Advertisement

The Salmonella selC locus contains a pathogenicity island mediating intramacrophage survival

Anne‐Béatrice Blanc‐Potard, Eduardo A. Groisman

Author Affiliations

  1. Anne‐Béatrice Blanc‐Potard1 and
  2. Eduardo A. Groisman*,1
  1. 1 Department of Molecular Microbiology, Washington University School of Medicine, 660 S.Euclid Ave, Campus Box 8230, St Louis, MO, 63110, USA
  1. *Corresponding author. E-mail: groisman{at}borcim.wustl.edu
View Full Text

Abstract

Pathogenicity islands are chromosomal clusters of horizontally acquired virulence genes that are often found at tRNA loci. The selC tRNA locus of Escherichia coli has served as the site of integration of two distinct pathogenicity islands which are responsible for converting benign strains into uro‐ and enteropathogens. Because virulence genes are targeted to the selC locus of E.coli, we investigated the homologous region of the Salmonella typhimurium chromosome for the presence of horizontally acquired sequences. At this site, we identified a 17 kb DNA segment that is both unique to Salmonella and necessary for virulence. This segment harbors a gene, mgtC, that is required for intramacrophage survival and growth in low Mg2+ media. The mgtC locus is regulated by the PhoP/PhoQ two‐component system, a major regulator of virulence functions present in both pathogenic and non‐pathogenic bacterial species. Cumulatively, our experiments indicate that the ability to replicate in low Mg2+ environments is necessary for Salmonella virulence, and suggest that a similar mechanism is responsible for the dissemination and acquisition of pathogenicity islands in enteric bacteria.

Introduction

Pathogenicity islands are chromosomal clusters of virulence genes present in pathogenic organisms but absent from related non‐pathogenic bacteria (Groisman and Ochman, 1996; Hacker et al., 1997). Pathogenicity islands constitute a major driving force in the evolution of bacterial pathogens because their acquisition often determines the virulence properties of a microorganism. For example, enteropathogenic strains of Escherichia coli harbor a 35 kb pathogenicity island, termed LEE, that encodes the ability to form attaching and effacing lesions in intestinal epithelial cells (McDaniel et al., 1995), and introduction of LEE into a benign strain of E.coli renders it capable of producing these lesions (McDaniel and Kaper, 1997).

In contrast, Salmonella typhimurium has a complex life cycle in infected animals and requires a large number of virulence genes. While many of these virulence determinants are also present in non‐pathogenic species, several others are encoded in regions of the chromosome that are specific to Salmonella (Groisman and Ochman, 1994). For example, the vast majority of the genes conferring the ability to invade epithelial cells reside within SPI‐1, a 40 kb pathogenicity island at 63 minutes in the S.typhimurium chromosome (Mills et al., 1995; Galán, 1996). A second pathogenicity island, SPI‐2, harbors genes that are indispensable for survival within macrophages and to cause systemic disease (Ochman et al., 1996; Shea et al., 1996). In addition to SPI‐1 and SPI‐2, other Salmonella‐specific regions have been implicated in virulence: For example, the pagC and msgA genes at 25 minutes are required for intramacrophage survival (Gunn et al., 1995; Hohmann et al., 1995) and the spv genes in the large virulence plasmid are necessary for replication in extraintestinal tissues (Gulig et al., 1993). Despite their foreign origin, many of these virulence determinants are regulated by PhoP/PhoQ (Miller et al., 1989; Bajaj et al., 1996; Heithoff et al., 1997), a two‐component system that is present in both pathogens and non‐pathogens (Groisman et al., 1989). The PhoP/PhoQ regulatory system governs the adaptation to low Mg2+ environments (García Véscovi et al., 1996; Soncini et al., 1996) and, in S.typhimurium, it is essential for intramacrophage survival and virulence in mice (Fields et al., 1986, 1989; Miller et al., 1989).

Pathogenicity islands are often located next to tRNA genes (Groisman and Ochman, 1996; Hacker et al., 1997). For example, the PAI‐2, PAI‐4 and PAI‐5 pathogenicity islands are integrated at the tRNAleuX, tRNApheV and tRNApheR genes of E.coli, respectively (Blum et al., 1994; Swenson et al., 1996), and the SPI‐2 island of S.typhimurium is adjacent to the tRNAVal locus (Hensel et al., 1997). In E.coli, the selC tRNA locus has been targeted by two different pathogenicity islands, LEE and PAI‐1, which determine whether E.coli is rendered an entero‐ or a uropathogen (Blum et al., 1994; McDaniel et al., 1995). Moreover, selC has been shown also to be the attachment site of the retronphage ΦR73 in E.coli (Inouye et al., 1991).

Because two pathogenicity islands reside at the selC locus of E.coli, we investigated the selC locus of S.typhimurium for the presence of horizontally acquired sequences. In this study, we identify a new pathogenicity island that is necessary for intramacrophage survival and establish that the ability to grow in Mg2+‐limiting environments is essential for Salmonella virulence.

Results

A Salmonella‐specific DNA segment is present at the selC locus of Salmonella typhimurium

The selC tRNA gene of E.coli is the site of insertion of two pathogenicity islands (Blum et al., 1994; McDaniel et al., 1995) and the retronphage ΦR73 (Inouye et al., 1991; Sun et al., 1991). In E.coli K‐12, orf307, an open reading frame of unknown function, is located 2 kb downstream of the selC gene (Figure 1A). However, the selC and orf307 genes are >70 kb apart in uropathogenic strains of E.coli harboring the PAI‐1 pathogenicity island (Blum et al., 1994), >35 kb apart in enteropathogenic E.coli carrying the LEE island (McDaniel et al., 1995) and >12.7 kb apart in ΦR73 lysogens (Inouye et al., 1991; Sun et al., 1991). Thus, we hypothesized that S.typhimurium may contain horizontally acquired sequences at the selC locus and that these sequences might contribute to the virulence properties of this intracellular pathogen.

Figure 1.

Physical and genetic maps of the S.typhimurium and E.coli K‐12 selC–orf307 chromosomal regions. (A) The E.coli K‐12 map is based on reported nucleotide sequences (DDBJ/EMBL/GenBank accession number L10328). The S.typhimurium map is based on restriction analysis of plasmid pEG9106, which carries both selC and mgtCB genes, and on sequence data on the mgtCB locus (DDBJ/EMBL/GenBank accession number J05728). A, AvaI; E, EcoRI; H, HindIII. The distance between selC and the beginning of orf307 is 17 kb in S.typhimurium whereas it is only 2 kb in E.coli K‐12. The positions of the 4 kb EcoRI–HindIII and 780 bp fragments used as probes in Southern hybridization experiments are shown. The orientation of the various open reading frames is indicated by arrows. (B) Nucleotide sequence upstream and downstream of the selC gene. Alignment of the corresponding DNA sequences of the S.typhimurium (DDBJ/EMBL/GenBank accession number Y13864) and E.coli K‐12 chromosomes (DDBJ/EMBL/GenBank accession number L10328). The selC gene, which is 100% identical between S.typhimurium and E.coli K‐12, is indicated by bold letters. The discontinuity between Salmonella and E.coli K‐12 sequences starts 11 bp downstream of the selC gene. (C) Nucleotide sequence near the orf307 gene. Alignment of the corresponding DNA sequences of the S.typhimurium (DDBJ/EMBL/GenBank accession number J05728) and E.coli K‐12 chromosomes (DDBJ/EMBL/GenBank accession number L10328). The 5′ end of the orf307 gene is indicated in bold letters. The discontinuity between Salmonella and E.coli K‐12 sequences appears 30–50 bp upstream from the beginning of orf307. Alignments were performed with the GCG program GAP. The position of the primers (selC‐F, selC‐R, F05‐F and E07‐R) used in our studies is indicated by horizontal lines.

To determine the distance between the selC and orf307 genes in S.typhimurium, we carried out PCR reactions using primers corresponding to selC and orf307 (Figure 1B and C). The selC primer was based on a 32 bp segment of the E.coli K‐12 selC tRNA gene because the sequence of the Salmonella selC gene was not available and tRNA genes are highly conserved. We searched the Salmonella genome for orf307 homologs and identified a 94 bp segment upstream of the mgtCB operon that was 91% identical to the 5′ end of orf307 in E.coli K‐12. The orf307 primer corresponded to a 21 bp portion of this segment that is 100% conserved between the Salmonella and E.coli K‐12 genomes. [Both mgtCB and selC map to the 82 minute region of the S.typhimurium chromosome (Sanderson et al., 1995).] When E.coli K‐12 DNA was used as template, the expected 2 kb PCR product was obtained; however, no fragment could be amplified with Salmonella DNA as template (data not shown). These results suggested that the distance between selC and orf307 may be too large in Salmonella to be amplified under the PCR conditions used.

We examined the possibility that the DNA segment located between selC and orf307 harbored sequences specific to Salmonella by investigating 10 bacterial species for the presence of the mgtCB operon. Using gene‐specific probes, mgtC‐ and mgtB‐hybridizing sequences were not detected in E.coli K‐12 and seven other bacterial species (Figure 2). However, the hybridization patterns obtained with these probes were not identical: mgtC‐hybridizing sequences were also found in Klebsiella pneumoniae and mgtB‐hybridizing sequences were present in Enterobacter aerogenes; yet, Yersinia enterocolitica DNA gave positive signals with both probes (Figure 2). In contrast, the mgtA gene, which maps to a different chromosomal location than mgtCB and is known to be present in E.coli K‐12 (Tao et al., 1995), was detected in all enteric species tested (Figure 2). (The mgtA and mgtB genes of S.typhimurium encode similar Mg2+ transporters and are 57% identical at the DNA level.) The narrow and sporadic distribution of the mgtC and mgtB genes suggests that the DNA segment harboring the mgtCB operon was acquired by horizontal gene transfer. This DNA segment was acquired prior to the diversification of all extant serovars of Salmonella because mgtCB‐hybridizing sequences were detected in all eight subspecies encompassing the genus Salmonella (Boyd et al., 1996; data not shown).

Figure 2.

Distribution of mgtA, mgtB and mgtC sequences among bacterial species. Southern hybridization experiments were carried out using PCR‐generated, gene‐specific probes for the S.typhimurium mgtA, mgtB and mgtC genes, and total DNA from the bacterial species indicated in the figure. + indicates the presence of a positive hybridization signal in the designated species. Relationships among species follow those presented in Ahmad et al. (1990) and Ochman and Groisman (1994).

Analysis of the junctions of the Salmonella‐specific segment harboring mgtCB

To determine the genetic structure of the selCorf307 region in S.typhimurium, we isolated plasmids containing both selC and mgtCB genes. First, we screened a library by colony hybridization using an mgtB‐specific probe and recovered six positive clones. Two of these clones also gave positive signals with a selC‐specific probe, indicating that the distance between the mgtB and selC genes of Salmonella must be <31 kb, because this is the maximun insert size obtained by the mini‐Mu cloning system used to generate the plasmid library (Groisman and Casadaban, 1986).

We constructed restriction maps of mgtCB+ selC+ and mgtCB+ selC plasmids and compared them with the published sequence of the mgtCB locus. We established that selC and orf307 are separated by 17 kb in S.typhimurium whereas these genes are only 2 kb apart in the E.coli K‐12 chromosome (Figure 1A). To determine whether the Salmonella‐specific sequences extended to the selC end of the 17 kb region, we investigated the DNA from 10 different bacterial species for the presence of sequences hybridizing to a 4 kb EcoRI–HindIII restriction fragment located in the middle of the 17 kb region and to a 780 bp fragment located 260 bp downstream of the selC gene (Figure 1A). Apart from hybridizing to the 17 kb region of S.typhimurium, the 4 kb probe detected a second weak band in S.typhimurium that was similar in intensity to a single band detected in Citrobacter freundii (data not shown). On the other hand, sequences hybridizing to the selC‐proximal probe were not detected in bacteria other than S.typhimurium (data not shown). Cumulatively, these results support the notion that the 17 kb region is specific to Salmonella.

To define the borders of the 17 kb region, we determined the nucleotide sequence of the Salmonella chromosome upstream and downstream of the selC gene and upstream of the mgtC gene. The Salmonella and E.coli K‐12 chromosomes exhibit 70% DNA sequence identity in the 420 bp region immediately preceding selC and 78% identity in a 870 bp segment upstream of mgtC which includes orf307. The discontinuity between the Salmonella and E.coli K‐12 chromosomes starts 11 bp downstream of the selC gene (Figure 1B) and extends for at least 1 kb. In the mgtCB end, the discontinuity starts 30–50 bp upstream of the putative initiation codon for orf307 (Figure 1C) and extends for at least 4.5 kb. The G+C contents of the 1 kb dowstream of the selC gene and the 4.5 kb region upstream of orf307 are 39.8 and 49.3%, respectively, which are lower than the overall 52–54% G+C content estimated for the S.typhimurium chromosome. The incorporation of the 17 kb region into the S.typhimurium chromosome might have occurred by a mechanism similar to that resulting in the integration of the LEE pathogenicity island in E.coli, because microorganisms that harbor these islands at the selC locus are deleted for orf394 (McDaniel et al., 1995; data not shown), an open reading frame present between selC and orf307 in E.coli K‐12.

The mgtCB operon is necessary for virulence

Salmonella typhimurium causes a lethal infection in mice that resembles typhoid fever of humans. We hypothesized that the 17 kb region harboring the mgtCB operon may constitute a pathogenicity island and be required for virulence because: (i) mgtC and mgtB genes display a sporadic distribution in enteric bacteria (Figure 2); (ii) this region is tightly linked to the selC gene, the site of insertion of two pathogenicity islands in E.coli (Blum et al., 1994; McDaniel et al., 1995); and (iii) transcription of the mgtCB operon is governed by PhoP/PhoQ (García Véscovi et al., 1996; Soncini et al., 1996), the major regulator of virulence functions in Salmonella (García Véscovi et al., 1994; Groisman and Heffron, 1995).

To examine the virulence role of the 17 kb region, a strain harboring a mutation in mgtCB was used to inoculate BALB/c mice by the intraperitoneal (i.p.) route: 40–50% of the animals survived at doses that were 15–150 times the median lethal dose (LD50) of the wild‐type strain (i.e. 10 organisms; Table I). While an LD50 could not be clearly determined, the mgtCB mutant was not as attenuated as strains harboring mutations in the phoP locus (Miller et al., 1989), which is known to control expression of several other virulence determinants in Salmonella. A mutant harboring a transposon insertion in the mgtB gene was as virulent as the wild‐type parent, suggesting that the virulence defect of the mgtCB mutant is due to the absence of mgtC. Despite the mgtA gene being regulated also by the PhoP/PhoQ system, an mgtA mutant remained virulent in mice. On the other hand, a strain harboring mutations in both mgtA and mgtCB genes did not cause a lethal infection when inoculated at doses up to 1500 times the LD50 of the wild‐type strain (Table I). These results demonstrate that the mgtCB operon is required for virulence in Salmonella, and define a new pathogenicity island that we have designated SPI‐3.

View this table:
Table 1. Mouse virulence properties of mgt mutants of S.typhimurium

The mgtC gene is required for intramacrophage survival

We reasoned that the mgtCB mutant might be defective for intramacrophage survival because the mgtCB operon is transcriptionally activated by the PhoP protein (García Véscovi et al., 1996; Soncini et al., 1996) and phoP mutants are unable to survive within macrophages (Fields et al., 1986, 1989). We investigated strains harboring MudJ transposon insertions in the mgtA and mgtCB genes (Figure 3A) for their ability to replicate within the macrophage‐like cell line J774. The mgtCB mutant behaved like the phoP strain: it failed to grow in J774 cells (Figure 4A). In contrast, the mgtA mutant is virulent in mice (Table I) and survived in macrophages to the same extent as the wild‐type strain (Figure 4A). These results demonstrate that the mgtCB locus is necessary for replication within macrophages.

Figure 3.

Structure of chromosomal DNA of mgt mutants and of plasmids harboring mgtCB genes. (A) Position of MudJ insertions in the EG9521, EG9527, EG9821 and EG9823 mutants. Strain EG9527, which carries a MudJ insertion in mgtC is phenotypically mgtC mgtB. (B) Structure of mgtC+‐ and/or mgtB+‐containing plasmids used in this study. The orientation of the various open reading frames is indicated by arrows.

Figure 4.

The mgtC gene of S.typhimurium is required for intramacrophage survival. (A) Intramacrophage survival properties of 14028s (wild‐type), MS7953s (phoP7953::Tn10), EG9521 (mgtA9226::MudJ), EG9527 (mgtCB9232::MudJ) and EG9821 (mgtB11::MudJ) in J774 cells. Macrophage survival was determined as described in Materials and methods 18 h after infection. For each mutant, the percentage of survival relative to the wild‐type strain is presented. (B) Intramacrophage survival of 14028s (wild‐type) and EG9527 (mgtCB9232::MudJ) harboring plasmids with mgtC+ and/or mgtB+ genes. The percentage of macrophage survival relative to the wild‐type strain harboring pBR322 is presented. Plasmids used for complementation tests, mgtCB+ (pEG9091), mgtB+ (pEG9092) and mgtC+ (pEG9094), are derivatives of pBR322 (Figure 3B). Values represent the mean of three independent experiments ± SD.

The mgtC and mgtB genes form a bicistronic operon (Snavely et al., 1991a; Tao et al., 1995); therefore, a MudJ transposon insertion in the mgtC gene is predicted to affect expression of the downstream mgtB gene (Figure 3A). To define whether mgtC, mgtB or both genes are required for intramacrophage survival, we examined the ability of plasmids containing the mgtC and/or mgtB genes to complement the mgtCB mutant (Figure 3B). A plasmid carrying the mgtC gene restored the ability to survive in macrophages in the mgtCB mutant but one harboring the mgtB gene did not (Figure 4B). As expected, a plasmid with the entire mgtCB operon rescued the macrophage survival defect of the mgtCB strain (Figure 4C).

That the macrophage survival defect of the mgtCB::MudJ mutant is due to the lack of MgtC protein is substantiated further by the phenotype of mgtB mutants, which replicated within macrophages but to a lesser extent than the wild‐type strain (Figure 4A). The mild defect displayed by the mgtB mutant is probably due to an indirect effect on mgtC expression because the mgtC‐containing plasmid could restore wild‐type macrophage survival to the mgtB mutant (data not shown). Taken together, these results establish that the mgtC gene is essential for intramacrophage survival.

Identification of a defined medium that reproduces the macrophage survival phenotype of the mgt mutants

We have established previously that both mgtA and mgtCB are required for optimal growth in liquid N‐minimal medium supplemented with 10 μM Mg2+ (Soncini et al., 1996); yet, mgtA and mgtCB mutants differ in their ability to survive within macrophages (Figure 4A). We identified a defined medium that reproduces the macrophage survival phenotype of these mutants: when grown in NCE medium supplemented with 10 μM Mg2+, the mgtCB and phoP mutants reached an optical density (OD) that was lower than the ODs achieved by the wild‐type and mgtA strains (Figure 5A). The mgtB mutant, which is slightly defective for macrophage survival, reached an OD that was in between the ODs of the wild‐type and mgtCB strains (Figure 5A). The growth defect of the phoP, mgtCB and mgtB mutants is due to a low Mg2+ concentration in the medium because these mutants grew like the wild‐type strain in the presence of 10 mM Mg2+ (Figure 5B). Equivalent levels of Ca2+ and Mn2+ could not rescue the mutants, even though these divalent cations can modulate expression of PhoP‐activated genes (García Véscovi et al., 1996).

Figure 5.

mgtCB mutants of Salmonella are defective for growth in low Mg2+ NCE liquid medium. Optical density (OD600) achieved by 14028s (wild‐type), MS7953s (phoP7953::Tn10), EG9521 (mgtA9226::MudJ), EG9527 (mgtCB9232::MudJ) and EG9821(mgtB11::MudJ) at different times following inoculation in NCE medium containing 10 μM or 10 mM MgCl2. The data correspond to a representative experiment done in duplicate. Error bars are indicated only if larger than the symbols.

While there are several differences between NCE and N‐minimal media, a lower pH in NCE medium (i.e. pH 7.0 versus pH 7.5 in N‐minimal medium) may account for the ability of the mgtA mutant to grow like the wild‐type strain in NCE medium. When the pH of N‐minimal medium was lowered to 7.0, the mgtA mutant grew like wild‐type Salmonella (data not shown). In contrast, the mgtCB mutant was defective in N‐minimal medium at both pH 7.0 and 7.5. These results demonstrate that the mgtCB locus is essential for optimal growth in Mg2+‐limiting environments.

The Salmonella mgtCB locus confers growth in low Mg2+ upon wild‐type E.coli

We reason that wild‐type E.coli would be impaired for growth in low Mg2+ because this bacterial species is missing the mgtCB operon (Figure 2) and mgtCB is required for growth in low Mg2+ (Figure 5A). Wild‐type E.coli K‐12 reached an OD that was lower than that of wild‐type S.typhimurium and similar to the OD of the Salmonella mgtCB mutant when grown in NCE liquid medium supplemented with 10 μM Mg2+ (three different E.coli K‐12 strains exhibited the same behavior; data not shown). The limited growth of E.coli was due to the low Mg2+ concentration in NCE medium because E.coli and Salmonella reached the same OD in 10 mM Mg2+.

The absence of mgtCB appears to be responsible for the limited growth of E.coli in low Mg2+ medium because a plasmid harboring the Salmonella mgtCB+ operon allowed E.coli to reach the same OD as that of wild‐type Salmonella or the mgtCB mutant transformed with the mgtCB+ plasmid (Figure 6). An mgtC+‐containing plasmid conferred partial growth in low Mg2+, but a plasmid carrying the mgtB+ gene had no effect. These results further support the notion that the mgtCB locus is essential for optimal growth in low Mg2+. Apart from mgtC, several Salmonella‐specific genes are required for intramacrophage survival (Gunn et al., 1995; Ochman et al., 1996), and, as predicted, the mgtCB+ plasmid could not confer macrophage survival properties upon E.coli (data not shown).

Figure 6.

The mgtCB genes allow E.coli K‐12 to grow in low Mg2+ NCE medium. OD600 of S.typhimurium 14028s (wild‐type) and EG9527 (mgtCB9232::MudJ), and E.coli MG1655 (wild‐type) harboring pEG9091 (mgtCB+), pEG9092 (mgtB+), pEG9094 (mgtC+) or pBR322 following 20 h in NCE liquid medium containing 10 μM MgCl2. Values represent the mean of at least three independent experiments ± SD.

Mg2+ rescues growth of mgtCB mutants in macrophages

If the macrophage survival defect of mgtCB mutants is due to their inability to grow in low Mg2+ environments, one would anticipate that addition of Mg2+ to host cells might restore intramacrophage survival. Indeed, growth of both mgtCB and phoP mutants was improved when Mg2+ was added to the tissue culture medium after bacteria had been internalized by the macrophage (Figure 7). Mg2+ exerted its action after phagocytosis because it could rescue the mutants even when added 4 h after infection of J774 cells. Moreover, the effect was specific to the phoP and mgtCB mutants because growth of a purB strain (a purine auxotroph that cannot replicate in macrophages) was not affected by the addition of Mg2+. These results indicate that the Salmonella‐containing vacuole is an Mg2+‐limiting environment, which had been suggested previously by the transcriptional induction of PhoP‐activated genes both in host cells (Garcia‐del Portillo et al., 1992; Heithoff et al., 1997) and in low Mg2+ media in vitro (García Véscovi et al., 1996; Soncini et al., 1996).

Figure 7.

Effect of Mg2+ addition on intramacrophage survival of phoP and mgtCB mutants. Survival properties of 14028s (wild‐type), MS7953s (phoP7953::Tn10), EG9527 (mgtCB9232::MudJ) and EG9652 (purB877::Tn10) in J774 cells in the absence or presence of Mg2+ (25 mM) added to the DMEM, after internalization of the bacteria by the macrophages. The purB mutant is a purine auxotroph that is defective for intramacrophage survival and could not be rescued by the addition of Mg2+. Values represent the mean of at least three independent experiments ± SD.

Discussion

Pathogenicity islands are chromosomal clusters of horizontally acquired virulence genes that are often found at tRNA loci (Groisman and Ochman, 1996; Hacker et al., 1997). These clusters typically encode complete functional units and their incorporation into a benign strain can render it pathogenic. We have now identified a pathogenicity island in S.typhimurium that mediates intramacrophage survival, virulence in mice and growth in low Mg2+.

The evolution of Salmonella virulence

Two pathogenicity islands have been identified in Salmonella: SPI‐1, at 63 minutes in the S.typhimurium chromosome, governs the ability to invade epithelial cells (Mills et al., 1995; Galán, 1996); and SPI‐2, at 31 minutes, is necessary for intramacrophage survival (Ochman et al., 1996; Shea et al., 1996). We have now established that the 17 kb region harboring the mgtCB operon at 82 minutes constitutes a third pathogenicity island, designated SPI‐3, because: (i) the mgtC gene is necessary for full virulence in mice and intramacrophage survival (Table I; Figure 4); (ii) the mgtCB sequences exhibit a restricted distribution in eubacteria (Figure 2); and (iii) SPI‐3 is located immediately downstream of selC (Figure 1), a tRNA gene that is the site of insertion of two pathogenicity islands in E.coli (Blum et al., 1994; McDaniel et al., 1995).

The identification of SPI‐3 provides the first example for the incorporation of horizontally acquired sequences at the selC gene in two different bacterial species. Whereas SPI‐3 inserted 11 bp downstream of selC in S.typhimurium (Figure 1B), the PAI‐1 and LEE pathogenicity islands are present 16 bp 3′ to the selC gene in E.coli (Blum et al., 1994; McDaniel et al., 1995) (Table II). In contrast to PAI‐1, which is flanked by short direct repeats (Blum et al., 1994), no repeats are found at the insertion sites of SPI‐3 (Figure 1) and LEE (McDaniel et al., 1995). The utilization of selC as integration target in two different species suggests that a similar mechanism is responsible for the incorporation of horizontally acquired sequences at this tRNA locus. This mechanism may be phage mediated because selC is the attachment site for the retronphage ΦR73 of E.coli (Inouye et al., 1991; Sun et al., 1991) and an ΦR73‐related integrase is encoded within the pathogenicity island PAI‐1 of E.coli (Hacker et al., 1997).

View this table:
Table 2. Horizontally acquired sequences at the selC locus

The nutritional environment of the Salmonella‐containing phagosome

To replicate within phagocytic cells, Salmonella must adapt to the microbicidal and nutrient‐poor environment of the phagosome. This requires the manufacture of nutrients not available from host tissues and the co‐ordinate regulation of genes that protect the bacterium from the oxygen‐dependent and ‐independent killing mechanisms of the phagocytic cell. The Salmonella‐containing phagosome is limiting for purines, pyrimidines, histidine and methionine because auxotrophs for these compounds are not able to replicate in macrophages (Fields et al., 1986). This environment appears also to be low in glutamine because strains defective in both the biosynthesis and transport of glutamine are unable to grow within macrophages (Klose and Mekalanos, 1997). Our experiments now indicate that the Mg2+ concentration in the phagosome is also limiting since phoP and mgtCB mutants cannot survive in macrophages (Figure 4A), are impaired for growth in low Mg2+ liquid media (Figure 5A), and in both cases growth can be ameliorated by the addition of Mg2+ (Figures 5B and 7). That the Salmonella‐containing phagosome might be low in Mg2+ was suggested previously by the transcriptional induction of PhoP‐activated genes in host cells (Garcia‐del Portillo et al., 1992; Heithoff et al., 1997) and in low Mg2+ media in vitro (García Véscovi et al., 1996; Soncini et al., 1996), but a direct biochemical determination of the Mg2+ concentration in Salmonella‐containing phagosomes has yet to be reported.

Consistent with the macrophage survival defect, the mgtCB mutant was attenuated for virulence in mice (Table I). This suggests that acquisition of the mgtCB operon was important in the development of Salmonella as an intracellular pathogen because it allowed replication in Mg2+‐limiting environments. On the other hand, E.coli is an extracellular commensal of mammals that lacks mgtCB and exhibits limited growth in low Mg2+. The Salmonella mgtCB operon enabled wild‐type E.coli K‐12 to grow in low Mg2+ liquid media (Figure 6), but, as predicted, it did not confer intramacrophage survival upon this species since additional Salmonella‐specific genes are required for this ability (Gunn et al., 1995; Ochman et al., 1996). Because the addition of Mg2+ to the macrophage did not fully restore intramacrophage survival to the mgtCB mutant, we cannot rule out the possibility of a factor(s) unrelated to the ability to grow in low Mg2+ contributing to the macrophage replication defect of this mutant.

Function of the MgtC protein

The macrophage survival defect of the mgtCB mutant is due to the lack of mgtC (rather than mgtB) because this mutant was fully rescued by a plasmid carrying mgtC+ (Figure 4B). The MgtC protein is predicted to localize to the inner membrane where it could interact with the Mg2+ transport protein MgtB (Snavely et al., 1991a). However, the MgtC protein is not necessary for membrane insertion or transport function of the MgtB protein (Tao et al., 1995), and our results now indicate that MgtC mediates macrophage survival in the absence of the MgtB (Figure 4B) and MgtA proteins (data not shown). Thus, the MgtC protein could be a Mg2+ transporter that functions independently of the Mg2+ uptake proteins encoded by the mgtA and mgtB genes.

PhoP/PhoQ: a global regulator of virulence functions

The acquisition of pathogenicity islands offers a rapid way of evolving new functions; however, the incorporated sequences, even when they encode their specific regulators, must coordinate their expression with that of the recipient genome. In addition to the mgtCB operon, several virulence genes of foreign origin are regulated by the PhoP/PhoQ regulatory system (Figure 8), which is present in pathogenic and non‐pathogenic species (Groisman et al., 1989). These genes include: invasion determinants encoded within the SPI‐1 island (Bajaj et al., 1996); the pagC gene, required for intramacrophage survival (Hohmann et al., 1995); iviVI‐A, a gene that is induced specifically in the spleen of Salmonella‐infected mice (Heithoff et al., 1997); and the plasmid‐encoded spvB gene (Heithoff et al., 1997), necessary to cause systemic disease.

Figure 8.

The PhoP/PhoQ system regulates expression of horizontally acquired virulence genes. The prgH locus resides within the SPI‐1 invasion island and the mgtCB operon is part of the SPI‐3 island described in this study. + and − indicate PhoP‐activated and ‐repressed loci, respectively.

Horizontally acquired virulence genes may be regulated by the PhoP/PhoQ regulatory system to ensure their proper temporal and spatial expression. The concentration of extracellular Mg2+ is the specific regulatory signal that controls the PhoP/PhoQ system (García Véscovi et al., 1996). During the course of infection, Salmonella is exposed to millimolar levels of Mg2+ in extracellular fluids (Reinhart, 1988) and micromolar concentrations of this divalent cation in host cell vacuoles (Garcia‐del Portillo et al., 1992). Thus, the PhoP/PhoQ system contributes to Salmonella pathogenicity by both conferring the ability to grow in Mg2+‐limiting environments and co‐ordinating the expression of different virulence determinants.

Materials and methods

Bacterial strains and growth conditions

Bacterial strains used is this study are listed in Table III. All S.typhimurium strains are derived from 14028s, except for MM197, MM198 and TT282 which are derived from LT2. Strains EG9652, EG9821 and EG9823 were constructed by phage P22‐mediated transduction (Davis et al., 1980) using lysates grown on TT282, MM197 and MM198, respectively. The position of the MudJ transposon in strains EG9521, EG9527, EG9821 and EG9823 is presented in Figure 3A. Bacteria were grown at 37°C in Luria Broth (LB; Miller, 1972), in NCE‐minimal medium (Maloy, 1990) supplemented with 0.1% casamino acids, 38 mM glycerol and either 10 μM or 10 mM MgCl2, or in N‐minimal medium (Snavely et al., 1991b) supplemented with 0.1% casamino acids, 38 mM glycerol and 10 μM MgCl2. Ampicillin and kanamycin were each used at 50 μg/ml.

View this table:
Table 3. Bacterial strains and plasmids

The ability of the different bacterial strains to grow in low Mg2+ liquid media was evaluated as follows: overnight cultures grown in NCE‐minimal medium supplemented with 10 mM MgCl2 were washed three times with Mg2+‐free medium and diluted 1:200 (Figure 5) or 1:50 (Figure 6) in culture media containing 10 μM or 10 mM MgCl2. The optical density at 600 nm (OD600) was measured at different times (Figure 5) or after 20 h of incubation (Figure 6).

Plasmid construction

Plasmids used in this study are listed in Table III. Plasmid DNA was introduced into bacterial strains by electroporation using a Bio‐Rad apparatus as recommended by the manufacturer. Recombinant DNA techniques were performed according to standard protocols (Sambrook et al., 1989). Plasmid pTT39 is a pBluescript derivative with a BamHI–ClaI insert containing the entire mgtCB operon (Tao et al., 1995). pTT39d contains only the mgtB gene because of a SmaI deletion that removes the 3′ end of the mgtC gene (Tao et al., 1995; Figure 3B). Plasmids pEG9091 and pEG9092 were constructed by cloning the BamHI–ClaI inserts of pTT39 and pTT39d, respectively, between the BamHI and ClaI sites of plasmid pBR322 (Figure 3B). pEG9094 is a derivative of pEG9091 that contains only the entire mgtC gene because the HindIII–ClaI segment of the mgtB gene has been deleted (Figure 3B). As predicted, pEG9091 and pEG9092 allowed growth of EG9798 (corA mgtA mgtCB triple mutant) in LB medium whereas pEG9094‐containing EG9798 cells required the addition of 100 mM Mg2+ to the LB for growth (Hmiel et al., 1989).

Cloning of the selC–mgtCB chromosomal region

A library from wild‐type strain 14028s was constructed by the in vivo cloning technique using the mini‐Mu replicon element Mud5005 (Groisman and Casadaban, 1986). This method allows the isolation of plasmids with inserts of up to 31 kb. Kanamycin‐resistant transductants were screened by colony hybridization for the presence of the mgtB gene using a labeled PCR fragment corresponding to the coding region of mgtB as probe. Six positive clones were obtained and tested for the presence of the selC gene by hybridizing with an oligonucleotide probe corresponding to 32 bp of the E.coli K‐12 selC gene, and two positive clones were recovered. One of these clones harbored a plasmid, pEG9106, that contains an insert of ∼20 kb and was used for subsequent molecular analysis.

Molecular biological techniques

PCR reactions were carried out on purified chromosomal DNA with Taq polymerase according to the manufacturer's protocol (Gibco BRL). The primer selC‐F, 5′‐ATCCAGTTGGGGCCGCCAGCGGTCCCGGGCAG‐3′, is complementary to the E.coli K‐12 selC gene. The primer E07‐R, 5′‐TTTCTGGTGGAACCCATTTTT‐3′, corresponds to position 78–98 of the mgtCB published sequence [accession number J05728 (Snavely et al., 1991a)]. The sequences of these two primers are 100% identical in the E.coli K‐12 and S.typhimurium chromosomes (Figure 1B and C).

Colony hybridization experiments were performed as described (Buluwela et al., 1989) using either PCR‐generated DNA fragments labeled with [α‐32P]dCTP and the Ready To Go kit (Pharmacia Biotech) or oligonucleotides labeled with [γ‐32P]ATP by the T4 polynucleotide kinase (New England Biolabs) as described (Sambrook et al., 1989). Southern hybridization analysis was carried out using chromosomal DNA digested by EcoRI, size fractionated in 1% agarose gels and transferred to nylon membrane by capillarity (Sambrook et al., 1989). Hybridization was performed at 65°C as described (Groisman et al., 1993) with probes corresponding to PCR‐generated fragments corresponding to the coding regions of the mgtA, mgtB and mgtC genes, to a 780 bp fragment (Figure 1A) amplified with primers 5′‐ACTTACAGGCTCATCCTTTCTC‐3′ and 5′‐AACGTAAGGCTATAGTGCCT‐3′ or to a 4 kb EcoRI–HindIII fragment from the 17 kb region (Figure 1A). Prior to autoradiography, washes (two for 15 min at room temperature and one for 5–10 min at 65°C) were performed in 1× SSC, 0.1% SDS.

DNA sequencing was carried out on purified pEG9106 DNA (Qiagen kit) using the dye terminator cycle sequencing kit with AmpliTaq DNA polymerase (Perkin Elmer) and an ABI 373 sequencer. The sequence around the selC gene was determined using primers complementary to the E.coli K‐12 selC sequence: primer selC‐R (5′‐TGCCCGGGACCGCTGGCGGCCCCAACTGGATTTG‐3′) was used to sequence the region upstream of selC and primer selC‐F (see above) was used to sequence the region downstream of selC. The sequence upstream of the mgtC gene was determined using primer F05‐F (5′‐AGCTCTGTTATGATGCCTGC‐3′), which corresponds to position 129–148 of the published mgtCB sequence (Snavely et al., 1991a). The DNA sequence generated in this work was determined completely on both strands and has been submitted to the EMBL database under accession numbers Y13864 and AJ000509. DNA sequence alignments were conducted with the GCG software packages (University of Wisconsin Biotechnology Center, Madison, WI).

Macrophage survival assay

Macrophage survival assays were conducted with the macrophage‐like cell line J774 essentially as described (Buchmeier and Heffron, 1989). Briefly, 5×105 macrophages in Dulbecco‘s modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 60 μM 6‐thioguanine were allowed to adhere for 24 h in duplicate 24‐well plates. Bacteria were added at a ratio of 10 bacteria per macrophage, and the plates were centrifuged at 1000 r.p.m. for 10 min at room temperature. The cells were incubated for 20 min at 37°C to permit phagocytosis, and free bacteria were removed by three washes with phosphate‐buffered saline (PBS). For the time 0 sample, wells were treated immediately by aspirating the medium and adding 200 μl of 1% Triton X‐100 and 800 μl of PBS. The content of each well was then transferred to a new well, appropriate dilutions were made in PBS and the number of bacteria was quantitated by plating for colony‐forming units (c.f.u.) on LB agar. For the 18 h samples, DMEM supplemented with 10% FBS, 60 μM 6‐thioguanine and 12 μg/ml gentamycin was added, and the cells were incubated at 37°C. When indicated, 25 mM MgCl2 was added to the DMEM supplemented with gentamycin. After 18 h of infection, wells were washed twice with PBS and were treated with Triton X‐100 as indicated above. The percentage survival was obtained by dividing the number of bacteria recovered after 18 h by the number of bacteria present at time 0 and multiplying by 100. All experiments were done in duplicate on at least three independent occasions.

Mouse virulence assays

Virulence assays were performed with 7‐ to 8‐week‐old female BALB/c mice inoculated i.p. with 100 μl of bacteria diluted in PBS (4–10 mice were used at each dose per mutant strain). The number of organisms injected was quantitated by plating for c.f.u. on LB agar. Viability was recorded for at least 5 weeks. Bacterial strains used for mouse virulence assays were freshly made transductants to ensure that all strains had an identical genetic background and did not accumulate mutations that would interfere with their ability to cause a lethal infection.

Acknowledgements

We thank Fernando Soncini, who originally conducted the macrophage survival assays with the mgtCB mutant; Felix Solomon for technical assistance; Oname Burlingame for generation of the plasmid library; Michael Maguire for plasmids and strains; and Howard Ochman, Henry Huang, David Sibley and three anonymous referees for comments on an earlier version of this manuscript. This work was supported by NIH grant GM54900 to E.A.G. who is the recipient of a Research Career Development Award from the NIH. A.‐B.B.‐P. received financial support from the Philippe Foundation.

References

View Abstract