsa43H was at 65% identical to C. stercorarium xylosidase/arabinofuranosidase (XylA), which was particularly interesting in that at an optimal pH of 7.0 it exhibited xylosidase and arabinofuranosidase activity within the single GH43 domain [44].
Xyn10D was the only endoxylanase identified in the B. proteoclasticus cytosol and contains an N-terminal GH10 catalytic domain, as well a C-terminal CE10 domain, and the full-length protein is 97% identical to the B. fibrisolvens H17c(SA) XynB, now designated as a strain of B. proteoclasticus [4,45]. When expressed in E. coli, XynB was able to cleave the backbone of xylooligomers close to sites of arabinose substitution, which are known to negatively influence the activity of many endoxylanases upon the xylan backbone [11]. A feature of Xyn10D is that similar to Xsa43H, it contains a novel N-terminal CE10 domain implying that both enzymes might possess novel catalytic activity within the rumen microbiome.
Following the removal of xylooligomer substituents, the β-xylosidase Xyl3A that contains two non-homologous GH3 functional domains ( Figure 5) is expected to complete the hydrolysis of xylobiose and xylooligosaccharides to xylose monomers that can then enter the central metabolic pathways. Xyl3A is 53% identical and 71% similar to Bxl3B from Clostridium stercorarium NCIMB 11754, which hydrolyzes xylose from the non-reducing end of xylotriose, liberates only small amounts of xylose from synthetic-α-arabinofuranoside, and does not release xylose from intact arabinoxylan or arabinose substituted xylooligosaccharides. Xyl3A was the only enzyme of its type identified in the B. proteoclasticus cytosol. Taken together these data suggest that the B. proteoclasticus Xyl3A may act exclusively on the non-reducing ends of short chain xylooligosaccharides and is a vital component of the cytosolic xylooligosaccharide degrading capability of B. proteoclasticus.

Conclusions
Proteomic examination of the B. proteoclasticus cytosolic proteome has provided further insight into the xylan degrading enzyme system that has evolved to enable B. proteoclasticus to degrade and utilize xylan [3,4], and provides a clearer understanding of the metabolic processes involved in plant cell-wall degradation by the fibrolytic rumen microbiota. Efficient xylan utilization requires the synergistic action of a variety of enzymes, including acetyl-xylan-and ferulic acid esterases, α-L-arabinofuranosidases, α-D-glucuronidases, endo-xylanases and β-xylosidases. This study confirms that B. proteoclasticus possesses the enzymes required for the cytosolic degradation of internalized substituted oligosaccharides that are predicted to be derived from extracellular xylan degradation [9,10,13,36,37], and supports the notion that B. proteoclasticus makes an important contribution to ruminant metabolism.
This study also extends our understanding of the enzymes that are likely to play important roles in xylan degradation in the rumen. The use of exogenous fibrolytic enzymes as feed additives is an emerging technology that holds promise as a means of enhancing forage utilization and improving ruminant productivity [46][47][48]. Application of fibrolytic enzymes to rumen forage prior to consumption has resulted in increased voluntary intake, milk production, and average daily weight gain as a consequence of increased forage digestibility [49][50][51][52][53][54][55]. Nonetheless, results are inconsistent and not always positive [56][57][58][59]. Commercially prepared fibrolytic enzymes for ruminant applications are cocktails of crude enzyme extracts that usually contain specified levels of xylanase or cellulase activities and are assessed primarily on their capacity to degrade plant cell walls in vitro [47,60], rather than their specific formulations or the suitability of these formulations for their intended purposes. In particular, it has been suggested that side-chain hydrolyzing enzymes, such as ferulic acid esterases, acetyl-xylan esterases and arabinofuranosidases, should be incorporated [47]. This study has provided a clearer understanding of the specific enzymes that are utilized by a prominent hemicellulose degrading rumen microbe, and may assist in achieving the more rational and targeted design of multi-enzyme products aimed at enhancing the cell wall degradation of rumen forages.