Identification of a histidyl residue in the active center of endoglucanase D from Clostridium thermocellum.

Diethylpyrocarbonate modification of endoglucanase D from Clostridium thermocellum, cloned in Escherichia coli, resulted in a rapid but partial (maximally 70-80%) loss of activity. The second-order rate constant of inactivation proved to be exceptionally high (3210 M-1.min-1). A 3-fold reduction of the kcat and a 2-fold increase of the Km for 2'-chloro-4'-nitrophenyl beta-cellobioside were observed. Spectrophotometric analysis indicate the presence of one rapidly (k = 0.45 min-1) and two slower (k = 0.23 min-1) reacting histidyl residues. In the presence of 50 mM methyl beta-cellotrioside, the rate of inactivation was reduced 16-fold, and the kinetics of modification were compatible with the protection of 1 histidyl residue. Since peptide analysis was inconclusive, identification of the critical residue was attempted by site-directed mutagenesis. Each of the 12 histidyl residues present in the endoglucanase D sequence was mutated into either Ala or Ser. Seven of the mutant enzymes had specific activities lower than 50% of the wild-type. Only in the case of the Ser-516 mutant, however, was the residual activity not affected by diethyl pyrocarbonate. These findings suggest an important functional or structural role for His-516 in the wild-type enzyme.

min") reacting histidyl residues. In the presence of 50 mM methyl &cellotrioside, the rate of inactivation was reduced 16-fold, and the kinetics of modification were compatible with the protection of 1 histidyl residue.
Since peptide analysis was inconclusive, identification of the critical residue was attempted by site-directed mutagenesis. Each of the 12 histidyl residues present in the endoglucanase D sequence was mutated into either Ala or Ser. Seven of the mutant enzymes had specific activities lower than 50% of the wild-type. Only in the case of the Ser-516 mutant, however, was the residual activity not affected by diethyl pyrocarbonate. These findings suggest an important functional or structural role for His-516 in the wild-type enzyme.
More than 50 genes from various organisms (fungi, bacteria, and plants) involved in cellulose or hemicellulose hydrolysis have been cloned and sequenced (Bkguin, 1990). This has provided a wealth of information on the encoded endoglucanases (EC 3.2.1.4) and cellobiohydrolases (EC 3.2.1.91). Hydrophobic cluster analysis has indicated that these enzymes can be classified into six distinct structural families (Henrissat et al., 1989).
As illustrated in the same study, alignments point to a limited number of conserved residues, likely to play an important role either in catalysis or in maintaining the enzymes' structural integrity. Only occasionally have these residues been identified by chemical modification, site-directed mutagenesis, or x-ray diffraction studies. Chemical modification and spectrophotometric analysis have revealed critical tryp-* Work carried out in Paris was supported by Grant EN3B-0082-F from the Commission of European Communities, by Contract MRES-87-C-0396 from the French Ministry of Science and Higher Education, and by research funds from the University of Paris 7. The costs of publication of this article were defrayed in part, by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 645272; To whom correspondence should be addressed: Tel.: 32-91-tophan and carboxylic acid residues in Schizophyllum commune endoglucanase I (Clarke andYaguchi, 1985, 1986;Clarke, 1987). Similarly, in cellobiohydrolase I from Trichoderma reesei an essential glutamic acid residue has been identified (Tomme and Claeyssens, 1989). For the core protein of cellobiohydrolase 11, from the same organism, the threedimensional structure has been determined recently by x-ray diffraction studies and catalytically important amino acid residues localized, using data collected with a ligand diffused into the crystal (Rouvinen et al., 1990). Essential glutamic acid residues in endoglucanases from various Bacillus sp. were studied recently by site-directed mutagenesis (Baird et al., 1990).
Clostridium thermocellum, a Gram-positive, thermophilic bacterium, produces a very active cellulase complex, termed "cellulosome" (Lamed et al., 1983). In this complex, 14-18 different components, many endowed with endoglucanase activity, have been identified. At least 15 distinct genes were clones in Escherichia coli and seven have been sequenced (Bkguin, 1990). The corresponding enzymes can be ordered into three of the cellulase families defined by Henrissat et al. (1989). Because it is easily purified from a hyperproducing clone and readily crystallizes (Joliff et al., 1986b and1986c), endoglucanase D was chosen for structural and functional studies. No information concerning critical or essential residues is available, neither for this nor for any other enzyme of the corresponding cellulase family E (Henrissat et al., 1989).
We present evidence, based on chemical modification and site-directed mutagenesis research, that a histidyl residue plays an important role.
Enzyme Purification-The soluble cytoplasmic fraction from E.
Enzyme Modifications-Endoglucanase D (3 p~ in PC buffer), incubated at 23 "C in the absence or presence of a competitive ligand (50 mM methyl p-cellotrioside), was treated with freshly prepared (EtOCO),O solution (0.08-1.2 mM) made up in ice-cold ethanol. The concentration of the histidine reagent in the commercial stock solution was determined spectrophotometrically at 242 nm (e = 3200 M". cm") (Ovadi et al., 1967). The final ethanol concentration did not exceed 2% (v/v) and did affect neither the activity nor the stability of the enzyme. From the enzyme-modifier mixtures 5-p1 samples were quenched at various time intervals in 400 pl of ice-cold PC buffer containing 1.5 mM N-acetylimidazole. Residual activity, expressed as percentage of this of unmodified enzyme, was then measured after addition of 400 p1 of 2 mM CNPC as described above.
Difference spectra (230-350 nm) of modified uersus native protein were recorded with a double-beam spectrophotometer Uvikon 810 (Kontron, Switzerland) equipped with two thermostated copper cuvette holders. (EtOCO),O, dissolved in ethanol, was added to the sample cuvette, whereas equivalent volumes of the alcohol were added to the reference. The number of modified His was calculated from the absorbance increase at 242 nm (Ovadi et al., 1967;Shina and Brewer, 1985).
Isolation and Analysis of Modified Peptides-60-pl samples of endoglucanase D (30-40 p~ in PC buffer) were treated for 10 min with 2 mM (EtOCO),O (23 'C), in the presence or absence of a ligand, and inactivation was followed as described above. After adding 90 p1 of 0.1 M Tris/HCl buffer, pH 8.0, containing 5 M guanidine HC1 and 3.5 pg of subtilisin to the modified protein samples, the mixtures were incubated for another 9 min at 33 "C. The resulting peptides (-150 pg) were analyzed by reversed-phase liquid chromatography (Waters chromatographic system) on a C4-Vydac 214 TP-54 column (0.46 X 25 cm) using a linear gradient of 0-70% acetonitrile in 0.05% trifluoroacetic acid at a flow rate of 1.5 ml/min and monitoring at 214 and 242 nm (Waters variable-wavelength detector). By comparing the 242 tracings of proteolysates of untreated samples with those where modification was reversed (2 M hydroxylamine, 2 min, 33 "C), peptide fractions containing N-ethoxycarbonylated His were identified.
Site-directed Mutagenesis-Mutagenic oligonucleotides were prepared using a MilliGen/Biosearch DNA synthesizer. His codons were mutated to Ala or Ser, depending on the likelihood of second-site hybridization as determined by computer search. The mismatched nucleotides were flanked on each side by 6-10 bases (total length, 15-20-mers). Mutants of His-65, His-197, His-222, His-286, His-445, His-492, and His-516 were constructed according to the phosphorothioate-based strategy of Taylor et al. (1985), using the mutagenesis kit of Amersham (UK). The pCTH6030 template was obtained by recloning, at the HincII site of M13mp8 (Messing and Vieira, 1982), a 1.7-kb HincII fragment which encodes the catalytic domain of endoglucanase D (Chauvaux et al., 1990). The coding sequence was fused in frame with the 5' end of the lacZ' gene carried by the vector. Endoglucanase activity in cells infected by recombinant phage was detected using the carboxymethylcellulose-Congo Red assay (Teather and Wood, 1982;Cornet et al., 1983).
Mutants of His-167, His-170, His-174, His-187, and His-269 were obtained according to the strategy of Kunkel et al. (1987); i.e. the uracil-substituted template strand, obtained after growth in a dutung-strain, was counter-selected by transformation of a wild-type strain. A mutagenesis kit was used according to the instructions of the supplier (Bio-Rad). Since the CJ236 dutung-host does not carry the amber suppressor required for growth of M13mp8, the 1.7kb fragment containing the catalytic site of endoglucanase D was recloned in M13mp19 (Yanisch-Perron et al., 1985). The coding sequence of the gene was fused in frame with lacZ' at the level of the Hind111 site of the polylinker.
Mutations were identified by sequencing (Sanger et al., 1977) and, for all mutants showing less than 50% of the wild-type activity, the entire, 1.7-kb fragment was checked for other mutations.

Preparation of Crude Extracts from Clones Expressing Mutant
Proteins-3 ml of a 100-fold diluted overnight culture of E. coli TG1 (Wain-Hobson et al., 1985) (absorbance at 600 nm, -0.05) in 2XL medium (2% tryptone, 1% yeast extract, 1% NaC1) was infected with a fresh plaque of recombinant phage. After incubating the culture with vigorous aeration for 5 h at 37 "C, the cells were centrifuged and the supernatant was kept as phage stock. 500 p1 of the latter were used to infect 50 ml of a 100-fold diluted overnight culture of E. coli TG1 in 2XL medium. After 4 h of growth under aeration at 37 "C and centrifugation, cells were harvested, resuspended in 10 ml of PC buffer, and disrupted by sonication. The extracts were cleared by centrifugation (5 min; 3000 X g) and kept in aliquots at -20 'C. Immunoblotting Assay of Endoglucanase D-Enzyme concentrations in crude extracts of strains expressing mutant genes were estimated from Western blots. 40 pg of protein from each crude extract, containing 20-120 ng of antigen, was loaded on a 10% SDSpolyacrylamide gel. After electrophoresis, proteins were transferred to nitrocellulose (Towbin et al., 1979), and the blot was probed with anti-endoglucanase D antiserum preadsorbed with crude extract of E. coli TGl(pUC8) (GrBpinet et al., 1988), followed by 1251-labeled protein A. Bands located by autoradiography were cut out, weighed, and counted by y-ray scintillation. Values were corrected for nonspecifically bound radioactivity by substracting the number of counts per mg of nitrocellulose detected in a weighed sample from a blank area of the blot. The actual amount of endoglucanase D antigen was estimated from a standard curve obtained with purified protein.
Modification of Mutant Proteins with (EtOCO)zO-Protein concentration in the crude extracts (measured according to Sedmak and Grossberg, 1977, using bovine serum albumin as a standard) was adjusted with PC buffer to 1.85 mg/ml, and (EtOCO),O was added to a final concentration of 5.5 mM. After incubating at room temperature for 0-8 min, 100-pl samples were withdrawn and quenched by diluting into 400 pl of PC buffer containing 2.5 mM N-acetylimidazole. The activity was determined as described above.

RESULTS
Inactivation Kinetics and Specificity of Modification-Endoglucanase D (3-4 WM) was rapidly inactivated by excess (EtOC0)20 (0.08-1.2 mM) and pseudo-first-order kinetics applied for the early part of the inactivation curves (Fig. 1A). However, even in the presence of high initial modifier concentrations and after prolonged incubations, the N-ethoxycarbonylated enzyme retained 20-30% of its activity. An intermediary enzyme-(EtOCO)zO complex was probably not formed (Fig. 1B), since the pseudo-first-order rate constants (kapp) depended linearly on the modifier concentration. The second-order rate constant for inactivation (3210 min" exceeded most published values (e.g. Lundblad and Noyes, 1984;Van Grysperre et al., 1988).
Treatment of modified endoglucanase D with 50 mM hydroxylamine resulted in complete reactivation of the enzyme within 20 min (Fig. 2B). Since this reagent is known to reverse N-ethoxycarbonylation of modified Tyr and His residues only (Burnstein et al., 1974), reactivation excludes the possible involvement of Cys or Lys residues (Melchior and Fahrney, 1970;Miles, 1977). Incubation of endoglucanase D with tetranitromethane, in 5-or 10-fold molar excess over the total amount of Tyr, did not affect enzymatic activity. Furthermore, under native conditions and using several specific reagents (mercury compounds, dithio derivatives, or maleimides), no inactivation due to the modification of thiol groups was observed.' P. Tomme, S. Chauvaux, P. BBguin, J. Millet, J.-P. Aubert, and M. Claeyssens, unpublished data.

C. thermocellum Endoglucanase D 10315
Specific reaction was confirmed by difference spectra analysis of native and modified endoglucanase D (Fig. 3). The absorbance maximum at 240-243 nm, characteristic for Nethoxycarbonylated His, disappeared upon hydroxylamine treatment (Ovadi et al., 1967). A discrete minimum apparent at 290 nm could be due to the perturbation of a Trp or Tyr residue in the proximity of a modified His (De Boeck et al., 1984). the modification reaction (Ovadi et al., 1967;Shina and Brewer, 1985). Under the conditions used ( Fig. 2 A ) , endoglucanase D, incubated in the absence of inhibitor, was converted to a form with 30% residual activity. The pseudo-firstorder constant of inactivation was calculated as 0.45 rnin".

Number of Modified H i s Residues and Protection by
During the first 10 min, the kinetics of modification paralleled inactivation (Fig. 2 A ) and were compatible with the reaction of 3 residues. Slow reaction of a fourth residue became apparent after longer incubation times (>30 min) but caused no further activity loss. Assuming that the modification of the first residue occurred at the initial inactivation rate, 0.45 rnin", curve-fitting applied to all experimental data (Fig. 2 A ) yielded an average rate constant of 0.23 min" for the other 2 His residues.
In the presence of methyl P-cellotrioside, inactivation was 16-fold reduced ( k = 0.028 min"). The degree of modification was compatible with the protection of 1 His residue (Fig. 2 A ) . Similarly as above, data could be fitted assuming 1 His residue to react with a pseudo-first-order rate constant of 0.028 min" and the other 2 residues at 0.17 min-l. Catalytic Properties of Modified Endoglucanase D-The catalytic parameters for CNPC hydrolysis determined for modified endoglucanase D are reported in Table I. Compared to the values for intact enzyme. K,,, was increased %fold and kcat was three times lower. Since there is only partial activity loss and changes appear in both parameters, this suggests the presence of homogeneously modified endoglucanase D with reduced catalytic efficiency. The effect of the modification on the activation energy (AAG = 4.6 kJ/mol) could be typical for the loss of an uncharged hydrogen bond in the transition state complex in the modified enzyme (Fersht, 1987).
Identification of Modified His Residues by Peptide Mapping-Proteolysates of endoglucanase D, modified in the presence or absence of 50 mM methyl P-cellotrioside, were analyzed by reversed-phase liquid chromatography (Tomme and Claeyssens, 1989). Since comparison of the chromatograms did not lead to the detection of specifically labeled fractions, differential peptide mapping was attempted using proteolysates before and after hydroxylamine treatment (Fig. 4, A and  B ) . This led to the identification of three peptides in accordance with the number of modified His residues (see above). The partial sequences determined were compatible with modification of residue His-286 in peptide I (Z75NFGGFIMPEN(EHD)287) and of His-65 in I1 ("'NAALDAISHV''). Peptide I11 was probably blocked as it  5 mM, pH 6.4), and assayed for residual activity (circles). The number of modified His residues (squares) was determined as described under "Materials and Methods." Curves were fitted using the first-order rate constants given in the text. B, reactivation (23 "C) with 50 mM hydroxylamine.

TABLE I Kinetic parameters of intact and diethyl pyrocarbonate-modified
endoglucanase D Activity measurements were performed at 25 "C, pH 6.4, with CNPC (0.08-1.0 mM) as described under "Materials and Methods." Kinetic parameters were derived from ( S ) / u uersus (S) plots (Hanes, 1932). proved recalcitrant to NHp-terminal sequencing.

Site-directed Mutagenesis of His Residues-Each of the 12
His residues present in the enzyme was changed by sitedirected mutagenesis into Ala or Ser. The activity of the mutated proteins was measured in cell extracts. In order to correct for variations in the rate of expression between the different mutants, the concentration of endoglucanase D in the extracts was assayed by Western blotting. Since endoglucanase D antigen was estimated after SDS-polyacrylamide gel electrophoresis, this could be limited to full-size protein, precluding possible interference by degradation products. Furthermore, since the assay was normalized for denatured wildtype protein, artefacts due to the loss of conformational epitopes by the mutated proteins could be avoided. The results of the mutagenesis experiments are gathered in Table 11. Mutations of His-65, His-167, His-170, His-187, His-269 did not result in significant loss of activity relative to isogenic wild-type constructions. However, the residual activity of proteins mutated in His-174, His-197, His-222, His-286, His-445, His-492, and His-516 ranged from 2 to 54%.

FIG. 4.
Reversed-phase liquid chromatagraphy of modified endoglucanase D proteolysates before (A) and after ( B ) treatment with hydroxylamine. Endoglucanase D was modified and processed as described under "Materials and Methods." Peptides I, 11, and 111 were collected for further amino acid sequence analysis.
Mutants Ala-197 and Ser-516 were thermolabile, but at 30 "C both enzymes were stable and their activity could be compared with that of the wild-type.
Since the identity of the (EtOCO)zO-sensitive histidyl residue in the native protein could not unequivocally be established by these activity assays, the inactivation of the various mutated enzymes by the same reagent was investigated. Indeed, the enzyme mutated at the critical target should be resistant to the reagent. Contrary to the wild-type, and to all other His mutants (data not shown), only the activity of the enzyme bearing the His-516 -Ser mutation was not affected by (EtOCO)zO (Fig. 5).

DISCUSSION
All the evidence presented above point to the presence of a His residue in the catalytic center of endoglucanase D from C. thermocellum. His-specific modification of endoglucanase D with (EtOCO),O resulted in 70-80% loss of activity. Inactivation of the enzyme was prevented in the presence of methyl P-cellotrioside, suggestive for modification of a critical histidine within the active center. The identification of His-516 as the critical residue was based on site-directed mutagenesis studies. The specific activity of the His-516 + Ser The percentage residual activity (CNPC, 60 "C) of each mutant is expressed relative to the protein produced by the isogenic wildtype strain (pCT6030 and pCT6090, respectively). Endoglucanase activity present in extracts of infected cells was related to the amount of full-size endoglucanase D detected by immunoblotting as described under "Materials and Methods." * Thermosensitive proteins encoded by mutants pCTH6036 and pCTH60312 were assayed at 30 "C and activities compared with this of wild-type enzyme measured under the same conditions. were treated with the modifying reagent and residual activity was assayed at 30 "C as described under "Materials and Methods." mutant was reduced by 75%, similar to the effect observed upon N-ethoxycarbonylation of native enzyme. Furthermore, the residual activity in this mutant proved insensitive to (EtOCO),O modification. Peptide analysis allowed the identification of the 2 other residues modifiable with diethyl pyrocarbonate, His-65 and His-286. Mutation of His-65 to Ala did not result in significant inactivation; mutation of His-286 to Ala resulted in 81% inactivation (Table II), but the residual activity of the mutated protein was still sensitive to (EtOCO),O modification. Since (EtOC0)ZO treatment did not inactivate the enzyme carrying the His-516 + Ser mutation, it may be that replacement of His-286 by Ala was more harmful to the structure of the protein than N-ethoxycarbonylation.
The exact role of His-516 remains to be defined. It does not seem absolutely essential for catalysis, since the mutant is still enzymatically active and changes in kinetic parameters for a substrate such as CNPC are small. It could tentatively be suggested that in the wild-type enzyme this histidyl residue is implicated in the formation of a salt bridge or hydrogen bond to the substrate transition state (Fersht, 1987). Alternatively, the lower thermostability of the Ser-516 mutant could point to a role in the maintenance of an active enzyme conformation. The presence of critical carboxylic acid groups  (Joliff et al., 1986a) and in at least two of the other cellulases: Ps. flu. EGA, P. fluorescens endoglucanase A (Hall and Gilbert, 1988); P. amer. EG, Persea americana (avocado) endoglucanase (Tucker et al., 1987); and Cel. fi. CenB, Cellulomonas fimi endoglucanase B (Meinke et al., 1991). His residues in endoglucanase D are numbered as shown. Numbers at the start of the alignments refer to the positions of the amino acids in the sequence of the polypeptides including the signals. Dashes indicate gaps left to improve the alignment.
in the 500-550 region was shown by chemical modification3 and mutagenesis4 studies and is consistent with the hypothesis that this sequence is indeed important for activity.
As expected for a catalytically or structurally important region, His-516 and the neighboring amino acids are conserved in related cellulases from family E (Henrissat et al., 1989). None of the other His residues is conserved in these enzymes (Fig. 6). As shown here for endoglucanase D, however, mutations of several of the variable sites to Ala or Ser lead to significant activity losses. Indeed, the strongest inactivation (98%) was observed for mutant His-174, a residue which is only conserved in P. fluorescens endoglucanase A. The reduced activity of the mutants could indicate that the corresponding sites can only tolerate a limited change in amino acids residues. Alternatively, some of the interactions responsible for the structure and function of the enzymes may vary between the different proteins even if they belong to the same family.