Directed evolution of GH43 β-xylosidase XylBH43 thermal stability and L186 saturation mutagenesis

  • Sanjay K. Singh
  • Chamroeun Heng
  • Jay D. Braker
  • Victor J. Chan
  • Charles C. Lee
  • Douglas B. Jordan
  • Ling Yuan
  • Kurt Wagschal

DOI: 10.1007/s10295-013-1377-0

Cite this article as:
Singh, S.K., Heng, C., Braker, J.D. et al. J Ind Microbiol Biotechnol (2014) 41: 489. doi:10.1007/s10295-013-1377-0


Directed evolution of β-xylosidase XylBH43 using a single round of gene shuffling identified three mutations, R45K, M69P, and L186Y, that affect thermal stability parameter Kt0.5 by −1.8 ± 0.1, 1.7 ± 0.3, and 3.2 ± 0.4 °C, respectively. In addition, a cluster of four mutations near hairpin loop-D83 improved Kt0.5 by ~3 °C; none of the individual amino acid changes measurably affect Kt0.5. Saturation mutagenesis of L186 identified the variant L186K as having the most improved Kt0.5 value, by 8.1 ± 0.3 °C. The L186Y mutation was found to be additive, resulting in Kt0.5 increasing by up to 8.8 ± 0.3 °C when several beneficial mutations were combined. While kcat of xylobiose and 4-nitrophenyl-β-d-xylopyranoside were found to be depressed from 8 to 83 % in the thermally improved mutants, Km, Kss (substrate inhibition), and Ki (product inhibition) values generally increased, resulting in lessened substrate and xylose inhibition.


Glycosyl hydrolase Directed evolution Gene shuffling Thermal stability Protein engineering 


The half-life of a glycosidic bond in nature is on the order of 5 × 106 years [32], underlying the ubiquity of cellulose and hemicelluloses as plant cell wall structural components and key repositories of biologically-fixated carbon. After cellulose, hemicelluloses are the second most abundant plant cell wall polymer type; they are heterogeneous in nature as far as monosaccharide constituents, main-chain and branched-chain monosaccharide linkage types, linkages between hemicelluloses and lignin, and further modifications such as acetylation and feruylation. One of the most common hemicelluloses is arabinoxylan, composed mainly of a xylose backbone with arabinose branch points, and abundantly present in agricultural crop residues including maize and all cereal grasses [5].

Glycosyl hydrolases (GH’s), both ubiquitous and unending in variety like their substrates, catalyze glycosyl polymer deconstruction with such efficiency that they comprise some of the best enzyme catalysts known. A combination of enzyme types, primarily GH’s, are necessary for complete arabinoxylan biomass enzymic deconstruction; these include β-d-xylosidases (EC, responsible for hydrolysis of xylooligosaccharides, predominately xylobiose (X2) to xylose [8, 9]. The β-xylosidases XylBH43 [27] and SXA [13], used for molecular breeding in this study, are members of glycosyl hydrolase family 43 (GH43), whose members include endo- and exo-α-l-arabinases (EC, α-l-arabinofuranosidases (EC, xylanases (EC, and galactan 1,3-β-galactosidases (EC ( [3]. SXA and XylBH43 are among the top five most efficient GH43 xylosidases reported for the hydrolysis of natural substrate X2 [16]. However, the GH43 xylosidases studied to date lose activity precipitously above 50 °C [16], while in biomass saccharification reactors, a higher thermal stability than that of the natural enzymes thus far encountered may be desired for faster reaction times, reduced matrix viscosity, and lower microbial contamination [6]. In nature, thermal stability is selected for directly by organisms pioneering thermally extreme environments. In established populations, the evolution of protein function likely occurs mostly under selection pressure for function within a temperature range, while also evolving indirectly for thermal stability since this confers increased protein structural stability. The latter in turn enables and promotes evolution by allowing proper folding, despite significant mutation [2]. Thus, evolution-based protein engineering of properties other than thermal stability may benefit by engineering thermally improved variants as starting materials.

Strategies to obtain thermally improved enzymes, which are often most effective when combined [6], include directed evolution [24], rational engineering, and screening thermophilic organisms. We show here that screening a small, ~3,840-member library created by a single round of gene shuffling resulted in identifying three amino acid residues and a cluster of mutations near loop-D83 that significantly affect thermal stability. Furthermore, we combined some of the mutations, resulting in an 8.8 ± 0.5 °C increase in Kt0.5. Also identified were mutants with altered pH and inhibitor profiles, potentially desirous characteristics for xylosidase use in mechanistic studies and commercial biomass saccharification.

Materials and methods

DNA shuffled library

The DNA sequence of XylBH43 (PDB ID 1YRZ) was synthesized (Genscript USA Inc., Piscataway, NJ) with internal NdeI and XhoI restriction sites, removed as described previously [7], and subcloned into the pET29b(+) expression plasmid (EMD Chemicals, Inc., Gibbstown, NJ). For use in DNA shuffling, the DNA sequence of SXA (PDB ID 3C2U) was optimized by inspection for maximal identity with that of XylBH43, with the constraint of maintaining the native amino acid sequence (Supplementary Fig. S1), and was then synthesized and subcloned into pET29b(+). DNA shuffling was performed as described [24, 36], with some modifications. Parental DNA was amplified using primers 5′-CTCAGCTTCCTTTCGGGCTTTGTT-3′ and 5′-TGTGAGCGGATAACAATTCCCCTCTAG-3′, which correspond to regions of the pET29b(+) vector that are upstream and downstream of the NdeI and XhoI restriction sites. We mixed 5 μg of DNA from each parent, and subjected them to DNAse I digestion at 37 °C using supplied buffer in the presence of 1 mM Mg2+ and 0.16 U DNAse I in a final volume of 100 μl. The reaction was allowed to proceed for 14 min, and then a 20-μl aliquot was removed and quenched by mixing with 5 μl 500 mM EDTA on ice every 2 min thereafter. The polymerase chain reaction (PCR) reassembly reaction mixture contained 30 μl of fragments (100–200 bp), 5 μl of 10× Pfu DNA polymerase buffer, 2 μl of 10 mM dNTP mix, and 2.5 U of Pfu DNA polymerase, in a final volume of 50 μl. Reactions were performed as follows: 96 °C for 1.5 min, 35 cycles of denaturation at 94 °C for 30 s, nine hybridization steps separated by 3 °C from 65 to 41 °C for 1.5 min each, elongation step 72 °C for 4 min, and the final extension at 72 °C for 10 min. Fragments were amplified using the primers 5′-AGAAATAATTTTGTTTAACTTTAAGAAGGAGATATACATATG-3′ and 5′-GTGGTGGTGGTGGTGCTCGAG-3′, which contain the NdeI and XhoI restriction sites, respectively. In this step, 5 μl of reassembled product was amplified using 5 U of Pfu DNA polymerase, 2 μl of 10 mM dNTP mix, and the supplied buffer in a 100 μl reaction. The PCR protocol was: 95 °C for 2 min; 25 cycles of 1 min denaturation at 95 °C, hybridization step at 55 °C for 1 min, elongation at 72 °C for 4 min, and a final extension step at 72 °C for 10 min. DNA was purified using a 0.8 % agarose gel and Zymo Gel DNA purification kit (Zymo Research Corp., Irvine CA), and subcloned into the NdeI and XhoI restriction sites of the pET29b(+) expression plasmid. Electroporation was performed using electrocompetent E. coli BL21(DE3) cells according to the manufacturer’s protocol (Lucigen Corp., Middleton, WI, USA).

L186 saturation mutagenesis

We constructed a gene library encoding variants containing all the possible amino acids at position 186, following the methods described by Pattanaik et al [20]. Two primer pairs (Supplementary Table 1) were designed to randomize position 186 in the nucleotide sequence. PfuUltra High-Fidelity DNA polymerase (Agilent Technologies, USA) was used in sPCR to minimize random point mutations. PCR using pET29b(+) XylBH43-L186 as a template was performed with a Peltier Thermal Cycler (MJ Research, USA) using the following program: preheat 94 °C for 2 min; 16 cycles of 94 °C for 30 s, 55 °C for 1 min and 68 °C for 8 min; followed by incubation at 68 °C for 1 h. The resulting randomized PCR products were purified using Wizard SV Gel and a PCR clean-up system (Promega, USA), and then treated with DpnI (New England Biolabs, USA) for 1 h to eliminate the parental plasmid. The DpnI enzyme was then heat inactivated at 75 °C for 15 min. The resulting plasmid library was transformed into chemically competent TB1 cells. Plasmids were isolated from the overnight cultures of randomly selected colonies using Wizard Plus SV miniprep kit (Promega, USA) and sequenced to verify the mutation. Plasmids encoding variants with all 20 amino acids at position 186 were identified from 90 randomly selected bacterial colonies.

Thermal stability library screening

The XylBH43/SXA shuffled library was plated onto LB agar containing 30 μg/ml kanamycin (LBkan) using 22 × 22 cm Q-trays (Molecular Devices LLC, Sunnyvale, CA), and incubated overnight at 30 °C. A Q-pix colony picker was used to transfer ~3,800 colonies to Greiner Fluorotrac 200 384-well microplates containing LBkan and the Novagen Overnight Express™ autoinduction system. A whole-cell, in vivo 1° screen was performed using the substrate 4-methylumbelliferyl-β-d-xylopyranoside (MUX), as described previously [28], and ~1,300 clones with active enzyme expression were rearrayed into 96-well hit plates. Two rounds of a 2° screen were performed using crude enzyme lysates prepared using the hit plates as described [7], with a thermal challenge at temperatures of 55 °C for 1 h in round 1, and 57 °C for 1 h with an additional 45 min at 65 °C in round 2, to obtain 10 unique clones for further testing. For hit verification, we performed protein expression and purification from the 2° screen hits using Ni–NTA affinity chromatography followed by size-exclusion chromatography, as previously described [7]. Finally, we determined the temperature at which activity is halved over a 1 h incubation period (Kt0.5), as described below.

Kt0.5 determination

Hydrolytic activity rates before and after thermal treatment were assessed by measuring the rate of p-nitrophenyl leaving the group appearance, using a wavelength of 400 nm and 4 mM 4NPX substrate. Enzyme concentration was adjusted prior to the thermal challenge, such that initial rates were normalized to ~50 μM/min at 25 °C. A thermocycler (MJ Research, Watertown MA) was used to establish a temperature gradient of 20–25 °C, depending on the enzyme being tested, and the selected enzyme was incubated at various temperatures for 1 h in an assay buffer consisting of 100 mM phosphate, pH 6.5, and 1 mg/ml bovine serum albumin (BSA). After the thermal challenge, the temperature was lowered to 0–4 °C for 10 min, and then allowed to refold at 30 °C for at least 10 min. Kt0.5 was calculated by fitting the data to Eq. 1 (see “Equations”), where a is an exponential term, t is the temperature, and Kt0.5 is the temperature where p is half of P.

Activity versus pH profile

The effect of pH on Vmax for 4NPX hydrolysis was determined using buffers containing 1 mg/ml BSA, with 100 mM succinate being used for pH 3–7, 100 mM phosphate for pH 6.5–8, and 100 mM pyrophosphate for pH 7.5–9. Endpoint assays were performed in quadruplicate using 32.5 nM enzyme, 6.4 mM 4NPX; the reactions were induced at 25 °C for 15 min, and then quenched by adding an equal volume of Na2CO3 before spectrophotometric analysis at 400 nm.

Xylobiose hydrolysis analysis and kinetics

Xylobiose (X2) was obtained from Wako Chemicals (Richmond, VA). All other reagents were of high quality. Water was purified by a Milli-Q Academic A10 unit (Millipore; Billerica, MA). A DX500 Dionex HPLC system equipped with an ED40 electrochemical detector (pulsed amperometry) was used for oligosaccharide separation and detection (Dionex; Sunnyvale, CA). Reactions used to calculate kcat and Km values contained varied concentrations of X2 (1.20–48.4 mM) in 100 mM sodium succinate, pH 6.0 (ionic strength = 0.3 M), in a final volume of 0.3 mL, and were at 25 °C for all enzymes tested, except for XylBH43 and L186K. These two were also tested at elevated temperatures corresponding to their respective Kt0.5 values of 51 and 59 °C. Reactions were initiated with 7 μl of appropriately diluted enzyme. Progress was monitored discontinuously by quenching 150 μl of the reaction with 150 μl 200 mM sodium phosphate buffer, with the pH adjusted so that the final quenched reaction was pH ~10.7, followed by HPLC analysis. The time zero reaction mixtures contained no enzymes. Product analysis was performed on a Dionex HPLC. Separation of the products and reactant was achieved using a CarboPac PA-1 column with a separation step (15 mM NaOH), followed by a column washing step (250 mM sodium acetate in 100 mM NaOH) and a column reequilibration step (15 mM NaOH). Data were fitted to Eq. 2. Reactions used to calculate KssX2 values were performed as described previously, where an enzyme-coupled spectrophotometric assay is used for xylose quantitation [26]. The reaction conditions were 25 °C in 100 mM phosphate buffer, with a pH of 6.5, 1 mg/ml BSA, and 1.25–1.75 nM enzyme, and contained eight different X2 concentrations ranging from 1.88–120 mM in a final volume of 30 μL. Data were fitted to Eq. 3.

Enzyme kinetics, and inhibition constants using 4NPX

Michaelis–Menten kinetic parameters were obtained by continuous monitoring of hydrolyzed nitrophenyl chromophores, as previously described [25], using a reaction buffer containing 100 mM sodium phosphate, 0.1 % BSA, pH 6.5 and 25 °C, and 4NPX concentrations of 80–9,600 μM. Monosaccharide inhibition constants were determined, as previously described [7], using 4NPX and xylose concentrations of 0, 75, 150, and 300 mM, and the data were fitted to either Eq. 4 or Eq. 5.


$$p = \frac{P}{{1 + \left( {\frac{t}{{K_{t}^{0.5} }}} \right)^{a} }}$$
$$\frac{v}{{E_{\text{T}} }} = \frac{{k_{\text{cat}} \times S}}{{K_{\text{m}} + S}}$$
$$\frac{v}{{E_{\text{T}} }} = \frac{{k_{\text{cat}} \times S}}{{K_{\text{m}} + S\left( {1 + \frac{S}{{K_{\text{ss}} }}} \right)}}$$
$$\frac{v}{{E_{\text{T}} }} = \frac{{k_{\text{cat}} \times S}}{{K_{\text{m}} \left( {1 + \frac{I}{{K_{\text{i}} }}} \right) + S\left( {1 + \frac{S}{{K_{\text{ss}} }}} \right)}}$$
$$\frac{v}{{E_{\text{T}} }} = \frac{{k_{\text{cat}} \times S}}{{K_{\text{m}} \left( {1 + \frac{I}{{K_{\text{i}} }}} \right) + S}}$$

Data were fitted to equations using either the computer program Grafit (Erithacus Software Ltd., Surrey, UK) or GraphPad Prism 5 (GraphPad Software, LaJolla, CA, USA). Symbol definitions are the following: ν is the observed initial (steady-state) rate of catalysis; ET is the total enzyme concentration; kcat is the turnover number of catalysis; S is the substrate concentration; Km is the Michaelis constant; I is the inhibitor concentration; Ki is the dissociation constant for I from EI; Kss is the dissociation constant for S from ESS (substrate inhibition/activation); p is the determined parameter at a single temperature; P is the temperature-independent value of the parameter; t is the temperature; Kt0.5 is the temperature where p is half of P; a is an exponential term.


Shuffled library construction and screening

The native DNA sequences of SXA and XylBH43 exhibit 56 % identity, while family shuffling [4] is typically performed using genes with >60 % identity to increase the cross-over frequency during recombination [35]. Thus, to maximize overall DNA homology and thereby increase recombination events [24], SXA was optimized for shuffling with XylBH43 by aligning the DNA sequences and changing the individual SXA codons to match XylBH43 codons, with the constraint that the amino acid sequence must be retained. Using this strategy, the overall DNA sequence identity with XylBH43 increased by 17 %, from 56 to 73 % for optimized SXA (Supplementary Fig. 1) used to generate a DNA family shuffled library. About 3,840 library members were initially winnowed using an in vivo, active/inactive library purification 1° screen, wherein genotype and phenotype remain conveniently linked [28]. This screen selects for a combination of sufficient protein expression coupled with enzyme activity acting on MUX; in our study, it resulted in the identification of ~1,300 active clones (~33 % active) that were then rearrayed into 96-well hit plates. After the thermal challenge, 10 clones were selected for hit verification entailing protein expression and purification, followed by determination of Kt0.5. Hit verification confirmed five XylBH43 variants (4B9, 2B10, 5F1, 6F11, and 10C4) with ΔKt0.5 values that were 3.0 ± 0.3 to 5.6 ± 0.4 °C higher than XylBH43, and containing from 5–11 amino acid changes compared to XylBH43 (Table 1). These variants comprise 15 different amino acid substitutions in total from SXA into XylBH43, resulting from a single round of gene shuffling.
Table 1

Thermal stability hit amino acid changes and L186Y additivity

Shuffled hit/ΔKt0.5 (°C)a

XylBH43 amino acid changes

Addition of L186Y mutant/ΔKt0.5 (°C)a

4B9/3.0 ± 0.3

5: V22A; C78D; H82A; T85K; Y87W

4B9:L186Y/6.8 ± 0.1

2B10/4.6 ± 0.6

7: V22A; M69P; N70D; C78D; H82A; T85K; Y87W

2B10:L186Y/8.8 ± 0.5

5F1/4.7 ± 0.1

11: V22A; S54A; S55H; T58S; M69P; N70D; C78D; H82A; T85K; Y87W; H419N

5F1:L186Y/7.7 ± 0.01

6F11/2.8 ± 0.2

6: V22A; R45K; K48V; C78D; Q185A; L186Y

Not applicable

10C4/5.6 ± 0.4

6: M69P; N70D; H82A; T85K; Q185A; L186Y

Not applicable

aXylBH43 Kt0.5 = 51.0 ± 0.3 °C

Thermal stability Kt0.5 determinations

Site-directed mutagenesis was used to create the corresponding single mutation species for all 15 mutations in Table 1, except S54A and S55H, which were tested as the double mutant (Table 2). When tested singly, three mutations significantly affected thermal stability compared to wild-type XylBH43 (Kt0.5 = 51.0 ± 0.3 °C): M69P ΔKt0.5 = 1.7 ± 0.3 °C, L186Y ΔKt0.5 = 3.2 ± 0.4 °C, and R45K ΔKt0.5 = −1.8 ± 0.1. The L186Y mutation, having the largest individual positive ΔKt0.5 value, was present in the shuffled hits 6F11 and 10C4. Shuffled hit 4B9 contained the fewest number of mutations (five), comprised largely of a cluster of four mutations (C78D, H82A, T85K, and Y87W) near a two-residue (DG) hairpin loop, termed loop-D83 (Fig. 1), that in combination appeared to increase Kt0.5 by ~3 °C, but when tested singly did not significantly affect Kt0.5. Shuffled hits 2B10 and 5F1 had in common with 4B9 the loop-D83 mutations, and additionally contained the M69P mutation. Site-directed mutagenesis creation of 4B9:L186Y, 2B10:L186Y, and 5F1:L186Y (Table 1) for additivity analysis resulted in Kt0.5 increases of 3.8 ± 0.3, 4.2 ± 0.8, and 3.0 ± 0.1 °C, respectively, due to the addition of the L186Y mutation compared to a Kt0.5 increase of 3.2 ± 0.4 °C for L186Y in isolation.
Table 2

Site-directed mutant thermal stability

Site-directed mutant

ΔKt0.5 (°C)a

Site-directed mutant

ΔKt0.5 (°C)a


0.6 ± 0.3


0.0 ± 0.6


−1.8 ± 0.1


−0.5 ± 0.2


−0.8 ± 0.1


0.6 ± 0.1


−0.3 ± 0.2


−0.3 ± 0.5


0.2 ± 0.1


−0.1 ± 0.5


1.7 ± 0.3


3.2 ± 0.4


−0.4 ± 0.3


0.5 ± 0.2


1.9 ± 0.4


aXylBH43 Kt0.5 = 51.0 ± 0.3 °C

Fig. 1

Location of mutations affecting ΔKt0.5, with structure representation created using PyMol (Schrödinger LLC)

Saturation mutagenesis of L186 resulted in all variants being active (Table 3). Most L186 changes increased Kt0.5, with L186K showing the largest increase (8.1 ± 0.3 °C). Four mutations lowered Kt0.5, with reductions ranging from −0.42 ± 0.2 °C (L186V) to −1.9 ± 0.1 °C (L186I). The Lys mutation was introduced into the corresponding position of the parental enzyme SXA (SXA:Y184K), resulting in reduced thermal stability; ΔKt0.5 = −2.4 ± 0.1 °C (data not shown).
Table 3

L186 saturation mutagenesis thermal stability

L186 mutant

ΔKt0.5 (°C)a

L186 mutant

ΔKt0.5 (°C)a

Lys (K)

8.1 ± 0.3

Tyr (Y)

3.2 ± 0.4

Gln (Q)

6.7 ± 0.2

Gly (G)

3.2 ± 0.2

Asn (N)

5.9 ± 0.2

Ala (A)

2.2 ± 0.2

His (H)

5.7 ± 0.3

Thr (T)

1.7 ± 0.2

Cys (C)

5.2 ± 0.2

Trp (W)

1.2 ± 0.2

Ser (S)

5.0 ± 0.2

Leu (L)


Glu (E)

4.4 ± 0.1

Val (V)

−0.4 ± 0.2

Phe (F)

4.4 ± 0.2

Asp (D)

−0.9 ± 0.2

Arg (R)

4.4 ± 0.3

Pro (P)

−1.5 ± 0.2

Met (M)

4.2 ± 0.1

Ile (I)

−1.9 ± 0.1

aXylBH43 Kt0.5 = 51.0 ± 0.3 °C

pH curves

Concerning the pH optima of 4NPX hydrolysis, the acidic limbs of the pH curves of the XylBH43 variants are shifted ~0.5–1.0 unit closer to the acidic side relative to XylBH43 (Fig. 2). In particular, the mutants 10C4 (curve not shown), L186E (curve not shown), and L186R maintain >80 % maximal activity at pH 5, whereas XylBH43 activity is <60 % of the maximal activity at this pH. L186R also has a particularly broad pH optimum (~2 pH units) from ~pH 5.5–7.5 (Fig. 2). The basic limbs of five of the mutants (L186V, D, P, I, and T) are significantly shifted toward the acidic side (Fig. 2, circled). They have <40 % relative activities at pH 8, and all of these except L186T are also the only variants with Kt0.5 values lower than wild-type L186 (Table 3).
Fig. 2

pH curves at 25 °C for 4NPX hydrolysis by XylBH43 and select variants. Filled square xylBH43, filled triangle 2B10, filled circle L186R

Kinetic parameters acting on X2

The thermal stability hits (Table 1), the L186Y mutant (Table 2), the L186Y additivity constructs (Table 1), and the top six most thermally stable L186 site-directed mutants (Table 3) were evaluated for X2 catalysis at pH 6 and 25 °C (Table 4). The determined kcatX2 values were seen to decrease by a factor of 0.17–0.92, while the KmX2 values increased for all variants tested except L186Q, for which KmX2 decreased by ~½; the L186Q variant also had the largest observed kcatX2 decrease of 83 %. The determined KssX2 parameters (substrate inhibition) were increased for all variants except 2B10. For variant L186H, the lack of appreciable substrate inhibition at the highest X2 concentrations tested (Fig. 4b) led to poor fitting of this parameter using Eq. 3 (KssX2 = 361 ± 467 mM).
Table 4

Kinetic parameters for catalyzed reactions acting on X2


kcatX2 (s−1)a

kcatX2/KmX2 (s−1 mM−1)a

KmX2 (mM)a

KssX2 (mM)b

XylBH43 (25 °C)

91.9 ± 1.4

17.3 ± 0.3

5.3 ± 0.2

130 ± 10

XylBH43c (51 °C)

267 ± 4.1

30.7 ± 1.1

8.7 ± 0.4


L186K (25 °C)

54.1 ± 0.9

6.5 ± 0.2

8.3 ± 0.3

700 ± 160

L186Kd (59 °C)

262 ± 4.6

11.4 ± 0.3

23 ± 0.9



15.8 ± 0.2

6.0 ± 0.3

2.6 ± 0.1



73.4 ± 1.3

3.1 ± 0.1

23.5 ± 0.9



49.7 ± 0.6

1.2 ± 0.01

40.3 ± 0.8

360 ± 470


73.4 ± 1.4

9.5 ± 0.3

7.8 ± 0.4



60.1 ± 0.6

6.7 ± 0.2

9.0 ± 0.3



68.8 ± 0.7

8.9 ± 0.2

7.7 ± 0.2

330 ± 30


73.6 ± 05

10.8 ± 0.2

6.8 ± 0.1

260 ± 60


64.2 ± 1.7

5.5 ± 0.3

11.7 ± 0.8



73.5 ± 0.7

10.3 ± 0.2

7.2 ± 0.2

140 ± 30


56.4 ± 0.2

5.6 ± 0.04

10.2 ± 0.1



84.2 ± 0.5

14.2 ± 0.2

5.9 ± 0.1

200 ± 20


72.6 ± 0.8

7.6 ± 0.2

9.6 ± 0.3



60.4 ± 0.7

5.9 ± 0.1

10.2 ± 0.3

470 ± 80


54.2 ± 0.3

5.4 ± 0.1

10.0 ± 0.1

320 ± 170

ND not determined

aReactions were in 100 mM sodium succinate, pH 6.0, I = 0.3 M, 25 °C; the analysis was performed using HPAEC/PAD, and the data were fitted using Eq. 2

bReactions were in 100 mM sodium phosphate, pH 6.5, 25 °C; the analysis was performed using a spectrophotometric enzyme-coupled assay, and the data were fitted using Eq. 3

cReactions were in 100 mM sodium succinate, pH 6.0, I = 0.3 M, 51 °C; the analysis was performed using HPAEC/PAD, and the data were fitted using Eq. 2

dReactions were in 100 mM sodium succinate, pH 6.0, I = 0.3 M, 59 °C; the analysis was performed using HPAEC/PAD, and the data were fitted using Eq. 2

For XylBH43, the kcatX2 increased 2.9-fold to 267 ± 4 s−1 when determined at its Kt0.5 (51 °C), compared to at 25 °C, and KmX2 increased 1.6-fold to 8.7 ± 0.4 mM, resulting in kcatX2/KmX2 nearly doubling to 30.7 ± 1.1 s−1 mM−1 (Table 4). For mutant L186K, the kcatX2 determined at its Kt0.5 temp (59 °C), instead of at 25 °C, increased 4.9-fold to 262 ± 5 s−1, and KmX2 increased 2.8-fold to 23 ± 0.9 mM, such that kcatX2/KmX2 also nearly doubled to 11.4 ± 0.3 s−1 mM−1 (Table 4).

Kinetic parameters acting on 4NPX

Kinetic parameters of 4NPX hydrolysis and inhibition by xylose were determined for L186H, L186K and L186Y (Table 5; Fig. 4a). The kcat4NPX values ranged from 25 to 79 % of those seen for XylBH43. The Km4NPX, Kixylose, and Kss4NPX values are larger for all variants, up to 2-fold for Km4NPX and Kixylose. The Kss4NPX values are larger by ~3-fold for L186Y and L186K, and no substrate inhibition was observed for L186H at the highest 4NPX concentration employed.
Table 5

Kinetic parameters of XylBH43 wild-type and selected mutant-catalyzed reactions acting on 4NPX


kcat4NPX (s−1)

kcat4NPX/Km4NPX (s−1 mM−1)

Km4NPX (mM)

Kixylose (mM)

Kss4NPX (mM)


17.9 ± 1.3

61.7 ± 13.5

0.3 ± 0.1

30.2 ± 5.0

11.5 ± 2.4


14.1 ± 0.9

34.4 ± 6.3

0.4 ± 0.1

38.5 ± 5.1

31.4 ± 9.6


6.4 ± 0.5

9.8 ± 2.0

0.7 ± 0.1

59.3 ± 9.0

31.5 ± 12.3


4.4 ± 0.1

9.3 ± 0.9

0.5 ± 0.1

58.0 ± 7.2


Reactions were in 100 mM sodium phosphate, pH 6.5, 0.1 % BSA, 25 °C; data were fitted using Eq. 4

aN/A = not applicable to the 4NPX concentrations tested; data were fitted using Eq. 5


In gene shuffling of closely-related enzymes, multiple mutations can be tolerated, leading to relatively high library quality [4, 35]. Indeed, one of the selectants (5F1) contained numerous (11) amino acid mutations. Screening a small, ~3,840 member library in a single round of family shuffling identified three individual mutations (R45K, M69P, L186Y) (Table 2) and a cluster of four mutations near loop-D83 that significantly affected thermal stability. Other recent examples of GH43-enzyme engineering are broadening the substrate acceptance of an arabinofuranosidase to include endo-xylanase activity [19], and lowering inhibitor affinities of β-xylosidase SXA [7, 15].

The domain architecture of GH 43 enzymes consist mainly of two types: either a single, α/β barrel catalytic domain, or an N-terminal catalytic domain combined to a similar-sized C-terminal β-sandwich domain with carbohydrate binding properties [34]. XylBH43 and SXA are of the latter type, and all of the mutations in the hits occurred in the catalytic domain, except H419N on the β-sandwich non-catalytic domain (Fig. 1). Figure 3 shows mutation locations and the amino acid alignment for XylBH43, SXA, and six other closely-related GH43 enzymes (45–55 % identity), consisting of the most efficient GH43 xylosidases reported to date for xylooligosaccharide hydrolysis, with kcatX2 values ranging from 40 s−1 for the Bacillus pumilus enzyme (Bp), to the highest reported kcatX2 of 420 s−1 for the Lactobacillus brevis enzyme (Lb) at 25 °C [16]. Interestingly, the mutations occurred mainly in loop regions or in β-sheets immediately before and after loop regions, suggesting that increasing loop rigidity may be a common underlying mechanism for increased thermal stability. Structural rigidity can confer thermal stability [1, 11], and amino acid B-values (organized by the XylBH43 crystal structure data from PDB 1YRZ for each amino acid using the program B-fitter [21]; see Supplementary Table 2) are a measure of electron density smearing that have been used successfully to identify regions for thermal stability improvement [12, 21, 22]. Thermally destabilizing mutations R45K (ΔKt0.5 = −1.8 ± 0.1 °C) and K48V (ΔKt0.5 = −0.8 ± 0.1 °C) are on a six-residue loop (R45-W50; Fig. 1), separating two β-strands. R45 (B = 33.67, 4th highest Arg B-value of 23 Arg residues having an average B-value = 24.52) and K48 (B = 32.74, 6th highest Lys B-value of 27 Lys residues having an average B-value = 27.52) have the largest B-factors of the shuffled mutations, ranking 41 and 37 out of 522 amino acids, respectively. M69 is located 15 Å from the active site carboxylates on the longest (24 residue) extended loop (S54-P77; Fig. 1), and XylBH43 is the only highly efficient GH43 xylosidase reported that contains Met at this position, otherwise occupied by Pro (Fig. 3). Stabilization has previously been shown by Pro substitution into loop regions, by making the loop more rigid and lowering the number of conformations and thus entropy of the unfolded state [1, 18, 29].
Fig. 3

Sequence alignment of XylBH43 and high-activity GH43 xylosidase homologs. Filled circle location of amino acid change in shuffled hits; asterisk active site catalytic Glu residue; XylBH43 PDB ID 1YRZ; SXA PDB ID 3C2U; LbLactobacillus brevis ATCC 367, GenBank ABJ65333.1; BsBacillus subtilis 168, GenBank CAB13642.2; BpBacillus pumilus, GenBank CAA29235.1; GsGeobacillus stearothermophilus, PDB ID 2EXH; AmAlkaliphilus metalliredigens, GenBank ABR49445.1; CaClostridium acetobutylicum, GenBank AAK81382

The thermal stability increases observed for hits 4B9, 2B10, and 5F1, containing the cluster of mutations C78D, H82A, T85K, and Y87W (Table 1), were not ascribable to single amino acid changes, as tested by variants in Table 2 where only T85K modestly improved Kt0.5. However, together these four mutations appear to increase Kt0.5 by ~3 °C. H82A and T85K comprise the last and first amino acids, respectively, of β-strands having an intervening two-residue (D83G84) hairpin loop, termed loop-D83 (Figs. 1, 3), and the increased thermal stability may have arisen from stabilizing this loop. Recently, the Kt0.5 of a polyol dehydrogenase was increased by 7 °C in a chimera created by exchanging a loop region from a less thermostable homologous enzyme [33]. Residues H82A and T85K have moderately-high B-factors (ranked 71/522 and 73/522, respectively), while loop residue D83 has a high B-factor, ranked 17 out of 522 amino acids. Hit 4B9 is of particular interest for further thermal improvement, since the kinetic parameters determined for X2 were little affected for this variant (Table 4). Variants 2B10 and 5F1 have in common the additional mutations M69P and N70D, as compared to 4B9, and the observed Kt0.5 increase of ~1.7 °C for these two hits compared to 4B9 is consistent with M69P simple additivity.

Mutation L186Y in the selected hits (Table 1) occurs on a 12-residue, active-site loop near catalytic general acid Glu188 (Fig. 1). Most 10C4 stabilization is likely due to the L186Y and M69P mutations, and assuming the stability conferred by all the mutations in 10C4 are simply additive, their sum of ΔKt0.5 = 4.5 ± 0.8 °C can account for the observed ΔKt0.5of 5.6 ± 0.4 °C. Most 6F11 stabilization appears due to the L186Y mutation, though the stability is somewhat more than expected based on L186Y alone, since 6F11 also contains the destabilizing mutations R45K and K48V. Saturation mutagenesis [31] of L186 resulted in all but four mutants with Kt0.5 values greater than that of XylBH43, with the largest effect seen in L186K, having an ΔKt0.5 = 8.1 ± 0.3 °C (Table 3). The effect of amino acid mutation is, in general, context-dependent [1], and when Lys was introduced into the corresponding position of the parental enzyme SXA, a decrease in Kt0.5 of 2.4 °C was observed. The breadth of the effect on the Kt0.5 of the L186 saturation mutagenesis, ranging from −1.9 to 8.1 °C, supports the saturation mutagenesis of other residues that appear to slightly perturb Kt0.5, such as V22A, K48V, H82A, T85K, and H419N (Table 2).

Combining several mutations, each of which confers an incremental improvement in thermal stability, can be a way to achieve significant gains in protein thermal stability [17, 18]. The L186Y mutation alone increased Kt0.5 by 3.2 ± 0.4 °C, and its additivity (within experimental error) to hits 4B9, 2B10, and 5F11 was demonstrated, with Kt0.5 increasing by 3.8 ± 0.3, 4.2 ± 0.8, and 3.0 ± 0.1 °C, respectively, thereby generating variants with combined Kt0.5 increases of 6.8 ± 0.1 to 8.8 ± 0.5 °C (Table 1). Simple additivity toward thermal stability can be anticipated for mutations not in direct contact [23, 30]; for example, an early directed-evolution thermal stability study showed eight of nine mutations individually contributing to the thermal stability of an esterase [10].

Kinetic parameter kcatX2 decreased in the mutant xylosidases, ranging from a minimal change for hit 4B9 (8 % decrease, ΔKt0.5 = 3.0 °C) to a significant decrease for L186Q (80 % decrease, ΔKt0.5 = 6.7 °C), and parameters KmX2, Km, Kixylose and KssX2 increased varyingly and in tandem for the mutants except L186Q, where KmX2 was decreased by ~1/2 (Tables 4, 5). While in general, a higher Kt0.5 would allow thermal compatibility with other thermally stable enzymes in a cocktail, the effects of concurrent changes in kcat, Km, Ki, and Kss during selection on the efficiency of industrial saccharification are process-specific. The effect of a lowered kcat would be mitigated under saccharification reactor conditions owing to the increased turnover of the thermally stabilized enzymes at higher temperatures. This was demonstrated here, where XylBH43 and L186Kwere tested at their respective Kt0.5 values for turnover efficiency of natural substrate X2, and the determined kcatX2 values were similarly ~260 s−1 (Table 4). GH43 xylosidases are considered attractive candidates for enzyme cocktails since catalysis occurs without an enzyme-bound intermediate that is susceptible at high substrate loadings to transferase-type activity (e.g., the transxylosylation of X2 yielding xylotriose [14]), conditions that may occur in commercial saccharification. However, GH43 xylosidases are prone to substrate and product inhibition. As a result, whereas an increased Km is deleterious for processes requiring high turnover at low substrate concentrations, the observed concurrent deterioration of Km can both be ameliorated at higher substrate concentrations and be a positive attribute owing to the in-tandem decrease in product (Ki) and substrate (Kss) inhibition. The effect of increased Ki on 4NPX hydrolysis was tested at 300 mM xylose and resulted in the convergence of turnovers for XylBH43 and L186Y (Table 5; Fig. 4a), while the effect of increased Kss was tested at up to 120 mM natural substrate X2, wherein the turnovers of XylBH43 and L186Y, K, and H converged (Table 4; Fig. 4b).
Fig. 4

Xylose and xylobiose inhibition of XylBH43 and select variants. Filled square wild-type; open square L186Y; filled triangle L186K; filled circle L186H. a 4NPX hydrolysis in the presence of 300 mM xylose; b X2 hydrolysis; reaction conditions were 100 mM phosphate, pH 6.5, 0.1 % BSA, 25 °C


This work was supported by funding from the United States Department of Agriculture, CRIS 5325-41000-049-00 (C.H., V.J.C., C.C.L., K.W.) and CRIS 3620-41000-118-00D (D.B.J. and J.D.B.). The mention of firm names or trade products does not imply that they are endorsed or recommended by the US Department of Agriculture over other firms or similar products not mentioned. The USDA is an equal opportunity provider and employer.

Supplementary material

10295_2013_1377_MOESM1_ESM.docx (11 kb)
Supplementary Table 1 (DOCX 10 kb)
10295_2013_1377_MOESM2_ESM.docx (25 kb)
Supplementary Table 2 (DOCX 24 kb)
10295_2013_1377_MOESM3_ESM.docx (11 kb)
Supplementary Fig. 1 (DOCX 10 kb)

Copyright information

© Springer-Verlag (outside the USA) 2013

Authors and Affiliations

  • Sanjay K. Singh
    • 2
  • Chamroeun Heng
    • 1
  • Jay D. Braker
    • 3
  • Victor J. Chan
    • 1
  • Charles C. Lee
    • 1
  • Douglas B. Jordan
    • 3
  • Ling Yuan
    • 2
  • Kurt Wagschal
    • 1
  1. 1.USDA Agricultural Research ServiceWestern Regional Research CenterAlbanyUSA
  2. 2.Department of Plant and Soil Sciences, Kentucky Tobacco Research and Development CenterUniversity of KentuckyLexingtonUSA
  3. 3.USDA Agricultural Research ServiceNational Center for Agricultural Utilization ResearchPeoriaUSA

Personalised recommendations