FEMS Microbiology Letters 100 (1992) 261-268 © 1992 Federation of European Microbiological Societies 0378-1097/92/$05.00 Published by Elsevier
261
FEMSLE 80034
The autolytic ('suicidase') system of Enterococcus hirae: From lysine depletion autolysis to biochemical and molecular studies of the two muramidases of Enterococcus hirae ATCC 9790 G e r a l d D. S h o c k m a n Department of Microbiology and Immunology, Temple University School of Medicine, Philadelphia, Pennsylvania, USA Received 18 May 1992 Accepted 1 July 1992
Key words: Autolysin; Muramidase; Enterococcus hirae; Penicillin-binding protein
1. S U M M A R Y Autolysis of Enterococcus hirae ATCC 9790 is the result of the action of endogenous enzymes that hydrolyze bonds in the protective and shape-maintaining cell wall peptidoglycan. It is thought that these potentially suicidal enzymes play a positive role(s) in wall growth and division and are expressed as autolysins when cell wall assembly a n d / o r repair are inhibited. E. hirae possesses two potentially autolytic enzymes, both of which are muramidases. Although they hydrolyze the same bond as hen egg-white lysozyme, both are high-molecular-mass, complex enzymes. Muramidase-1 is synthesized as a zymogen, requiring protease activation. It is a glucoenzyme that is also multiply nucleotidylated with an unusual nucleotide, 5-mercaptouridine monophosphate. Muramidase-2 is almost certainly a product of a separate gene. The deduced amino acid
sequence of a cloned gene for extracellular muramidase-2 showed several unusual features. It appears to be a two-, or perhaps three-domain protein with a putative glycosidase-active site near the N-terminal end and six 45-amino-acid-long repeats at the C-terminal end which are presumed to be involved with high-affinity binding to the insoluble peptidoglycan substrate. Muramidase-2 binds penicillin with low affinity. The presence of several amino acid groupings characteristic of serine-active site /3-1actam-interactive proteins is consistent with the possible presence of a penicillin-binding, third domain. Indirect evidence consistent with a role(s) for these enzymes in cell wall growth and division has been obtained. However, proof of such role(s) awaits modern genetic, molecular, and biochemical analyses.
2. I N T R O D U C T I O N Correspondence to: G.D. Shockman, Department of Microbiology and Immunology, Temple University School of Medicine, Philadelphia, Pennsylvania, USA.
In the late 1940s and early 1950s, well before the advent of amino acid analyzers, the nutritional requirements of fastidious bacteria, primar-
262
ily lactobacilli and streptococci, were used to quantify the amino acid (and vitamin) content of various foods and other materials. Attempting to increase the reliability and precision of microbiological amino acid assays, Dr. G. Toennies and collaborators carefully examined many of the variables of turbidimetric methods rather than the then widely used acidimetric assays [1-3]. For various reasons he selected as the test organism what was then known as Streptococcus faecalis ATCC 9790, which later became Streptococcus faecium, Enterococcus faecium and, most recently, Enterococcus hirae. As an organic chemist, his choice of a salt-tolerant Enterococcus was fortuitous, as it permitted an increase in the buffer concentration of the chemically defined medium used [4] to 0.3 M sodium phosphate, thus greatly reducing the growth (and later, autolysis) inhibitory effects of the vast amounts of lactic acid metabolically generated by the fermentation of glucose during growth of this organism. Much later, two observations were found to be directly related to cell wall metabolism of E.
hirae. The first observation was that although exhaustion of growth-limiting quantities of most of the amino acids required by this organism (e.g. valine, threonine, leucine, isoleucine, histidine, or methionine) signalled the end of the exponentialgrowth phase, in the absence of further protein synthesis, culture turbidity continued to increase, postexponentially, at a continuously decreasing rate [5,6]. Later, this phenomenon was shown to be due to continued cell wall accumulation [6-9], resulting in bacteria that possessed a larger amount of cell wall in the form of a thickened wall structure rather than an increase in wall surface area [10,11]. The exception was lysine depletion. In this case, rather than an increase in culture turbidity, lysine exhaustion was followed by a reasonably rapid, pronounced decrease in culture turbidity [6,12]. The use of a highly buffered culture medium and maintenance of a nearly neutral pH was essential for the observation of cellular autolysis. The reason for this exception only became obvious several years later as an indirect result of bacterial cell wall chemistry studies from several
laboratories throughout the world (see ref. 13 for an excellent detailed summary, or ref. 14 for a more recent, less detailed review). Lysine was the only nutritionally required amino acid examined that was a component of the cell wall peptidoglycan of E. hirae, in addition to being a protein amino acid! Both phenomena, wall thickening and cellular autolysis, are examples of unbalanced growth [15]. Thus, deprivation of an amino acid that was a component of both protein and bacterial wall peptidoglycan resulted in autolysis. This led to the proposal that the attainment of osmotic fragility and bacterial autolysis induced by treatments with antibiotics that specifically inhibited further wall peptidoglycan biosynthesis, was not dependent on the continued synthesis of protein and other cytoplasmic macromolecules. Therefore, bacteria did not become osmotically fragile by simply 'growing out' of the protection of their walls, but this process required the active participation of an enzyme activity(ies) that hydrolyzed preexisting bonds in the peptidoglycan that were essential for maintaining the integrity of, and protective functions of, the wall [6,11]. Thus, the statement in many texts that /3-1actam and other cell wall antibiotics are lethal only on growing bacteria is not quite true! More precisely, the requirement is that the drug must interrupt ongoing assembly of new wall peptidoglycan into an expanding surface.
3. AUTOLYSIS OF E. HIRAE Subsequent observations showed that exponentially growing and dividing cells of E. hirae were prone to autolysis which could be observed not only by the depletion of lysine but by placing bacteria from an exponentially growing culture into a culture medium lacking one or more nutritionally required precursors of peptidoglycan, including glutamic acid and glucose [10,16,17], or more simply, into a buffer of appropriate pH and concentration. These data led us to ask why growing and dividing bacteria should carry around with themselves such potentially suicidal enzyme activities. This question led us to the hypothesis
263 that autolytic peptidoglycan hydrolases play some role in peptidoglycan assembly, bacterial wall growth, and cell division [11], an idea that was also proposed by others [18,19] based on data obtained with other bacterial species. It was also rather obvious that such 'suicidases' must be extremely well regulated and integrated with wall biosynthesis, so that hydrolysis of bonds in the peptidoglycan exoskeleton was rapidly followed by biosynthetic processes that, among other things, repaired the potential damage. The possession of one, or usually more, autolytic peptidoglycan hydrolase activities is not unique to E. hirae. A variety of Gram-positive and Gram-negative species, including Escherichia coli [20] have been shown to have such activities. Thus a variety of subsequent data from this and other laboratories (see refs. 21-23 for reviews) are consistent with, support, and extend the hypothesis that peptidoglycan hydrolases, which, at least in some cases, are potential suicidases, are related to, and perhaps essential for, the processes that result in expansion of the protective wall surface to enclose an increasing volume of cytoplasmic contents. In addition, because the wall peptidoglycan is the cellular component primarily responsible for maintaining the shape of the bacterial cells, which subtly changes during the cell division cycle, the broad general hypothesis has been modified to include possible roles for these enzymes in hydrolysis of a small number of bonds at appropriate topological sites. Those strategic cuts would engineer modifications of the previously assembled structure, resulting in small changes in shape (morphogenesis) of the exoskeleton. The potential suicidal nature of these activities, plus the postulated action at precise times and topological locations, implies that these enzyme activities must be precisely regulated and coupled with cell wall biosynthetic, or at least repair, processes. For such positive roles only a relatively small number of well regulated enzyme molecules per bacterial cell are required. We know of many bacterial species and strains that fail to exhibit a gross autolytic behavior when exposed to conditions that result in rapid lysis of closely related or similar bacterial species. Examples of such a con-
trast are many Group A streptococci and some viridans streptococci such as Streptococcus mutans, which are resistant to autolysis when exposed to/3-1actam or other cell wall antibiotics in contrast to the rapid autolysis of E. hirae ATCC 9790. The recent demonstration that a Group A streptococcus resistant to autolysis possessed a low but detectable level of peptidoglycan hydrolase activity [24] is consistent with the hypothesis that small amounts of such activities may be essential and play a positive role in surface growth. Certainly the existence of autolytic defective mutants of a variety of both Gram-positive and Gram-negative bacterial species suggests that only relatively low levels of peptidoglycan hydrolase activities may be essential. In support of this hypothesis of essentiality is the complete lack of success in isolating truly autolytically negative, otherwise isogenic strains of any bacterial species. The presence in many species of more than one and, in some instances, multiple peptidoglycan hydrolases, is also consistent with one or more essential functions of such enzymes.
4. T H E P E P T I D O G L Y C A N OF E. H I R A E ATCC 9790
HYDROLASES
In contrast to a variety of both Gram-positive and Gram-negative bacteria which contain peptidoglycan hydrolases of two or more specificities, a broad variety of data are completely consistent with the presence of only one specificity of peptidoglycan hydrolase in E. hirae (summarized in ref. 22). The only detectable bond cleaved is the fl, 1-4 bond between N-acetylmuramic acid and N-acetylglucosamine by the action of a lysozymelike, N-acetylmuramoylhydrolase (muramidase). No evidence of the presence of amidase, endopeptidase (except for a fl-lactam sensitive DDcarboxypeptidase [25,26], whose action does not affect the two- or three-dimensional integrity of the peptidoglycan structure) or glucosaminidase activities was obtained. The presence of only muramidase activity is consistent with the belief that this organism is a relatively simple model of wall surface growth and degradation, since (i) it can only divide in one plane, (ii) it is comparatively
264 simple in shape, consisting only of poles and lacking cylindrical wall, (iii) it enlarges its wall surface area at only a limited number of nascent septal sites [27-29], and (iv) it conserves assembled wall for a number of generations [30,31] and fails to shed cell wall by the process known as turnover [32]. However, the observation of Kawamura and Shockman [33] that E. hirae possessed two separate and biochemically distinct peptidoglycan hydrolases as well as a growing assortment of data which show that, in contrast to hen eggwhite lysozyme, both muramidases are highmolecular-mass complex proteins indicate that this model is far from simple. 4.1. Some properties of muramidase-1 Muramidase-1 is present as a high-molecularmass (130-kDa) zymogen that can be proteolytically cleaved by any of a broad variety of proteases to an 87-kDa active form [34]. It is a glycoprotein that is multiply substituted with oligosaccharide containing only glucose units [34], thus, it is a glucoenzyme. In addition to possessing covalently linked glucose, the polypeptide also possesses about 12 phosphodiester-linked monomeric nucleotides [35]. Direct probe mass spectroscopy was used to identify this nucleotide as 5-mercaptouridine monophosphate [35]. Phosphodiesterase treatment of native muramidase-1 resulted in the removal of about half of the nucleotides and phosphorus and was accompanied by about a 50% reduction in enzyme activity (Dolinger and Shockman, unpublished data). These last data suggest that nucleotidylation plays a role in enzyme activity, although the exact role remains to be elucidated. Muramidase-1 has an unusual mechanism of hydrolysis. Studies of hydrolysis of soluble linear peptidoglycan chains produced and secreted by penicillin-inhibited cells of autolytic defective strains of E. hirae [36] showed that the enzyme binds to the non-reducing ends of these 42- to 47-disaccharide-peptide-long chains and sequentially (processively) hydrolyzes the bonds between N-acetylmuramic acid and N-acetylglucosamine along the chains yielding virtually only one product - - monomeric disaccharide-peptide units [37,38].
4.2. Some properties of muramidase-2 The presence of a second peptidoglycan hydrolase in extracts of E. hirae was observed as an activity that both rapidly dissolved walls of Micrococcus luteus and failed to bind to concanavalin A (ConA)-Sepharose [33]. This activity was not observed earlier, as it has little ability to dissolve its homologous substrate (wails of E. hirae), although it appears to hydrolyze some glycosidic bonds in the latter substrate. Further studies showed that the activity that failed to bind to ConA-Sepharose and probably is not a glycoprotein, is a muramidase [39] (muramidase-2) that substantially differs in substrate specificity [33,40] from muramidase-1. Although it is remotely possible that muramidase-2 is a non-glucosylated, a n d / o r non-nucleotidylated, form of muramidase-1, substantial evidence exists that each is a product of separate genes [41]. Muramidase-2 also appears to be a complex protein. After sodium dodecylsulfatepolyacrylamide gel electrophoresis (SDS-PAGE), two polypeptide bands with apparent molecular masses of about 125 and 70 kDa were shown to possess peptidoglycan hydrolase activity and to have the ability to bind radioactively labelled pelficillin with low affinity [39]. These two polypeptides appeared to have the same electrophoretic mobility on SDS-PAGE as penicillin binding proteins (PBPs) 1 and 5 present in membrane preparations of E. hirae. However, recently obtained immunochemical and molecular data clearly demonstrate that PBP 5 and the 70-kDa form of muramidase-2 are separate and distinct proteins [41]. Recently [42] we successfully cloned in E. coli and sequenced a gene for the extracellular form of muramidase-2. The deduced amino acid sequence of the 666-amino-acid-long open reading frame showed several interesting features: (i) a 49-amino-acid-long putative signal sequence with a potential protease cleavage site between Ala 49 and Asp 50. This cleavage site was confirmed by the virtual identity of the determined N-terminal amino acid sequence of mature extracellular muramidase-2 with the deduced amino acid sequence; and (ii) the presence of six 45-aminoacid-long highly homologous repeat units, sepa-
265 rated by intervening sequences highly enriched for serine and threonine (approx. 51% serine plus threonine), was found at the C-terminal end of the polypeptide. Although these repeats have little homology with repeat units found at the Cterminal ends of a variety of different bacterial proteins, including those in the pneumococcal amidase gene [43] and the muramidase genes of several pneumococcal phages [44], substantive homology was observed with four contiguous Cterminal 68-amino-acid-long repeats of an autolysin of unspecified enzymatic specificity of S. faecalis [45]. The amino acid sequence of the S. faecalis autolysin showed other areas of substantive homology with muramidase-2, including the putative signal peptide and much of the remaining polypeptide, except for a 104-amino-acid-long insertion between amino acids 64 and 65 of the E. hirae muramidase-2, and for the serine- plus threonine-enriched sequences [46]. Analogous to the interpretations of Garcla and co-workers [43,44] that the highly conserved pneumococcal repeats may be involved in the binding of these enzymes to wall of pneumococci possessing choline-containing wall teichoic acid, it seems likely that the six 45-amino-acid-long repeats of muramidase-2 are somehow involved with the high-affinity binding [47] of this enzyme to the wall peptidoglycan of E. hirae. Further analysis of the derived amino acid sequence of muramidase-2 showed the presence of an SXXK-active site motif and of several amino acid groupings such as SGN and KSG, that are characteristic motifs of serine /3-1actamases and PBPs [48]. Although the spacings of the motifs in muramidase-2 are not inconsistent with those found in other penicillin-interactive proteins, the shape and distribution of hydrophobic and hydrophilic clusters along the amino acid sequences of both muramidase-2 and the S. faecalis autolysin differ from those found in a broad assortment of serine fi-lactamases and in a number of PBPs [46]. Thus it is possible that the mechanism of penicillin interaction may differ from that of the typical serine-active site penicillin-interactive protein [49].
4.3. Possible functions and regulation of the two potentially suicidal muramidases of E. hirae Such potentially suicidal activities that are present in growing and dividing bacteria must be exquisitely well regulated and integrated with wall and other biosynthetic processes. It is also clear that endogenous peptidoglycan hydrolases are responsible for the lysis of cells of some species and strains of bacteria after antibiotic or nutritional inhibition of peptidoglycan assembly. However, a number of observations are consistent with, but do not prove, a well regulated role for such activities in wall growth and division. For the E. hirae system we have established that: (i) wall autolytic activity is associated with newly assembled wall, at nascent septa [50] in growing and dividing bacteria; (ii) inhibition of protein synthesis either by the use of antibiotics or nutritionally results in a rapid (and reversible) decrease in autolytic capacity although enzyme activity(ies) remains present [51]; (iii) rates of cellular autolysis correlate with growth rates of cultures [52]; (iv) bacterial cells from synchronized cultures show variations in rates of autolysis that can be correlated with the division cycle [52,53]; (v) membrane-associated ligands such as acylated lipoteichoic acid and certain lipids, notably cardiolipin, have been shown to inhibit autolysis of cells and isolated walls [54, 55]; and (vi) additional potential levels of regulation include substrate specificity, topological location of enzyme molecules a n d / o r susceptible bonds, the processive hydrolytic mechanism of muramidase-1, proteinase activation of the latent form of muramidase-1, and the state of protein folding of either of the enzymes. The presence of two large and complex muramidases in a bacterial cell of comparatively simple shape and mode of division implies that hydrolytic function may be essential to the bacterium. In 'normal' growth situations, whatever they may be in natural environments, the two activities may have separate and distinct, and perhaps complementary, functions in wall growth and division. However, deletion of one activity may not be lethal. The other activity might be
266
able to fully or partially take over that function. Certainly the presence of two separate and distinct gene products helps to explain our difficulties in obtaining truly autolysis-negative mutants! We hope that the power of modern genetic, molecular and biochemical methods will help us to solve the questions of functions and essentiality of these peptidoglycan hydrolases in the near future.
ACKNOWLEDGEMENT The work in the author's laboratory was supported by U.S. Public Health Service grant AI05044 from the National Institutes of Health.
REFERENCES [1] Toennies, G. and Gallant, D.L. (1948) J. Biol. Chem. 174, 451-463. [2] Toennies, G. and Gallant, D.L. (1949) Growth 13, 7-20. [3] Toennies, G. and Gallant, D.L. (1949) Growth 13, 21-25. [4] Shockman, G.D. (1963) In: Analytical Microbiology (Kavanagh, F., Ed.), pp. 567-673. Academic Press, New York. [5] Toennies, G. and Shockman, G.D. (1953) Arch. Biochem. Biophys. 45, 447-458. [6] Shockman, G.D., Kolb, B.J.J. and Toennies, G. (1958) J. Biol. Chem. 230, 961-977. [7] Shockman, G.D. (1959) J. Biol. Chem. 234, 2340-2342. [8] Toennies, G., Bakay, B. and Shockman, G.D. (1959) J. Biol. Chem. 234, 3269-3275. [9] Toennies, G. and Shockman, G.D. (1963) Biochemistry 2, 294-296. [10] Shockman, G.D. (1963) Trans. NY Aead. Sci. Ser. 11, 26, 182-195. [11] Shockman, G.D. (1960) Bacteriol. Rev. 29, 345-358. [12] Toennies, G. and Gallant, D.L. (1949) J. Biol. Chem. 177, 831-839. [13] Salton, M.R.J. (1964) The Bacterial Cell Wall. Elsevier, Amsterdam. [14] Hammond, S.M., Lambert, P.A. and Rycroft, A.N. (1984) The Bacterial Cell Surface. Croom Helm, London and Sydney. [15] Cohen, S.S. and Barrier, H.D. (1956) J. Bacteriol. 71, 588-597. [16] Shockman, G.D., Conover, M.J., Kolb, J.J., Riley, L.S. and Toennies, G. (1961) J. Bacteriol. 81, 44-50. [17] Shockman, G.D., Conover, M.J., Kolb, J.J., Phillips, P.M., Riley, L.S. and Toennies, G. (1961) J. Bacteriol. 81, 36-43.
[18] Mitchell, P. and Moyle, J. (1957) J. Gen. Microbiol. 16, 184-194. [19] Weidel, W., Frank, H. and Martin, H.H. (1960) J. Gcn. Microbiol. 22, 158-166. [20] H61tje, J.-V. and Tuomanen, E.I. (1991) J. Gen. Microbiol. 137, 441-454. [21] Doyle, R.J. and Koch, A.L. (1987) Crit. Rev. Microbiol. 15, 169-222. [22] Daneo-Moore, L. and Shockman, G.D. (1977) In: The Synthesis, Assembly and Turnover of Cell Surface Components (Poste, G. and Nicoloson, G.L., Eds.), pp. 597715. North Holland Publishing Co., Amsterdam. [23] Shockman, G.D. and Barrett, J.F. (1983) Annu. Rev. Microbiol. 37, 501-527. [24] McDowell, T.D. and Lemanski, C.L. (1988) J. Bacteriol. 170, 1783-1788. [25] Coyette, J., Perkins, H.R., Polacheck, I., Shockman, G.D. and Ghuysen, J.-M. (1974) Eur. J. Biochem. 44, 459-468. [26] Kariyama, R., Massidda, O., Daneo-Moore, L. and Shockman, G.D. (1990) J. Bacteriol. 172, 3718-3724. [27] Cole, R.M. and Hahn, J.J. (1962) Science 135, 722-723. [28] Cole, R.M. (1964) Bacteriol. Rev. 29, 326-344. [29] Higgins, M.L. and Shockman, G.D. (1970) J. Bacteriol. 101,643-648. [30] Higgins, M.L. and Shockman, G.D. (1970) J. Bacteriol. 103, 244-254. [31] Higgins, M.L., Pooley, H.M. and Shockman, G.D. (1971) J. Bacteriol. 105, 1175-1183. [32] Boothby, D., Daneo-Moore, L., Higgins, M.L., Coyette, J. and Shockman, G.D. (1973) J. Biol. Chem. 248, 21612169. [33] Kawamura, T. and Shockman, G.D. (1983) FEMS Microbiol. Lett. 19, 65-69. [34] Kawamura, T. and Shockman, G.D. (1983) J. Biol. Chem. 258, 9514-9521. [35] Dolinger, D.L., Schramm, V.L. and Shockman, G.D. (1988) Proc. Natl. Acad. Sci. USA 85, 6667-6671. [36] Barrett, J.F. and Shockman, G.D. (1984) J. Bacteriol. 159, 511-519. [37] Barrett, J.F., Schramm, V.L. and Shockman, G.D. (1984) J. Bacteriol. 159, 520-526. [38] Barrett, J.F., Dolinger, D.L., Schramm, V.L. and Shockman, G.D. (1984) J. Biol. Chem. 259, 11818-11827. [39] Dolinger, D.L., Daneo-Moore, L. and Shockman, G.D. (1989) J. Bacteriol. 17l, 4355-4361. [40] Shockman, G.D., Dolinger, D.L. and Daneo-Moore, L. (1988) In: Antibiotic Inhibition of Bacterial Cell Surface Assembly and Function (Actor, P., Daneo-Moore, L., Higgins, M.L., Salton, M.R.J. and Shockman, G.D., Eds.), pp. 195-210. American Society for Microbiology, Washington. [41] Shockman, G.D., Chu, C.-P., Kariyama, R., Tepper, L.K. and Daneo-Moore, L. (1992) In: Proceedings of the FEMS Symposium Bacterial Growth and Lysis: Metabolism and Structure of the Bacterial Sacculus (de Pedro, M.A., L6ffelhardt, W. and H61tje, J.-V., Eds.). Plenum, New York, in press.
267 [42] Chu, C.-P., Kariyama, R., Daneo-Moore, L. and Shockman, G.D. (1992) J. Bacteriol. 174, 1619-1625. [43] Garcla, E., Garcla, J.L., Garcla, P., Arraras, A., Sanchez-Puelles, J.M. and Ldpez, R. (1988) Proc. Natl. Acad. Sci. USA 85,914-918. [44] Garcla, E., Garcia, J.L., Garcla, P., Sanchez-Puelles, J.M. and Ldpez, R. (1990) Gene 86, 81-88. [45] Beliveau, C., Potvin, C., Trudel, J., Asselin, A. and Bellemare, G. (1991) J. Bacteriol. 173, 5619-5623. [46] Joris, B., Englebert, S., Chu, C.-P., Kariyama, R., Daneo-Moore, L., Shockman, G.D. and Ghuysen, J.-M. (1992) FEMS Microbiol. Lett. 91,257-264. [47] Kariyama, R. and Shockman, G.D. (1992) J. Bacteriol. 174, 3236-3241. [48] Ghuysen, J.-M. (1991) Annu. Rev. Microbiol. 45, 37-67. [49] Kariyama, R. and Shockman, G.D. (1992) In: Proceedings of the FEMS Symposium - Bacterial Growth and
[50] [51] [52] [53] [54]
[55]
Lysis: Metabolism and Structure of the Bacterial Sacculus (de Pedro, M.A., L6ffelhardt, W. and H61tje, J.-V., Eds.). Plenum, New York, in press. Higgins, M.L., Pooley, H.M. and Shockman, G.D. (1970) J. Bacteriol. 103, 504-512. Pooley, H.M. and Shockman, G.D. (1970) J. Bacteriol. 103, 457-466. Hinks, R.P., Daneo-Moore, L. and Shockman, G.D. (1978) J. Bacteriol. 133, 822-829. Hinks, R.P., Daneo-Moore, L. and Shockman, G.D. (1978) J. Bacteriol. 134, 1074-1080. Cleveland, R.F., H61tje, J.-V., Wicken, A.J., Tomasz, A., Daneo-Moore, L. and Shockman, G.D. (1975) Biochem. Biophys. Res. Commun. 67, 1128-1135. Cleveland, R.F., Wicken, A.J., Daneo-Moore, L. and Shockman, G.D. (1976) J. Bacteriol. 126, 192-197.
261
FEMSLE 80034
The autolytic ('suicidase') system of Enterococcus hirae: From lysine depletion autolysis to biochemical and molecular studies of the two muramidases of Enterococcus hirae ATCC 9790 G e r a l d D. S h o c k m a n Department of Microbiology and Immunology, Temple University School of Medicine, Philadelphia, Pennsylvania, USA Received 18 May 1992 Accepted 1 July 1992
Key words: Autolysin; Muramidase; Enterococcus hirae; Penicillin-binding protein
1. S U M M A R Y Autolysis of Enterococcus hirae ATCC 9790 is the result of the action of endogenous enzymes that hydrolyze bonds in the protective and shape-maintaining cell wall peptidoglycan. It is thought that these potentially suicidal enzymes play a positive role(s) in wall growth and division and are expressed as autolysins when cell wall assembly a n d / o r repair are inhibited. E. hirae possesses two potentially autolytic enzymes, both of which are muramidases. Although they hydrolyze the same bond as hen egg-white lysozyme, both are high-molecular-mass, complex enzymes. Muramidase-1 is synthesized as a zymogen, requiring protease activation. It is a glucoenzyme that is also multiply nucleotidylated with an unusual nucleotide, 5-mercaptouridine monophosphate. Muramidase-2 is almost certainly a product of a separate gene. The deduced amino acid
sequence of a cloned gene for extracellular muramidase-2 showed several unusual features. It appears to be a two-, or perhaps three-domain protein with a putative glycosidase-active site near the N-terminal end and six 45-amino-acid-long repeats at the C-terminal end which are presumed to be involved with high-affinity binding to the insoluble peptidoglycan substrate. Muramidase-2 binds penicillin with low affinity. The presence of several amino acid groupings characteristic of serine-active site /3-1actam-interactive proteins is consistent with the possible presence of a penicillin-binding, third domain. Indirect evidence consistent with a role(s) for these enzymes in cell wall growth and division has been obtained. However, proof of such role(s) awaits modern genetic, molecular, and biochemical analyses.
2. I N T R O D U C T I O N Correspondence to: G.D. Shockman, Department of Microbiology and Immunology, Temple University School of Medicine, Philadelphia, Pennsylvania, USA.
In the late 1940s and early 1950s, well before the advent of amino acid analyzers, the nutritional requirements of fastidious bacteria, primar-
262
ily lactobacilli and streptococci, were used to quantify the amino acid (and vitamin) content of various foods and other materials. Attempting to increase the reliability and precision of microbiological amino acid assays, Dr. G. Toennies and collaborators carefully examined many of the variables of turbidimetric methods rather than the then widely used acidimetric assays [1-3]. For various reasons he selected as the test organism what was then known as Streptococcus faecalis ATCC 9790, which later became Streptococcus faecium, Enterococcus faecium and, most recently, Enterococcus hirae. As an organic chemist, his choice of a salt-tolerant Enterococcus was fortuitous, as it permitted an increase in the buffer concentration of the chemically defined medium used [4] to 0.3 M sodium phosphate, thus greatly reducing the growth (and later, autolysis) inhibitory effects of the vast amounts of lactic acid metabolically generated by the fermentation of glucose during growth of this organism. Much later, two observations were found to be directly related to cell wall metabolism of E.
hirae. The first observation was that although exhaustion of growth-limiting quantities of most of the amino acids required by this organism (e.g. valine, threonine, leucine, isoleucine, histidine, or methionine) signalled the end of the exponentialgrowth phase, in the absence of further protein synthesis, culture turbidity continued to increase, postexponentially, at a continuously decreasing rate [5,6]. Later, this phenomenon was shown to be due to continued cell wall accumulation [6-9], resulting in bacteria that possessed a larger amount of cell wall in the form of a thickened wall structure rather than an increase in wall surface area [10,11]. The exception was lysine depletion. In this case, rather than an increase in culture turbidity, lysine exhaustion was followed by a reasonably rapid, pronounced decrease in culture turbidity [6,12]. The use of a highly buffered culture medium and maintenance of a nearly neutral pH was essential for the observation of cellular autolysis. The reason for this exception only became obvious several years later as an indirect result of bacterial cell wall chemistry studies from several
laboratories throughout the world (see ref. 13 for an excellent detailed summary, or ref. 14 for a more recent, less detailed review). Lysine was the only nutritionally required amino acid examined that was a component of the cell wall peptidoglycan of E. hirae, in addition to being a protein amino acid! Both phenomena, wall thickening and cellular autolysis, are examples of unbalanced growth [15]. Thus, deprivation of an amino acid that was a component of both protein and bacterial wall peptidoglycan resulted in autolysis. This led to the proposal that the attainment of osmotic fragility and bacterial autolysis induced by treatments with antibiotics that specifically inhibited further wall peptidoglycan biosynthesis, was not dependent on the continued synthesis of protein and other cytoplasmic macromolecules. Therefore, bacteria did not become osmotically fragile by simply 'growing out' of the protection of their walls, but this process required the active participation of an enzyme activity(ies) that hydrolyzed preexisting bonds in the peptidoglycan that were essential for maintaining the integrity of, and protective functions of, the wall [6,11]. Thus, the statement in many texts that /3-1actam and other cell wall antibiotics are lethal only on growing bacteria is not quite true! More precisely, the requirement is that the drug must interrupt ongoing assembly of new wall peptidoglycan into an expanding surface.
3. AUTOLYSIS OF E. HIRAE Subsequent observations showed that exponentially growing and dividing cells of E. hirae were prone to autolysis which could be observed not only by the depletion of lysine but by placing bacteria from an exponentially growing culture into a culture medium lacking one or more nutritionally required precursors of peptidoglycan, including glutamic acid and glucose [10,16,17], or more simply, into a buffer of appropriate pH and concentration. These data led us to ask why growing and dividing bacteria should carry around with themselves such potentially suicidal enzyme activities. This question led us to the hypothesis
263 that autolytic peptidoglycan hydrolases play some role in peptidoglycan assembly, bacterial wall growth, and cell division [11], an idea that was also proposed by others [18,19] based on data obtained with other bacterial species. It was also rather obvious that such 'suicidases' must be extremely well regulated and integrated with wall biosynthesis, so that hydrolysis of bonds in the peptidoglycan exoskeleton was rapidly followed by biosynthetic processes that, among other things, repaired the potential damage. The possession of one, or usually more, autolytic peptidoglycan hydrolase activities is not unique to E. hirae. A variety of Gram-positive and Gram-negative species, including Escherichia coli [20] have been shown to have such activities. Thus a variety of subsequent data from this and other laboratories (see refs. 21-23 for reviews) are consistent with, support, and extend the hypothesis that peptidoglycan hydrolases, which, at least in some cases, are potential suicidases, are related to, and perhaps essential for, the processes that result in expansion of the protective wall surface to enclose an increasing volume of cytoplasmic contents. In addition, because the wall peptidoglycan is the cellular component primarily responsible for maintaining the shape of the bacterial cells, which subtly changes during the cell division cycle, the broad general hypothesis has been modified to include possible roles for these enzymes in hydrolysis of a small number of bonds at appropriate topological sites. Those strategic cuts would engineer modifications of the previously assembled structure, resulting in small changes in shape (morphogenesis) of the exoskeleton. The potential suicidal nature of these activities, plus the postulated action at precise times and topological locations, implies that these enzyme activities must be precisely regulated and coupled with cell wall biosynthetic, or at least repair, processes. For such positive roles only a relatively small number of well regulated enzyme molecules per bacterial cell are required. We know of many bacterial species and strains that fail to exhibit a gross autolytic behavior when exposed to conditions that result in rapid lysis of closely related or similar bacterial species. Examples of such a con-
trast are many Group A streptococci and some viridans streptococci such as Streptococcus mutans, which are resistant to autolysis when exposed to/3-1actam or other cell wall antibiotics in contrast to the rapid autolysis of E. hirae ATCC 9790. The recent demonstration that a Group A streptococcus resistant to autolysis possessed a low but detectable level of peptidoglycan hydrolase activity [24] is consistent with the hypothesis that small amounts of such activities may be essential and play a positive role in surface growth. Certainly the existence of autolytic defective mutants of a variety of both Gram-positive and Gram-negative bacterial species suggests that only relatively low levels of peptidoglycan hydrolase activities may be essential. In support of this hypothesis of essentiality is the complete lack of success in isolating truly autolytically negative, otherwise isogenic strains of any bacterial species. The presence in many species of more than one and, in some instances, multiple peptidoglycan hydrolases, is also consistent with one or more essential functions of such enzymes.
4. T H E P E P T I D O G L Y C A N OF E. H I R A E ATCC 9790
HYDROLASES
In contrast to a variety of both Gram-positive and Gram-negative bacteria which contain peptidoglycan hydrolases of two or more specificities, a broad variety of data are completely consistent with the presence of only one specificity of peptidoglycan hydrolase in E. hirae (summarized in ref. 22). The only detectable bond cleaved is the fl, 1-4 bond between N-acetylmuramic acid and N-acetylglucosamine by the action of a lysozymelike, N-acetylmuramoylhydrolase (muramidase). No evidence of the presence of amidase, endopeptidase (except for a fl-lactam sensitive DDcarboxypeptidase [25,26], whose action does not affect the two- or three-dimensional integrity of the peptidoglycan structure) or glucosaminidase activities was obtained. The presence of only muramidase activity is consistent with the belief that this organism is a relatively simple model of wall surface growth and degradation, since (i) it can only divide in one plane, (ii) it is comparatively
264 simple in shape, consisting only of poles and lacking cylindrical wall, (iii) it enlarges its wall surface area at only a limited number of nascent septal sites [27-29], and (iv) it conserves assembled wall for a number of generations [30,31] and fails to shed cell wall by the process known as turnover [32]. However, the observation of Kawamura and Shockman [33] that E. hirae possessed two separate and biochemically distinct peptidoglycan hydrolases as well as a growing assortment of data which show that, in contrast to hen eggwhite lysozyme, both muramidases are highmolecular-mass complex proteins indicate that this model is far from simple. 4.1. Some properties of muramidase-1 Muramidase-1 is present as a high-molecularmass (130-kDa) zymogen that can be proteolytically cleaved by any of a broad variety of proteases to an 87-kDa active form [34]. It is a glycoprotein that is multiply substituted with oligosaccharide containing only glucose units [34], thus, it is a glucoenzyme. In addition to possessing covalently linked glucose, the polypeptide also possesses about 12 phosphodiester-linked monomeric nucleotides [35]. Direct probe mass spectroscopy was used to identify this nucleotide as 5-mercaptouridine monophosphate [35]. Phosphodiesterase treatment of native muramidase-1 resulted in the removal of about half of the nucleotides and phosphorus and was accompanied by about a 50% reduction in enzyme activity (Dolinger and Shockman, unpublished data). These last data suggest that nucleotidylation plays a role in enzyme activity, although the exact role remains to be elucidated. Muramidase-1 has an unusual mechanism of hydrolysis. Studies of hydrolysis of soluble linear peptidoglycan chains produced and secreted by penicillin-inhibited cells of autolytic defective strains of E. hirae [36] showed that the enzyme binds to the non-reducing ends of these 42- to 47-disaccharide-peptide-long chains and sequentially (processively) hydrolyzes the bonds between N-acetylmuramic acid and N-acetylglucosamine along the chains yielding virtually only one product - - monomeric disaccharide-peptide units [37,38].
4.2. Some properties of muramidase-2 The presence of a second peptidoglycan hydrolase in extracts of E. hirae was observed as an activity that both rapidly dissolved walls of Micrococcus luteus and failed to bind to concanavalin A (ConA)-Sepharose [33]. This activity was not observed earlier, as it has little ability to dissolve its homologous substrate (wails of E. hirae), although it appears to hydrolyze some glycosidic bonds in the latter substrate. Further studies showed that the activity that failed to bind to ConA-Sepharose and probably is not a glycoprotein, is a muramidase [39] (muramidase-2) that substantially differs in substrate specificity [33,40] from muramidase-1. Although it is remotely possible that muramidase-2 is a non-glucosylated, a n d / o r non-nucleotidylated, form of muramidase-1, substantial evidence exists that each is a product of separate genes [41]. Muramidase-2 also appears to be a complex protein. After sodium dodecylsulfatepolyacrylamide gel electrophoresis (SDS-PAGE), two polypeptide bands with apparent molecular masses of about 125 and 70 kDa were shown to possess peptidoglycan hydrolase activity and to have the ability to bind radioactively labelled pelficillin with low affinity [39]. These two polypeptides appeared to have the same electrophoretic mobility on SDS-PAGE as penicillin binding proteins (PBPs) 1 and 5 present in membrane preparations of E. hirae. However, recently obtained immunochemical and molecular data clearly demonstrate that PBP 5 and the 70-kDa form of muramidase-2 are separate and distinct proteins [41]. Recently [42] we successfully cloned in E. coli and sequenced a gene for the extracellular form of muramidase-2. The deduced amino acid sequence of the 666-amino-acid-long open reading frame showed several interesting features: (i) a 49-amino-acid-long putative signal sequence with a potential protease cleavage site between Ala 49 and Asp 50. This cleavage site was confirmed by the virtual identity of the determined N-terminal amino acid sequence of mature extracellular muramidase-2 with the deduced amino acid sequence; and (ii) the presence of six 45-aminoacid-long highly homologous repeat units, sepa-
265 rated by intervening sequences highly enriched for serine and threonine (approx. 51% serine plus threonine), was found at the C-terminal end of the polypeptide. Although these repeats have little homology with repeat units found at the Cterminal ends of a variety of different bacterial proteins, including those in the pneumococcal amidase gene [43] and the muramidase genes of several pneumococcal phages [44], substantive homology was observed with four contiguous Cterminal 68-amino-acid-long repeats of an autolysin of unspecified enzymatic specificity of S. faecalis [45]. The amino acid sequence of the S. faecalis autolysin showed other areas of substantive homology with muramidase-2, including the putative signal peptide and much of the remaining polypeptide, except for a 104-amino-acid-long insertion between amino acids 64 and 65 of the E. hirae muramidase-2, and for the serine- plus threonine-enriched sequences [46]. Analogous to the interpretations of Garcla and co-workers [43,44] that the highly conserved pneumococcal repeats may be involved in the binding of these enzymes to wall of pneumococci possessing choline-containing wall teichoic acid, it seems likely that the six 45-amino-acid-long repeats of muramidase-2 are somehow involved with the high-affinity binding [47] of this enzyme to the wall peptidoglycan of E. hirae. Further analysis of the derived amino acid sequence of muramidase-2 showed the presence of an SXXK-active site motif and of several amino acid groupings such as SGN and KSG, that are characteristic motifs of serine /3-1actamases and PBPs [48]. Although the spacings of the motifs in muramidase-2 are not inconsistent with those found in other penicillin-interactive proteins, the shape and distribution of hydrophobic and hydrophilic clusters along the amino acid sequences of both muramidase-2 and the S. faecalis autolysin differ from those found in a broad assortment of serine fi-lactamases and in a number of PBPs [46]. Thus it is possible that the mechanism of penicillin interaction may differ from that of the typical serine-active site penicillin-interactive protein [49].
4.3. Possible functions and regulation of the two potentially suicidal muramidases of E. hirae Such potentially suicidal activities that are present in growing and dividing bacteria must be exquisitely well regulated and integrated with wall and other biosynthetic processes. It is also clear that endogenous peptidoglycan hydrolases are responsible for the lysis of cells of some species and strains of bacteria after antibiotic or nutritional inhibition of peptidoglycan assembly. However, a number of observations are consistent with, but do not prove, a well regulated role for such activities in wall growth and division. For the E. hirae system we have established that: (i) wall autolytic activity is associated with newly assembled wall, at nascent septa [50] in growing and dividing bacteria; (ii) inhibition of protein synthesis either by the use of antibiotics or nutritionally results in a rapid (and reversible) decrease in autolytic capacity although enzyme activity(ies) remains present [51]; (iii) rates of cellular autolysis correlate with growth rates of cultures [52]; (iv) bacterial cells from synchronized cultures show variations in rates of autolysis that can be correlated with the division cycle [52,53]; (v) membrane-associated ligands such as acylated lipoteichoic acid and certain lipids, notably cardiolipin, have been shown to inhibit autolysis of cells and isolated walls [54, 55]; and (vi) additional potential levels of regulation include substrate specificity, topological location of enzyme molecules a n d / o r susceptible bonds, the processive hydrolytic mechanism of muramidase-1, proteinase activation of the latent form of muramidase-1, and the state of protein folding of either of the enzymes. The presence of two large and complex muramidases in a bacterial cell of comparatively simple shape and mode of division implies that hydrolytic function may be essential to the bacterium. In 'normal' growth situations, whatever they may be in natural environments, the two activities may have separate and distinct, and perhaps complementary, functions in wall growth and division. However, deletion of one activity may not be lethal. The other activity might be
266
able to fully or partially take over that function. Certainly the presence of two separate and distinct gene products helps to explain our difficulties in obtaining truly autolysis-negative mutants! We hope that the power of modern genetic, molecular and biochemical methods will help us to solve the questions of functions and essentiality of these peptidoglycan hydrolases in the near future.
ACKNOWLEDGEMENT The work in the author's laboratory was supported by U.S. Public Health Service grant AI05044 from the National Institutes of Health.
REFERENCES [1] Toennies, G. and Gallant, D.L. (1948) J. Biol. Chem. 174, 451-463. [2] Toennies, G. and Gallant, D.L. (1949) Growth 13, 7-20. [3] Toennies, G. and Gallant, D.L. (1949) Growth 13, 21-25. [4] Shockman, G.D. (1963) In: Analytical Microbiology (Kavanagh, F., Ed.), pp. 567-673. Academic Press, New York. [5] Toennies, G. and Shockman, G.D. (1953) Arch. Biochem. Biophys. 45, 447-458. [6] Shockman, G.D., Kolb, B.J.J. and Toennies, G. (1958) J. Biol. Chem. 230, 961-977. [7] Shockman, G.D. (1959) J. Biol. Chem. 234, 2340-2342. [8] Toennies, G., Bakay, B. and Shockman, G.D. (1959) J. Biol. Chem. 234, 3269-3275. [9] Toennies, G. and Shockman, G.D. (1963) Biochemistry 2, 294-296. [10] Shockman, G.D. (1963) Trans. NY Aead. Sci. Ser. 11, 26, 182-195. [11] Shockman, G.D. (1960) Bacteriol. Rev. 29, 345-358. [12] Toennies, G. and Gallant, D.L. (1949) J. Biol. Chem. 177, 831-839. [13] Salton, M.R.J. (1964) The Bacterial Cell Wall. Elsevier, Amsterdam. [14] Hammond, S.M., Lambert, P.A. and Rycroft, A.N. (1984) The Bacterial Cell Surface. Croom Helm, London and Sydney. [15] Cohen, S.S. and Barrier, H.D. (1956) J. Bacteriol. 71, 588-597. [16] Shockman, G.D., Conover, M.J., Kolb, J.J., Riley, L.S. and Toennies, G. (1961) J. Bacteriol. 81, 44-50. [17] Shockman, G.D., Conover, M.J., Kolb, J.J., Phillips, P.M., Riley, L.S. and Toennies, G. (1961) J. Bacteriol. 81, 36-43.
[18] Mitchell, P. and Moyle, J. (1957) J. Gen. Microbiol. 16, 184-194. [19] Weidel, W., Frank, H. and Martin, H.H. (1960) J. Gcn. Microbiol. 22, 158-166. [20] H61tje, J.-V. and Tuomanen, E.I. (1991) J. Gen. Microbiol. 137, 441-454. [21] Doyle, R.J. and Koch, A.L. (1987) Crit. Rev. Microbiol. 15, 169-222. [22] Daneo-Moore, L. and Shockman, G.D. (1977) In: The Synthesis, Assembly and Turnover of Cell Surface Components (Poste, G. and Nicoloson, G.L., Eds.), pp. 597715. North Holland Publishing Co., Amsterdam. [23] Shockman, G.D. and Barrett, J.F. (1983) Annu. Rev. Microbiol. 37, 501-527. [24] McDowell, T.D. and Lemanski, C.L. (1988) J. Bacteriol. 170, 1783-1788. [25] Coyette, J., Perkins, H.R., Polacheck, I., Shockman, G.D. and Ghuysen, J.-M. (1974) Eur. J. Biochem. 44, 459-468. [26] Kariyama, R., Massidda, O., Daneo-Moore, L. and Shockman, G.D. (1990) J. Bacteriol. 172, 3718-3724. [27] Cole, R.M. and Hahn, J.J. (1962) Science 135, 722-723. [28] Cole, R.M. (1964) Bacteriol. Rev. 29, 326-344. [29] Higgins, M.L. and Shockman, G.D. (1970) J. Bacteriol. 101,643-648. [30] Higgins, M.L. and Shockman, G.D. (1970) J. Bacteriol. 103, 244-254. [31] Higgins, M.L., Pooley, H.M. and Shockman, G.D. (1971) J. Bacteriol. 105, 1175-1183. [32] Boothby, D., Daneo-Moore, L., Higgins, M.L., Coyette, J. and Shockman, G.D. (1973) J. Biol. Chem. 248, 21612169. [33] Kawamura, T. and Shockman, G.D. (1983) FEMS Microbiol. Lett. 19, 65-69. [34] Kawamura, T. and Shockman, G.D. (1983) J. Biol. Chem. 258, 9514-9521. [35] Dolinger, D.L., Schramm, V.L. and Shockman, G.D. (1988) Proc. Natl. Acad. Sci. USA 85, 6667-6671. [36] Barrett, J.F. and Shockman, G.D. (1984) J. Bacteriol. 159, 511-519. [37] Barrett, J.F., Schramm, V.L. and Shockman, G.D. (1984) J. Bacteriol. 159, 520-526. [38] Barrett, J.F., Dolinger, D.L., Schramm, V.L. and Shockman, G.D. (1984) J. Biol. Chem. 259, 11818-11827. [39] Dolinger, D.L., Daneo-Moore, L. and Shockman, G.D. (1989) J. Bacteriol. 17l, 4355-4361. [40] Shockman, G.D., Dolinger, D.L. and Daneo-Moore, L. (1988) In: Antibiotic Inhibition of Bacterial Cell Surface Assembly and Function (Actor, P., Daneo-Moore, L., Higgins, M.L., Salton, M.R.J. and Shockman, G.D., Eds.), pp. 195-210. American Society for Microbiology, Washington. [41] Shockman, G.D., Chu, C.-P., Kariyama, R., Tepper, L.K. and Daneo-Moore, L. (1992) In: Proceedings of the FEMS Symposium Bacterial Growth and Lysis: Metabolism and Structure of the Bacterial Sacculus (de Pedro, M.A., L6ffelhardt, W. and H61tje, J.-V., Eds.). Plenum, New York, in press.
267 [42] Chu, C.-P., Kariyama, R., Daneo-Moore, L. and Shockman, G.D. (1992) J. Bacteriol. 174, 1619-1625. [43] Garcla, E., Garcla, J.L., Garcla, P., Arraras, A., Sanchez-Puelles, J.M. and Ldpez, R. (1988) Proc. Natl. Acad. Sci. USA 85,914-918. [44] Garcla, E., Garcia, J.L., Garcla, P., Sanchez-Puelles, J.M. and Ldpez, R. (1990) Gene 86, 81-88. [45] Beliveau, C., Potvin, C., Trudel, J., Asselin, A. and Bellemare, G. (1991) J. Bacteriol. 173, 5619-5623. [46] Joris, B., Englebert, S., Chu, C.-P., Kariyama, R., Daneo-Moore, L., Shockman, G.D. and Ghuysen, J.-M. (1992) FEMS Microbiol. Lett. 91,257-264. [47] Kariyama, R. and Shockman, G.D. (1992) J. Bacteriol. 174, 3236-3241. [48] Ghuysen, J.-M. (1991) Annu. Rev. Microbiol. 45, 37-67. [49] Kariyama, R. and Shockman, G.D. (1992) In: Proceedings of the FEMS Symposium - Bacterial Growth and
[50] [51] [52] [53] [54]
[55]
Lysis: Metabolism and Structure of the Bacterial Sacculus (de Pedro, M.A., L6ffelhardt, W. and H61tje, J.-V., Eds.). Plenum, New York, in press. Higgins, M.L., Pooley, H.M. and Shockman, G.D. (1970) J. Bacteriol. 103, 504-512. Pooley, H.M. and Shockman, G.D. (1970) J. Bacteriol. 103, 457-466. Hinks, R.P., Daneo-Moore, L. and Shockman, G.D. (1978) J. Bacteriol. 133, 822-829. Hinks, R.P., Daneo-Moore, L. and Shockman, G.D. (1978) J. Bacteriol. 134, 1074-1080. Cleveland, R.F., H61tje, J.-V., Wicken, A.J., Tomasz, A., Daneo-Moore, L. and Shockman, G.D. (1975) Biochem. Biophys. Res. Commun. 67, 1128-1135. Cleveland, R.F., Wicken, A.J., Daneo-Moore, L. and Shockman, G.D. (1976) J. Bacteriol. 126, 192-197.
