Antibody Response of Children with Enteropathogenic Escherichia ...

269

Antibody Response of Children with Enteropathogenic Escherichia coli Infection to the Bundle-Forming Pilus and Locus of Enterocyte Effacement– Encoded Virulence Determinants Marina B. Martinez, Carla R. Taddei, Alejandro Ruiz-Tagle, Luiz R. Trabulsi, and Jorge A. Giroń

Faculdade de Cieˆncias Farmaceûticas, and Instituto de Cieˆncias Biome´dicas, Universidade de Saõ Paulo, Saõ Paulo, Brazil; Centro de Investigaciones en Ciencias Microbiolo´gicas, Beneme´rita Universidad Autońoma de Puebla, Puebla, Mexico

Enteropathogenic Escherichia coli (EPEC) express a plasmid-encoded type IV pilus termed bundle-forming pilus, which is associated with the formation of bacterial microcolonies on cultured epithelial cells. Bacterial attachment and effacement of the enterocyte brush border membrane is attributed to a surface outer membrane protein adhesin termed intimin and EPEC-secreted proteins EspA, EspB, and EspD. Except for intimin, production in vivo or antibody response against these virulence determinants during natural EPEC infections in young children has not been demonstrated. Antibody responses against BfpA, intimin, EspA, and EspB were investigated in Brazilian children naturally infected with EPEC. Generally, IgG antibodies against BfpA and EspB were the most commonly found, followed by antiEspA and intimin antibodies. Thus, bundle-forming pilus and locus of enterocyte attachment–encoded products are produced in vivo during natural EPEC infections and elicit an immune response against heterologous EPEC virulence determinants. These findings have important implications in the immunoprophylaxis against EPEC infections.

Acute diarrhea is the most important syndrome associated with nutritional aggravation, particularly among infants !6 months to 1 year of age living in developing countries [1]. Enteropathogenic Escherichia coli (EPEC) is an important cause of acute diarrhea in this population [2]. When EPEC reach the small intestine, they attach intimately to the enterocyte brush border membrane, triggering a profound rearrangement of cellular cytoskeleton proteins accompanied by protein phosphorylation, pedestal formation, and destruction of surface microvilli. This histopathology constitutes the so-called attaching and effacing lesion (A/E) [1]. Several chromosomal genes located in the locus of enterocyte effacement (LEE) [3] are involved in the manifestation of the A/E lesion. The eae codes for a 94-kDa outer membrane protein termed intimin, which mediates intimate contact with epithelial cells; tir encodes an intimin translocated receptor called Tir [1]; the espA, espB, and espD genes encode EPEC-secreted proteins (Esps) EspA, EspB, and EspD, involved in signal transduction; and the esc and sep genes encode a type III secretion system [4]. EPEC adheres to the small

Received 24 March 1998; revised 19 August 1998. Parents gave full consent for participation of the children in the study, who underwent physical examination and laboratory testing. Financial support: CNPq-USP, Brazil (to M.B.M.); Conacyt (Mexico) Project 3485P-M9607 (to J.A.G.). Reprints or correspondence: Dr. Marina Baquerizo Martinez, Faculdade de Cieˆncias Farmaceûticas, Depto. de Analises Clıńicas e Toxicologia, Universidade de Saõ Paulo, Av. Lineu Prestes, 580, bl. 17. Cidade Universita´ria, Butanta˜, Saõ Paulo, Brazil 05508-900 ([email protected]). The Journal of Infectious Diseases 1999; 179:269–74 q 1999 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/99/7901-0039$02.00

bowel of children or tissue culture cells, forming tight microcolonies, a pattern termed localized adherence (LA) [1]. The LA phenotype is mediated by a type IV pilus, called the bundleforming pilus (BFP) [5], which is encoded on the 90-kb EPEC adherence factor (EAF) plasmid [6]. The BFP is composed of a structural bundlin subunit BfpA of 19.5 kDa. The BFP forms interbridging structures within the adhering microcolony, tethering bacteria to each other [5]. Experimental infection of human volunteers with EPEC showed that they developed IgG antibodies against a 94-kDa outer membrane protein [6] that was initially thought to be the EAF determinant and later was identified as intimin [1]. IgA and oligosaccharide fractions from human colostrum were shown to inhibit LA of EPEC to HEp-2 or HeLa cells and to recognize intimin [7–10]. However, none of the studies above noted reactivity to BFP, probably because production of BFP is induced under defined growth conditions [5] and the methods then available did not assay for BFP or Esps. A recent study reported that human convalescent serum from a volunteer experimentally infected with EPEC recognized Esps [4]. Nevertheless, little is known about the antibody response elicited against BFP and Esps during natural infections in young children or whether these antibodies are protective. Recently, Donnenberg et al. [11] observed that human volunteers rechallenged with homologous or heterologous EPEC strains developed an IgG response mainly against lipopolysaccharide O antigen and intimin but not to BfpA. In the present study, we evaluated the antibody response of Brazilian children naturally infected with EPEC against BfpA, intimin, and the recently described virulence determinants EspA and EspB. These data are important

270

Martinez et al.

JID 1999;179 (January)

in terms of vaccine development and pose the possibility of including these surface proteins in vaccines to immunize mothers and protect neonates against EPEC infection.

taining Esps were concentrated 500-fold in a filtration apparatus (Amicon, Beverly, MA) under nitrogen pressure. The Esps were reacted with human sera (diluted 1:50) as described above.

Materials and Methods

Results

Bacterial strains and growth conditions. EPEC strains E2348/ 69 (O127:H6), B171 (O111:NM), and ICB34 (O55:H6) were used as positive controls of LA, for production of BFP and Esp, and for hybridization with eae, bfpA, and EAF probes. JPN15, a plasmidless derivative of E2348/69 [6], E. coli DH5a, and ICB34 grown in Luria broth (LB) were used as negative controls of BFP expression. Generally, EPEC strains do not express BFP in LB [5]. CVD206 is an eae deletion mutant derived from E2348/69 [10]. The strains were grown at 377C on Luria agar plates for DNA hybridization and in LB for adherence assays to HEp-2 cells [2, 9, 12]. For assessment of BFP expression, EPEC strains were grown in Dulbecco’s MEM (Sigma, St. Louis) complemented with 10% fetal bovine serum [12]. Patients, controls, and antisera. We selected 26 children (ranging from 2 months to 3 years of age) with acute diarrhea who shed E. coli belonging to EPEC serogroups and who attended the Saõ Paulo University Hospital in Saõ Paulo. This hospital provides free medical care to urban children of lower socioeconomic status living in Saõ Paulo. Blood was obtained from the children 10 days after the onset of acute diarrhea. Unfortunately, we did not have sera from these children before infection. As negative controls, we used 25 sera obtained from children without diarrhea who attended the same hospital because of illnesses other than intestinal infections. They were in the same age range as the patients with diarrhea and had the same socioeconomic status. All sera were collected and kept at 2207C for further testing. Rabbit polyclonal antisera. Antiserum raised against BFP was used to assess BFP production [5]. Anti-EspA and anti-EspB antisera were a gift of G. Frankel (Imperial College of Science, Technology, and Medicine, London). Identification of EPEC and colony blot hybridizations. Stool samples were obtained from the children and processed for isolation of E. coli as reported before [2]. DNA hybridizations with the eae, bfpA, and EAF probes were done under high-stringency conditions as previously described [2, 12]. Localized adherence assay. Adhesion of EPEC to HEp-2 cell monolayers after 3 h of infection was done as previously described [2, 9]. Immunoblottings. Whole cell extracts of ICB34 and E2348/69, grown in MEM, were reacted with human sera diluted 1:100 to determine the presence of BfpA and intimin-reacting antibodies as previously described [4, 6, 12]. Rabbit anti-BFP antiserum was used as positive control. Goat anti-human IgG and IgA, and anti-rabbit IgG conjugated to horseradish peroxidase, were used as secondary antibodies. The reaction was developed with a mixture of diaminobenzidine and 3% hydrogen peroxide (Sigma). To confirm the presence of BfpA-reacting antibodies in the children’s sera, blots containing BFP purified from EPEC B171 were also used. Isolation of Esps. Esps were prepared from E2348/69 and ICB34 grown overnight at 377C in Dulbecco’s MEM without fetal bovine serum as previously described [4]. Filtered supernatants con-

Characterization of E. coli isolates and BFP production. Before testing for antibodies against EPEC antigens in the children’s sera, we confirmed the presence of EPEC in their stools and that the isolates produced BFP. Among the E. coli isolates, we found the following numbers of EPEC serogroups: O55, 7; O86, 1; O111, 4; O119, 8; O125, 1; O127, 1; O128, 3; and O142, 1. The results of DNA probe hybridizations, LA assays, and expression of BFP of these isolates are shown in table 1. A high correlation between positivity with eae, EAF, and bfpA probes was observed. This association has been reported previously [2, 12]. Except for 1 eae1 O128 isolate, the E. coli from the latter four serogroups were negative with eae, EAF, and bfpA probes. It has already been shown that some of these serogroups are not considered EPEC; rather, they belong to such other diarrheogenic E. coli groups as enterotoxigenic, enteroaggregative, diffusely adhering, or avirulent isolates [1]. Among 17 bfpA1 isolates, 14 (82.3%) produced the bundlin subunit (19.5 kDa) (table 1; figure 1A). We were unable to demonstrate BfpA production in 3 bfpA1 isolates. An EAFnegative O111 E. coli isolate from patient 11 produced bundlin subunit, which suggests that the BFP may be produced in the presence or absence of the EAF determinant. The production of BfpA by ICB34 was demonstrated by immunoblotting, and this strain was used to screen the children’s sera for reactivity against this antigen. Antibody response to EPEC antigens. Among 17 children who had EPEC that hybridized to bfpA plasmid probes (EAF and bfpA), 13 (76.4%) had sera that contained specific antibodies against BfpA. Figure 1B depicts the reactivity of sera from patients to BfpA (see also table 1). Two children who harbored E. coli isolates that did not hybridize to EPEC probes had sera with antibodies to BfpA (table 1). It is possible that these children had been infected with EPEC in the past. We determined the class of the BfpA-reacting immunoglobulins by using specific goat anti-human IgG and IgA conjugates. The sera contained primarily IgG (figure 1) and not IgA anti-EPEC antibodies (data not shown). When ICB34 was grown in LB (a nonpermissive condition for BfpA expression) or when whole cell extracts of JPN15 or DH5a were used as the antigens, no 19.5-kDa protein reacted with the children’s sera (data not shown). Furthermore, the 13 sera reacting with BfpA in whole cell extracts of ICB34 reacted with purified BFP obtained from B171 (data not shown). Among the 25 sera obtained from control children, 3 reacted weakly with the bundlin subunit, suggesting that these 3 children might have been exposed to EPEC before. Reactivity to intimin depended on the strain that was used


Antibody Response against EPEC Virulence Determinants

271

Table 1. Characteristics of strains isolated from children with diarrhea and reactivity of their sera against BfpA, intimin, EspA, and EspB. Bacterial isolates

Patient 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

DNA probes

Reactivity of patients’ serum

EPEC serotype

eae

EAF

bfpA

LA at 3h

BfpA expression

Anti-BfpA ICB34

Anti-intimin ICB34

Anti-intimin E2348

Anti-EspA ICB34

Anti-EspA E2348

Anti-EspB ICB34

Anti-EspB E2348

O55 O55 O55 O55 O55 O55 O55 O86 O111 O111 O111 O111 O119 O119 O119 O119 O119 O119 O119 O119 O125 O127 O128 O128 O128 O142

1 1 1 2 1 1 1 2 1 1 1 1 1 1 1 1 1 2 1 1 2 2 2 1 2 2

1 1 1 2 1 1 1 2 1 1 2 1 1 1 1 1 1 2 1 1 2 2 2 2 2 2

1 1 1 2 1 1 1 2 1 1 1 1 1 1 1 1 1 2 1 1 2 2 2 2 2 2

ND 1 1 2 1 1 ND ND ND ND 1 1 ND ND 1 ND ND 2 1 ND ND 2 ND ND ND ND

1 1 1 2 1 1 2 2 1 1 1 1 2 1 2 1 1 2 1 1 2 2 2 2 2 2

1 1 2 1 1 1 2 2 1 1 1 2 2 1 1 1 1 1 1 1 2 2 2 2 2 2

1 1 2 1 1 1 1 1 2 2 1 2 2 2 1 2 2 2 1 2 2 2 2 2 2 2

1 2 2 1/2 2 2 1 2 2 2 2 2 2 2 1/2 2 2 2 2 2 1 1/2 2 2 2 1

2 2 2 1 1 2 1 1/2 1 2 1 2 2 2 1 2 1 2 1 1 2 1 2 1 1 1

2 2 1/2 1/2 1 1 1 2 1/2 2 1 1 2 2 1 1/2 2 2 1/2 1 1/2 1 1/2 1 1/2 1/2

1 2 1 1 1 2 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1

2 1/2 1/2 1 1 1 1 2 1/2 2 1 2 2 2 1 1/2 1/2 2 1 1 2 1 1 1 1 1

NOTE. 1: positive reaction; 2: negative reaction; 1/2: weak reaction; ND: not done.

as the source of the antigen. For example, 8 sera from children harboring bfpA1 EPEC recognized intimin from ICB34 (figure 1B), whereas intimin from E2348/69 was recognized by 3 of the sera (table 1). These data suggest that these children were previously exposed to other bacteria expressing intimin-related antigens [13]. Other proteins in the ranges of 27–30, 35–39, and 55–67 kDa were also detected in whole cell extracts of ICB34 (figure 1B), but their identity is unknown. It is possible that the 35- to 39-kDa proteins correspond to Esps (EspB and EspD) [4] or outer membrane proteins and the 55- to 67-kDa proteins to flagella proteins. Some of the children who harbored non-EPEC E. coli also recognized some the EPEC antigens tested, suggesting previous exposures to these bacteria (table 1). To confirm the identity of the 94-kDa protein recognized by the children’s sera, we reacted these sera with whole cell extracts of CVD206 (an eae deletion mutant derived from E2348/69 unable to express intimin) (figure 1C). None of the 94-kDa protein–reacting sera recognized a 94-kDa protein in CVD206, confirming that the 94-kDa protein in ICB34 is intimin. There was a marked difference in the reactivity to EPEC antigens between sera from patients and controls. In contrast to sera from children with diarrhea, which commonly detected BfpA, intimin, and possibly Esps, the control sera did not recognize these antigens. Most control sera (figure 1D) detected a band of ∼33 kDa, which may well correspond to an outer membrane protein.

Reactivity to secreted proteins. Reactivity of children’s sera to Esps depended on the strain used for isolation of these proteins. For example, among 17 children harboring bfpA1 EPEC, 8 reacted with EspA (25 kDa) obtained from ICB34 (figure 1E), whereas 11 reacted to EspA produced by E2348/69 (table 1). Similarly, 14 children reacted with EspB (38 kDa) secreted by ICB34 (figure 1D), while 11 children reacted with EspB of E2348/69 (table 1). In some instances, a doublet band of ∼38–40 kDa was detected, and it is possible that these proteins correspond to EspB (38 kDa) and EspD (39 kDa), which are usually present in Esp extracts and comigrate very closely in polyacrylamide gels [4]. However, reactivity to EspA and EspB was also observed in sera from children who shed non-EPEC strains, suggesting that these children were previously exposed to these antigens. Discussion The aim of the present study was to determine if during the course of natural EPEC infections in children, specific antibodies against intimin and newly described virulence determinants, such as BFP or Esps, are produced; to do so we analyzed the reactivity of convalescent sera against these antigens in immunoblots. The presence of antibodies against bacterial antigens is commonly considered a marker of the production of virulence factors in vivo [2, 6]. To achieve this, we first demonstrated that the EPEC strains isolated from the chil-

272

Martinez et al.


Figure 1. Bundle-forming pilus (BFP) production and IgG reactivity of sera from children with EPEC infection to EPEC antigens. A, Demonstration of BfpA (19.5 kDa) production by EPEC isolates. Lane 1, ICB34; lanes 2–13, isolates 1, 2, 3, 4, 5, 6, 7, 9, 10, 11, 12, 13. B, Whole cell lysate of ICB34 reacted with sera of children with diarrhea and anti-human IgG conjugate. Lane nos. indicate serum from patients 1, 4, 5, 8, 10, 14, 2, 15, 19, 7, 12, 18, 17, and 16, and rabbit anti-BFP serum. C, Whole cell lysate of CVD206 (eae deletion mutant) reacted with sera of children with diarrhea and anti-human IgG conjugate. Lane nos. indicate serum from patients 1, 4, 5, 8, 10, 14, 2, 15, 19, 7, 12, 18, 17, and 16. D, Whole cell lysate of ICB34 reacted with sera of children without diarrhea (controls) and anti-human IgG conjugate. Lane nos. indicate serum from control children 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39. E, Reactivity of sera of children with diarrhea to EPEC-secreted proteins obtained from supernatant of ICB34. Lane nos. indicate serum from patients 17, 24, 25, 20, 26, 7, and 1, controls 30 and 31, and rabbit anti-EspA and anti-EspB.


Antibody Response against EPEC Virulence Determinants

dren with diarrhea contained the genetic virulence traits that characterize EPEC, such as the presence of the EAF plasmid and eae and bfpA genes. Among 17 bfpA1 EPEC isolates found, 14 produced BfpA that reacted with rabbit anti-BFP serum. It is possible that the remaining 3 BfpA-negative isolates lack an accessory or regulatory gene involved in production of BfpA or that these isolates produce antigenically different BFPs [12]. No BfpA production was observed in non-EPEC isolates. To determine if the EPEC-infected children developed antibodies against BfpA, intimin, and Esps, sera were collected 10 days after the onset of diarrhea. Unfortunately, no sera were available before the onset of diarrhea. Nevertheless, 76.4% of the children infected with bfpA1 EPEC had circulating IgG but not IgA antibodies against BfpA. Furthermore, the BfpAreacting sera recognized purified BFP from B171 (O111:NM), supporting the specificity of these antibodies. Interestingly, only 3 of the sera from controls reacted, albeit weakly, with the BfpA subunit. We cannot rule out the possibility that these children had been infected with EPEC in the past and thus contained circulating antibodies against bundlin. Recently Bieber et al. [14] reported that human volunteers experimentally infected with EPEC B171-8 carrying inactivated bfpA, bfpT (perA), or bfpF had significantly less diarrhea than did those receiving the wild type strain. This suggests that BFP is required for virulence of EPEC in vivo. The degree of reactivity to Esps and intimin differed markedly depending on the serotype of the strain used as the source of the antigen. While 8 of the sera from patients recognized intimin produced by ICB34, E2348/69-derived intimin was recognized only by 3 of the sera. Recently, a family of related intimins was reported among A/E1 E. coli [13], which could explain the variability in recognition of heterologous intimins by the different sera used here. In relation to reactivity to Esps, a larger number of sera reacted with EspB produced by ICB34 than with EspB produced by E2348/69. On the basis of available data on DNA and protein sequencing of Esps, it is likely predictable that as with the intimin family, subfamilies of Esps exist among attaching and effacing pathogens. The implications of this variability in terms of widespread protection deserve further attention. The fact that control sera from healthy children did not react against any of the antigens sought is intriguing, considering that all children live in an area in which EPEC is endemic and therefore would be likely exposed to EPEC infections. It is possible that anti-EPEC antibodies are elicited only shortly after infection and do not remain circulating for long periods. Barros et al. [15] reported that most children !30 days old with EPEC infection developed anti-EPEC antibodies after discharge from the hospital and not during the acute phase of the illness. The development of protective antibodies during the course of infection and thereafter remains an interesting issue to address. The role of intimin in experimental EPEC infection has been demonstrated [10]. Other studies have shown that

273

human volunteers experimentally infected with EPEC elicited antibodies against intimin [6] and that human colostrum may contain anti-intimin IgAs [7–9]. However, the importance of BFP and some LEE-encoded products in the pathogenesis of EPEC in young children has not been fully addressed. Several investigations have provided valuable information to the understanding of EPEC pathogenicity and its interaction with the immune system [6–10]. At the time those studies were done, the BFP and EPEC-secreted products had not been discovered, and the methods then available did not assay for BFP or Esps. Antibody response against EPEC antigens was investigated with EPEC strains grown under conditions that are nonpermissive for expression of BFP or Esps. Thus, it was not revealed whether the BFP or Esps are produced in vivo or if antibodies are elicited against these important EPEC determinants. Recently, the effect of prior experimental human EPEC infection on illness following homologous and heterologous rechallenge was studied by Donnenberg et al. [11]. These authors observed that there was no significant effect of prior infection on the incidence of diarrhea, although severity of disease was milder in the homologous-rechallenge study group. Disease severity was inversely correlated with the level of prechallenge serum IgG against the O127 lipopolysaccharide. These authors noted that 30%–50% of human volunteers responded with IgG against intimin, but no reactivity to BfpA was observed. In contrast, we demonstrate here that young children infected by EPEC develop an IgG response to bundlin, intimin, and Esps. EPEC is a pathogen of 0- to 2-year-old children living in areas of the world with inadequate sanitary or nutritional conditions [1]. Thus, interpretation of antibody responses in adult volunteers living in developed countries should be done with caution. Physiologic age-related factors, normal flora, and nutritional status are clearly distinct between children and adults and may selectively influence the establishment of disease and expression of virulence factors. The data presented here confirm the widespread immunogenicity and antigenic variability of BFP and other LEE-encoded virulence determinants among heterologous EPEC. Furthermore, the data strongly suggest that specific antibodies are elicited during naturally occurring infections with EPEC among young infants and that these antigens are produced in vivo. Further studies are needed to elucidate whether anti-BFP, -EspA, -EspB, and -intimin antibodies are protective. Acknowledgments We thank Taˆnia A. T. Gomes and Luzinete A. Silva for performing DNA probe hybridizations, Eloisa C. Souza and the University Hospital of University of Saõ Paulo for providing the sera of children, and Gad Frankel for providing anti-EspA and anti-EspB antisera. References 1. Nataro JP, Kaper JB. Diarrheogenic Escherichia coli. Clin Microbiol Rev 1998; 5:109–14.

274

Martinez et al.

2. Gomes TAT, Rassi V, MacDonald KL, et al. Enteropathogens associated with acute diarrheal disease in urban infants in Saõ Paulo, Brazil. J Infect Dis 1991; 164:331–7. 3. McDaniel TK, Jarvis KJ, Donneberg MS, Kaper JB. A genetic locus of enterocyte attachment and effacement conserve among diverse enterobacterial pathogens. Proc Natl Acad Sci USA 1995; 92:1664–8. 4. Jarvis KG, Giroń JA, Jerse AE, McDaniel TK, Donnenberg MS, Kaper JB. Enteropathogenic Escherichia coli contains a specialized secretion system necessary for the export of proteins involved in attaching and effacing lesion formation. Proc Natl Acad Sci USA 1995; 92:7996–8000. 5. Giroń JA, Ho ASY, Schoolnik GK. An inducible bundle-forming pilus of enteropathogenic Escherichia coli. Science 1991; 254:710–3. 6. Levine MM, Nataro JP, Karch H, et al. The diarrheal response of humans to some serotypes of enteropathogenic Escherichia coli is dependent on a plasmid encoding an enteroadhesiveness factor. J Infect Dis 1985; 152: 550–9. 7. Camara LM, Carbonare SB, Silva MLM, Carneiro-Sampaio MMS. Inhibition of enteropathogenic Escherichia coli (EPEC) adhesion to HeLa cells by human colostrum: detection of specific sIgA related to EPEC outermembrane proteins. Int Arch Allergy Immunol 1994; 103:307–10. 8. Costa-Carvalho BT, Bertipaglia A, Sole D, Naspitz CK, Scaletsky ICA. Detection of immunoglobulin (IgG and IgA) anti-outer-membrane proteins of enteropathogenic Escherichia coli (EPEC) in saliva, colostrum, breast

9.

10. 11.

12.

13.

14.

15.


milk, serum, cord blood, and amniotic fluid. Study of inhibition of localized adherence of EPEC to HeLa cells. Acta Paediatr 1994; 83:870–3. Cravioto A, Tello A, Villafań H, Ruiz J, del Vedovo S, Neeser JR. Inhibition of localized adhesion of enteropathogenic Escherichia coli to HEp-2 cells by immunoglobulin and oligosaccharide fractions of human colostrum and breast milk. J Infect Dis 1991; 163:1247–55. Donnenberg MS, Yu J, Kaper JB. The role of the eaeA gene in experimental enteropathogenic Escherichia coli infection. J Clin Invest 1993; 92:1412–7. Donnenberg MS, Tacket CO, Losonsky G, et al. Effect of prior experimental human enteropathogenic Escherichia coli infection on illness following homologous and heterologous rechallenge. Infect Immun 1998; 66:52–8. Giroń JA, Donnenberg MS, Martin WC, Jarvis KG, Kaper JB. Distribution of the bundle-forming pilus structural gene (bfpA) among enteropathogenic Escherichia coli (EPEC). J Infect Dis 1993; 168:1037–41. Agin TS, Wolf MK. Identification of a family on intimins common to Escherichia coli causing attaching-effacing lesions in rabbits, humans, and swine. Infect Immun 1997; 65:320–6. Bieber D, Ramer SW, Wu CY, et al. Type IV pili, transient bacterial aggregates, and virulence in enteropathogenic Escherichia coli. Science 1998; 280:2114–8. Barros HC, Ramos SRTS, Trabulsi LR, Silva MM. Inhibition of enteropathogenic Escherichia coli adhesion to HeLa cells by serum of infants with diarrhea and cord serum. Braz J Med Biol Res 1995; 28:83–7.