Direct comparison of gluco-oligosaccharide oxidase variants and glucose oxidase: substrate range and H2O2 stability

Glucose oxidase (GO) activity is generally restricted to glucose and is susceptible to inactivation by H2O2. By comparison, the Y300A variant of gluco-oligosaccharide oxidase (GOOX) from Sarocladium strictum showed broader substrate range and higher H2O2 stability. Specifically, Y300A exhibited up to 40 times higher activity on all tested sugars except glucose, compared to GO. Moreover, fusion of the Y300A variant to a family 22 carbohydrate binding module from Clostridium thermocellum (CtCBM22A) nearly doubled its catalytic efficiency on glucose, while retaining significant activity on oligosaccharides. In the presence of 200 mM of H2O2, the recombinant CtCBM22A_Y300A retained 80% of activity on glucose and 100% of activity on cellobiose, the preferred substrate for this enzyme. By contrast, a commercial glucose oxidase reported to contain ≤0.1 units catalase/ mg protein, retained 60% activity on glucose under the same conditions. GOOX variants appear to undergo a different mechanism of inactivation, as a loss of histidine instead of methionine was observed after H2O2 incubation. The addition of CtCBM22A also promoted functional binding of the fusion enzyme to xylan, facilitating its simultaneous purification and immobilization using edible oat spelt xylan, which might benefit the usage of this enzyme preparation in food and baking applications.

Scientific RepoRts | 6:37356 | DOI: 10.1038/srep37356 oxidize the anomeric carbon of carbohydrate substrates through an FAD reduction mechanism; however, in these cases, the FAD cofactor is bi-covalently linked through cysteine and histidine residues. Furthermore, these enzymes display a comparatively broad substrate profile, ranging from xylose, and galactose, to malto-, cello-and xylo-oligosaccharides 14,16,18 . Similar to MnCO, GOOX oxidizes oligosaccharides more effectively than monosaccharides 16 . To date, thirteen GOOX variants have been created through site-directed mutagenesis 16,19 , where one variant (Y300A) shows higher activity on monosaccharides (including glucose and xylose) compared to the wild-type GOOX. The Y300A variant also retains activity on oligosaccharides and displays reduced substrate inhibition 19 .
MnCO activity has already been reported to increase dough elasticity and consistency 20 and was tested as a substitute for GO in baking applications 21 . Nevertheless, MnCO was inactivated by H 2 O 2 during substrate turnover 22 , although the extent of H 2 O 2 inactivation was not reported.
The comparatively broad substrate range of Y300A along with reduced substrate inhibition make it a suitable alternative to GO particularly in applications that comprise mixed carbohydrates. Accordingly, this study directly compared the substrate profile and H 2 O 2 inactivation of the A. niger GO and the Y300A variant of GOOX. The Y300A variant was also fused to CtCBM22A, a carbohydrate-binding module from Clostridium thermocellum, generating CtCBM22A_Y300A to facilitate enzyme purification through binding to edible oat fibre.
To prepare insoluble OSX, 2 g OSX was suspended in 200 mL of 50 mM Tris-HCl pH 8.0 for 48 h at room temperature and then washed three times. The washed OSX was then filtered through a 0.45-μ m nylon membrane, and the dry weight of the retentate was measured.
CBM fusion synthesis. The CtCBM22A sequence with its native connecting loops (10 amino acids at the N-terminus and 7 amino acids at the C-terminus), together with a TP-rich linker (SRGGGTATPTPTPTPTP) at the C-terminus, was synthesized by NZYTech (Lisbon, Portugal) after codon optimization for Pichia pastoris expression. The synthesized CBM sequence was fused with that of Y300A in pPICZα A. The fusion also contains a His 6 -tag at the C-terminus to facilitate purification. DNA sequences were confirmed at the Center for Applied Genomics (the Hospital for Sick Children, Toronto). The synthetic gene was deposited in the GenBank database under the accession number of KC707634.
His-tag purification of CtCBM22A_Y300A. The confirmed plasmid was transformed into P. pastoris KM71H following the manufacturer's instructions (Life Technologies, USA). Expression screening by immuno-colony blot, cell growth, protein expression and purification were conducted following previously reported methods 23 . Briefly, cells were grown at 15 °C and 250 rpm for 5 days, and 0.5% methanol was added every 24 h to induce recombinant protein expression. Culture supernatant was desalted and mixed with Ni-NTA resin (Qiagen, USA), bound proteins were then eluted with 250 mM imidazole, followed by centrifugal buffer exchange with 50 mM Tris-HCl pH 8.0.
Protein concentration was measured by the Bradford method (Bio-Rad Laboratories, USA), and verified by band densitometry on SDS-PAGE gels using known quantities of bovine serum albumin as a reference and ImageJ for image analysis (http://rsbweb.nih.gov/ij/). OSX purification of CtCBM22A_Y300A. After five days of induction with 0.5% methanol, 250 mL P. pastoris culture supernatants comprising CtCBM22A_Y300A were harvested by centrifugation for 10 min at 2,000 × g (repeated three times), then filtered through a 0.45 μ m membrane. Half of the supernatant was then exchanged to 50 mM Tris-HCl pH 8.0 using 30 kDa MWCO Vivaspin 20 centrifugal concentrators (GE healthcare, UK). Insoluble OSX was then added at 1.25% (w/v) to 40 mL filtered supernatant and buffer exchanged supernatant. Resulting solutions were mixed for 24 h at 4 °C using a 360-degree rotator set at 8 rpm. Protein-bound OSX and unbound (flow-through) fractions were then collected by vacuum-filtration using a 0.45 μ m membrane. Protein-bound OSX was washed 3 times with 50 mM Tris-HCl pH 8.0 and then stored in this buffer. All fractions were kept at 4 o C.
OSX binding efficiency. The protein in 700 μ L of filtered supernatant, buffer-exchanged supernatant or flow-through fractions after OSX addition was separately precipitated with 10% trichloroacetic acid. The precipitate was washed thoroughly with ice-cold acetone, and the pellet was suspended in 40 μ L of SDS loading dye. The proteins in 35-40 mg wet weight of bound OSX were extracted with 40 μ L of the denaturing solution (10% SDS and 10% beta-mercaptoethanol) for 10 min at 100 o C. Precipitated as well as extracted proteins were then analyzed using 10% SDS-PAGE gels.
The effect of Ca 2+ on CtCBM22A binding to xylan was investigated by incubating the fusion enzyme with OSX overnight at 4 o C in the presence of 5 mM CaCl 2 ; the amount of unbound and bound protein was then evaluated weekly by SDS-PAGE. Binding affinity on soluble polymeric substrates. Binding of CtCBM22A_Y300A on soluble OSX, arabinoxylan, beech wood xylan, wheat bran hemicellulose and propoxylated wheat bran hemicellulose was evaluated by affinity gel electrophoresis 24 . Briefly, 0.01% of each polysaccharide was added to 7.5% (w/v) acrylamide in a Tris-glycine buffer (25 mM Tris, 250 mM glycine pH 8.3) prior to polymerization to produce native polyacrylamide gels. Protein samples (5 μ g) were loaded into each gel followed by electrophoresis using ice-cold Tris-glycine buffer.
Specific activity, enzyme kinetics and thermostability measurements. The production of oxidized sugars was determined by continuously measuring the co-production of H 2 O 2 at 37 °C for 15 min using a chromogenic horseradish peroxidase assay 23 . Activity measurements were performed using 16 nM of each GOOX variant and 0.5 mM substrate; kinetic measurements were performed using the following substrate concentrations: 0.05-300 mM glucose; 0.05-1,200 mM xylose; 0.05-20 mM cellobiose; 0.01-10 mM cellotriose, xylobiose and xylotriose; 0.01-4 mM of longer cello-and xylo-oligosaccharides. GraphPad Prism5 software (GraphPad Software, USA) was used for curve fitting and deriving kinetic parameters.
To compare activities of GO and GOOX variants, 24 nM of each enzyme were separately assayed in triplicate with 10 mM of xylose, maltose and cellobiose; 1 mM of glucose and cellotetraose; and 0.6 g/L of mixed XOS. GOOX enzymes were assayed in 50 mM Tris-HCl pH 8.0 while GO was assayed in 50 mM sodium acetate pH 5.0. Thermostability tests were performed by incubating GOOX enzymes (16 nM) at temperatures between 30 and 54 °C for 1 h, and then measuring residual activity (initial rate) on 0.5 mM cellobiose. The effect of H 2 O 2 on GO activity was similarly determined, except that the buffer was replaced by 50 mM sodium acetate (pH 5.0) and 1 mM glucose was used. Substrate concentration was kept low to avoid any contribution from the de novo release of H 2 O 2 . Following 5-h incubation with glucose, or 2-h incubation with cellobiose, reactions were filtered through 10 K Omega TM Nanosep centrifugal devices (Pall Corp., USA) to separate reaction products from each enzyme. To facilitate the high-performance anion-exchange chromatography (HPAEC) analyses, 100 U catalase was added to reactions containing glucose to prevent H 2 O 2 interference, and then incubated for 1 h at 25 °C before filtration through 10 K Omega TM Nanosep centrifugal devices.
To test for stability, 16 nM GOOX enzymes and GO were also pre-incubated with 200 mM H 2 O 2 for 30 min at 25 °C in the dark in an Eppendorf ThermoMixer with shaking (300 rpm). The H 2 O 2 was then degraded using catalase (2 × 100 U) prior to addition of 1 mM glucose to measure residual activity. Following a 2-h incubation with glucose at 37 °C, each sample was filtered using a 10 K Omega TM Nanosep centrifugal device prior to analysis by HPAEC-PAD.
Potential effects of residual cultivation media on H 2 O 2 decomposition was tested in triplicate by incubating 1 μ L of culture supernatant or medium with 74 μ L of different H 2 O 2 concentrations (up to 150 μ M) in 96-well plates at 25 °C for 5 h in the dark. The remaining amount of H 2 O 2 was measured using the chromogenic horseradish peroxidase assay.
Total amino acid analysis. Both Y300A and CtCBM22A_Y300A (1,000 pmoles) were incubated with 200 mM H 2 O 2 in the presence of 1 mM glucose for 5 h in the dark. The samples were then washed intensively with water using a 10 K Omega TM Nanosep centrifugal device, and treated with 6 N HCl containing 1% phenol for 24 h at 110 °C. The amino acid composition was analyzed at the SPARC BioCentre (the Hospital for Sick Children, Toronto).

Results and Discussion
Broader substrate specificity of Y300A. The GOOX variant Y300A displays a similar substrate profile to the wild-type enzyme; however, it demonstrates higher activity on glucose 16 and lower substrate inhibition 19 . Therefore, Y300A was chosen for the current comparative analysis with GO. The substrate preference of GO and Y300A was assayed using glucose, xylose, maltose, cellobiose, cellotetraose, and XOS. On tested mono-and oligosaccharides, GO exhibited only 2.5% activity relative to Y300A, with the exception of glucose, for which GO activity was 100% higher than Y300A (Fig. 1). The different substrate preference of Y300A and GO is likely attributed to the open active site of GOOX 25 in comparison with the narrow binding pocket of GO, which is specifically suited for glucose 26 . Improving glucose oxidation by Y300A. The yield of CtCBM22A_Y300A recombinantly expressed in P. pastoris KM71H was approximately 200 mg/L. Fusing CtCBM22A to Y300A also increased k cat values of the enzyme on all sugars, up to 67% in the case of cellotriose, from 670 min −1 to 1,120 min −1 (Table 1). A similar effect was previously seen when CtCBM22A was fused to the N-terminus of the wild-type GOOX 23 . The K m values of CtCBM22A_Y300A on monosaccharides including glucose and xylose were also decreased by around 45%, resulting in higher catalytic efficiency (Table 1). For instance, k cat /K m values of CtCBM22A_Y300A on glucose and xylose (173 and 27 mM −1 min −1 , respectively) are about 100% higher than those of Y300A (Table 1), and more than 600% higher than those reported for wild-type GOOX 19 . As CtCBM22A does not bind xylose and glucose 27 , the observed improvement in K m and catalysis on monosaccharides, as well as slight thermal activation at 40 °C Scientific RepoRts | 6:37356 | DOI: 10.1038/srep37356 (Fig. 2) by CtCBM22A_Y300A, is likely due to a subtle structural change induced in the Y300A variant upon fusion to CtCBM22A, as previously proposed for GOOX 23 .
As expected, the catalytic efficiency of the fusion enzyme on glucose was lower than those reported for A. niger GO (i.e., 1,800 28 and 2,280 mM −1 min −1 29 ). However, the significantly lower K m values of CtCBM22A_Y300A on glucose (5.4 mM) compared to those reported for GO (~ 30 28 and ~50 mM 29 ) may offer advantages to applications containing low glucose concentrations.
Enhanced xylan binding through CtCBM22A fusion. The presence of different xylans including OSX, arabinoxylan, beech wood xylan and wheat bran xylan significantly reduced the migration of CtCBM22A_Y300A in native polyacrylamide gels, indicating strong xylan binding by the fusion enzyme (Fig. 3). Reduced migration was more obvious on buffer-soluble oat spelt xylan and propoxylated wheat bran xylan, as shown by sharp bands at or near the loading wells (Fig. 3). There was no difference in migration between the CtCBM22A fusion of Y300A (Fig. 3) and the previously reported wild-type fusion 23 , suggesting that the tyrosine substitution did not impact hemicellulose binding.     wild-type GOOX on xylan 23 , insoluble OSX was chosen for ability to immobilize and recover CtCBM22A_Y300A directly from culture supernatant while providing additional dietary fibre in food applications 30 .
Insoluble OSX was directly added to filtered culture supernatant containing CtCBM22A_Y300A or the supernatant that had been exchanged to 50 mM Tris buffer pH 8.0. The amount of protein that remained bound to insoluble OSX (B fraction) was then evaluated by SDS-PAGE (Fig. 4). The original amount of protein in supernatant solutions before loading (L fraction), those in the first flow-through (F fraction) and flow-through during washing (W fraction) were also analyzed. Low amounts of protein in flow-through solutions were detected, indicating that most of the protein bound to OSX (Fig. 4). Furthermore, there was a similar amount of bound protein from filtered culture supernatant and buffer-exchanged supernatant samples, suggesting that the culture supernatant did not interfere with OSX binding of CtCBM22A_Y300A. Therefore, OSX can be used to recover CtCBM22A_Y300A directly from culture supernatant without buffer-exchange, facilitating the simultaneous purification and immobilization of the enzyme.
Although the filtered, spent culture did not impact CtCBM22A_Y300A binding to OSX, enzyme activity towards cellobiose was reduced by approximately 50% (Fig. 4). Fresh induction media did not interfere with the assay (Fig. S1), which suggests that the reduced activity resulted in part by components released by the Pichia transformant.
Further analysis of CtCBM22A_Y300A from buffer exchanged culture medium and immobilized to OSX confirmed that immobilization did not reduce enzyme activity on cellobiose (Fig. 4). This suggests that cellobiose did not compete with OSX for CtCBM22A binding, which is consistent with the high K a of CtCBM22A towards xylo-oligosaccharides and particularly OSX 27 , and with our earlier characterization of CtCBM22A_GOOX 23 .
Binding between CtCBM22A_Y300A and OSX remained stable in 2 M of NaCl, even after 12 weeks at 4 °C (data not shown). However, up to 10% of the protein desorbed from OSX during storage at room temperature, of which about 50% bound back to the OSX upon the addition of 5 mM CaCl 2 (Fig. S2) (Fig. 5). The higher turnover number and better glucose binding by the fusion enzyme (Table 1) might have outweighed activity losses resulting from H 2 O 2 inactivation. No significant difference in H 2 O 2 tolerance was observed between Y300A and GO (Fig. 5), which displayed decreasing glucose oxidation with increasing H 2 O 2 concentration and retained approximately 60% activity at 200 mM H 2 O 2 . It should be noted, however, that the commercial GO preparation used herein contains a very low but detectable amount of catalase (≤ 0.1 U/mg of GO, as reported by Sigma). An earlier report shows that GO retains approximately 20% activity after 20-min incubation with 100 mM H 2 O 2 12 . This difference in H 2 O 2 tolerance of GO samples may reflect differences in activity measurement (i.e., oxygen consumption versus gluconic acid production herein) or improvement in commercial preparations of GO.
When retaining the same reaction conditions but substituting glucose for cellobiose, H 2 O 2 inactivation was not detected for any GOOX variant (Fig. 6). The catalytic efficiency of Y300A and CtCBM22A_Y300A on cellobiose (3,400 and 4,000 min −1 mM −1 , respectively) is an order of magnitude higher than that on glucose (98 and 173 min −1 mM −1 , respectively) ( Table 1). Presumably then, GOOX variants were able to oxidize its preferable substrate, cellobiose, before H 2 O 2 inactivation took effect.
Pre-incubation of GO and GOOX variants with 200 mM H 2 O 2 for 30 min prior to addition of glucose did not impact enzyme activity (Fig. 7). This suggests that reduced forms of GO and GOOX are more susceptible to inactivation by H 2 O 2 , as previously reported for GO 12 .
Apparent mode of H 2 O 2 inactivation. The oxidation of key methionine residues in the presence of H 2 O 2 has been proposed to cause GO inactivation 12 . The total amino acid analysis of GOOX variants, however, revealed significant loss of histidine in Y300A and its CtCBM22A fusion following incubation with H 2 O 2 (Table S1). The broader active site as well as the double-covalent flavinylation in GOOX might contribute to this potential difference in H 2 O 2 inactivation. Notably, histidine residues H70 and H138 are positioned in the GOOX active site, of which the N δ1 atom of H70 binds to the 8α -methyl group of the isoalloxazine ring of FAD while H138 interacts with FAD via a hydrogen bond (Fig. S3). The oxidation of either these histidine residues would affect FAD binding  and electrochemical properties. For instance, an H70A substitution, which led to removal of the 8α -N1-histidyl bond, retained only 0.5% catalytic efficiency and about 55% redox potential, compared with the wild-type 31 . Histidine residues can be oxidized by H 2 O 2 to form a mixture of products including Asp and 2-oxo-histidine 32 . Since H 2 O 2 treatment only slightly increased the (Asp + Asn) content in Y300A and CtCBM22A_Y300A samples (Table S1), 2-oxo-histidine is probably the major product of oxidation. As histidine loss was 6.7 and 9.3 mol/mol of protein in Y300A and its CtCBM22A fusion, respectively (Table S1), other histidine residues were also modified. Besides the two exposed histidine residues in the active site and six in the 6x-his tag, Y300A also has four histidine residues on the surface, and the fusion provides two additional exposed histidines. Loss of tyrosine in the H 2 O 2 -treated samples was also detected, particularly in CtCBM22A_Y300A, which has three tyrosine residues in the binding cleft out of its eleven surface-exposed tyrosine residues.

Conclusion
GOOX variants Y300A and CtCBM22A_Y300A demonstrate potential benefit in applications characterized by relatively high H 2 O 2 concentrations and mixed sugars. Both Y300A and CtCBM22A_Y300A have broader substrate profiles than the commonly used GO, and CtCBM22A_Y300A displayed less H 2 O 2 inactivation than GO when assayed with glucose. Furthermore, both GOOX variants showed no H 2 O 2 inactivation on cellobiose, their favorable substrate. CtCBM22A_Y300A is the best variant of GOOX to date for glucose oxidation (as well as xylose oxidation) while retaining considerable activity on oligosaccharides. The presence of CtCBM22A also permits enzyme binding to OSX, facilitating simultaneous purification and immobilization of the enzyme, and ancillary fibre enrichment in potential food applications.