Redesign of the Substrate-binding Site of Hen Egg White Lysozyme Based on the Molecular Evolution of C-type Lysozymes*

On the basis of the molecular evolution of hen egg white, human, and turkey lysozymes, three replace- ments (Trpez with Tyr, Asns7 with Gly, and Asp''' with Gly) were introduced into the active-site cleft of hen egg white lysozyme by site-directed mutagenesis. The replacement of TrpeZ with Tyr led to enhanced bacteriolytic activity at pH 6.2 and a lower binding constant for chitotriose. The fluorescence spectral properties of this mutant hen egg white lysozyme were found to be similar to those of human lysozyme, which contains Tyr at position 62. The replacement of Asns7 with Gly had little effect on the enzymatic activity and binding constant for chitotriose. However, the combination of AsnS7 -.) Gly (N37G) replacement with Asp"' -.) Gly (DlOlG) and Trpa2 + Tyr (W62Y) conversions enhanced bacteriolytic activity much more than each single mutation and restored hydrolytic activity toward glycol chitin. Consequently, the mutant lysozyme containing triple replacements (N37G, W62Y, and DlOlG) showed about %fold higher bacteriolytic activity than the wild-type hen lysozyme at pH 6.2, which is close to the optimum pH of the wild-type enzyme. C-type lysozymes from many vertebrates and some inver-tebrates

the substrate-binding region among C-type lysozymes. However, a number of substitutions that may modulate enzymatic activity have been observed (JollBs and JollBs, 1984). For example, of hen egg white lysozyme is replaced with Tyr in human and rat lysozymes, which exhibit 2-4-fold enhanced bacteriolytic activity (Mulvey et al., 1973(Mulvey et al., , 1974. We converted Trp" of hen egg white lysozyme to Tyr by sitedirected mutagenesis and found that the mutated lysozyme showed enhanced bacteriolytic activity (Kumagai et al., 1987;Kumagai and Miura, 1989).
Although human lysozyme shows 4-fold higher lytic activity than does hen egg white enzyme, the mutant hen lysozyme in which TI$* was replaced with Tyr showed at best twice the lytic activity of the wild-type hen enzyme. Another difference in substrate-binding sites between human and hen enzymes is the replacement of in the hen enzyme with Gly. Asp"' is a conserved amino acid residue in C-type lysozymes. In turkey lysozyme, however, Asplo' is replaced with Gly, and the lysozyme also exhibits 1.5-fold higher lytic activity (LaRue and Speck, 1970). Fig. 1 shows amino acid substitutions that are found in substrate-binding sites of the three C-type lysozymes.
It is believed that Asp"' interacts with the A and B sugar residues at the A and B subsites by hydrogen bonding and that Trp6* directly interacts with GlcNAc residues bound to subsites B and C. On the other hand, is thought to contribute to the binding of sugar residues at the F subsite, on the basis of model-building studies (Blake et al., 1967).
We then introduced these three replacements found in nature either alone or in combination into the active site of hen egg white lysozyme by site-directed mutagenesis. It was found that the mutant lysozyme in which all three replacements (N37G, W62Y, and DlOlG) had been introduced showed the highest lytic activity. We report the findings in this paper, together with the fluorescence spectroscopic properties of the mutated lysozymes, and the effects of these replacements on their ligand-binding activities.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases, DNA-modifying enzymes, and DNA-sequencing kits were purchased from Takara Shuzo (Kyoto) and Toyobo (Osaka). Micrococcus lysodeikticus cells were from Sigma. Hen egg white lysozyme, glycol chitin, and N-acetylchitooligosaccharides were from Seikagaku Kogyo (Tokyo). S-Sepharose (fast flow) was obtained from Pharmacia. CM-Toyopearl650M and TSK gel ODs-1MA columns were products of Tosoh (Tokyo). Other chemicals were of reagent grade.
Site-directed Mutagenesis-A 20-base and two 19-base mutation primers synthesized by phosphoramidate chemistry were used to replace the A d 7 codon, AAC, with the Gly codon, GGT; the Trps2 codon, TGG, with the Tyr codon, TAT; and the Asplo' codon, GAT, with the Gly codon, GGT. Site-directed mutagenesis was carried out by the methods described by Morinaga et al. (1984) using a doublestranded plasmid, pKK-1, which contains a hen egg white lysozyme Redesign of the Active Site of Hen Lysozyme 4609 FIG. 1. Comparison of amino acid sequence and the active-site residues of C-type lysozymes. Numbering of amino acid residues is based on that of hen egg white lysozyme. For human and turkey lysozymes, only residues that are different relative to hen egg white lysozyme are indicated. The amino acid residues of hen egg white lysozyme whose side chains are expected to interact with the substrate are underlined, and among these, the uncommon residues of the three lysozymes are double underlined.
cDNA (Kumagai et al., 1987;Kumagai and Miura, 1989). The cDNAs carrying double and triple mutations were constructed by utilizing the unique Sac1 and HinfI sites of the lysozyme cDNA. The mutations were confirmed by DNA sequencing analysis.
Expression of the Mutant Hen Egg White Lysozymes-For construction of the expression plasmids, the mutated hen lysozyme cDNAs were inserted into the Sal1 site of pYG-100, as previously described (Kumagai and Miura, 1989). The expression plasmids were introduced into Saccharomyces cerevisiue AH22 strain (MAT a, leu2, his4, cir+) and the Leu+ transformants were grown in yeast minimal medium supplemented with histidine (20 pglml). After cultivation of the transformanta at 30 "C for 120 h, the mutant lysozymes secreted into the growth medium were purified by cation-exchange chromatography on S-Sepharose (fast flow) followed by a CM-Toyopearl 650M column. The size and homogeneity of the purified enzymes were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis according to the method of Laemmli (1970). The enzyme concentration was determined spectrophotometrically (El% at 280 nm = 26.4 for hen lysozyme and 25.7 for human lysozyme), and the molar absorption coefficients of the mutant lysozyme containing W62Y were adjusted by a factor of 0.89 (Kumagai et al., 1987).
Confirmation of Replacement of Trp" with Tyr in the Mutant Hen Lysozyme by Peptide Mapping Analysis-The wild-type hen lysozyme and the W62Y mutant lysozyme were denatured with 6 M guanidine hydrochloride and reduced with 8-mercaptoethanol, After S-carboxymethylation, the derivatives were digested with ~-l-tosylamido-2phenylethyl chloromethyl ketone-treated trypsin (1% lysozyme by weight) at 37 "C for 10 h. The tryptic peptides were separated by reverse-phase high performance liquid chromatography (TSK gel ODs-12OA 4.6 X 250 mm, Tosoh). The elution was performed by applying a linear gradient of 1-40% acetonitrile in 0.1% concentrated hydrochloric acid for 100 min at a flow rate of 0.8 ml/min. Peptide peaks were detected at both 210 and 280 nm. The peptides containing Trp and Tyr were collected and subjected to amino acid sequence analysis on an Applied Biosystems 473A Protein Sequencer.
The high performance liquid chromatography analysis patterns for the wild-type and the W62Y mutant enzymes appeared to be exactly the same, except that one peak of the mutant detected at 280 nm was eluted at an earlier retention time than the corresponding peak of the wild-type. Sequence analysis of this peptide peak clearly showed the replacement of Trp with Tyr at position 62 of the W62Y mutant enzyme (data not shown).
Enzymatic Assays-Bacteriolytic activity of lysozyme was assayed by the method of Parry et al. (1965) with slight modification. Lysozyme was added to 1 ml of a suspension of M. lysodeikticus cells in 50 mM sodium phosphate buffer (pH 6.2) or in 0.1 M NaCl and 50 " " " " " " " mM Tris-HC1 (pH 7.5) at 25 "C. Absorbance at 540 nm was monitored on a Hitachi U-3210 spectrophotometer against a water blank. The decrease in absorbance at 540 nm after 1 min was measured and compared with the wild-type and mutant lysozymes. Hydrolytic activity toward glycol chitin was measured following the reducing group procedure (Imoto and Yagishita, 1971). Glycol chitin (0.05% (w/v)) was incubated with or without lysozyme at 25 "C for 1 h in 20 mM sodium acetate buffer (pH 5.0), whose ionic strength was adjusted to 0.1 with NaC1, or in 0.1 M NaCl and 50 mM Tris-HC1 (pH 7.5). Fluorescence Spectra-A Hitachi 850 fluorescence spectrophotometer was used for the measurement of fluorescence spectra. The excitation wavelength was 280 nm, and the protein concentration was 0.012 mg/ml. FWative quantum yields were determined in 0.1 M NaCl and 50 mM Tris-HC1 (pH 7.5) at 25 "C by comparing the protein emission spectrum with that of tryptophan solution in the same buffer (Mulvey et al., 1973). Because the quantum efficiencies of tryptophan in water and the above assay buffer were within 2%, we assumed the quantum efficiency of tryptophan to be 0.2 (Teale and Weber, 1957).
Binding Constants of N-Acetylchitooligosacchrides-The fluorescence spectra of a series of enzyme-inhibitor complex solutions were measured with (G~CNAC)~, (GlcNAc)z, and GlcNAc in the assay buffer at 25 "C. Five concentrations of GlcNAc were used from 0 to 0.13 X 1O-l M. The six of (GlcNAc)z'were from 0 to 7 X 1O-l M. The seven of (GlcNAc), were from 0 to 1 X M. The data are analyzed by the derived Scatchard plot of Mulvey et al. (1973) where F E and F, . are the fluorescence intensities of the free enzyme and the saturated complex, respectively, and F is the fluorescence intensity in the presence of I mol of the added inhibitors. The binding constants of K. were evaluated by the slopes in the plots of (

RESULTS
Fluorescence Spectra of the W62Y Mutant Lysozyme-The fluorescence emission spectra of the W62Y mutant lysozyme and its complex with the oligomer of N-acetylglucosamine are shown in Fig. 2. The emission spectrum of the wild-type lysozyme is included for comparison. The emission maximum for the mutant lysozyme was found to be 336 nm, which is at a slightly shorter wavelength than that of the wild-type enzyme (341 nm). When (GICNAC)~ is bound to the wild-type lysozymes at pH 7.5, the fluorescence emission is enhanced and its maximum is blue-shifted from 341 to 335 nm. On the other hand, the quantum yield of the W62Y mutant was found to be decreased and estimated to be 0.047 according to the procedure of Mulvey et al. (1973). This value is significantly lower than that of the wild-type enzyme (0.06). The magnitude of the fluorescence enhancement of the W62Y mutant on (G~CNAC)~ binding was smaller than that of the wild-type enzyme, and its emission maximum was scarcely blue-shifted. These spectral features seem to be closely related to those of human lysozyme, which contains a Tyr residue at position 62 of hen egg white lysozyme (Mulvey et al., 1973).
Binding Constants of N-Acetylglucosamine Oligomers to the W62Y Mutant Lysozyme-The binding constants of (GlcNAc)& (GlcNAc)z, and GlcNAc to the wild-type and the W62Y mutant lysozyme at pH 7.5 and 25 "C were estimated from the changes in the fluorescence spectra upon ligand binding and are summarized in Table I. The binding constants for GlcNAc to both enzymes were found to be essentially unchanged. On the other hand, those of (GlcNAc)z and (G~CNAC)~ to the W62Y mutant lysozyme decreased significantly. It was shown by X-ray crystallography that GlcNAc binds to the C subsite, (GlcNAc), to the B and C subsites, and (G1cNAc)a to the A, B, and C subsites in the substratebinding sites of the wild-type lysozyme. If it is assumed that the binding mode of these GlcNAc residues to the mutant lysozymes is the same as that of the wild-type enzyme, the binding of GlcNAc monomer to the C subsite seems to be unchanged upon conversion of Trp6' to Tyr. However, since the free energy change on the binding of (GlcNAc)z to the mutant lysozyme was found to be very close to that of (G~CNAC)~ (Table I), the replacement of Trp6' may mainly affect the binding of GlcNAc to the B subsite.
Enzymatic Activities of the Single and Multiple Mutated Lysozymes-We have already reported the enhanced bacteriolytic activity of the W62Y mutant lysozyme (Kumagai et al., 1987). Fig. 3 summarizes the bacteriolytic activities of wild-type, three single mutants (N37G, W62Y, and DlOlG), three double mutants (N37G W62Y, W62Y DlOlG, and N37G DlOlG), one triple mutant (N37G W62Y DlOlG), and human lysozyme. At pH 6.2, replacement of TrpeZ with Tyr (Kumagai et al., 1987) and Asp"' with Gly enhanced the bacteriolytic activity. The double mutants exhibited much more enhanced activities. Although a single replacement of with Gly alone did not increase the activity, the combination of the N37G mutation with either W62Y or DlOlG gave remarkably enhanced activity. The mutant N37G W62Y contains the same set of amino acid residues involved in substrate binding by human lysozyme. The triple mutant (N37G W62Y DlOlOG) showed about 3-fold higher lytic activity. However, this was still lower than that of human lysozyme.
The lytic activities of these mutants at pH 7.5 were lower than that of the wild-type enzyme at pH 7.5, except for one double mutant (N37G DlOlG). It is noteworthy that the activities of mutants containing W62Y were greatly reduced at pH 7.5. Fig. 4 compares the hydrolytic activities toward glycol chitin of wild-type, the mutant lysozymes described above, and human lysozyme. At pH 5.0, which is the optimum pH for the wild-type enzyme, introduction of single mutations reduced the activities of the mutant enzymes. On the other hand, double mutations restored the hydrolytic activities on glycol chitin. In particular, the replacement of Ama7 with Gly of the single mutant enzymes seemed to be very effective in recovering their activities. As a consequence, two double mutants involving conversion of Asna7 to Gly showed almost the same activity as the wild-type enzyme. In addition, the triple mutant (N37G W62G DlOlG) exhibited about 90% of the wildtype lysozyme activity. As observed with bacteriolytic activities, most mutants of the enzymes showed reduced activities toward glycol chitin at pH 7.5.
Binding of (GlcNAc)s to the Mutant Lysozymes-The fluorescence emission spectra of the mutant enzymes containing conversions of Ama7 to Gly and Asp"' to Gly were essentially toward glycol chitin. The activities were measured in 20 mM sodium acetate buffer (pH 5.0), whose ionic strength was adjusted to 0.1 with NaCl, or in 0.1 M NaCl and Tris-HC1 (pH 7.5), as described under "Experimental Procedures." Relative activities are expressed by taking the activity of the wild-type hen lysozyme at pH 5.0 to be 100.

Comparison of binding constants of (GlcNA& to the wild-type hen, the mutant hen, and human lysozymes
Binding constants of the inhibitor were determined by fluorescence spectral changes as described under "Experimental Procedures."

hf-1
Wild-type 7.5 Human 2.6 X 104 7.0 X 10' the same as that of the wild-type enzyme. On the other hand, double or triple mutants containing Tyr at position 62 showed the same emission spectra as the W62Y mutant. Therefore, conversion of Asp''' and seem not to influence fluorescence. Table I1 summarizes the binding constants of (GlcNAc), to the mutant lysozymes measured by changes in the fluorescence spectra. As clearly shown in the table, the conversion of Am3', which is probably located near the F subsite, did not affect the binding of (GlcNAc), to sites A, B, and C. On the other hand, conversion of Asp"' and Trp6', which comprise the A, B, and C subsites, led to large reductions in the binding constants. Analysis of changes in free energy on binding to single and double mutant enzymes revealed that the effects of conversion of Asp'" and Trp" on ligand binding are interdependent. The estimated free energy change on replacement of Asp"' with Gly was almost consistent with that reported by Kirsch et al. (1989).

DISCUSSION
Hen egg white lysozyme contains six tryptophanyl residues. Of these, three tryptophans (Trp6', Trp6,, and Trp'") are located in the substrate-binding cleft and are believed to interact with sugar rings placed in subsites B, C, and D.
Conversion of Trp6' to Tyr by site-directed mutagenesis markedly altered the fluorescence properties of the protein, which may be explained in terms of the spectroscopic features of Trp6' and Trp'''.
Trp6' is the tryptophan most exposed to solvent and is very susceptible to chemical reagents (Hayashi et al., 1965). A chemical modification study suggested that Trp6' provides 35-38% of the fluorescence of the lysozyme . Trp6' is expected to emit at a longer wavelength. The other major fluorophore of the enzyme is Trp'", which is Site of Hen Lysozyme 461 1 partially buried and may emit at a shorter wavelength. The slight blue shift of the emission maximum upon (GlcNAc), binding is due to the removal of Trp6' from its exposed, solvent-accessible region to a less polar environment. The replacement of Trpsz with Tyr led to a reduction of the quantum yield and blue shift of the emission maximum. Since the contribution of the Tyr residue to fluorescence is probably very small, Trp'" may dominate the fluorescence spectrum of the Tyr6' mutant. This is further supported by the fact that the fluorescence of the W62Y mutant was found to be more similar to that of human lysozyme, in which Trp6' is replaced with Tyr, and Trpa3 and Trp'" are conserved. The quantum yield of the W62Y mutant was between those of the wild-type hen and human lysozymes. The mutant showed fluorescence emission maximum at the same wavelength as human lysozyme, which is shorter than that of the wild-type hen enzyme.
Studies of the binding of ligands to the W62Y mutant showed that the mutant binds (GlcNAc), and (GICNAC)~ less strongly than does the wild-type lysozyme. On the other hand, the GlcNAc binding constant to the mutant was almost identical with that of the wild-type enzyme. Analysis by x-ray crystallography suggests that Trp6' interacts with the sugar residue at subsite B by van der Waals contact and with the sugar C-6 hydroxyl group at subsite C through hydrogen bonding. However, studies on the lysozyme-GlcNAc complex by NMR (Cassels et al., 1978) and the binding capability of 6-deoxy-GlcNAc to the enzyme have shown that such a hydrogen bond would be too weak for complex formation. The observation that there is little difference between interactions of GlcNAc with the wild-type and the mutant lysozyme suggests a small contribution of hydrogen bonding between Trp" and GlcNAc residue for the inhibitor binding. Alternatively, the reduced binding constants for (GlcNAc), and (GlcNAc), to the W62Y mutant may be due to the difference in nonpolar interactions with Tyr and Trp residues. Of course, the possibility that the binding mode of these inhibitors and the topographies of their binding sites could be altered by the replacement of Trp with Tyr cannot be ruled out.
On the basis of molecular evolution of three C-type lysozymes, multiple mutations were introduced into the substratebinding region of hen egg white lysozyme, and unique features of the mutant enzymes were found. The replacement of Am3' with Gly did not affect the binding of (GlcNAc), to the enzymes, which is consistent with the observation that (GlcNAc), binds to the A, B, and C subsites. On the other hand, mutations of Asp'" and Trp6, led to a significant reduction in the binding constant. Judging from the analysis of Asp'" and Trp6' double mutations, the effects of these mutations on (GlcNAc), binding were not independent, probably due to subtle structural changes in the A, B, and C subsites. With the exception of mutations enhanced bacteriolytic activity but reduced hydrolytic activity on glycol chitin and the binding constant of (GlcNAc),. However, it should be noted that replacement of Asn37 with Gly in combination with the Asp"' and Trp6' mutations gave an even greater enhancement of bacteriolytic activity and restored the hydrolytic activity on glycol chitin. This observation will provide some insight into the contribution of the F subsite to substrate binding, which has previously been possible only by model building (Blake et al., 1967) or energy calculation (Pincus and Scheraga, 1979).
Most of the mutant lysozymes showed enhanced bacteriolytic activity and reduced hydrolytic activity on glycol chitin. In preliminary kinetic experiments using small synthetic substrates, we have found a reduced K, but increased kc, of the Tyr6, mutant (data not shown). The cell surface of M. lyso-deikticus cells is highly negatively charged, and lysozyme is a very basic protein. On binding of the lysozyme to the cells, electrostatic interaction between them may be dominant, and the efficiency of hydrolysis of the -GlcNAcpl-4MurNAclinkage may directly reflect the lytic process. Alternatively, about 40% of -GlcNAc~1-4MurNAc-linkages are anchored to the peptidoglycan layer through short peptides linked to lactic acid groups of MurNAc residues in the M. lysodeikticus cell wall. Therefore, the conformation of the -GlcNAc@l-4MurNAc-polymer may be more rigid than the free chitin or glycol chitin polymers. Slightly altered topographies of substrate-binding sites of the mutant lysozymes may more readily accommodate the rigid structure of -GlcNAc@l-4-MurNAcregions in the M. lysodeikticus cell wall.