Penicillin-binding proteins of pathogenic Escherichia coli strains

R.C.C. Ferreira a;*, S.O.P. Costa a, L.C.S. Ferreira b, L.R. Trabulsi c

a Laboratoèrio de Geneètica de Microrganismos, Departamento de Microbiologia, Universidade de Saìo Paulo, ICB II, Av. Prof. Lineu Prestes,1374, 1³ andar, Cidade Universitaèria, Saìo Paulo, SP 05508-900, Brazil

b Laboratoèrio de Fisiologia Celular, Instituto de Biof|èsica Carlos Chagas Filho, CCS-UFRJ, Cidade Universitaèria, Rio de Janeiro,RJ 21949-900, Brazil

c Laboratoèrio Especial de Microbiologia, Instituto Butantan, Saìo Paulo, SP, Brazil

Received 22 July 1998; received in revised form 22 September 1998; accepted 28 September 1998

Abstract

The penicillin-binding proteins of 11 pathogenic Escherichia coli strains, including enteropathogenic, enterotoxigenic,enteroinvasive, enteroaggregative, and enterohemorrhagic E. coli, were detected in gels following the labeling of isolated cellenvelopes with [3H]benzylpenicillin. The electrophoretic profiles, sensitivities to and morphological changes induced by L-lactam antibiotics showed that the penicillin-binding proteins of most pathogenic E. coli possess structural and physiologicalfunctions similar to those of E. coli K12. z 1998 Published by Elsevier Science B.V. All rights reserved.

Keywords: Penicillin-binding protein; Pathogenic Escherichia coli ; L-Lactam antibiotic ; Enteropathogenic Escherichia coli

1. Introduction

Escherichia coli is a member of the normal humanintestinal £ora but also represents an importantcause of diseases such as diarrhea causing thousandsof deaths every year, mainly among children in de-veloping countries [1]. Five major categories of diar-rheagenic E. coli have been de¢ned based on thespeci¢c molecular strategies used by these pathogensto cause disease [2]. Enterotoxigenic E. coli (ETEC)produces heat-stable and heat-labile enterotoxinsand a variety of colonization factors which were re-

sponsible for the binding of bacteria to enterocytereceptors. Enteroinvasive E. coli (EIEC) invadesthe intestinal epithelium causing local necrosis andhemorrhagic diarrhea. Enteropathogenic E. coli(EPEC), a frequent cause of persistent diarrhea, at-taches to enterocytes and promotes the e¡acement ofmicrovilli. Enterohemorrhagic E. coli (EHEC), likeEPEC, displays the attachment and e¡acement phe-notype but additionally produces a Shiga-like toxincausing hemorrhagic diarrhea and, in some cases,kidney failure, the hemolytic uremic syndrome. En-teroaggregative E. coli (EAEC) has a bio¢lm-basedadherence pattern and causes diarrhea through se-creted cytotoxins. Besides the great diversity of vir-ulence-associated traits, serological and molecular bi-ology techniques have demonstrated that pathogenicE. coli strains show a signi¢cant diversity in theirgenetic composition [3,4].

0378-1097 / 98 / $19.00 ß 1998 Published by Elsevier Science B.V. All rights reserved.PII: S 0 3 7 8 - 1 0 9 7 ( 9 8 ) 0 0 4 5 9 - 5

FEMSLE 8460 5-11-98

* Corresponding author. Tel. : +55 (21) 5601484;Fax: +55 (21) 280-8193; E-mail: ritacafe@usp.br

FEMS Microbiology Letters 168 (1998) 313^317

The ¢nal stages of peptidoglycan biosynthesisare catalyzed by a set of cytoplasmic membraneproteins, some of which represent the killing targetsof L-lactam antibiotics, the penicillin-binding pro-teins (PBPs) [5,6]. PBPs are present in all bacteriawith a cell wall but they may vary in number,size, relative amounts, and a¤nity for L-lactam anti-biotics, frequently following taxonomic lines [7,8].PBPs perform essential physiological roles in thebacterial metabolism such as control of cell shape,septum formation and osmotic pressure resistance[8,9].

So far, studies relating to E. coli PBPs have fo-cused only on K12 laboratory strains, which maydi¡er signi¢cantly in genetic and antigenic composi-tion from their pathogenic relatives. In this work webegan structural and physiological studies of thePBPs from diarrheagenic E. coli strains.

2. Materials and methods

2.1. Bacterial strains and growth conditions

Eleven diarrheagenic E. coli strains were analyzedin this work. All strains were considered sensitive tothe tested L-lactam antibiotics in disk di¡usion sus-ceptibility tests carried out in LB agar plates andusing E. coli C600 as a reference strain. One repre-sentative strain of each serotype, representing themajor diarrheagenic E. coli groups, was chosen asfollows: serotype O157:H43 (ETEC); serotypeO129:H3 (EIEC); O157:H7 (EHEC); O111:H12(EAEC), and the O127:H6 prototype EPEC strainE-2348/E [10]. Six EPEC strains, belonging to di¡er-ent serotypes of signi¢cant epidemiological relevancein Brazil, were also included in this study:O111ab:H2, O86:H34, O55:H6, O142:H34 andO142:H6 [11]. All pathogenic E. coli strains wereisolated from hospitalized diarrheic infants in SaìoPaulo, Brazil, except O157:H7 which was from theUSA and was kindly given by Dr. T.S. Whittam(University of Pennsylvania, USA). The strains didnot produce L-lactamase, as evaluated by the sensi-tivity to L-lactam antibiotics and lack of reactionwith nitrocephin (Oxoid, Hampshire, UK), a chro-mogenic cephalosporin. Bacteria were grown as aer-ated cultures in LB medium at 37³C.

2.2. Detection of PBPs and SDS-PAGE

The isolated cell envelopes were suspended inphosphate bu¡er and aliquots, containing approxi-mately 250 Wg of total protein, were labeled with[3H]benzylpenicillin (21 Ci/mmol; Amersham, LittleChalfont, UK) for 20 min at room temperature aspreviously described [5,12,13]. [3H]Benzylpenicillinwas used at a ¢nal concentration of 14 Wg ml31

which, in our hands, was enough to saturate allPBPs of E. coli K12 C600 strain. Proteins were frac-tionated in 10% (w/v) acrylamide gels by use of aMini Protean II slab gel electrophoresis system(Bio-Rad Laboratories, Richmond, CA, USA). Poly-acrylamide gels were treated with 1 M sodium sali-cylate, dried, and placed against Kodak XAR-5 ¢lmsat 370³C for 6^10 days. Some E. coli strains weretreated with 1 mM sodium metaperiodate in 100 mMsodium acetate (pH 4.5) for 15 min at room temper-ature in the dark before being subjected to the cellenvelope extraction procedure in order to reduce theamount of LPS in the Sarkosyl-soluble fraction.

2.3. Identi¢cation of L-lactam lethal targets

The lethal targets of L-lactam antibiotics were de-termined by competition binding assays betweennon-radioactive L-lactam antibiotics (penicillin G,mecillinam and cephalexin) and [3H]benzylpenicillinto the PBPs of pathogenic E. coli strains, as previ-ously described [5]. Non-radioactive L-lactams weretested at their minimal inhibitory concentrations(MIC).

2.4. Microscopic observations

The morphological e¡ects induced by di¡erent L-lactam antibiotics on exponential phase E. coli cellswere observed in a Zeiss phase contrast microscope.

2.5. Determination of MICs

The MIC for each L-lactam antibiotic was deter-mined in microdilution broth susceptibility tests us-ing 96-well microtiter plates containing two-fold se-rial dilutions of the tested antibiotics in LB medium(200 Wl) and seeded with approximately 103 cells (10Wl) of each strain. Plates were visually inspected for

FEMSLE 8460 5-11-98

R.C.C. Ferreira et al. / FEMS Microbiology Letters 168 (1998) 313^317314

bacterial growth following overnight incubation at37³C. Strains were tested in duplicate in at leasttwo independent experiments.

3. Results

The PBP electrophoretic pro¢les of ¢ve pathogenicE. coli strains, representing ¢ve major diarrheagenicgroups for humans, were determined in £uorogramsafter labeling cell envelopes with[3H]benzylpenicillin. Strains 48/63 (O129:H3,EIEC), E2348/69 (O127:H6, EPEC), 30(O157:H43, ETEC), 4A (O157:H7, EHEC), and132 (O111:H12, EAEC) possessed similar PBP pro-¢les when compared to the E. coli K12 C600 strain(Fig. 1). Five major bands with molecular massesranging from 94 to 42 kDa, corresponding to themajor E. coli K12 PBPs (PBP1a and 1b and PBP5and 6 were resolved as single bands in minigels),were detected in all strains. Except for theO157:H7 strain, which showed weakly[3H]benzylpenicillin-labeled bands corresponding toPBP2 and PBP3, all pathogenic E. coli strains hadPBP levels similar to those found in E. coli C600. Asixth band, probably corresponding to PBP7, wasalso detected in most of the isolates (not shown).Several pathogenic E. coli strains expressed largeamounts of lipopolysaccharide (LPS) which were re-

leased during the Sarkosyl solubilization treatmentand interfered with the resolution and molecularmass determination of co-migrating PBPs in SDS-PAGE. Therefore, most of the strains were subjectedto sodium metaperiodate treatment before isolationof cell envelopes.

The morphological e¡ects induced by the L-lactamantibiotics showed that some PBPs of pathogenic E.coli strains have physiological functions similar tothose described for E. coli K12 strains. Therefore,

FEMSLE 8460 5-11-98

Table 1Interactions of L-lactam antibiotics with the PBPs of pathogenic E. coli strains

Serotype Straina Penicillin G Mecillinam Cephalexin

MICb MEc MIC ME MIC ME

K12 C600 10 F,B 0.25 R 5 FEIECO129:H3 48/63 20 F,B 0.25 R 5 FEPECO127:H6 E2348/69 20 F,B 1 R 10 FETECO157:H43 30 10 F,B 0.25 R 5 FEHECO157:H7 4A 10 F,B 0.1 R 2.5 FEAECO111:H12 132 10 F,B 0.25 R 5 F

aLaboratory strain collection reference code.bMIC (Wg ml31), based on growth of strains in liquid medium following overnight incubation at 37³C.cME: morphological e¡ects induced by the tested antibiotics at MIC. F: ¢laments; R: round cells ; B: cells and ¢laments with bulges.

Fig. 1. The PBP electrophoretic pro¢les of pathogenic E. colistrains. The PBP pro¢les of di¡erent E. coli strains were deter-mined in £uorograms following labeling with [3H]benzylpenicillin.The samples were as follows: 1, strain C600 (K12); 2, strain 48/63 (O129:H3, EIEC); 3, strain E2348/69 (O127:H6, EPEC);4, strain 30 (O157:H43, ETEC); 5, strain 4A (O157:H7, EHEC);6, strain 132 (O111:H12, EAEC). The position of PBPs and mo-lecular mass markers (phosphorylase B, 97.4 kDa; bovine serumalbumin, 68 kDa; and ovalbumin, 43 kDa) are indicated on theleft and right sides of the ¢gure, respectively.

R.C.C. Ferreira et al. / FEMS Microbiology Letters 168 (1998) 313^317 315

cephalexin, with increased a¤nity for PBP3, and me-cillinam, which binds speci¢cally to PBP2 over awide concentration range, induced ¢lament andovoid cell formation, respectively, at their corre-sponding MICs in all strains tested (Table 1). Com-petition labeling assays with mecillinam and cepha-lexin con¢rmed that, at MICs, their major killingtargets are PBP2 and PBP3 (data not shown).O157:H7, which showed a reduced expression and/or reduced a¤nity for [3H]benzylpenicillin of PBP2and PBP3, had a reproducibly higher sensitivity tomecillinam and cephalexin when compared to theother strains including E. coli C600 (Table 1).

The previously described phenotypic and geno-typic variability among EPEC strains of distinct se-rotypes [14^16] led us to determine the PBP pro¢lesof strains belonging to seven di¡erent serotypes, inaddition to the O127:H6 (E2348/69) strain. Asshown in Fig. 2, EPEC strains belonging to serotypesO111ab:H2, O127:H40, O86:H34, O55:H6,O142:H34, O142:H6 and O119:H2 share a similarset of PBPs. The altered migration patterns of PBPs1, 2 and 3 observed in some EPEC strains could beattributed, at least in part, to residual amounts ofLPS not e¤ciently removed from the cell envelopesduring treatment with sodium metaperiodate. Likethe other E. coli strains, all EPEC strains showedthe same pattern of morphological changes, andcorresponding putative lethal targets, induced bymecillinam, cephalexin or penicillin G (data notshown).

4. Discussion

The PBPs of E. coli K12 were initially reportedalmost 25 years ago and, since their discovery, sev-eral groups have dedicated considerable attention totheir enzymatic activities, structural properties, phys-iological roles and interactions with L-lactam antibi-otics [6,8,9]. Nonetheless, in contrast to other bacte-rial species [7,12,13], the PBP contents of pathogenicE. coli strains have not been studied. In this work wedescribed the PBP pro¢les of 11 pathogenic E. colistrains representing the major diarrheagenic groupsfor humans. All strains possessed a similar set ofPBPs, based on number and molecular masses.Moreover, preliminary experiments have shownthat the physiological roles of at least PBP2 and 3are similar to those found in E. coli K12.

As observed in E. coli K12 strains, the high mo-lecular mass PBPs of pathogenic E. coli strains rep-resent the main killing targets of L-lactam antibiot-ics. The MICs for penicillin G, mecillinam andcephalexin corresponded to concentrations able topartially or totally inhibit the binding of radioactivepenicillin to their target PBPs. These results indi-cated also that the permeability barrier mediatedby the outer membrane or production of basal levelsof type I L-lactamases have little e¡ect upon suscept-ibility in these strains. Nonetheless, some strainssuch as E2348/69 (O127:H6), which showed repro-ducibly lower L-lactam susceptibility levels, and theO157:H7 strain, which showed lower MICs for me-cillinam and cephalexin, might have essential PBPswith altered expression and/or a¤nity to L-lactams.The reduced labeling of PBP2 and 3 of the O157:H7strain, the most outstanding quantitative di¡erenceobserved among the PBP pro¢les of the strains ana-lyzed, probably re£ects the reduced concentration ofthese proteins in the cell envelope since this strainwas reproducibly more sensitive to PBP2- andPBP3-speci¢c antibiotics but not to penicillin G orother L-lactam antibiotics which bind preferentiallyto PBP1 (Table 1 and unpublished observations).

PBP2 and 3 of pathogenic E. coli strains probablyperform the same physiological functions describedfor their corresponding counterparts in E. coli K12[5,9]. Mecillinam caused formation of osmoticallystable round cells and showed high a¤nity toPBP2. Cephalexin inhibited cell division and bound

FEMSLE 8460 5-11-98

Fig. 2. The PBP electrophoretic pro¢les of di¡erent serotypes ofenteropathogenic E. coli (EPEC). The PBP pro¢les of di¡erentEPEC serotypes were determined in £uorograms following label-ing with [3H]benzylpenicillin. The samples were as follows: 1,strain 18 (O111ab:H2); 2, strain E2348/69 (O127:H6); 3, strain10 (O127:H40); 4, strain 24 (O86:H34); 5, strain 21 (O55:H6);6, strain 22 (O142:H34); 7, strain 19 (O142:H6); 8, strain 109(O119:H2). The position of PBPs and molecular mass markers(phosphorylase B, 97.4 kDa; bovine serum albumin, 68 kDa;and ovalbumin, 43 kDa) are indicated on the left and right sidesof the ¢gure, respectively.

R.C.C. Ferreira et al. / FEMS Microbiology Letters 168 (1998) 313^317316

preferentially to PBP3. In spite of the reducedamounts of PBP2 and 3 in the cell envelope of theO157:H7, these proteins were responsible for theseessential functions in cell wall metabolism based onthe induced morphological changes.

During the last few years an increasing amount ofevidence has indicated that pathogenic E. colistrains, in contrast to their K12 relatives, can occupyniches and survive the host immune responses due totheir £exible genomes and expression of diverse vir-ulence-associated traits such as cell envelope proteinsand LPS composition. The present work has shownthat the PBP content and, probably, the cell wallmetabolism are rather conserved among the labora-tory and di¡erent diarrheagenic E. coli strains. How-ever, such observations do not rule out that, underspeci¢c growth conditions, regulation of the cell wallmetabolism, mediated by control of the enzymaticactivity and/or stability of PBPs, in pathogenic E.coli strains might present features not shared by lab-oratory strains.

References

[1] Levine, M.M. (1987) Escherichia coli that cause diarrhea: en-terotoxigenic, enteropathogenic, enteroinvasive, enterohemor-rhagic and enteroadherent. J. Infect. Dis. 155, 377^389.

[2] Nataro, J.P. and Kaper, J.P. (1998) Diarrheagenic Escherichiacoli. Clin. Microbiol. Rev. 11, 142^201.

[3] Caugant, D.A., Levin, B.R., Orskov, I., Orskov, F., Eden,C.S. and Selander, R.K. (1985) Genetic diversity in relationto serotype in Escherichia coli. Infect. Immun. 49, 407^413.

[4] Orskov, F. and I., Orskov, (1992) Escherichia coli serotypingand disease in man and animals. Can. J. Microbiol. 38, 699^704.

[5] Spratt, B.G. (1977) Properties of the penicillin-binding pro-teins of Escherichia coli K12. Eur. J. Biochem. 72, 341^352.

[6] Park, J.T. (1987) Murein synthesis. In: Escherichia coli andSalmonella typhimurium. Cellular and Molecular Biology(Neidhardt, F.C. et al., Eds.), pp. 663^671. American Societyfor Microbiology, Washington, DC.

[7] Georgopapadakou, N.H. and Liu, F.Y. (1980) Penicillin-bind-ing proteins in bacteria. Antimicrob. Agents Chemother. 18,148^157.

[8] Spratt, B.G. and Cromie, K.D. (1988) Penicillin-binding pro-teins of gram-negative bacteria. Rev. Infect. Dis. 10, 699^711.

[9] Spratt, B.G. (1975) Distinct penicillin binding proteins in-volved in the division, elongation, and shape of Escherichiacoli K12. Proc. Natl. Acad. Sci. USA 72, 2999^3003.

[10] Baldini, M.M., Nataro, J.P. and Kaper, J.B. (1986) Localiza-tion of a determinant for Hep-2 adherence by enteropatho-genic Escherichia coli. Infect. Immun. 52, 334^336.

[11] Toledo, M.R.F., Alvariza, M.C.B., Murahovischi, J., Ramos,S.T.T.S. and Trabulsi, L.R. (1983) Enteropathogenic Esche-richia coli serotypes and endemic diarrhea in infants. Infect.Immun. 39, 586^589.

[12] Ferreira, R.C.C., Park, J.T. and Ferreira, L.C.S. (1994) Plas-mid regulation and temperature-sensitive behavior of the Yer-sinia pestis penicillin-binding proteins. Infect. Immun. 62,2404^2408.

[13] Ferreira, R.C.C., Park, J.T., Camelo, D.C., Almeida, D.F.and Ferreira, L.C.S. (1995) Interactions of Yersinia pestis pen-icillin-binding proteins with L-lactam antibiotics. Antimicrob.Agents Chemother. 39, 1853^1855.

[14] Campos, L.C., Whittam, T.S., Gomes, T.A.T., Andrade,J.R.C. and Trabulsi, L.R. (1994) Escherichia coli O111 in-cludes several clones of diarrheagenic strains with di¡erentvirulence properties. Infect. Immun. 62, 3282^3288.

[15] Gonc°alves, A., Campos, L.C., Campos, T.A.T., Rodrigues, J.,Sperandio, V., Whittam, T. and Trabulsi, L.R. (1997) Viru-lence properties and clonal structure of strains of Escherichiacoli O119 serotypes. Infect. Immun. 65, 2034^2040.

[16] Whittam, T.S. and McGraw, A. (1996) Clonal analysis ofEPEC serogroups. Rev. Microbiol. 27, 7^16.

FEMSLE 8460 5-11-98

R.C.C. Ferreira et al. / FEMS Microbiology Letters 168 (1998) 313^317 317