A multifunctional GH39 glycoside hydrolase from the anaerobic gut fungus Orpinomyces sp. strain C1A

Transcriptomics-guided selection of a GH39 enzyme for cloning and characterization

As a part of an extensive transcriptomic analysis of lignocellulosic biomass degradation by the anaerobic fungal isolate Orpinomyces sp. strain C1A (Couger et al., 2015), the most highly transcribed gene annotated as a β-xylosidase was selected for cloning and biochemical characterization. The selected m.21910 transcript (GenBank accession number KT997999) was annotated as member of the GH39 CAZyme family based on the presence of the conserved protein domain pfam01229 (Glyco_hydro_39) family. When strain C1A was grown on different substrates (glucose, Corn Stover, Energy Cane, Switchgrass, and Sorghum), m.21910 constituted 58–84% of the transcriptional activity (i.e., normalized FPKM values) of all GH39 transcripts (n = 9), and 5.7–18.2% of the transcriptional activities of all C1A genes putatively annotated as β-xylosidases (members of GH39 and GH43, n = 41) (Couger et al., 2015). The gene encoding for Bgxg1 protein was previously identified in the genome of strain C1A (GenBank contig accession number ASRE01002650.1, range: 2,346–3,460, see GCA_000412615.1 for whole genome). The ctg7180000059688.1 gene consists of 1,115 bp and no introns (refer to IMG gene ID 2518718918 for a visual representation of the gene, https://img.jgi.doe.gov/cgi-bin/m/main.cgi?section=TaxonDetail&page=taxonDetail&taxon_oid=2518645524). The protein product is predicted to be extracellular and non-cellulosomal, based on the presence of a signal peptide, and the absence of a CBM fungal dockerin domain, respectively.

Bgxg1 sequence analysis and phylogeny

To determine the phylogenetic affiliation of Bgxg1 and the overall topology and global phylogeny of GH39 CAZymes, GH39 β-xylosidase sequences available in CAZY database (http://www.cazy.org/GH39_all.html) (n = 1, 145 total GH39 sequences, retrieved October 28, 2015, edited to remove α-iduronidases and duplicates, resulting in n = 200 β-xylosidases), in addition to Bgxg1, were aligned using Clustal Omega (Sievers et al., 2011). The generated alignment was used to construct a maximum likelihood tree in RAxML (Stamatakis, 2014), which was subsequently visualized and annotated using Mega6 (Sievers et al., 2011; Tamura et al., 2013).

Synthesis, cloning, expression, and purification of Bgxg1 protein

Bgxg1 expression and purification

Ten milliliters of overnight cultures of BL21(DE3)pLysS E. coli cells transformed with pET28a(+)-bgxg1 were used to inoculate 1 L LB broth, containing kanamycin (15 µg/mL) and chloramphenicol (34 µg/mL). The culture was incubated at 37 °C with shaking at 200 rpm until an OD₆₀₀ = 0.6 was reached. Isopropyl-β-D-thiogalactopyranoside (IPTG, 1 mM final concentration) was then added to induce protein production, and the culture was gently shaken at room temperature overnight. Cells were then pelleted by centrifugation (6,000× g, 10 min, 4 °C) and the pellets were collected and stored at −20 °C.

Preliminary small-scale experiments indicated that the protein is expressed in the inclusion body fraction (Fig. S2). Inclusion body extraction was initiated by incubating the cultures in B-Per Cell Lysis Reagent (Thermo Scientific, Grand Island, NY, USA) (10 ml per 500 ml of culture) for 15 min at room temperature with gentle shaking to lyse the cells. The homogenate was centrifuged (10,000× g, 30 min, 4 °C) and the inclusion body extraction procedure (Grassick et al., 2004) was conducted on the cell pellet as follows: the pellet was resuspended in a urea-based inclusion body extraction buffer (20% glycerol, 8 M urea, 50 mM sodium monobasic phosphate, 500 mM sodium chloride, pH 8.0) for 30 min at room temperature with gentle shaking. The homogenate was centrifuged (10,000× g, 30 min, 4 °C) and the resultant supernatant containing target inclusion body proteins was subsequently utilized for refolding and purification procedures.

Recombinant protein refolding was achieved using slow dialysis as previously described (Grassick et al., 2004). In brief, inclusion body extract was incubated with EDTA (1 mM final concentration) and β-mercaptoethanol (100 mM final concentration) for 2 h at room temperature with gentle shaking, transferred to dialysis tubing (NMWL: 12,000–14,000 Da), and placed for 3 h into inclusion body exchange buffer (20% glycerol, 8 M urea, 50 mM sodium monobasic phosphate, 500 mM sodium chloride, 1 mM EDTA, pH 8.0) for removal of the β-mercaptoethanol. The buffer was refreshed and dialyzed for an additional 3 h. The dialysis tubing was then placed into a low-urea refolding buffer (2 M urea, 50 mM sodium monobasic phosphate, 500 mM sodium chloride, 1 mM EDTA, 3 mM reduced glutathione, 0.9 mM oxidized glutathione, pH 8.0) and dialyzed overnight, followed by a no-urea refolding buffer (50 mM sodium monobasic phosphate, 500 mM sodium chloride, 1 mM EDTA, 3 mM reduced glutathione, 0.9 mM oxidized glutathione, pH 8.0) for 36 h.

Following dialysis, the contents of the tubing were centrifuged to remove insoluble, precipitated proteins (15,000× g, 15 min, 4 °C). The supernatant, containing refolded soluble protein, was then exposed to a nickel-nitriloacetic acid (Ni-NTA, 1:1 ratio) slurry (UBPBio, Aurora, CO, USA), packed in a glass frit column (25 × 200 mm, 98 mL volume Kimble-Chase Kontes Flex Column, Vineland, NJ, USA), and allowed to incubate at 4 °C for 1 h on an orbital shaker. Protein purification followed as detailed previously (Morrison, Wright & John, 2012). Samples were concentrated using Amicon Ultra-15 Centrifugal Filter Units (NMWL 30 kDa; Millipore) and protein concentration was determined using a Qubit Fluorimeter (Thermo Scientific) in reference to standard protein concentrations. Protein refolding was checked as activity against PNPX, as described below. An SDS-PAGE gel was run to check protein size and purity, as previously described (Laemmli, 1970; Morrison, Wright & John, 2012).

Biochemical characterization of Bgxg1 (enzyme activity assays)

Bgxg1 modeling

Homology modeling by Iterative Threading ASSEmbly Refinement (I-TASSER) (Roy, Kucukural & Zhang, 2010; Yang et al., 2015; Zhang, 2008), was conducted to generate a three-dimensional model of Bgxg1 using Thermoanaerobacterium saccharolyticumβ-xylosidase (PBD entry 1UHV) as a template. PyMOL was used to align the Bgxg1 structural prediction to that of Thermoanaerobacterium saccharolyticum (PBD entry 1UHV) to examine and speculate the impact of variations in amino acids residue on the enzyme’s active site topology and putative substrate binding capacities (PyMol, 2014).

Bgxg1 phylogenetic affiliation

Phylogenetic analysis grouped all GH39 sequences into 4 phylogenetically-resolved and bootstrap-supported clades (Classes I–IV in Fig. 1). Orpinomyces sp. strain C1A Bgxg1 protein belonged to Class III, forming a well-supported cluster with GH39 proteins from the anaerobic fungus Piromyces sp. strain E2, as well as GH39 proteins from the bacterial genera Clostridium and Teredinibacter (70–74% sequence identities) (Fig. 1). To our knowledge, none of the GH39 proteins within this specific cluster, or in the entire Class III GH39, has been biochemically characterized.

Physiological characterization

SDS-PAGE results show that the Bgxg1 protein is consistent with the predicted size of 42.7 kDa (protein predicted molecular weight is 39.6 KDa + 0.996 kDa linker + 2.101 kDa double histidine tag) (Fig. S3).

The thermal and pH ranges and optima were determined by conducting assays at a range of temperatures and pH’s, as described above. Bgxg1 exhibited activity in a wide range of pH (3–8) and temperatures (25–60 °C), with optimal activity at pH 6 and 39 °C (Figs. 2A and 2B). The thermal and pH stabilities of Bgxg1 were examined by conducting activity assays post-stress (pH or thermal)-incubations as described above. Bgxg1 retained more than 80% of its specific activity post-application of pH stress ranging between 6 and 11 (Fig. 2C), and 60% of its specific activity post application of pH stress of 4, 5, and 12 (Fig. 2C). Further, Bgxg1 retained ≥70% of its specific activity across the broad range of temperature stressors applied (4–70 °C) (Fig. 2D). In addition, exposure to pH stress from 6–11 and temperature stress from 4–70 °C did not produce results that were significantly different from the optimal conditions (p-value > 0.05, 95% confidence interval, Fig. 2).

Substrate specificities and kinetics

To date, all characterized GH39 enzymes exhibit a narrow substrate range (β-xylosidase or α-L-iduronidase) (Table S1). As predicted by sequence analysis, Bgxg1 exhibited β-xylosidase activity (11.5 ± 1.2 U/mg, Table 1), strong as compared to previously reported β-xylosidase activities (Table S1). While having a β-xylosidase activity greater than the majority of the field in Table S1, Bgxg1 does have lower β-xylosidase activity than other anaerobic fungi Neocallimastix frontalis (16 U/mg, Hebraud & Fevre, 1990), Neocallimastix patriciarum (30.4 U/mg, Zhu, Cheng & Forsberg, 1994), Piromyces communis (28 U/mg, Hebraud & Fevre, 1988), and Sphaeomonas communis (27 U/mg, Hebraud & Fevre, 1988) (Table 1, Table S1); however, these are not of the same GH-family. Interestingly, in addition to β-xylosidase activity, Bgxg1 also exhibited strong β-glucosidase (73.4 ± 7.15 U/mg), β-galactosidase (54.6 ± 2.26 U/mg), and weak xylanase (10.8 ± 1.25 U/mg) activities (Table 1), as compared to reported activities from previously characterized enzymes (Tables S2–S4). Our extensive literature review identified 63 enzymes that have been biochemically-characterized to have β-glucosidase activity and, of these, only seven have a reported specific activity higher than that of Bgxg1 (Table S2), including Candida peltata (108 U/mg, Saha & Bothast, 1996), Thermoascus aurantiacus (232 U/mg, Tong, Cole & Shepherd, 1980), and Trichoderma reesei (768 U/mg, Takashima et al., 1998). Similarly, we only identified three β-galactosidase with a reported higher activity than Bgxg1 (Table S3), including Alicyclobacillus acidocaldarius (229 U/mg, Yuan et al., 2008), Bifidobacterium adolescentis (526 U/mg, Hinz et al., 2004), and Thermus aquaticus (1,750 U/mg, Ulrich, Temple & Mcfeters, 1972). On the other hand, the xylanase activity of Bgxg1 is relatively weak, with many previously reported xylanases exhibiting a much higher specific activity (Table S4), such as Aspergillus niger (19 U/mg, 25 U/mg, 35 U/mg, and 48 U/mg, John, Schmidt & Schmidt, 1979) and Trichoderma reesei (46 U/mg, Takashima et al., 1998). Bgxg1 exhibited no detectable exoglucanase, endoglucanase, mannanase, arabinosidase, acetyl xylan esterase, cellobiohydrolase, mannosidase, or α-glucuronidase activities.

In addition to its high β-xylosidase, β-glucosidase, and β-galactosidase specific activities, Bgxg1 exhibited remarkably high affinities towards all examined substrates, with K_m values (calculated via extrapolation through Lineweaver–Burke plot) in the low nM range for PNPG and PNPGal, the low µM range for PNPX (Table 2, Tables S1–S4).

Substrate competition studies

Substrate competition studies were conducted using a variable concentration of an unlabeled substrate (acting as an inhibitor) and a fixed concentration of a chromophore (PNP-based) substrate (Table 3). The results strongly suggest the occurrence of cross-substrate competitive inhibition between xylobiose and cellobiose (Table 3), since the presence of increasing concentrations of a single substrate lowers the specific activity and increases the K_m of the enzyme towards the other substrate, whilst not affecting its V_max (K_m and V_max calculated via extrapolation through Lineweaver–Burke plot). This pattern strongly indicates that a single active site is responsible for the observed activities (Table 3), a conclusion that is in agreement with the lack of identifiable additional domains other than pfam01229 in Bgxg1, as well as with the structural modeling data described below.

In single substrate assays, Bgxg1 was capable of converting cellobiose to glucose and xylobiose to xylose at a very fast rate (Figs. 3A and 3B). This reaction occurs more quickly for xylobiose, as a stable maximal xylose concentration is reached after only 1 min of incubation (Fig. 3B), compared to 15 min for glucose release from cellobiose (Fig. 3A). However, the extent of sugar release at the conclusion of the experiment was higher in cellobiose incubations (Fig. 3A) than xylobiose incubations (Fig. 3B). Competition studies using equimolar concentrations of both substrates revealed the preference of Bgxg1 for xylobiose, since a higher proportion of xylose rather than glucose was detected within the first 15 min of the incubation (Fig. 3C). Nevertheless, the final concentrations of sugars released after 60 min of incubation did not differ when comparing single substrate versus competition experiments (Figs. 3A–3C). Similar to the patterns observed in single substrate assays, Bgxg1 reduced a larger amount of cellobiose to glucose than xylobiose to xylose in competition experiments (Fig. 3C), which is consistent with the higher affinity (lower K_m value) of Bgxg1 for PNPG (12.5 nM) over PNPX (4.85 µM) (Table 2).

Structure activity predictions

The Bgxg1 protein sequence was submitted to I-TASSER for structural prediction by Iterative Threading ASSEmbly Refinement (Roy, Kucukural & Zhang, 2010; Yang et al., 2015; Zhang, 2008) utilizing the β-xylosidase originating from Thermoanaerobacterium saccharolyticum (PBD entry 1UHV) as a template for model creation (Roy, Kucukural & Zhang, 2010; Yang et al., 2004; Yang et al., 2015; Zhang, 2008). Bgxg1 is predicted to have three distinct domains: a catalytic (α∕β)₈ barrel fold domain (position 26–307), a small α-helical domain (position 1–25), and a β sandwich domain (position 308–344) (Fig. S4). Overall, the structure is predicted to contain 11 β-sheets (8 in (α∕β)₈-barrel, 3 in β-sandwich), and 10 α-helices (8 in (α∕β)₈-barrel, 2 in α-domain). The catalytic (α∕β)₈-barrel fold domain is predicted to consist of eight parallel β-sheets (β1–β8), and eight parallel α-helices (α1–α8). Consistent with β-xylosidases of Thermoanaerobacterium saccharolyticum (1UHV) and Geobacillus stearothermophilus (1PX8), the active site pocket of Bgxg1 is predicted to be located on the upper side of the (α∕β)₈-barrel (Czjzek et al., 2005; Yang et al., 2004) (Fig. S4). Alignment and structural predictions identified the conservation of the general acid–base active site residue Glu127 in the C-terminal of β3, as part of the GH39-conserved Asn126-Glu127-Pro128 motif as well as the nucleophilic residue Glu225 in β6 (Fig. 4) (Czjzek et al., 2005; Yang et al., 2004).

Using the predicted model we sought to infer structural differences potentially responsible for the observed relaxed substrate specificities in Bgxg1 by investigating the amino acid conservation patterns between Bgxg1 and all structurally and/or biochemically-characterized β-xylosidases. These enzymes are: Thermoanaerobacterium saccharolyticumβ-xylosidase (Yang et al., 2004), Geobacillus stearothermophilusβ-xylosidase (Bhalla, Bischoff & Sani, 2014; Czjzek et al., 2005), and Bacillus halodurans C-125 protein BH1068 (Wagschal et al., 2008), all of which belong to Class II (Fig. 1), as well as Caulobacter crescentus CcXynB2 (Correa et al., 2012), which belongs to Class I (Fig. 1). All of these enzymes have previously been reported to possess β-xylosidase activity (Bhalla, Bischoff & Sani, 2014; Correa et al., 2012; Czjzek et al., 2005; Wagschal et al., 2008; Yang et al., 2004). We focused on 25 amino acids in two groups: (i) those previously shown to be important for β-xylosidase activity (Czjzek et al., 2005; Yang et al., 2004) (this group includes (in addition to the conserved general acid–base and nucleophilic active sites described above) amino acids providing the tight hydrogen bonding necessary to stabilize the xylosyl-enzyme intermediate formed during the reaction, such as Arg52, His54, Asn159, His228, Tyr230, Glu278, Trp315, Glu322, and Glu323 (locations refer to position in 1UHV)), as well as (ii) those physically interacting with the active site as deduced by the predicted Bgxg1 model (Fig. S3A) (this group includes Val46, Val81, Ile124, Trp125, Gly130, Thr131, Trp132, Phe139, Pro162, Cys163, Tyr164, Ser165, Lys171, His192, Asn242, and Lys247 (locations refer to position in Bgxg1)). Of these 25 amino acids, 15 differed between Bgxg1 and the four other proteins. Five of these 15 amino acids were not conserved amongst any of the five sequences studied and so were not further investigated (Fig. 4). Therefore, 10 distinct differences (eight substitutions and two deletions) between Bgxg1 on one hand and the four biochemically-characterized β-xylosidases on the other were identified (Table 4). These differences that are predicted to exist in or around the active site of Bgxg1 would putatively impact the size, charge, and/or polarity within the active site (Table 4, Fig. S4).

The expanded substrate specificity observed in this study could be a unique trait in Bgxg1, or it could be specific to all GH39 CAZymes of anaerobic fungi (e.g., Class III-C), or to the entire Class III β-xylosidases. Based on the above speculations about the amino acids potentially responsible for Bgxg1 relaxed specificity, we further investigated the conservation of these 10 amino acid changes (Table 4) within class III of GH39 proteins. Bgxg1 (as well as other GH39 proteins encoded in C1A genome), all three GH39 proteins from the Piromyces genome (accession numbers shown in Fig. 1), and all additional sequences from Class III-C belonging to the genera Clostridium and Teredinibacter were found to encode 9 of the 10 observed amino acid substitutions (Table 4). However, within the broader Class III, little similarity in key amino acids was observed between Bgxg1 sequences and β-xylosidases belonging to Class III-A, III-B, or III-D (Table 4). Collectively, these results putatively suggest that the observed relaxed specificity in Bgxg1 could be exclusive to Class III-C β-xylosidases.