Introduction

This mini review concerns properties and practical aspects involving xylan 1,4-β-D-xylosidase (EC 3.2.1.37), hereafter referred to as β-xylosidase, that are subjects of investigation in the authors’ laboratories, including discovery and improvement of catalytically efficient enzyme to fully deconstruct 1,4-β-xylooligosaccharides to D-xylose monosaccharides at the industrial level for subsequent processes such as fermentation to biofuels and other bioproducts. Therefore, properties and applications will not be addressed for α-D-xylosidase (EC 3.2.1.-; Lovering et al. 2005; Okuyama et al. 2006; Yoshikawa et al. 1994), xylan 1,3-β-xylosidase (EC 3.2.1.72; Umemoto et al. 2008), oligosaccharide reducing-end xylanase (EC 3.2.1.156; Honda and Kitaoka 2004; Fushinobu et al. 2005), and xylose-utilizing glycosynthase (Ducros et al. 2003; Kim et al. 2005; Shaikh and Withers 2008).

Currently, β-xylosidase is used on the industrial scale when it is included in xylanase cocktails for deinking recycled paper (Marques et al. 2003), processing wood pulp (Tsujibo et al. 2001), improving bread dough baking and nutritional quality (Dornez et al. 2007), hydrolysis of bitter xylosylated compounds from grape juice during extraction and liberation of aroma derived from xylosylated compounds of grapes during wine making (Manzanares et al. 1999), and hydrolysis of xylan to d-xylose residues for subsequent reduction to xylitol (Polizeli et al. 2005). For these applications, large quantities of the enzymes are prepared from fungal and bacterial sources such as Aspergillus niger, Trichoderma reesei, and Bacillus sp. among others (Polizeli et al. 2005). Whereas the latter processes can be large in scale, they are dwarfed by the potential use of β-xylosidase, other hemicellulases, and cellulases in processes that hydrolyze lignocellulosic biomass to monosaccharides for fermentation to biofuels as well as conversion to other value-added products (Werpy and Petersen 2004). Therefore, a major concern is the discovery of cost-efficient β-xylosidase (and other enzymes) which includes positive enzyme attributes such as high k cat and k cat/K m, low affinity for monosaccharide inhibitors, good stabilities to pH and temperature, low levels of adsorption/inactivation by biomass feedstocks, and low cost of enzyme production.

Fundamental properties of β-xylosidase

By definition of the Nomenclature Committee of the International Union of Biochemistry and Molecular Biology (http://www.chem.qmul.ac.uk/iubmb/), xylan 1,4-β-d-xylosidase catalyzes hydrolysis of single xylosyl residues from the non-reducing end of 1,4-β-d-xylooligosaccharides, including 1,4-β-d-xylobiose. According to the CAZy database (http://www.cazy.org/; Cantarel et al. 2009), β-xylosidase is currently grouped into six glycoside hydrolase families (GH3, GH30, GH39, GH43, GH52, and GH54) based on amino acid sequence similarities (Table 1). Based on structural folds, GH30 and GH39 are grouped into clan GH-A, having fold (α/β)8, GH43 is in clan GH-F, having fold fivefold β-propeller, and the other families are unassigned. The rapid rate of three-dimensional structure determinations by X-ray crystallography will undoubtedly fill this knowledge gap with time. High-quality crystals that diffract beyond 2 Å are often obtained with glycoside hydrolases, including β-xylosidase (Brunzelle et al. 2008), under simple conditions (e.g., polyethylene glycol as precipitant). Site-directed mutagenesis has been used to create inactive enzyme for the preparation of crystals of β-xylosidase in complex with substrate (Brüx et al. 2006), which provided structural models that are most useful for understanding the organization of the active site. As an alternative to substrates, β-xylosidase has been co-crystallized with an amino alcohol inhibitor which binds in the active site binding pocket (Brunzelle et al. 2008). For many glycoside hydrolases including GH43 β-xylosidase, small positional changes occur upon binding substrate according to structural overlays (Brunzelle et al. 2008). This constancy portends well for obtaining good homology models for newly discovered β-xylosidase. Conveniently, such three-dimensional models can be readily obtained by submittal of an amino acid sequence to automated services such as the Swiss-Model Workspace (Arnold et al. 2006).

Table 1 CAZy classification of 1,4-β-xylosidase
Full size table

Thorough examination of glycoside hydrolase reactions and experimental means for their study are available (Sinnott 2007; Rempel and Withers 2008; Yip and Withers 2006; Withers 2001; Rye and Withers 2000). Importantly, for certain mechanistic and enzyme performance considerations, β-xylosidases (and most other glycosidases) catalyze hydrolysis of substrates by either of two canonical reaction mechanisms (Koshland 1953). Both reaction types employ two amino acid side chains containing carboxyl groups (Asp or Glu). In the inversion mechanism (or single displacement mechanism), the β-anomeric configuration of the substrate’s non-reducing end xylosyl glycone residue is, upon hydrolysis, released as α-xylose. This hydrolysis mechanism entails a single transition state on the enzyme, implicating addition of the nucleophilic water molecule (activated by general base Asp) to C1 of the glycone in concert with proton delivery (by general acid Glu) to the aglycone alkoxide leaving group, facilitating its departure from C1 of the glycone (Fig. 1a). In the retaining mechanism (or double displacement mechanism), the substrate’s nonreducing end xylosyl glycone residue is, upon hydrolysis, released as β-xylose. This β-xylosidase mechanism comprises two transition states (Fig. 1b), one TS for the xylosylation step forming an ester bond between the enzyme nucleophile (Asp or Glu) and C1 of the glycone with simultaneous proton delivery (by general acid Glu) to the aglycone alkoxide leaving group, promoting its departure and one TS for the dexylosylation step in which the ester bond is hydrolyzed.

Fig. 1
figure 1

Reaction mechanisms of β-xylosidase. a Inverting. b Retaining

Full size image

Unlike the rapidly interconverting anomeric conformations of arabinofuranose, for example, d-xylose has the relatively lengthy half-life for mutarotation of 1 h at 25°C (Jordan et al. 2007b), and this makes 1H NMR the method of choice for the determination of the anomeric conformation of the hydrolysis product d-xylose. In addition, two other methods have been reported: HPLC separation of the two anomers of d-xylose (Braun et al. 1993) and a coupled enzyme system employing α-stereospecific d-xylose isomerase and sorbitol dehydrogenase (Kersters-Hilderson et al. 1976).

Identification of the catalytic base and catalytic acid for β-xylosidase operating through an inverting mechanism (GH43) has relied heavily on X-ray structural information, which implicates two of the three Asp and Glu residues of the active site (Brüx et al. 2006; Brunzelle et al. 2008). Mutation of the three highly conserved residues to nonreactive residues (Gly or Ala) produces inactive (>99.99%) enzyme (Shallom et al. 2005; Jordan et al. 2007b, 2009). However, one of these, an Asp, is in a poor position for serving as either catalytic base or acid; it is considered to aid catalysis by raising the pK a of the catalytic acid and in stabilizing the position of the glycone moiety of substrate. The X-ray structures implicate the other active site Asp residue as the catalytic base and the active site Glu as the catalytic acid. Biochemical evidence supports the identities of the catalytic acid and base. Mutation of the catalytic acid to either Gly or Ala changes the k cat/K m pH profile so that the pK a 7 attributed to the catalytic acid is lost (Shallom et al. 2005; Jordan and Braker 2010). Loss of activity by mutation of the catalytic base from Asp to Gly can be partially reversed by including azide in reaction mixtures; presumably, the azide serves as a nucleophile adding to substrate C1 or it serves as a catalytic base to activate an active site water to serve as the nucleophile (Shallom et al. 2005).

Identification of the enzyme nucleophile and catalytic acid/base of retaining β-xylosidase has depended largely on X-ray structures. In the case of a retaining GH39 β-xylosidase, substantiating biochemical evidence supporting the identity of the enzyme nucleophile has been accomplished by using a suicide substrate, containing a fluorinated xylosyl residue that can xylosylate the enzyme but has a very slow rate of dexylosylation (Vocadlo et al. 1998). The xylosylated amino acid residue was identified using mass spectrometry. Identification of the catalytic acid/base residue of a retaining GH39 β-xylosidase was secured by mutagenesis of the likely candidate Glu to Ala (Bravman et al. 2001). The inactivated mutant enzyme was confirmed as the general acid/base by experiments demonstrating burst kinetics, pH shift, and azide rescue.

Retaining glycosidases exhibit, to varying extents, transglycosylation activity in which a nucleophile (e.g., sugar), in competition with water, attacks C1 of the glycosyl group of the glycosylated enzyme intermediate, whereas inverting glycosidases lack this activity. This is an important distinction since transglycosylation could seriously diminish the performance efficiency of retaining β-xylosidase in its function of catalyzing hydrolysis reactions. Among β-xylosidase families, only GH43 operates via an inverting mechanism (Table 1). It is likely that characterization of new β-xylosidases will uncover inverting enzymes that belong to existing GH families or define new ones.

Practical considerations: discovery of efficient β-xylosidase from nature

While β-xylosidase broadly impacts society in its numerous applications that include glycosynthase activity, our chief concern is the use of β-xylosidase for complete hydrolysis of xylooligosaccharides to d-xylose monosaccharide for the multiple uses mentioned or referred to in the “Introduction.” For hydrolysis, highly active enzyme acting on xylooligosaccharides (k cat, k cat/K m) is desired. Also, it is desirable to identify highly active β-xylosidases for operating in different niches of practice which encompass enzymes with diverse properties including tolerance of pH, temperature, and other factors.

To begin with, we survey the literature for β-xylosidase acting on xylobiose and xylooligosaccharides (Table 2) since these are the enzyme’s natural substrates and the most relevant from a commercial application perspective. Despite its importance to the oeuvre, immediately obvious is the shortness of the list of β-xylosidase with kinetic parameters reported for substrate 1,4-β-d-xylobiose (X2); all but three of these were determined in recent years by the authors’ laboratories who will continue to add to the list. There are dozens of enzymes with kinetic parameters reported for the artificial, relatively labile substrate 4-nitrophenyl-β-d-xylopyranoside (4NPX) without kinetic parameters for X2, and three of these are listed in Table 2. Of the latter, two have qualitative experimental results (TLC) indicating that they act on X2 (Smaali et al. 2006), demonstrating that the enzyme fulfills a major requirement for being categorized as β-xylosidase (EC 3.2.1.37). The other enzyme acts on 4NPX, but not X2; it has been designated as an arabinofuranosidase (EC 3.2.1.55; Wagschal et al. 2007).

Table 2 Kinetic parameters of β-xylosidase acting on xylobiose and 4NPX
Full size table

Among those with kinetic parameters listed for acting on 4NPX and X2 (1,4-β-d-xylobiose), there is the qualitative correlation that all enzymes that act on X2, act on 4NPX. Quantitatively, it is a different matter as there is only slight correlation between k cat and k cat/K m values of β-xylosidase acting on X2 and 4NPX: ratios of \( k_{\rm{cat}}^{{\rm{X}}2}/k_{\rm{cat}}^{4{\rm{NPX}}} \) vary from 0.035 to 5.8. Ratios of (k cat/K m)X2/(k cat/K m)4NPX vary from 0.0034 to 2.0.

β-xylosidase from Selenomonas ruminantium (hereafter referred to as SXA) is the most active β-xylosidase acting on X2 currently reported (Table 2). Kinetic parameters k cat and 1/K m of SXA acting on X2 and 4NPX have different patterns of dependence on pH and temperature (Fig. 2a), whereas the kinetic parameter k cat/K m, which reports free enzyme, has similar patterns (Fig. 2a). Relative kinetic parameters of \( k_{\rm{cat}}^{{\rm{X}}2}/k_{\rm{cat}}^{4{\rm{NPX}}} \), (k cat/K m)X2/(k cat/K m)4NPX and (1/K m)X2/(1/K m)4NPX vary considerably with temperature and pH (Fig. 2b); the dependence of (k cat/K m)X2/(k cat/K m)4NPX on pH changes the least with pH because it reports mainly free enzyme. Therefore, 4NPX is a poor surrogate for X2 in predicting catalytic activities of multiple β-xylosidases acting on X2. Similarly, 4NPX is a poor surrogate for predicting X2 activity of β-xylosidase under changing conditions of pH and temperature.

Fig. 2
figure 2

pH and temperature dependence of kinetic parameters of SXA acting on 4NPX and X2. a Kinetic parameters were determined at 25°C and varied pH (Jordan et al. 2007b, 2009; Jordan and Braker 2007) or at pH 5.3 and varied temperature (Jordan and Braker 2009). Values of pH-dependent parameters are relative to the value at pH 7.0 and 25°C: \( k_{\rm{cat}}^{4{\rm{NPX}}} = 12.7\,{{\hbox{s}}^{ - 1}} \), \( {k_{\rm{cat}}}/K_{\rm{m}}^{4{\rm{NPX}}} = 34.7\,{\hbox{m}}{{\hbox{M}}^{ - 1}}\,{{\hbox{s}}^{ - 1}} \), and \( 1/K_{\rm{m}}^{4{\rm{NPX}}} = 2.65\,{\hbox{m}}{{\hbox{M}}^{ - 1}} \) (Jordan et al. 2007b, 2009) and \( k_{\rm{cat}}^{{\rm{X}}2} = 135\,{{\hbox{s}}^{ - 1}} \), \( {k_{\rm{cat}}}/K_{\rm{m}}^{{\rm{X}}2} = 62.1\,{\hbox{m}}{{\hbox{M}}^{ - 1}}\,{{\hbox{s}}^{ - 1}} \), and \( 1/K_{\rm{m}}^{{\rm{X}}2} = 0.464\,{\hbox{m}}{{\hbox{M}}^{ - 1}} \) (Jordan 2008). Values for the temperature-dependent parameters are relative to the value at 25°C and pH 5.3: \( k_{\rm{cat}}^{4{\rm{NPX}}} = 32.4\,{{\hbox{s}}^{ - 1}} \), \( {k_{\rm{cat}}}/K_{\rm{m}}^{4{\rm{NPX}}} = 42.2\,{\hbox{m}}{{\hbox{M}}^{ - 1}}\,{{\hbox{s}}^{ - 1}} \), and \( 1/K_{\rm{m}}^{4{\rm{NPX}}} = 1.3\,{\hbox{m}}{{\hbox{M}}^{ - 1}} \) (Jordan et al. 2007b, 2009) and \( k_{\rm{cat}}^{{\rm{X}}2} = 188\,{{\hbox{s}}^{ - 1}} \), \( {k_{\rm{cat}}}/K_{\rm{m}}^{{\rm{X}}2} = 87.3\,{\hbox{m}}{{\hbox{M}}^{ - 1}}\,{{\hbox{s}}^{ - 1}} \), and \( 1/K_{\rm{m}}^{{\rm{X}}2} = 0.463\,{\hbox{m}}{{\hbox{M}}^{ - 1}} \) (Jordan 2008). b Ratios of kinetic parameters (4NPX/X2) are plotted versus temperature. Parameters were determined at pH 5.3. c Ratios of kinetic parameters (4NPX/X2) are plotted versus pH. Parameters were determined at 25°C

Full size image

Clearly, there is a need to extend the list of Table 2 to include kinetic parameters of additional β-xylosidases acting on X2 and xylooligosaccharides. Facilitating this goal, there are five convenient, quantitative methods for determining d-xylose produced from β-xylosidase acting on X2 and larger xylooligosaccharides that have been reported. X2, 1,4-β-d-xylotriose (X3), 1,4-β-d-xylotetraose (X4), 1,4-β-d-xylopentaose (X5), 1,4-β-d-xylohexaose (X6), and the other reagents and equipment needed for the determinations are available commercially. Unlike 4NPX in which reaction progress can be monitored continuously and discontinuously, reaction progress of β-xylosidase acting on xylooligosaccharides is currently limited to discontinuous monitoring. Historically, the first of these methods for d-xylose quantification is the use of D-xylose isomerase and sorbitol reductase as coupling enzymes (Van Doorslaer et al. 1985). The second method relies on pyranose oxidase and peroxidase as the coupling enzymes (Wagschal et al. 2005). The third method uses colorimetric detection of reducing sugars (Rasmussen et al. 2006). The fourth method depends on HPLC separation of reaction product d-xylose which is monitored using pulsed amperometry and electrochemical detection (Jordan 2008). The fifth method uses xylose mutarotase and β-d-xylose dehydrogenase as coupling enzymes (Jordan and Braker 2009).

An example of performance considerations

Hydrolysis of lignocellulose to substituent monosaccharides can be achieved by chemical means. However, high yields of monosaccharides are withheld, in part, due to dehydration reactions that generate furfural from d-xylose and hydroxymethylfurfural from d-glucose (Larsson et al. 1999; Luo et al. 2002). The aldehydes and other side products are toxic to the fermenting organisms (Palmqvist et al. 1999; Klinke et al. 2004), preventing efficient ethanol production; furthermore, they are corrosive to the reactors. For these and other considerations, it is desirable to employ milder chemical methods for pretreatment of the biomass that produce lower levels of the toxins, but are sufficient to expose cellulose and hemicellulose fibers for subsequent saccharification by enzymatic means. Although the literature is replete with daunting statements on the requirement of many enzyme functions for achieving complete hydrolysis of xylan to free its constituent monosaccharides, it is clear that the requirements depend on the lignocellulosic materials and on the pretreatment method used prior to enzymatic hydrolysis. For sparsely substituted arabinoxylan, depleted of ester substituents following pretreatment, four enzymes (two xylanases, a β-xylosidase and an arabinofuranosidase) are sufficient to completely liberate the d-xylose content (Adelsberger et al. 2004). For such material, rather than being just one hemicellulase of many, β-xylosidase would in fact carry the greatest work load in terms of the number of glycosidic bonds cleaved, as well as relieving product inhibition of xylanase activities, making its performance of primary importance.

Inspection of Table 2 where k cat values are presented in descending order reveals that SXA is the best catalyst known for hydrolysis of X2, and β-xylosidase from Bacillus pumilus is second best. Likewise, larger xylooligosaccharides are accepted as substrate better by SXA than by the B. pumilus β-xylosidase. Thus, from k cat and k cat/K m values determined for SXA at pH 5.3 and 25°C (Jordan 2008) and k cat and k cat/K m values determined for the B. pumilus β-xylosidase at its pH optimum of pH 7.15 at 25°C (Van Doorslaer et al. 1985), kinetic parameters are higher for SXA than the B. pumilus β-xylosidase by factors of 10 (X2), 32 (X3), 31 (X4), 20 (X5), and 27 (X6) for k cat and factors of 14 (X2), 16 (X3), 15 (X4), 10 (X5), and 13 (X6) for k cat/K m. \( k_{\rm{cat}}^{{\rm{X}}2} \) for SXA acting on xylobiose is 185 s−1 at 25°C (Table 2) and 390 s−1 at 35°C (Jordan and Braker 2009). This puts SXA in the company of other highly active glycosidases acting on their natural substrates with k cat values of 1,000 s−1 (Lairson and Withers 2004; Konstantinidis et al. 1993). However, SXA is clearly slower than certain glycosidases such as the Bacillus cereus β-amylase acting on maltohexaose (k cat = 1,200 s−1 at 25°C; Yamaguchi et al. 1996), suggesting that there may be room for improvement of SXA k cat. Suitable for serving roles in biomass saccharification processes, SXA is stable at pH 5 and above and at temperatures of 50°C and below, and it is only weakly inhibited by ethanol (Jordan et al. 2007a).

Early on, it was recognized that a potential Achilles’ heel of SXA performance in saccharification processes is its relatively high affinities for d-xylose (pH-independent K i = 3.5 mM) and for d-glucose (pH-independent K i = 17.7 mM), which could accumulate to high levels in saccharification reactions (approaching 1 M) and inhibit SXA catalysis (Jordan and Braker 2007). Xylanase and β-xylosidase can operate synergistically in depolymerizing xylan to d-xylose (La Grange et al. 2001), presumably through a mechanism where xylanase releases small xylooligosaccharides from xylan, which serve as good substrates for β-xylosidase, and β-xylosidase catalyzes their conversion to d-xylose. Xylanase is inhibited by xylobiose (K i = 4.8 mM; Williams et al. 2000), so inhibition of the β-xylosidase by d-xylose accumulation could lead to the accumulation of xylobiose and inhibition of xylanase. Other relevant enzymes such as cellulase and β-glucosidase (Xiao et al. 2004) exhibit inhibition by glucose, making this a general problem affecting multiple steps of the saccharification process. SXA and other lignocellulosic enzymes may function effectively in simultaneous saccharification and fermentation (SSF) processes in which saccharification enzymes and the fermenting organism operate simultaneously in a single reactor since potentially inhibiting monosaccharides are consumed by the fermenting organism as they are generated by the saccharification enzymes. However, even in the SSF process, performance of SXA could benefit from lower affinities to monosaccharides.

In addressing the perceived problem, a series of experiments were carried out to define the binding properties of d-xylose, d-glucose, and other monosaccharides to the active site of SXA (Jordan and Braker 2007). Assisting the analysis of monosaccharide binding, SXA has the simplest active site possible for an efficient glycoside hydrolase by comprising two subsites: subsite −1 for binding substrate glycone, subsite +1 for binding substrate aglycone, and for substrates larger than X2, the reducing end extends into solvent (Jordan et al. 2007b; Brunzelle et al. 2008).

In the resulting model of Fig. 3 (Jordan and Braker 2007; Jordan et al. 2009), catalysis (Jordan et al. 2007b) and binding of monosaccharides is governed by pK a 5 and pK a 7, the former for the catalytic base (D14) and the latter for the catalytic acid (E186). When both carboxylate groups are protonated (D14HE186H), SXA is catalytically inactive and binds monosaccharides weakly. In its monoprotonated form (D14−E186H), SXA is catalytically active and can bind certain monosaccharides (e.g., d-glucose and d-xylose) to either subsite −1 or subsite +1, providing single occupation, and certain monosaccharides (e.g., l-arabinose and d-ribose) can bind to both subsite −1 and subsite +1, providing double occupation. Also, in its monoprotonated state, certain monosaccharides (e.g., d-xylose and d-ribose, but not d-glucose) can bind in subsite −1 simultaneously with occupation of subsite +1 by the xylosyl moiety of 4NPX (and presumably that of X2, but not the arabinofuranosyl moiety of 4-nitrophenyl-α-l-arabinofuranoside (4NPA). In its doubly deprotonated state (D14−E186−), SXA is catalytically inactive. However, its active site binds monosaccharides (e.g., d-glucose, d-xylose, L-arabinose, and d-ribose) and the glycone moieties of substrates 4NPX and X2, but not 4NPA within subsite +1. 4NPA does not bind to subsite −1 or subsite +1 of D14−E186− SXA.

Fig. 3
figure 3

Binding of monosaccharide inhibitors (I and X), amino alcohols (A), and substrate (S) to SXA (E). At the top, a model of the two-subsite active site of SXA in its dianionic state (D14−E186−) is shown. In the equilibrium model, E is shown in two pH-dependent forms: catalytically active, monoanionic form D14−E186H and catalytically inactive, dianionic form D14−E186−. Amino alcohols bind only to the D14−E186− form of enzyme and only to subsite −1. E_S and E_I designations indicate that the substrate and inhibitor bind in subsite +1 and not in subsite −1 of the D14−E186− enzyme, whereas both subsites +1 and −1 can be occupied by I, X, and S in the D14−E186H enzyme. Ligands do not bind to the diprotonated form of enzyme (D14HE186H). The model is based mainly on Jordan and Braker (2007) and Jordan et al. (2009)

Full size image

The model of Fig. 3 includes the binding of amino alcohols (e.g., ethanolamine, Tris, and triethanolamine) to the active site of SXA. The cationic form of the amines has affinity only for subsite −1 of the doubly deprotonated active site (D14−E186−). The neutral amine binds 1,000-fold more weakly. Small amino alcohols such as ethanolamine (but not the larger triethanolamine) bind to subsite −1 and accommodate simultaneous binding of monosaccharides in subsite +1. Thus, by saturating subsite −1 with a small amino alcohol, binding affinity of a monosaccharide to subsite +1 can be determined. Even though amino alcohols bind in subsite −1 of the SXA active site, inhibition patterns can appear competitive, noncompetitive, or uncompetitive, depending on the substrate and size of the amino alcohol (Jordan et al. 2009). Amino alcohols have been known as active site inhibitors of glycoside hydrolases for over half a century, and they provide a powerful means for determining the binding modes of monosaccharides, particularly within the relatively simple active site of SXA and its mutants (see below). As an aside, there are many examples in the recent literature where amino alcohols (e.g., Tris) have been used, unwarily, as pH buffers in determination of kinetic and equilibrium binding parameters of glycosidases (including β-xylosidase), and this has seriously diminished the value of the work.

Another method that has been developed for the analysis of SXA that applies specifically to β-xylosidases is comparison of kinetic parameters (particularly k cat) for β-xylosidase acting on 4NPX and 4NPA (Jordan and Li 2007; Fan et al. 2010; Jordan and Braker 2010). Many β-xylosidases, particularly those of GH43, catalyze hydrolysis of 4NPA, 4NPX, and xylooligosaccharides, owing to the three-dimensional structural similarity of their glycones (Jordan and Li 2007) and the tenfold faster spontaneous hydrolysis rate of 4NPA over that of 4NPX (Jordan and Braker 2009). The energy required to flatten a portion of pyranose rings to their reactive conformation exceeds that needed for furanose rings by about 7 kcal/mol (Durette and Horton 1971). Therefore, mutants of SXA that cause a decrease in the ratio \( k_{\rm{cat}}^{4{\rm{NPX}}}/k_{\rm{cat}}^{4{\rm{NPA}}} \) can assign the native residue as positively influencing the distortion state of the xylosyl glycone toward hydrolysis. SXA mutations that increase the ratio \( k_{\rm{cat}}^{4{\rm{NPX}}}/k_{\rm{cat}}^{4{\rm{NPA}}} \) can assign the native SXA residue as positively influencing other factors of catalysis such as C1 migration, nucleophilic water activation, and protonation of the leaving group. This provides an independent method for determining whether an SXA mutation influences subsite −1 or subsite +1 of the active site.

Having some understanding of monosaccharide binding to SXA and experimental tools for their study in place, but no a priori conceptions on engineering lower inhibitor affinity, we set out to prepare mutants of SXA by error-prone PCR in a directed evolution effort (Yuan et al. 2005) and search for SXA variants with lower affinities for d-xylose and d-glucose (Fan et al. 2010).

Modification of efficient β-xylosidase for improved performance

Directed evolution selection of SXA with improved K i values for d-xylose was achieved via a two-tier high-throughput screen where the active/inactive primary screen comprises an Escherichia coli transformant, whole cell spectrofluorometric assay of general utility that identifies active β-xylosidases capable of hydrolyzing 4-methylumbelliferyl xylopyranoside or the arabinosyl cogener. The cell-free secondary screen selects for improved 4NPX hydrolysis activity in the presence of d-xylose; the A 0 (activity without inhibitor)/A i (activity with inhibitor) of normalized primary screen selectants provides the selection data (Fan et al. 2010). Using this strategy, the single-site mutation of SXA (W145G) was selected for characterization. For the SXA variant, \( K_{\rm{i}}^{{\rm{D - xylose}}} \) is improved threefold from the wild-type 6.5 to 22 mM for W145G. \( K_{\rm{i}}^{{\rm{D - glucose}}} \) was raised twofold from the wild-type 49 to 91 mM for W145G. Counteracting the performance improvement with respect to inhibitors are the similar values for \( k_{\rm{cat}}^{{\rm{D - xylobiose}}} \) and the threefold larger value of \( K_{\rm{m}}^{{\rm{D - xylobiose}}} \) of W145G-SXA (\( k_{\rm{cat}}^{{\rm{D - xylobiose}}}155\,{{\hbox{s}}^{ - 1}} \), \( K_{\rm{m}}^{{\rm{D - xylobiose}}}4.5\,{\hbox{mM}} \)) compared with that of wild-type SXA (\( k_{\rm{cat}}^{{\rm{D - xylobiose}}}174\,{{\hbox{s}}^{ - 1}} \), \( K_{\rm{m}}^{{\rm{D - xylobiose}}}1.6\,{\hbox{mM}} \)). Thus, the mutation offers no overall performance advantage for SXA acting on xylooligosaccharides in the presence of d-xylose and d-glucose. Interestingly, W145 resides 6 Å from the active site and has contacts with two residues of subsite +1 and one of subsite −1. By using methods described in the previous section, it was found that wild-type SXA and W145G have similar K d values for triethanolamine (specific ligand for subsite −1), W145G has a threefold larger K i for d-xylose than wild-type SXA in the presence of saturating triethanolamine (so d-xylose can only bind in subsite +1), and the W145G mutation causes a twofold increase in the ratio \( k_{\rm{cat}}^{4{\rm{NPX}}}/k_{\rm{cat}}^{4{\rm{NPA}}} \). All three of these effects of W145G point to influences on subsite + 1 (and not subsite −1) in lowering the affinity for d-xylose (and by inference d-glucose). As this is the first report to date on engineering β-xylosidase in the laboratory for improved hydrolase performance, definition of process and biomass-specific selection targets such as improved k cat and thermal stability, coupled with the availability of a straightforward primary screen, should facilitate the execution of future β-xylosidase engineering campaigns.

β-Xylosidase has more commonly been the subject of “synergy engineering” which involves optimizing a mixture of several GH’s with differing linkage hydrolysis specificities, as well as differing heteromeric enzyme quaternary structures (Wen et al. 2009). A high throughput method for lignocellulosic digestibility (Chundawat et al. 2008) has recently been used to optimize six core GH activities including β-xylosidase for hydrolysis of ammonia fiber expansion pretreated corn stover (Gao et al. 2010), and a minimal enzyme cocktail including β-xylosidase has been reported for wheat arabinoxylan (Sørensen et al. 2007). In efforts to create single polypeptides containing GHs with multiple linkage hydrolysis specificities, chimeric β-xylosidase fusion proteins have been created, including bifunctional enzymes having xylanase and either α-arabinofuranosidase activity or β-xylosidase activity (Fan et al. 2008), trifunctional enzymes with and without carbohydrate binding domains having xylanase, α-arabinofuranosidase, and β-xylosidase activities (Fan et al. 2009a), and a chimera possessing xylanase, endoglucanase, α-arabinofuranosidase, and β-xylosidase activities (Fan et al. 2009b). Such chimeras were shown to maintain parental levels of catalytic activities and synergistic properties, and they may be particularly useful for GH expression in-plant (recently reviewed by Taylor et al. 2008) since a single gene can encode multiple catalytic and binding domains of known stoichiometry. In a somewhat removed area of study, an example of anthropogenic selective evolution and potential refining of function in response to the advent of agriculture has recently been hypothesized for a β-xylosidase of a host-specialized plant pathogen (Brunner et al. 2009).

Examples of other GH enzymes that have been recent subjects of protein engineering efforts for practical applications include a high-activity barley α-amylase (Wong et al. 2004), an α-amylase active on raw starch (Wong et al. 2003), several examples for shifting the pH optima and pH stability of xylanases (Umemoto et al. 2009; Chen et al. 2001; De Lemos et al. 2005; Fenel et al. 2006; Liu et al. 2009), decreased affinity of an endoxylanase for inhibitory cereal proteins (Bourgois et al. 2007), β-glucosidase thermostability enhancement (Lopez-Camacho et al. 1996), low-temperature catalysis for a β-glucosidase (Lebbink et al. 2000), creation of a thermostable fungal class II cellobiohydrolases (Heinzelman et al. 2009), and improved catalytic efficiency of an endo-β-1,4-glucanase (Lin et al. 2009).

Conclusions

β-xylosidase is a critical component of industrial enzyme-based processes for the utilization of renewable biomass feedstocks, yet at present, few examples exist of the characterization of the enzyme acting on xylooligosaccharides, its natural substrates. With an avalanche of DNA sequence information in hand and more deposited regularly, there are many hundreds to thousands of putative β-xylosidase genes known with larger numbers sure to follow. Particularly intriguing are putative β-xylosidases arising from genome sequencing projects of organisms inhabiting diverse environments such as extremophiles. However, in order for us to tap this biotic wealth of genetic information, it is critical going forward that widespread β-xylosidase discovery efforts include the determination of kinetic parameters for the enzyme acting on its natural substrates (perhaps by using the currently available methods referenced above) so that the breadth of β-xylosidase catalytic capability generated by natural processes is genuinely sussed out. In addition to finding β-xylosidases with high k cat and k cat/K m values acting on xylooligosaccharides, β-xylosidases with lower affinities for d-xylose and d-glucose and better stabilities versus extremes of pH and temperature could be uncovered. For example, the β-xylosidase from Scytalidium thermophilum is reported to be uninhibited by 200 mM d-xylose (Zanoelo et al. 2004), though there is currently no information on its k cat and k cat/K m values when acting on xylooligosaccharides. In executing more extensive surveys of β-xylosidases, it is anticipated that in addition to the potential discovery of commercially relevant β-xylosidases straightaway, natural substrate catalytic parameters coupled with three-dimensional structures and amino acid sequence information will enable and inform future efforts in the area of single-activity engineering of β-xylosidase, which curiously is yet in its infancy for this key enzyme needed for enzymatic deconstruction of biomass.