INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Some strains of Escherichia coli and related members of the family Enterobacteriaceae produce antibacterial proteins called colicins (6, 34, 41). Colicins are active on sensitive strains of the same family and preferably on strains within the same species (32, 33). Besides colicin synthesis, colicinogenic strains code for immunity proteins that specifically inhibit the action of the colicin types they produce. Colicin synthesis is believed to provide producer bacteria a selective advantage over noncolicinogenic, sensitive strains (35). Genes for colicin synthesis are plasmid encoded, with the possible exception of bacteriocin 28b of Serratia marcescens (44).

Growth-inhibitory effects of colicins are limited to sensitive bacterial strains which possess specific outer membrane receptor proteins. Colicins bind to bacterial receptors whose primary function is often to facilitate the uptake of nutrients (e.g., vitamin B₁₂, ferric siderophores). In this respect, colicins resemble bacteriophages, which also appropriate receptor proteins to infect a sensitive bacterium. Several outer membrane receptors were identified as colicin receptors (e.g., BtuB, FepA, FhuA, Tsx, and OmpA). Some colicins show additional dependence on lipopolysaccharide molecules and outer membrane porins (12, 40). This complex dependence on outer membrane components is characteristic for colicins taken up by the Tol system. Ton-dependent colicins bind to individual outer membrane proteins, with the exception of colicins 5 and 10, which show dependence on both the Tsx and TolC outer membrane proteins (28, 29). Receptor binding is followed by translocation of the colicin molecule through the cell envelope by either the Ton or the Tol system (4, 43).

Colicin Js is a polypeptide toxin (molecular mass, 10.4 kDa) originally identified as a product of Shigella sonnei (1). Compared to other colicins, colicin Js has a unique antimicrobial spectrum, being active only on enteroinvasive serotypes of E. coli and Shigella strains able to produce a positive reaction in the Serény test, an experimental keratoconjunctivitis in rabbits or guinea pigs. Enteroinvasive E. coli (EIEC) serotypes not sensitive to colicin Js were reported to be negative in this enteroinvasiveness test (19).

Some Shigella and EIEC strains cause a bacillary dysentery characterized by bacterial invasion of the colonic and rectal mucosa. The enteroinvasiveness phenotype of these strains is associated with a 230-kb virulence plasmid coding for the majority of genes that contribute to the enteroinvasiveness phenotype (38). Entry of bacteria into host cells is followed by lysis of the internalized vacuole, bacterial growth in the cytoplasm, and infection of adjacent cells in the intestinal mucosa (37, 39). The enteroinvasiveness phenotype is regulated by the virF-virB regulatory cascade, which is expressed at 37°C and repressed at 30°C (11).

This communication describes the identification of EIEC genes coding for colicin Js sensitivity, iron and temperature regulation of these genes, and comparisons of the primary structures of the genes from two EIEC strains and one Shigella strain.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Media. Bacterial strains were grown at 37°C in TY medium containing (per liter) 8 g of Bacto Tryptone (Difco Laboratories), 5 g of yeast extract, and 5 g of NaCl (pH 7). For selection and maintenance of plasmids, we added (per milliliter of liquid medium or 1.5% [wt/vol] TY agar) 25 µg of chloramphenicol, 100 µg of ampicillin, or 25 µg of kanamycin. Isopropyl--D-thiogalactopyranoside (IPTG) and 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal) were used at 0.5 mM and 80 µg ml¹, respectively. The iron-chelating compound 2,2'-dipyridyl was added to TY plates at 0.2 mM. Campylobacter strains were grown on Campylobacter CSM Selective Medium plates (Becton Dickinson Biosciences, Sparks, Md.).

Bacterial strains and plasmids. The bacterial strains and plasmids used in this study are listed in Table 1. Colicin Js producer strain S. sonnei type 7, colicin Js-sensitive S. sonnei 17 (colicin type 6), and EIEC strain O164 were kindly provided by J. marda, Brno, Czech Republic. Strains of the genera Acinetobacter, Campylobacter, Enterobacter, Pasteurella, Klebsiella, Morganella, Pseudomonas, Salmonella, Serratia, and Shigella, as well as EIEC and enteropathogenic E. coli strains, were from the strain collection of B. Murray, University of Texas Houston Medical School. Mannheimia, Yersinia, two Pseudomonas strains, and one Pasteurella strain were from the Czech Collection of Microorganisms, Brno, Czech Republic. Vibrio cholerae and some Salmonella strains were from the strain collection of this laboratory. The E. coli, EIEC, and Shigella strains used in this study are shown in Table 1. Campylobacter strains were grown at 42°C on TY or CSM plates under microaerophilic conditions generated by the CampyPak Microaerophilic System (Becton Dickinson Biosciences).

Crude colicin preparations. Cells from the TY cultures of colicinogenic strains (producers of colicins Js, 5, 10, E2, and U) induced by mitomycin C (0.5 µg ml¹; Sigma) were harvested, resuspended in distilled water, washed, and sonicated. The sonicates were used as crude colicins. E. coli LMG194, containing a plasmid (pDS83) with the cja gene under the control of the lac promoter, was induced at an optical density at 600 nm (OD₆₀₀) of 0.4 by addition of 1 mM IPTG and then cultivated for an additional 4 h at 37°C. Cells were subsequently harvested, washed, and lysed.

Determination of sensitivity to colicin Js and colicin activity assays. Colicin Js producer bacteria were inoculated onto agar plates with a single stab and subsequently grown for 48 h at 37°C. After that, plates were exposed to chloroform vapor for 30 min to lyse the producer bacteria and then overlaid with 100 µl of colicin Js-sensitive bacteria in 3 ml of 0.75% (wt/vol) TY agar. After overnight cultivation at 37°C, bacteria sensitive to colicin Js formed a zone of growth inhibition around the colicin Js producer.

TnphoA transposon mutagenesis. EIEC strain O164 was grown to an OD₆₀₀ of 0.5 and mixed with TnphoA at a multiplicity of infection equal to 1. Phages were allowed to attach to sensitive bacteria for 30 min at 30°C, and the bacteria were subsequently grown for 4 h at 30°C. Aliquots were plated on TY plates supplemented with kanamycin. Colicin Js-resistant colonies were isolated, and their resistance phenotype was tested with Tol-dependent colicin E2 and Ton-dependent colicin 5 or 10. Strain O164Tn10, which is resistant to colicin Js, was used for further characterization. Genomic DNA of EIEC strain O164Tn10 with a TnphoA insertion was digested with BamHI, cutting once inside TnphoA, as well as in the flanking bacterial DNA. To allow PCR amplification, ends of digested DNA were ligated to each other and ligation products were used for subsequent PCR amplification with primers GW472 and GW473 by using priming sites on TnphoA and directing DNA synthesis toward the outside of TnphoA. The PCR product was cloned in pCR2.1 and sequenced, and the point of TnphoA insertion was identified.

Construction of a genomic library of EIEC strain O164 in E. coli VCS257 and screening for colicin Js sensitivity. Genomic DNA of EIEC strain O164 was partially digested with restriction enzyme Sau3AI, and 25- to 50-kb fragments were isolated and ligated into cosmid pBeloBAC11. Two microliters of ligation solution was mixed with 25 µl of Gigapack III Gold Packaging Extract (Stratagene) and incubated at room temperature (22°C) for 2 h. Subsequently, 500 µl of autoclaved SM buffer (5.8 g of NaCl, 2.0 g of MgSO₄ · 7H₂O, 50 ml of 1 M Tris-HCl [pH 7.5], and 5.0 ml of 2% [wt/vol] gelatin in 1 liter of deionized water) was added to tubes together with 20 µl of chloroform. The cosmid packaging reaction mixture was then diluted 10-fold, and 25 µl was mixed with 25 µl of VCS257 bacteria, resuspended in 10 mM MgSO₄ at an OD₆₀₀ of 0.5, and cultivated at room temperature for 30 min. For expression of antibiotic resistance, 200 µl of TY medium was subsequently added and bacteria were cultivated for 1 h at 37°C, inoculated onto TY plates supplemented with chloramphenicol, and incubated overnight at 37°C. Individual colonies (more than 1,000) were picked, and their sensitivity to colicin Js was tested by streaking them across the region of a TY plate containing dried crude colicin Js. Strains showing inhibition of growth in the colicin Js-containing region were tested for colicin Js sensitivity by using a colicin activity assay.

Southern blot analyses. Chromosomal or plasmid DNA was digested with EcoRI or HindIII and electrophoresed on a 0.8% agarose gel. DNA was subsequently transferred to a nylon membrane by a standard capillary method. The probes used in Southern blot analyses were prepared by using a Gene Images CDP-Star detection module (Amersham Pharmacia Biotech), and the fluorescein-labeled probes were detected with anti-fluorescein-alkaline phosphatase conjugate. Blocking, hybridization, and detection of hybridized probe were performed in accordance with the manufacturer's recommendations.

Restriction analysis. Standard methods were used for DNA isolation, restriction endonuclease analysis, ligation, and transformation of plasmid DNA (36).

Recombinant DNA methods. Plasmids were isolated and cells were transformed by standard techniques. PCR products were cloned into vector pCR2.1 or pCR2.1-TOPO in accordance with the manufacturer's (Invitrogen) recommendations. Cloning in other vectors was done after restriction digestion of both plasmid (cosmid) DNA and target DNA. Insert DNA was sequenced by using the Taq Dye-deoxy Terminator method and a model 377 DNA sequencing system (Applied Biosystems, Foster City, Calif.). The PCR primers used in this study are shown in Table 2.

In vitro transposition. For cosmid mutagenesis, an in vitro Tn7 transposition system (GPS-1 Genome Priming System; New England Biolabs) was used in accordance with the manufacturer's recommendations.

Computer-assisted sequence analysis. Computer-assisted sequence analysis was performed by using programs in the Genetics Computer Group (Madison, Wis.) software package. ProtParam at ExPASy was used for calculations of theoretical polypeptide molecular weight, isoelectric point, etc. Signal sequence prediction (signalp) and protein localization (psort) programs were used.

Nucleotide sequence accession numbers. The nucleotide sequences reported in this study have been deposited in the GenBank database under accession numbers AF283288 to AF283294.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of the region conferring colicin Js sensitivity. A colicin Js-sensitive strain, EIEC strain O164, was subjected to TnphoA transposon mutagenesis, and colicin Js-resistant colonies were isolated. Strains tolerant to colicin Js, defective in translocation of colicin Js (e.g., exbB mutants), were excluded by testing for sensitivity to colicins 5 and 10. A strain specifically resistant to colicin Js was identified. The point of the TnphoA insertion was identified by DNA sequencing of plasmid pDS144 (see Materials and Methods and Table 1). The insertion was localized 26 nucleotides upstream of the start codon of senB (23). The senB gene codes for the TieB product, which may have some role in enterotoxin production by EIEC strains. However, cloning of senB into pCR2.1 (pDS154) was not associated with acquisition of colicin Js sensitivity.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Restriction map of the 40-kb DNA insert of pDS150. Restriction sites for EcoRI and PstI are indicated. EcoRI and PstI fragments were subcloned, and the ends of inserts were sequenced. Hits with respect to E. coli chromosomal DNA, E. coli and Shigella IS sequences, and E. coli plasmids pB171 and pColIb-P9 are indicated. Subcloned DNA fragments of pDS150 in pBSSK(+) are shown (pDS173-175, pDS177, pDS179-180, and pDS186). The cjrABC operon, together with the downstream senB gene, is flanked by IS-like sequences.

View larger version (12K): [in this window] [in a new window]   FIG. 2.   Genetic organization of the cjrABC gene operon. Black (gray) circles indicate Tn7 (TnphoA) insertions leading to resistance to colicin Js, while the open circle represents a Tn7 insertion not affecting sensitivity to colicin Js. Restriction sites for EcoRI, HindIII, and PstI are indicated. Insertion of Tn7 into pDS150 24 bp upstream of the Fur box (pDS219) did not interfere with colicin Js sensitivity, but insertion of Tn7 9 bp downstream from the Fur box (pDS202) resulted in complete loss of colicin Js sensitivity. Downstream from the cjr genes is located senB. Note that the cjr and senB genes are flanked by IS-like sequences.

Cloning of cjr genes. To confirm the role of the cjr genes as colicin Js receptor determinants and to identify the gene(s) coding for the Js receptor, cjr genes were subcloned into two separate compatible plasmids. Since a role for downstream gene cjrC in colicin Js sensitivity was more probable than one for cjrA and cjrB, cjrC was cloned in a separate plasmid, pPD101, resulting in pDS215, while the cjrAB genes were cloned in pCR2.1-TOPO (pDS213). Clones containing cjrC or cjrAB alone did not acquire sensitivity to colicin Js. However, E. coli BL21 with pDS213 and pDS215 was fully sensitive to colicin Js, indicating that the cjrC gene must be present with cjrA, cjrB, or both. Deletions in pDS213 were prepared in cjrA (pDS222; 0.1-kb intragenic deletion of cjrA) or cjrB (pDS221; 0.3-kb intragenic deletion cjrB) and tested with cjrC. The results are summarized in Table 4 and show that the cjrB and cjrC genes are sufficient for colicin Js sensitivity.

Uptake of colicin Js by sensitive bacteria. To test whether colicin Js is taken up by the Ton or Tol system, we examined the sensitivities of different E. coli mutants to colicin Js. Because of the unique antimicrobial spectrum of colicin Js, E. coli mutants were transformed with pDS146, a cosmid with a DNA fragment from EIEC strain O164 that confers sensitivity to colicin Js. The results are summarized in Table 5. An exbB mutant was insensitive to colicin Js, while tonB mutants and all tested tol mutants were sensitive to an extent similar to that of control strain E. coli 5K(pDS146). Introduction of the E. coli K-12-derived tonB gene (cloned on plasmid pDS282) into strain E. coli GUC6 tonB(pDS146) resulted in a 10-fold decrease in colicin Js sensitivity (Table 5). This fact is consistent with decreased iron starvation as a result of complementation of tonB function.

Sensitivities of indicator strains to colicin Js at different temperatures. Expression of genes specific for EIEC and Shigella strains was shown to be temperature regulated (11). The enteroinvasiveness phenotype is observed at 37°C but not at 30°C. We tested the sensitivities of colicin Js indicator strains to colicin Js at different temperatures. The results are summarized in Table 6. S. sonnei 17 was most sensitive to colicin Js at 37°C, less sensitive at 30°C, and even less sensitive at 22°C. The decrease in colicin Js sensitivity was most obvious in the decrease in the dilution causing complete growth inhibition of the indicator bacteria (clear zones of growth inhibition). At 22°C, no clear zones were formed. Similar results were obtained with EIEC strain O164, where the decrease in sensitivity was even greater. EIEC strain O143 showed results similar to those of EIEC strain O164, with lower initial sensitivity to colicin Js. On the other hand, the sensitivity of E. coli 5K to colicin U was not changed when it was grown at 37 or at 22°C. E. coli 5K(pDS146) showed the same sensitivity to colicin Js at all of the temperatures tested, indicating that the uptake of colicin Js was not regulated in this non-EIEC strain and that the growth-inhibitory activity is not affected by temperature.

Role of the virB gene in regulation of sensitivity to colicin Js. Regulation of colicin Js sensitivity by temperature was not observed in the E. coli 5K strain with the cjr operon on cosmid pDS146. Since many of the genes responsible for the enteroinvasiveness phenotype were shown to be regulated by the VirB protein (11), we tested the sensitivities of strains carrying cloned virB to colicin Js at different temperatures. By using PCR amplification, we cloned virB from EIEC strain O143(pDS253) (Tables 1 and 2). When cloned in the pCR2.1-TOPO plasmid, the virB gene was oriented under the control of the T7 promoter. Although the EIEC strains did not contain T7 RNA polymerase, "background" transcription of the virB genes in this high-copy-number plasmid was sufficient to produce increased colicin Js sensitivity. The EIEC O143 strain grown at 25°C was at least 2 orders of magnitude less sensitive to colicin Js than when it was grown at 37°C. The same strain with the virB gene in plasmid pDS253 grown at 25°C was as sensitive to colicin Js as EIEC strain O143 grown at 37°C. No effect of pDS253 was observed for EIEC strain O143 grown at 37°C. The sensitivity to colicin Js was thus regulated by the virB gene, and the regulatory effect of cloned virB was detectable only when bacteria were grown at room temperature. At this temperature, the VirF-VirB regulatory cascade of enteroinvasive bacteria was shown to be turned off (11). Sequencing of the virB gene from EIEC strain O143 (accession no. AF283290) revealed a DNA sequence identical to that previously described for virB from Shigella flexneri (2).

Sensitivity of E. coli strains to colicin Js under iron-depleted conditions. To test the hypothesis that cjr genes are regulated by the Fur protein, we determined the sensitivities of bacteria to colicin Js under standard and iron-depleted conditions (Table 8). The presence of the iron-binding compound dipyridyl increased the colicin Js sensitivity of both EIEC strains, despite their different sensitivities to colicin Js in the presence of iron. Under iron-limited conditions, they reached the same sensitivity. The same sensitivity to colicin Js under iron-depleted conditions was observed also for E. coli strains BL21 and 5K transformed with cosmids with the cjr operon (pDS146 and pDS153, respectively). The difference between the sensitivities of EIEC strains O164 and O143 to colicin Js was thus likely due to a difference between the basal levels of iron regulation of the cjr operon in these strains. Increased colicin Js sensitivity was observed for E. coli BN4020 fur(pDS146) under standard conditions (Table 8), indicating negative transcriptional regulation of the cjr operon by the Fur protein.

Sensitivities of different gram-negative pathogenic bacterial strains to colicin Js. Since cjrBC showed homology to protein products from other bacteria, we tested the sensitivities of some gram-negative bacteria to colicin Js (Table 9). Despite the similarity of CjrBC proteins to gene products of other bacteria, no bacterial strains other than EIEC and Shigella strains were found to be sensitive to Js. The action of colicin Js seems to be specific for EIEC and Shigella strains, resembling the narrow inhibition spectrum of other colicins. Three out of six EIEC strains sensitive to colicin Js were as highly sensitive to colicin Js as EIEC strain O164; the other three were less sensitive to colicin Js (as sensitive as EIEC strain O143). EIEC strains thus appear to have two types of colicin Js sensitivity phenotype (data not shown).

Sequence homology of CjrB and CjrC from two EIEC strains and S. flexneri The cjrBC genes from EIEC strain O143 and S. flexneri #1 were PCR amplified by using the TonBU-TonBL and HasRU-HasRL primer pairs (Table 2), respectively. The PCR products were subsequently sequenced, and the sequences were compared to the cjrBC sequence from EIEC strain O164. Since PCR primers recognizing the first 27 and last 24 nucleotides of cjrB were used for cjrB amplification from bacterial DNA, minor changes in primer binding sequences would not be detected. The cjrB gene from EIEC strain O143 (accession no. AF283291) differed from cjrB of EIEC strain O164 by one base pair, which does not change the amino acid sequence of CjrB. cjrB from S. flexneri #1 (accession no. AF283292) differed from the cjrB gene from EIEC strain O164 by two base pairs, but this does not change any of the amino acids of CjrB. In the case of cjrC, primers recognizing the first 15 nucleotides and primers binding downstream of cjrC were used for PCR amplification. The cjrC gene from EIEC strain O143 (accession no. AF283293) was completely identical to cjrC of EIEC strain O164. cjrC from S. flexneri #1 (accession no. AF283294) differed from cjrC of EIEC strain O164 by two nucleotides, changing one amino acid (T734 to S734). Since the CjrB and CjrC proteins from EIEC strains O164 and O143 were identical, the difference between the sensitivities of the two strains to colicin Js was not due to differences between their cjrCB genes.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Independent approaches identified the region involved in colicin Js sensitivity near and upstream from the senB gene. senB was previously identified in a cosmid clone with DNA from EIEC strain EI34 (23). The senA gene, located on the same cosmid, was shown to be encoded on the large virulence plasmid in EIEC and Shigella strains (23). Although the distance between the senA and senB genes is not known, a polar effect on senA of a TnphoA insertion into the senB gene was proposed (23). However, we did not find the senA gene in an 8-kb region downstream from senB (data not shown). This might reflect a more than 8-kb distance between the sen genes, or it might be explained as a result of EIEC strain differences. The hybridization pattern of the cjrB gene was found to be the same as that of the plasmid-encoded virB gene and different from that of the chromosome-encoded tonB gene. These data are consistent with the localization of the cjr operon on the large virulence plasmid of this enteroinvasive strain. The cjr genes were found to be flanked on both ends by direct repeats and sequences producing significant similarity to bacterial insertion sequences, suggesting DNA rearrangements near the cjr genes.

Tn7 in vitro mutagenesis of cosmid DNA revealed three cjr genes in a 4-kb region. Insertion of Tn7 into any of these genes or into the promoter region upstream of cjrA resulted in complete loss of sensitivity to colicin Js. Since cloning of cjr genes showed that cjrB and cjrC are sufficient to mediate colicin Js sensitivity, insertion of Tn7 into cjrA likely led to colicin Js resistance due to a polar effect. CjrA was homologous to an iron-regulated hypothetical protein from P. aeruginosa. The function of CjrA protein remains unknown. The presence of an N-terminal lipoprotein signal sequence in CjrA suggests that the protein is exported from the cell cytoplasm. Based on sequence predictions, CjrA might be an inner membrane lipoprotein.

CjrB showed similarities to TonB proteins from some gram-negative pathogens, and in accordance with this, the CjrB protein was predicted to be a periplasmic protein anchored to the inner membrane. Colicin Js uptake requires at least one component of the Ton system, protein ExbB. Since E. coli tonB strains with cjrBC genes are fully sensitive to colicin Js, CjrB appears to be an EIEC-specific TonB protein homolog.

CjrC is similar to outer membrane receptors involved in siderophore or heme binding by gram-negative bacteria. Based on sequence similarities and sequence prediction, CjrC appears to be an outer membrane receptor for colicin Js. Ton-dependent colicins (group B) were shown to have a pentapeptide sequence near the N terminus called the TonB box. This sequence was proposed to be responsible for interaction with the TonB protein and for Ton-dependent translocation through the cell envelope. The TonB box of colicin B is a DTMVV sequence, and its introduction into the colicin U molecule resulted in TonB-dependent uptake of colicin U (30). Similar sequences were also identified near the N termini of TonB-dependent outer membrane proteins, and their function in the receptor protein-TonB interaction was described (5, 7). The absence of the TonB box in either the colicin Js polypeptide or the CjrC protein might be explained by their functional tonB independence. Different amino acid residues of colicin Js and CjrC might be involved in the interaction with CjrB. However, no significant homology was found within colicin Js and the N terminus of CjrC.

The sensitivity of EIEC and Shigella strains to colicin Js is temperature dependent, being greater when the bacteria are cultured at 37°C. The invasion phenotype of EIEC and Shigella strains is also temperature regulated, being expressed only when the bacteria are cultivated at 37°C and disappearing at 30°C. The invasion genes located on a large virulence plasmid are regulated by the products of the virF and virB genes (2). The virB gene is activated by the virF gene product, and the VirB protein positively regulates the transcription of the majority of invasion genes (11). E. coli strains with cosmid DNA conferring colicin Js sensitivity did not show this type of temperature-dependent sensitivity, indicating that the regulatory component is specific for EIEC and Shigella strains. Indeed, introduction of the virB gene expressed from the plasmid promoter into EIEC strain O143 increased the sensitivity of this strain to colicin Js at room temperature to the same extent as culturing of the bacteria at 37°C. Moreover, colicin Js sensitivity was increased by virB even for E. coli strains with cjr genes on a cosmid. The positive regulation of the cjr operon by the virF-virB regulatory cascade might suggest the involvement of cjr genes in the invasiveness phenotype. The effect of VirB in the transcriptional activation of cjr genes was blocked by the insertion of Tn7 24 bp upstream from the Fur box. Since the VirB protein is believed to be a DNA binding protein that activates promoters by directly binding to them (11), the VirB binding site(s) may be upstream from the Fur box.

In addition to temperature regulation, transcription of cjr genes is negatively regulated by the level of available iron. This regulation is mediated by the Fur protein. The Fur binding site before the cjr genes differed by four nucleotides from the consensus Fur binding palindromic sequence (9). More recently, the Fur consensus box was reinterpreted as a combination of three repeats of the 5'NAT(A/T)AT3' motif (13). In this case, the cjr Fur box differs by three nucleotides from the Fur consensus sequence. The iron-Fur-regulated genes code for iron acquisition systems and, in many bacterial pathogens, for bacterial toxins or other virulence factors, e.g., shigella enterotoxin 1 (15). Moreover, many other genes were found to be regulated by the Fur protein, e.g., genes involved in acid shock response, defense against oxygen radicals, metabolic pathways, etc. (14). The function of the cjr operon, despite the role in colicin Js uptake, is unknown. The iron and temperature regulation of the cjrABC genes suggests the involvement of these genes in iron metabolism and/or in the enteroinvasiveness phenotype. However, genes with functions other than iron metabolism may be regulated by the Fur protein (14) and some VirB-regulated genes might be dispensable for the invasiveness phenotype.

The cjrB and cjrC genes from two different serotypes of EIEC strains and from S. flexneri are very similar to each other, differing by one or two base pairs per gene. The cjrB genes in all three strains and cjrC in two EIEC strains code for identical proteins, while cjrC from S. flexneri #1 differs by one amino acid residue. Both the cjrB and cjrC genes from EIEC strains are more related to each other than to cjrBC from Shigella.

Among the 120 strains of 16 species of gram-negative bacteria, only EIEC and Shigella strains were found to be sensitive to colicin Js. The presence of the cjr operon is thus likely to be specific for bacteria with the enteroinvasiveness phenotype. The high specificity of the lethal action of colicin Js, which is restricted to EIEC and Shigella strains, resembles the narrow antimicrobial spectra described for other colicins (41). Among the sensitive strains, two types of colicin Js sensitivity phenotype, differing in sensitivity to colicin Js by 1 order of magnitude, were found. EIEC strains O164 and O143, which differ in colicin Js sensitivity, code for identical CjrB and CjrC proteins and have identical sequences in the promoter regions. The different levels of sensitivity of the two strains to colicin Js might be due to differences in the regulation of cjr genes and/or other differences between the EIEC strains.