Roles of DNA Adenine Methylation in Regulating Bacterial Gene Expression and Virulence

Background.

Based on the organization of 10 amino acid domains present in MTases, Dam is classified in the α group (Fig. 1) (46). Dam homologues are widespread among enteric bacteria, including Escherichia coli, Salmonella spp., Serratia marcescens, Yersinia spp., and Vibrio cholerae, cholerae, but are also present in disparate genera, including Neisseria among others (Table 1 and Fig. 1). Dam methylation is not essential for viability of E. coli (3); however, recent data indicate that Dam is an essential gene in Vibrio cholerae and Yersinia pseudotuberculosis (31, 45), similar to results showing that the CcrM methylase is essential in Caulobacter crescentus, Brucella abortus, Rhizobium meliloti, and Agrobacterium tumefaciens (32, 70, 73, 93). Certain α-group methylases, including the DpnII methylase, share significant sequence identity with Dam (32% for DpnII) and methylate GATC sites like Dam but are part of restriction-modification systems (Fig. 1).

Functions.

Adenine methylation can alter the interactions of regulatory proteins with DNA, either by a direct steric effect or by an indirect effect on DNA structure (18, 61, 62). Initial studies with dam mutants showed that Dam regulates the expression of certain genes in E. coli including trpR (60), Tn10 transposase (68), and dnaA (13) as well as phage genes including mom of Mu (24). Methylation of a GATC site(s) within the consensus RNA polymerase binding site inhibits (trpR and Tn10 transposase) or enhances (dnaA) transcription, by altering the interaction with the transcription apparatus.

As discussed above, methylation can alter the affinity of regulatory proteins for DNA. Conversely, DNA binding proteins have been shown to inhibit methylation of specific DNA sequences. For example, the SeqA protein involved in the timing of DNA replication binds specifically to hemimethylated DNA sequences near the origin of replication (oriC), thereby sequestering oriC from Dam methylation for a part of the cell cycle and maintaining it in a hemimethylated state (41). Other regulatory proteins bind nonmethylated DNAs with highest affinity, protect specific DNA sequences from methylation, and form DNA methylation patterns (DMPs), which are present in certain eukaryotes as well (see Fig. 2) (28). DMPs are formed when regulatory factors bind to DNA target sites that overlap or are near methylation sites and inhibit their methylation (12).

Regulation by hemimethylation.

Work carried out with E. coli Dam indicates that it acts as an efficient de novo methylase, methylating both nonmethylated and hemimethylated GATC sites with similar efficiency (82). Dam plays an important role in regulating the timing and targeting (51) of a number of cellular functions including DNA replication (9, 34, 41, 72), segregation of chromosomal DNA (52, 58), mismatch repair (29, 48, 71), and transposition (19, 66, 68, 77, 89). In all of these events, hemimethylated GATC sites, present immediately following DNA replication, control the binding of proteins to specific DNA target sites. For example, DNA replication is controlled in part by SeqA, which binds specifically to hemimethylated GATC sites near the origin of replication and delays their methylation (34, 41). Segregation of newly replicated DNA may occur by binding of hemimethylated DNA to membrane-bound factors. In methyl-directed mismatch repair, MutH binds to hemimethylated DNA and cleaves the nonmethylated strand (1). Certain transposases, including Tn10 transposase, bind with highest affinity to hemimethylated binding sites, limiting transposition to a time immediately following DNA replication (68).

Regulation by DNA methylation patterns.

Dam also plays pivotal roles in controlling gene expression by the formation of DMPs. DMPs have long been known to be present in eukaryotes and appear to regulate gene expression (5, 92). The first reported DMPs which regulate gene expression in prokaryotes are within the pyelonephritis-associated pilus (pap) operon of uropathogenic E. coli (7).

Most nonmethylated GATC sites are found within noncoding regions that likely have a regulatory function. For example, the GutR repressor binds in the upstream regulatory region of the glucitol (gut) operon, blocking methylation of a GATC site designated GATC −44.5 within the GutR binding domain and forming a specific DMP (85). In the presence of glucitol, GutR no longer blocks DNA methylation, indicating that it no longer binds gut regulatory DNA. This is an example of environmental control of a DMP. Notably, although catabolite gene activator protein (CAP) also binds to a site overlapping GATC −44.5, it does not protect this site from methylation.

It appears that most, if not all, DMPs are formed by the binding of regulatory proteins such as GutR (85), Lrp (12), histone-like nucleoid-structuring protein (H-NS) (91), and OxyR (22) to upstream regulatory DNA sequences. In these examples, purified proteins have been shown to block methylation of GATC sites that are contained in or near the DNA recognition sequence in vitro. Inhibition of methylation could occur by direct steric occlusion of Dam binding or by alterations in DNA conformation which change the configuration of the Dam target (GATC) site (61).

Analysis of the E. coli chromosome has shown that there are at least 50 GATC sites that are stably undermethylated (67, 75). The locations of these sites may vary depending upon environmental conditions (23), which can alter the expression and/or binding of regulatory factors that bind to DNA and specifically block DNA methylation. Analysis of the Salmonella chromosome by pulsed-field gel electrophoresis has shown that, similar to E. coli, specific DNA methylation patterns are present (26). Although many nonmethylated GATC sites have been identified in E. coli, methylation in only a subset has been shown to control the expression of linked genes. For example, GutR blocks methylation of GATC −44.5 in the gut operon, but methylation of GATC −44.5 does not alter the binding of GutR to gut DNA in vitro nor does it alter gut expression in vivo (85). The CarP regulatory protein and integration host factor (IHF) protect a GATC site in the carAB operon, but it is not clear if methylation controls CarAB expression (16). In contrast, methylation of regulatory GATC sites in the pap and agn operons directly controls expression of Pap pili and the Ag43 outer membrane protein, respectively (12, 22, 27). This occurs by reduction of the affinities of the Lrp and OxyR proteins for pap and agn regulatory DNAs, respectively. In pap and related methylation-controlled operons, methylation of two GATC sites spaced 102 bp apart regulates Lrp binding, whereas in agn, methylation of three closely spaced GATC sites inhibits OxyR binding (22). In these instances, there is a mutual competition between the methylase and the DNA binding protein, forming a DMP which heritably controls gene expression and provides a form of cellular memory.

E. coli.

Dam regulates the expression of a large group of pilus operons that play important roles in virulence in urinary tract infections (e.g., Pap, Prf, and S pili) and diarrheal diseases (e.g., Afa, CS31a, and K88 pili) (30, 36, 49, 53, 87). All of these pilus-adhesin operons share common regulatory features including control by Lrp, the presence of a promoter proximal GATC site (GATC^prox) located within one set of Lrp binding sites and a promoter distal site (GATC^dist) within a second set of Lrp binding sites, inter-GATC spacing of 102 or 103 bp, and a homologue to the PapI regulatory protein that binds to Lrp, increasing its affinity for the promoter distal Lrp sites which then helps activate transcription (7, 8, 11, 12, 33, 55, 56, 86) (Fig. 2).

In the pap operon, DNA methylation directly regulates the switch between pilus expression (phase ON) and nonexpression (phase OFF) by dictating the binding of Lrp (Fig. 2). At low PapI levels, Lrp binds with high affinity to promoter proximal sites, blocks transcription from the papBA promoter, and inhibits methylation of GATC^prox (90). However, the promoter distal GATC^dist site is not bound by Lrp and thus becomes methylated. Methylation of GATC^dist inhibits movement of Lrp to the distal set of sites and thus is presumed to lock cells in the phase-OFF state until DNA replication generates a hemimethylated GATC^dist site which binds Lrp with a higher affinity (36). Methylation of the promoter proximal GATC^prox site is required for the expression of Pap pili (12). Mutations in the Lrp binding sites near GATC^prox result in a phase-locked ON transcription phenotype that is Dam and PapI independent (55). These results indicate that methylation of GATC^prox may help displace Lrp from its promoter proximal DNA binding sites that overlap GATC^prox, with the aid of PapI (33). Binding of Lrp-PapI at the promoter distal GATC^dist site blocks its methylation, forming a DNA methylation pattern that is characteristic of cells expressing pili. The cell environment controls the pap DNA methylation pattern since in poor carbon sources the cyclic AMP level is high and stimulates PapI expression via a cyclic AMP-CAP binding site in the pap regulatory region (2). PapI facilitates movement of Lrp to the GATC^dist site, and Lrp blocks methylation of GATC^dist and helps activate pap transcription (84). Other members of the Pap family, including sfa (87), daa (87), fae (30), and clp (49) and pef in Salmonella serovar Typhimurium (53), have been shown to be regulated by DNA methylation patterns as well (36).

Salmonella.

Torreblanca and Casadesus first identified genes regulated by Dam in Salmonella serovar Typhimurium using a genetic approach (78). One of these genes mapped to the pSLT virulence plasmid and was later shown to be finP, which expresses an antisense RNA controlling the F-type pili required for conjugative plasmid transfer (79). The result is that under conditions of low levels of Dam, transfer of the Salmonella pSLT plasmid is elevated. The physiologic connection between Dam-controlled pilus expression and mating is not yet clear but could function to coordinate mating with virulence plasmid replication or to enable environmental control of mating (79). Dam also regulates the expression of plasmid-encoded fimbriae (Pef) encoded by pSLT by a mechanism that shares features with pap (53). Work from Heffron's laboratory indicates that Pef may play a role in Salmonella virulence (83). Like pap, expression of Pef fimbriae is turned off in the absence of Dam since methylation of a promoter-proximal GATC site of pef is essential for transcription.

Recently, Dam was shown to be essential for the virulence of Salmonella serovar Typhimurium in a murine model of typhoid fever (21, 25, 26). Dam⁻ Salmonella shows reduced M-cell cytotoxicity and invasion of enterocytes but appears to grow normally within cells (21). In the absence of Dam, serovar Typhimurium is avirulent when given orally and intraperitoneally and fails to kill mice at 10,000 times the lethal dose required to kill half of the animals (LD₅₀). Dam-deficient Salmonella colonizes Peyer's patches in a manner similar to that of wild-type bacteria but attains only very low numbers in systemic tissues and is totally cleared from mice after about 4 weeks. The failure of dam mutants to cause disease is not the result of defects in mismatch repair since mutS and mutL serovar Typhimurium is fully virulent (21, 26).

Why is Dam⁻ Salmonella avirulent? We hypothesize that dam mutant Salmonella is markedly attenuated as a result of dysregulation of gene expression. Dam-deficient Salmonella serovar Typhimurium up-regulates the expression of over 35 genes that are induced during infection (26), including spvB, a cytotoxin which causes apoptosis of macrophages (39). In contrast, Dam positively regulates the secretion of the SipA, SipB, and SipC proteins coded for by the Salmonella pathogenicity island type 1 virulence locus (21). Thus, in the absence of Dam, virulence factors such as SpvB are predicted to be overexpressed (the SpvB protein was recently shown to be ectopically expressed at very high levels in the absence of Dam [D Guiney, unpublished results]), whereas other factors such as SipABC are underexpressed. We hypothesize that this combination of overexpression and underexpression of virulence proteins inhibits virulence (25, 45). If this is correct, then overexpression of Dam might also block virulence since negatively regulated factors such as SpvB would be underexpressed and positively regulated factors such as SipABC would be overexpressed. In fact, overexpression of Dam reduces the virulence of Salmonella serovar Typhimurium 10,000-fold (25). As predicted, the protein profiles from Dam⁻ and Dam-overproducing Salmonella strains are different from each other and from wild-type Salmonella (25). In addition, Dam⁻ Salmonella releases a high level of outer membrane vesicles, suggesting an instability defect in the outer membrane. Since OmpA is a highly immunogenic protein, vesicle release by dam mutants might contribute to their efficacy as live attenuated vaccines (J. Casadesus, personal communication). These data support the hypothesis that Dam is a global regulator of virulence genes in Salmonella and that Dam levels regulate virulence.

Both Dam-deficient and Dam-overproducing Salmonella serovar Typhimurium strains are highly effective vaccines against salmonellosis. As few as 90 Dam⁻ bacteria administered intraperitoneally provide significant protection (21) and oral vaccination of mice with 10⁹ Dam⁻ Salmonella bacteria completely protects them from challenge with 10⁹ wild-type Salmonella bacteria (10,000 times the oral LD₅₀) (26). Vaccination with Dam⁻ Salmonella protects mice against challenge with other S. enterica serovars including Enteritidis and Dublin (25). Additionally, Dam⁻ Salmonella conferred cross-protective immunity in chickens (20a). This cross-protection could occur as a result of aberrant expression of a number of virulence determinants of serovar Typhimurium, some of which might be shared with other Salmonella serovars. The dysregulation of expression of Salmonella virulence determinants not only could disrupt the normal pathogenic cycle but also may enable the host immune system to mount an effective response. This response could be elicited to Dam-controlled bacterial antigens which are normally under temporal and spatial control and not easily detected. Consistent with this hypothesis, S. enterica serovar Typhimurium overproducing Dam conferred significant protection against homologous Salmonella (25) although not to the same extent as Dam-deficient Salmonella. This discrepancy could be due, in part, to the differences in antigen expression observed between these two vaccines (see above). Dam is also essential for full virulence of S. enterica serovar Enteritidis (25), which can invade the yolk sac and contaminate eggs. Because Dam⁻ S. enterica serovars Typhimurium and Enteritidis are highly attenuated, it seems likely that Dam might also be essential for virulence of S. enterica serovar Typhi, the causative agent of typhoid fever.

Other pathogens

Although Dam from E. coli and Salmonella spp. is not essential for growth, dam is an essential gene in V. cholerae and Y. pseudotuberculosis (31) (Table 1). Since V. cholerae has two chromosomes (81), it is possible that Dam plays the same roles as those in E. coli and Salmonella, but additionally, Dam may coordinate the timing and segregation of the two chromosomes. Recent data indicate that the virulence of Dam⁻ Erwinia chrysanthemi is greatly reduced for African violets and lettuce, two of its hosts (C.-H. Yang and N. Keen, unpublished data). Thus, Dam is important for the virulence of both animal and plant pathogens.

Overproduction of Dam inhibits the colonization of V. cholerae based on a suckling mouse model (31). Moreover, overexpression of Dam in Y. pseudotuberculosis greatly attenuates virulence in a murine model (>6,000-fold) and alters the protein expression profile. Oral immunization of mice with Dam-overproducing Y. pseudotuberculosis protects against challenge with at least 1,000 times the LD_50. Thus, it appears that the Dam-based vaccination strategy developed using S. enterica serovar Typhimurium can be extended to other enteric pathogens.

In contrast to the essentiality of Dam for virulence and/or growth in enteric bacteria, all pathogenic isolates of group B Neisseria meningitidis have a mutation within dam (dam gene replacement, or drg) and thus do not express Dam (14). These meningococci are deficient in methyl-directed mismatch repair, as expected, and show high rates of phase variation of a neuraminic acid capsule controlled by the polysialyltransferase (siaD) gene. Frameshift mutations within a poly-deoxycytosine repeat in the siaD gene coding sequence regulate the on-off expression of capsule. A high rate of capsular phase variation was observed only in dam mutant meningococci. Thus, selection for Dam-deficient meningococci may occur as a result of increased phase variation rates.

We are only beginning to understand the roles that Dam plays in regulating the interactions between bacterial pathogens and their hosts that contribute to virulence. Many bacterial pathogens, including S. enterica serovar Typhi, Shigella spp. (dysentery), pathogenic E. coli including O157:H7 (hemolytic-uremic syndrome), Haemophilus influenzae (pneumonia and otitis media), and Legionella pneumophila (pneumonia), have been reported to contain dam homologues and/or Dam activity (25, 38, 45) (Table 1 and Fig. 1). It seems likely that Dam is important for the pathogenesis of a number of diseases. An important next step will be to determine the mechanism by which Dam controls virulence and whether there are any unifying concepts that can be gleaned from the analysis of genes regulated by Dam in a number of pathogens.

Alteration in DMPs caused by fluctuations in the binding of regulatory proteins such as Lrp or OxyR to DNA target sites can be viewed as a novel mechanism for global regulation. The relevance of DMPs to pathogenesis could be determined by identification of all Dam-regulated virulence genes in a pathogen by microarray analysis and matching them to the nonmethylated GATC sites in the chromosome (75, 88). The first part of this analysis has recently been carried out in E. coli, with the data indicating that the mRNA levels of about 50 to 60 genes are significantly altered in the absence of Dam (N. Bourquard, G. W. Hatfield, and D. A. Low, unpublished data).

Background.

In contrast to Dam, CcrM is classified in the β group of DNA methylases (Table 1). CcrM was originally discovered as a cell cycle-regulated MTase in Caulobacter (95). CcrM is 49% identical to the HinfI MTase from H. influenzae, which shares the same recognition sequence but is part of a retriction-modification system. CcrM homologues are found in the α-proteobacteria, including the plant pathogen A. tumefaciens and symbiont Rhizobium meliloti and B. abortus (64), which causes brucellosis in cattle and humans (70). All CcrM homologues tested appear to be essential for cell viability, similar to the essential roles of MTases in mammalian cells and Dam in V. cholerae and Y. pseudotuberculosis. Moreover, based on complementation analysis of Caulobacter and Rhizobium, the CcrM proteins are functionally interchangeable (64).

Functions.

The CcrM methylase from C. crescentus plays an essential role in the cell cycle of this developmentally programmed bacterium. CcrM, in contrast to Dam, appears to be a “maintenance” methylase with preference for hemimethylated DNA over nonmethylated DNA (4). CcrM methylase is essential for the viability of α-proteobacteria including C. crescentus, B. abortus, R. meliloti, and A. tumefaciens (32, 70, 73, 93).

Initial evidence that CcrM regulates gene expression came from analysis of the ccrM gene itself. Methylation of GAnTC sites within the ccrM promoter inhibits transcription. Mutation of these CcrM target sites prevents shutdown of methylase gene transcription after cell division (74). Analysis of the complete genome sequence of C. crescentus showed that CcrM target sites were less abundant than predicted from random occurrence and had a bias to intergenic regions (54). These data support the hypothesis that CcrM is a global regulator of gene expression. Indeed, recent microarray analysis by Lucy Shapiro and colleagues indicates that approximately 100 genes in C. crescentus are affected by depletion of CcrM (L. Shapiro, unpublished data).

Roles in virulence.

CcrM may play an important role in the virulence of B. abortus based on the analysis of bacteria overexpressing the methylase (70). Overexpression of CcrM on recombinant plasmids inhibited growth of Brucella within murine peritoneal macrophages. This attenuation did not appear to be due to alterations in bacterial growth rates or alterations in cell morphology or DNA replication initiation at lower CcrM expression levels. It thus appears that the defect in intracellular replication is due to some other effect of CcrM on cellular function such as regulation of a gene(s) required for adapting to the intracellular environment (64).