Enzyme expression and purification
The resulting recombinant enzyme was expressed and described in the 2-3 results and discussion. The purified enzyme (> 90% homogeneity) was analyzed by SDS-PAGE (Figure 3-2a) and mass spectrometry (Figure 3-2b), and gave mol wt values of ca. 50 kDa and 50045 (major peak), respectively.
The theoretical mol wt is 49445 Da. The discrepancy between the measured and calculated mol wt values is likely because of protein glycosylation in Pichia. This argument is somewhat supported by the presence of a series of mass peaks (50045, 50208, 50369; Figure 3-2b) with differences of 161-163 Da, which may indicate the glycosylation of protein.
pH profiles of the wild-type and the E223G mutant
It is common that the catalytic reactivity of many retaining glycohydrolases is mainly mediated by two active site carboxylates (aspartate and/or glutamate). A bell-shaped activity profile reflected two apparent pKa is often observed. The pH-dependent activity assay showed that the purified enzyme had optimum Abf activity at pH 2.8–3.2 (Figure 3-3). The pH-dependent activity curve of recombinant Abf exhibited a bell shape, with two apparent pKa values of 1.8 ± 0.1 and 4.2 ± 0.1. It is of interest that both pKa values were about 2 pKa units lower than those found for other glycohydrolases.
Mutagenic study
The first protein structure of a family 54 α-L-arabinofuranosidase, from A.
kawachii, was recently resolved [70]. This structure is valuable for structural simulation of enzymes in the same family. Based on sequence comparisons, our enzyme has 73% identity and > 85% homology to the A. kawachii
α-L-arabinofuranosidase. By inspection of the protein structure (Figure 3-4), residues appearing in the active site of A. kawachii α-L-arabinofuranosidase and T. koningii (shown in parentheses) may be listed: C176 (C178), D219 (D221), E221 (E223), D297 (D299), and S299 (S301). All these residues are highly conserved in 12 enzymes of the GH54 family (Figure 3-5). The carboxyl groups of E221 (E223 in T. koningii Abf) and D297 (D299 in Abf) are located on either side of the anomeric C-1 carbon of enzyme-bound arabinofuranose, strongly suggesting that these residues serve as nucleophile and general acid/base residue, respectively. To confirm the catalytic roles of these residues, an extensive mutagenesis study was performed on E223, D299, and 22 other Asp or Glu residues (D95, D120, D160, E163, D170, D181, E186, D191, E198, D221, D238, E289, E310, D320, E323, E407, D410, D429, E436, D437, D482, and D490), which are highly conserved in GH54 family members. We changed the putative catalytic residues, aspartate (D) and glutamate (E), to asparagine (N), glutamine (Q) or glycine (G) by site-directed mutagenesis. All crude mutant enzymes with mutations in D170, D221, E223, D299 or E310 largely lost enzyme activity with pNPAF as substrates (Appendix 7). Some of these mutant enzymes were further purified and studied. Kinetic parameters are summarized (Table 3-1). The Km values of mutant enzymes, ranging from 0.22–0.32 mM, are quite similar to that of wild-type Abf, except for the mutants D299N, and E310G, where the Km
values were 0.05 mM, and 3.0 mM, respectively. The kcat values of the mutant enzymes D170N, D221N, D299N, D299G, and E310G decreased (with respect to the wild-type Abf) by factors of 31, 7000, 1300, 262, and 70,
hydrolase family 54 members, has been postulated to act as a nucleophile, this study uses site-directed mutagenesis for direct confirmation of this function.
Residue D221 (D219 in the A. kawachii enzyme) engages in strong hydrogen bonding with the C3 and C5 sugar hydroxyl groups [70]. The large decreases in activity observed when these residues are mutated presumably occurs because the substrate is now incorrectly oriented in the active site.
The residues D170 and E310 are more than 20 Ǻ distant from the active site.
The activity loss of the E310G mutant results in part from a 10-fold increase in Km value, compared to the unmutated enzyme. A perturbation of general protein structure by this mutation should perhaps be considered.
One unexpected finding is that the E223G mutant still possesses extremely high activity. Its kcat (22 s–1) and kcat/Km (59460 M–1s–1) values are nearly identical to those of wild type Abf. After careful inspection of the active-site structure of A. kawachii enzyme, the Asp-189 (Asp-191 in our Abf) was found to locate closely to the nucleophile, E221 (E223 in our Abf). The orientation and distance (4–5 Ǻ) between Asp-189 and the C-1 position of the substrate is perfect for an inverting-type of catalysis to take place. It is very likely that when the site-chain of the nucleophile is replaced by proton (ie, E223G), more space will be gained and, therefore, allow water molecule to diffuse deeply into the active site, while it is no room around the nucleophile if the Glu is mutated into Gln. This hypothesis has been tested by analyzing the activity of double mutant (D191N/E223G) and the stereochemistry of E223G catalysis. The results from this study will be published elsewhere.
In many glycohydrolases, exogenous nucleophiles such as azide, formate, and other anions have been shown to enhance the catalytic activity of enzymes mutant in residues that serve as nucleophiles or provide general acid/base functions [88-91]. Activity enhancement of a mutant by addition of a nucleophile (for example, azide) and formation of a stereospecific product (α- or β-glycosyl azide) offers a useful technique for identifying essential residues
of glycosidases. In this study, however, addition of high concentrations of azide (up to 2 M) did not rescue the activities of the D299G, D299N, E223Q or E223G mutants, presumably because the active site was located relatively deep in Abf that the azide ion could not diffuse to the site. Similar observations were reported previously in a study on the A. kawachii IFO4308 enzyme [70]. Nevertheless, comparing the pH activity profiles (kcat vs pH) of the wild-type Abf and the D299N mutant (Figure 3-3) may provide an insight into the essential function of D299. The activity of the D299N mutant was almost constant in the range of pH 3–6.5 and with a trend of decrease at pH
<2.4, indicating the absence of the second pKa point. Also, the low Km value (0.05 mM) of the D299N mutant suggests accumulation of the glycosyl-enzyme intermediate, whose hydrolysis is then accelerated by the general acid/base catalytic action of the enzyme. Unfortunately, owing to the various degree of post-glycosylation on the recombinant Abf, an unequivocal result of mass spectrometric analysis in attempt to show the presence of the glycosyl-enzyme intermediate was unsuccessful. In sum, the results of the site-directed mutagenesis and kinetic studies, in combination with structure analysis, point to E223 and D299 as the essential nucleophile and general acid/base residues of Abf, respectively. The function of D299 will be further discussed below.
Transglycosylation and product partition
A time-course 1H NMR study is commonly employed to examine
arabinofuranosidases, an arabinofuranoside product may be expected to undergo fast mutarotation to form four arabinose tautomers. These are α- and β-L-arabinofuranosides and α- and β-L-arabinopyranosides [57]. To overcome this limitation, a method involving transglycosylation of the Abf using methanol as glycosyl acceptor was employed. The advantage of this strategy is the formation of a methyl glycoside that cannot mutarotate. For most retaining enzymes, although the formation of a covalent enzyme intermediate is expected, it is difficult to detect as the lifetime is short. In the past, useful 2-fluoroglycoside products have been obtained using a specific glycosyl-enzyme trapping technique [62, 92-94]. However, perhaps because 2-fluoroarabinofuranoside is difficult to synthesize, this strategy has not yet been used in any study of α-L-arabinofuranosidase. We used an alternative, indirect, method. If a constant chemical bias may be noted in different enzyme products, the formation of a common intermediate in an enzymatic reaction may be inferred. Our study showed that the Abf exhibited strong transglycosylation activity when methanol was used as an arabinofuranosyl acceptor. Here, a suitable amount of pNPAF, pCPAF and CNPAF were enzymatically hydrolyzed in acetate buffer (pH 4.1) containing 12% (v/v) methanol. The solutions were then dried and exchanged with D2O several times before 1H-NMR analysis. For all three reactions, the 1H-NMR spectra (measured at 25 ºC) of the sugar moieties (in the range of 3-6 ppm) were nearly identical. In principle, five different end products with arabinosyl ring structures should be observed (Figure 3-6a). According to the literature [80] and our study, the C1 protons of each sugar ring were assigned as follows:
methyl-α-arabinofuranoside (4.86 ppm, J = 1.03 Hz), α-arabinofuranose (5.17 ppm), β-arabinofuranose (5.22 ppm, J = 3.8 Hz), α-arabinopyranose (4.43 ppm, J = 5.9 Hz), and β-arabinopyranose (5.16 ppm, J = 3.24 Hz) (Figure 3-6b). Based on peak assignment and the integration of the C1 proton on the sugar ring, the ratio of methyl-α-arabinofuranoside/arabinose was calculated
from each spectrum. Regardless of the substrates, these ratios were nearly constant, and averaged 1.04 ± 0.02 (1.04 for pCPAF, 1.06 for pNPAF, and 1.03 for CNPAF). This suggests that a common intermediate, most likely an arabinosyl-enzyme structure, occurs in the reaction pathways. As the product of transglycosylation in this experiment is methyl-α-arabinofuranoside, we can confirm that the Abf is indeed a retaining enzyme.
Substrate reactivity and Brønsted plot
Kinetic assessment of substrates bearing different leaving phenols is a common strategy in studying mechanistic actions of glycohydrolases [95].
For glycohydrolases with two-step mechanisms (formation of a glycosyl-enzyme intermediate in a glycosylation step followed by breakdown of the intermediate in a deglycosylation step), the aglycon moiety is cleaved in the glycosylation step of the reaction (Figure 3-1). Therefore, the reaction rate of the first step dictates the ease by which the leaving group may be released from the substrate. A strong correlation between the pKa of the phenol leaving group and the activity of the enzyme should be observed if the first step is rate-limiting. To determine the rate-limiting step of hydrolysis catalyzed by the recombinant Abf, we worked to prepare
aryl-α-L-arabinofuranosides bearing different leaving phenols (pKa values 5.15 to 9.99) and to perform steady-state kinetic analysis using these substrates. The reaction temperature was held at 25 ºC to reduce spontaneous substrate hydrolysis. Although Abf is specific with regard to the glycon moiety of the
relationship, which has been shown to be useful in understanding the mechanism of enzyme action [51, 96-97]. Based on the kcat values, an extended Brønsted plot was constructed by plotting the logarithm of kcat against the pKa of the leaving phenol (Figure 3-7a). A plot with a slightly downward trend was obtained, with a slope (βlg value) of –0.18. That the detailed structures of the leaving phenols do not affect the kinetic parameters of the reaction may indicate that the dearabinosylation step (breakdown of the arabinosyl-enzyme intermediate) is the rate-limiting step. Alternatively, although it may be considered unlikely, the low Brønsted constant could indicate that the arabinosylation step is the slow step and that an early transition state is attained. However, the very weak reaction inhibition shown by MAF (Ki > 16 mM) minimized this possibility. If the transition state were substrate-like, strong inhibition would be expected. We further studied substrate reactivity using the D299G mutant enzyme. A new plot with βlg = –1.3 was obtained (Figure 3-7a). For substrates with pKa > 6, the absence of general acid/base catalysis resulted in a strong correlation between the pKa of the phenol leaving group and enzyme activity. Clearly, formation of the arabinosyl-enzyme intermediate is now the rate-limiting step.
However, with 2,5-DNPAF (a good substrate, with pKa = 5.15), the log kcat
value is clearly lower than expected if the enzyme reaction were to proceed as outlined above. This indicates that the dearabinosylation step becomes at least partially rate-limiting when 2,5-DNPAF breakdown is catalyzed by the D299G enzyme. In addition to alteration of catalytic properties, the D299G mutation also affects enzyme Km values with different substrates (Table 2).
The Km values decrease as the pKa values of the leaving phenols change in the substrates. For the wild-type enzyme, such changes in Km values are not obvious. These data showed that as the ability of the aglycon moiety to depart the enzyme increased, the more glycosyl-enzyme intermediate accumulated. This may be appreciated if it is noted that Km can be
represented as [E][S]/Σ[ES]. Thus, the more the glycosyl-enzyme intermediate accumulates, the lower the Km value. Such behavior may be expected in mutants affected in the general acid/base residue, as a good leaving group elevates the rate of the first catalytic step. The rate of the second step remains slow as the basic residue, which (in unmutated enzymes) activates the water molecule, is missing.
The kcat/Km values are informative with respect to the first irreversible step. For retaining glycoside hydrolases, the Brønsted relationship obtained by plotting kcat/Km values of substrates against the pKa values of the leaving phenols provides information about the glycosylation step [54, 61]. As may be seen (Figure 3-7b), the Brønsted constants (βlg) are –0.19 and –1.2 for wild-type Abf and the D299G mutant, respectively. This suggests that the arabinosylation step catalyzed by the wild-type Abf enzyme is not sensitive to the leaving abilities of different phenols, as the general acid/base residue protonates the oxygen of the glycosidic bond, thus imparting constant leaving abilities to different phenols. When the general acid/base residue is absent, however, the catalytic efficiency (kcat/Km) for the tested substrate became highly sensitive to the leaving ability of phenol groups (βlg = –1.2). A typical kinetic consequence of the general acid/base mutation is that for substrates requiring strong acid assistance (such as pCPAF and mNPAF), the first reaction step is much slower (12000–14000-fold decreases were noted in this study) than is the case with substrates that need less acid assistance (such as 2,5-DNPAF). The kinetic behavior displayed by the D299G mutant and the pH activity profile of D299N, when compared with data from the
Table 3-1. Michaelis–Menten parameters for the hydrolysis of pNPAF by wild-type and mutant enzymes at pH 4.1 and 25 °C.
Enzyme kcat (s–1) Km (mM) kcat/Km (M–1s–1)
wild-type 21 0.32 65625
D170N 0.68 0.30 2250
D221N 0.003 0.22 14
E223Q 0.0005 0.88 0.57
E223G 22 0.37 59460
D299N 0.016 0.05 320
D299G 0.08 0.21 400
E310G 0.3 3.0 100
The kinetic measurements for wild-type and E223G were within 10%
standard error, whereas for other mutants with much weaker activity, the errors were within 20%.
Table 3-2. Km and kcat values of wild-type Abf using various aryl-α-L-arabinofuranosides at pH 6.5
Unit: kcat / Km (M–1s–1), kcat (s–1)
Km, mM kcat kcat / Km log kcat log (kcat / Km) Substrate pKa
WT D299G WT D299G WT D299G WT D299G WT D299G
2,5-DNPAF 5.15 0.28 0.10 8.25 0.69 29542 6761 0.92 –0.16 4.47 3.83 CNPAF 6.45 0.42 0.20 6.52 0.12 15633 572 0.81 –0.94 4.19 2.76 pNPAF 7.18 0.31 0.21 2.84 0.014 9215 64 0.45 –1.86 3.96 1.81 mNPAF 8.39 0.54 0.65 2.02 0.0003 7369 0.46 0.31 –3.52 3.87 –0.33 pCPAF 8.49 0.30 0.69 1.70 0.0004 5592 0.58 0.23 –3.41 3.75 –0.25 PAF 9.99 0.28 – 1.23 – 4334 – 0.09 – 3.64 –
*: The D299G mutant enzyme is almost inactive towards PAF.
HO O
Figure 3-1. Proposed reaction mechanism of a retaining α-L-arabinofuranosidase.
kDa M 1
97.0
66.0
45.0
(a) (b) 30.0
20
Figure 3-2. SDS-PAGE (a) and mass spectrometry (b) analysis of the recombinant α-L-arabinofuranosidase. Lanes: M, markers; 1, recombinant wild-type Abf.
8 6
4 2
0
0.030
0.025
0.020
0.015
0.010 k of D299N, s cat-1
8 6
4 2
0 40
30
20
10
0
pH k of WT Abf, s cat-1
Figure 3-3. pH activity profiles of wild-type Abf (○) and the D299N mutant enzyme (●). The kcat values of wild-type and D299N mutant were measured at the final pH values: 1.9, 2.0, 2.4, 3.3, 3.9, 4.2, 5.5, 6.5.
Figure 3-4. The active site of the GH54-family enzyme from Aspergillus kawachii IFO4308 (1WD4) with arabinofuranose in place. The corresponding amino acids in the Trichoderma koningii Abf are labeled in parentheses.
170 181 186 191 221 223 289 299 310
U38661 166: YAVLDGTHYNGACCFDYGNAETNSRDTGN 194 215:GPWIMADLENGL 226 286:MSKEGAIILGIGGDNSNGGQGTFYEGV AB085904 164: YAVLDGTHYNDACCFDYGNAETSSTDTGA 192 213:GPWIMVDMENNL 224 284:MSKEGAIILGIGGDNSNGAQGTFYEGV Z69252 166: YAVLDGTHYNGACCFDYGNAETNSRDTGN 194 215:GPWIMADLENGL 226 286:MSKEGAIILGIGGDNSNGAQGTFYEGV AF367026 172: YAVLDGTHYNDACCFDYGNAETSSTDTGN 200 221:GPWVMADLENGL 232 292:MSKEGAIILGIGGDNSNGAQGTFYEGV AB073861 172: YAVLDGTHYNSACCFDYGNAEVSNTDTGN 200 221:GPWIMADLENGL 232 292:MSKEGAIILGIGGDNSNGAQGTFYEGV AB073860 172: YAVLDGTHYNSACCFDYGNAEVSNTDTGN 200 221:GPWIMADLENGL 232 292:MSKEGAIILGIGGDNSNGAQGTFYEGV L23502 164: YAVLDGTHYNDACCFDYGNAETSSTDTGA 192 213:GPWIMVDMENNL 224 284:MSKEGAIILGIGGDNSNGAQGTFYEGV U39942 164: YAVLDGTHYNDACCFDYGNWQTSSTDTGA 192 213:GPWLMVDMENNL 224 284:MSKEGAIILGIGGDNSNGAQGTFYEGV Y13759 170: YAVLDGTHYNDGCCFDYGNAETSSLDTGN 198 219:GPWIMADLENGL 230 291:MSKEGAIILGIGGDNSNGAQGTFYEGA AY495375 166: YAVLDGTHYNDACCFDYGNAEISNTDTGN 194 215:GPWLMADLENGL 226 286:MSLEGAIILGIGGDNSNGAQGTFYEGV AJ310126 166: YAVLDGTHYNGACCFDYGNAETSSTDTGN 194 215:GPWIMADLENGL 226 285:MKKEGAIILGIGGDNSNGAQGTFYEGV AF306764 167: YAVLDGTHYNGGCCFDYGNAETNNLDTGN 195 216:GPWVMADLENGL 227 287:MSKEGAIILGIGGDNSNGAQGTFTEGA
Figure 3-5. Data from a multialignment exercise, using partial sequences, of family GH54 α-L-arabinofuranosidases. Biology WorkBench 3.2 CLUSTALW (San Diego Supercomputer Center, CA, USA) software was used. All enzyme sequences were derived from published gene sequences.
GenBank accession details are: U38661 from Hypocrea koningii G-39, AB085904 from A. kawachii IFO 4308, Z69252 from Hypocrea jecorina RutC-30, AF367026 from Penicillium purpurogenum, AB073861 from Aspergillus oryzae RIB40, AB073860 from Aspergillus oryzae HL15, L23502 from Aspergillus niger, U39942 from Aspergillus niger, Y13759 from Emericella nidulans argB2, AY495375 from Aureobasidium pullulans, AJ310126 from Fusarium oxysporum f. sp. Dianthi, and AF306764 from Cochliobolus carbonum.
O
Figure 3-6. Stereochemical properties and common intermediates of Abf catalysis. (a) Enzymatic reactions, using various substrates, in the presence of methanol. (b) A partial NMR spectrum (chemical shift 3.4–5.4 ppm) of the end-products. Peak assignment is given in the text.
10 8
6 4
2
0
-2
-4
pKa of phenols
log kcat
(a)
10 8
6 4
6
4
2
0
-2
pKa of phenols
log k /Kcatm
(b)