The Mechanism of Soluble Peptidoglycan Hydrolysis by an Autolytic Muramidase A PROCESSIVE

The action of purified N-acetylmuramoylhydrolase (muramidase, EC 3.2.1.17) of Streptococcus faecium ATCC 9790 on linear, uncross-linked, soluble, peptidoglycan chains produced by the same organism in the presence of benzylpenicillin was characterized as a processive exodisaccharidase. Specific labels, one ([“CIGal) added to the nonreducing ends of chains, and the other (3H from [‘H]N&H4) incorporated into the reducing ends of the chains, were used to establish that an enzyme molecule binds at the nonreducing terminus and sequentially hydrolyzes the glycosidic bonds, re- leasing disaccharide-peptide units. An enzyme molecule remains bound to a chain, and is not released at a detectable rate, until hydrolysis of that chain is complete. Reaction rates increased with the length of the polymer chain to give a maximum of 9 1 bonds cleaved/ min/enzyme molecule for hydrolysis of a continuous polymeric substrate. The relationship between hydro- lytic rate and glycan chain length is consistent with hydrolysis of bonds within the chain followed by slow release of enzyme from the distal, reducing terminus. This mechanism was experimentally confirmed by analysis of product formation during hydrolysis with stoichiometric mixtures of enzyme and soluble pepti- doglycan chains. Kinetic analyses showed an apparent K,,,


The Mechanism of Soluble Peptidoglycan Hydrolysis by an Autolytic Muramidase
A PROCESSIVE EXODISACCHARIDASE* (Received for publication, December 5, 1983, andin revised form, May 12, 1984) John F. Barrett$$, David L. Dolinger$, Vern L. Schrammq, and Gerald D. ShockmanSII From the *Department of Microbiology and Immunology and the TDepartment of Biochemistry, Tempk University School of Medicine, Philadelphia, Pennsylvania 19140 The action of purified N-acetylmuramoylhydrolase (muramidase, EC 3.2.1.17) of Streptococcus faecium ATCC 9790 on linear, uncross-linked, soluble, peptidoglycan chains produced by the same organism in the presence of benzylpenicillin was characterized as a processive exodisaccharidase. Specific labels, one (["CIGal) added to the nonreducing ends of chains, and the other (3H from ['H]N&H4) incorporated into the reducing ends of the chains, were used to establish that an enzyme molecule binds at the nonreducing terminus and sequentially hydrolyzes the glycosidic bonds, releasing disaccharide-peptide units. An enzyme molecule remains bound to a chain, and is not released at a detectable rate, until hydrolysis of that chain is complete. Reaction rates increased with the length of the polymer chain to give a maximum of 9 1 bonds cleaved/ min/enzyme molecule for hydrolysis of a continuous polymeric substrate. The relationship between hydrolytic rate and glycan chain length is consistent with hydrolysis of bonds within the chain followed by slow release of enzyme from the distal, reducing terminus. This mechanism was experimentally confirmed by analysis of product formation during hydrolysis with stoichiometric mixtures of enzyme and soluble peptidoglycan chains. Kinetic analyses showed an apparent K,,, of 0.17 PM for the enzyme, independent of substrate polymer length. The dissociation constant for the initial enzyme-substrate complex was calculated to be 1.5 nM. Kinetic analyses are consistent with one catalytic site per enzyme molecule. The koaJK,,, value of 9 X lo6 initial hydrolytic events when long chains are hydrolyzed. The kinetic and physical properties of this muramidase are highly consistent with its location outside of the cellular permeability barrier and its ability to remain with and hydrolyze appropriate bonds in the cell wall in such sin environment. "1 s-l is near the limit imposed by diffusion for the Autolytic peptidoglycan hydrolases (autolysins) are endogenous bacterial enzymes that hydrolyze specific bonds within insoluble cell walls, thereby causing the walls to lose their *This research was supported by United States Public Health Service Research Grant AI 05044. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Predoctoral trainee supported by United States Public Health Service National Research Service Award Training Grant AI 07101.
Present address, Department of Biology, Washington University, St.
)I To whom correspondence should be addressed.
structural integrity and the cells to lyse unless external osmotic protection is provided (1). There have been numerous studies of the substrate specificities of such enzymes (see Ref. 2 for a recent review), and enzymes that hydrolyze the peptide, amide, and glycosidic bonds in wall peptidoglycans have been described (1-3). Several bacterial species have been found to contain enzymes that hydrolyze more than one type of bond within its peptidoglycan. However, there have been few studies of the mechanism by which bacterial autolysins attack their substrates. Such studies (4-7) have been limited by difficulties in obtaining pure enzymes and by insolubility of the chemically complex wall substrate. Most studies of the peptidoglycan hydrolases have used either (i) insoluble substrates, such as walls themselves, or insoluble peptidoglycan residues (obtained by treatment of walls with strong acids or bases); (ii) low-molecular-weight peptidoglycan fragments consisting of structural units of the wall peptidoglycan, such as DSP1' or peptide cross-linked dimers of DSP, units (obtained from enzymatic hydrolysates of walls), or (iii) lowmolecular-weight oligosaccharides or peptides obtained via chemical synthesis or chemical hydrolysis of analogous substrates, such as chitin. The latter type of substrates have been very successfully used for the detailed examination of the mechanism of action of HEWL.
Recently it has become possible to isolate and purify speptidoglycans produced and secreted by penicillin-treated bacteria (8-12). These soluble polymers contain the bonds normally hydrolyzed by their respective autolysins in the same (or a very similar) relationship to nearby substituents as the homologous, insoluble cell wall. Thus, they can be regarded as "natural" substrates for autolysins, except for the absence of the peptide cross-bridges normally hydrolyzed by the plactam-sensitive peptidoglycan hydrolases.
As described elsewhere (13), high-molecular-weight s-peptidoglycans, produced by Streptococcus faecium ATCC 9790 in the presence of penicillin, have been isolated, purified, and chemically characterized. These polymers were shown to contain glycan strands of approximately uniform length that are fully substituted with peptides and thus consist of linear chains of DSP, units. Mur, muramic acid. concanavalin A-Sepharose 4B-purified N-acetylmuramoylhydrolase (muramidase, EC 3.2.1.17) of S. faecium (14), producing DSP, as virtually its only product (15). Here, using a variety of techniques, including selective radiolabeling of the reducing and nonreducing ends of the chains, we demonstrate that this enzyme is a processive, exodisaccharidase that commences hydrolysis from the nonreducing ends of the s-peptidoglycan chains. The kinetic and rate constants have been defined for the initial interaction of enzyme and s-peptidoglycan, for the hydrolysis of internal glycosidic bonds, and for the release of the enzyme from the limiting s-peptidoglycan.

MATERIALS AND METHODS
Experimental Procedures-The organisms used, the preparation and characterization of s-peptidoglycans, SF muramidase, gel filtration, paper chromatography, determination of radioactivity, and chemicals and reagents have been described (13-15). The s-peptidoglycans are soluble, linear, uncross-linked chains consisting of 44-47 8,l-l-linked DSP, units with MurNAc at the reducing terminus. These s-peptidoglycan chains are secreted into the medium by benzylpenicillin-inhibited cells (13). Unless otherwise indicated, the substrate for SF muramidase was s-peptidoglycan produced by Lyt-14 (an autolysis-deficient strain of S. faecium (16)).
Preparation of Standard Markers for Paper Chromatography-Preparative amounts of DSP, units (monomers) and the glycanlinked DSP, (dimer) were obtained from hydrolysis of Lyt-14 speptidoglycan prepared from cells exposed to 50 Fg/ml of benzylpenicillin in wall medium. Unlabeled s-peptidoglycan (1 mg) was hydrolyzed with HEWL (20 units/ml) in 0.5 ml of 0.5 M Na acetate, pH was hydrolyzed to produce labeled DSP, and DSP,. After desalting 6.0, for 30 min. In some experiments, radiolabeled s-peptidoglycan hydrolysates over Bio-Gel P-2 (Bio-Rad Laboratories, 38 X 1.25 cm), standards were separated by descending paper chromatography in Solvent I (butanokacetic acid:water, 4:1:5, v/v/v, upper phase) on Whatman 1 paper, for 72 h, and identified by comparison with known standards.
Labeling of s-Peptidoglycan with ["CIGal-Galactosyltransferase has been shown to be capable of transferring ["CIGal from UDP- (1 unit is defined as transferring 1.0 pmol of Gal from UDP-Gal to D-G~c per minute), 15 pmol of 3-(N-morpholino)propanesulfonic acid, 13 mM MnC12, in a final volume of 230 pl. After incubation for 2 h at 37 "C, the s-peptidoglycan was separated from unincorporated UDP-["CIGal by gel filtration over a Sephadex G-100 column. Fractions containing [14C]Gal + s-peptidoglycan were pooled, lyophilized, and dissofved in 2.0 ml of 0.02% NaN3, pH 7.2 (adjusted with 0.01 M NH40H). Galactosyltransferase was removed by affinity chromatography of the s-peptidoglycan on a vancomycin-Sepharose 4B column Preparation of s-Peptidoglycans of Various Molecular Sizess-Peptidoglycan was subject to partial enzymatic hydrolysis by HEWL (10 mg of Lyt-14 s-peptidoglycan, 0.1 mg of HEWL in 0.1 ml of 0.5 M Na acetate, pH 6.0, at 37 "C for 5 min). The reaction was stopped by heating (100 'C, 12 min) and the hydrolysate was fractionated on Sephadex G-100 into five fractions on the basis of the elution profile. Each fraction was rechromatographed on Ultrogel AcA 34, and each of the peaks, eluted with water, was pooled, lyophilized, and stored at -70 'C. The average glycan chain length of ( l 8 ) . 2 A. R. Zeiger, personal communication. each preparation was determined by the ratio of reducing groups before and after acid hydrolysis (19).

Hydrolysis of s-Peptidoglycan by SF Muramidase-s-Peptidogly-
cans of various molecular sizes (8-46 DSP, units) were incubated with SF muramidase (in various amounts from 10 to 1000 units) using the optimal conditions described (15). At intervals, samples were taken and monitored for reducing groups (19).
Establishing the Concentration of SF Muramidme-Purified SF muramidase, which was over 90% homogeneous as determined by sodium dodecyl sulfate-gel electrophoresis, was dialyzed against 2 mM K phosphate, pH 6.8. Equal volumes of dialysate and enzyme were dried to constant weight at I10 "C and weighed on a Cahn analytical balance. Samples of the same solution were used to establish standard curves of protein and enzyme activity. Experiments which depend on absolute enzyme concentration use these values for quantity of enzyme.  products of HEWL action on Micrococcus luteus s-peptidoglycan resolved in Solvent I (9, 15) include the disaccharide GlcNAc-MurNAc and the glycan disaccharide dimer GlcNAc-MurNAc-GlcNAc-MurNAc, DSP, and DSP2, in addition to larger oligosaccharides and oligosaccharide-peptides (DSP,).

Separation and Identification
Heterogeneity of products is due to the nature of the M . luteus s-peptidoglycan, which is only 40-50% substituted with peptides on MurNAc residues, presumably due in part to the action of the penicillin-insensitive amidase of M. luteus (9,20). Because all of the MurNAc residues of s-peptidoglycan of S. faecium are substituted with peptides, HEWL hydrolysis produced only two detectable low-molecular-weight products, DSP, and DSP,, plus longer oligosaccharide-peptides near the origin (15). In contrast to HEWL action, SF muramidase hydrolysis of Lyt-14 s-peptidoglycan resulted in the production of a large amount of DSP, and very little DSP, (15). Radiolabeled s-peptidoglycan was prepared from the two autolysis-defective strains of S. fwciurn   (13)) to contain "C (from ['4C]Glc) in the glycan portion and 3H (from [3H]Lys) in the peptide portion. Complete hydrolysis of these polymers by the SF muramidase or partial hydrolysis by HEWL resulted in products that contained both 3H and 14C and had the same paper chromatographic mobilities as DSP, and DSP, (15). No evidence was obtained for the production of disaccharides or tetrasaccharides, indicating that, within the limits of detection, all MurNAc residues of the Lyt-14 and Aut-3 s-peptidoglycan were substituted with peptides.

Kinetics of Hydrolysis of s-Peptidoglycan Chains of Different Chain Lengths by the SF Murumidase-Lineweaver-Burk
plots of hydrolysis of s-peptidoglycan chains of 13 & 1, 22 k 2, 28 f 2, and 46 f 2 DSPl units by the SF muramidase showed decreased maximum velocities and decreased apparent K , values with decreasing chain length when substrate concentration was expressed in terms of DSP, units (Fig. 2, inset). However, plots of the same data in which substrate concentration is expressed in terms of s-peptidoglycan chains (irrespective of chain length), resulted in similar K, values (average K, = 0.175 k 0.03 PM; Fig. 2). In contrast, V,, was strongly dependent on the degree of polymerization size of the substrate, with turnover numbers ranging from 7 to 19 mol of DSPl released min-' mol-' of enzyme (Fig.  2). To determine the true K,, the substrate concentration was expressed in terms of free substrate concentration, while taking into account any depletion of free substrate by the enzymesubstrate complex (21). Lineweaver-Burk plots of the hydrolyses of s-peptidoglycans by the SF muramidase were used to generate apparent K , values. By using data generated from the hydrolyses of s-peptidoglycans of various lengths with various concentrations of enzyme, the relationship of K , to enzyme concentration and to chain length can be established.  molecular-weight products during hydrolysis by SF muramidase (Fig. 2) clearly distinguishes the SF muramidase from HEWL, which has been characterized as an endoglycosidase (22)(23)(24)(25). To determine whether the SF muramidase is processive (catalyzing multiple hydrolytic events before enzyme release), several experiments were conducted, including (i) competition experiments in which substrate was added to enzyme which was prebound to substrate under conditions where the catalytic rate is insignificant (0 "C); and (ii) determination of the initial hydrolytic site of the SF muramidase on the s-peptidoglycan by the use of end-labeled polymers.
In the experiment shown in Fig. 4 (bottom panel), latent (zymogen form) SF muramidase was prebound to excess [3H] s-peptidoglycan. Following proteinase activation of the latent enzyme, an equal amount of ['4C]s-peptidoglycan was added, followed by hydrolysis at a permissive temperature. Samples were taken at intervals, the reactions were stopped (100 "C, 7 min), and samples were analyzed by paper chromatography in Solvent I. Preincubation with [3H]s-peptidoglycan resulted in the production of only [3H]DSPl units for the first 2 min of incubation (Fig. 4, bottom). In contrast, release of ["C] DSP, was observed only after a delay of approximately 2 min, after which release of [3H]DSPl and ["C]DSP1 paralleled each other. This initial release of product from prebound substrate clearly demonstrates preferential hydrolysis of prebound substrates and indicates that free DSP-, s-peptidoglycan does not readily exchange with existing enzyme s-peptidoglycan complexes.
As a control for these competition experiments, a known endoglycosidase, HEWL, was tested in a similar experiment using the same substrates. The experimental design was identical to that for the SF muramidase, but with the modification of assaying for both DSP, and DSPz products, since the random pattern of hydrolysis by HEWL results primarily in these products (9,(21)(22)(23)(24). In contrast to the SF muramidase, HEWL hydrolysis did not show a preferential release of 3Hlabeled products from the substrate to which enzyme was prebound (Fig. 4, upper p a n e l ) and instead resulted in the parallel release of both 3H-and "C-labeled products throughout the incubation.
Several variations on the type of the experiments described in Fig. 4 were performed. For example: (i) latent SF muramidase was first activated with bovine plasma albumin (containing activating proteinase) before prebinding to [ 3 H ]~peptidoglycan at 0 "C; (ii) latent SF muramidase was prebound to ['4C]s-peptidoglycan at 0 "C before activation and exposure to [3H]s-peptidoglycan. The results of both of these experiments were quantitatively indistinguishable from those shown in the lower panel of Fig. 4. In summary, all experiments demonstrated that the SF muramidase preferentially hydrolyzed the glycan chain of speptidoglycan to which it was prebound. Furthermore, the SF muramidase maintained its initial enzyme-substrate complex for approximately 2 min (Fig. 4). To examine further the sequential nature of hydrolysis, preactivated SF muramidase was prebound to excess ["CIS-peptidoglycan at 0 "C, 1-to 30fold excess unlabeled s-peptidoglycan was added, and the reaction mixtures were rapidly brought to 37 "C. Release of ["CIDSP, units occurred at equal rates independent of the presence of varying amounts of competing, unlabeled substrate. The independent rate of [14C]DSPI formation continued for about 2 min. After 2 min, the rate of release of ["C] DSP, units depended on the concentration of unlabeled speptidoglycan added. As the concentration of the unlabeled 7t 6- GIu-speptidoglycan were incubated in 1.0 ml of 10 m M Na phosphate, pH 6.8, at 0 "C for 5 min. Bovine plasma albumin (75 pg/ml) was then added and, after 3 min at 0 "C (to allow proteinase activation of the zymogen form), [14C]Glus-peptidoglycan was added and the temperature was rapidly brought to 37 "C. The final concentrations of reactants in 1.2 ml of 10 mM Na phosphate, pH 6.8, were: 2.8 nM SF muramidase, 174 nM ['HIS-peptidoglycan chains (DSP,) and 174 nM ["CIS-peptidoglycan chains. In the experiment shown in the upper panel, HEWL (5 units/ ml) was added instead of the SF muramidase. A second control experiment consisted of SF muramidase being prebound to both [3H]s-peptidoglycan and ['4C]s-peptidoglycan (bottom, solid line). At timed intervals, during incubation at 37 "C, samples were removed, the reaction stopped (100 "C for 7 min), and the products of hydrolysis identified by paper chromatography as described in Fig. 1 competing s-peptidoglycan increased, the time at which the change in rate of [14C]DSP1 release occurred remained the same, but the rate of release of ["CIDSP, decreased as the concentration of unlabeled competing s-peptidoglycan was increased (Fig. 5). The constant period (2 min) of irreversible binding of the SF muramidase to prebound substrate (Figs. 4  and 5 ) is consistent with the observed turnover number of 21 mol of DSP, cleaved/min/mol of enzyme, as calculated from kinetic analyses of the DSP46 polymer under these assay conditions (Fig. 2).

.
Site of Binding and Direction of Hydrolysis-The results described above suggested a single binding site for the enzyme nonreducing end of a glycan chain. Therefore, s-peptidoglycan was radiolabeled with 3Hat the reducing terminus by reduction of the terminal MurNAc residue with ['HINaBH,. Similarly, s-peptidoglycan was radiolabeled by the addition of [14C]Gal to the nonreducing terminus via the action of galactosyltransferase (17). The hydrolysis of doubly labeled speptidoglycan by the SF muramidase in a molar ratio of enzyme:substrate of approximately 5 resulted in the rapid and preferential production of [I4C]Gal + DSP, (and DSP2) (Fig.  6). The initial burst of 14C + DSPl and DSPz indicated that the SF muramidase preferentially hydrolyzes bonds in the speptidoglycan at the nonreducing terminus. A delayed burst on an s-peptidoglycan chain, most likely at the reducing or of 3H-labeled products was observed, with a midpoint at 2 min after the reaction was initiated. The difference in release of [14C]Gal-labeled, and 3H-labeled product is consistent with the competition studies that indicated a period of approximately 2 min when SF muramidase, prebound to s-peptidoglycan, could not be displaced (Figs. 4 and 5). The release of 3H-labeled products a t approximately 2 min is also consistent with the apparent turnover number (-2l/min) obtained from the kinetic data (Fig. 2). This result indicates that there is no significant change in kinetic properties of the enzyme as a result of introducing ["CIGal at the nonreducing end or 3H at the reducing end of s-peptidoglycan. The initial products released from [14C]Gal + s-peptidoglycan included a substantial amount (up to 35% of the total 14C released) of [14C]Gal + DSP, (inset, Fig. 6). Production of the dimer could be due to the presence of galactose at the nonreducing terminus interfering, to some degree, with the first catalytic cleavage in some of the enzyme-substrate complexes, as all other data (15) suggest that 95% of total products from the natural substrate are DSPl units. Similarly, the presence of muramicitol formed by reduction of the reducing end of the polymer (inset, Fig. 6) could have been responsible for the production of DSP, units from the terminal of the s-peptidoglycan, which is normally the reducing end. Time Course of Product Release-The detailed time course of s-peptidoglycan hydrolysis by SF muramidase was established by using approximately a 1:6 molar ratio of enzyme-tosubstrate polymer and following the extent of DSP, release at frequent time intervals. The reaction was slowed by following s-peptidoglycan hydrolysis at 30 "C rather than 37 "C. A rapid burst of product was observed in the first minute, followed by a 2lh-min lag before the second mole of s-peptidoglycan chain was hydrolyzed (Fig. 7). The second mole of s-peptidoglycan polymer was hydrolyzed in approximately 1 min, followed by a second lag period. The rate of hydrolysis between lag periods was approximately 60 catalytic events/ enzyme molecule/min, consistent with the predicted turnover number of 91 catalytic events/min at 37 "C for a continuous polymer of DSP units (see Fig. 3B). This result indicates that the observed dependence of V,,, on s-peptidoglycan chain length (Figs. 2 and 3B) is due to the slow rate of enzyme release from the limit digest fragment and that the enzyme cleaves the DSP units at a uniform rate until approaching the limit digest fragment.
The results of Fig. 7 also provided the stoichiometry of enzyme catalytic sites. Following the first burst of product formation (after 1 min, Fig. 7), 0.12 nmol of s-peptidoglycan (DSP,) was released by 0.14 nmol of enzyme. The enzyme thus catalyzed the hydrolysis of 0.86 mol of s-peptidoglycan/ catalytic cycle/polypeptide chain.

DISCUSSION
Several muramidases, such as the well-studied HEWL, are known to randomly hydrolyze &1,4 linkages between MurNAc and GlcNAc (22)(23)(24) and, in the case of several lysozymes, to carry out transglycosylations (24). The data provided in this and previous studies (13, 15) demonstrate that the purified endogenous SF muramidase is a specific exodisaccharidase. Comparisons of the action of the SF muramidase with that of HEWL on s-peptidoglycans showed that, in contrast to HEWL, SF muramidase hydrolysis resulted in only one major product, DSPI, even after short incubations ( Fig. 4;  peptidoglycan (0.4 nmol of chains), labeled with 6 and 51% efficiency by ["CIGal and 'H, respectively (as described under "Materials and Methods"), were mixed with 833 units of activated SF muramidase (0.56 nmol) and 10 mM Na phosphate, pH 6.8, in a total volume of 600 pl, and incubated at 0 "C. This represents an enzyme:substrate ratio of 1.4:l. The reaction mixture was quickly brought to 37 "C, and samples of 100 pl were taken at timed intervals. Following heat inactivation at 100 "C for 5 min, the products were separated in Solvent I and quantified by detection of radioactivity. Total ["CIGal + products (m) precede release of 'H-labeled products ( Latent SF muramidase (0.14 nmol) was preincubated at 0 "C for 5 min with 0.8 nmol of ["C]glucose-DSPle and converted from the zymogen by 5 min of incubation at 0 "C with 0.4 pg/ml of trypsin. The total reaction mixture (0.5 ml) was rapidly heated to 30 "C, and, at the indicated intervals, samples (20 pl) were taken and boiled to terminate the reaction. The amount of DSP, formed was established by chromatography and liquid scintillation counting as described under "Materials and Methods" and in the legend tQ Fig. l. substrate; however, the apparent K,,, remained constant when substrate concentration was expressed in terms of s-peptidoglycan chains (Fig. 2). These observations clearly distinguish the actions of SF muramidase and HEWL. For example, HEWL hydrolysis of oligosaccharides greater than six saccharides (which is the minimum for maximum productive binding by HEWL (26)) shows little difference in kinetic constants as a function of polymer size (27). This relatively constant value of K, in terms of substrate molarity but not in DSPl units (Fig. 2) suggested a unique binding site on speptidoglycan chains. This site appears to be the nonreducing end of the polymer. There are several possible explanations for the observed difference in apparent V,, with different polymer sizes. These include: (i) that internal chain cleavages are slower on shorter chains; (ii) that there is a decreased cleavage rate as the terminal residue is approached; or (iii) that the initial cleavage is slow compared to the others.
T o distinguish between initial binding at an internal residue or at a terminus, a series of competition experiments was carried out, in which the SF muramidase was prebound to labeled s-peptidoglycan under conditions where no significant catalysis occurs. Once bound to a glycan chain, the SF muramidase sequentially hydrolyzed the entire glycan chain to DSPl before becoming available to combine with another glycan chain (Figs. 4,5, and 7). Both the zymogen and active form of the SF muramidase bind tightly to the s-peptidoglycan, similar to the binding of SF muramidase to intact cell walls (28,29). The time (-2 min) required to completely hydrolyze a glycan chain in competition experiments is consistent with the kinetically derived turnover number of the enzyme, suggesting that initial binding is at a terminus, with sequential hydrolysis of nearly the entire s-peptidoglycan chain occurring before dissociation of the enzyme-substrate complex. SF muramidase hydrolysis of peptidoglycan chains labeled extrinsically at the reducing terminus with 3H (from [3H]NaBH4), and at the nonreducing terminus by the addition of ["CIGal, showed preferential release of 14C-labeled products from the nonreducing terminus, and a 2-min delay in release of 3H-labeled products from the reducing terminus (Fig. 6). In summary, the kinetic and labeling experiments indicated that the SF muramidase binds to the nonreducing terminus of the s-peptidoglycan followed by the sequential hydrolysis of sensitive bonds and the release of DSP1. Enzyme is not released from the complex until hydrolysis is complete. Thus, it can be concluded tht SF muramidase is a processive exodisaccharidase. The observation that the DSP2 moiety is not a good substrate for the SF muramidase (having a 10to 100-fold lower reactivity than the D S P~l e n g t h s-peptidoglycan substrate) and that DSPz is present in significant amounts following extensive hydrolysis of s-peptidoglycan indicates that the final DSP2 unit is not usually cleaved. The long lag period which follows complete hydrolysis of a s-peptidoglycan chain indicates that hydrolysis is rapid but release of the final DSP, unit is slow, becoming a rate-limiting factor in the overall steady-state pattern when enzyme is present at a low molar ratio with respect to DSP,. A provisional model for this action is shown in Fig. 8.
The calculation of rate constants for this mechanism and the determination of binding site stoichiometries requires the concentration of active (catalytic) sites in SF muramidase preparations. Active site titration with ['4C]Gal-labeled substrate and analysis of the initial burst of product at saturating substrate concentration (Fig. 7) provides the titration of catalytic sites per unit of enzyme activity. The molar ratio of product released in the initial burst to the number of active sites is approximately one catalytic site per molecule.
The inability of the SF muramidase to show transglycosylase activity (15) and the absence of product inhibition by the DSP, simplify the construction of a model to be provisionally assigned for the mechanism of action of the SF muramidase. Similar cases have been presented by Bailey and French (30).
A kinetic representation of the mechanism of action of the SF muramidase can be diagrammed, as shown in Scheme 1, where DSP, is the only form of the substrate to complex with E. The consequence of k,, kg, and k? being negligible compared to kl and k3 (no dissociation of the E-DSP, complex) results in the progression of the SF muramidase along the s-peptidoglycan chain, releasing DSPl (P). The last fragment to be released is usually DSP2, since approximately 5% of the total radioactivity of labeled DSP,, is found as DSP, following hydrolysis by SF muramidase (15). However, when the reducing end is converted to the muramicitol by borohydride reduction, a substantial fraction is released as the DSP, muramicitol (Fig. 6) (Fig. 2). Extrapolation of the rate to a continuous polymer gives a turnover number of 91 events min", and this value is confirmed experimentally by the burst kinetics shown in Fig. 7. This value provides a reasonable estimate of k3. The kat/K,,, value is 9 X lo6 M" s" using the K,,, of 0.17 PM (Fig. 3). This value is near the diffusioncontrolled limit of -5 X lo' M" s" for the second-order rate usually obtained for the second-order rate constant for interaction between small molecules and enzymes (32). The subsequent cleavages are not subject to diffusion control, since enzyme and substrate are in close proximity, and the release of enzyme from the substrate cannot be experimentally demonstrated ( e g . Figs. 4 and 6). The comparable constant has been reported to be 1.7 X lo4 M-' s-l for HEWL using (GlcNAc-MurNAc)s as the ligand (22). These calculations indicate that the muramidase is approximately 500 times more efficient in forming product from long peptidoglycans than is HEWL. The evolution of the catalytic potential has maximized the turnover number while retaining sufficiently tight binding to prevent dissociation of the enzyme from its normal substrate of cell wall polymer. The K , value for SF muramidase is constant as a function of chain length and can be used to calculate the dissociation constant for the enzyme-substrate complex during its action on an extended polymer. The K , is described by the constants (k, + k3)/kl and the dissociation constant, Kd = k2/k1. Kd can thus be estimated to be 1.5 X lo-' M for the reaction E + DSP, E.DSP, where n is large. The first-order rate constant for dissociation, k,, is approximately 0.013 s-', consistent with the results of Figs. 4 and 5 which demonstrate that the polymer does not readily dissociate once the Michaelis complex is formed. These constants are the limiting values for an s-peptidoglycan substrate which is continuous, DSP, where n is large. The value of 91 min-' for k3 in large DSP polymers may also be correct for hydrolysis of smaller speptidoglycans, as indicated by the linear plot of Fig. 3B. In this mechanism, hydrolysis of DSP, from the polymer always occurs at 91 min" and release of the final DSP, is slow and constant for any starting s-peptidoglycan. Release of DSP, is shown by a distinct rate constant, in the reaction mechanism (Scheme I), since hydrolysis of every peptidoglycan chain ends with the same E .DSP, complex. The value of k, can be estimated from Equation 1. & = (n -2)/V,.
-(n -2)/k3 (1) where n is the length of the DSP, s-peptidoglycan and V,,, is the maximum rate at saturating concentrations of s-peptidoglycans of various n values (Fig. 2). Equation 1 can be rearranged to predict that a double-reciprocal plot of apparent VmaX as a function of chain length should give a linear plot as shown in Equation 2. The slope of this line yields $, the average time for dissociation of the E.DSP2 complex. Analysis of the data from Fig.  3B, according to Equation 2, gives a value for k, of 1.5 min; the average residence time of DSPz on SF muramidase under normal assay conditions (37 "C). The observed lag period of 2.5 min at 30 "C ( Fig. 7) is consistent with the calculated lag period of 1.5 min at 37 "C from the results of Figs. 2 and 3. A mcchanism where the rate of each hydrolytic event is uniformly dependent on chain length can be eliminated by the results of Fig. 7 which clearly demonstrates the lag period for release of DSP,. The possibility that enzyme attaches to the nonreducing end (or elsewhere) and requires 1.5 min to begin hydrolysis at the nonreducing end is eliminated by results of Figs. 4-6, which demonstrate no lag in the initial phases of hydrolysis.
Although the SF muramidase has been known for over 15 years (33), its reaction kinetics were not previously characterized. The present studies establish that the enzyme recognizes the nonreducing end of peptidoglycan chains and does not initiate cleavage at internal glycan residues. The enzymesubstrate complex has a large commitment to catalysis, with the substrate release rate being nearly insignificant compared to the initiation of hydrolysis. These factors cause the enzyme to act continuously on a single s-peptidoglycan chain. Enzyme release does not occur until all glycan bonds have been cleaved, with the exception of the final DSP, unit. The results suggest relatively rapid hydrolysis of internal glycan residues followed by a pause as the last DSP units are released or hydrolyzed and released.
In U~U Q , this enzyme attaches tightly to the cell wall (28).
The kinetic properties of high binding affinity (-lom9 M for the dissociation constant) and lack of release during catalysis are well suited to the enzyme location outside of the cellpermeability barrier and to its proposed role as an exoskeletalremodeling enzyme (1). Although the turnover rate of the enzyme is relatively slow, any increase in rate would necessitate a decreased affinity, since the V,.,/K,,, value is near diffusion limits. A lowered affinity would be unsuitable for a cell wall-remodeling enzyme, since it would risk loss of the enzyme from the cell wall to the medium.