|Home | About | Journals | Submit | Contact Us | Français|
Shiga toxin 2 (Stx2), one of the principal virulence factors of enterohemorrhagic Escherichia coli, is encoded by 933W, a lambda-like prophage. 933W prophage induction contributes to Stx2 production, and here, we provide evidence that Dam methyltransferase is essential for maintenance of 933W lysogeny. Our findings are consistent with the idea that the 933W prophage has a relatively low threshold for induction, which may promote Stx2 production during infection.
Enterohemorrhagic Escherichia coli (EHEC) serotype O157:H7 bacteria are a major cause of food- and waterborne illness in the United States, Europe, and Japan (5, 7, 25). These bacteria are highly infectious and produce potent Shiga toxins that account for the most severe clinical manifestations of EHEC infection, including the hemolytic uremic syndrome. In the sequenced EHEC strain EDL933, the stx1 and stx2 genes are located within lambdoid prophages designated 933J (also referred to as 933V ) and 933W, respectively (21). The former prophage is thought to be defective (20). Induction of Stx-encoding prophages can markedly increase Stx production and trigger phage-mediated cell lysis, thereby providing a mechanism for Stx release (18, 27). Thus, understanding processes that regulate Stx prophage lysogeny can provide insight into EHEC pathogenicity. In E. coli K-12, dam is known to influence lambda lysogeny. Here, we investigated the influence of dam on the maintenance of Stx-encoding prophages.
The dam gene encodes a DNA methyltransferase that methylates the adenine in the sequence GATC in double-stranded DNA (9, 28). Dam methylation in nonpathogenic E. coli K-12 has been studied extensively (9, 28). Among other traits, dam mutants exhibit single-stranded interruptions in their DNA (12). These single-strand interruptions are a consequence of MutHLS DNA mismatch repair and are converted into double-strand breaks which require homologous recombination to restore genomic integrity (11, 19). The presence of these breaks induces the SOS response (14, 22), likely explaining the increased induction of prophage lambda observed in E. coli K-12 dam lysogens (13). Unlike E. coli K-12, EHEC strain EDL933 contains several putative Dam-like methyltransferases, but Campellone et al. (2) recently demonstrated that deletion of a chromosomal dam gene that is 99% identical to that in E. coli K-12 (21) was sufficient to abrogate DNA methylation in this strain.
We used lambda Red recombination to inactivate the dam gene of the prototype EHEC strain EDL933, as recently described for construction of dam mutants in TUV93-0, a stx1- and stx2-deficient derivative of EDL933 (2). Unexpectedly, the frequency of dam mutants in the EDL933 background was much lower than that which we observed in the TUV93-0 background. Most of the candidate EDL933 dam mutants proved to be false positives. For example, when fragments of plasmid pKM212 (Δdam::kan) (2) were used to transform EDL933, only 1 out of the 10 Kanr colonies tested by PCR showed deletion of the dam gene (Fig. (Fig.1).1). The other nine candidates were presumably plasmid transformants. We chose to further analyze this Δdam::kan strain (designated GM7284) as well as KM69, an independently derived EDL933 dam deletion mutant which was made in two steps. First, a SacI-SphI digest of pKM213 (Δdam::cat-sacB) was used to generate a deletion of the dam gene (2), and second, a SacI-SphI digest of pKM210 was used to generate an in-frame (markerless) deletion of the dam gene by selection of a sucrose-resistant Cams colony (indicative of loss of the cat-sacB cassette ). The deletion was confirmed by PCR (2). These mutants lacked methylation in GATC sequences, as determined by digestion with the restriction endonucleases DpnI and Sau3A, which do not cut unmethylated DNA (data not shown). As has been observed in E. coli K-12 dam mutants, each of these EDL933 dam mutants showed an increase in frequency of spontaneous mutation to rifampin resistance relative to EDL933 and a false-positive candidate. The fractions of overnight cultures showing spontaneous mutation to rifampin resistance (per 108 cells, ± standard errors) are as follows: for EDL933/pTP223, 0.125 ± 0.78; for KM68 (Δdam::cat sacB), 2.67 ± 0.56; for KM69 (Δdam), 2.83 ± 0.81; for GM7284 (Δdam::Kn), 2.83 ± 0.81; and for false-positive candidate no. 1 (from Fig. Fig.1),1), 0.121 ± 0.004. (Determinations were done in triplicate.) Also, as reported for E. coli K-12 dam mutants, the EDL933 dam mutants exhibited heterogeneous morphology with many cells showing filamentation (2).
We initially suspected that dam might influence 933W lysogeny because we found that there was no detectable Stx2 in enzyme-linked immunosorbent assays (23, 24) of mitomycin C-treated cell lysates of the dam mutant strains (data not shown). In contrast, Stx1 was detectable in these lysates. Furthermore, PCR analyses revealed the loss of stx2 but not stx1 from both dam mutants; both stx1 and stx2 were present in the false-positive-candidate dam mutants mentioned above (Fig. (Fig.1).1). We performed microarray studies to investigate whether the 933W prophage transcriptome was lost from KM69. We plotted microarray signals from EDL933 and KM69 (Δdam) against each other, which revealed a cluster of genes (bottom right of Fig. Fig.2)2) that are expressed in EDL933 but not KM69 (Δdam), and 55/56 of these are located in bacteriophage 933W (the other is dam). We confirmed our suspicion that the prophage was excised from KM69 by using a PCR assay that demonstrated that the attB site in this strain was unoccupied; furthermore, we were unable to amplify intW-wrbA from KM69 (data not shown). Thus, the data confirm the loss of bacteriophage 933W, and not just the stx2 gene, in KM69 and suggest that dam may be required to maintain 933W lysogeny.
Our understanding of the relationship between Dam methylation, SOS induction, mismatch repair, and prophage induction is much more extensive for E. coli K-12 than for EDL933. E. coli K-12 was used, therefore, to test the hypothesis that dam mutants are nonviable due to enhanced 933W prophage sensitivity to induction by mismatch repair-induced SOS signaling. A 933W lysogen of E. coli K-12 strain MM294 (16) was isolated using a phage lysate from EDL933 and its identity confirmed by PCR using primers specific for the 933W immunity region (data not shown). An MM294(lambda+) lysogen was also constructed for use as a control. The lambda phage used to prepare this lysogen produced turbid plaques at both 30°C and 42°C, and the MM294(lambda+) lysogen was inducible with mitomycin C treatment, indicating that the lambda phage used did not harbor cI857 or ind mutations. The MM294(lambda+) and MM294(933W) strains were transduced to Camr with a P1vir lysate propagated on a dam-13::Tn9 donor. The number of transductants with MM294(933W) was more than 1,000-fold lower than that with MM294(lambda+) or with nonlysogenic MM294 (data not shown). The four Camr transductants obtained from MM294(933W) were shown to contain Tn9 in the dam gene and characterized further (Table (Table1).1). Two of the dam::Tn9 transductants (GM7259 and GM7261) did not produce bacteriophage plaques either spontaneously or after mitomycin C induction, did not contain the 933W prophage, as determined by PCR, and were sensitive to 2-aminopurine (2-AP), indicating an active mismatch repair system. Ordinarily, sensitivity to 2-AP is indicative of dam mutants, and these strains can become resistant when DNA mismatch repair is abrogated by mutation in the mutH, mutL, or mutS genes (6). Thus, these two transductants appear to be dam mutants that have lost the 933W prophage. The third transductant (GM7258) was 2-AP sensitive and contained the 933W prophage, but this phage produced pinpoint plaques, suggesting that it had acquired a mutation in either the host or the prophage, compromising phage development. The fourth transductant (GM7260) produced wild-type levels of spontaneously induced phage but was resistant to 2-AP, suggesting that mismatch repair was compromised in this strain. Strain GM7260 has a strong mutator phenotype which is complemented by a plasmid-borne mutS+ gene. Inactivation of mismatch repair in dam mutants abolishes the number of detectable double-strand breaks (19), the processing of which signals SOS induction. Thus, none of the four Camr transductants isolated proved to have both a fully functional mismatch repair system and a wild-type 933W prophage, suggesting that dam is essential for maintenance of 933W lysogeny.
The data obtained with both EDL933 and E. coli K-12 (933W) are consistent with the hypothesis that the level of SOS induction in dam mutants generated by mismatch repair-mediated DNA breaks is sufficient to induce the 933W prophage, leading to phage-mediated cell lysis. In contrast, in E. coli lambda+ lysogens, only a small fraction of the dam population shows full SOS induction (14) and undergoes prophage induction and cell lysis (13), consistent with our finding that the transduction frequencies were the same in the lambda lysogen and the nonlysogen. Thus, our observations suggest that 933W induces more easily than lambda at a given level of SOS induction. Indeed, Livny and Friedman (8) showed that at a given level of inducing signal, a greater fraction of lysogens with Stx-encoding prophages are induced than lysogens with non-Stx-encoding prophages. A possible explanation for the “hair-trigger” induction of prophage 933W is the observation that, unlike other lambdoid phages, which have three operators on the left side (OL), prophage 933W has only two such operators (26). The lack of a third OL in 933W precludes a lambda-like model in which interactions between cI repressors at OR and OL are critical for repression (4).
Dam has also been reported to influence the lysogeny of other prophages. Increased excision of the defective prophage ST64B from a dam mutant of Salmonella enterica is also due to enhanced SOS regulon expression (1). In this case, however, there was also a direct effect on the transcription of genes putatively involved in phage induction due to the presence of dam sites in the regulatory regions of these genes. Increased prophage excision in dam mutants may not be a general phenomenon, however, as Alonso et al. (1) found that of four prophages in S. enterica tested, only ST64B was affected.
The virulence of S. enterica and some other pathogenic bacteria is greatly reduced by a dam mutation and has led to the proposal that dam strains can be used as vaccines (3, 10). An alternative approach would be to inactivate the Dam methyltransferase in vivo by the use of a small molecule as a therapeutic. A similar strategy for EHEC would seem unwise, since inhibition of Dam in intestinal EHEC would likely lead to both increased induction of prophage 933W and Shiga toxin production.
Finally, our conclusion that loss of dam leads to inviability of EDL933 through prophage induction is a caution for studies where the ability to delete a particular gene is often used to determine if the gene is essential or not to the viability of the organism. In this case, Dam does not perform an essential function but the cells die due to an indirect cause.
Complete microarray data for these strains are available at http://users.umassmed.edu/martin.marinus/arrays/index.html.
This work was not supported by any funding agency. K.C.M., A.L.O., and M.G.M. volunteered their time to perform experiments and used scarce overhead funds to buy needed supplies.
We thank W. D. Rupp, M. Meselson, and J. Leong for providing bacterial strains and bacteriophages and J. Leong for suggestions that improved the manuscript.
Published ahead of print on 2 November 2007.