Biochemical and kinetic characterisation of a novel xylooligosaccharide-upregulated GH43 β-d-xylosidase/α-l-arabinofuranosidase (BXA43) from the probiotic Bifidobacterium animalis subsp. lactis BB-12

The Bifidobacterium animalis subsp. lactis BB-12 gene BIF_00092, assigned to encode a β-d-xylosidase (BXA43) of glycoside hydrolase family 43 (GH43), was cloned with a C-terminal His-tag and expressed in Escherichia coli. BXA43 was purified to homogeneity from the cell lysate and found to be a dual-specificity exo-hydrolase active on para-nitrophenyl-β-d-xylopyranoside (pNPX), para-nitrophenyl-α-L-arabinofuranoside (pNPA), β-(1 → 4)-xylopyranosyl oligomers (XOS) of degree of polymerisation (DP) 2–4, and birchwood xylan. A phylogenetic tree of the 92 characterised GH43 enzymes displayed five distinct groups (I − V) showing specificity differences. BXA43 belonged to group IV and had an activity ratio for pNPA:pNPX of 1:25. BXA43 was stable below 40°C and at pH 4.0–8.0 and showed maximum activity at pH 5.5 and 50°C. Km and kcat for pNPX were 15.6 ± 4.2 mM and 60.6 ± 10.8 s-1, respectively, and substrate inhibition became apparent above 18 mM pNPX. Similar kinetic parameters and catalytic efficiency values were reported for β-d-xylosidase (XynB3) from Geobacillus stearothermophilus T‒6 also belonging to group IV. The activity of BXA43 for xylooligosaccharides increased with the size and was 2.3 and 5.6 fold higher, respectively for xylobiose and xylotetraose compared to pNPX. BXA43 showed clearly metal inhibition for Zn2+ and Ag+, which is different to its close homologues. Multiple sequence alignment and homology modelling indicated that Arg505Tyr506 present in BXA43 are probably important for binding to xylotetraose at subsite +3 and occur only in GH43 from the Bifidobacterium genus.


Introduction
Prebiotics are commonly non-digestible oligosaccharides which improve the composition of the gut microbiota thus eliciting beneficial health effects (Macfarlane et al. 2006;van den Broek et al. 2008). With a growing market for prebiotic containing foods there is increasing interest in understanding how prebiotics function at the molecular level. Two approved prebiotics are fructooligosaccharides (FOS) and galacto-oligosaccharides (GOS), while β-(1 → 4) linked xylo-oligosaccharides (XOS) with degree of polymerisation (DP) of 2-10 are considered emerging prebiotics (Roberfroid, 2007). Similarly to the approved prebiotics, FOS and GOS, XOS enhance growth of probiotic Bifidobacterium and Lactobacillus species, while suppressing Bacteroides species (Mäkeläinen et al. 2010) and pathogens, e.g. Clostridium species (Rycroft et al. 2001). Notably, XOS can lower the amount of secondary bile acids associated with potential tumour-promoting activity (Moure et al. 2006).
XOS are obtained by hydrolysis of xylans, which are linear β-(1 → 4) linked polysaccharides, typically decorated at the 2-and/or 3-position by mono-or di-substitution with α-L-arabinofuranosyl residues and a substitution pattern varying with the botanical origin (Ebringerova 2006). Probiotic bacteria can possess intracellular xylosidases degrading XOS to D-xylose, which in bifidobacteria is metabolised via the D-fructose-6-P shunt, also referred to as the bifid shunt (Ruas-Madiedo et al. 2005). We found a novel β-D-xylosidase BXA43 of glycoside hydrolase family 43 (GH43) (Cantarel et al. 2009) that was highly up-regulated in Bifidobacterium animalis subsp. lactis  cultures grown with XOS as the sole carbon source and probably plays an important role in the XOS catabolism (Gilad et al. 2010).
The present study focuses on characterisation of the GH43 β-D-xylosidase/α-L-arabinofuranosidase BXA43 from Bifidobacterium animalis subsp. lactis BB-12 (BB-12) which was 10-30 fold upregulated in BB-12 grown on XOS as demonstrated by qPCR and DNA microarray analyses (Gilad et al. 2010). This dual-function GH43 was suggested to be important in the catabolism of XOS taken up by BB-12 to D-xylose. A comprehensive phylogenetic analysis of GH43 as well as a homology model provides a basis for the comparative analysis of the enzymatic properties of BXA43.

Cloning
The BXA43 gene from BB-12 (locus tag BIF_00092, NCBI accession ADC85541), annotated to encode a β-D-xylosidase, was isolated and amplified from chromosomal DNA (provided by Chr. Hansen A/S, Hørshom, Denmark) using upstream primers designed to precede the native Shine-Dalgarno sequence and downstream primers complementary to the 3'terminus and excluding the stop codon; BamHI and XhoI sites are underlined.

Effect of pH and temperature on stability and activity
The stability of 9.7 nM BXA43 was determined at pH 3-10 by incubation at 4°C for 24 h in Britton-Robinson's universal buffers (Britton and Robinson, 1931) followed by measurement of the residual activity towards pNPX (see below). Thermostability of 9.7 nM BXA43 was determined by measuring activity towards pNPX after 10 min incubation at 20-65°C in 50 mM Na-acetate, pH 5.2. The temperature dependence of activity towards pNPX (see below) was determined in the same range. Half-lives of 9.8 nM BXA43 at 40-60°C were determined in 50 mM Na-acetate, pH 5.2 based on residual activity in aliquots (20 μl) removed at appropriate time intervals, followed by immediate addition of ice-cold 50 mM Na-acetate pH 5.2 (20 μl). Samples were kept on ice until assayed. Half-lives were calculated according to t ½ = ln(2) / Ae -Ea / RT , where A is the preexponential factor, E a the activation energy, R the gas constant, and T the temperature in Kelvin.
To determine kinetic parameters, initial rates of hydrolysis were obtained from five time points (0-16 min) and 14 pNPX concentrations (0.25-80 mM) using 7.8 nM BXA43 at the above conditions (in duplicates of four independent reactions). Kinetic parameters, k cat (V max ), K m , and K i were calculated by curve fitting the data to the Michaelis-Menten equation describing substrate in- Activity towards 1% (w/v) birchwood xylan, oat spelt xylan and rye arabinoxylan (all Megazyme) by 78 nM BXA43 was assessed at the above conditions (50 μl) after overnight incubation by thin layer chromatography (TLC; Silica gel 60 F254) of aliquots (1 μl) with arabinose and xylose as reference and developed in 8:2:1 ethylacetate:isopropanol:water for approx. 45 min and dried, followed by staining the carbohydrates with 2% (w/v) orcinol.
Dual-specificity β-D-xylosidase/α-L-arabinofuranosidases are found in GH43 groups II, III, and IV and BXA43 belongs to group IV. The dual specificity β-D-xylosidase/α-L-arabinofuranosidase XylC from B. adolescentis ATCC 15703 is the closest well characterised relative (87% similarity, 79% identity; UniProt accession A1A0H6; Lagaert et al. 2011), but BXA43 is also similar to another characterised group IV β-D-xylosidase/α-L-arabinofuranosidase SXA from Selenomonas ruminantium GA192 (68% similarity, 52% identity; UniProt accession O52575; Jordan et al. 2007). The crystal structure of a complex with xylobiose (PDB: 2EXH, 2EXJ, Brüx et al. 2006) is available for the group IV β-D-xylosidase XynB3 from Geobacillus stearothermophilus T-6 (66% similarity, 50% identity; UniProt accession Q09LX0). A multiple sequence alignment (Additional file 1: Figure S1) shows conservation in BXA43 of the general base and acid catalytic residues Asp 14 and Glu 187 as well as Asp 127 proposed to modulate pK a of the catalytic acid (Nurizzo et al. 2002), and various residues involved in substrate binding, including His 250 at subsite −1 (Brüx et al. 2006). It also reveals that BXA43 Tyr 506 and Arg 505 , which might be involved in the positioning of Tyr 506 , are not conserved in XynB3. These structural  features may be important for the specificity of BXA43 in particular towards xylooligosaccharides.
Recombinant BXA43 was produced in E. coli and purified from the cell lysate in a final yield of 2.2 mg/L culture. It migrated in SDS-PAGE as a single band of approximately 62 kDa in excellent agreement with the theoretical mass of 61,774 Da. BXA43 was stable for 24 h at 4°C at pH 4-9 (Figure 2A). After 10 min at 45°C and 57°C, BXA43 retained 100% and 25% activity, respectively ( Figure 2B). The t ½ values of inactivation of 12 d at 40°C, 25.5 h at 45°C, 141 min at 50°C, 19.5 min at 55°C, and 2.5 min at 60°C indicated a linear correlation as described by the Arrhenius equation with activation energy, E a = 384 kJ/mol ( Figure 2C). These data suggest that BXA43 is very stable in the gut, its natural ecological niche.
BXA43 showed maximum activity towards pNPX at pH 4.0-5.5 and a temperature optimum of 50°C ( Figure 2B). The kinetic parameters as derived from initial rates for hydrolysis of pNPX was K m = 15.6 ± 4.2 mM and k cat = 60.6 ± 10.8 s -1 assuming the presence of uncompetitive substrate inhibition with K i = 29.6 ± 8.5 mM (Figure 3). BXA43 hydrolysed xylooligosaccharides and the specific activity for xylobiose (X 2 ) was 2.3 fold higher than for pNPX and increased further for xylotriose (X 3 ) and xylotetraose (X 4 ) ( Table 1). The activity towards pNPA was 4% of the value for pNPX, whereas paranitrophenyl-β-D-galactopyranoside (pNPG) was not a substrate (Table 1). TLC indicated release by 20 μM BXA43 of xylose from birchwood xylan, but not from oat spelt xylan or rye arabinoxylan after 24 h incubation. The activity of BXA43 for pNPX decreased by 81% in the presence of 1 mM Zn 2+ , but only by 9% in 1 mM Ag + , while Ca 2+ , Mg 2+ or Ni 2+ did not affect the activity.

Discussion
BXA43 of Bifidobacterium animalis subsp. lactis BB-12 from GH43 was produced recombinantly in E. coli BL21 (DE3) and showed K m of 15.6 ± 4.6 mM, k cat of 60.6 ± 10.8 s -1 and the catalytic efficiency k cat /K m of 3.9 s -1 mM -1 similar to k cat of 57 s -1 and k cat /K m of 3.3 s -1 mM -1 reported for β-xylosidase (XynB3) from Geobacillus stearothermophilus T-6 (Brüx et al. 2006). By contrast, β-xylosidase/α-L-arabinofuranosidase from S. ruminantium GA192 has much lower K m = 0.38 mM (Brunzelle et al. 2008), and a β-xylosidase from Bacillus pumilus IPS has K m = 3.9 mM (Xu et al. 1991). Remarkably, V max of BXA43 of 166 U mg -1 was 11-fold higher than V max of 15 U mg -1 obtained for the GH43 enzyme from the thermophile Clostridium stercorarium F-9 (Suryani et al. 2004). BXA43 showed decreased rate of hydrolysis above 18 mM pNPX indicating apparent substrate inhibition (K i = 29.6 ± 8.5 mM), which however, was not a result of product condensation as verified by TLC (data not shown) consistent with no reports of condensation activity of GH43 enzymes. By contrast, the retaining Thermoanaerobacter ethanolicus β-xylosidase/ α-L-arabinofuranosidase from GH3 was subject to substrate inhibition at low substrate concentration of 0.5 mM pNPX (Mai et al. 2000).
As BXA43 has 5°C lower temperature optimum than the group IV enzyme deXA isolated from a compost starter mixture (Wagschal et al. 2009a) the thermal halflives determined of BXA43 (141 min at 50°C; 19.5 min at 55°C; 2.5 min at 60°C) and deXA (630 min at 49°C; 234 min at 51°C; 47 min at 53°C) are comparable. BXA43 was inhibited by 1 mM Zn 2+ and less sensitive to Ag + , while a GH43 β-D-xylosidase from Talaromyces thermophilus was inhibited by both Zn 2+ , Cu 2+ and Hg 2+ (Guerfali et al. 2008). Remarkably, 1 mM Zn 2+ had no effect on activity of the β-D-xylosidase/α-L-arabinofuranosidase from G. stearothermophilus T-6, which has 66% sequence identity and similar specific activity to BXA43, whereas 1 mM Ag + inactivated this enzyme (Shallom et al. 2005). Thus, distinctly different metal ion sensitivity is observed among GH43 enzymes.  BXA43 hydrolysed X 2 , X 3 , and X 4 in line with a suggested model for XOS utilisation in BB-12 that includes uptake via an ATP binding cassette (ABC) oligosaccharide transport system whose components were identified in the secreted (secretome) and the membrane proteomes of BB-12 grown on XOS; BB-12 is incapable of growing on xylose as the sole carbon source (Gilad et al. 2011;. While an increasing specific activity for X 2 , X 3 , and X 4 (Table 1) is seen for BXA43, this was not reported for GbtXyl43A from G. thermoleovorans IT-08 (Wagschal et al. 2009b) and deAX from an uncultured bacterium (Wagschal et al. 2009a) which both belong to group IV and have the highest specific activity on X 2 or X 3 , respectively. This can possibly be explained by the uniqueness of the BXA43 Tyr 506 residue, found only within the Bifidobacterium and Lactobacillus GH43 enzymes. This residue may stack onto substrate at subsite +3 in BXA43, as supported by homology modelling (Figure 4). Interestingly, the group IV XylC from B. adolescentis ATCC 15703 that is hypothesised to play an important role in the efficient conversion of X 2 to xylose (Lagaert et al. 2011), possesses the corresponding tyrosine residue, has slightly higher specific activity on X 2 (23.7 s -1 mM -1 ) than X 4 (17.0 s -1 mM -1 ), and the highest specific activity observed towards xylohexaose (X 6 , 24.6 s -1 mM -1 ). Notably, B. adolescentis ATCC 15703 also harbours a GH120 β-D-xylosidase (XylB), not found in BB-12, which shows a drastic increase in specific activity from X 2 (0.7 s -1 mM -1 ) to X 4 (17.6 s -1 mM -1 , Lagaert et al. 2011).
A phylogenetic tree (Figure 1) generated for characterised GH43 enzymes, as annotated by the CAZy database (Cantarel et al. 2009) of which approximately 35% have been studied and 13 have a known pNPX:pNPA activity ratio (Table 2), has five clusters (groups I − V) (Figure 1) representing distinct substrate specificities, as opposed to a previously reported phylogenetic tree (Qian et al. 2003) that displayed four groups. The two phylogenetic trees, however, are based on different dataset and are analysed with different purposes, therefore their group numbering is not directly comparable. The newly updated tree provides more Figure 4 Model of BXA43 (green) superimposed with the structure of the close homolog XynB3 (grey, 2EXH), showing the two catalytic residues D 14 and E 187 (in blue; BXA43 numbering). Xylotetraose (yellow) was manually fitted from the structure of the complex with BsAXH-m2,3 on Bacillus subtilis (grey, 3C7G) (Vandermarliere et al., 2009) occupying the +1, +2, +3, and +4 subsites and illustrating a predicted +3 subsite involvement of Y 506 (red). Arg 505 in BXA43 may play a role in the positioning of Tyr 506 .