Genome-wide identification of novel genomic islands that contribute to Salmonella virulence in mouse systemic infection


  • Editor: Rob Delahay

Correspondence: Nobuhiko Okada, Department of Microbiology, School of Pharmacy, Kitasato University, 5-9-1 Shirokane, Minatoku, Tokyo 108-8641, Japan. Tel.: +81 3 5791 6256; fax: +81 3 3444 4831; e-mail:


Salmonella pathogenicity islands are inserted into the genome by horizontal gene transfer and are required for expression of full virulence. Here, we performed tRNA scanning of the genome of Salmonella enterica serovar Typhimurium and compared it with that of nonpathogenic Escherichia coli in order to identify genomic islands that contribute to Salmonella virulence. Using deletion analysis, we identified four genomic islands that are required for virulence in the mouse infection model. One of the newly identified pathogenicity islands was the pheV-tRNA-located genomic island, which is comprised of 26 126 bp, and encodes 22 putative genes, including STM3117–STM3138. We also showed that the pheV tRNA-located genomic island is widely distributed among different nontyphoid Salmonella serovars. Furthermore, genes including STM3118–STM3121 were identified as novel virulence-associated genes within the pheV-tRNA-located genomic island. These results indicate that a Salmonella-specific pheV-tRNA genomic island is involved in Salmonella pathogenesis among the nontyphoid Salmonella serovars.


A large number of bacteria acquire a number of genes required for replication and survival in various environments during their evolution. In general, bacterial genomes consist of a core sequence with additional strains and species-specific sequences inserted within the core genome. Genomic islands are large segments of DNA acquired by horizontal gene transfer, have a G+C content different from that of the host core genome, and often contain mobile genetic elements. Pathogenicity islands are one type of the genomic islands that encode genes involved in virulence, which distinguish pathogenic bacteria from related nonpathogenic bacteria (reviewed in Hacker et al., 1997). Because genomic islands are frequently located adjacent to tRNA genes, it is thought that the conserved structures in tRNA facilitate the integration and excision of genomic islands (reviewed in Hacker & Kaper, 2000).

Salmonella enterica is an important pathogen diverged from the Escherichia coli lineage, which evolved with the acquisition of a large number of virulence genes that are required during various stages of pathogenesis (reviewed in Ochman et al., 2000), strongly suggesting that Salmonella-specific genes contribute to virulence. In Salmonella, at least five Salmonella pathogenicity islands (SPIs) have been characterized among S. enterica serovars. SPI-1 and SPI-2 encode type III secretion systems and are required for different stages of infection. SPI-1 is essential for invasion of host cells and the induction of macrophage apoptosis (Galan, 2001). In contrast, SPI-2 confers survival and replication within macrophages and is required for systemic infection in the mouse model (Hensel et al., 1997). The primary virulence genes encoded by SPI-3, mgtCB, are high-affinity magnesium transport systems, and are necessary for macrophage survival and full virulence in mice (Blanc-Potard et al., 1999). SPI-4 is required for intestinal infection in mice (Morgan et al., 2004). Furthermore, effector proteins, sopB and pipAB, translocated by SPI-1 and SPI-2 type III secretion systems, respectively, are encoded by SPI-5 (Wood et al., 1998; Tsolis et al., 1999). Interestingly, SPI-2–SPI-5 locate adjacent to tRNA on the chromosome (Figueroa-Bossi & Bossi, 1999).

Based on these observations, we devised an approach to identify pathogen-specific genomic islands at tRNA in S. enterica serovar Typhimurium (S. Typhimurium). Hansen-Wester & Hensel (2002) have characterized Salmonella-specific DNA segments that show characteristics of pathogenicity islands. However, none of the tRNA-associated islands required for virulence in mice have been identified so far. Thus, in order to identify the novel genomic islands required for Salmonella virulence, we selected nine tRNA-associated regions specific for S. enterica, except for SPI-2–SPI-5, or Gifsy-1 prophage, based upon the genome sequence from the S. Typhimurium strain LT2 (McClelland et al., 2001). In this study, we found that four out of nine genomic islands are required for virulence in the mouse infection model. Using deletion analysis, we further determined the genes within the pheV-tRNA genomic island, which are necessary for the full virulence of S. Typhimurium.

Materials and methods

Bacterial strains, plasmids, and growth conditions

The bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were grown in Luria–Bertani (LB) medium or on LB agar under conditions for selection for resistance to ampicillin (100 μg mL−1), chloramphenicol (25 μg mL−1), nalidixic acid (50 μg mL−1), kanamycin (25 μg mL−1), or spectinomycin (50 μg mL−1) as appropriate. Phage P22-mediated transduction for Salmonella has been described previously (Sternberg & Maurer, 1991).

Table 1.   Bacterial strains and plasmids used in this study
NameRelevant characteristicsReferences
  1. ATCC, American Type Culture Collection.

Bacterial strains
14028S. Typhimurium, wild-typeATCC
SH100Spontaneous derivative of 14028, NalrGotoh et al. (2003)
DH5αE. coli: recA1 endA1 gyrA96 thi-1 hsdR17 supE44 (lacXYA-argR)U169 deoR (80 dlac(lacZ)M15)Invitrogen
S17.1λpirE. coli: recA thi pro hsdR RP4-2-Tc:Mu:kan Tn7pir, Smr, TprMiller & Mekalanos (1988)
pGEM-T EasyTA cloning vector, AmprPromega
pACYC184p15A-based low copy number plasmid, Cmr, TcrNew England BioLabs
pMW118pSC101-based low copy number plasmid, AprNippon Gene
pWM91pir-dependent sucrose selection vector, AprMetcalf et al. (1996)
pMW-STM3118STM3118 in pMW118, AprThis study
pMW-STM3119STM3119 in pMW118, AprThis study
pAC-STM3120STM3120 in pACYC184, CmrThis study
pAC-STM3121STM3121 in pACYC184, CmrThis study

For construction of the complementing plasmids, pMW-STM3118, pMW-STM3119, pAC-STM3120, and pAC-STM3121, the corresponding genes were amplified from the genomic DNA of strain ATCC 14028, and were cloned into pMW118 or pACYC184. All constructs were verified by DNA sequencing.

Molecular techniques

Molecular methods (DNA preparation, PCR, ligation, transformation, DNA cloning, and DNA–DNA hybridization) were applied as described previously (Haneda et al., 2001) according to the standard methods. TaKaRa Ex taq DNA Polymerase (Takara) was used for PCR experiments as recommended by the manufacturer. All primers used are listed in the Supporting Information, Table S1.

Mutant construction

Deletion mutations in Salmonella genomic islands were constructed by an allele exchange using suicide vector pWM91 (Metcalf et al., 1996). To delete the complete genomic island, c. 1 kb of flanking DNAs corresponding to the 5′and 3′ ends of the desired deletion were amplified by PCR. PCR products digested with BamHI and BamHI-digested ΩKm2 element (Perez-Casal et al., 1991) were ligated and cloned into pGEM-T Easy (Promega), generating the plasmids pGEM-tRNA-Ωkm2. The NotI-digested fragments of pGEM-tRNA-Ωkm2 were inserted in the same site of pMW91 (Metcalf et al., 1996). The resulting plasmids were transferred from E. coli S17.1λpir to S. Typhimurium strain SH100 by conjugation. Exoconjugants were selected as described previously (Gotoh et al., 2003). The PCR-based λ Red recombinase system was also used for deletions of tRNA genomic islands (Datsenko & Wanner, 2000). Approximately, 1 kb of flanking DNAs corresponding to the 5′and 3′ ends of the desired deletion, were amplified by PCR. PCR products were digested with SpeI or XhoI. After ligation with these fragments with the SpeI and XhoI-digested FRT-franked kan-resistance gene, which was amplified by PCR using plasmid pKD4 (Datsenko & Wanner, 2000) as a DNA template, the ligated fragment was amplified by PCR. The resulting PCR product was introduced into S. Typhimurium strain SH100 carrying pKD46 (Datsenko & Wanner, 2000) by electroporation. To disrupt genes on the genomic islands, a nonpolar mutation was generated by insertion of the SmaI-digested kan-resistance gene cassette from pUC18K (Menard et al., 1993) into the ORF of the cloned genes on the plasmid. The disrupted genes were amplified by PCR, and the PCR products were introduced into S. Typhimurium strain SH100 carrying pKD46 by electroporation. All disrupted genes were transferred by phage P22 transduction into S. Typhimurium strain SH100 and verified by PCR.

Mouse mixed infections

For the competitive index (CI) assay (Beuzon & Holden, 2001), female BALB/c mice (5–6 weeks old) were used for the mouse infection studies and were housed at Kitasato University according to the standard Laboratory Animal Care Advisory Committee guidelines. Mice were inoculated by intraperitoneal infection with 100 μL of inoculum containing a total of 1 × 105 bacteria, composed of an equal number of the wild-type and mutant bacteria to be tested. At 48 h after infection, mice were sacrificed by carbon oxide inhalation and the spleens were homogenized in cold phosphate-buffered saline by mechanical disruption. The number of each bacterial strain in the spleen was determined by plating a dilution series of the lysate onto LB agar and LB agar with appropriate antibiotics. Each CI value was calculated as described previously (Miki et al., 2004). The CI assays were analyzed by Student's t-test for statistical significance.


Identification of tRNA genomic islands required for virulence of S. Typhimurium in vivo

tRNA-encoding genes are known as a hot spot for the integration of foreign DNAs. Thus, in order to identify the genomic islands required for Salmonella virulence, we selected 14 genomic islands, which were adjacent to tRNA genes and contained >10 ORFs, specifically present in the sequenced S. Typhimurium strain LT2 (McClelland et al., 2001; also reviewed in Porwollik & McClelland, 2003). Five of 14 genomic islands, including SPI-2, SPI-3, SPI-4, and SPI-5, and Gifsy-1 prophage, have already been characterized, but the remaining nine genomic islands, including aspV-, thrW-, argU-, STM1247-, pro-2, glyU-, pheV-, pheU-, and leuX-tRNA, consist of gene clusters that have an unknown function in Salmonella virulence in vivo (Fig. 1a).

Figure 1.

 Genomic islands required for Salmonella virulence. (a) Location of the nine S. Typhimurium genomic islands within the Salmonella chromosome. The genome of ATCC 14028 has not been sequenced, and may differ in size and include integration of phage genomes relative to that of LT2. Black boxes indicate the genomic islands required for the establishment of a mouse systemic infection (see also Table 2). Parentheses indicate genes consisting of the genomic island in LT2. (b) Confirmation of insertions of genomic islands in S. Typhimurium strain 14028 by PCR. PCR products from both LT2 and ATCC 14028 genomes revealed that nine genomic islands are inserted in the same site of LT2. None of the LT2 tRNA-associated islands were amplified from Escherichia coli K-12 strain DH5α genome by PCR using the same primers.

Because S. Typhimurium laboratory strain LT2 is attenuated due to an altered rpoS allele (Wilmes-Riesenberg et al., 1997), we used a virulent strain ATCC 14028 to determine the contribution to virulence of nine tRNA-associated genomic islands. To elucidate whether the tRNA-associated genomic insertions were present at the same site between LT2 and 14028 strains, both 3′ and 5′ flanking regions of the genomic islands were amplified by PCR. PCR primers were selected for one amplicon consisting of the terminal region within the tRNA-associated island, the tRNA gene, and the proximal region outside the tRNA-associated island. A genomic DNA from E. coli K-12 strain DH5α was used for negative control. The results indicated that the genomic insertions examined are conserved between LT2 and 14028 strains (Fig. 1b).

To analyze the role of these nine genomic islands on virulence, deletion mutants were constructed in S. Typhimurium virulent strain 14028, and their virulence was assessed by mouse mixed infections using the CI assay. The control experiment with the wild-type 14028 and its spontaneous nalidixic acid-resistance-derivative strain SH100 resulted in a CI of 1.04 (± 0.21). As shown in Table 2, competition experiments with the wild-type SH100 and nine mutant strains showed that four mutants containing deletions in aspV-tRNA (ΔSTM0266–0307), pro2-tRNA (ΔSTM2230–2245), pheV-tRNA (ΔSTM3117–3138), and leuX-tRNA (ΔSTM4488–4510)associated genomic islands, were attenuated in their virulence for mice. These results suggest that four genomic islands function as pathogenicity islands in S. Typhimurium. Next, we further characterized the pheV-tRNA island because all ORFs encoded in this region are prospective genes with unknown function.

Table 2.   CI values of tRNA locus deletion mutants of S. Typhimurium
SH100 (wt) vs.STM genes deletedCI value*No. of mice
  • *

    The CI value was calculated as the output ratio of mutant to SH100 [wt (wild type)] divided by the input ratio of the deletion mutants to the wt.

  • Significantly different (P<0.05) from CI values obtained from SH100 vs. 14028.

14028 (wt) 1.04 ± 0.216
ΔaspVSTM0266–STM03070.49 ± 0.215
ΔthrWSTM0324–STM03631.33 ± 0.804
ΔargUSTM0554–STM05671.14 ± 0.403
ΔSTM1247STM1239–STM12690.92 ± 0.263
Δpro2STM2230–STM22450.55 ± 0.145
ΔglyUSTM3025–STM30361.27 ± 0.853
ΔpheVSTM3117–STM31380.29 ± 0.198
ΔpheUSTM4305–STM43202.95 ± 0.934
ΔleuXSTM4488–STM45100.41 ± 0.295

Genetic organization of a novel pathogenicity island integrated to the pheV-tRNA

In silico analysis using the S. Typhimurium sequenced strain LT2 revealed that the pheV-tRNA island consists of 26 126 bp and the G+C content of the region is 48%, which is lower than the average 53% G+C content of the LT2 chromosome (McClelland et al., 2001). The 22 predicted ORFs include genes from STM3117 to STM3138 (Fig. 2a). Genomic-based comparative analysis of the pheV-tRNA region between S. Typhimurium and other enteropathogenic bacteria, including Shigella flexneri (Jin et al., 2002), enterohemorrhagic E. coli (Hayashi et al., 2001), and uropathogenenic E. coli (Welch et al., 2002) revealed that heterogenous pathogenicity islands, which differ in size and gene complement, are present among these pathogenic bacteria (data not shown).

Figure 2.

 Genetic maps of pheV-tRNA islands on the genome of S. Typhimurium strain LT2 and serovar Typhi strain CT18 (a). Gene numbers assigned by the sequencing projects are indicated for LT2 and CT18. Arrows represent the size and orientation of genes or putative ORFs. Genes homologous to both strains LT2 and CT18 are indicated as open arrows. Genes specific for S. Typhimurium are indicated as dotted arrows. Genes specific for S. Typhi are indicated as diagonal arrows. (b) The distribution of STM3119 in different Salmonella serovars and enteropathogenic bacteria. The gene-specific probe was used to assess the distribution of STM3119 DNA sequence in Salmonella serovars and other enteropathogenic bacteria using a Southern dot blot assay.

To elucidate whether the pheV-tRNA-associated pathogenicity islands were distributed among different Salmonella serovars, we compared the genetic organization of other S. enterica sequenced strains including serovar Typhi strains CT18 and Ty2, serovar Paratyphi A strain ATCC9150, serovar Chleraesuis strain SC-B67, serovar Enteritidis strain PT4, and serovar Gallinarum strain 287/91. Sequences were obtained from the Salmonella comparative sequencing blast server ( From these sequence data, we found that the pheV-tRNA pathogenicity islands were highly conserved among S. enterica serovars Typhimurium, Choleraesuis, Enteritidis, and Gallinarum, but seven ORFs corresponding to STM3117–STM3123 were completely replaced in the different DNA sequences on the genome of S. enterica serovars Typhi and Paratyphi A (Fig. 2a). Furthermore, the distribution of the DNA sequences was analyzed by Southern hybridization with the genomic DNA from seven Salmonella serovars and five different enteropathogenic bacteria using the STM3119 sequence as a probe DNA. As shown in Fig. 2b, the STM3119-specific sequence was present in nontyphoid Salmonella serovars, but absent in serovar Typhi or other enteropathogenic bacteria tested.

Identification of genes involved in virulence within the pheV-tRNA genomic island

To identify the genes on the pheV-tRNA genomic island required for Salmonella virulence, we first constructed two large deletion mutants of strain 14028 lacking either the left (STM3117–STM3129) or right (STM3130–STM3138) region of the island and determined the effect on virulence by mouse mixed infections. The results showed that while a deletion from STM3130–3138 had no effect on virulence, a deletion from STM3117 to STM3129 attenuated to the same level as a deletion of the entire pheV-tRNA island (Fig. 3). These results indicate that the virulence determinants are located in STM3117–STM3129. Next, we constructed the null mutation in each of the 13 genes from STM3117 to STM3129 and determined the virulence of these mutants in mouse mixed infections. CI analysis by mixed infections showed that mutation in genes STM3118, STM3119, STM3120, and STM3121 had a significant effect on virulence for BALB/c mice (Fig. 3). Successful complementation was achieved for ΔSTM3118 and ΔSTM3119 by the expression of the wild-type STM3118 and STM3119 genes (cloned from strain ATCC 14028) on a low-copy-number plasmid, respectively (Fig. 3). The defect of ΔSTM3120 and ΔSTM3121 was partially restored by the complementation of the wild-type STM3120 and STM3121 genes (cloned from strain ATCC 14028) on a low-copy-number plasmid (Fig. 3). These results suggest that the virulence defect conferred by the deletion mutation in each gene from STM3118 to STM3121 is due to a loss of each gene function.

Figure 3.

 Genes from STM3118 to STM3121are required for virulence in mouse systemic infection by S. Typhimurium. A genetic map of the pheV-tRNA island on the genome of S. Typhimurium is indicated (top). A series of deletion mutants is represented by a solid bar with the corresponding results of the CI assays. Deletions on the chromosome of S. Typhimurium are indicated as open boxes. *Significantly different from the CI of S. Typhimurium ATCC 14028 (wild type) vs. isogenic spontaneous NalR strain SH100 (P<0.05).+Plasmid, the CI values from the wild-type strain vs. a single-gene deletion mutants containing a wild-type copy of the deleted gene on the plasmid are indicated. ND, not done.


Genome sequence analyses of bacteria have revealed that the function of a high proportion of genes remains unknown. Genomic islands are thought to play a role in creating a large novel gene pool (Hsiao et al., 2005) and are often acquired as clusters at tRNA loci by horizontal gene transfer. Thus, tRNAs appear to be good indicators of genomic islands as well as of pathogenicity islands. In this study, we identified four tRNA-associated genomic islands of S. Typhimurium, which are involved in Salmonella pathogenesis.

The aspV-tRNA genomic island, which contains 42 putative ORFs (from STM0266 to STM0307) in the 47-kb DNA region, has been previously described as SPI-6 or SCI (Salmonella enterica centisome 7 genomic island) in S. Typhimurium strains SR11 and ATCC 14028 (Folkesson et al., 1999, 2002). This region encodes the saf fimbrial operon (STM0299–STM0303), the sinR transcriptional regulator (STM0304) in the LysR-family, and the invasin-like pagN (STM0305), which is involved in host cell adhesion and invasion (Folkesson et al., 2002; Lambert & Smith, 2008). In contrast to a previous report that deletion of the entire SCI locus has no effect on mouse systemic infection (Folkesson et al., 2002), we have shown that aspV-tRNA genomic islands contribute to Salmonella pathogenesis using mouse mixed infection. Recently, two other genes on the aspV-tRNA genomic island of strain ATCC 14028 have been identified as virulence-associated genes, including sciS (STM0285), which contributes to limited intracellular replication and virulence in mice (Parsons & Heffron, 2005), and STM0272, which encodes a putative ClpB-like protease required for bacterial growth within macrophages (Klumpp & Fuchs, 2007), suggesting that the lack of these genes results in an attenuation of Salmonella virulence.

pro2-tRNA-associated genomic island of strain LT2 encode 16 putative ORFs (from STM2230 to STM2245) in a 15-kb region, as has been described previously (Hansen-Wester & Hensel, 2002). Several virulence-associated factors encoded by genes located on pro2-tRNA genomic island of strain ATCC 14028 genome, including sspH2 (STM2241) that encodes an effector protein of the SPI-2 type III secretion system (Hansen-Wester & Hensel, 2002), and oafA, which encodes an O-antigen (lipopolysaccharide) acetylase (Slauch et al., 1996), might be essential for Salmonella pathogenesis. In addition, many prophage-like genes present in this island remain uncharacterized.

The 34-kb DNA region, located at leuX-tRNA on the genome of S. Typhimurium strain LT2, contains 23 putative ORFs (from STM4488 to STM4510). Interestingly, the leuX-tRNA genomic island is the most variable region among Salmonella serovars (Bishop et al., 2005). For example, the genes located in this locus in the S. Typhi, termed SPI-10, encode a P4-like phage, the sef and pef fimbrial operons, IS element remnants, and genes of unknown function, but those in serovar Typhimurium are replaced by unrelated genes. In the leuX-tRNA region of S. Typhimurium, only uxuR (STM4507), which is the transcriptional repressor belonging to the uxu regulon involved in sugar metabolism in γ purple bacteria (Rodionov et al., 2000) has been characterized.

The pheV-tRNA genomic island on the genome of S. Typhimurium strain LT2 encodes 22 putative ORFs. Analyses of pheV-tRNA genomic islands in silico reveal that this is a hot spot for divergence within Salmonella and other enteropathogenic bacteria. Probably, the presence of the sequences homologous to the attP attachment site sequence of P4-like phages (GAGTCCGGCCTTGGCACCA) in the pheV-tRNA regions of E. coli, Shigella, and Yersinia as well as Salmonella may drive the hypervariability of this region (Manson & Gilmore, 2006). In addition, the genetic structure at the pheV-tRNA region of S. Typhimurium is partly different from that of serovar Typhi. Genes from STM3117 to STM3123 in S. Typhimurium located between the duplicated attP homologous sequences are completely replaced by genes from STY3273 to STY3292 of the pheV-tRNA genomic island of S. Typhi, indicating that the phage attachment site sequence plays a role as drivers of diversity in S. enterica serovars.

Our results demonstrate that four virulence-associated genes, STM3118, STM3119, STM3120, and STM3121, which are required for Salmonella mouse systemic infection, are encoded on the pheV-tRNA genomic island of S. Typhimurium strain ATCC 14028. It has been reported that peptides of STM3117–STM3120 have been identified previously from a proteomic analysis of S. Typhimurium strain ATCC 14028 isolated from the spleen of infected mice (Becker et al., 2006) and that expression of STM3117, STM3118, and STM3119 have been induced during Salmonella infection of macrophages (Shi et al., 2006). STM3117, STM3118, and STM3119 were annotated as putative lactoylglutathion lyase, putative acetyl-CoA hydrase, and a monoamine oxidase, respectively. It is therefore likely that STM3117–STM3119 function in biosynthesis and modification of the peptidoglycan layer of the cell wall (Shi et al., 2006). STM3120 encodes a possible citrate lyase and STM3121, termed stmR (Morrow et al., 1999), is a putative transcriptional regulators of the LysR family.

The importance of these genes in the pheV-tRNA genomic island for Salmonella pathogenesis has recently been described in other Salmonella serovars. In fact, cat2, which is identical to STM3118 in S. Enteritidis is required for growth within chicken macrophages, while the role of other genes in this region in the virulence of S. Enteritidis remains unknown (Zhao et al., 2002). In addition, the transposon insertion mutations in genes identical to STM3118, STM3120, and STM3121 in S. Gallinarum-attenuated virulence in a chicken infection model (Shah et al., 2005). Furthermore, STM3118–STM3121 in S. Typhimurium are also conserved in the pigment (pgm) locus in part of the 102-kb pathogenicity island of Yersinia pestis containing ripA (Y2385), ripB (Y2384), Y2383, and Y2382. It was recently shown that RipA is essential for Y. pestis to survive in interferon γ-activated macrophages (Pujol et al., 2005). These data show that Salmonella and Yersinia possess a common defense mechanism for activated macrophages.


This work was supported in part by Grants-in-Aid for Scientific Research (C) (17590398 and 18590435) and for Young Scientists (B) (17790291) from the Japanese Ministry of Education, Culture, Sports, Sciences, and Technology.