The aerobactin iron transport system genes in Shigella flexneri are present within a pathogenicity island


  • Steven A. Vokes,

    1. Department of Microbiology and Institute for Cellular and Molecular Biology, University of Texas at Austin, Austin, TX 78712-1095, USA.
    Search for more papers by this author
  • Stephanie A. Reeves,

    1. Department of Microbiology and Institute for Cellular and Molecular Biology, University of Texas at Austin, Austin, TX 78712-1095, USA.
    Search for more papers by this author
  • Alfredo G. Torres,

    1. Department of Microbiology and Institute for Cellular and Molecular Biology, University of Texas at Austin, Austin, TX 78712-1095, USA.
    Search for more papers by this author
  • Shelley M. Payne

    1. Department of Microbiology and Institute for Cellular and Molecular Biology, University of Texas at Austin, Austin, TX 78712-1095, USA.
    Search for more papers by this author

Shelley M. Payne. E-mail; Tel. (+1) 512 471 9258; Fax (+1) 512 471 7088.


Genes encoding the synthesis and transport of aerobactin, a hydroxamate siderophore associated with increased virulence of enteric bacteria, were mapped within a pathogenicity island in Shigella flexneri. The island, designated SHI-2 for Shigella pathogenicity island 2, was located downstream of selC, the site of insertion of pathogenicity islands in several other enteric pathogens. DNA sequence analysis revealed the presence of multiple insertion sequences upstream and downstream of the aerobactin genes and an integrase gene that was nearly identical to an int gene found in Escherichia coli O157:H7. SHI-2 sequences adjacent to selC were similar to sequences at the junction between selC and pathogenicity islands found in E. coli O157:H7 and in enteropathogenic E. coli, but the junctions between the island and downstream yic genes were variable. SHI-2 also encoded immunity to the normally plasmid-encoded colicins I and V, suggesting a common origin for the aerobactin genes in both S. flexneri and E. coli pColV. Polymerase chain reaction and Southern hybridization data indicate that SHI-2 is present in the same location in Shigella sonnei, but the aerobactin genes are not located within SHI-2 in Shigella boydii or enteroinvasive E. coli. Shigella dysenteriae type 1 strains do not produce aerobactin but do contain sequences downstream of selC that are homologous to SHI-2. The presence of the aerobactin genes on plasmids in E. coli pColV and Salmonella, on a pathogenicity island in S. flexneri and S. sonnei and in a different chromosomal location in S. boydii and some E. coli suggests that these virulence-enhancing genes are mobile, and they may constitute an island within an island in S. flexneri.


The acquisition of iron is a problem common to human bacterial pathogens. At least two different strategies are used by bacteria to compete with the host for the limited supply of this essential element. One mechanism is the expression of receptors for host iron complexes such as transferrin, lactoferrin and haemoglobin, which enable some pathogens to use these as a source of iron (Morton and Williams, 1990; Hanson et al., 1992; Cornelissen and Sparling, 1994). Another mechanism is the synthesis and secretion of siderophores, low-molecular-weight, high-affinity, iron-binding compounds that can remove iron from host sources and facilitate its uptake by the bacterium (Neilands et al., 1987; Crosa, 1989).

Among the Enterobacteriaceae, iron uptake by direct utilization of host sources and via siderophores has been observed. Growth on haem or haemoglobin as the sole iron source occurs in some isolates of Yersinia spp. (Hornung et al., 1996), E. coli (Law et al., 1992; Torres and Payne, 1997) and Shigella (Wyckoff et al., 1998). Siderophores are produced by most of these bacteria, but there is variation in the type of siderophore produced. Analysis of the Shigella spp., which multiply within colonic epithelial cells and produce dysentery, and clinical isolates of E. coli, which produce diseases ranging from mild diarrhoea to septicaemia and meningitis, revealed the presence of two different siderophore-mediated iron transport systems. The catechol siderophore enterobactin is produced by E. coli (Rogers, 1973; Earhart, 1996) and by some, but not all, Shigella (Perry and San Clemente, 1979; Payne et al., 1983). A second siderophore, aerobactin, is synthesized by Shigella flexneri and Shigella boydii (Lawlor and Payne, 1984). This hydroxamate is also synthesized by some Shigella sonnei and E. coli clinical isolates (Payne, 1988).

Mapping of several iron transport loci suggests that horizontal transmission of the genes has occurred. In S. dysenteriae type 1, genes encoding the haem transport system are contained on a 9.1 kb region located between two open reading frames (ORFs) of the E. coli K-12 map (Wyckoff et al., 1998). These genes are present in two distantly related lineages of the E. coli and Shigella group but not in other, more closely related strains, suggesting that there was more than one occurrence of acquisition or loss of these genes (Torres and Payne, 1997; Wyckoff et al., 1998). An iron acquisition system of Yersinia spp. that maps to the high pathogenicity island has also been found in some pathogenic E. coli (Bearden et al., 1998; Buchrieser et al., 1998; Schubert et al., 1998). Similarly, the aerobactin genes have characteristics of transmissible elements. They are found on plasmids, including pColV (Williams, 1979) and F1me (Colonna et al., 1985) in certain strains of E. coli and Salmonella, respectively, but are chromosomal in Shigella spp. (Lawlor and Payne, 1984) and in other E. coli and Salmonella isolates (McDougall and Neilands, 1984; Marolda et al., 1987).

This study was undertaken to map the aerobactin genes in Shigella and analyse the chromosomal region in which they are found in order to understand better the mechanism for the distribution and possible transmission of these genes among pathogenic bacteria.


Genetic organization of the S. flexneri aerobactin island

Cosmid clones containing the aerobactin biosynthesis and transport genes have been isolated previously from SA100 (Lawlor et al., 1987; Marolda et al., 1987). To characterize the genes surrounding this locus, the cosmid DNA was subcloned and the sequence determined. Analysis of the DNA sequence (GenBank accession no. AF097520) indicated the genetic organization shown in Fig. 1. Based on the similarities between this region and pathogenicity islands described below, we designated this 30 kb S. flexneri region SHI-2 for Shigella pathogenicity island 2.

Figure 1.

. Map of the S. flexneri pathogenicity island. The sequence of the DNA surrounding the aerobactin genes in S. flexneri SA100 was determined and the locations of the open reading frames inferred from the sequence analysis. The solid black boxes indicate sequences present in E. coli K-12 and S. flexneri. The patterned boxes indicate the size and position of ORFs within the island. o indicates reading frames whose direction of transcription is left to right, and r indicates those on the opposite strand. The DNA sequence between o35 and yicM has not been determined. The following letters below the line indicate relevant restriction sites: E, EcoRI; H, HindIII; P, PstI; S, SalI; K, KpnI; N, NcoI; Sc, SacI. Only the EcoRI and HindIII sites are shown in full. The two bars below the map indicate the probes used for Southern hybridization. The numbered arrows indicate the approximate positions of primers used for PCR amplification of the various junction fragments. Primer 10 is in yicK, which is found upstream of yicM in E. coli K-12 but is not present in S. flexneri.

The first gene within this cluster, int, has homology to the bacteriophage P4-like integrases (Table 1). The SHI-2 int is almost identical to the int gene recently described in a pathogenicity island, termed LEE for locus of enterocyte effacement, of E. coli O157:H7 (Perna et al., 1998). However, the homology between SHI-2 and E. coli O157:H7 LEE ends immediately 3′ of the int genes (Fig. 2B), indicating that these two strains do not contain the same island. Because the S. flexneri int gene has homology to the integrase associated with retronphage φR73 (Inouye et al., 1991) (Table 1), the strain was analysed for the presence of retronphage msDNA. No msDNA was detected in SA100, although it was detected in strains known to carry the retronphage (data not shown).

Table 1. . ORFs within the S. flexneri pathogenicity island (Ohtsubo and Matsutani, 1990; Ghosal et al., 1979; Hillen and Schollmeier, 1983; Krone et al., 1987). a. Location is given as nucleotide position, numbering from the first base in the island 3′ of selC.Thumbnail image of
Figure 2.

. A. Comparison of the junctions between selC and the pathogenicity island of S. flexneri SA100 and the LEEs from EPEC (McDaniel et al., 1995) and EHEC (Perna et al., 1998) strains. The 3′ end of the selC sequence of E. coli K-12 is also shown. The insertion point is considered the point at which the sequences in the strains containing islands diverge from E. coli K-12 and is shown as nucleotide 1 in the SA100 island sequence. B. Comparison of the 3′ end of the int sequences of SHI-2 and the EHEC LEE.

Downstream of int are a number of open reading frames (ORFs) with homology to insertion sequences and transposases (Fig. 1, Table 1[link]). These include copies of IS1, IS3, IS629, part of IS1150 and two copies of IS2, one on each side of the aerobactin locus (Fig. 1, Table 1[link]). The genes for aerobactin synthesis and the aerobactin receptor were mapped within this region, and sequencing within this operon indicated that the S. flexneri DNA sequence was essentially the same as that of the pColV-K30 aerobactin genes (Table 1). To determine whether the presence of aerobactin genes in the Shigella chromosome is the consequence of insertion of the ColV plasmid, the sequences flanking the aerobactin operons were compared. Polymerase chain reaction (PCR) and sequence analysis of the DNA flanking the aerobactin operons showed that S. flexneri differs from pColV-K30 (Fig. 1, Table 2[link]); there was no PCR amplification of sequences upstream of pColV iucA when primers derived from the S. flexneri sequence more than 1 kb upstream were used (Table 2). Amplification with a primer derived from the iucA sequence (Fig. 1, primer 7) and a primer beginning 426 bases upstream of the iucA start codon (primer 6) produced a 640 bp product in SA100 and pColV (Table 2). Primer pairs 5,7 and 4,7 amplified the expected 1879 bp and 2829 bp fragments in S. flexneri, but no products were observed with either primer pair when pColV DNA was used (Table 2). These PCR results are in agreement with earlier studies using Southern hybridization that showed conservation of the aerobactin genes, but not the flanking DNA, in S. flexneri and E. coli pColV (Lawlor et al., 1987; Marolda et al., 1987).

Table 2. . Conservation of sequences flanking iuc in aerobactin-producing strainsa. a. The presence of aerobactin genes was determined by Southern hybridization and confirmed by bioassays (Lawlor and Payne, 1984; Lawlor et al., 1987) for aerobactin synthesis and transport.b. Location of primer sequences shown in Fig. 1. + indicates amplification of a DNA fragment of the size predicted from the SA100 sequence; − indicates no amplification.Thumbnail image of

Association between aerobactin and colicin immunity genes

Although the genes adjacent to the S. flexneri aerobactin genes are not identical to those of pColV, there is a common feature to both regions. Both pColV and sequences upstream of the S. flexneri aerobactin operon encode immunity to colicin V (Table 3). The S. flexneri cosmid (pKLS971) and subclones pSAV3 and pJLG1 (Fig. 1) encode immunity to colicins V and Ib and to a colicin produced by S. flexneri SA100 (Table 3). An ORF, designated imm, is required for protection against the colicins; deletion of an NcoI fragment encompassing this gene (Fig. 1, pSAV3NcoΔ) eliminates protection against ColV, ColIb or the S. flexneri colicin (Table 3). Unlike the plasmid-encoded ColV and ColI immunity genes, however, the S. flexneri gene encoding colicin immunity is not closely linked to the colicin synthesis genes. The S. flexneri cosmid pKLS971, which contains ≈ 10 kb upstream and downstream of the immunity gene, was tested for colicin production, but no detectable colicin was produced by E. coli strains carrying this cosmid (data not shown). The colicin encoded by S. flexneri has not been characterized but, like colicins V and I, its receptor is the Cir protein. A cir mutant, JK458, was resistant to colicins V and I and the S. flexneri colicin, while the parent Cir+ strain JK360 was sensitive to all three (Table 3).

Table 3. . Identification of genes within Shigella spp. and E. coli encoding immunity or resistance to colicins. a. Strains producing the indicated colicins were stabbed into agar and overlaid with the strain to be tested for sensitivity; + indicates a zone of inhibition > 5 mm around the stab.Thumbnail image of

Additional aerobactin-producing strains of Shigella were tested for sensitivity to these colicins (Table 3). S. boydii 0-1392, like S. flexneri SA100, was not sensitive to colicins V and Ib or to the S. flexneri colicin. This may reflect the presence of immunity genes in S. boydii, or this strain may lack the receptor for these colicins. The S. sonnei strain, PB66, was sensitive to all three colicins and lacked sequences homologous to S. flexneri imm (data not shown). Therefore, there is a linkage between aerobactin genes and a colicin immunity gene in S. flexneri and pColV strains, but not in the other strains tested.

Association between SHI-2 and the selC tRNA gene

To obtain additional evidence about the possible mechanism by which the aerobactin genes might have spread within the Enterobacteriaceae, the junction between the aerobactin region and sequences common to the S. flexneri and E. coli K-12 chromosomes was analysed. The junction at the 5′ end of the island was found to be immediately downstream of the selC gene (Fig. 2). This location is the site of insertion of several pathogenicity islands in other enteric pathogens, including the LEE in enteropathogenic (EPEC) (McDaniel et al., 1995) and O157:H7 enterohaemorrhagic (EHEC) E. coli (Perna et al., 1998), Pai I of uropathogenic strains (Blum et al., 1994) and SPI-3 in Salmonella enteritidis (Blanc-Potard and Groisman, 1997). Comparison of the junction sequences among the E. coli and Shigella strains (Fig. 2A) indicates that the SHI-2 junction has homology to the EPEC and EHEC LEEs. The sequences immediately downstream of selC in S. flexneri are most closely related to those of the EPEC LEE, while the first gene in the island, int, is homologous to the EHEC int gene (Fig. 2A). The EHEC LEE appears to contain a deletion in the sequence between selC and int compared with the EPEC and S. flexneri islands (Fig. 2A). The S. flexneri DNA downstream of the int gene was not homologous to the sequence downstream of the EHEC int (Fig. 2B) or to any other sequences in the DNA database. Thus, this DNA sequence analysis indicates that the mechanism of insertion of the LEEs and SHI-2 may have common features, but the genes within the islands are distinct.

The junction at the 3′ end of the island was also analysed and compared with other islands inserted at selC. The presence of LEE in EPEC and EHEC strains is associated with deletions in yicK and yicL, which map downstream of selC in E. coli K-12 (McDaniel et al., 1995; Perna et al., 1998). S. flexneri also lacks yicK and yicL sequences, but the deletion in Shigella appears to be larger than in the EPEC or EHEC and extends into nlpA, the gene downstream of yicL (Table 4 and Fig. 1[link]).

Table 4. . Sequences present downstream of selC in Shigella and E. coli strains. a. Approximate locations of primer pairs 1,2 (selC–int ) and 1,10 (selC–yicK ) are shown in Fig. 1; yicL, nlpA and yicM were detected by amplification with primer pairs SVO209, SV6; ECS20, ECS17; and ECS21, ECS9 respectively. + indicates amplification of a DNA fragment of the expected size (except as noted); − indicates no amplification; ND, not determined.b. The presence of aerobactin genes was determined by Southern hybridization and confirmed by bioassays (Lawlor and Payne, 1984; Lawlor et al., 1987) to test for aerobactin synthesis and transport.Thumbnail image of

Presence of SHI-2 in other Shigella and E. coli strains

To determine whether SHI-2 was present in other Shigella and E. coli and to determine whether the chromosomal aerobactin genes are always found within SHI-2, we used PCR amplification and Southern hybridization of selected island and junction fragments. The results of these assays, which are shown in Tables 2 and 4 and Fig. 3 and discussed below, suggested the genomic organizations depicted in Fig. 4.

Figure 3.

. Presence of sequences homologous to S. flexneri int and iuc in Shigella and E. coli. Genomic DNA was digested with EcoRI, and the fragments were hybridized to a 866 bp internal fragment of the int gene (A) or a 2068 bp probe spanning the EcoRI site upstream of the aerobactin promoter (B). Locations of the probes and EcoRI sites are shown in Fig. 1. Lanes contain DNA from: lane1, S. flexneri SA100; lane 2, S. sonnei PB66; lane 3, S. boydii 0-1392; lane 4, EIEC strain 1107-81; lane 5, S. dysenteriae Ubon 378; lane 6, S. dysenteriae 0-4576; lane 7, E. coli W3110. Numbers to the left indicate sizes and positions of molecular weight markers.

Figure 4.

. Comparison of organization of aerobactin genes and their association with the selC locus in Shigella spp. and pathogenic E. coli. The locations of the aerobactin genes and adjacent sequences, int, selC,yicK, yicL, nlpA, yicM, were determined by Southern hybridization and PCR analysis. Patterned boxes indicate sequences conserved among different strains. Boxes with diagonal lines indicate the aerobactin genes, the hatched box represents int and dotted boxes indicate conserved ORFs adjacent to the aerobactin genes. The black boxes represent sequences found in E. coli K-12, including selC, yicK, yicL, yicM and nlpA. The dotted lines indicate distances of unknown length separating the genes of interest.

The presence of the aerobactin genes in Shigella and E. coli isolates was determined initially by Southern hybridization with an iuc probe, and by chemical and bioassays for aerobactin synthesis and transport (Table 2). Three primer pairs that amplify the DNA sequences upstream of the aerobactin operon in S. flexneri (Fig. 1, primer pairs 4,7, 5,7 and 6,7) and one that amplifies downstream sequence (Fig. 1, primer pair 8,9) were used to analyse the sequences flanking the aerobactin operon in those strains that contained aerobactin genes. As expected, the primer pair located within the aerobactin genes (pair 6,7) amplified DNA from all the aerobactin-producing strains (Table 2). The S. sonnei and S. boydii DNA was also amplified with primers that amplified the region between the aerobactin gene iucA and the upstream genes in the S. flexneri island (Table 2, primers 5,7). The DNA sequences of the fragments amplified by this primer pair were determined; the sequences shared > 90% homology among the three Shigella species (data not shown), indicating that the ORFs orf24 and rorf25 upstream of the aerobactin genes are conserved in these species. The third set of primers (primer pair 4,7) failed to amplify either the S. sonnei or the S. boydii DNA (Table 2). Primer 4 is derived from sequences within the IS1 upstream of aerobactin. Failure to amplify with this primer pair is consistent with earlier studies of Shigella spp. aerobactin genes, showing that the IS1 upstream of aerobactin in S. flexneri was absent in S. sonnei and S. boydii (Lawlor and Payne, 1984). The downstream primers amplified a fragment in the S. flexneri and S. sonnei only, indicating that orf35 was present in both these species, but not in the others. DNA from the aerobactin-positive E. coli strains could not be amplified with any of the primer pairs other than 6,7. Therefore, the DNA flanking the aerobactin genes in these E. coli strains is not the same as that in S. flexneri and S. sonnei (Table 2).

Additional PCR analyses were used to determine whether SHI-2, or any other pathogenicity island, was located downstream of selC in strains other than S. flexneri. Primer pair 1,2 (Fig. 1) amplified a 553 bp selC–int fragment when the int sequence was located immediately downstream of selC, thus identifying strains that potentially have SHI-2 at this site; primer pair 1,10 (Fig. 1) amplified the 736 bp selCyicK junction in strains such as E. coli K-12 that lack an island or other insertion downstream of selC. Among the aerobactin-positive strains tested, both S. flexneri and S. sonnei had int sequences downstream of selC (Table 4). The selC–int primers amplified a 553 bp fragment in both species, and the selC–yicK primers did not amplify these DNA (Table 4). Amplification with the selC–int primers indicates that sequences homologous to the S. flexneri int are located immediately downstream of selC in S. sonnei as well. In contrast, the aerobactin island was not found downstream of selC in S. boydii or in the aerobactin-positive EIEC strain 1107-81. No amplification was detected in S. boydii or EIEC using the primer pair that amplified the selC–int junction (Table 4), but the selC–yicK primers amplified a DNA fragment from these strains that was the same size as that from E. coli K-12 strain W3110 (Table 4). These data indicate that there is no insertion downstream of selC in these strains and that the aerobactin genes are located at a different site in the chromosome. Interestingly, the S. dysenteriae type 1 strains were found to contain an insertion at selC that has int sequences in common with S. flexneri. This strain does not contain aerobactin genes, however, so the insertion is not identical to that found in S. flexneri (Table 4).

To obtain additional information about the linkage between the aerobactin genes and the SHI-2 int gene, Southern hybridizations were performed using two probes, an internal sequence of the int gene and a fragment that spanned the EcoRI site upstream of iucA (Fig. 1). In S. flexneri, both the int and iuc probes hybridized to the 13 kb EcoRI fragment of the pathogenicity island DNA in S. flexneri, and the iuc probe hybridized to a second, 8.5 kb, EcoRI fragment that contains the aerobactin genes (Fig. 3, lanes A1 and B1). The same hybridization pattern was observed with S. sonnei (Fig. 3, lanes A2 and B2), although the larger of the two EcoRI fragments is slightly smaller in S. sonnei than in S. flexneri. Similarly, the int and iuc probes hybridized to fragments of the same size when the S. sonnei DNA was digested with BamH1, and the fragment was smaller than that observed with S. flexneri (data not shown). Because both the int and iuc probes hybridized to S. sonnei DNA restriction fragments of the same size, it is likely that the aerobactin and int sequences are physically linked and constitute a pathogenicity island in S. sonnei, as they do in S. flexneri. However, the islands in these two species are not identical, as indicated by the absences of an IS1 and imm gene upstream of the aerobactin genes in S. sonnei (Table 2 and data not shown). Furthermore, the junctions between SHI-2 and the downstream yic genes are different in S. flexneri and S. sonnei. nlpA is present downstream of the island in S. sonnei, whereas nlpA sequences are deleted in S. flexneri (Table 4).

Chromosomal DNA from S. boydii, S. dysenteriae, EIEC and E. coli K-12 was also hybridized with the S. flexneri int and iuc probes to determine whether these sequences were present and, if so, were the int and iuc genes linked in any of these strains (Fig. 3). Hybridization with the int sequence was found only in S. dysenteriae strains, while the iuc probe hybridized to sequences in S. boydii and the EIEC strain (Fig. 3). The iuc probe hybridized to a single EcoRI fragment of the EIEC strain, verifying that the DNA upstream of the EcoRI site in the aerobactin promoter region is different in S. flexneri and in the EIEC strain. Two bands were detected when S. boydii DNA was hybridized to the iuc probe (Fig. 3, lane B3). The larger band was smaller than that detected either in S. flexneri or in S. sonnei, and this fragment did not hybridize to the int probe. The hybridization and PCR data together suggest that the S. sonnei aerobactin genes are found at the same site and are on an island similar to the S. flexneri aerobactin island. The S. boydii aerobactin genes have sequences homologous to S. flexneri upstream, but the genes are located at different sites in the two species. In the pathogenic E. coli strains examined, neither the surrounding sequences nor the map locations are the same as in S. flexneri. Thus, the aerobactin genes, which are widespread among Enterobacteriaceae, are not restricted to a single chromosomal location and are found in a variety of different genetic contexts (Fig. 4). It is likely that these genes are mobile and, in S. flexneri and S. sonnei, they have become associated with a pathogenicity island that maps at selC.


Pathogenicity islands are common features of enteric bacterial pathogens (Groisman and Ochman, 1996). These are defined by Hacker et al. (1997) as regions of the chromosome that (i) carry virulence genes; (ii) are present in pathogenic strains and absent or sporadically distributed in less pathogenic strains; (iii) have a different G + C content from host bacterial DNA; (iv) occupy large chromosomal regions; (v) represent distinct genetic units; (vi) are associated with tRNA genes or insertion sequences; (vii) contain potential mobility genes such as IS elements or integrases; and viii) are relatively unstable. Studies reported here, along with previous analyses of the S. flexneri aerobactin genes, indicate that they are in a chromosomal region that has these characteristics. The aerobactin genes are found in some, but not all, strains of Shigella and E. coli and are more often associated with highly pathogenic strains (Lawlor and Payne, 1984; Valvano et al., 1986). The aerobactin genes are found within a 30 kb region just downstream of a tRNA gene, selC, and there are multiple IS elements and an integrase gene in the region. The int sequence has homology to int genes found in other pathogenicity islands inserted near selC. The G + C content of the portion of the SHI-2 that has been sequenced is 46%, slightly lower than the 51% G + C content of the rest of the chromosome. Omitting from this analysis the sequences of the IS elements, which may have transposed onto this region after acquisition of the island, yields a G + C content of 43%. Spontaneous deletions of the aerobactin genes have been observed in S. flexneri (Lawlor et al., 1987), indicating instability of the region. Thus, the aerobactin region appears to fit within the category of pathogenicity islands. This island is designated SHI-2 (Shigella pathogenicity island 2), as it is distinct from a previously described S. flexneri pathogenicity island, named she, that encodes a homologue of the IgA protease-like family of proteins (Rajakumar et al., 1998). Although the precise site of the she island has not been reported, it does not map to the same NotI fragment as selC (Rajakumar et al., 1998). SHI-2 contains the aerobactin operon and a colicin immunity gene and also has several novel ORFs. It is possible that one or more of these novel genes is required for pathogenicity, although S. flexneri virulence genes other than aerobactin have not been reported to map near selC.

Although SHI-2 is similar to other pathogenicity islands, it is not identical to any of the previously described islands found downstream of selC or elsewhere in the chromosome. The sequence most closely related to SHI-2 at the selC junction is the EPEC LEE pathogenicity island, while the SHI-2 int gene is closely related to the EHEC int gene. The homology to the EHEC pathogenicity island appears to be restricted to the int gene, with the sequences diverging immediately 3′ of the int coding region. The EPEC and EHEC LEEs include genes for attaching and effacing lesions (McDaniel et al., 1995), genes that are not found in S. flexneri SA100 (data not shown). One of the distinct phenotypes associated with SHI-2, aerobactin synthesis and transport, is absent from the EHEC strains and, when present in other pathogenic E. coli strains, it maps at a different location. Similarly, a colicin immunity gene has not been reported within other pathogenicity islands. Thus, while there are common features among these enterobacterial islands, the islands themselves are distinctly different.

Moss et al. (1999) have characterized SHI-2 in M90T, a serotype 5a strain of S. flexneri. The genetic organization and DNA sequence of the SA100 and M90T islands are almost identical at the left end, i.e. from selC through the aerobactin locus. However, the sequences downstream of iutA in the two strains are distinct, and the island is ≈ 30kb in SA100, rather than the 23.8 kb observed in M90T (Moss et al., 1999). Downstream of the aerobactin genes, the M90T island contains a copy of IS600 but no other ORFs, whereas the SA100 island contains a copy of IS2 and additional sequences of unknown function. Similarly, we found that the island in S. sonnei is closely related, but not identical, to the SHI-2 in S. flexneri SA100. The S. sonnei island lacks some of the sequences found in SA100, and the junctions between the islands and the downstream K-12-like sequences are distinct in the two species. These differences are indicative of the mosaic structure of this island and may indicate that the right-hand end of the island is unstable. Genes or insertion sequences both within and adjacent to the island may have been gained or lost in different isolates.

The aerobactin genes are found in at least three different locations in the E. coli–Shigella group: (i) they are plasmid-encoded in ColV strains (Williams, 1979); (ii) they map to the selC island in S. flexneri and S. sonnei; and (iii) they are found in at least one other chromosomal location in the S. boydii and EIEC strains. The mobility of the genes may be related to the presence of insertion sequences flanking the genes. Copies of IS1 are found on either side of the pColV aerobactin genes (McDougall and Neilands, 1984), and the S. flexneri aerobactin genes are flanked by IS2 elements. Movement of these genes does not appear to result from a simple transposition event, however, as the sequences immediately upstream and downstream of the genes and the position and type of associated insertion sequences are different in each case. In contrast to the divergence of sequences flanking the aerobactin operon, the DNA sequences of the aerobactin genes are highly conserved.

The synthesis of a colicin and the presence of a colicin immunity gene upstream of the S. flexneri aerobactin genes suggest a common origin of the S. flexneri and pColV aerobactin genes. The S. flexneri colicin has not been characterized, but it requires the same receptor, Cir, as that used by colicins V and I, and the putative colicin immunity gene renders E. coli insensitive to colicins V and I as well as to the S. flexneri colicin. Although the putative immunity ORF is approximately the same size as other immunity genes, no significant homology between this sequence and known colicin immunity genes was noted in either the DNA or amino acid sequence. Also, the position of the colicin synthesis and immunity genes relative to the aerobactin genes is different in pColV than in Shigella. The immunity genes are tightly linked to the corresponding colicin synthesis genes on pColV but map at a distance from the aerobactin cluster (Ambrozic et al., 1998). The lack of similarity in the immunity gene DNA sequences and the different genetic organizations of the regions surrounding the aerobactin genes indicate that the aerobactin and immunity genes were not transferred as a block between the ColV plasmid and the Shigella chromosome.

The observation that the aerobactin genes are found in a variety of different locations and are more highly conserved than the flanking sequences suggests that these genes are highly mobile and may be acquired by additional human or animal pathogens. In the case of S. flexneri, the genes have become associated with a pathogenicity island and, thus, have effectively created an island within an island. The surrounding island has a number of features in common with, but distinctly different from, islands found at the same site in other pathogens and, at least in S. dysenteriae, the island exists independently of the aerobactin genes.

Experimental procedures

Strains and plasmids

Strains and plasmids used in this study are listed in Table 5. Strains were routinely grown in L broth or on L agar. Antibiotics were added at standard concentrations to maintain plasmids. Bacteria were grown in a low-iron, Tris-buffered minimal medium as described previously (Lawlor et al., 1987) to assay for the production of siderophores.

Table 5. . Strains and plasmids. a. Crb, Congo red binding; Iuc, aerobactin biosynthesis; Iut, aerobactin transport.b. Clinical isolate from the Texas Department of Health.Thumbnail image of

DNA sequencing, sequence analysis and amplification of sequences by PCR

DNA sequencing was performed using an ABI Prism 377 automatic sequencer. The GenBank accession number for this sequence is AF097520. Routine DNA sequence analysis was performed using macvector (Olson, 1994) (Oxford Molecular). Homologies to proteins and genes were analysed using the blastxblosum62 and blastn programs, respectively, through the National Center for Biotechnology Information (Altschul et al., 1990; 1997; Gish and States, 1993). To determine the DNA sequence upstream of the sequence contained in pKLS971, inverse PCR was performed on SA100 genomic DNA. The DNA was digested with Sau3AI, purified using the Geneclean II Kit (BIO 101) and ligated under conditions favouring circularization. Inverse PCR was performed in a reaction containing 2.5 mM MgCl2, 5 U of Taq DNA polymerase (Qiagen) and 1 μM each of primers 2 and 3 (Fig. 1). The PCR reaction consisted of 30 cycles with 1 min at 94°C, 1 min at 65°C and 3 min at 72°C. The PCR product was isolated and sequenced directly.

Genomic PCR was performed under a variety of conditions depending on the length of the product and the melting temperature of the primers.

The sequences of the primers and their exact positions are (numbers indicate locations of the 5′ and 3′ nucleotides, numbering from the start of the island downstream of selC): primer 1, 5′-ATCCAGTTGGGGCCGCCAGCGGTCCCGGGCAG-3′ (in selC) (Blanc-Potard and Groisman, 1997); primer 2, 5′-CTCGCGAGCATCGGCTAGCGTTACATCGGGGT-3′ (486–455); primer 3, 5′-GTGAAGCCAGGAAACTCCTCGCTGCTGGAGGC-3′ (494–525); primer 4, 5′-TACCACGACCTCAAAGGCCG-3′ (11592–11611); primer 5, 5′-GGCTCG-

CCAATGCCCTGATA-3′ (12542–12561); primer 6, 5′-CAGCCCTAGCAGGGTAAAG-3′ (13781–13800); primer 7, 5′-GCCGTGACCAGATGAGCAGG-3′ (14402–14421); primer 8, 5′-TTCACCGTGCCATAAGAGCC-3′ (in o35); primer 9, 5′-TATGGAGGTATGCAGGCTGC-3′ (in o35); primer 10, 5′-GTGAGATCAAGTATTTTTGATGGAGTGGTAGC-3′ (in E. coli yicK, not present in SA100) (Blanc-Potard and Groisman, 1997).

Primer ECS9, 5′-ACAGAACCTGCTGCAATG-3′ (in yicN ); primer ECS17, 5′-GAATTCTGCTGGCAGGTT-3′ (in nlpA); primer ECS20, 5′-CGACTTCGGGTGATTGAT-3′ (in nlpA); primer ECS21, 5′-CGAAATGCCTAAATCCTG-3′ (in yicM); primer SVO209, 5′-TGCCATCTTCCTTGGTATTCTCTGTGGTATCG-3′ (in yicL); primer SV6, 5′-CGATAATCGTCGGTGAGAGGAATTGCAGCA-3′ (in yicL).

Colicin and siderophore assays

To detect colicin synthesis or sensitivity, L plates were stabbed with a colony of the colicin-producing strain and incubated overnight. The plates were then inverted over chloroform-saturated Whatman no. 1 disks for 15 min, dried upright for 30 min and overlaid with 3 ml of 0.7% agar containing 100 μl of a fully grown culture of the colicin indicator strain. The plates were incubated at 37°C overnight, and colicin sensitivity was determined by the presence of a clear zone around the site of the stab.

The presence of hydroxamate siderophores was detected by the ferric perchlorate assay (Atkin et al., 1970). The synthesis and transport of aerobactin was confirmed by bioassays as described previously (Lawlor et al., 1987).

Southern hybridization

Genomic DNA was isolated with DNAzol reagent (Molecular Research) as described by the manufacturer. Southern hybridizations were performed according to the procedure of Maniatis et al. (1982). Probe labelling, hybridization and detection with CSPD reagent were performed as described in the Genius II System (Boehringer Mannheim).


This work was supported by PHS grant AI16935. We thank Tom Whittam, James Kaper, Richard Calendar, Kathy Lawlor, Jennifer Gordon and Erich Six for providing strains or plasmids. Jeremy Moss, Arturo Zychlinsky and Eduardo Groisman generously shared unpublished data so that the two manuscripts could be published together. We also thank Erich Six for helpful discussions and suggestions. León Eidels, Laura Runyen-Janecky, Elizabeth Wyckoff and Melissa Mann provided expert editorial guidance.