45
BACrERIOLOGICAL REVIEWS, Sept. 1974, p. 291-335 Vol. 38, No. 3 Copyright 0 1974 American Society for Microbiology Printed in U.S.A. Interaction of Penicillin with the Bacterial Cell: Penicillin-Binding Proteins and Penicillin-Sensitive Enzymes PETER M. BLUMBERG' AND JACK L. STROMINGER The Biological Laboratories, Harvard University, Cambridge, Massachusetts 02138 INTRODUCTION . ........................................................ 291 General ............... ......................................... 291 Cell Waill Structure and Synthesis ....................... ........................ 292 Identification of the Target of Penicillin Action .......... ........................ 293 PENICILLIN-BINDING COMPONENTS .............. .......................... 296 What Is Actually Bound? ....................................................... 298 How Much Penicillin Do Cells Bind? ................. ........................... 298 Is the Penicillin Bound Covalently? ................... .......................... 299 To What Group Does Penicillin Bind?..... 299 Under What Conditions Can Penicillin Be Released from the PBCs?........ ........ 300 What Is the Physical Nature of the PBCs? ............. .......................... 300 Are PBCs Located in Peripheral Membranes, Mesosomes, or Both? ...... ........ 301 Do Organisms Have One or Multiple Proteins Which Bind Penicillin? ..... ...... 301 Isolation of PBCs by Covalent Affinity Chromatography ........ ................. 303 What Is the Relationship of the PBCs to the Killing Site? ....... ................. 304 Biochemical Evidence for the Involvement of PBCs in Peptidoglycan Synthesis . . 305 PENICILLIN-SENSITIVE ENZYMES ................ ........................... 306 Transpeptidase . ........................................................ 306 Transpeptidase assayed with natural acceptor in a coupled system ............ 306 Transpeptidase assayed with unnatural acceptor in a coupled system ..... ..... 307 Transpeptidase assayed in a simple, uncoupled system ....... ................. 307 D-Alanine Carboxypeptidase .................................................... 308 E. coli carboxypeptidase ...................................................... 309 Carboxypeptidases from Bacillus ................... .......................... 309 Streptomyces carboxypeptidases .................... .......................... 309 Endopeptidase . ........................................................ 313 Other Activities . ........................................................ 314 Number of Enzymes Corresponding to the Different Penicillin-Sensitive Enzy- matic Activities ........................................................ 314 Relationship ofthe Penicillin-Sensitive Enzymes to the PBCs ....... ............. 316 Physiological Functions of the Penicillin-Sensitive Enzymes ....... .............. 316 Physiological Evidence for the Existence of Multiple Transpeptidases ..... ...... 318 Kinetics of Penicillin Inhibition ................................................. 319 Models for the Interaction of Penicillin with Its Target ........ .................. 321 MECHANISM OF KILLING BY PENICILLINS .......... ....................... 324 Mutant Analysis . ....................................................... 324 Evidence That Penicillin Kills by Inhibiting Cross-Linking . ...................... 324 Killing by Amidino Penicillin ................................................... 325 Transfer of Penicillin Side Chain ................................................. 325 CONTROL OF CELL WALL SYNTHESIS ............. .......................... 326 SUMMARY ................ ....................................... 327 LITERATURE CITED ....................................................... 327 INTRODUCTION classes, the penicillins and cephalosporins (see Fig. 1). The central feature of the antibiotic General nucleus is the highly strained beta-lactam ring. The beta-lactam antibiotics stand out among As discussed below, this ring may be that part antimicrobial agents for-their wide spectrum of of the molecule which reacts with the penicillin activity and for their remarkably low toxicity to target. The side chains attached to the nucleus animals. These substances are composed of two are of importance because of their ease of chemical modification and the profound effects 'Present address: Department of Biology, Massachusetts which they exert on the properties of the mole- Institute of Technology, Cambridge, Mass. 02139. cule (162). Different side chains can, for exam- 291 on August 25, 2019 by guest http://mmbr.asm.org/ Downloaded from

Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

Embed Size (px)

Citation preview

Page 1: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BACrERIOLOGICAL REVIEWS, Sept. 1974, p. 291-335 Vol. 38, No. 3Copyright 0 1974 American Society for Microbiology Printed in U.S.A.

Interaction of Penicillin with the Bacterial Cell:Penicillin-Binding Proteins and Penicillin-Sensitive Enzymes

PETER M. BLUMBERG' AND JACK L. STROMINGERThe Biological Laboratories, Harvard University, Cambridge, Massachusetts 02138

INTRODUCTION......................................................... 291General ............... ......................................... 291Cell Waill Structure and Synthesis ....................... ........................ 292Identification of the Target of Penicillin Action .......... ........................ 293

PENICILLIN-BINDING COMPONENTS .............. .......................... 296What Is Actually Bound? ....................................................... 298How Much Penicillin Do Cells Bind? ................. ........................... 298Is the Penicillin Bound Covalently? ................... .......................... 299To What Group Does PenicillinBind?..... 299Under What Conditions Can Penicillin Be Released from the PBCs?........ ........ 300What Is the Physical Nature ofthe PBCs? ............. .......................... 300Are PBCs Located in Peripheral Membranes, Mesosomes, or Both? ...... ........ 301Do Organisms Have One or Multiple Proteins Which Bind Penicillin? ..... ...... 301Isolation ofPBCs by Covalent Affinity Chromatography ........ ................. 303What Is the Relationship of the PBCs to the Killing Site? ....... ................. 304Biochemical Evidence for the Involvement ofPBCs in Peptidoglycan Synthesis . . 305

PENICILLIN-SENSITIVE ENZYMES ................ ........................... 306Transpeptidase......................................................... 306Transpeptidase assayed with natural acceptor in a coupled system ............ 306Transpeptidase assayed with unnatural acceptor in a coupled system ..... ..... 307Transpeptidase assayed in a simple, uncoupled system ....... ................. 307

D-Alanine Carboxypeptidase .................................................... 308E. coli carboxypeptidase ...................................................... 309Carboxypeptidases from Bacillus ................... .......................... 309Streptomyces carboxypeptidases .................... .......................... 309

Endopeptidase......................................................... 313Other Activities......................................................... 314Number of Enzymes Corresponding to the Different Penicillin-Sensitive Enzy-

matic Activities ........................................................ 314Relationship ofthe Penicillin-Sensitive Enzymes to the PBCs ....... ............. 316Physiological Functions ofthe Penicillin-Sensitive Enzymes ....... .............. 316Physiological Evidence for the Existence of Multiple Transpeptidases ..... ...... 318Kinetics of Penicillin Inhibition ................................................. 319Models for the Interaction ofPenicillin with Its Target ........ .................. 321

MECHANISM OF KILLING BY PENICILLINS .......... ....................... 324Mutant Analysis........................................................ 324Evidence That Penicillin Kills by Inhibiting Cross-Linking ....................... 324Killing by Amidino Penicillin ................................................... 325Transfer of Penicillin Side Chain ................................................. 325

CONTROL OF CELL WALL SYNTHESIS ............. .......................... 326SUMMARY ................ ....................................... 327LITERATURE CITED ....................................................... 327

INTRODUCTION classes, the penicillins and cephalosporins (seeFig. 1). The central feature of the antibiotic

General nucleus is the highly strained beta-lactam ring.The beta-lactam antibiotics stand out among As discussed below, this ring may be that part

antimicrobial agents for-their wide spectrum of of the molecule which reacts with the penicillinactivity and for their remarkably low toxicity to target. The side chains attached to the nucleusanimals. These substances are composed of two are of importance because of their ease of

chemical modification and the profound effects'Present address: Department of Biology, Massachusetts which they exert on the properties of the mole-

Institute of Technology, Cambridge, Mass. 02139. cule (162). Different side chains can, for exam-291

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 2: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

Penicillin

R-CO-HN-C -C ""CH3

-N CH-COOH

Cepholosporin

H H/sRI-CO-HN-C C CH

~CN CH2-R?0' \C COOH

FIG. 1. Structures ofpenicillin and cephalosporin.

ple, render the penicillin resistant to degrada-tion by penicillinase, tolerant of gastric acidity,or able to penetrate the outer envelope ofgram-negative organisms.

Since the introduction of penicillin into gen-eral medical practice, a continual problem hasbeen the emergence of drug-resistant strains ofbacteria (201). In many cases this resistance hasbeen shown to arise from production of beta-lac-tamase (penicillinase), an enzyme which de-grades the antibiotic. In other cases, the intrin-sic sensitivity of the organism is what appears tohave been altered. Synthesis of novel penicillinderivatives has fortunately kept pace so far withthe appearance of penicillin-resistant bacterialstrains. In order to maintain such progress, a

detailed understanding of the mode of action ofpenicillin in killing bacteria, and the nature ofthe interaction between penicillin and its recep-tor should be valuable.The remarkably low toxicity of penicillin for

animals suggests that the drug inhibits somebacterial structure or function without analoguein higher organisms. In fact, the target ofpenicillin has been demonstrated to be thebacterial cell wall, a unique microbial structure.(A trivial exception to this generalization exists[204]. The cephalosporin 7-(5-benzyl-thi-oacetamido)-cephem-3-ylmethyl-N-dimethyldi-thiocarbamate-4-carboxylic acid also inhibitsprotein synthesis and thus is active against bothfungi and bacterial spheroplasts. Apparently,the cephalosporin decomposes to liberate thedimethyldithiocarbamate in the 3-position onthe molecule. This hydrolysis product is respon-sible for the unusual antibiotic behavior.) Thecell wall is a giant macromolecule which en-velopes the organism, supporting the bacterialcell membrane against lysis caused by the dif-ference in osmolarity between the cell cyto-plasm and the culture medium, which is rela-

tively hypotonic. Because cell wall synthesizedin the presence of penicillin is weak and unableto provide the cell with this needed osmoticsupport, penicillin lyses growing bacteria. Con-sistent with this explanation, penicillin does notkill cells growing in sufficiently hypertonic me-dium. Likewise, bacteria are protected from theaction of penicillin in the absence of cell growth,in which case new cell wall synthesis is not re-quired.The action of penicillin as an inhibitor of cell

wall biosynthesis offers a second importantreason for studying the interaction of this drugwith the bacterial cell. Many fundamental, butpoorly understood, phenomena are related tocell wall synthesis: morphogenesis-the cellwall is responsible for maintaining the shape ofthe bacterium. What regulates the shape as-sumed by the cell wall? Cell division-divisionis by definition the formation of a septumcontaining cell wall and cell membrane to yieldtwo daughter cells. Membrane proteins-manyof the proteins involved in cell wall synthesis aremembrane bound, including most of the peni-cillin receptors. Regulation of complex syn-thetic pathways-cell wall synthesis involvesthe interplay of large numbers of enzymes and isclosely related as well to the synthesis of theother macromolecules in the cell. Penicillinprovides a valuable tool for investigation ofsome of these problems.

Cell Wall Structure and SynthesisThe bacterial cell wall is a highly complex

structure consisting of multiple classes of poly-mers. These polymers include the peptidogly-can, teichoic acids, attached and secreted poly-saccharides or proteins, or both, and, ingram-negative organisms, lipopolysaccharide.The state of knowledge regarding these struc-tures has been ably discussed in recent reviews,which cover cell walls in general (94, 174, 199,208, 227), peptidoglycan (83, 86, 93, 210, 217,226, 244), teichoic acids (4, 7, 8, 9), and lipo-polysaccharide (129, 136, 165). Among thesestructures, the peptidoglycan is of particularimportance. Its integrity is required for mainte-nance of cell shape in bacteria. Moreover, thepeptidoglycan is the one cell wall constituentuniversally distributed in bacteria (except inextreme halophiles and in mycoplasma, both ofwhich exist in very specialized environments),and, unlike most of the other cell wall polymers,it is essential for cell survival under normalgrowth conditions.The precise composition of the peptidoglycan

varies with the bacterial species. Typical struc-

292 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 3: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

tures are those shown in Fig. 2. The peptidogly-can, also known as murein or mucopeptide,possesses a heteroglycan backbone of alternat-ing residues of N-acetylglucosamine and N-acetylmuramic acid. The N-acetyl-muramicacid residues are substituted by peptide chainswhich are cross-linked to give a mesh-likecharacter to the peptidoglycan. The net result isto knit the entire bacterial cell wall into a"bag-shaped macromolecule" (236).

Synthesis of peptidoglycan can be dividedconceptually into three stages (217, 218, 220).The first step, catalyzed on the inside of the cellby cytoplasmic enzymes, is the synthesis of thesoluble peptidoglycan precursor uridine 5'-diphosphate (UDP)-N-acetylmuramyl-pen-tapeptide (Fig. 3). This step is followed bytransfer of the N-acetylmuramyl-pentapeptideand N-acetylglucosamine to a lipid carrier,C55-isoprenyl alcohol, in the membranes, form-ing a subunit of the glycan polymer (Fig. 4). Themembrane-bound disaccharide pentapeptide isnow modified in a species-specific fashion. Suchmodification may include substitution of acarboxyl group on glutamic or diaminopimelicacid residues or attachment of a peptide sidechain (e.g., the pentaglycine chain attached tothe lysine residue of the pentapeptide in Staph-ylococcus aureus [Fig. 4]).

In the last stage, the modified disaccharide-pentapeptide residue is transferred to a glycanacceptor on the outside of the cell to form linearpeptidoglycan. Finally, the peptide chains ofthe linear peptidoglycan are cross-linked in areaction catalyzed by a transpeptidase. It is thislast step which is the penicillin-sensitive reac-tion in cell wall synthesis. The cross-bridge isformed between the carboxyl group of the pe-

MurNac

/ L-ala-D-Glu-meso-DAP-D-Ala

GIcNac

GlcNac

MurNac

L-Ala-D-Glu-meso-DAP-D-Ala MurNac

GlcNac D-Ala-meso-DAP-D-Glu-L-Ala

GlcNac

MurNac

L-.,la-,-lu-meso DAP-D-Ala

A

nultimate D-alanine in the pentapeptide on onechain and an amino group in a nearby chain, forexample, that of the diaminopimelic acid resi-due (Fig. 5a). When a peptide cross-bridge ispresent, e.g., the pentaglycine chain in S. au-reus, the amino group of the cross-bridge is usedinstead (Fig. 5b). The driving force for thecross-linking is provided by release of the termi-nal D-alanine from the pentapeptide chain andprobably by the insolubility of the cross-linkedproduct. By using this transpeptidation mecha-nism, the organism avoids the energetic prob-lem that would be involved in achieving netsynthesis of peptide bonds outside the cell.

In some organisms, a D-alanine carboxypepti-dase activity is present in addition to thetranspeptidase. This activity, found in Esche-richia coli and Bacillus subtilis but not in S.aureus, cleaves the terminal D-alanine from thepentapeptide chain of several different sub-strates. Its function is presently only conjec-tural. Because the tetrapeptide resulting fromthe action of the carboxypeptidase cannot func-tion as a donor in transpeptidation, one hypoth-esis is that its function may be to decrease theamount of cross-linking in the cell wall oforganisms possessing this enzyme.

Identification of the Target of PenicillinAction

The effect of penicillin on cell wall structureand morphology has been known since the 1940s(59, 84, 85). Duguid's early analysis of themorphological effects showed great perspicacity(59). He observed not only that penicillin re-quired cell growth for activity, but also that celldeath appeared to be caused by cell lysis: "The

/ / /(Gy5 MurNAc/

\ / L-Ala GlcNAcD-Glu-NH2

GlcNAc L-LysD- Ala MurNAc

MurNAc 0Gy5-Glu-NH2(Gly)5 / L-Ala / \L-Lys

0-A-la-NH2 ,GlcNAc D-Ala/ L-Lys / 0-AWla

GIcNAc 0-Ala oMurNAc

-Ala-H

I > D-Glu- NH2

/5 sL-LysAcIcNAc 0-AaI

/ (Gly)I

B ,'FIG. 2. Structure of the peptidoglycan. A, Peptidoglycan of Escherichia coli. B, Peptidoglycan of

Staphylococcus aureus. MurNac, N-acetylmuramic acid; GlcNac, N-acetylglucosamine.

VOL. 38, 1974 293

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 4: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

294 BLUMBERG AND STROMINGER

UTP + N-acetylglucosamine-1-PPP

UDP-GIcNAc

phosphoenolpyruvate-UDP-GIcNAc-pyruvate enol ether

h-JTPNHUDP-MurNAc

L-alan ne

ATP D-glutamic acid

L-alanine j L-lysine

D-alanine UDP-Mu rNAc-L-Ala-D- G-Glu-L-Lys

ATP

ATP D-alanyl-D-alanine

U D P-Mu rNAc-L-Ala - D-GIu -L-Lys- D-Ala-D-Ala(U D P-acetylmu ra myl - pentapeptide)

BAcTERIOL. REV.

CH2OH

2Z2~UDP

HO

NHCOCH3 NH2

CH3-CH-CO NH-CH-CONH - CH-(CH2)-CO H-CH-CO H-CH-CO NH- CH-COOHJCH3 bi3

FIG. 3. First stage in cell wall synthesis: formation of UDP-N-acetylmuramyl-pentapeptide (structureshown).

UDP-MurNAc-Pentapeptide lUMP

r, P-Phosp-opMurNAc (-pentapeptide)-P-P-Phospholipid

P-P-Phospholipid UDP-GlcNAc

GlcNAc-MurNAc-(deca- UDP

peptide amide)-AcceptorGIcNAc-MurNAc(-pentapeptide)-P-P-Phospholipid

I ATP, NH,Acceptor /

GIcNAc-MurNAc(-decapeptide amide)- GlcNAc-MurNAc(-p.ntapeptide amide)-P-P-Phospholipid P-P-Phospholipid

tRNA Glycyl-tRNAFIG. 4. Second stage in cell wall synthesis. The modification reactions illustrated are those which occur in

CH3

Staphylococcus aureus. The lipid moiety is a [C,,]isoprenyl alcohol [H(CH,-C CH-CH2)1,0H].

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 5: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

GlcNAc-MurNAc L -Ala * D -Glu - meso-DAP D -Alaa D -Ala

GlcNAc-MurNAc * L-Ala D-Glu * meso -DAP D-Ala * D-Ala

GlcNAc-MurNAc' L-Ala D-GlU * meso-DAP D-Ala

GlkNAc-MurNAc L-Ala D-GIu *r eso-DAP D-Ala D-Ala + D-Ala

/0 A

Peptidoglycan polymer

MurNAcI

L-Ala

Peptidoglycan polymer

MurNAc

L-Ala

D- lu D-GluI I

L-Lys-Gly-Gly-Gly-Gly-Gly--, L-Lys-I \ I

D-Ala Ad D-AlaI'D-AI

D-Ala D-Ala

Peptidoglycan polymer Peptidoglycan polymer

MurNAc MurNAc

L-Ala L-Alal I

D-Glu D-GluLLys-'-,Gl- GI D-GLysG

I Y-G111-G1)e-G IYGGL4Ys-GlYy-GD-Ala / -- -l

D-Ala B

FIG. 5. Final stage in cell wall synthesis: cross-linking of peptidoglycan polymers by transpeptidation. A,Escherichia coli; B, Staphylococcus aureus.

morphological changes produced by the lowerpenicillin concentrations ... suggest that peni-cillin at these concentrations interferes specifi-cally with the formation of the outer supporting

cell wall, while otherwise allowing growth toproceed until the organism finally bursts itsdefective envelope and so undergoes lysis." Adecade later, these observations had been con-

-Gly-Gly-Gly-Gly-Gly

transpeptidase

VOL. 38, 1974 295

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 6: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

firmed and extended by the finding that suchpenicillin-treated cells could be protected fromthe action of the antibiotic by the presence of anosmotic support, either sucrose or sodium chlo-ride (97, 116, 117). This result demonstrated a

causal link between the observed changes in cellmorphology, lysis, and cell death.

Elucidation of the specific lesion in cell wallsynthesis caused by penicillin followed from a

different line of investigation. Early importantobservations were that labile esters, apparentlycontaining reducing sugars (181), and acid solu-ble absorbing materials, presumably free nu-

cleotides (155), accumulated in cells inhibitedby penicillin. Determination of the structure ofthese nucleotide-sugar compounds (175-177,216) revealed that they included UDP-N-acetyl-muramyl-pentapeptide. This finding suggestedthat penicillin blocked a stage in cell wallbiosynthesis subsequent to synthesis of thisnucleotide precursor (182, 221). These stagesare now known to include both glycan polymeri-zation and peptide cross-linking. Martin andco-workers (137-140) in studying the composi-tion of the peptidoglycan of the cell wall ofProteus mirabilis were led at first to the hypoth-esis that penicillin was inhibiting cross-linking,though later analyses led them to doubt theirearlier conclusion (109). Wise and Park (243)and Tipper and Strominger (228, 229) on thebasis of different studies of the effects of penicil-lin on peptidoglycan synthesis in S. aureus

concluded that the terminal reaction was a

transpeptidation and that penicillin specificallyblocked this reaction. (However, the generalityof this conclusion has not been established).This conclusion that penicillin blocked cross-

linking was supported by in vitro studies.Transfer of sugar-peptide from nucleotide pre-cursor to lipid and acceptor was not inhibitedby large excesses of penicillin (33, 142, 143, 147);the sole remaining reaction which penicillinmight inhibit in the biosynthetic sequence was

thus transpeptidation. Finally, development ofan in vitro transpeptidation assay permitted thedirect demonstration of penicillin action (2, 105,106).The model of transpeptidation and the pro-

posed mode of action of penicillin which evolvedfrom the above studies were as follows (222, 228,229): the transpeptidase was hypothesized toreact with the peptide bond between the termi-nal D-alanines in the pentapeptide chain ofuncross-linked peptidoglycan. An acyl-enzymeintermediate would be formed and D-alaninereleased. The amino group from the prospectivecross-bridge would next displace the enzymefrom the acyl-enzyme intermediate, regenerat-

ing free enzyme and forming a peptide cross-bridge (Fig. 6).

Penicillin was hypothesized to be an analogueof the terminal D-alanyl-D-alanine in the pen-tapeptide chain (Fig. 7). The CO-N bond in thehighly strained beta-lactam ring would corre-spond to the peptide bond cleaved during trans-peptidation, and penicillin might, in fact, be ananalogue of the transition state in peptide bondcleavage. The transpeptidase would react withpenicillin to split the beta-lactam ring to form apenicilloyl-enzyme complex, analogous to thepostulated acyl-enzyme intermediate. However,since the penicilloyl-enzyme complex is stable,the transpeptidase would be inactivated (Fig.6).The above brief description of cell wall syn-

thesis does not emphasize the complexity of thephenomenon. Regulation of wall synthesis istied into deoxyribonucleic acid (DNA) replica-tion and the cell cycle. A specific cell shape ismaintained during growth. The amount of wallsynthesis is varied with the growth rate. Thelevels of degradative and synthetic enzymes arepreserved in careful balance.

Recently, evidence of this complexity hasbegun to appear from several laboratories (219).Multiple penicillin targets exist. How many arethere? What are their relationships to eachother and their roles in cell wall metabolism?Furthermore, the precise manner in which peni-cillin inhibits its various targets has never beenfully elucidated. Is penicillin actually a sub-strate analogue or might it act at an allostericsite, as recently suggested (123)? Does penicillinalways act in the same way? The discussionwhich follows will examine the evidence regard-ing these questions.

PENICILLIN-BINDING COMPONENTSPenicillin was first demonstrated to bind to

bacteria in the late 1940s (50, 51, 134, 135, 203).Its interaction with the cell has subsequentlybeen studied in considerable detail (1, 6, 44-46,49-51, 55, 58, 63-65, 67, 69, 75, 76, 110, 111,113-115, 134, 135, 157, 160, 179, 185, 188, 189,192, 197, 200, 202, 203, 209, 211, 225). An earlydiscussion of the subject which showed greatinsight, but which is now out of date, is that ofCooper (48). The goal of the penicillin-bindingstudies has been twofold, to elucidate the in-teraction of penicillin with the drug receptor(s)and to investigate the roles these receptors playin the metabolism of the cell. The state ofunderstanding of both questions is still incom-plete. However, the application of modern bio-chemical techniques has led to dramatic prog-ress over the last few years. It has been possible

296 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 7: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

A /CONH-D-AIa-C-NH-D-Ala0 IO. UOOH

B

+H2O A

B

Transpeptidase A

CONH-D-Ala B /NHCO-Gly-l

CROSS-LINKED GLYCOPEPTIDE

D-ALAN I NECARBOXYPEPTI DASE

TRANSPEPTI DATI ONREACTION

FIG. 6. Proposed mechanism of transpeptidation and its relationship to penicilloylation and D-alaninecarboxypeptidase activity. A represents the end of the main peptide chain of the glycan strand. B represents theend of the pentaglycine substituent from an adjacent strand. If the acyl enzyme intermediate can react withwater instead of the acceptor (left), the enzyme would be regenerated and the substrate released. The overallreaction would be the hydrolysis of the terminal D-alanine residue of the substrate (D-alanine carboxypeptidaseactivity).

HH H

H _.c c

H

H I7/

H H

FIG. 7. Dreiding stereomodels of penicillin (left)and of the D-alanyl-D-alanine end of the peptidogly-can strand (right). Arrows indicate the position of theCO-N bond in the beta-lactam ring of penicillin andof the CO-N bond in D-alanyl-D-alanine at the end ofthe peptidoglycan strand (228).

to demonstrate multiple targets for penicillin.The roles of some of these targets have beenidentified. Furthermore, the penicillin-bindingcomponents have been isolated either as a

mixture, or, for one of the components, in pureform. By using this material, the interaction ofpenicillin with its receptors has been shown tobe much more complicated than previouslybelieved.The amount of penicillin which bacteria bind

is very small. As a result, early efforts to detectthe removal of penicillin from culture by bacte-ria were unsuccessful. Specific binding wasobscured by the background loss of penicillindue to decomposition. Demonstration of bind-ing therefore had to await the synthesis ofradioactive penicillin, labeled in one of twoways. Early studies often used 35S-labeled peni-cillin produced biosynthetically by growth ofthe penicillin-producing mold in the presence ofradioactive sulfate. The chief advantage of thismaterial was its high specific activity, up to 300mCi/mmol (209). This permitted studies whichwould not have been possible with material oflower specific activity (209). Its major drawbackwas the relatively short half-life of 35S, only 90days. The alternative strategy was to use peni-cillin radioactively labeled with 14C in the sidechain attached to the penicillin nucleus. Thisproduct, unlike the 35S-labeled material, is nowavailable commercially. Unfortunately, its spe-

R-CONH

PENICILLINACTI ON

VOL. 38, 1974 297

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 8: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

cific activity has generally been in the range of25 mCi/mmol (the theoretical maximal activityfor substitution by a single atom of "4C is 60mCi/matom), which has been inconvenientlylow for many studies of penicillin binding.

What Is Actually Bound?In general, the entire penicillin molecule

appears to react with its target, and a covalentbond appears to be formed between the car-bonyl group of the beta-lactam ring and anunidentified group on the penicillin-bindingcomponents (PBCs) (see later section). In addi-tion to this specific binding, certain penicillindegradation products may also bind to cells.Although no good comparisons have been pub-lished, cells appear to bind comparable quan-tities of penicillin regardless of whether the siteof labeling of the antibiotic is in the side chainor in the drug nucleus (64, 114). Further evi-dence that the entire molecule is bound comesfrom studies of the products liberated uponrelease of the penicillin bound to the bindingcomponents. The 35S-labeled compound re-leased from the PBCs of M. pyogenes by alka-line treatment co-chromatographed with peni-cilloic acid (209). Moreover, derivatives anddegradation products of this material behavedchromatographically in the same way as did thecomparable products arising from penicilloicacid. More recent studies with the membrane-bound PBCs of B. subtilis showed that theproduct of reversal by hydroxylamine co-chromatographed with authentic penicilloyl hy-droxymate in four solvent systems (114). Afterreversal by ethanethiol and methylation withdiazomethane, one of the two products co-chromatographed with authentic a-ethylthio-fl-methylpenicilloate (114). The other had slightlyslower mobility in all four solvent systemsexamined. The significance of this latter obser-vation is not yet known.The above conclusion that the entire penicil-

lin molecule is bound ought not to be general-ized to the case of cephalosporins, because it isknown that hydrolysis of the beta-lactam bondof the cephalosporin leads to expulsion of theside chain in the 3-position (171). A comparablereaction may occur after cephalosporin bindingto PBC, but this matter has not been investi-gated.An early question was whether the binding of

radioactive penicillin which was observed was aproperty of the penicillin per se or whether itwas due to an impurity in the radioactivematerial. The above work strongly implied thatpenicillin itself bound to the PBC (the penicil-

loic acid released by alkali from M. pyogenespresumably was a degradation product of thebound material, because penicilloic acid itselfdoes not bind [49]). Further evidence was thatuptake of the radioactive material was pre-vented by prior addition of pure, unlabeledpenicillin (45, 50, 157, 202).A characteristic of this specific uptake of

radioactive penicillin was that after the PBCsbecame saturated, subsequent addition of ra-dioactive penicillin had no effect on the amountof binding. However, in addition to this specificbinding, a second binding phase was also some-times observed, which was proportional to theamount of penicillin added and which could notreadily be saturated. This second phase, unlikethe specific uptake, was insensitive to suchtreatments as boiling of the PBCs. As shown byCooper et al. and others (46, 49, 197), suchnonspecific binding was due to degradationproducts in the penicillin which accumulatedwith time; the nonspecific binding could beeliminated by repurification of the radioactivepenicillin immediately before use. Although theidentity of the penicillin degradation productsresponsible for such nonspecific binding has notbeen completely established, the most likelycandidate is penicillenic acid. Unlike the otherdegradation products which Daniel and John-son examined, penicillenic acid can compete forbinding with radioactive penicillin (55). Fur-thermore, it can also be shown to bind nonspe-cifically to proteins (55, 119).

How Much Penicillin Do Cells Bind?Determinations of the amount of penicillin

bound by bacteria vary with the growth state ofthe organisms, their treatment, etc. As a generalrule, gram-positive organisms bind 4 to 15 nmolof penicillin per g (dry weight) (48, 225). Thegram-negative organism E. coli may bind 5- to10-fold less. Such numbers correspond to 100 to10,000 molecules of penicillin per cell (47, 64,67, 134, 191, 225); the range of values commonlyreported is 1,000 to 4,000 (64).The major uncertainty in these values lies in

determinations of the number of cells per g (dryweight). For example, Suginaka et al. (225) andRogers (197) both reported S. aureus H to bindspecifically 5.7 nmol of penicillin per g (dryweight). However, the number of molecules percell were reported as 4,000 and 100, respectively.In the first case, the number of cells wasdetermined from a viable count. As recognizedby the authors, this value may be low becauseclumps and chains of organisms are counted asa single organism. Rogers, in contrast, counted

298 BACMIOL. RWv.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 9: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

total cells. This latter method includes non-via-ble particles and, in addition, tends to giveartifactually high values (149).A second cause for uncertainty, in the case of

organisms which bind penicillin only at higherconcentrations and in small amounts, is thatthe background arising from nonspecific bind-ing may be substantial. E. coli, which bindsone-tenth as much penicillin as does B. subtilis,is a typical example (225). Here, 50% of thebinding is nonspecific, and special care must betaken to determine the appropriate blank.

Is the Penicillin Bound Covalently?No general study has been made of bacterial

proteins which bind penicillin tightly, but in areadily reversible fashion; the only work alongsuch lines has been the analysis of cell wallenzymes which have later turned out to bereversibly inhibited by penicillins (see sectionon penicillin-sensitive enzymes). Instead, efforthas concentrated on that firm penicillin bindingwhich requires relatively drastic treatment forreversal. The evidence indicates that this bind-ing is covalent. The nature of the covalent bond,however, is not certain.Bound radioactive penicillin cannot be eluted

from either whole cells or membranes by wash-ing with nonradioactive penicillin (44, 64, 114,134, 202) or released by treatment with penicil-linase (114). Conceivably the binding could benoncovalent, but very tight. However, protein-denaturing agents also do not induce penicillinrelease. Thus, treatment with sodium dodecylsulfate (15, 16, 209) at 100 C, boiling (16, 46),exposure to 8 M urea (16, 225) or 6M guanidine-hydrochloride (16) or phenol extraction (44,202) have no effect. Evidence that such proce-dures actually do denature the PBCs is theability of these treatments to prevent subse-quent specific binding of radioactive penicillin.Similarly, Pronase digestion of the purifiedD-alanine carboxypeptidase-penicillin complexfrom Bacillus stearothermophilus did not re-lease intact penicillin (16). Further evidence fora covalent linkage of penicillin to the PBCs isformation of an a-ethylthio-penicilloate upontreatment of the PBC-bound penicillin withethanethiol under conditions where the ethane-thiol would not have been predicted to reactwith intact penicillin (114).

To What Group Does Penicillin Bind?Early workers (209) believed that the penicil-

lin might be bound as an ester because of itsalkali lability. More recently the bond between

the penicillin and the PBCs has been suggestedto be a thioester, at least in the case of the B.subtilis binding components. The major evi-dence for this conclusion is the characteristics ofreversal of the penicillin binding by variousagents. In particular, hydroxylamine and H202release penicillin under conditions where nor-mal amide and ester linkages allegedly are notcleaved (114, 225). Furthermore, treatmentwith ethanethiol is reported to lead to penicillinrelease (114, 225). Consistent with this explana-tion, prior reaction of the PBCs with thiolreagents prevents the subsequent binding ofradioactive penicillin (114). In addition, withthe purified D-alanine carboxypeptidase of B.subtilis, which accounts for 70% of the totalpenicillin binding in that organism (15), bind-ing of penicillin blocks one of the four sulfhydrylgroups titratable with 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB) (233). Likewise,thiol reagents inactivate the enzyme (232).Although the above evidence is suggestive, it

is by no means persuasive. The concentrationsof thiol reagents employed to block binding bypenicillin are grossly excessive. Whereas 1 mMiodoacetamide suffices to block normal sulfhy-dryl groups, Lawrence et al. (114), for example,used 200 times more to prevent penicillin bind-ing. At such concentrations, the reagent is nolonger specific. Although the reason for the highlevels of reagent which were necessary may bethat the sulfhydryl group is deeply buried on theenzyme so that it is relatively inaccessible to thereagent, alternatively it may be that the alkyl-ating reagent is reacting with a less readilyalkylated residue than cysteine. Moreover, theability of penicillin to protect a sulfhydryl groupin the native enzyme from DTNB implies onlythat the group is in the penicillin binding site,not that it is covalently bound to penicillin.Furthermore, the hydroxylamine-induced re-lease may be an enzymatically catalyzed reac-tion rather than a chemical hydrolysis (16).Hydroxylamine release of the bound penicillinfrom B. subtilis membranes was prevented byboiling of the membranes after binding of thepenicillin (114). An initial interpretation wasthat the denaturing treatment rendered thepenicillin-enzyme bond inaccessible to the hy-droxylamine. More recent studies exclude thisexplanation. If denaturation is effected byagents which unfold proteins, e.g., sodium do-decyl sulfate, hydroxylamine still fails to bringabout release (16). Apparently, undenaturedbinding components are required. The release ofbound penicillin by ethanethiol may likewiserequire active binding components, because its

VOL. 38, 1974 299

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 10: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

action also is blocked by prior boiling of themembranes after binding of radioactive penicil-lin (114). Tentatively, then, the conclusion isthat a sulfhydryl group may be present in thepenicillin binding site of the B. subtilis carboxy-peptidase. This group may or may not beinvolved in covalent binding. Although a num-ber of the experiments discussed above could berepeated and extended, the availability of puri-fied enzymes and PBCs should permit a moredefinitive approach, viz, the isolation and char-acterization of peptides containing bound peni-cillin.

Under What Conditions Can Penicillin BeReleased from the PBCs?

Specific release of penicillin from PBCs oc-curs upon treatment with H202, ethanethiol,and hydroxylamine as discussed above. Thehydroxylamine treatment is of particular notebecause it does not destroy the ability of thePBCs to combine specifically with penicillin(14). Similarly, in the case of the B. subtiliscarboxypeptidase, hydroxylamine restored ac-tivity to the penicillin-inactivated enzyme(115). This finding was crucial for the laterdevelopment of covalent affinity chromatogra-phy as a means of purifying PBCs (describedbelow) (14).Even more mild conditions may lead to

release of penicillin and regeneration of en-zymatic activities. During storage at 6 C in 0.05M tris(hydroxymethyl)aminomethane-hydro-chloride, pH 7.5, 1 mM MgCl2, 1 mM 2-mercap-toethanol, in the presence of penicillinase, en-zymatic activity slowly returned to penicillin-inactivated D-alanine carboxypeptidase from B.subtilis. The rate was dramatically dependenton the penicillin derivative employed. Negligi-ble activity returned when penicillin G or ceph-alothin was used. In contrast, the half-time forhydrolysis with cloxacillin was 68 h and with6-aminopenicillanic acid was 14 h (13).This phenomenon has subsequently been in-

vestigated in more detail (16). The D-alaninecarboxypeptidases from B. subtilis and Bacillusstearothermophilus both release their cova-lently bound penicillin. The rate of releasevaries with the organism, the specific penicillinderivative, and the temperature. At 37 C, thehalf-time of release of penicillin G is 200 min forthe B. subtilis enzyme and 50 to 60 min for theB. stearothermophilus enzyme. The rate ishighly temperature dependent. At 55 C, thehalf-time for release of penicillin G from the B.stearothermophilus enzyme is only 10 min. Thisrelease has the same pH optimum as the

enzymatically catalyzed carboxypeptidase reac-tion.The nature of the penicillin released by the

enzyme is not known. It is neither benzyl-penicillin, benzylpenicilloic acid, benzylpenil-loic acid (the decarboxylation product of peni-cilloic acid), benzylpenicillenic acid (a rear-rangement product of benzylpenicillin impli-cated in nonspecific binding), benzylpenillicacid (an acid-catalyzed rearrangement productof benzylpenicillin), nor phenylacetic acid (thepenicillin side chain).As was the case with the hydroxylamine

release, the spontaneous release of penicillinwould seem to be enzymatically catalyzed;release is prevented by denaturation of theenzyme. An important unresolved question iswhether the enzyme simply labilizes the peni-cillin-enzyme bond, or whether it plays a moredirect role, e.g., by transferring penicillin toacceptors such as hydroxylamine or ethane-thiol. The determination of the structure of thematerial released from the enzyme hopefullymay distinguish these alternatives.

Penicillinases long have been hypothesized tohave evolved from PBCs (190, 228). The demon-stration that PBCs can degrade penicillin, al-beit at a rate 103 to 105 times more slowly thanthe usual beta-lactamases (16, 35), would beconsistent with this hypothesis.

Multiple proteins are responsible for penicil-lin binding in most organisms (see later sec-tion). The lability of the penicillin-protein bondvaries with the particular binding component.Of the five components in B. subtilis mem-branes, component I is the most susceptible tospontaneous cleavage (15). With hydroxyla-mine-induced cleavage, component I reactsmost readily, component V more slowly, andcomponents II and IV more slowly still. Detailedcomparisons of rates have not, however, beenmade (14).

What Is the Physical Nature of the PBCs?The penicillin-binding components are pro-

teins found either entirely or predominantly inthe cell membrane. The binding componentshave been isolated from B. subtilis, S. aureus,and B. stearothermophilus by covalent affinitychromatography (14, 245). In the case of the B.subtilis carboxypeptidase, only one of two puri-fied PBCs studied in any detail, no lipid orcarbohydrate covalently bound to the proteincould be demonstrated (232). The amino acidcomposition of the B. subtilis carboxypeptidasewas relatively typical (232).The molecular weights of PBCs have been

300 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 11: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

determined by polyacrylamide gel electrophore-sis in the presence of sodium dodecyl sulfate.The values range from a molecular weight of50,000 for carboxypeptidases from bacilli toapproximately 120,000 (15, 245). Such valuesare relevant to speculation on the origin ofpenicillinases, because the molecular weights ofpenicillinases with one exception-the R46 pen-icillinase-have been reported to be belowabout 30,000 (35, 36, 196). Therefore, if the twoclasses of proteins are related, extensive dele-tion in the PBC gene must be postulated.Much evidence points to the PBCs being

located in the membrane. Vigorous disruptionof cells led to liberation of the PBCs into the cellsupernatant (51). However, if the cells werebroken more gently, most of the PBCs remainedin the bacterial cell wall fraction instead (75).The difference in appearance of cell wall frac-tions in the two preparations suggested that, inthe latter case, retention of the cytoplasmicmembrane in the wall fraction was responsiblefor preservation of its penicillin-binding ability.This indeed appears to be the explanation. Purewalls do not bind penicillin (58). Moreover,Cooper (45) showed that the distribution ofPBC upon fractionation of cell lysates cor-related with distribution of lipid phosphate.Furthermore, the specific binding activity of thelipid fraction was 7 to 12 times that of the wholeorganisms. Such a value is consistent with thePBCs all being located in the membrane. Adifficulty in excluding the presence of a smallfraction of soluble PBC was that a certainpercentage of the membranes was not recoveredduring centrifugation. These contaminate thesupernatant fraction and may account for thesmall amount of specific penicillin bindingwhich Cooper reported finding in that fraction.

Similar conclusions were reached by Sugi-naka et al. (225). Within an experimental errorof 20%, the same amount of penicillin was foundregardless of whether penicillin binding towhole cells was measured, whether penicillinwas first bound to the cells and membraneswere subsequently prepared from them, orwhether the membranes were prepared first andpenicillin was then bound. As before, the quali-fication on these results is that the 20% experi-mental error could obscure binding by a biologi-cally significant PBC in the cytoplasm. Thereis, however, no evidence for such an enzyme.

Little is known about the orientation of thePBCs in the membrane, viz, whether they areall on the outer surface or partly located onboth surfaces. At least 50% of the D-alaninecarboxypeptidase, the major binding compo-nent of B. subtilis, is on the outer surface (215).

This conclusion is based on the ability ofpenicillin covalently coupled to Sepharose toinhibit the enzyme in protoplasts. Becausesteric interference might be expected to preventcomplete inhibition, the location of the other50% cannot be inferred by this method.

Are PBCs Located in PeripheralMembranes, Mesosomes, or Both?

Mesosomes are often attached to cell wallsepta, which in turn are believed to be the majorsites of cell wall synthesis (32, 70, 77, 206). Apossible prediction is that some or all PBCsmight therefore be localized in the mesosomes.The evidence on this question is unfortunatelyboth inadequate and conflicting.Duerksen (58) reports that 95% of the label in

cells of Bacillus cereus treated with radioactivepenicillin was released upon conversion of thecells to protoplasts by treatment with lysozyme.Because mesosomes are released from the cellby such treatment, this finding would implythat the PBCs are located exclusively in themesosomes. However, in the body of the paper,it appears that release of 58% of the label is amore typical value. Moreover, part of thismaterial is dialyzable. It may be due to freepenicillin trapped by the cell wall, hydrolysis ofthe penicilloyl-PBC bond, or proteolysis of thepenicillin-PBC complex. The strongest state-ment that can be drawn from these findings isthat a portion of the PBCs may be located in themesosomes.A similar conclusion was reached by Forsberg

and Ward for distribution of the D-alaninecarboxypeptidase of Bacillus licheniformis (80).Here, the specific activity of this enzyme, themajor PBC in those bacilli examined, was 65%as great in mesosomes as in peripheral mem-branes.

Do Organisms Have One or MultipleProteins Which Bind Penicillin?

A major change in thinking about how peni-cillin acts on cells has been necessitated by therecent finding of multiple PBCs. The earliestevidence for multiple binding components wasthe suggestion of Mohberg and Johnson (157)that synnematin B (cephalosporin N) did notreact with all the penicillin-binding sites of S.aureus which are sensitive to penicillin G.However, the evidence presented was weak.Competition experiments where radioactivepenicillin G and nonradioactive synnematin Bwere added simultaneously indicated that asmall portion of the penicillin sites (0.4 nmol/gof cells) were not subject to competition by

301VoL 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 12: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

synnematin. Although the authors tried to ex-clude the possibility that this noncompetitivebinding was due to penicillin degradation prod-ucts, nonspecific binding still remains a proba-ble explanation for their findings.Much stronger evidence for multiple PBCs

came by inference from the demonstration thatthe D-alanine carboxypeptidase of B. subtiliswas distinct from the cephalothin killing siteand was most likely not vital for cell metabo-lism (13). In growing cells, exposure to lowlevels of cephalothin rendered the cells osmotic-ally fragile under conditions where at least 85%of the carboxypeptidase activity remained.Conversely, exposure of growing cells to 6-aminopenicillanic acid could inhibit the car-boxypeptidase 95% under conditions in whichthe growth rate of the organism was altered by10% at most. These results implied the exist-ence of two binding components, a cephalothin-sensitive killing site (transpeptidase) and arelatively cephalothin-resistant carboxypepti-dase. Whether separate proteins accounted forthese sites or whether a single protein somehowwas differentiated so as to exhibit these differ-ent behaviors was not determined.Evidence for the former explanation, the

existence of distinct proteins, was provided bySuginaka et al. (225). Column chromatographyand isoelectric focusing of solubilized mem-branes to which radioactive penicillin had beenbound revealed multiple peaks of radioactivity.As discussed by the authors, the interpretationof these multiple peaks is unclear. Althoughthey could represent separate proteins, theymight also be artifacts caused by variations incharge or state of aggregation of a single protein.Indeed, the purified D-alanine carboxypeptidasefrom B. subtilis was subsequently shown to givemultiple peaks upon isoelectric focusing in thepresence of detergent (J. N. Umbreit, PH.D.thesis, Harvard University, Cambridge, Mass.,1972). Nonetheless, in the case of B. subtilismembranes, greatly different sensitivities tocephalothin of the bands found upon isoelectricfocusing strongly argued for the existence of atleast two distinct PBCs. The D-alanine carboxy-peptidase would be among the cephalothin-resistant bands; the killing site would presum-ably be among the cephalothin-sensitive bands.Development of less ambiguous separation

techniques greatly clarified the issue of multiplePBCs. Polyacrylamide gel electrophoresis in so-dium dodecyl sulfate of solubilized membranesto which penicillin had been bound revealedmultiple peaks-five in B. subtilis, three in B.cereus (or B. stearothermophilus (245)), and

two in S. aureus (15) (Fig. 8, 9). For the B.subtilis components, the proportions, molecularweights, and kinetics of binding of penicillinsand cephalosporins were determined (15). Com-ponent V, which accounted for 70% of the totalpenicillin bound, was identified as the D-alaninecarboxypeptidase. The evidence was that (i) itbound penicillins and cephalosporins at thesame rate as did the carboxypeptidase and (ii)that it had the same molecular weight as thepurified carboxypeptidase (232). Three of theother components, I, II, and IV, were cephalo-thin sensitive. The rates of antibiotic binding ofall three of these components were within afactor of three of that estimated for the penicil-lin killing site for the series of four beta-lactam

2000iCONTROL

1600

1200

I U m IZ

400-

3020 30540 jdye

$ ~ ~~~~~~~~position

2000 CEPHALOTHIN1i 2000 PREBOUND

FIG. 8. Penicillin-binding components of Bacillussubtilis. After [("CJpenicillin G was bound, mem-branes were solubilized in sodium dodecyl sulfate(SDS) and subjected to SDS gel electrophoresis.Radioactivity in 1-mm gel slices is shown. ComponentV is the D-alanine carboxypeptidase. The failure ofcomponents III and V to react with cephalothin underthe conditions used is shown below; prior treatmentwith cephalothin does not affect labeling of thesecomponents with [14C]penicillin G (15).

302 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 13: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

B. cereus

V 200

.G 0 .1

S-oureus 400-

400-20-

2 -200

20 Ss

0 20 30 40

10 20 30 40 50 60 70

Fraction NumberFIG. 9. Penicillin-binding components of Bacillus cereus and Staphylococcus aureus. After binding of

[14Cjpenicillin G, membranes were solubilized in sodium dodecyl sulfate (SDS) and subjected to SDS gelelectrophoresis (15).

antibiotics examined. Consequently, all threeare candidates for the penicillin killing site.The multiple PBCs revealed by SDS gels did

not appear to be artifacts of the gel procedure(15). Proteolysis was unlikely because the sam-ples were boiled immediately after addition ofthe sodium dodecyl sulfate gel applicationbuffer. Furthermore, treatment of the mem-branes with the protease inhibitor phenylmeth-ane sulfonyl fluoride had no effect on the gelpattern. Aggregation caused by formation ofdisulfide bridges between proteins likewiseseemed unlikely. Peaks eluted from gels andre-electrophoresed migrated with the same mo-bility as before, and the patterns obtained werequite reproducible from one experiment to thenext. Further confidence in the validity of theseresults is provided by isolation of the PBCs bycovalent affinity chromatography (14).

Isolation of PBCs by Covalent AffinityChromatography

The finding that neutral hydroxylaminecould restore the activity of penicillin-inac-tivated D-alanine carboxypeptidase (115) and

the development of convenient methods forcoupling ligands to Sepharose (54) laid the stagefor the isolation of PBCs by covalent affinitychromatography (14). The affinity resin wasprepared by coupling 6-aminopenicillanic acidvia its free amino group to a long succinyl-diaminodipropylamine side chain attached toSepharose. Membranes were solubilized withnonionic detergent and then reacted with theSepharose-linked penicillin. After penicilloyla-tion of the PBCs had occurred, the unboundproteins were eluted by extensive washing of theSepharose in the presence of high salt. ThePBCs could then be recovered free from con-taminating protein by elution with hydroxyla-mine.

In the case of B. subtilis, the same pattern offive components was obtained by affinity chro-matography as had been found earlier uponelectrophoresis of whole membranes to whichradioactive penicillin had been bound. Becauseprotein rather than radioactivity was beingmonitored, the protein could be reduced prior tosodium dodecyl sulfate gel electrophoresis. Thisfinding eliminated the possibility mentionedabove that some of the binding components

303VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 14: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

observed might have been artifacts induced byformation of disulfide bridges.

Yields of the binding components were ingeneral excellent. Carboxypeptidase was ob-tained in up to 80% yield. Those of componentsI, II, and IV were 50 to 110% of this. ComponentIII, which bound penicillins quite poorly (15),was obtained in much poorer yield, approxi-mately 10%. The PBCs did not appear to bedenatured by the isolation procedure. All stillbound pencillin. For all components, approxi-mately one molecule of penicillin appeared tohave been bound per molecule of protein (12).Although the above procedure gave a mixture

of five PBCs, a minor modification permittedpurification of the B. subtilis carboxypeptidase(Fig. 10). If the membranes were treated withlow levels of cephalothin before addition of theSepharose-linked penicillin, binding to the af-finity resin of the cephalothin-sensitive compo-nents , II, and IV was blocked. Only compo-nents III and V were therefore contained in thehydroxylamine eluant from the affinity column.The customary low yield of component III in theaffinity chromatography step, together withfurther loss during concentration of the eluantfrom the affinity column, reduced the level ofcontamination of the carboxypeptidase to lessthan 1%. Because the carboxypeptidase is amajor membrane protein (15, 232), 0.75% of thetotal protein in the membrane, this procedurehas now made available for study a PBC inquantities of the order of hundreds of milli-grams. In addition, the procedure has also beenapplied for purifying the D-alanine carboxypep-tidase from the thermophilic organism B. stea-rothermophilus (245).A possible criticism of the demonstration of

multiple PBCs in B. subtilis is that conceivablythe multiplicity of components could havearisen from proteolysis of the binding compo-nent of highest molecular weight. Such a sug-gestion seems most unlikely. The proportions ofthe components are relatively constant. Therates at which they bind penicillins are dramat-ically different, up to four orders of magnitude.The carboxypeptidase, the component of lowestmolecular weight, has a unique N-terminalamino acid. Nonetheless, although nonspecificproteolysis seems very unlikely, the possibilityof specific proteolysis cannot be excluded.Proinsulin is converted to insulin, trypsinogento trypsin. Such a process could be involved incontrol of cell wall synthesis. Comparison of thetryptic maps of the isolated binding compo-nents would resolve the matter in a straightfor-ward fashion.

FIG. 10. Penicillin-binding components of Bacillussubtilis isolated by covalent affinity chromatography.The isolated material was subjected to sodium do-decyl sulfate gel electrophoresis, then stained withCoomassie brilliant blue. A, The five penicillin bind-ing components (compare with the radioactive tracein Fig. 8). B, Purified D-alanine carboxypeptidase(component V) prepared by the affinity chromatogra-phy method.

Affinity chromatography on a cephalosporinC column has also been used as one step in apurification of the penicillinase from B. licheni-formis (52). In that case, however, the enzyme isnot covalently linked to the column and iseluted by a change in the pH of the buffer.

What Is the Relationship ofthe PBCs to theKilling Site?

Tremendous effort has been devoted to studyof the PBCs on the assumption that analysis ofthe PBCs may lead to a better understanding ofthe interaction of penicillin with its killing site.However, despite considerable effort, the rela-tionship between PBC and killing site is stillindeterminate. Basically, for a variety of wild-type organisms, the sensitivity to penicillins ofthe PBCs correlates with the sensitivity of thewhole organisms. On the other hand, in penicil-lin-resistant mutants isolated in the laboratory,the resistance of the organisms is not reflectedby the PBCs.

Early observations that penicillin-sensitivestaphylococci bound penicillin whereas resist-

304 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 15: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

ant strains did not (50, 202) led to a closerexamination of this problem. Eagle and col-leagues (63, 64, 67) found that with four strainsof cocci plus the rod-shaped E. coli, theamounts of penicillin bound at low penicillinconcentrations were related to the sensitivity ofthe organism. The more sensitive organismsbound more. In addition, if all of the organismswere exposed to penicillin at the lethal dose(LD)99.9 concentration, then the four coccibound comparable amounts. The low bindingby E. coli could be explained by the presence inthat organism of penicillinase.A different type of experiment carried out by

Edwards and Park (69) correlated the sensitiv-ity of the intact organism to penicillins and thesensitivity of the PBCs. They looked at thebinding to one organism, S. aureus H, of avariety of penicillins and cephalosporins. Be-cause the only radioactive penicillin commer-cially available is penicillin G, the procedurethey used was first to bind the nonradioactivepencillin at different concentrations and thenmeasure residual sites by reacting with labeledpenicillin G. All the beta-lactam antibioticsexamined saturated the penicillin G bindingsites at approximately the minimal inhibitoryconcentration for the particular antibiotic.These findings, together with those on the

five binding components of B. subtilis, are

consistent with the notion that some, but notall, PBCs may be targets for the killing action ofpenicillin. This evidence, however, is restrictedto a narrow range of organisms, gram-positivecocci and a single gram-positive rod. No ade-quate analysis of gram-negative organisms or

other gram-positive organisms has been made.Moreover, due to the presence of multiplePBCs, questions of correlation are only mean-ingful in terms of the correlation in sensitiv-ity of the cell and a given PBC.Examination of penicillin-resistant mutants

derived in the laboratory (65, 66, 179) did notprovide the expected support for the hypothesisthat the PBCs were the target for the lethalaction of penicillin. Although resistance was notdue to an increased rate of inactivation of thepenicillin, a good correlation between resistanceand binding was not observed. For the four cocciexamined by Eagle (65), one organism decreasedthe amount of penicillin it would bind while itspenicillin resistance remained unaltered.Another increased its amount of binding whenits penicillin resistance increased. This study,like much of the early work, was less informa-tive than it would otherwise have been becausethe reaction of penicillin was analyzed in terms

of total binding. The rate of saturation of PBCsby penicillin is the important factor, notwhether the total amount of penicillin bound atsaturation increases or decreases.

In a more recent study, mutants of S. aureusH resistant to up to 105 times the penicillinconcentration lethal for the parent strain wereisolated (179). They still bound large amountsof penicillin at very low penicillin concentra-tions (0.1 Ag/ml for some) although, in fact, theamounts were only 50% of that bound by theparent. The authors hypothesize an explanationfor their data in terms of (i) "functional versusnonfunctional transpeptidase molecules," (ii)"variations in accessibility to penicillin," and(iii) substrate protection of transpeptidase.A study by Sabath et al. (207) sheds further

light on the problem. Strains of S. aureus 104times more resistant to methicillin than thewild type were isolated. The pH of the mediumhad a dramatic effect on this resistance. At pH5.2 the resistance was not expressed. In con-trast, pH had little effect on the sensitivity ofthe wild-type organism to the antibiotic. Theinterpretation is that this "intrinsic resistance"shown by the mutants is due not to an altera-tion in the target enzyme, but to a change in itsaccessibility. A similar situation may exist inthe mutants isolated by Park et al. (179).The conclusion that can be drawn from the

above studies with mutants is that the situationis quite complex. The results can well beexplained by the presence of multiple PBCs,only some of which are lethal targets for penicil-lin. In such cases, binding by the nonlethaltarget may obscure the relevant reaction. Thispossibility had in fact been considered as oneexplanation for the results (65). At that time,the techniques were not available to test thehypothesis. A second possibility is that someportion of the PBCs may be maintained in acryptic state. If a cryptic site were the killingsite and if penicillin binding to it could beobserved only, for example, after disruption ofthe cells, then detection of alterations of it inmutant cells would not be observed by thetechniques ordinarily used. A third possibility,which cannot be ignored, is that none of thePBCs may be the lethal target in some microor-ganisms. The case of Bacillus megaterium (seelater section) may provide an example.

Biochemical Evidence for the Involvementof PBCs in Peptidoglycan Synthesis

If PBCs are involved in cell wall synthesis, aprediction is that if cells are saturated withpenicillin and the penicillin is then removed,

VOL. 38, 1974 305

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 16: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

cell wall synthesis should reflect the blockage ofthe PBCs. Only two studies of limited scopehave examined this question, both using thesame organism, S. aureus strain H.

Rogers (197) examined the effect of priorbinding of penicillin followed by removal on thetotal amount of peptidoglycan synthesized (hedid not look at the degree of cross-linking). Hefound that prior reaction with penicillin greatlysensitized the amount of total cell wall synthe-sis to inhibition upon readdition of penicillin.He interpreted this and other results to indicatethat most of the penicillin binding in S. aureuswas not important, but that a small number ofsometimes cryptic penicillin-sensitive sitesplayed a vital role.The second study (179, 180) showed that cells

of S. aureus grown in the presence of penicillinshowed subsequent inhibition of the totalamount of cell wall synthesized. Moreover,although experimental details were not given,exposure of cells in buffer to penicillin followedby washing of the cells led to partial inhibitionof cell wall cross-linking. This latter resultwould suggest that one of the PBCs is a trans-peptidase in this organism.

In this sort of experiment direct and indirecteffects of the penicillin treatment must bedifferentiated. If cells become "sick" during thepenicillin treatment, the subsequent effect oncell wall synthesis may reflect this disturbancerather than the irreversible binding of thepenicillin. Moreover, peptidoglycan synthesisand cross-linking are different phenomena; therole of the PBCs in these two processes may wellbe different. Finally, results in S. aureus cannotbe generalized either to gram-positive bacilli orto gram-negative organisms.

PENICILLIN-SENSITIVE ENZYMESThe above discussion deals with one approach

to the study of the mode of action of penicillins,namely, the analysis of proteins which irreversi-bly bind beta-lactam antibiotics. The discoverythat penicillin inhibited cell wall biosynthesisled to a second line of investigation in the studyof the enzymatic activities involved in cell wallmetabolism and their sensitivities to penicil-lins. As was the case with the PBCs, cell wallenzymology has turned out to be far morecomplex than had initially been believed. In all,three different penicillin-sensitive enzymaticactivities have now been reported. Severalmajor issues in the field at the present time canbe defined. (i) What relationship do the differ-ent enzymatic activities have to each other? Arethey catalyzed by the same or separate en-

zymes? (ii) How many proteins (enzymes) in agiven organism catalyze any one of these activi-ties? (iii) What is the function of the differenttypes of penicillin-sensitive enzymes? (iv) Whatrelationship do the different activities have tothe mode of killing by penicillin; which one orones are the killing site? (v) How does penicillininteract with these enzymes?The different penicillin-sensitive enzymatic

activities found in bacteria include (i) transpep-tidases, assayed either with natural or withartificial acceptors; (ii) D-alanine carboxypepti-dases; and (iii) endopeptidases, enzymes whichcleave cross-linked peptide dimers of the pepti-doglycan. Of these activities, the first is ofparticular interest because penicillin is thoughtto kill bacteria by inhibiting peptidoglycancross-linking.

TranspeptidaseTranspeptidase assayed with natural ac-

ceptor in a coupled system. This first assaymost closely approximates the in vivo transpep-tidation reaction. After addition of substrate,UDP-acetylmuramyl-pentapeptide, the reac-tion proceeds through the complex series ofpolymerization steps which culminate in pep-tide cross-linking and release of the terminalD-alanine from the pentapeptide chain (Fig. 3,4, and 5). Although for many years such asystem had only been demonstrable with thegram-negative organisms E. coli and Salmo-nella (2, 3, 105, 106), similar reactions have nowbeen carried out with B. megaterium (239, 240),Micrococcus luteus (152), S. aureus (154),Sporosarcinia ureae (128), and B. stearother-mophilus (127).The course of the reaction can be monitored

by release of the C-terminal D-alanine from thepentapeptide. The reaction product is appliedto paper and chromatographed in isobutyricacid-1 N ammonia (5:3) to separate substrate,degradation products, newly formed peptido-glycan, and liberated D-alanine. Under optimalconditions, D-alanine release is coupled to pep-tidoglycan cross-linking; 1 mol of D-alanine isreleased per mole of acetylmuramyl-pentapep-tide incorporated into the peptidoglycan. Analy-sis of the reaction products confirms the occur-rence of cross-linking. Lysozyme digestion ofthe products leads to the release of dimers ofdisaccharide-peptide. In contrast, if cross-link-ing is prevented by penicillin, only monomers ofthe disaccharide-peptide can be obtained.

Several factors seem relatively important inobtaining an in vitro peptidoglycan synthesiz-ing system of the sort described above. First of

306 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 17: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

all, simplicity in the cell wall structure is amajor advantage. Because the coupled systemrequires the activity of all those enzymes func-tioning in the biosynthetic pathway beyond thelevel of the UDP-acetylmuramyl-pentapeptide,each modification of the pentapeptide which isrequired for eventual cross-linking reduces thechances of success. Thus, the two best-studiedpeptidoglycan-synthesizing systems are in E.coli and B. megaterium, neither of which hasmodifications of the basic pentapeptide struc-ture in its peptidoglycan. Conversely, in B.subtilis, where the carboxyl group of the diami-nopimelic acid residue of the pentapeptide isamidated (235), it has not been possible so far todemonstrate in vitro cross-linking, althoughsynthesis of linear peptidoglycan can be ob-tained (115). The existence of modificationreactions is not an insuperable barrier, however.Recently cross-linking has been demonstratedboth in a system from B. stearothermophiluswhich includes an amidation reaction (127) andalso in the even more complicated system in S.ureae (128), where the addition of glycine andD-glutamic acid in the cross bridge precedescross-linking.A second important factor is a fairly low level

of autolytic and other degradative enzymes. Forexample, although autolysins cause only minorproblems in assays with E. coli, they cause

rapid breakdown of the newly formed peptido-glycan in B. subtilis (115). Degradation of thesubstrate must also be avoided. D-Alanine car-boxypeptidases release the terminal D-alaninefrom UDP-acetylmuramyl-pentapeptide, ren-

dering it incapable of acting as a donor intranspeptidation. As a further drawback, thisD-alanine release by the carboxypeptidase canobscure the specific D-alanine release catalyzedby the transpeptidase. In such cases, the extentof transpeptidation must be monitored by themore laborious procedure of lysozyme digestionfollowed by analysis of dimer formation. Thus,transpeptidation is far more readily assayed inE. coli K-12, which has low levels of D-alaninecarboxypeptidase, than in E. coli B, whichpossesses high levels (105, 106).

Thirdly, the gentle preparation of the appro-priate membrane fraction is critical. Differentmeans of breakage can affect specific activitiesby more than a factor of ten (193, 194, 239).Grinding with glass beads is apparently fairlygentle; rupture in a French press is more severe(193). Furthermore, the membrane fractionscomposed of larger membrane pieces appear tobe more active (154, 193). Presumably, exces-sive fragmentation of the membrane leads to

disorganization of the enzyme system. In accordwith this general concept, complete peptidogly-can synthesis was obtained in what was essen-tially the cell wall fraction of S. aureus and M.luteus (152, 154). It was suggested that themembrane attached to the cell wall might beprecisely that active at the growing points fornew cell wall synthesis.The different transpeptidase activities as-

sayed by means of the coupled cross-linkingreaction have all proved to be penicillin sensi-tive. However, although the E. coli (3, 105, 106),S. aureus (154), and M. luteus (152) systems areirreversibly inhibited by penicillins (i.e., thetranspeptidases presumably are PBCs), thisresult cannot be generalized. The transpepti-dase of the B. megaterium system is reported tobe reversibly inhibited by penicillins (239, 240).In this organism, therefore, the transpeptidasemay not be a PBC.Transpeptidase assayed with unnatural

acceptor in a coupled system. A major diffi-culty inherent in any coupled reaction as anassay for transpeptidase is that it cannot beused to assay fractionated membranes. Conse-quently, it cannot provide an assay useful forpurification. From this point of view, transpep-tidation reactions utilizing simpler unnaturalacceptors are potentially of more value.Two such reactions have been described. In

B. megaterium, inclusion of a D-amino acid atmoderately high concentration in the peptido-glycan synthesizing system leads to incorpora-tion of the D-amino acid into the newly synthe-sized peptidoglycan (239, 240). Analysis of theproducts revealed that the D-amino acid (diami-nopimelic acid was used customarily) was incor-porated via a transpeptidation in the normalfashion. Because the enzyme recognized theD-center of the amino acid, dimer formationoccurred in the presence of DD-diaminopimelicacid, which possesses two such centers. Theproducts formed are shown in Fig. 11.This system, although slightly simpler than

the normal transpeptidation assay, is still acoupled system. The simultaneous presence ofthe diaminopimelic acid during synthesis oflinear peptidoglycan is required. The activity ispenicillin sensitive. Of the inhibition, aboutone-half is reversible. The rest is irreversible(see later section).Transpeptidase assayed in a simple, un-

coupled system. An important development hasbeen the demonstration of transpeptidase reac-tions catalyzed by purified D-alanine carboxy-peptidases from Streptomyces spp. (60, 89-92,184, 186) and B. stearothermophilus (195, 245).

307VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 18: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

(-GlcNAc-MurNAc-)n

Ala

Glu

Dap

I +EAla

Ala

(-GIcNAc-Mu rNAc-)n

Ala

Glu

DapD|p +AlaAla

(-GIcNAc-MurNAc-)n

Ala

Glu

+ second unit

(-GIcNAc-Mu rNAc-)n

Ala

GluI I

Dap Dap

Ala - 3P Ala

or

+ Ala

(-GlcNAc-Mu rNAc-) n

Ala

Glu

(-GlcNAc-MurNAc-) n

Ala

Glu+ AlaI I

Dap < Dapl l

Ala -- Ala

FIG. 11. Transpeptidation reactions in particulate enzyme from Bacillus megaterium. Three separatetranspeptidation reactions can occur. (A) Diaminopimelic acid (Dap) is incorporated at the end of the peptidesubunit replacing D-alanine. Either meso- or DD-diaminopimelic acid can be utilized (but not the LL form). (B)If DD-diaminopimelic acid is the substrate in reaction one, a second subunit can interact with the first one, withthe two amino groups ofDap forming a bridge between the two peptidoglycan units. (C) If meso-diaminopimelicacid is the substrate in reaction one, a different transpeptidation results in a normal cross-link between the twostrands and a diaminopimelic acid terminal residue.

These assays employ synthetic and naturalpeptides which function in an uncoupled systemboth as donors and acceptors in the reactions.These activities were discovered in enzymes

which were first isolated as D-alanine carboxy-peptidases and have given rise to extensivediscussion of the physiological functions of such"DD-carboxypeptidase transpeptidases." Theyare discussed below together with the carboxy-peptidases.

D-Alanine CarboxypeptidaseIn addition to the transpeptidases, a second

penicillin-sensitive activity which has attracted

considerable attention is that of the D-alaninecarboxypeptidases, first encountered in E. coli(2, 105). These enzymes specifically hydrolyzethe peptide bond between the two terminalD-alanine residues of UDP-acetylmuramyl-L-ala-D-glu-meso-Dap-D-ala-D-ala (Fig. 12). Sofar, carboxypeptidases which utilize the corre-

sponding L-lysine-containing substrate have notbeen identified. Upon its discovery, it was

postulated that the D-alanine carboxypeptidasemight be an "uncoupled" transpeptidase (105,115), where water rather than an amino group

displaced the enzyme from the postulated acyl-enzyme intermediate (see Fig. 12). Other possi-

(A)

(B)

(C)

308 BACMMROL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 19: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

a)Natural substrate

UDP-MurNAc- L-Ala - D- Glu -meso- Dap-D-Ala -D-Ala

UDP - MurNAc-L-Ala - D-Glu- meso-Dap-D-+

D-Ala

H20 mine acceptor (e.g. glycine)

-Ala UDP- MbrNAc-L-Ala-D-Glu-meso-Dap-D-Ala-Gly

D-Ala

b) Synthetic substrate

N. N- diacetyl - L- Lys - D-Ala - D- Ala

H2 amine acceptor (e.g. glycine)

N,N-diacetyl-L-Lys -D-Ala NN -diacetyl-L-Lys-D-Ala-Gly

D-Ala D-Ala

FIG. 12. D-Alanine carboxypeptidase and model transpeptidase reactions.

ble roles have also been suggested.Three classes of carboxypeptidases have been

examined. Some of their properties are summa-rized in Table 1. Because these enzymes are theonly penicillin-sensitive enzymes which so farhave been purified to homogeneity, they havebeen the object of most of the detailed study ofthe interaction of penicillin and substrates withpenicillin-sensitive enzymes.E. coli carboxypeptidase. The soluble E. coli

carboxypeptidase has only been partially puri-fied (107). Unlike the other carboxypeptidases,it may have a requirement for the nucleotidemoiety of its substrate; the enzyme is only 20%as active on acetylmuramyl-pentapeptide as onUDP-acetylmuramyl-pentapeptide (107). Thisreduction in activity may or may not be anartifact of metal ion concentration. The enzymealso stands out for its sensitivity to penicillins.Penicillin G at a concentration of 0.002 gg/mlcauses 50% inhibition.Carboxypeptidases from Bacillus. The B.

subtilis carboxypeptidsase by contrast is lo-cated in the membrane and has been purified tohomogeneity by two procedures: by conven-tional enzyme purification techniques after sol-ubilization in detergent (231, 232) and by cova-lent affinity chromatography (14). The enzymehas been extensively studied both as a PBC andas a penicillin-sensitive enzyme. Unlike the E.coli carboxypeptidase, it does not appear torecognize the nucleotide portion of its substrate,because it possesses comparable activity on

UDP-acetylmuramyl-pentapeptide and acetyl-muramyl-pentapeptide (115). In addition, theB. subtilis enzyme can cleave the syntheticsubstrate N,N-diacetyl-lysyl-D-alanyl-D-ala-nine used to assay the Streptomyces enzymes(see below). The B. subtilis carboxypeptidasedisplays intermediate sensitivity to penicillins.It is inhibited by 0.3 Aig/ml of penicillin G (13).However, it is only inhibited by 150 jg/ml ofcloxacillin.The B. stearothermophilus carboxypeptidase

resembles the B. subtilis enzyme in most re-spects. It is distinguished, however, by itsunusual temperature stability. It should benoted that its purification by covalent affinitychromatography, at least as reported, is lesssatisfactory than that of the B. subtilis enzyme.The low increase in specific activity suggeststhat partial inactivation of the enzyme mayoccur during purification. Like the Strep-tomyces carboxypeptidases discussed below,the B. stearothermophilus enzyme can carry outa single transpeptidation reaction by usingUDP-acetylmuramyl- L-ala- D-glu-meso-Dap- D-ala- D-ala as donor and glycine or D-alanine, butnot diaminopimelic acid, as acceptor (11, 195)(Fig. 12). At an acceptor-donor ratio of 100:1,the percentage of donor transpeptidated wasgreater than 60%. Indeed, at a 1:1 ratio ofD-alanine to donor, 5% transpeptidation stilloccurred.Streptomyces carboxypeptidases. The

Streptomyces spp. apparently secrete into the

309VoL 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 20: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

~LO -4etw co

a)n F

'-4e CfD

_I CO m C14eC C)

koCY

LO u oF e)

eO CO )<-H )C

0

4)~~~~a

al

b bIt 00

=~~~~~~~~c W

C~~~~

co-

0

0 4.

X0 O8 W0Of0 s b

0i c

o C.')@"0 > ^

0-8. e0 .

' g0A0

0.0

* B__

.~~~~~~~~~0

at) t¢ c

mQ U

-oC._

c)._

0J0o

0Q

0

0Y.

0

"0

C0

0

qa)

"0

C.X

QO00w

*- 0

00

enZ

BACERIOL. REV.

culture medium enzymes possessing D-alaninecarboxypeptidase activity on the synthetic sub-strate N,N-diacetyl-lysyl-D-alanyl-D-alanine(88, 120-123). Three of the four enzymes exam-ined (from four different Streptomyces species)are also capable of acting as transpeptidases byusing as donor N,N-diacetyl-L-lysyl-D-alanyl-D-alanine and as acceptor D-alanine, glycine, ordiaminopimelic acid. Many interesting observa-tions have been made during study of theseenzymes. Their properties are summarized inTable 2.Determination of carboxypeptidase activity on

an extensive series of synthetic peptides of thegeneral structure X - R, - R2 - R1 (OH) re-veals certain substrate requirements of the en-zymes (120, 121, 123):

(i) The C-terminal amino acid (Rj) should bea free acid, although there is some activity withthe amide. This residue must be either glycineor else a D-amino acid. The specificity for theD-amino acid is relatively low.

(ii) The R2 residue should be D-alanine. Thereis some activity with glycine, but none withother D- or L-amino acids. The strictest specific-ity occurs at this residue.

(iii) The R, residue must be an L-amino acidand substituted on its a-amino group, i.e., Xcannot be a hydrogen atom. It should alsopossess a long side chain. Thus, R, cannot behydrogen or acetyl, i.e., D-alanyl-D-alanine andacetyl-D-alanyl-D-alanine are not substrates.For some strains, namely Streptomyces albusG, K11, and R61, any additional terminalamino group on the side chain of R, (e.g., onlysine) should be acylated. For strain G, but notR61, the enzyme is also active if this aminogroup is in an a position to a free carboxyl group(e.g., as in diaminopimelic acid). In contrast,with the R39 enzyme, acylation of the aminogroup reduces the activity. For all four enzymes,substitution of the e-amino group of lysine at R,with pentaglycine gives a highly active sub-strate.The enzymatic efficiency of the four carboxy-

peptidases on the different substrates is afunction both of the Michaelis constant (Km),for binding the substrate, and the catalyticconstant Vmax. In the case of strains R39 andalbus G, good substrates bind to the enzymewith lower Km than do poor substrates. Thevalues for Vmax are less affected. For strains R61and K11, the situation is reversed. All thesubstrates bind with high Kms; the good sub-strates are distinguished by having muchgreater values of Vmax (123). These results areinterpreted to indicate that the binding surface

310

0.)0

0.-0.0C.s0C!

C!p

Co

CL0

0.

0WUt3

I

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 21: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

9 :OX III-1

H

_~~ 0

0'V

03 X

311VOL 38, 1974

t0

0

0.0

.

0.

0

0.W

cn

12,

10

40

0C

.0

1._

S.. 1.~bI 0

OD

0-

._3

._

.w .=0

-u

300V 0

o co co

Z >4 >4 >4

bd5b3

9 o o o> Z Z z

1-

0)

._4

6.;O

. _

a d t:0 0

00 I" 0 GoA VV

104C-48

cN 0_- C14V-4 m*

- 0

Go

-4- -000r

0 0R 0II

1'

1-u!t

_I¢;I__j _t_F

>,II-0_ _1

0

650.

tCZ

t (#!)T1¢1

N :

¢C6

11 .5¢

IS

++ +

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 22: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

for the R61 and K11 enzymes is relativelynonspecific, with good substrates inducing a

conformational change leading to catalytic ac-

tivity. Conversely, the R39 and albus G en-

zymes would have a much more specific bindingsurface. Actually, as shown in Table 3, thesedifferences in the range of Km and Vmax for thedifferent enzymes are not always too great.Peptide analogues of the N,N-diacetyl-L-

lysyl-D-alanyl-D-alanine function in some cases

as competitive inhibitors of the albus G and R61enzymes (164). Such studies suggest that thetwo C-terminal residues of the tripeptide func-tion to promote binding to the enzyme. Thethird residue has little effect on binding.Rather, the side chain is crucial for promotingcatalytic activity. In addition, it has beensuggested that the normal configuration of thepeptide bond in the C-terminal dipeptide dur-ing binding to the enzyme is cis. This conclusionis based on one observation: that the acetyl-racemic-cyclodiaminoadipic acid lactam acts as

a relatively good inhibitor of the R61 and albusG enzymes. This compound is an alanyl-alanineanalogue, constrained to a cis configuration byits ring formation. These conclusions ought notbe generalized to the R39 enzyme. This enzyme

was not inhibited by any of the peptide inhibi-tors examined; indeed, some of the peptideswere good substrates for the enzyme.

(i) Transpeptidase activity with amino acidacceptors: In addition to D-alanine carboxypep-tidase activity, the R39, R61, and K11 enzymes,but not the albus G enzyme, also possess

transpeptidase activity. This activity was firstshown by using N,N-diacetyl-L-lysyl-D-alanyl-D-alanine as donor and D-alanine or glycine as

acceptor (186). In the presence of acceptor,three reactions occurred. Donor was cleaved bythe carboxypeptidase activity, donor was trans-ferred to acceptor, and the transpeptidationproduct was cleaved by the carboxypeptidaseactivity. The proportion of donor converted to

transpeptidation product under optimal condi-tions was quite high, greater than 60% for eitherR61 or R39 at an acceptor to donor ratio of 100: 1(i.e., 0.15 M acceptor). At lower acceptor-donorratios, the amount of transpeptidation whichoccurred was still substantial: At 10:1, it was 55and 40% for the R61 and R39 enzymes, respec-tively. Indeed, 5% transpeptidation still oc-

curred with the R61 enzyme at a 1:125 ratio.Several factors affected the relative activities

of the enzymes as transpeptidases and carboxy-peptidases. High pH favored transpeptidation(82, 89, 92). Likewise, a 50% reduction in watercontent in the assay mixture by ethylene glycoland glycerol (65:35, vol/vol) decreased hydrol-ysis of donor tripeptide considerably, from 60 to25%, while having little effect on transpeptida-tion. In addition, as will be discussed below,high concentrations of transpeptidation accep-

tor not only enhanced transpeptidation butoften decreased hydrolysis of donor by a greateramount than could be explained by competitionbetween hydrolysis and transpeptidation.

(ii) Transpeptidase activity with dipeptideacceptors: The structure of the cell wall ofStreptomyces strains R61 and K11 differs fromthat of strain R39 (which indeed suggests thatstrain R39 may have been improperly classifiedas Streptomyces [89]). Whereas the cross-

bridge in the former strains extends throughLL-diaminopimelic acid and glycine, that in thelatter is a direct linkage via meso-diaminopi-melic acid (see Table 2). In accord with thisdifference in cell wall structure, the acceptorspecificity for the R61 and K11 enzymes differsfrom that of the R39 enzyme.Although both the R39 and R61 enzymes

could use as acceptor the amino acids glycine,D-alanine, and meso-diaminopimelic acid, onlythe R61 enzyme was active with a wide range ofcompounds (184). Dipeptides with N-terminalglycine were most active. Those with N-termi-nal D-alanine were less so. Good acceptors had a

TABLE 3. Catalytic efficiencies on different substrates of DD-carboxypeptidases from Streptomycesa

Expt and K. (mM) Range of Vmax Range of Refstrain K. (jimol/mg per h) Vmax

Expt 1Albus G 0.3-15 50 9-100 11 120R61 10-36 3.6 1.7-890 520 120

Expt 2R39 0.2-2.4 12 40-3,000 75 123K11 8-30 3.8 9-2,000 222 123

a Different sets of substrates were employed in the two experiments, so comparison between experiments isnot justified.

312 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 23: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

C-terminal amino acid which neither was toobulky nor possessed a D-asymmetric center inthe group involved in the peptide bond. Thus,the best acceptors were either glycyl-L-alanine,glycyl-glycine, or E-glycyl-L-lysine. Perhaps alittle surprisingly, a-glycyl-a'-acetyl-LL-diaminopimelic acid, the analogue of the natu-ral acceptor in the cell wall, gave four- tofivefold less transpeptidation than the aboveacceptors. A large number of other acceptorsalso functioned with the R61 enzyme. Theseincluded the lactam of meso-diaminopimelicacid, 2-amino-2-deoxyhexuronic acid, D-cyclO-serine, and 6-aminopenicillanic acid (a pooracceptor).

(iii) Transpeptidase activity with cell wallpeptides as acceptors: The use of natural cellwall peptides as acceptors for the R61 enzymehas not been reported. However, although di-peptides did not function as acceptors for theR39 enzyme, analogues of the natural cell wallpeptide acceptor were active (89). Here, speci-ficity was directed at the tripeptide sequenceL-alanyl-y-D-glutamyl-(L)-meso-diaminopimelicacid. Substitution either of the N-terminus withthe disaccharide f-1,4-N-acetyl-glucosaminyl-N-acetylmuramic acid or else of the C-terminuswith D-alanine had little effect. In contrast,amidation of the a-carboxyl on the glutamicacid or of the D-carboxyl on the diaminopimelicacid residues had profound effects. The lattersubstitution rendered the compound completelyinactive. The former converted it to the naturalcell wall tetrapeptide in this organism andcaused major changes in its influence on hydrol-ysis and transpeptidation. Transpeptidation byusing natural amidated donor as well as accep-tor has not yet been reported. However, at highconcentrations of peptide, the R39 enzyme doescatalyze dimer formation between two mole-cules of the pentapeptide L-alanyl-'y-D-gluta-myl-L-meso-diaminopimelyl-D-alanyl-D-alanine(92). This peptide differs from the one naturallyfound in the wall only in its lack of amidesubstitution on the a-carboxyl of the glutamicacid.

Multiple observations indicated that the na-ture and concentration of the acceptor affectedthe efficiency of the R39-enzyme both as acarboxypeptidase and as a transpeptidase (89).(i) With L-alanyl-y-D-glutamyl-(L)-meso-diaminopimelic acid as acceptor, increasing theconcentration of acceptor above a 9: 1 acceptorto donor ratio had no further effect on the extentof transpeptidation. It did, however, considera-bly decrease the extent of donor hydrolysis. (ii)Amide substitution of the a-carboxyl of the

glutamic acid in the above tripeptide acceptorcaused transpeptidation to occur only within anarrow range of acceptor-donor ratios (1 to 3: 1was optimal, depending on the donor concentra-tion used). Higher ratios strongly inhibited bothtranspeptidation and hydrolysis. (iii) The semi-amidated peptide dimer (Fig. 13) did notitself serve as an acceptor. However, it stronglyinhibited both the carboxypeptidase activity ofthe R39 enzyme and also its transpeptidaseactivity with glutamic-amidated tetrapeptideas acceptor.

Basically similar although less extensive evi-dence exists for such effects by acceptor on theR61 enzyme (82, 184). High concentrations ofdipeptide acceptor inhibited both transpeptida-tion and donor hydrolysis. The relative sensitiv-ity of the two processes depended on the specificacceptor examined. When diaminopimelic acidwas used as acceptor instead, no such inhibitionoccurred.

In addition to the purified soluble DD-carboxy-peptidase-transpeptidase discussed above,Streptomyces R61 also possesses a similar mem-brane-bound transpeptidase activity (60, 92).This activity differs from that of the solubleenzyme in that in the membrane it acts exclu-sively as a transpeptidase. Carboxypeptidaseactivity appears, however, upon solubilizationof the activity in 2 M urea, 10 mMethylenediaminetetraacetic acid. In addition,the membrane activity differs from that of thesoluble enzyme in the efficiency of transpepti-dation obtained with different acceptors, beingmore efficient with some substrates and lessefficient with others.

EndopeptidaseSeveral endopeptidases have been found in E.

coli (17, 98, 183). These enzymes cleave thedisaccharide-tetrapeptide dimer between theD-alanine and the diaminopimelic acid residues

L-Ala- D-Glu-(OH)

L -(OH)L-Ala- D-Glu-(OH)

DAPL D-Ala- - (amide)

DAP-(OH)

D

FIG. 13. Structure of the semiamidated peptideaffecting the activity of the DD-carboxypeptidase-transpeptidase from Streptomyces strain R39.

313VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 24: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

314 BLUMBERG AND

(Fig. 14). Such endopeptidases thus act asautolytic enzymes which antagonize the actionof the transpeptidase. At least one endopepti-dase activity in E. coli is penicillin sensitive(17); a second activity unaffected by the antibi-otic has also been reported (98).

It has been noted that the endopeptidasesubstrate formally resembles that of a D-alaninecarboxypeptidase, in that both possess a freecarboxyl group in a position alpha to the pep-tide bond which is cleaved. The free COOHgroup in the cross-linked dimer is that on theD-asymmetric center of the diaminopimelic acidresidue (Fig. 14). In agreement with this resem-blance in substrate, the B. stearothermophilus(245), Streptomyces (82, 120, 123), and E. coli(17) carboxypeptidases also possess endopepti-dase activity. The levels of activity as endopep-tidases are less than as carboxypeptidases,however. The relative enzymatic efficienciesrange from 17 to 0.15%.

Other ActivitiesThe presence in E. coli of a "penicillin

sensitive glycosidase" was also reported (98; J.V. HiSltje, Ph.D. thesis, Eberhard-KarlsUniversitat, TUbingen, Federal Republic of Ger-many, 1970). However, the degree of sensitivityof this enzyme to penicillin is not very great.

NHCOCH3 CH20OHo

0CH20H 0

HC-CH3

NHCOCH3 CH20H COOH 0IH

OH OH0 0 (L) HC-CH3

OH20H 0 NHCOCH3 00

HC-CH3 HN

CO (D) HC-COOHHN (OH2)2

(L) HC-CH3 CO

CO H2N HN

HN (D) HC-(CH2)3-CH (L)

(D) HC-COOH HOOC CO

(CH2)2 HN

CO H3C- C (D)HN NH CO

(L) CH-(CH2)3-CH (D) \

CO HOOCHN

CH3-CH (D)COOH

)STROMINGER BACTERIOL. REv.

The data show that penicillin at 1,000 U/ml(600 gg/ml) inhibits 20%; at 2,500 U/ml (1,500,gg/ml), it inhibits 40%. These values are suffi-ciently high so that the specificity of the reac-tion with penicillin is questionable. Because theenzyme apparently is very labile, bovine serumalbumin is added for stabilization. The authorsattempt to explain the relative insensitivity ofthe enzyme to penicillin in terms of a nonspe-cific interference with the action of penicillin bythe bovine serum albumin. However, the valid-ity of this explanation is dubious.Halobacterium salinarium lacks the peptido-

glycan characteristic of normal bacteria (112).Nevertheless, the organism has been reported tobe sensitive to relatively high concentrations ofpenicillin (158). This finding suggests that peni-cillin may be able, at high concentrations, toinhibit some activity other than those withwhich it is normally associated.

Number of Enzymes Corresponding to theDifferent Penicillin-Sensitive Enzymatic

ActivitiesMany organisms are now known to contain

several different penicillin-sensitive activities.E. coli has three: (i) transpeptidase assayed in acoupled, cross-linking reaction with naturalacceptor, (ii) D-alanine carboxypeptidase, and

L- Ala

D-Glu

meso-DAP

I D-Ala

I D-Ala

CH20H00 O-CUP)

AC HO0O NHCOCH3

HO-CH3CO

HN

(L) HG-CH3

00

HN

(D) HC -COOH

(CH2)2

0O

H2N HN

(D) HO-(CH2)3-OH (L)HOOC 00

HN

H30-OH (D)

001-*-

NH

(D) HC-CH3

COOH

FIG. 14. Comparison of substrates for endopeptidase and D-alanine carboxypeptidase.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 25: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

(iii) endopeptidase. In B. megaterium, likewise,three different activities are present: (i) trans-peptidase assayed in a coupled cross-linkingreaction with natural acceptor, (ii) transpepti-dase assayed in a coupled system with unnatu-ral acceptor, and (iii) D-alanine carboxypepti-dase. The presence of these multiple activitiesemphasizes the potential complexity in theinteraction of penicillin with the bacterial cell.They further raise the important question of thenumber of proteins which are responsible for thedifferent activities. Resolution of this questionis hampered by the small number of theseenzymes which so far have been purified. Oneapproach consequently has been to compare theresponse to penicillins of the different activitiesin question. Two questions have been exam-ined: is the inhibition reversible, and what isthe profile of sensitivity to different beta-lactamantibiotics?By these criteria, B. megaterium would ap-

pear to have at least five separate enzymes(Table 4). The transpeptidase activity assayedby cross-linking with natural acceptors (Table4) may be the product of two enzymes, because30% of the activity was resistant to concentra-tions of cloxacillin ranging from 0.1 to 2.5ug/ml. Likewise, the transpeptidase activityassayed with diaminopimelic acid as acceptor(Table 4) may be the product of two separateenzymes, because one-half of the activity isinhibited irreversibly by cloxacillin. Similarly,two D-alanine carboxypeptidases are present.The one active on the unnatural lysine-contain-ing substrate is penicillin insensitive. The en-zyme active on the natural substrate is penicil-

lin sensitive. The data do not indicate, however,whether this latter activity could be the same asthe cloxacillin-resistant portion of transpepti-dase assayed with natural acceptor (Table 4).

In the S. aureus transpeptidase system (154),transpeptidation is inhibited 60% by penicillinG at 0.5 ,ug/ml. The remaining synthesis isconsiderably more refractory. Half-inhibitionrequires perhaps 25 ,g/ml. This result mightimply two separate transpeptidases in this orga-nism (possibly corresponding to the two PBCsin S. aureus).

In E. coli, the transpeptidase is irreversiblyinhibited (106, 107). Moreover, the sensitivity ofthe carboxypeptidase to penicillins is at least100 times greater than is that of the transpepti-dase. These enzymes therefore are presumablydifferent. Carboxypeptidase and endopeptidasein E. coli have been suggested to be the same onthe basis of their simultaneous presence in apartially purified carboxypeptidase preparation(17). Further purification, however, suggeststhat the situation is more complex. Multipleactivity peaks were found on column chroma-tography. At least three types of enzyme werepresent, having (i) carboxypeptidase activityonly, (ii) endopeptidase activity only, and (iii)both types of activity (224).

In Streptomyces, as discussed elsewhere, thesoluble and membrane-bound transpeptidasesprobably differ. Among other evidence, the twoenzymes show dramatically different profiles ofsensitivity to various penicillins (60, 90). Like-wise, the D-alanine carboxypeptidase of B.subtilis appears to differ from the penicillinkilling site, presumably the transpeptidase (13).

TABLE 4. Transpeptidase and D-alanine carboxypeptidase activities in B. megaterium membranesa

Inhibition by Reversibility Inhibition by Inhibition byDeterminationcloxacillin of inhibition penicillin G cepholathinby cloxacillin

TranspeptidaseCross-linking with natural ac- 70% sensitive at 0.1 jg/ml Yes ND + +

ceptors30% resistant to 2.5 gg/ml ND ND

Diaminopimelic acid as accep- 88% at 0.5 gg/ml 50% 100% at 3,000 jiglml" +tor

CarboxypeptidaseAssayed on natural substratec 50% at 500 jg/ml No 100 jig/ml NDAssayed on lysine-containing Insensitive Insensitive Insensitive

substrate

a Data are from reference 240; ND, not determined.° Due to the presence of penicillinase in the membrane preparation, high concentrations of penicillin G were

required to block these reversibly inhibited activities.¢ The natural substrate was UDP-acetylmuramyl-L-ala-D-glu-meso-Dap-D-ala-D-ala. In the lysine-containing

substrate, diaminopimelic acid was replaced by L-lysine.

VOL. 38, 1974 315

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 26: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

On the other hand, multiple activities arepossessed by several purified enzymes. Thepurified Streptomyces R61, K11, and R39 car-boxypeptidases (82, 89, 92, 184, 186) and thepurified B. stearothermophilus carboxypepti-dase (245) also exhibit "transpeptidase" andendopeptidase activity.Much of the above evidence for the existence

of multiple enzymes is susceptible to differentinterpretations. In cases where activities arepartially inhibitable, the assignment of multi-ple enzymes is least convincing. Because theassays are customarily performed on heteroge-neous membrane preparations, differences inthe location of the enzyme in the membranecould result in shielding of a certain proportionof the sites from penicillin. Likewise, artifacts inthe determination of penicillin inhibition couldgenerate the appearance of separate enzymes(see later section). Moreover, if penicillin actedas an allosteric inhibitor, complete bindingmight result in only partial inhibition of ahomogeneous enzyme.Another consideration is that some change in

state or structure of one enzyme with onepattern of antibiotic sensitivities might lead toa new activity with a new pattern of interactionwith penicillins. If the active site of a transpep-tidase became accessible to water, it might beconverted to a carboxypeptidase. The kineticsof penicillin inhibition might also be altered.Such an effect has been assumed (60, 92) toaccount for the difference in penicillin sensitivi-ties of the soluble and particulate transpepti-dases of Streptomyces. However, this hypothe-sis remains unproven.To the contrary, the purified B. subtilis

carboxypeptidase closely resembles the mem-brane-bound, unpurified enzyme; penicillinsensitivities for the enzyme in the two statesagree with each other within a factor of two(233). A possibly more significant type of altera-tion could be proteolytic degradation. Becausethe molecular weight of the B. subtilis carboxy-peptidase is only 40% of that of the PBC ofhighest molecular weight (I), removal of 60% ofcomponent I could easily change the proteinsufficiently to account for the difference of up to104 in the reactivities to various penicillinderivatives.

Consequently, separate important questionsare: (i) How many distinct proteins account forthe assayable activities? (ii) How many genescode for these distinct proteins? The answer tothe second question will require mutants defec-tive in the different enzymes or structuralcomparison of the purified, distinct proteins.

Relationship of the Penicillin-SensitiveEnzymes to the PBCs

The demonstration of multiple penicillin-sen-sitive activities is in accord with the existence ofmultiple PBCs. Both lines of evidence supportthe general conclusion that at least some of themultiple enzymatic activities are real ratherthan artifacts of assay. S. aureus may have twodistinguishable transpeptidase activities. Itlikewise has two binding components. B.subtilis and B. stearothermophilus have ceph-alothin-sensitive killing sites and a cephalothin-resistant carboxypeptidase; they similarly pos-sess both cephalothin-sensitive and cephalo-thin-resistant PBCs.The binding components of E. coli have not

been thoroughly investigated. However, in un-published experiments at least two have beenshown to exist. PBCs have not been analyzed inStreptomyces.The demonstration in B. megaterium that the

cross-linking activity is reversibly inhibited bypenicillins while some other activities are irre-versibly inhibited is very important. It empha-sizes that the penicillin killing site may not be aPBC. Furthermore, it underlines the dangers ingeneralizing results from one class of bacteria toanother.

Physiological Functions of thePenicillin-Sensitive Enzymes

The physiological role of no penicillin-sensi-tive enzyme has been determined in a convinc-ing fashion. The most thorough study has beendevoted to the carboxypeptidases. The cellwalls of organisms possessing carboxypepti-dases show a relatively low degree of cross-link-ing. The peptidoglycan of E. coli or B. subtilis iscross-linked to form dimers or trimers at most.In contrast, the peptidoglycan of an organismsuch as S. aureus, which lacks a carboxypepti-dase, is much more highly cross-linked. Becausethe product of the carboxypeptidase reaction isincapable of being a donor in cell wall cross-linking, the function of the enzyme may be toregulate the degree of linkage in the wall. Analternative hypothesis is that the carboxypepti-dase activity is a side reaction of the transpepti-dase.Evidence against this latter hypothesis is the

demonstration that cephalothin kills B. subtiliswithout inhibiting carboxypeptidase. Con-versely, at least 95% of the B. subtilis carboxy-peptidase could be inhibited without killing thecells. On the other hand, although inhibition ofthe carboxypeptidase was predicted to increase

316 BAGFUOL. Rzv.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 27: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

the degree of cross-linking of the peptidoglycan,this has not been verified. Within the accuracyof the experiments (±7%), no change in cross-linking was observed at concentrations of 6-aminopenicillanic acid which caused up to 95%inhibition of the carboxypeptidase (213). Sev-eral factors may account for these results. (i)Five percent of the carboxypeptidase activitymay be sufficient for the cell. This enzyme ispresent, after all, in much larger quantities inB. subtilis than are the putative transpepti-dases. (ii) Only a small fraction of the totalcarboxypeptidase may function in cell wallsynthesis. After residing only momentarily atthe site of peptidoglycan synthesis, the enzymemight be displaced and become functionallyinactive. Because the resistance of the enzymeto penicillins should be inversely proportional toits half-life (13), the physiologically releventcarboxypeptidase would be much more resist-ant than predicted. (iii) This putative smallfraction of carboxypeptidase at the growth zonemight be inaccessible to the penicillin andthereby protected.

Technically, the accuracy of the experimentsuffers from the use of dinitrophenylation ofunblocked amino groups on the diaminopimelicacid to determine the degree of cross-linking. Ifthe effects on cross-linking were small, theymight have been obscured by experimentalerror. More accurate determination should bepossible with a recently developed technique,deamination of the free amino groups by nitrousacid (78, 153).

In E. coli, the carboxypeptidase is reversiblyinhibited by penicillins. Consequently, al-though the Ki for the enzyme is considerablybelow the concentration required to kill thecells, inhibition of the enzyme in intact cellscannot be verified. Indeed, the enzyme may notbe accessible to penicillin. Much of it is cyto-plasmic, and it is not clear either that penicillinpenetrates the cytoplasmic membrane or that itdoes not suffer interference from the high pro-tein concentration in the cytoplasm.The properties of the E. coli enzymes support

the suggestion that the carboxypeptidase couldregulate cross-linking. The enzyme is five timesmore active on the soluble precursor UDP-acetylmuramyl-pentapeptide than it is on theacetylmuramyl-pentapeptide alone (107). Onuncross-linked cell walls, even less activitycould be shown. Moreover, it has been shownthat the UDP-acetylmuramyl-tetrapeptidewhich results from the action of the carboxypep-tidase can be incorporated into linear peptido-glycan (106). Experimental questions which

would be of interest to answer are (i) whetherUDP-acetylmuramyl-tetrapeptide is present inappreciable amounts in E. coli and (ii) whetherthe ratio of UDP-acetylmuramyl-tetrapeptideto UDP-acetylmuramyl-pentapeptide is af-fected by penicillin.The membrane-bound transpeptidase assay-

able in Streptomyces R61 (60, 92) is quiteplausibly a killing site for penicillin. The en-zyme and organism differ by no more than afactor of five in their sensitivities to penicillin.Whether other potential lethal targets for peni-cillin exist in this organism is not known.The role of the soluble DD-carboxypeptidase-

(endopeptidase)-transpeptidase in Strepto-myces is more of a puzzle. It is excreted into themedium and, unlike the case of the membrane-bound transpeptidase, there is no correlationbetween its sensitivity to different penicillinsin strains R39, R61, and K11 and that of the or-ganism. Moreover, whereas certain peptideanalogues of the substrate inhibit this enzyme,they are not inhibitory for the organism (164).Likewise, no transpeptidase activity can bedemonstrated in one of the "DD-carboxypep-tidase-transpeptidases," that from Strepto-myces albus G. Consequently, the soluble car-boxypeptidase itself would appear to be distinctfrom the physiological transpeptidase.The question arises, however, whether this

enzyme could be a solubilized form of themembrane-bound transpeptidase. Inadequateevidence exists to resolve this question, al-though it has been stated without qualificationthat the two enzymes are related (60). Thestrongest evidence against the hypothesis is thedisparity in penicillin sensitivities of the twoenzymes (see preceding section). Likewise, theenzymes differ somewhat in the extent of trans-peptidation they can perform with differentacceptors (see earlier section). On the otherhand, the membrane-bound transpeptidaseupon solubilization of 2 M urea acquired somecarboxypeptidase activity. Moreover, the speci-ficity for acceptors, particularly for the R39enzyme and the kinetics of inhibition of trans-peptidation at high acceptor concentrations (82,92) strongly suggest that the carboxypeptidasespossess an acceptor binding site.Most known membrane enzymes, except for

those which have been solubilized by proteol-ysis, are water soluble only in the presence ofdetergent. If the membrane-bound transpepti-dase of Streptomyces is related to the solublecarboxypeptidase-transpeptidase found in theculture medium of these organisms, it seems astrong possibility that the latter may be a

317Voi- 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 28: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

proteolytic fragment of the former. This possi-bility is perhaps strengthened by the fact thatmaximal activity in the culture filtrates fromwhich the carboxypeptidases were prepared wasobtained only after 50 to 135 h of incubation (87,120, 123). An important, related question iswhether the release of these enzymes from thecells is by chance or whether it is the result of aspecific process. If it is a specific process, thenwhat physiological function does the enzymehave for the organism?The presence of a binding site for acceptor on

the enzyme would be compatible either with aproteolytic origin for these enzymes or with anevolutionary origin from a transpeptidase gene.Alternatively, if the physiological function ofthe enzyme is neither as transpeptidase nor ascarboxypeptidase, but rather as endopeptidase,then it would need to possess binding sites forboth the "donor" and "acceptor" portions of thecross-linked dimer substrate. Studies of thecarboxypeptidase from Streptomyces have al-ready contributed to our knowledge of enzymesof cell wall metabolism, and resolution of thequestions raised surely will significantly en-hance our understanding of this interestingbiochemical process.

If indeed there are two transpeptidase activi-ties in S. aureus (see earlier section), they maynot both be needed for cell survival. Thispossibility is one interpretation which could bedrawn from the report (180) that highly penicil-lin-resistant S. aureus mutants still bound peni-cillin at low concentrations. This binding wouldhave to be to the more sensitive of the twoputative transpeptidases, because binding wasaccomplished with penicillin at a concentrationof 0.1 gg/ml. However, it has not yet been shownthat the two S. aureus binding componentscorrespond to two transpeptidase activities, norhas the sensitivity of the separated PBCs in themutants been reported.

In B. megaterium, the transpeptidase activityassayed with diaminopimelic acid as acceptormay be essential to the viability of the organismbut, if so, it is not the most sensitive vital targetfor the action of beta-lactam antibiotics. Ceph-alosporins such as 7-aminocephalosporanic acidand cephalothin killed cells at concentration atwhich little inhibition of the transpeptidationassayed with diaminopimelic acid as acceptoroccurred (12% inhibition in the case of the7-aminocephalosporanic acid) (240). The cross-linking activity, in contrast, showed better,although not excellent, agreement.

In E. coli, the cross-linking activity has thesensitivity to ampicillin, cephalothin, and

methicillin expected for the killing site (217).Although the enzyme is 10-fold more sensitiveto penicillin G than is the whole organism, thisdifference is attributed to the relativeimpermeability to this antibiotic of the outermembrane of E. coli. Although this explanationis probably correct, it should be noted that E.coli K-12 may be more permeable to penicillinG than ampicillin (196). The physiological roleof the endopeptidase of E. coli has not beenstudied.The study of transpeptidase activities in

different organisms clarifies both their role inthe cell and their relationship to the killing site.First, multiple transpeptidases seem to exist.Consequently, the assay of one, as is the case ineither E. coli or Streptomyces, does not indicatewithout supporting evidence that it is either theonly transpeptidase activity in that organism orthat it is the killing site. If multiple transpepti-dases are found, inhibition of merely the mostsensitive enzyme may be lethal. Alternatively,inhibition of the most resistant may be neces-sary. This latter possibility would help explainresults such as those found by Park et al. (180).Resistance of the second component would beobscured by the continued sensitivity of thefirst.

Physiological Evidence for the Existence ofMultiple Transpeptidases

The biochemical results agree with the phys-iological evidence that led to the initial hypoth-esis that multiple transpeptidases exist (212,219, 225). In E. coli, low concentrations ofpenicillin are known to cause filamentation (59,84, 85). Higher concentrations may inhibit elon-gation as well (212). With a penicillin-resistantmutant of B. licheniformis, conversely, elonga-tion was inhibited by nonlethal concentrationsof penicillin (99). In addition, structural differ-ences apparently exist between cell walls fromthe ends and sides of the rod-shaped bacteriumB. subtilis as evidenced by differential sensitiv-ity to autolytic enzymes (73, 74). Moreover,intrinsic resistance to penicillin, unlike that tosuch drugs as streptomycin, is acquired only ina stepwise fashion (56, 57).

Separate transpeptidases could function forcell wall elongation, for septum formation, andfor corner formation in rod-shaped organisms.This hypothesis would provide a rationale forthe multiple PBCs (14, 15) and transpeptidaseactivities. Likewise, it would account for theinability to obtain mutants resistant in a singlestep to a high level of penicillin. If threetranspeptidases had sensitivites of x < y < z,

318 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 29: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

then a single-step mutation would only yieldmutants with resistance < y. So far, such ideasremain speculation.

Kinetics of Penicillin InhibitionA variety of kinetic mechanisms for inhibition

of different enzymes by penicillins have beendescribed. In some cases penicillin acts as areversible, competitive inhibitor. The soluble E.coli carboxypeptidase, which is inhibited in thismanner, is extraordinarily sensitive to penicil-lins (107). Representative K1 values are 1.6 x10-8 M for penicillin G, 6.4 x 10- M for ampi-cillin, and 3 x 10-6 M for cephalothin. Penicil-lins which either lacked the side chain, viz., 6-amino-penicillanic acid, or which had a hydro-lyzed beta-lactam bond. viz., penicilloic acid,still retained some activity. The Ki values ofthose compounds were 2.5 x 10-6 M and 5 x10-6 M, respectively.The second common type of inhibition is

irreversible inactivation caused by covalentbinding of the penicillin to the enzyme. Thismode of action is exemplified by the B. subtiliscarboxypeptidase (13, 233). Here, the reactionbetween penicillin and enzyme is believed tooccur in a two-step process.

K,E + P = E-P--S--EP (1)

The penicillin binds to the enzyme in a reversi-ble complex; irreversible inactivation, probablyby acylation of the enzyme, follows. For such areactionIn (active enzyme per total enzyme)

= -k3.P-t/(K, + P). (2)The values for k. and K, have been determinedfor 10 beta-lactam antibiotics (233). The sensi-tivity of the carboxypeptidase to inhibitionappeared to depend primarily on its K, for theparticular beta-lactam antibiotic. Valuesranged from 28 to 0.1 mM. In contrast the ratesof acylation, k,, were relatively constant, 42-2 x10-2 s-1. The actual numbers should be viewedwith some caution, however. The values of k,3correspond to one-half of the enzyme beingconverted from E-P to EP* in 1.7 to 35 s. Suchtimes are of the same order as those required formixing, degradation of free penicillin by penicil-linase, etc. Indeed, the data plotted to deter-mine K, and k3 do not distinguish beyond doubtbetween the postulated mechanism and thesimplistic model

E + P -EP* (3)although the two-step mechanism is in allprobability correct.

In any case, the concentrations of beta-lac-tam antibiotics customarily employed whenmeasuring rates of binding are far below theirK, values (reversible) for the enzyme. Conse-quently, the rate of inhibition can be approxi-mated asIn (active enzyme per total enzyme)

= -klPP*PP- t (4)where kapp = (kIK,). For the one-step mecha-nism (equation 3), an identical relation wouldhold, where kapp, = ka.

Briefly, equation 4 indicates that inactivationof the enzyme by acylation is proportional totime as well as to penicillin concentration,provided that the penicillin concentration issufficiently greater than that of the enzyme, sothat it remains constant over the course of thereaction. Consequently, when irreversible inac-tivation takes place, discussion of enzyme sensi-tivity merely in terms of penicillin concentra-tion is relatively meaningless. Rates of inactiva-tion are most easily determined by first bindingpenicillin to the enzyme for a fixed amount oftime, then destroying free penicillin with peni-cillinase, and finally assaying residual en-zymatic activity.Although inhibition of carboxypeptidase in

vitro is proportional to the exponential of thepenicillin concentration, inhibition of the en-zyme in growing cells is proportional to theconcentration instead (13):active enzyme per total enzyme

(5)1 + kaps * PP generation time.

The explanation is that in growing cultures theenzyme is being synthesized exponentially aswell as inactivated exponentially. The combina-tion of the two effects yields a nonexponentialterm dependent on the ratios of the relative rateconstants.The kinetics of irreversible inactivation have

been demonstrated only with the carboxypepti-dases from B. subtilis and B. stearothermo-philus (13, 233, 245). Here, it was also confirmedthat the inactivation of the enzyme was ac-companied by the physical binding of penicillin(15, 232). Inhibition of a number of other en-zymes, including the E. coli (106) and M. luteus(152) transpeptidases, is not reversed by peni-cillinase; but the kinetics of binding have notbeen studied, and there is no direct proof forthe assumption that they are acylated.The carboxypeptidases isolated from the pen-

icillin-sensitive strains of Streptomyces R61and K11 were inhibited by penicillin in acompetitive manner (120, 123) (but see below).

319VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 30: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

However, that from strain R39 appeared to havebeen inhibited in a partially competitive fash-ion (123). The kinetics were mistakenly termed"noncompetitive" in the paper. As plotted, thedata in fact show that both Vmax and Km werealtered in the presence of penicillin. Moreover,additional data points would be desirable, be-cause lines indicating strictly competitive inhi-bition might be drawn through the points illus-trated. If in fact both Km and /max were altered,the binding of penicillin at the substrate sitemight distort the enzyme so as to result in adecrease in its catalytic efficiency (Vmax). Sucha phenomenon has been seen in the inhibition ofpenicillinases by some penicillins. Other possi-ble problems are that approximately compara-ble distortions in kinetics would have beenobtained if penicillin were causing irreversibleinactivation of the enzyme (cf. 233), althoughthe possibility of such inactivation was renderedfairly unlikely by extensive controls. Likewise,reversible inactivation would lead to similardistortions, provided that release of bound peni-cillin occurred slowly. Although the rate ofrelease has not been reported for the R39enzyme, for the R61 enzyme it is very slow, witha half-time of several hours (see below). Somehydrolysis of penicillin due to instability underthe conditions of assay could have led to adistortion of the kinetics. Consequently, thenature of the inhibition of the R39 enzyme mustbe considered unresolved. However, confirma-tion that inhibition is partially competitive andan explanation of the phenomenon would be ofmajor importance in shaping thinking about thenature of the interaction of penicillin with itsreceptors.An additional type of inhibition which may

occur is irreversible inhibition that is not ac-companied by irreversible binding of penicillin.After inhibition of the D-alanine carboxypepti-dase of E. coli by penicillin G, only 66% of theenzymatic activity was restored by addition ofpenicillinase (107). Although the authors sug-gest that penicilloic acid might account for theresidual inhibition, this does not appear to bethe case. The penicilloic acid concentrationshould only be 0.008 jg/ml; 50% inhibition ofthe enzyme by penicilloic acid only occurs at 2jig/ml, a concentration 250-fold higher. Like-wise, because the E. coli carboxypeptidase doesnot bind penicillins irreversibly (114), modelsinvolving partial binding or hydrolysis are un-likely. One alternative which has precedent in arelated system is that the presence of penicillinmay convert the enzyme to a particularly labilestate, in which it then proceeds to denature.

Such a phenomenon occurs when the S. aureuspenicillinase is exposed to methicillin (35, 36,96). Moreover, as has been shown recently,penicillin binding renders the StreptomycesR61 enzyme more readily denaturable (92).

Irreversible inhibition is not always irreversi-ble (see earlier section). Release of penicillin Gfrom the B. subtilis D-alanine carboxypeptidaseat 37 C occurs with a half-time of 200 min (16).At 55 C, release from the B. stearothermophiluscarboxypeptidase is even faster (half-time of 10min). Because the incubation period for assay ofenzymatic activity is often long compared withthese release times, misleading kinetic resultscan be obtained. In the case of the B. subtiliscarboxypeptidase, 6-aminopenicillanic acidcaused partial inhibition at low concentrations.Very much higher concentrations did not in-hibit the remaining activity (115). This resultsuggested the existence of two carboxypepti-dases, one sensitive to 6-aminopenicillanic acidand the other resistant. In reality, there is onlyone enzyme. The "resistant fraction" was anartifact of release of 6-aminopenicillanic acidduring the course of the assay.Such artifacts do not account for the partial

reversibility of the inhibition by penicillin of thetranspeptidase activity assayed with diamino-pimelic acid as acceptor in B. megaterium.Unlike the above example, the proportion ofactivity in the B. megaterium system which wasreversible was independent of the length of timefor which the penicillin-inhibited enzyme wasexposed to penicillinase before assay (240). Ifhydrolysis were occurring, a greater proportionof the activity would appear to be reversible atlater times.

In B. subtilis, the kinetics of penicillin bind-ing to the PBCs I, IV, and V were thosepredicted for irreversible binding (15). How-ever, the binding curve of component II wasbiphasic for certain penicillins, viz., 6-aminopenicillanic acid and penicillin G. Thereason for this behavior is not yet known. Atrivial explanation is that there were actuallytwo proteins in the component II peak obtainedon sodium dodecyl sulfate polyacrylamide gels.However, this possibility seems unlikely in thatthe components would have needed identicalmolecular weights and rates of binding for anumber of penicillins, and they would have hadto be present in equal amounts. Other possibili-ties are either negative cooperativity in penicil-lin binding or else a difference in the accessibil-ity to these penicillins unique to component II.In any case, the behavior of component IIindicates that an electrophoretically homogene-

320 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 31: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

ous protein may bind penicillin at two separaterates. Consequently, a biphasic penicillin bind-ing curve is not necessarily evidence for multi-ple proteins which bind penicillin or are inhib-ited by it.Considerable care must be taken to exclude

the possibility of acylation in "competitively"inhibited enzymes. Because covalent bindingcan be reversed, the actual reaction scheme is

E+P EP+E + P (6)

where E = enzyme, P = penicillin, EP =

penicilloylated enzyme, and P* is the unidenti-fied produce of release. Provided that P >> E (sothat the decrease in the concentration of P as itis converted to P* can be neglected), then in thesteady state

(E)(P) = 2= K,. (7)

(EP) k,

An apparent K, can thus be determined. For theB. subtilis carboxypeptidase at 25 C (13, 16),the values for penicillin G are k, = 1.1 x10'3/min, k, = 4 x 104/M/min, K,* = 2.7 x10- 8 M: for 6-aminopenicillanic acid, k2 = 5.5 x

10-3/min, k1 = 7.6 x 102/M/min, K,* = 7.2 x

10-6 M. These values of K,* closely resemblethe values of K, for reversibly inhibited carboxy-peptidases. Moreover, although the B. subtilis(233) and B. stearothermophilus (16) carboxy-peptidases are acylated, the kinetics of inhibi-tion are competitive with respect to substrate.Thus, the existence of competitive kinetics inthe absence of supporting data does not sufficeto distinguish between mechanisms involvingcovalent and noncovalent inhibition.The necessity for caution is emphasized by

the behavior of the Streptomyces R-61 carboxy-peptidase. Penicillin binds to this enzymeslowly (accompanied by a conformationalchange in the enzyme). Penicillin is also re-leased very slowly (half-time of 145 min) (92,163). The reaction scheme is

E+P= =EP KD= * (8)

At 25 C, in 10 mM sodium phosphate, pH 7.0,kt = 1.08 x O6/M/min, k, = 4.8 x 10- 3/min, KD= 4.5 x 10- 9 M. These values closely resemblethose for an acylated carboxypeptidase, thatfrom B. subtilis. Demonstration that acylationin fact is not occurring is thus imperative. Ispenicillin released from the Streptomyces R61carboxypeptidase as intact penicillin or as a

rearrangement product, as takes place with theB. stearothermophilus enzyme? Is bound radio-

active penicillin immediately released by so-dium dodecyl sulfate?

Models for the Interaction ofPenicillin withIts Target

A number of models have been proposedwhich suggest an analogy between penicillinand various portions of the peptidoglycan. Be-fore it was known that penicillin inhibitedcross-linking, Collins and Richmond proposedthat penicillin G was an analogue of N-acetyl-muramic acid (43). The subsequent demonstra-tion of the effect of penicillin on transpeptida-tion deprived this hypothesis of much of itsattraction. Subsequently, Wise and Park ar-gued that penicillin was an analogue of theL-alanyl-y-D-glutamyl portion of acetylmuramyl-pentapeptide (243). However, this hypothesiswas rendered unlikely by the finding that thea-carboxyl of the glutamate in the pentapeptideof S. aureus was amidated.Tipper and Strominger (228) proposed that

penicillin was an analogue of the terminalacyl-D-alanyl-D-alanine in the pentapeptidechain (see Fig. 7). More specifically, the fixedring structure common to penicillins was hy-pothesized to be a structural analogue of thatconformation of acyl-D-alanyl-D-alanine whichbinds to the active site of the enzyme. In thismodel, the reactive CO-N bond in the highlystrained beta-lactam ring would correspond inposition to the peptide bond cleaved duringtranspeptidation and would be ideally situatedto react with the acyl acceptor site of theenzyme. The two rings of the penicillin nucleusare in different planes, and it was furtherhypothesized that penicillin might more strictlybe an analogue of a transition state duringcleavage of the normal peptide substrate, atwhich the planar double-bond character of thepeptide bond had deteriorated into a nonplanarsingle-bonded state. Reaction of the transpepti-dase with the beta-lactam ring of penicillinwould result in formation of a penicilloyl en-zyme complex analogous to the postulated acylenzyme intermediate and would provide a ra-tional explanation for the acylation of PBCs bybeta-lactam antibiotics. The idea that penicil-lin was a transition state analogue was furtherexplored by Lee (118).

Several lines of evidence support this model.(i) Substrate protects the D-alanine carboxy-peptidase of B. subtilis against inactivation bypenicillin (233). (ii) The enzyme is inhibited bysulfhydryl reagents (e.g., DTNB, mercurials),and either substrate or bound penicillin pro-tects one sulfhydryl group on the native enzyme

321VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 32: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

against these reagents. (iii) The D-alanine car-boxypeptidase of B. stearothermophilus ap-pears to catalyze the transfer of bound penicil-lin to hydroxylamine (16). Such a reaction couldbest be explained if penicillin were bound at thecatalytic site of the enzyme.Although the most likely explanation is that

both penicillin and substrate are bound at thecatalytic site on the enzyme, containing anactive sulfhydryl group, the data are also com-patible with an alternative model in which thesulfhydryl is at an allosteric site. In this lattercase, competition would be due to alternativeconformational states of the enzyme (catalyti-cally active and inactive) induced by substrateor penicillin. Experimentally, these possibilitiescan be distinguished by demonstrating thatsubstrate and penicillin are bound at the samesite (e.g., by affinity labeling or by isolation ofthe hypothetical acyl enzyme intermediate) orby crystallographic studies of the enzyme withsubstrate or inhibitor bound, both difficultobjectives.

In addition, the substrate analogue hypothe-sis predicted that 6-methyl-penicillins and 7-methyl-cephalosporins, which would moreclosely resemble the D-alanyl-D-alanine, mightbe even better antibiotics than the unmodifiedmolecule (228). This prediction has not beensupported by experiment either in vivo (18, 102)or in vitro (240). However, the 7-methoxy sub-stituents on cephalosporins result in highlyactive antibiotics which also are active in vitroas inhibitors of the penicillin-sensitive enzymes(102, 150, 151, 214). Thus, before the relation-ship of these data to the proposed mechanismcan be evaluated, it is important to know whytwo bulky substituents on the beta-lactam ring(methyl and methoxy) have such different ef-fects on activity. There are presently insuffi-cient data relating to this point to evaluate itfurther. One important consideration may bethe effect of different substituents on the reac-tivity of the beta-lactam ring.The alternative hypothesis that penicillin

may exert its control at an allosteric site on theenzyme is based on studies of the StreptomycesDD-carboxypeptidase-transpeptidases. Severalexperimental observations led to this hypothe-sis. (i) Inhibition of the R39 enzyme by penicil-lin is partially competitive rather than competi-tive (123). (ii) The binding of penicillin to theR61 enzyme, monitored fluorimetrically, canbe competed with neither acceptor nor donor(92). (iii) Penicillin apparently binds to the R61enzyme under conditions where the enzyme ispartially denatured and inactive (viz., in 3.6 M

guanidine-hydrochloride) (163). (iv) The albusG carboxypeptidase is insensitive to inhibitionby penicillins, but it is inhibited by syntheticsubstrate analogues (122, 164).None of the above arguments is persuasive.

(i) The unusual kinetics reported for inhibitionof the R39 enzyme may have several explana-tions, which have been discussed above. (ii) Theclaim that transpeptidation donor cannot com-pete for binding with penicillin to the R61enzyme, determined by physical measurements(92, 163), is inconsistent with the kinetic mea-surements that donor and penicillin do compete(120). It is not clear from the published datawhether the amount of substrate present (5 xKi) would have been sufficiently high to com-pete with inhibitor. (iii) The binding of penicil-lin to inactivated enzyme does not indicate thatthe substrate and drug-binding sites are differ-ent. Rather, it implies that sufficient damage tothe enzyme has occurred so that certain cata-lytic functions are lost. Substrate may also bindto the partially denatured enzyme. (iv) Thefailure of penicillin to inhibit the S. albus Gcarboxypeptidase suggests only that this en-zyme has a stricter binding site than the otherenzymes. In fact it has the lowest Km forsubstrate of all of the Streptomyces carboxy-peptidases. An analogous case may be dihy-dropteroate synthetase in sulfonamide-sensitiveand -resistant bacteria. Sulfonamides are struc-tural analogues of p-aminobenzoic acid, a sub-strate for this enzyme. In the resistant orga-nisms, sulfonamides, the substrate analogues,bind to the enzyme much more poorly, but thebinding of substrate itself is unaffected (172,173). There are some obvious differences insome of the bond angles in substrate andpenicillin. If penicillin is an analogue of thetransition state of the terminal D-alanyl-D-ala-nine, and not of the ground state, it might fail tobind to a substrate binding site which requiredan unusually specific fit. Inhibition of thisenzyme by synthetic substrate analogues whichclosely resemble substrate is compatible withthis concept. It indicates that certain peptideanalogues, such as substrate itself, can fit thebinding site of this particular enzyme, some-thing which the putative transition state ana-logue cannot do. On the other hand, penicillinbinds to the penicillin-sensitive o-alanine car-boxypeptidases of E. coli and Streptomycesstrains R61, K11, and R39 much better thandoes substrate: the values of K, for penicillin Gare 1.6 x 10-8M, 7.5 x 10- M, 6 x 10"8M, and1.4 x 10-8 M, respectively. The values of Km are0.6 mM (UDP-acetylmuramyl-pentapeptide as

322 BAcNrRIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 33: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

substrate), 12 mM, 11 mM, and 0.8 mM (N,N-diacetyl-lysyl-D-alanyl-D-alanine as substrate),respectively. This result, that inhibitor bindsmuch better than substrate, is characteristic oftransition state analogues (124).The behavior of the D-alanine carboxypepti-

dase from B. stearothermophilus membraneswas interpreted to provide further support forthe allosteric model of penicillin action (11).First, penicillin gave only partial inhibition ofthe membrane activity, 75 to 80%. This partialinhibition was reflected in hyperbolic plots ofKm/V. Secondly, certain treatments, e.g., stor-age at 4 or -20 C, led to apparent partialdissociation of penicillin sensitivity from en-zyme activity. In general, the partial loss ofenzymatic activity was accompanied by dispro-portionately greater loss of the ability of penicil-lin to inhibit the enzyme. For example, storageat -20 C for 4 weeks led to 61% loss of activity,but only 10% of this remaining activity could beinhibited by penicillin. Consequently, the abso-lute penicillin-resistant activity had increasedfrom 20% of the original total to 34.6%. Simi-larly, after storage for 3 weeks at 0 C, themembrane activity had decreased to 43%. Itremained 80% inhibitable by penicillin. How-ever, if after storage the membranes were resus-

pended and sedimented by centrifugation, thenthe penicillin sensitivity was lost.The authors interpreted these complicated

results to indicate that penicillin bound to "atmost only a portion of the catalytic site," or,alternatively, that it acted as an allostericinhibitor. The possibility that there were twoD-alanine carboxypeptidases, which differed instability at 4 and -20 C and one of which was

penicillin insensitive, was considered, althoughdeemed unlikely. The authors argued that thismodel could not explain the increase in the totalpenicillin-resistant activity which occurredupon storage.The behavior of the purified D-alanine car-

boxypeptidase of B. stearothermophilus (albeitfrom a different strain) provides no evidence fordissociation of activity and inhibition (245), andthis purified enzyme is totally inhibitable bypenicillin. Consequently, the presence of twocarboxypeptidases and possible artifacts associ-ated with the membrane location of the unpuri-fied enzymes may account for the anomalousresults reported (11). Several D-alanine carboxy-peptidases are known to occur in both E. coli(224) and B. megaterium (240), and in the lattercase one of these is penicillin insensitive.

If penicillin is a substrate analogue, then theactive sites of carboxypeptidases can be divided

into two classes according to whether acylationoccurs or does not occur. (i) In the one studiedexample of the first class (B. subtilis), penicillinforms a weak reversible complex (K, = 10-2 to10-4 M) with the enzyme. Significant inhibitionresults because of acylation. (ii) In the secondclass, penicillin binds very tightly (K1 = 10-8M), but does not acylate. The large difference inthe values of K, in the two cases suggests onepossible explanation for the occurrence or lackof acylation. In the first class, it is possible thatthe penicillin must be distorted to bind to theenzyme, hence the high K1. This strain energysufficiently activates the beta-lactam bond toinduce acylation of an active group at thebinding site of the enzyme, probably but notnecessarily the same one involved in formationof the putative "acyl enzyme intermediate." Inthe second class, penicillin might fit the config-uration of the active site much better. Becausethe molecule remains undistorted, acylationwould not occur. Alternatively, of course, thesetwo types of carboxypeptidases may haveevolved with two different mechanisms, thesecond class having no acylation in either itsmechanism or its inhibition. In other proteases,they are obviously examples of those whichutilize covalent intermediates and those whichdo not (21).Thus, there are at the present time two

theories for the mode of inhibition of bacterialenzymes by penicillin-the substrate analoguehypothesis and the allosteric inhibitor hypothe-sis, neither of which is supported by substantialdata. It has been shown that some bacterialenzymes at least are inhibited by an acylationmechanism and that some catalytic activities(transfer of pencilloyl to hydroxylamine andrelease of penicillin in an altered form) occur atthis site, but it has not been decisively shownthat this is the same site at which catalysis ofsubstrate occurs. The protection by substrateagainst penicillin inactivation and the protec-tion of a sulfhydryl group from inactivation bysubstrate are most simply interpreted by thismodel, but these facts are also compatible withother models. The possibility that penicillinmay act at an allosteric site derives in part fromthe idea that there must be regulatory mech-anisms in cell wall synthesis and that penicillincould be an analogue of some natural regulator(19). It seems likely that the major regulation ofcell wall synthesis in most bacteria occurs atsome earlier step in the pathway (such as areaction involved in the synthesis of UDP-acetylmuramyl-pentapeptide), but that doesnot exclude the possibility that there may be

323VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 34: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

some modulation of cell wall synthesis at theterminal reactions, which are the reactionsinhibited by penicillin. In fact there is now

known a multiplicity of enzymes which areinhibited either reversibly or irreversibly. Itwould not be surprising to discover that the siteof inhibition, i.e., the substrate site or an

allosteric site, also was different in differentcases.

MECHANISM OF KILLING BYPENICILLINS

Mutant AnalysisOne way of better understanding the interac-

tion of penicillin with its targets and the role ofthe enzymes involved in cell wall synthesis is bythe examination of mutants. Two basic ap-proaches have been attempted. In the first,penicillin-resistant mutants have been isolated.The findings in gram-positive organisms so farhave been confusing and hard to interpret (seeearlier section). With gram-negative organismssuch as E. coli, penicillin resistance has gener-ally been attributable either to alterations inpenicillinase or permeability (19, 20, 29-31, 71,72, 125, 126, 159, 166-170). Mutants in thepenicillin-sensitive enzymes themselves havenot been found.The second approach has been the isolation of

conditional lysis mutants (28, 34, 95, 130-132,146, 156, 205, 241). The rationale here is thatmany of these should be temperature sensitivein the enzymes required for cell wall synthesis.The results have been generally disappointing.Almost all of the mutants isolated have beenmutated in the enzymes involved in synthesis ofthe soluble precursors of the peptidoglycan.However, one mutant in E. coli appears to havea thermolabile particulate D-alanine carboxy-peptidase (156). The soluble carboxypeptidaseshows no change. A second mutant appearsdefective in an enzyme involved in the forma-tion of lipid intermediate (156). More detailedcharacterization of the D-alanine carboxypepti-dase mutant should prove illuminating. Severalother mutants, which accumulate UDP-acetyl-muramyl-pentapeptide and in which no en-zymatic defect has been found so far, may alsobe of considerable interest.A related, but more direct, approach for

isolating mutants defective in the penicillin-sensitive enzymes of E. coli has recently beenattempted (108). Because penicillin-resistantmutants were generally caused by alterations inpermeability or penicillinase, whereas lysis mu-tants were generally altered in the wrong en-

zyme, a double-selection method was used.Mutants were isolated which were simultane-ously penicillin-resistant at low temperatureand lytic at high temperature. The reasoningwas that a certain proportion of the mutationswhich rendered vital penicillin-sensitive en-zymes resistant would simultaneously renderthe enzymes temperature labile. Most of themutants which became resistant due to changesin permeability or penicillinase would presum-ably not lyse and could be discarded. By meansof this technique, 500 mutants were isolated. Ofthese, a number possessed either hypo- orhyper-cross-linked peptidoglycan. Furthercharacterization of these mutants is in progress.

Evidence That Penicillin Kills by InhibitingCross-Linking

Penicillin is believed to kill bacteria by inhib-iting peptidoglycan cross-linking. However, theonly thorough studies that this is the mecha-nism of killing in vivo have been carried out inS. aureus (228, 229, 243). Here, good correlationwas shown between killing by penicillin andinhibition of cross-linking. Likewise, penicillininhibits cross-linking of B. megaterium cell wallin vivo (78). Whether this inhibition correlateswell with killing has not yet been reported.

In the presence of penicillin, Proteusmirabilis is converted to spherical forms (L-forms) which are viable in the presence ofsucrose. These L-forms are able to grow andmultiply and initially can revert to bacilli uponremoval of penicillin. The peptidoglycan com-position of these penicillin-induced L-forms ofP. mirabilis was similar in composition to thatof the wild type (138, 139). However, whereasthe L-form cell wall was completely brokendown by lysozyme, that of the parent rod wasonly partially degraded. This result led H. H.Martin to suggest that "penicillin ... preventedthe formation of certain cross-linkages withinthe mucopolymer which are indispensable forthe establishment of shape and mechanicalstability." Later studies from the same labora-tory cast doubt on this conclusion (109, 137).Incomplete digestion by lysozyme, rather thanthe presence of additional cross-linking, wasfound to account for the reduced breakdown.Moreover, in both parent and L-form, 32 to 34%of the diaminopimelic acid residues in the cellwall were susceptible to dinitrophenylation, i.e.,there was no difference in cross-linking. If theeffect of penicillin on P. mirabilis is due to aneffect on cross-linking, it would have to be dueto an effect on an extremely localized region ofthe wall, e.g., an effect on the enzymes responsi-

324 BACrERIOL. RWv.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 35: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

ble for the synthesis of the corners of the rod (cf.earlier section). Likewise, no change in thecross-linking of the cell walls of E. coli grown inthe presence of penicillin was observed in theone case where this has been studied (212).

In view of the great differences which existbetween bacteria both in their PBCs and in theenzymes which they possess, which are sensitiveto penicillin, generalization of the results ob-tained in S. aureus is obviously not justified. Invitro and in vivo studies in the bacilli and invitro studies in E. coli are consistent with thelethal action of penicillin being inhibition ofcross-linking. However, it is possible in thesecases that the inhibition of cross-linking is asubsidiary reaction which occurs in addition tothe actual action of the antibiotic.A still unexplained effect of penicillins on

peptidoglycan synthesis in B. megaterium couldbe a clue to an additional mechanism (240).When in vitro peptidoglycan synthesis wasinhibited by concentrations of cloxacillin justsufficient to prevent cross-linking, two uniden-tified lysozyme digestion products were formedwhich possessed mobilities greater than those ofknown disaccharide-peptides. The amountsproduced were very large, up to 65% of the totalpeptidoglycan synthesized. At only fivefold-higher concentrations of cloxacillin, the amountof these products decreased dramatically to20%o. What relation these products bear to thekilling action of cloxacillin is not known.

Killing by Amidino PenicillinIndeed, a new class of penicillins, amidino

penicillins, have a mode of killing different fromthat of normal penicillins. The basic structureof this class of derivatives is shown in Fig. 15.The distinguishing feature is the replacementby an amidino linkage of the usual acylaminolinkage attaching the side chain. The effect ofthe drug was striking in that gram-negativeorganisms such as E. coli were up to 100 timesmore sensitive than were such gram-positiveorganisms as S. aureus or B. subtilis. In the caseof a typical penicillin derivative, ampicillin, theratio is just the opposite. Furthermore, E. colitreated with low concentrations of amidinopenicillin formed large spherical bodies whichslowly lysed, regardless of whether sucrose waspresent or not. Ampicillin-treated cells, in con-trast, form small spheroplasts which are pro-tected by sucrose, but which lyse instantly if theosmotic support is removed (133, 148): At highconcentrations of the amidino penicillin, E. coliis lysed in the same manner as by ampicillin.

It thus appears that amidino penicillin func-

tions poorly as a penicillin which inhibits cross-linking. However, there must exist some otherenzymatic activity in gram-negative organismswhich is lacking in gram-positive organisms andwhich is exquisitely sensitive to the amidinopenicillin. The effects on cell morphology makeit likely, although not certain, that the inhib-ited activity is related to envelope synthesis.One possible candidate might be the enzymewhich attaches the lipoprotein (22-27, 100, 101,104, 141) to the peptidoglycan, presumably by atranspeptidation.

Recent work lends credence to this model(145, 178). E. coli transpeptidase, carboxypepti-dase, and endopeptidase were not inhibited bylethal concentrations of amidino penicillin.High concentrations of amidino penicillin hadno effect on the binding of ["4C ]penicillin G toE. coli membranes. Mutants resistant to theamidino penicillin showed no cross-resistance tonormal penicillin.

Transfer of Penicillin Side ChainStudies of the binding to gram-positive orga-

nisms of penicillins labeled in either the sidechain ("4C) or nucleus (35S) indicated that bothportions of the molecule were bound to thePBCs (see earlier section). However, severalpapers suggest that the problem should bere-examined in gram-negative organisms. E.coli possesses an amidase which cleaves the sidechain from the penicillin nucleus (5, 39-42,192). The activity of this amidase may beconsiderable, up to fourfold greater than thatcatalyzed by the penicillinase of the organism.A correlation has been reported between theactivity of penicillins against E. coli and thesusceptibility of the penicillins to the amidase(110); the highly active penicillins were unusu-ally good substrates. Under appropriate condi-tions, the E. coli amidase can synthesize amidebonds either directly or via a transpeptidation.Such an enzyme is therefore capable of transfer-ring the penicillin side chain to amino acceptorsin the cell. Such a reaction may actually occur.The binding of [4C ]penicillin G to P.

mirabilis was little inhibited by 6-aminopenicil-lanic acid (211). In contrast, a structural ana-

CH2 - CH2-CH2\ S ,CH3I~~"zN-CH=N-CH-CH CCUN2- CH2-CH2

CO-N CH

CO2HFIG. 15. Structure of the amidino penicillin

FL-1060, with which the studies described in the textwere carried out.

325VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 36: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

logue to the side chain, phenacetylglycine, pre-vented binding when present in only 10-foldexcess. Moreover, although phenacetylglycineper se had no effect on cell growth, it preventedformation of spheroplasts in penicillin-treatedcells. This result should be confirmed andextended.

CONTROL OF CELL WALL SYNTHESISSynthesis of the peptidoglycan is coupled to

that of the other macromolecules of the cell. Itwould therefore be expected to involve compli-cated regulatory pathways. Although this regu-lation is essentially not understood, the actionof penicillin could be intimately intertwinedwith these regulatory mechanisms. As alreadydiscussed, penicillin causes the accumulation ofUDP-acetylmuramyl-pentapeptide in some(but not all) organisms. An undefined mecha-nism must therefore exist by which the enzymesresponsible for utilizing this nucleotide aremodulated either by the degree of cross-linkingwhich occurs or else directly by the penicillin.

In a phenomenon which is quite closely re-lated, the amount of linear peptidoglycan whichbacteria will synthesize is strongly affected bythe action of penicillin. In such organisms as S.aureus, low concentrations of penicillin lead tosynthesis of uncross-linked peptidoglycan. Athigher concentrations, peptidoglycan synthesisvirtually ceases (229). Such mechanisms are notuniversal. In P. mirabilis, peptidoglycan syn-thesis continues independent of the antibioticconcentrations (138). Such effects of penicillinon the net synthesis of peptidoglycan are re-flected in in vitro assays. In the case of E. coli,methicillin and ampicillin inhibit total peptido-glycan synthesis in vitro; penicillin G does not(105). With M. luteus, penicillin G causesdramatic inhibition of synthesis (152). An unan-swered question is whether penicillin mightsometimes kill by inhibiting cell wall synthesisby such a control mechanism. If so, the anoma-lous results on the effect of penicillin on cross-linking in gram-negative organisms might beexplained.A third interesting example of control is the

so-called zone phenomenon described by Eagleet al. (61, 62, 68). For some organisms, but notothers, the rate at which penicillin kills the cellsis decreased if the concentration of penicillin isincreased beyond its optimal level. Two plausi-ble explanations for this phenomenon exist. (i)Because synthesis of uncross-linked cell wall atlow penicillin concentrations may be necessaryfor cell lysis, S. aureus H may be protected athigh concentrations of penicillin because pepti-

doglycan synthesis is totally inhibited. (ii) Bac-teria possess multiple PBCs, which vary in theiraffinities for penicillin. Inhibition of one compo-nent might lead to cell death, whereas inhibi-tion of a less sensitive binding component,which might possess an antagonistic activity,would reduce the rate of killing. The B. subtilissystem could provide an example of this secondmechanism. At low penicillin concentration, thetranspeptidase (presumably components I, II,and/or IV) is inhibited; higher penicillin con-centrations inhibit the carboxypeptidase (com-ponent V) as well. If the action of the carboxy-peptidase is indeed to limit the degree ofpeptidoglycan cross-linking, then its inhibitionmight well reduce the rate of bacteriolysis.

In some other organisms the antagonisticactivity inhibited by penicillin at the higherconcentration might be an autolysin such as theendopeptidase. Such an antagonistic actionbetween autolysins and the cell wall syntheticenzymes has been clearly demonstrated inPneumococcus (230). In this organism, the nor-mal autolytic activity could be suppressed ifcells were grown in the presence of ethanola-mine rather than choline. Under such condi-tions, the rate at which penicillin killed cellswas reduced 10-fold. Likewise, mutants of B.subtilis and B. licheniformis have been isolatedwhich are highly deficient in autolytic activity(79). These mutants are lysed by two inhibitorsof cell wall synthesis, vancomycin and cycloser-ine, only at concentrations 10- to 20-fold higherthan those required to lyse the wild type (198).An extremely interesting finding, which

should be pursued vigorously, is the demonstra-tion of mutant strains of Staphylococcus (10)and Pediococcus cerevisiae (237, 238, 242)which are penicillin dependent. Such organismsmay have resulted from antagonistic, penicillin-sensitive activities becoming sufficiently out ofbalande during mutation to penicillin resistanceso that partial inhibition of one activity becamenecessary for cell survival. Other models can ofcourse be visualized, and determination of theactual explanation may have great importancefor understanding the interaction of penicillinwith the bacterial cell.

Induction of penicillinase synthesis may offera promising and so far relatively unexploitedapproach to the study of regulation of cell wallsynthesis. Irreversibly bound penicillin in B.cereus (6, 53, 58, 187, 188, 189, 191) inducespenicillinase synthesis. Likewise, penicillinaseis induced in S. aureus (103) in the presence ofpenicillin. Either the physical binding of thepenicillin itself or perhaps an effect of penicillin

326 BACTEIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 37: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

in inhibiting peptidoglycan synthesis leads totransfer of information to the cytoplasm andactivation of penicillinase synthesis. Althoughthe entire receptor-transmitter complex mightbe components of the penicillinase system,another possibility is that some genetic ele-ments involved in control of cell wall synthesismay also be involved in regulation of penicillin-ase production (37, 38).

SUMMARYGreat strides have been made in the past 30

years in understanding the mode of action ofpenicillin. Its possible targets have been de-fined. Methods have been developed for isola-tion of the different penicillin-sensitive activi-ties. Several purified penicillin-sensitive en-zymes are presently available in large amounts.Future research will most likely focus on three

issues. How does penicillin actually interactwith its receptors? What are the in vivo func-tions of the different penicillin-sensitive en-zymes? How does regulation of peptidoglycansynthesis occur, and what role does penicillinplay in this process? Investigation of the modeof action is currently in a most exciting stage.Conflicting possibilities and hypotheses compelcritical analysis. Technical knowledge hag pro-gressed to a level where these hypotheses can betested adequately. The combined wealth ofideas and methods should promote rapid prog-ress.

LITERATURE CITED1. Anwar, R. A., P. M. Blumberg, and J. L.

Strominger. 1974. Penicillin binding compo-nents in Bacillus subtilis during sporulation.J. Bacteriol. 117:924-925.

2. Araki, Y., A. Shimada, and E. Ito. 1966. Effect ofpenicillin on cell wall mucopeptide synthesisin an Escherichia coli particulate system.Biochem. Biophys. Res. Commun.23:518-525.

3. Araki, Y., R. Shirai, A. Shimada, N. Ishimoto,and E. Ito. 1966. Enzymatic synthesis of cellwall mucopeptide in a particulate preparationof Escherichia coli. Biochem. Biophys. Res.Commun. 23:466-472.

4. Archibald, A. R., J. Baddiley, and N. L. Blum-son. 1968. The teichoic acids. Advan. Enzy-mol. 30:223-253.

5. Arcos, J. M., M. C. Ruiz, and J. D. Migica.1968. Beta-lactamase and penicillinacylasecoexist in Esclerichia coli. J. Bacteriol.96:1870-1872.

6. Aten, R. F., and R. A. Day. 1973. Penicillin-binding component of Bacillus cereus. J. Bac-teriol. 114:537-542.

7. Baddiley, J. 1968. Teichoic acids and the molec-

ular structure of bacterial walls. Proc. Roy.Soc. Ser. B Biol. Sci. 170:331-348.

8. Baddiley, J. 1970. Structure, biosynthesis, andfunction of teichoic acids. Accounts Chem.Res. 3:98-105.

9. Baddiley, J. 1972. Teichoic acids in cell wallsand membranes of bacteria, p. 35-77. In P. N.Campbell and F. Dickens (ed.), Essays inbiochemistry, vol. 8. Academic Press, NewYork.

10. Barber, M. 1953. Penicillin-resistant and peni-cillin-dependent staphylococcal variants. J.Gen. Microbiol. 8:111-115.

11. Barnett, H. J. 1973. D-alanine carboxypepti-dases of Bacillus stearothermophilus: solubili-zation of particulate enzymes and mechanismof action of penicillin. Biochim. Biophys. Acta304:332-352.

12. Blumberg, P. M. 1974. Penicillin binding com-ponents of bacterial cells and their relation-ship to the mechanism of penicillin action.Ann. N.Y. Acad. Sci. 235:310-325.

13. Blumberg, P. M., and J. L. Strominger. 1971.Inactivation of D-alanine carboxypeptidase bypenicillins and cephalosporins is not lethal inBacillus subtilis. Proc. Nat. Acad. Sci. U.S.A.68:2814-2817.

14. Blumberg, P. M., and J. L. Strominger. 1972.Isolation by covalent affinity chromatographyof the penicillin binding components frommembranes of Bacillus subtilis. Proc. Nat.Acad. Sci. U.S.A. 69:3751-3755.

15. Blumberg, P. M., and J. L. Strominger. 1972.Five penicillin binding components occur inBacillus subtilis membranes. J. Biol. Chem.247:8107-8113.

16. Blumberg, P. M., R. R. Yocum, E. Willoughby,and J. L. Strominger. 1974. Binding of ["4C]-penicillin G to the membrane-bound and thepurified D-alanine carboxypeptidases fromBacillus stearothermophilus and Bacillus sub-tilis and its reversal. J. Biol. Chem. (in press).

17. Bogdanovsky, D., E. Bricas, and P. Dezelee.1969. Sur l'identite de la mucoendopeptidaseet de la carboxypeptidase I d'Escherichia coli,enzymes hydrolysant des liaisons de configu-ration D-D et inhibee par la penicilline. C. R.Acad. Sci. (Paris) Ser. D 269:390-393.

18. Bohme, E. H. W., H. E. Applegate, B. Toeplitz,J. F. Dolfini, and J. Z. Gougoutas. 1971.6-Methyl penicillins and 7-methyl cephalo-sporins. J. Amer. Chem. Soc. 93:4324-4326.

19. Boman, H. G., K. G. Eriksson-Grennberg, J.Foldes, and E. B. Lindstroim. 1967. The regu-lation and possible evolution of a penicilli-nase-like enzyme in Escherichia coli, p.366-372. In V. V. Koningsberger and L. Bosch(ed.), Regulation of nucleic acid and proteinbiosynthesis. Amsterdam Elsevier Publ. Co.,Amsterdam.

20. Boman, H. G., K. G. Eriksson-Grennberg, S.Normark, and E. Matsson. 1968. Resistance ofEscherichia coli to penicillins. IV. Geneticstudy of mutants resistant to D, L-ampicillin

327VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 38: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

concentrations of 100 ug/ml. Genet. Res.12:169-185.

21. Boyer, P. D. (ed.) 1971. The enzymes, vol. III,Peptide bond hydrolysis (3rd ed.). AcademicPress Inc., New York.

22. Braun, V., and V. Bosch. 1972. Repetitive se-quences in the murein-lipoprotein of the cellwall of Escherichia coli. Proc. Nat. Acad. Sci.U.S.A. 69:970-974.

23. Braun, V., and V. Bosch. 1972. Sequence of themurein-lipoprotein and the attachment site ofthe lipid. Eur. J. Biochem. 28:51-69.

24. Braun, V., K. Hantke, H. Wolff, and G. Gerisch.1972. Degradation of the murein-lipoproteincomplex of Escherichia coli cell walls byDictyostelium amoebae. Eur. J. Biochem.27:116-125.

25. Braun, V., and K. Rehn. 1969. Chemical charac-terization, spatial distribution, and functionof a lipoprotein (murein-lipoprotein) of theEscherichia coli cell wall. The specific effect oftrypsin on the membrane structure. Eur. J.Biochem. 10:426-438.

26. Braun, V., and U. Sieglin. 1970. The covalentmurein-lipoprotein structure of the Esche-richia coli cell wall. The attachment site of thelipoprotein on the murein. Eur. J. Biochem.13:336-346.

27. Braun, V., and H. Wolff. 1970. The murein-lipo-protein linkage in the cell wall of Escherichiacoli. Eur. J. Biochem. 14:387-391.

28. Bukhari, A. I., and A. L. Taylor. 1971. Mutantsof Escherichia coli with a growth requirementfor either lysine or pyridoxine. J. Bacteriol.105:988-998.

29. Burman, L. G., and K. Nordstr6m. 1971. Colicintolerance induced by ampicillin or mutationto ampicillin resistance in a strain of Esche-richia coli K-12. J. Bacteriol. 106:1-13.

30. Burman, L. G., K. Nordstr6m, and G. D. Bloom.1972. Murein and the outer penetration bar-rier of Escherichia coli K-12, Proteus mirabi-lis, and Pseudomonas aeruginosa. J. Bacteriol.112:1364-1374.

31. Burman, L. G., K. Nordstrom, and H. G. Bo-man. 1968. Resistance of Escherichia coli topenicillins. V. Physiological comparison of twoisogenic strains, one with chromosomally andone with episomally mediated ampicillin re-sistance. J. Bacteriol. 96:438-446.

32. Chapman, G. B., and J. Hillier. 1953. Electronmicroscopy of ultra-thin sections of bacteria.I. Cellular division in Bacillus cereus. J. Bac-teriol. 66:362-373.

33. Chatterjee, A. N., and J. T. Park. 1964. Biosyn-thesis of cell wall mucopeptide by a particu-late fraction from Staphylococcus aureus.Proc. Nat. Acad. Sci. U.S.A. 51:9-16.

34. Chatterjee, A. N., and F. E. Young. 1972. Regu-lation of the bacterial cell wall: isolation andcharacterization of peptidoglycan mutants ofStaphylococcus aureus. J. Bacteriol.111:220-230.

35. Citri, N. 1971. Penicillinase and other beta-lac-

tamases, p. 23-46. In P. D. Boyer (ed.), Theenzymes, vol. 4, (3rd ed.). Academic PressInc., New York.

36. Citri, N., and M. R. Pollock. 1966. The biochem-istry and function of beta-lactamase (penicil-linase). Advan. Enzymol. 28:237-323.

37. Cohen, S., C. J. Gibson, and H. M. Sweeney.1972. Phenotypic suppression of methicillinresistance in Staphyloccus aureus by mutantnoninducible penicillinase plasmids. J. Bacte-riol. 112:682-689.

38. Cohen, S., and H. M. Sweeney. 1968. Constitu-tive penicillinase formation in Staphylococcusaureus owing to a mutation unlinked to thepenicillinase plasmid. J. Bacteriol.95:1368-1374.

39. Cole, M. 1969. Hydrolysis of penicillins andrelated compounds by the cell-bound penicil-lin acylase of Escherichia coli. Biochem. J.115:733-739.

40. Cole, M. 1969. Deacylation of acylamino com-pounds other than penicillins by the cell-bound penicillin acylase of Escherichia coli.Biochem. J. 115:741-745.

41. Cole, M. 1969. Penicillins and other acylaminocompounds synthesized by the cell-boundpenicillin acylase of Escherichia coli. Bio-chem. J. 115:747-756.

42. Cole, M. 1969. Factors affecting the synthesis ofampicillin and hydroxypenicillins by the cell-bound penicillin acylase of Escherichia coli.Biochem. J. 115:757-764.

43. Collins, J. F., and M. H. Richmond. 1962. Astructural similarity between N-acetyl-muramic acid and penicillin as a basis forantibiotic action. Nature (London)195:142-143.

44. Cooper, P. D. 1954. The site of action of penicil-lin. III. Effect of surface-active substances onpenicillin uptake by Staphylococcus aureus.Biochim. Biophys. Acta 13:433-438.

45. Cooper, P. D. 1954. The association of thepenicillin binding component of Staphylococ-cus aureus with a lipid fraction. J. Gen.Microbiol. 10:236-245.

46. Cooper, P. D. 1955. The site of action of penicil-lin: some properties of the penicillin-bindingcomponent of Staphylococcus aureus. J. Gen.Microbiol. 12:100-106.

47. Cooper, P. D. 1955. The site of action of penicil-lin: some changes in Staphylococcus aureusduring the first two hours growth in penicillinmedia. J. Gen. Microbiol. 13:22-38.

48. Cooper, P. D. 1956. Site of action of radiopeni-cillin. Bacteriol. Rev. 20:28-48.

49. Cooper, P. D., R. C. Clowes, and D. Rowley.1954. A note on the use of radioactive penicil-lin. J. Gen. Microbiol. 10:246-249.

50. Cooper, P. D., and D. Rowley. 1949. Investiga-tions with radioactive penicillin. Nature (Lon-don) 163:480-481.

51. Cooper, P. D., D. Rowley, and I. M. Dawson.1949. Location of radioactive penicillin inStaphylococcus aureus after contact with the

328 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 39: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

drug. Nature (London) 164:842-843.52. Crane, L. J., G. E. Bettinger, and J. 0. Lam-

pen. 1973. Affinity chromatography purifica-tion of penicillinase of Bacillus licheniformis749/c and its use to measure turnover of thecell bound enzyme. Biochem. Biophys. Res.Commun. 50:220-227.

53. Csanyi, V., G. Jacobi, and B. F. Straub. 1967.The regulation of penicillinase synthesis. Bio-chim. Biophys. Acta 145:470-484.

54. Cuatrecasas, P., and C. B. Anfinsen. 1971.Affinity chromatography, p. 345-378. In S. P.Colowick and N. 0. Kaplan (ed.), Methods inenzymology, vol. 22. Academic Press Inc.,New York.

55. Daniel, J. W., and M. J. Johnson. 1954. Proper-ties of the penicillin binding component ofMicrococcus pyogenes. J. Bacteriol.67:321-328.

56. Demerec, M. 1945. Production of staphylococcusstrains resistant to various concentrations ofpenicillin. Proc. Nat. Acad. Sci. U.S.A.31:16-24.

57. Demerec, M. 1948. Origin of bacterial resistanceto antibiotics. J. Bacteriol. 56:63-74.

58. Duerksen, J. D. 1964. Localization of the site offixation of the inducer, penicillin, in Bacilluscereus. Biochim. Biophys. Acta 87:123-140.

59. Duguid, J. P. 1946. The sensitivity of bacteria tothe action of penicillin. Edinburgh Med. J.53:401-412.

60. Dusart, J., A. Marquet, J.-M. Ghuysen, J. M.Frere, R. Moreno, M. Leyh-Bouille, K. John-son, C. Lucchi, H. R. Perkins, and M. Nieto.1973. DD-Carboxypeptidase-transpeptidaseand killing site of fl-lactam antibiotics inStreptomyces strains R39, R61, and K11.Antimicrob. Ag. Chemother. 3:181-187.

61. Eagle, H. 1948. A paradoxical zone phenomenonin the bactericidal action of penicillin in vitro.Science 107:44-45.

62. Eagle, H. 1951. Further observations on the zonephenomenon in the bactericidal action ofpenicillin. J. Bacteriol. 62:663-668.

63. Eagle, H. 1953. The binding of penicillin inrelation to its cytotoxic action, p. 1-9. Sympo-sium on growth inhibition and chemotherapy,Rome, 6th International Congress on Microbi-ology.

64. Eagle, H. 1954. The binding of penicillin inrelation to its cytotoxic action. I. Correlationbetween the penicillin sensitivity and combin-ing activity of intact bacteria and cell-freeextracts. J. Exp. Med. 99:207-226.

65. Eagle, H. 1954. The binding of penicillin inrelation to its cytotoxic action. II. The reactiv-ity with penicillin of resistant variants ofstreptococci, pneumococci, and staphylococci.J. Exp. Med. 100:103-115.

66. Eagle, H. 1954. The multiple mechanisms ofpenicillin resistance. J. Bacteriol. 68:610-616.

67. Eagle, H., M. Levy, and R. Fleischman. 1955.The binding of penicillin in relation to itscytotoxic action. IV. The amounts bound by

bacteria at ineffective, growth inhibitory, bac-tericidal, and maximally effective concentra-tions. J. Bacteriol. 69:167-172.

68. Eagle, H., and A. D. Musselman. 1948. The rateof bactericidal action of penicillin in vitro as afunction of its concentration, and its paradoxi-cally reduced activity at high concentrationsagainst certain organisms. J. Exp. Med.88:99-131.

69. Edwards, J. R., and J. T. Park. 1969. Correlationbetween growth inhibition and the binding ofvarious penicillins and cephalosporins toStaphylococcus aureus. J. Bacteriol. 99:459-462.

70. Ellar, D. J., D. G. Lundgren, and R. A. Slepecky.1967. Fine structure of Bacillus megateriumduring synchronous growth. J. Bacteriol.94:1189-1205.

71. Eriksson-Grennberg, K. G. 1968. Resistance ofEscherichia coli to penicillins. II. An improvedmapping of the amp A gene. Genet. Res.12:147-156.

72. Eriksson-Grennberg, K. G., H. G. Boman, J. A.T. Jansson, and S. Thoren. 1965. Resistance ofEscherichia coli to penicillins. I. Genetic studyof some ampicillin-resistant mutants. J. Bac-teriol. 90:54-62.

73. Fan, D. P., and B. E. Beckman. 1973. Structuraldifference between walls from hemisphericalcaps and partial septa of Bacillus subtilis. J.Bacteriol. 114:790-79'7.

74. Fan, D. P., M. C. Pelvit, and W. P. Cunning-ham. 1972. Structural difference betweenwalls from ends and sides of the rod-shapedbacterium Bacillus subtilis. J. Bacteriol.109:1266-1272.

75. Few, A. V., P. D. Cooper, and D. Rowley. 1952.Reaction of penicillin with the staphylococcalcell wall. Nature (London) 169:283-284.

76. Few, A. V., P. D. Cooper, and D. Rowley. 1953.Determination of radio-penicillin uptake byelectrodialysis. Biochem. J. 53:xix.

77. Fitz-James, P. 1960. Participation of the cyto-plasmic membrane in the growth and sporeformation of bacilli. J. Biophys. Biochem.Cytol. 8:507-528.

78. Fordham, W. D., and C. Gilvarg. 1974. Kineticsof cross-linking of peptidoglycan in Bacillusmegaterium. J. Biol. Chem. 249:2478-2482.

79. Forsberg, C., and H. J. Rogers. 1971. Autolyticenzymes in growth of bacteria. Nature (Lon-don) 229:272-273.

80. Forsberg, C. W., and J. B. Ward. 1972. N-acetyl-muramyl-L-alanine amidase of Bacillus li-cheniformis and its L-form. J. Bacteriol.110:878-888.

81. Frere, J. M., J. M. Ghuysen, H. R. Perkins, andM. Nieto. 1973. Molecular weight and aminoacid composition of the exocellular DD-car-boxypeptidase-transpeptidase of Strep-tomyces R61. Biochem. J. 135:463-468.

82. Frere, J. M., J. M. Ghuysen, H. R. Perkins, andM. Nieto. 1973. Kinetics of concomitanttransfer and hydrolysis reactions catalyzed by

329VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 40: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

330 BLUMBERG AN

the exocellular DD-carboxypeptidase-trans-peptidase of Streptomyces R61. Biochem. J.135:483-492.

83. Gale, E. F., E. Cundliffe, P. E. Reynolds, M. H.Richmond, and M. J. Waring. 1972. Inhibitorsof bacterial cell wall synthesis, p. 49-120. InThe molecular basis of antibiotic action. JohnWiley & Sons, Inc., New York.

84. Gardner, A. D. 1940. Morphological effects ofpenicillin on bacteria. Nature (London)146:837-838.

85. Gardner, A. D. 1945. Microscopical effect ofpenicillin on spores and vegetative cells ofbacilli. Lancet ii:658-659.

86. Ghuysen, J. M. 1968. Use of bacteriolytic en-zymes in determination of wall structure andtheir role in metabolism. Bacteriol. Rev.32:425-464.

87. Ghuysen, J. M., L. Dierickx, J. Coyette, M.Leyh-Bouille, M. Guinand, and J. N. Camp-bell. 1969. An improved technique for thepreparation of Streptomyces peptidases andN-acetylmuramyl-L-alanine amidase activeon bacterial wall peptidoglycans. Biochemis-try 8:213-222.

88. Ghuysen, J.-M., M. Leyh-Bouille, R. Bonaly, M.Nieto, H. R. Perkins, K. H. Schleifer, and 0.Kandler. 1970. Isolation of DD carboxypepti-dase from Streptomyces albus G culture fil-trates. Biochemistry 9:2955-2961.

89. Ghuysen, J.-M., M. Leyh-Bouille, J. N. Camp-bell, R. Moreno, J.-M. Frere, C. Duez, M.Nieto, and H. R. Perkins. 1973. Structure ofthe wall peptidoglycan of Streptomyces R39and the specificity profile of its exocellularDD-carboxypeptidase-transpeptidase for pep-tide acceptors. Biochemistry 12:1243-1251.

90. Ghuysen, J.-M., M. Leyh-Bouille, J. M. Frere, J.Dusart, K. Johnson, A. Marquet, and R.Moreno. 1972. The Streptomyces DD-carbox-ypeptidase-transpeptidase system as a modelfor the study of penicillin action. 3rd Interna-tional Symposium on Medicinal Chemistry,Milan, Italy.

91. Ghuysen, J.-M., M. Leyh-Bouille, J. M. Frere, J.Dusart, K. Johnson, M. Nakel, J. Coyette, H.R. Perkins, and M. Nieto. 1972. StreptomycesDD-carboxypeptidase-transpeptidases andmechanism of action of penicillin. In E. Mu-fnoz, F. Garc'lia-Ferrandiz, and D. Vazquez(ed.), Molecular mechanisms of antibioticaction on protein biosynthesis and mem-branes. Elsevier Scientific Publ. Co., NewYork.

92. Ghuysen, J.-M., M. Leyh-Bouille, J. M. Frere, J.Dusart, A. Marquet, H. R. Perkins, and M.Nieto. 1974. The penicillin receptor in Strep-tomyces. Ann. N.Y. Acad. Sci. 235:236-268.

93. Ghuysen, J.-M., J. L. Strominger, and D. J.Tipper. 1968. Bacterial cell walls, p. 53-104. InM. Florkin and E. H. Stotz (ed.), Comprehen-sive biochemistry 26A. American ElsevierPublishing Co., New York.

94. Glaser, L. 1973. Bacterial cell surface polysac-charides. Annu. Rev. Biochem. 42:92-112.

I) STROMINGER BACrERIOL. REV.

95. Good, C. M., and D. J. Tipper. 1972. Conditionalmutants of Staphylococcus aureus defective incell wall precursor synthesis. J. Bacteriol.111:231-241.

96. Gourevitch, A., T. A. Pursiano, and J. Lein.1962. Inactivation of staphylococcal penicil-linase by reaction with resistant penicillins.Nature (London) 195:496-497.

97. Hahn, F. E., and J. Ciak. 1957. Penicillin-induced lysis of Escherichia coli. Science125:119-120.

98. Hartmann, R., J.-V. Holtje, and U. Schwarz.1972. Targets of penicillin action in Es-cherichia coli. Nature (London) 235:426-429.

99. Highton, P. J., and D. G. Hobbs. 1971. Penicillinand cell wall synthesis: a study of Bacilluslicheniformis by electron microscopy. J. Bac-teriol. 106:646-658.

100. Hirashima, A., and M. Inouye. 1973. Specificbiosynthesis of an envelope protein of Esche-richia coli. Nature (London) 242:405-407.

101. Hirashima, A., H. C. Wu, P. S. Venkateswaran,and M. Inouye. 1973. Two forms of a struc-tural lipoprotein in the envelope of Esche-richia coli. Further characterization of the freeform. J. Biol. Chem. 248:5654-5659.

102. Ho, P. P. K., R. D. Towner, J. M. Indelicato, W.A. Spitzer, and G. A. Koppel. 1972. Biochemi-cal and microbiological studies on 6-sub-stituted penicillins. J. Antibiot. 25:627-628.

103. Imsande, J., J. W. Zyskind, and I. Mile. 1972.Regulation of staphylococcal penicillinasesynthesis. J. Bacteriol. 109:122-133.

104. Inouye, M., J. Shaw, and C. Shen. 1972. Theassembly of a structural lipoprotein in theenvelope of Escherichia coli. J. Biol. Chem.247:8154-8159.

105. Izaki, K., M. Matsuhashi, and J. L. Strominger.1966. Glycopeptide transpeptidase and D-ala-nine carboxypeptidase: penicillin-sensitiveenzymatic reactions. Proc. Nat. Acad. Sci.U.S.A. 55:656-663.

106. Izaki, K., M. Matsuhashi, and J. L. Strominger.1968. Biosynthesis of the peptidoglycan ofbacterial cell walls. XIII. Peptidoglycan trans-peptidase and D-alanine carboxypeptidase:penicillin-sensitive enzymatic reaction instrains of Escherichia coli. J. Biol. Chem.243:3180-3192.

107. Izaki, K., and J. L. Strominger. 1968. Biosynthe-sis of the peptidoglycan of bacterial cell walls.XIV. Purification and properties of two D-ala-nine carboxypeptidases from Escheribhia coli.J. Biol. Chem. 243:3193-3201.

108. Kamiryo, T., and J. L. Strominger. 1974. Peni-cillin-resistant temperature-sensitive mutantsof Escherichia coli which synthesize hypo- orhyper-cross-linked peptidoglycan. J. Bacte-riol. 117:568-577.

109. Katz, W., and H. H. Martin. 1970. Peptidecrosslinkage in cell wall murein of Proteusmirabilis and its penicillin-induced unstable.L-form. Biochem. Biophys. Res. Commun.39:744-749.

110. Kaufmann, W. 1964. The possible implication of

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 41: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

a bacterial enzyme in the biochemical mode ofaction of penicillins on gram negative bacte-ria. Biochem. Biophys. Res. Commun.14:458-462.

111. Kayser, F. H., E. J. Benner, R. Troy, and P. D.Hoeprich. 1971. Mode of resistance againstft-lactam antibiotics in stapylococci. Ann.N.Y. Acad. Sci. 182:106-117.

112. Larsen, H. 1967. Biochemical aspects of extremehalophilism. Advan. Microbial Physiol.1:97-132.

113. Lawrence, P. J., M. Rogolsky, and V. T. Hanh.1971. Binding of radioactive benzylpenicillinto sporulating Bacillus cultures: chemistryand fluctuations in specific binding capacity.J. Bacteriol. 108:662-667.

114. Lawrence, P. J., and J. L. Strominger. 1970.Biosynthesis of the peptidoglycan of bacterialcell walls. XV. The binding of radioactivepenicillin to the particulate enzyme prepara-tion of Bacillus subtilis and its reversal withhydroxylamine or thiols. J. Biol. Chem.245:3653-3659.

115. Lawrence, P. J., and J. L. Stominger. 1970.Biosynthesis of the peptidoglycan of bacterialcell walls. XVI. -The reversible fixation ofradioactive penicillin G to the D-alanine car-boxypeptidase of Bacillus subtilis. J. Biol.Chem. 245:3660-3666.

116. Lederberg, J. 1956. Bacterial protoplasts in-duced by penicillin. Proc. Nat. Acad. Sci.U.S.A. 42:574-577.

117. Lederberg, J., and J. St. Clair. 1958. Protoplastsand L-type growth of Escherichia coli. J.Bacteriol. 75:143-160.

118. Lee, B. 1971. Conformation of penicillin as atransition state analog of the substrate ofpeptidoglycan transpeptidase. J. Mol. Biol.61:463-469.

119. Levine, B. B. 1965. Immunochemical mech-anisms involved in penicillin hypersensitivityin experimental animals and in human beings.Fed. Proc. 24:45-50.

120. Leyh-Bouille, M., J. Coyette, J. M. Ghuysen, J.Idczak, H. R. Perkins, and M. Nieto. 1971.Penicillin-sensitive DD-carboxypeptidasefrom Streptomyces strain R61. Biochemistry10:2163-2170.

121. Leyh-Bouille, M., J.-M. Ghuysen, R. Bonaly, M.Nieto, H. R. Perkins, K. H. Schleifer, and 0.Kandler. 1970. Substrate requirements of theStreptomyces albus G DD-carboxypeptidase.Biochemistry 9:2961-2970.

122. Leyh-Bouille, M., J.-M. Ghuysen, M. Nieto, H.R. Perkins, K. H. Schleifer, and 0. Kandler.1970. On the Streptomyces albus G DD-car-boxypeptidase mechanism of action of penicil-lin, vancomycin, and ristocetin. Biochemistry9:2971-2975.

123. Leyh-Bouille, M., M. Nakel, J.-M. Frbre, K.Johnson, J.-M. Ghuysen, M. Nieto, and H. R.Perkins. 1972. Penicillin sensitive DD-car-boxypeptidases from Streptomyces strainsR39 and K11. Biochemistry 11:1290-1298.

124. Lienhard, G. E. 1973. Enzymatic catalysis and

transition-state theory. Science 180:149-154.125. Lindqvist, R. C., and K. Nordstrom. 1970. Re-

sistance of Escherichia coli to penicillins. VII.Purification and characterization of a penicil-linase mediated by the R factor R1. J. Bacte-riol. 101:232-239.

126. Lindstro5m, E. B., H. G. Boman, and B. B.Steele. 1970. Resistance of Escherichia coli topenicillins. VI. Purification and characteriza-tion of the chromosomally mediated penicil-linase present in amp A-containing strains. J.Bacteriol. 101:218-231.

127. Linnett, P. E., and J. L. Stominger. 1974.Amidation and cross-linking of the enzymati-cally synthesized peptidoglycan of Bacillusstearothermophilus. J. Biol. Chem. 249:2489-2496.

128. Linnett, P. E., and J. L. Stominger. 1974. Bio-synthesis and cross-linking of the -y-glutamylglycine containing peptidoglycan of vegetativecells of Sporosarcina ureae. J. Biol. Chem.249:2497-2506.

129. Luderitz, 0. 1970. Recent results on the bio-chemistry of the cell wall lipopolysaccharidesof Salmonella bacteria. Angew. Chem.9:649-663.

130. Lugtenberg, E. J. J., L. de Haas-Menger, and W.H. M. Ruyters. 1972. Murein synthesis andidentification of cell wall precursors of temper-ature-sensitive lysis mutants of Escherichiacoli. J. Bacteriol. 109:326-335.

131. Lugtenberg, E. J. J., and A. van Schijndel-vanDam. 1972. Temperature-sensitive mutants ofEscherichia coli K-12 with low activities of theL-alanine adding enzyme and the D-alanyl-D-alanine adding enzyme. J. Bacteriol.110:35-40.

132. Lugtenberg, E. J. J., and A. van Schijndel-vanDam. 1972. Temperature-sensitive mutants ofEscherichia coli K-12 with low activity of thediaminopimelic acid adding enzyme. J. Bacte-riol. 110:41-46.

133. Lund, F., and L. Tybring. 1972. 6ft-Amidinopenicillanic acids-a new group ofantibiotics. Nature N. Biol. 236:135-137.

134. Maass, E. A., and M. J. Johnson. 1949. Peni-cillin uptake by bacterial cells. J. Bacteriol.57:415-422.

135. Maass, E. A., and M J. Johnson. 1949. Therelations between bound penicillin and growthin Staphylococcus aureus. J. Bacteriol.58:361-366.

136. Makela, P. H., and B. A. D. Stocker. 1969.Genetics of polysaccharide biosynthesis.Annu. Rev. Genet. 3:291-322.

137. Martin, H. H. 1963. Zusammensetzung derStUtzmembran in der Zellwand von normalenZellen und Penicillin-Spharoplasten vonProtein mirabilis. Zentralbl. Bakteriol. Parasi-tenk. Infektionskr. Hyg. 191:409-414.

138. Martin, H. H. 1964. Composition of the mucopo-lymer in cell walls of the unstable and stableL-form of Proteus mirabilis. J. Gen. Microbiol.36:441-450.

139. Martin, H. H. 1964. Chemical composition of

331VoL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 42: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

cell wall mucopolymers from penicillin-induced spheroplasts and normal cells of Pro-teus mirabilis. Proc. 6th Int. Congr. Biochem.,New York (abstr.), p. 518.

140. Martin, H. H. 1966. Biochemistry of bacterialcell walls. Annu. Rev. Biochem. 35:457-484.

141. Martin, H. H., H. D. Heilmann, and H. J.Preusser. 1972. State of the rigid-layer in cellwalls of some gram-negative bacteria. Arch.Mikrobiol. 83:332-346.

142. Matsuhashi, M., C. P. Dietrich, and J. L.Strominger. 1965. Incorporation of glycine intothe cell wall glycopeptide in Staphylococcusaureus: role of sRNA and lipid intermediates.Proc. Nat. Acad. Sci. U.S.A. 54:587-594.

143. Matsuhashi, M., C. P. Dietrich, and J. L.Strominger. 1967. Biosynthesis of the peptido-glycan of bacterial cell walls. III. The role ofsoluble ribonucleic acid and of lipid interme-diates in glycine incorporation in Staphylococ-cus aureus. J. Biol. Chem. 242:3191-3206.

144. Matsuhashi, M., M. Furuyama, and B. Maruo.1969. Penicillin-sensitive D-alanine carboxy-peptidase in Anabaena variabilis, a blue-greenalga. Biochim. Biophys. Acta 184:670-671.

145. Matsuhashi, S., T. Kamiryo, P. M. Blumberg,P. Linnett, E. Willoughby, and J. L. Strom-inger. 1974. Study of the mechanism of actionand development of resistance to a new ami-dino penicillin. J. Bacteriol. 117:578-587.

146. Matsuzawa, H., M. Matsuhashi, A. Oka, and Y.Sugino. 1969. Genetic and biochemical studieson cell wall peptidoglycan synthesis in Esche-richia coli K-12. Biochem. Biophys. Res. Com-mun. 36:682-689.

147. Meadow, P. M., J. S. Anderson, and J. L.Strominger. 1964. Enzymatic polymerizationof UDP-acetylmuramyl-L-Ala-D-Glu-L-Lys-D-Ala-D-Ala and UDP-acetylglucosamine bya particulate enzyme from Staphylococcusaureus and its inhibition by antibiotics. Bio-chem. Biophys. Res. Commun. 14:382-387.

148. Melchior, N. H., J. Blom, L. Tybring, and A.Birch-Andersen. 1973. Light and electron mi-croscopy of the early response of Escherichiacoli to a 6#-amidinopenicillanic acid (FL1060). Acta Pathol. Microbiol. Scand. Sec. B81:393-407.

149. Meynell, G. G., and E. Meynell. 1970. Theoryand practice in experimental bacteriology,2nd ed. p. 18. Cambridge University Press.

150. Miller, A. K., E. Celozzi, B. A. Pelak, E. 0.Stapley, and D. Hendlin. 1972. Cephamycins,a new family of ,B-lactam antibiotics. III. Invitro studies. Antimicrob. Ag. Chemoth.2:281-286.

151. Miller, T. W., R. T. Goegelman, R. G. Weston, I.Putter, and F. J. Wolf. 1972. Cephamycins, anew family of fl-lactam antibiotics. II. Isola-tion and chemical characterization. Antimi-crob. Ag. Chemother. 2:132-135.

152. Mirelman, D., R. Bracha, and N. Sharon. 1972.Role of the penicillin-sensitive transpeptida-tion reaction in attachment of newly synthe-

sized peptidoglycan to cell walls of Micrococ-cus luteus. Proc. Nat. Acad. Sci. U.S.A.69:3355-3359.

153. Mirelman, D., R. Bracha, and N. Sharon. 1974.Studies on the elongation of bacterial cell wallpeptidoglycan and its inhibition by penicillin.Ann. N.Y. Acad. Sci. 235:326-347.

154. Mirelman, D., and N. Sharon. 1972. Biosynthe-sis of peptidoglycan by a cell wall preparationof Staphylococcus aureus and its inhibition bypenicillin. Biochem. Biophys. Res. Commun.46:1909-1917.

155. Mitchell, P., and J. Moyle. 1951. Relationshipsbetween cell growth, surface properties andnucleic acid production in normal and penicil-lin-treated Micrococcus pyogenes. J. Gen. Mi-crobiol. 5:421-438.

156. Miyakawa, T., H. Matsuzawa, M. Matsuhashi,and Y. Sugino. 1972. Cell wall peptidoglycanmutants of Escherichia coli K-12: existence oftwo clusters of genes, mra and mrb, for cellwall peptidoglycan biosynthesis. J. Bacteriol.112:950-958.

157. Mohberg, J., and M. J. Johnson. 1958. Compara-tive study of the binding of synnematin andpenicillin by bacterial cells. J. Bacteriol.76:385-392.

158. Mohr, V., and H. Larsen. 1963. On the structuraltransformations and lysis of Halobacteriumsalinarium in hypotonic and isotonic solu-tions. J. Gen. Microbiol. 31:267-280.

159. Monner, D. A., S. Jonsson, and H. G. Boman.1971. Ampicillin resistant mutants of Esche-richia coli K-12 with lipopolysaccharide alter-ations affecting mating ability and suscepti-bility to sex-specific bacteriophages. J. Bacte-riol. 107:420-432.

160. Muiioz, E., and A. Marquet. 1972. Cooperativebinding of penicillin G to cell membranes fromStreptomyces albus: possible relationshipwith the mechanism of resistance to and themechanism of action of the antibiotic. In E.Munioz, F. Garcia-Ferrandiz, and D. Vazquez,(ed.), Molecular mechanisms of antibioticaction on protein biosynthesis and mem-branes. Elsevier, New York.

161. Munioz, E., A. Marquet, V. Larraga, and J.Coyette. 1972. Isolation, partial characteriza-tion of the cytoplasmic membrane fraction ofStreptomyces albus G and DD-carboxypepti-dase localization. Arch. Mikrobiol.81:273-288.

162. Nayler, J. H. C. 1971. Structure-activity rela-tionships in semi-synthetic penicillins. Proc.Roy. Soc. Ser. B 179:357-367.

163. Nieto, M., H. R. Perkins, J.-M. Frere, and J.-M.Ghuysen. 1973. Fluorescence and circular di-chroism studies on the Streptomyces R61DD-carboxypeptidase-transpeptidase. Peni-cillin binding by the enzyme. Biochem. J.135:493-505.

164. Nieto, M., H. R. Perkins, M. Leyh-Bouille,J.-M. Frere, and J.-M. Ghuysen. 1973. Pep-tide inhibitors of Streptomyces DD-carboxy-

332 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 43: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

peptidases. Biochem. J. 131:163-171.165. Nikaido, H. 1968. Biosynthesis of lipopolysac-

charide in gram-negative enteric bacteria. Ad-van. Enzymol. 31:77-124.

166. Nordstrom, K., L. G. Burman, and K. G.Eriksson-Grennberg. 1970. Resistance ofEscherichia coli to penicillins. VIII. Physiol-ogy of a class II ampicillin-resistant mutant. J.Bacteriol. 101:659-668.

167. Ndrdstrom, K., K. G. Eriksson-Grennberg, andH. G. Boman. 1968. Resistance of Escherichiacoli to penicillins. III. AmpB, a locus affectingepisomally and chromosomally mediated re-sistance to ampicillin and chloramphenicolGenet. Res. 12:157-168.

168. Normark, S. 1969. Mutation in Escherichia coliK-12 mediating spherelike envelopes andchanged tolerance to ultraviolet irradiationand some antibiotics. J. Bacteriol.98:1274-1277.

169. Normark, S. 1970. Genetics of a chain-formingmutant of Escherichia coli. Transduction anddominance of the envA gene mediating in-creased penetration to some antibacterialagents. Genet. Res. 16:63-78.

170. Normark, S., H. G. Boman, and E. Matsson.1969. Mutant of Escherichia coli with anom-alous cell division and ability to decreaseepisomally and chromasomally mediated re-sistance to ampicillin and several other anti-biotics. J. Bacteriol. 97:1334-1342.

171. O'Callaghan, C. H., S. M. Kirby, A. Morris, R.E. Waller, and R. E. Duncombe. 1972. Corre-lation between hydrolysis of the f-lactambond of the cephalosporin nucleus and expul-sion of the 3-substituent. J. Bacteriol.110:988-991.

172. Ortiz, P. J. 1970. Dihydrofolate and dihydrop-teroate synthesis by partially purified en-zymes from wild-type and sulfonamide-resist-ant pneumococcus. Biochemistry 9:355-361.

173. Ortiz, P. J., and R. D. Hotchkiss. 1966. Theenzymatic synthesis of dihydrofolate anddihydropteroate in cell-free preparations fromwild-type and sulfonamide-resistant pneumo-coccus. Biochemistry 5:67-74.

174. Osborne, M. J. 1969. Structure and biosynthesisof the bacterial cell wall. Annu. Rev. Biochem.38:501-538.

175. Park, J. T. 1952. Uridine-5'-pyrophosphate de-rivatives. I. Isolation from Staphylococcusaureus. J. Biol. Chem. 194:877-884.

176. Park, J. T. 1952. Uridine-5'-pyrophosphate de-rivatives. II. A structure common to threederivatives. J. Biol. Chem. 194:885-895.

177. Park, J. T. 1952. Uridine-5'-pyrophosphate de-rivatives. III. Amino acid-containing deriva-tives. J. Biol. Chem. 194:897-904.

178. Park, J. T., and L. Burman. 1973. FL-1060: anew penicillin with a unique mode of action.Biochem. Biophys. Res. Commun. 51:863-868.

179. Park, J. T., J. R. Edwards, and E. M. Wise. 1974.In vivo studies on the uptake and binding of

beta-lactam antibiotics in relation to inhibitionof wall synthesis and cell death. Ann. N.Y.Acad. Sci. 235:300-309.

180. Park, J. T., M. E. Griffith, and I. Stevenson.1971. Resistance to penicillin in mutants of apenicillinase-negative organism, Staphylococ-cus aureus H. J. Bacteriol. 108:1154-1160.

181. Park, J. T., and M. J. Johnson. 1949. Accumula-tion of labile phosphate in Staphylococcusaureus grown in the presence of penicillin. J.Biol. Chem. 179:585-592.

182. Park, J. T., and J. L. Strominger. 1957. Mode ofaction of penicillin. Biochemical basis for themechanism of action of penicillin and for itsselective toxicity. Science 125:99-101.

183. Pelzer, H. 1963. Mucopeptidhydrolasen in Esch-erichia coli. B. I. Nachweis und Wirkungs-spezifitlat. Z. Naturforsch. 18b:950-956.

184. Perkins, H. R., M. Nieto, J.-M. Frere, M.Leyh-Bouille, and J.-M. Ghuysen. 1973.Streptomyces DD-carboxypeptidases astranspeptidases. The specificity for aminocompounds acting as carboxyl acceptors. Bio-chem. J. 131:707-718.

185. Perret, C. J. 1953. Note on the biosynthesis andisolation of ""S-labelled benzylpenicillin. J.Gen. Microbiol. 8:195-197.

186. Pollock, J. J., J.-M. Ghuysen, R. Linden, M. R.J. Salton, H. R. Perkins, M. Nieto, M. Leyh-Bouille, J. M. Frere, and K. Johnson. 1972.Transpeptidase activity of Streptomyces D-alanyl-D carboxypeptidases. Proc. Nat. Acad.Sci. U.S.A. 69:662-666.

187. Pollock, M. R. 1950. Penicillinase adaptation inBacillus cereus: adaptive enzyme formation inthe absence of free substrate. Brit. J. Exp.Pathol. 31:739-753.

188. Pollock, M. R. 1953. Penicillinase adaptationand fixation of penicillin sulphur by Bacilluscereus spores. J. Gen. Microbiol. 8:136-197.

189. Pollock, M. R. 1957. The activity and specificityof inducers of penicillinase production in Ba-cillus cereus strain NRRL 569. Biochem. J.66:419-428.

190. Pollock, M. R. 1971. The function and evolutionof penicillinase. Proc. Roy. Soc. Ser. B179:385-401.

191. Pollock, M. R., and C. J. Perret. 1951. Therelation between fixation of penicillin sulphurand penicillinase adaptation in Bacilluscereus. Brit J. Exp. Pathol. 32:387-396.

192. Pruess, D. L., and M. J. Johnson. 1965. En-zymatic deacylation of 35S-benzylpenicillin. J.Bacteriol. 90:380-383.

193. Reynolds, P. E. 1971. Peptidoglycan synthesis inbacilli. I. Effect of temperature on the in vitrosystem from Bacillus megaterium and Bacil-lus stearothermophilus. Biochim. Biophys.Acta 237:239-254.

194. Reynolds, P. E. 1971. Peptidoglycan synthesis inbacilli. II. Characteristics of protoplast mem-brane preparations. Biochim. Biophys. Acta237:255-272.

195. Reynolds, P. E., and H. J. Barnett. 1974. Trans-

333VOL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 44: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

BLUMBERG AND STROMINGER

peptidases and DD-carboxypeptidases inbacilli. Ann. N.Y. Acad. Sci. (in press).

196. Richmond, M. H., and R. B. Sykes. 1973. Thef-lactamases of gram-negative bacteria andtheir possible physiological role. Advan. Mi-crobial Physiol. 9:31-88.

197. Rogers, H. J. 1967. The inhibiton of mucopep-tide synthesis by benzylpenicillin in relationto irreversible fixation of the antibiotic byStaphylococci. Biochem. J. 103:90-102.

198. Rogers, H. J., and C. W. Forsberg. 1971. Role ofautolysins in the killing of bacteria by somebactericidal antibiotics. J. Bacteriol. 108:1235-1243.

199. Rogers, H. J., and H. R. Perkins. 1968. Cell wallsand membranes. Spon. London.

200. Rogolsky, M., P. J. Lawrence, and V. T. Hanh.1973. Binding of radioactive benzylpenicillinto asporogenous mutants of Bacillus subtilisduring postexponential growth. J. Bacteriol.114:220-227.

201. Rolinson, G. N. 1971. Bacterial resistance topenicillins and cephalosporins. Proc. Roy.Soc. Ser. B 179:403-410.

202. Rowley, D., P. D. Cooper, P. W. Roberts, and E.L. Smith. 1950. The site of action of penicillin.I. Uptake of penicillin on bacteria. Biochem.J. 46:157-161.

203. Rowley, D., J. Miller, S. Rowlands, and E.Lester-Smith. 1948. Studies with radioactivepenicillin. Nature (London) 161:1009-1010.

204. Russell, A. D., and R. H. Fountain. 1971.Aspects of the mechanism of action of somecephalosporins. J. Bacteriol. 106:65-69.

205. Russell, R. R. B. 1972. Temperature-sensitiveosmotic remedial mutants of Escherichia coli.J. Bacteriol. 112:661-665.

206. Ryter, A., and 0. E. Landman. 1964. Electronmicroscope study of the relationship betweenmesosome loss and the stable L state (orprotoplast state) in Bacillus subtilis. J. Bacte-riol. 88:457-467.

207. Sabath, L. D., S. J. Wallace, and D. A. Gerstein.1972. Suppression of intrinsic resistance tomethicillins and other penicillins in Staphylo-coccus aureus. Antimicrob. Ag. Chemother.2:350-355.

208. Salton, M. R. J. 1964. The bacterial cell wall.Elsevier Publishing Co., New York.

209. Schepartz, S. A., and M. J. Johnson. 1956. Thenature of the binding of penicillin by bacterialcells. J. Bacteriol. 71:84-90.

210. Schleifer, K. H., and 0. Kandler. 1972. Peptido-glycan types of bacterial cell walls and theirtaxonomic implications. Bacteriol. Rev.36:407-477.

211. Schmid, R., and R. Plapp. 1972. Binding of14C-penicillin G to Proteus mirabilis. Arch.Mikrobiol. 83:246-260.

212. Schwarz, U., A. Asmus, and H. Frank. 1969.Autolytic enzymes and cell division of Esche-richia coli. J. Mol. Biol. 41:419-429.

213. Sharpe, A., P. M. Blumberg, and J. L. Stroming-er. 1974. n-Alanine carboxypeptidase and cell

wall cross-linking in Bacillus subtilis. J. Bac-teriol. 117:926-927.

214. Stapley, E. O., M. Jackson, S. Hernandez, S. B.Zimmerman, S. A. Currie, S. Mochales, J. M.Mata, H. B. Woodruff, and D. Hendlin. 1972.Cephamycins, a new family of #-lactam an-tibiotics. I. Production by actinomycetes, in-cluding Streptomyces lactamdurans sp.n. An-timicrob. Ag. Chemother. 2:122-131.

215. Storm, D. R., P. M. Blumberg, and J. L.Strominger. 1974. Inhibition of the Bacillussubtilis membrane-bound D-alanine carboxy-peptidase by 6-aminopenicillanic acid cova-lently coupled to Sepharose. J. Bacteriol.117:783-785.

216. Strominger, J. L. 1959. The amino acid sequencein the uridine nucleotide-peptide from Staph-ylococcus aureus. C. R. Trav. Lab. Carlsberg31:181-192.

217. Strominger, J. L. 1970. Penicillin-sensitive en-zymatic reactions in bacterial cell wall synthe-sis. Harvey Lect. 64:179-213.

218. Strominger, J. L. 1973. The actions of penicillinand other antibiotics on bacterial cell wallsynthesis. Johns Hopkins Med. J. 133:63-81.

219. Strominger, J. L., P. M. Blumberg, H. Sugina-ka, J. Umbreit, and G. G. Wickus. 1971. Howpenicillin kills bacteria: progress and prob-lems. Proc. Roy. Soc. Ser. B. 179:369-383.

220. Strominger, J. L., K. Izaki, M. Matsuhashi, andD. J. Tipper. 1967. Peptidoglycan transpepti-dase and D-alanine carboxypeptidase: penicil-lin-sensitive enzymatic reactions. Fed. Proc.26:9-22.

221. Strominger, J. L., J. T. Park, and R. E. Thomp-son. 1959. Composition of the cell wall ofStaphylococcus aureus: its relation to themechanism of action of penicillin. J. Biol.Chem. 234:3263-3268.

222. Strominger, J. L., and D. J. Tipper. 1965.Bacterial cell wall synthesis and structure inrelation to the mechanism of action of penicil-lins and other antibacterial agents. Amer. J.Med. 39:708-721.

223. Strominger, J. L., and D. J. Tipper. Structure ofbacterial cell walls: the lysozyme substrate. InE. F. Osserman, R. E. Canfield, and S.Beychak (ed.), Symposium on Lysozyme (inpress).

224. Strominger, J. L., E. Willoughby, T. Kamiryo,P. M. Blumberg, and R. R. Yocum. 1974. Peni-cillin-sensitive enzymes and penicillin-bindingcomponents in bacterial cells. Ann. N.Y. Acad.Sci. 235:210-224.

225. Suginaka, H., P. M. Blumberg, and J. L. Strom-inger. 1972. Multiple penicillin-binding com-ponents in Bacillus subtilis, Bacillus cereus,Staphylococcus aureus, and Escherichia coli.J. Biol. Chem. 247:5279-5288.

226. Tipper, D. J. 1970. Structure and function ofpeptidoglycans. Int. J. Syst. Bacteriol.20:361-377.

227. Tipper, D. J. 1973. Bacterial cell walls, p.121-205, 331-346. In G. D. Fasman and S. N.

334 BACTERIOL. REV.

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from

Page 45: Interaction Penicillin with Bacterial Cell: Penicillin ... · PENICILLININTERACTIONWITHBACTERIALCELLS tures arethose shownin Fig. 2. Thepeptidogly-can, also known as murein or mucopeptide,

PENICILLIN INTERACTION WITH BACTERIAL CELLS

Timasheff (ed.), Subunits in biological sys-tems. Marcel Dekker Inc., New York.

228. Tipper, D. J., and J. L. Strominger. 1965.Mechanism of action of penicillins: a proposalbased on their structural similarity to acyl-D-alanyl-D-alanine. Proc. Nat. Acad. Sci.U.S.A. 54:1133-1141.

229. Tipper, D. J., and J. L. Strominger. 1968. Bio-synthesis of the peptidoglycan of bacterial cellwalls. XII. Inhibition of cross-linking by peni-cillins and cephalosporins: studies in Staphy-lococcus aureus in vivo. J. Biol. Chem.243:3169-3179.

230. Tomasz, A., A. Albino, and E. Zanati. 1970.Multiple antibiotic resistance in a bacteriumwith suppressed autolytic system. Nature(London) 227:138-140.

231. Umbreit, J. N., and J. L. Strominger. 1972.Solubilization, purification and characteriza-tion of a penicillin-sensitive D-alanyl-D-ala-nine carboxypeptidase from Bacillus subtilis.Fed. Proc. 31:414 (abstr.).

232. Umbreit, J. N., and J. L. Strominger. 1973.D-alanine carboxypeptidase from Bacillussubtilis membranes. I. Purification and char-acterization. J. Biol. Chem. 248:6759-6766.

233. Umbreit, J. N., and J. L. Strominger. 1973.D-alanine carboxypeptidase from Bacillussubtilis membranes. II. Interaction with peni-cillins and cephalosporins. J. Biol. Chem.248:6767-6771.

234. Umbreit, J. N., and J. L. Strominger. 1973.Relation of detergent HLB number to solubili-zation and stabilization of D-alanine carboxy-peptidase from Bacillus subtilis membranes.Proc. Nat. Acad. Sci. U.S.A. 70:2997-3001.

235. Warth, A. D., and J. L. Strominger. 1971.Structure of the peptidoglycan from vegeta-tive cell walls of Bacillus subtilis. Biochemis-try 10:4349-4358.

236. Weidel, W., and H. Pelzer. 1964. Bagshaped

macromolecules-a new outlook on bacterialcell walls. Advan. Enzymol. 26:193-233.

237. White, P. J. 1968. A strain of Pediococcuscerevisiae which requires methicillin forgrowth. J. Gen. Microbiol. 50:85-105.

238. White, P. J. 1968. A comparison of the cell wallsof Pediococcus cerevisiae and of a substrainthat requires methicillin for growth. J. Gen.Microbiol. 50:107-120.

239. Wickus, G. G., and J. L. Strominger. 1972.Penicillin-sensitive transpeptidation duringpeptidoglycan biosynthesis in cell-free prepa-rations from Bacillus megaterium. I. Incorpo-ration of free diaminopimelic acid into pepti-doglycan. J. Biol. Chem. 247:5297-5306.

240. Wickus, G. G., and J. L. Strominger. 1972.Penicillin-sensitive transpeptidation duringpeptidoglycan biosynthesis in cell-free prepa-rations from Bacillus megaterium. H. Effect ofpenicillins and cephalosporins on bacterialgrowth and in vivo transpeptidation. J. Biol.Chem. 247:5307-5311.

241. Wijsman, H. J. W. 1972. A genetic map ofseveral mutations affecting the mucopeptidelayer of Escherichia coli. Genet. Res. 20:65-74.

242. Wilkinson, B. J., and P. J. White. 1973. Wallsynthesis by Pediococcus cerevisiae and by asubstrain that requires methicillin for growth.J. Bacteriol. 116:507-509.

243. Wise, E. M., and J. T. Park. 1965. Penicillin: itsbasic site of action as an inhibitor of a peptidecross-linking reaction in cell wall mucopeptidesynthesis. Proc. Nat. Acad. Sci. U.S.A.54:75-81.

244. Work, E; 1969. Biochemistry of bacterial cellwalls. Lab. Pract. 18:831-838.

245. Yocum, R. R., P. M. Blumberg, and J. L. Strom-inger. 1974. Purification and characterizationof the thermophilic D-alanine carboxypepti-dase from membranes of Bacillus stearother-mophilus. J. Biol. Chem. (in press).

335VoL. 38, 1974

on August 25, 2019 by guest

http://mm

br.asm.org/

Dow

nloaded from