The putative iron transport system SitABCD encoded on SPI1 is required for full virulence of Salmonella typhimurium



Salmonella typhimurium is an invasive pathogen that causes diseases ranging from mild gastroenteritis to enteric fever. During the infection process, S. typhimurium induces a number of virulence genes required to circumvent host defences and/or acquire nutrients in the host. We have used the in vivo expression technology (IVET) system to select for S. typhimurium genes that are induced after invasion of a murine cultured cell line. We have characterized a putative iron transporter in Salmonella pathogenicity island 1, termed sitABCD. The sitABCD operon is induced under iron-deficient conditions in vitro and is repressed by Fur. This locus is induced in the animal specifically after invasion of the intestinal epithelium. We show that a sit null mutant is significantly attenuated in BALB/c mice, suggesting that SitABCD plays an important role in iron acquisition in the animal.


Salmonella spp. cause a range of diseases, from mild gastroenteritis to enteric fever in humans and animals. Infection of mice with Salmonella enterica serovar Typhimurium (S. typhimurium) approximates the human disease caused by S. typhi, and is a model for invasive salmonellosis (Miller et al., 1995). Salmonella spp. primarily infect the host orally via contaminated water or foodstuffs. During the early stages of S. typhimurium infection, the bacteria survive the acid barrier of the stomach to reach the small intestine, where they breach the intestinal epithelial barrier primarily through the M cells of the Peyer’s patch (Carter and Collins, 1974; Jones et al., 1993). Invasion of the intestinal epithelium is facilitated by the type III secretion system that is encoded on Salmonella Pathogenicity Island 1 (SPI1), a 40 kb gene cluster at centisome 63 of the chromosome (for review see Darwin and Miller, 1999; Galan and Collmer, 1999). After invasion, the bacteria survive and propagate within the Peyer’s patch and then spread to target organs, such as the liver and spleen (Carter and Collins, 1974). The ability to survive in macrophages is thought to be critical for S. typhimurium during these systemic stages of disease (Gulig et al., 1998).

Salmonella typhimurium induce numerous virulence factors to combat the host immune system and cause disease. These include conventional virulence determinants, e.g. type III secretion systems, which carry out distinct functions in the host. In addition, Salmonella must acquire scarce nutrients from the host environment, enabling them to survive and propagate at the site of infection. One essential nutrient is iron (Griffiths, 1987), which is required especially during rapid multiplication in systemic stages of infection (Gulig, 1996). Pathogenic bacteria must acquire iron in the presence of host-encoded high-affinity iron-binding proteins that act as a non-specific defence mechanism (Griffiths, 1987).

Salmonella typhimurium has evolved or acquired a number of iron uptake systems (Earhart, 1996). Iron(II) is transported, under anaerobic conditions, by the Feo system (Kammler et al., 1993). Iron(III) can be acquired using a variety of low molecular weight chelators termed siderophores. For example, enterobactin is produced and secreted into the environment. The enterobactin–iron(III) complex is subsequently transported into the cell by the Feo system (Earhart, 1996). Some strains of S. typhimurium can also produce aerobactin, which is usually plasmid encoded (Reissbrodt and Rabsch, 1988). In addition, S. typhimurium can transport a variety of siderophores, including ferrichrome, ferrioxamines, pyoverdines, myxochelins and coprogen, that are produced by other organisms (Luckey et al., 1972; Kingsley et al., 1995). Certain metabolites, such as 2,3-dihydroxybenzoic acid and dihydroxybenoylserines, can also act as siderophores (Earhart, 1996). Reissbrodt et al. (1997) have suggested that alpha-keto acids, which are excreted at high levels in low-iron conditions, can function in iron uptake in S. typhimurium. However, S. typhimurium is unable to utilize haem as an iron source (Reissbrodt et al., 1997) and is devoid of a ferric citrate transport system (Wagegg and Braun, 1981).

Although many of these systems have been well studied in vitro, it is not clear what the primary source of iron is during an infection. Mutations in several of these systems have been tested in animal models and have had little or no effect on virulence (Benjamin et al., 1985; Bäumler et al., 1996; Tsolis et al., 1996). The lack of an in vivo phenotype has been attributed to the redundancy of these systems. Very recently, Pattery et al. (1999) have reported the identification of a novel iron and pH-inducible putative ABC-type transport system, sfbABC. A mutation in sfbABC confers a dramatic virulence defect. However, it is not clear that sfbABC is an iron transport system. Most of the known siderophore receptor systems of S. typhimurium depend on the protein TonB to energize transport across the outer membrane. Tsolis et al. (1996) reported that tonB feo double mutants were largely unaffected in intraperitoneal (i.p.) LD50. Thus, even simultaneously affecting a number of known iron transport systems seems to have little effect on the virulence of S. typhimurium in the animal. This implies that other perhaps TonB-independent systems function or are capable of functioning during the infection process.

In vivo expression technology (IVET) is designed to select bacterial genes that are induced during an infection. The pIVET1 system is based on complementation of a purA auxotroph using transcriptional gene fusions to a promoterless purA gene. PurA is required during all stages of infection. Therefore, only those bacteria containing a transcriptionally active fusion can survive an IVET selection. The fusion construct is integrated into the chromosome by homologous recombination, thereby creating a merodiploid, where one copy of a promoter drives the fusion while the second copy controls the wild-type gene of interest. The vector also contains a promoterless lac operon 3′ to the promoterless purA, such that the transcriptional activity of any given fusion can be easily monitored (Mahan et al., 1993a; Slauch et al., 1997). Using IVET, we have identified the putative iron transport operon, sitABCD, encoded in SPI1 (Zhou et al., 1999). SitABCD is a member of a subfamily of ABC transport proteins that are identified by their periplasmic binding proteins (Boos and Lucht, 1996). We show here that sitABCD is preferentially induced in the animal during the systemic stages of the disease and that a sitA mutant is significantly attenuated in infection of BALB/c and C3H/HeN mice. Thus, sitABCD is required for full virulence of S. typhimurium and maybe a primary iron transport system in vivo.


IVET selection in tissue culture

Our initial goal was to identify genes that are induced during systemic stages of S. typhimurium infection in the animal. As one part of this analysis, we performed an IVET selection in cultured murine hepatocytes. The liver is a major target organ in the later stages of Salmonella infection (Carter and Collins, 1974), and, although the role of hepatocytes in infection is debatable (Gulig et al., 1998), S. typhimurium can apparently replicate in hepatocytes under some conditions (Conlan, 1997; Gulig et al., 1998). Prior to performing the selection, we established that (1) S. typhimurium can invade and replicate in tissue culture hepatocytes, and (2) under the conditions outlined in the Experimental procedures, there is a 100-fold selection for Pur+ bacteria in mixed infections of a purA mutant and wild type in these cells.

Starting with an integrated pIVET1 library (see Experimental procedures), we performed two rounds of IVET selection in cultured hepatocytes. In order to confirm that a selection for transcriptionally active promoters was taking place within the hepatocytes, the population of bacteria prior to inoculation was compared with that recovered after selection by monitoring the lac activity of the pIVET1 fusions on lactose MacConkey agar (Slauch et al., 1997), an iron-replete medium (see below). The preselection pool consisted of ≈ 21% Lac+ (red), 12% Lac+/– (pink or intermediate phenotype) and 67% Lac (white) colonies. After two rounds of selection in cultured hepatocytes, 80% of the recovered colonies were Lac+. This proves that there was a selection for transcriptionally active fusions in cultured hepatocytes. Seven per cent of the colonies recovered after selection were Lac. These Lac colonies presumably contain fusions to genes that are transcriptionally induced in the tissue culture system. Approximately 50 Lac colonies were chosen for further analysis. The DNA sequence of the chromosomal insert immediately upstream of the fusion joint was determined after recovering the original IVET fusion plasmid by P22 transduction (Mahan et al., 1993b). The 50 Lac colonies represented fusions to 38 independent genes. A number of fusions to genes in known virulence regulons such as sodA (manganese superoxide dismutase), pagJ (a PhoPQ-activated gene) and ssaE (in Salmonella pathogenicity island 2, SPI2) were identified. SPI2 is involved in the later stages of Salmonella virulence (Ochman et al., 1996; Shea et al., 1996). Thus, our selection was clearly targeting genes involved in S. typhimurium pathogenesis. We also identified a fusion in the second gene of an operon encoding an apparent ABC-type transporter that was not present in the Escherichia coli genome. Our preliminary sequence analysis revealed that this locus is situated in SPI1, located at centisome 63 in the Salmonella chromosome (Fig. 1; Darwin and Miller, 1999; Galan and Collmer, 1999). Recently, this locus has also been identified by Zhou et al. (1999) and designated sitABCD (SalmonellaIron Transporter). These authors provided evidence that this locus is involved in iron transport. Indeed, we also identified this locus in an IVET selection for genes induced under low-iron conditions (data not shown). Also, the putative SitABCD proteins show significant homology to the YfeABCD iron transport system of Yersinia pestis (Bearden et al., 1998).

Figure 1.

The sitABCD region of SPI1. The position of the sitB IVET fusion joint is 27 bp into the sitB gene. The sitA100::MudCm insertion is 662 bp into the sitA gene. H, HindIII. The DNA sequence of this region is contained in GenBank entries AF128999, U16278, AF013573 and AF148689.

The sitABCD operon is induced during the systemic stages of the disease

As the sit locus was identified using an IVET selection in a tissue culture system, we wanted to determine whether this operon was induced in the animal. To accomplish this, we performed a competition assay between the sitB IVET fusion strain (JS140, Lacin vitro) and a strain containing an IVET fusion that is constitutively expressed (JS145, Lac+in vitro). The expression level of this constitutive fusion is sufficient to allow this strain to propagate in any tissue in the animal (data not shown). If the sit locus is induced in the mouse, the sit fusion strain should be able to compete with the strain containing the constitutively expressed fusion and we should recover approximately equal numbers of the two strains from the animal. A 1:1 mixture of the two strains was administered i.p. and orally to two groups of six BALB/c mice each. In i.p. competition assays, there was no significant difference in the number of sit fusion and constitutive IVET fusion bacteria recovered from the liver [median competitive index (CI) = 1.20], suggesting that the sitB fusion is transcriptionally induced in this organ. In contrast, after oral inoculations, the sitB fusion strain was out-competed by the constitutive fusion strain in the small intestine (median CI = 0.30). This defect is further amplified in bacteria recovered from the spleens of the orally infected mice (median CI = 0.14), suggesting that the sitB fusion strain had diminished ability to survive initial gut colonization and gain access to the systemic tissue. Taken together, these results suggest that sitABCD was transcriptionally induced only after invasion of the intestinal mucosa.

The sitABCD operon is Fur regulated

Both the homology with known iron transport proteins and the identification of sitABCD in an IVET selection for genes induced under low-iron conditions led us to examine the regulation of this fusion in vitro under iron-starvation conditions. The sit operon is induced fivefold under iron-limiting conditions (Table 1). Many genes that are induced under iron-starvation conditions are regulated by Fur (Stojiljkovic et al., 1994). We transduced a fur null mutation into the strain containing the sitB fusion in an otherwise wild-type background and assayed β-galactosidase levels. The sit locus was constitutively expressed in the fur background, indicating that Fur acts as a repressor of sitABCD (Table 1). This is consistent with the results of Zhou et al. (1999), who also noted a potential ‘Fur Box’, or Fur-binding region upstream of the putative sit promoter. Given the location of sitABCD, we tested whether HilA or PhoPQ, known regulators of other SPI1 genes, had any role in sitABCD transcription. Expression of the sitB fusion was unaffected in strains containing mutations in these regulatory loci (Table 1). We also monitored the expression of the sitB fusion in response to a variety of other environmental conditions, such as changes in pH, osmolarity, temperature, magnesium concentration and conditions that induce the OxyR regulon (1 mM H2O2). The expression of sitABCD was also monitored after growth in conditions that induce expression of the SPI1 invasion system (Bajaj et al., 1996). None of these conditions had any significant effect on sitABCD expression (data not shown). Thus, sitABCD is primarily regulated by Fur in response to iron levels.

Table 1. Expression of the sitB IVET fusion.
 Lac phenotypebβ-Galactosidase activityc 
BackgroundaLow ironHigh ironLBLow ironHigh iron
  1. a . All strains are isogenic Pur + strains containing the sitB IVET fusion: JS141, JS146, JS147, JS143, JS144, JS148.

  2. b . Lactose MacConkey agar with 200 µ m 2, 2′-dipyridyl (low iron) or 40 µm FeSO4 (high iron).

  3. c . LB with 200 µ m 2, 2′-dipyridyl (low iron) or 40 µm FeSO4 (high iron). Units are defined as (µmol of ONP formed min−1) × 103/(OD600 × ml of cell suspension). Data presented as mean ± standard deviation, where n = 4.

Wild typeRedWhite9.4 ± 0.250.0 ± 0.99.2 ± 0.1
fur-1 RedRed65.6 ± 0.050.1 ± 2.760.0 ± 1.1
hilA339::kan RedWhite10.5 ± 0.8NDND
phoP102::MudCmRedWhite9.8 ± 0.7NDND
phoQ24 RedWhite10.5 ± 0.6NDND
sitA100::MudCmWhiteWhite7.2 ± 0.16.0 ± 1.28.9 ± 0.2

The sitABCD locus is required for full virulence

Our data indicated that the sitABCD operon was induced during the later stages of infection in the animal. We wanted to determine whether this locus had a significant role in pathogenesis by asking if a sit null mutation conferred a virulence defect. First, we isolated a MudCm insertion mutation in the sitA gene (see Experimental procedures). Sequence analysis showed that the insertion was 662 bp into the 917 bp sitA orf, 282 bp upstream of the pIVET1 fusion joint in sitB (Fig. 1). We recombined the sitA100::MudCm insertion onto the sitB pIVET1 fusion plasmid and integrated this construct into the chromosome of an otherwise wild-type strain such that the sitA100::MudCm was 5′ to the sitB fusion joint. We then monitored β-galactosidase activity produced from the fusion. The sitA100::MudCm insertion completely prevented induction of sitB in response to low iron (Table 1). Thus, the insertion mutation knocks out sitA and is polar on the downstream genes of the sit operon.

Using isogenic wild-type and sitA100::MudCm strains, we performed competition assays, inoculating BALB/c mice i.p. or orally. In i.p. competition assays, the sitA mutant was consistently out-competed by the wild-type strain in the spleens and livers of mice (Table 2). These assays were repeated six times in a total of 36 BALB/c mice, and the results are reproducible and statically significant. In oral competition assays, the sitA mutant was out-competed in bacteria recovered from the spleen and small intestine, which primarily comprises the gut-associated lymphoid tissue (Table 2). These data show that the sitA100::MudCm insertion confers a defect in growth or survival in the animal host.

Table 2. In vivo competition assays between sitA100::MudCm and wild-type strains.
Route of
Plasmid in
sitA100::MudCm b


Median CIc
No. of

  • a . All inoculum contained a 1:1 ratio of sitA100::MudCm (JS142) to isogenic wild type (JS107). Dose: 10 2 cfu i.p., 108 cfu oral, BALB/c; 105 cfu i.p., 109 cfu oral, C3H/HeN.

  • b

    . pAJ107 contains the entire sit locus and pWKS30 is the parent vector.

  • c

    . Competitive index = output(mutant/wild type)/input(mutant/wild type).

  • d

    . Student’s t-test comparing output versus inoculum.

  • e

    . These BALB/c and C3H/HeN mice were infected concurrently. The results of these BALB/c mice are not included in the data at the top of the table.

BALB/ci.p. Liver0.2326<0.005
BALB/cOral Small intestine0.3020<0.05
BALB/cei.p. Spleen0.266<0.05
C3H/HeNe  Spleen0.166<0.005
C3H/HeNOral Small intestine0.096<0.005

To further test if this attenuation phenotype was attributable to the sit locus, we inoculated a group of BALB/c mice i.p. with a 1:1 mixture of the wild-type and the sitA mutant strain. In these experiments, the mutant strain carried either a plasmid clone of the entire sitABCD operon or the parent vector. The attenuation phenotype was complemented by the presence of the sitABCD+ plasmid, whereas the parent pWKS30 vector was unable to alleviate the defect (Table 2). These data indicate that attenuation of the sitA100::MudCm strain is a result of loss of sitABCD function.

Finally, we performed both oral and i.p. competition assays using ityr C3H/HeN mice. The results (Table 2) suggested that the sitA100::MudCm strain is slightly more attenuated in the C3H/HeN mice compared with the itys BALB/c mice. However, although the sitA mutant is significantly attenuated in both mouse strains, the number of animals in this experiment was not large enough to statistically confirm a difference in the severity of the defect in BALB/c versus C3H/HeN mice.

The sit null mutation confers a growth defect under certain iron-limiting conditions

We wanted to know if the attenuation conferred by the sitA100::MudCm insertion is due to a simple inability to transport iron. To address this issue, we monitored the growth phenotype of the sitA100::MudCm and wild-type strains in LB or glucose minimal media containing various concentrations of the intracellular iron chelator, 2,2′-dipyridyl, or the extracellular iron chelator, DTPA. Both wild-type and mutant cultures reached approximately the same OD600 after 24 h when grown individually in LB with the intracellular iron chelator dipyridyl (Fig. 2). However, the sit mutant showed impaired growth compared with the wild type in Luria–Bertani (LB) containing certain concentrations of the extracellular iron chelator, DTPA. In contrast, 2,2′-dipyridyl, but not DTPA, specifically inhibited growth of the mutant at certain concentrations in glucose minimal medium (Fig. 2). Thus, the sitA null mutation does confer an in vitro growth phenotype under particular conditions. In separate experiments, we monitored growth of the sit mutant and wild type under conditions where both reach equivalent OD600 at 24 h. Both strains gave virtually identical growth curves. Thus, there is no apparent phenotype that is not reflected in the results in Fig. 2.

Figure 2.

In vitro growth phenotype of wild-type and sitA100::MudCm in iron-chelated media. The wild type is shown in filled circles with solid lines and the sitA::MudCm mutant is in open circles with dashed lines. Glucose minimal = E minimal medium with 0.4% glucose and 0.1% casamino acids. Two cultures are represented in each datum point. The error bars represent the actual OD600 of each separate culture, while the circle is the average of the two.

The sit locus is not involved in invasion

SPI1 encodes a type III secretion system required for invasion of cultured intestinal epithelial cells (Lee et al., 1992; Darwin and Miller, 1999; Galan and Collmer, 1999). To determine whether the sit locus played a invasion role, we performed invasion assays with wild-type and the sitA null strains in cultured hepatocytes and epithelial cells. Both wild-type and the sitA insertion mutant showed equal ability to invade cultured Henle407 epithelial cells and hepatocytes (data not shown). These data are consistent with the results obtained by Zhou et al. (1999). Thus, in contrast to the other loci encoded on SPI1, the sit locus is not involved in S. typhimurium invasion. Note that we are only examining initial invasion of the tissue culture cells in this experiment and therefore this result is not inconsistent with the defect in growth in the small intestine conferred by the sitA100::MudCm insertion.


We performed an IVET selection in cultured murine hepatocytes. Although we did not explicitly screen for hepatocyte-specific genes, this selection clearly targeted virulence regulons; fusions to pagJ (PhoP activated), ssaE (in SPI2) and sodA (superoxide dismutase) were isolated. We also identified a fusion to the sitABCD operon, located in SPI1. Zhou et al. (1999) recently reported the identification and molecular characterization of this operon and provided evidence that the system is involved in iron transport. The locus is highly homologous to the yfe locus in Y. pestis, which has been shown to transport iron and manganese and is required for full virulence in an animal model of plague (Bearden et al., 1998; Bearden and Perry, 1999). Zhou et al. (1999) reported that a sitBCD deletion mutation had no effect on virulence in an oral infection, although the authors do not state how this was determined. Here, we have shown that the sitABCD operon is preferentially expressed during the systemic stages of S. typhimurium infection in mice. We also provide evidence that a sit null mutation confers a virulence defect. Although this phenotype is subtle and was discovered using carefully controlled competition assays, it is reproducible and statistically significant.

Iron acquisition is critical for the growth of pathogens (Benjamin et al., 1985; Earhart, 1996), and the host actively reduces the availability of extracellular iron during an infection as a non-specific defence mechanism (Woolridge and Williams, 1993). The intracellular vacuole in which Salmonella reside is also limiting for iron (Garcia-del Portillo et al., 1992). Therefore, it is not surprising that Fur regulated, low-iron-induced genes such as sitABCD and sodA would answer an IVET selection in cultured hepatocytes and that these genes are also induced in the host.

S. typhimurium possess a number of iron uptake systems. However, the relative contribution of each of these systems to the infection process is not well understood. In the anaerobic environment of the small intestine, soluble iron(II) is apparently available for uptake by the feo system (Tsolis et al., 1996). Our result showing that the sitB IVET fusion strain does not compete well after oral infection is consistent with this hypothesis, implying that the Fur regulon is not induced in the small intestine. Tsolis et al. (1996) reported that a S. typhimurium feo mutant is only slightly impaired in small intestine colonization. However, the Fur regulon is induced in vitro in a feo mutant (Hantke, 1987). Thus, it is likely that the Fur–dependent iron(III) transport systems, once induced, can compensate for feo in the small intestine.

After invasion of the intestinal epithelium, iron presumably becomes limiting and the Fur regulon is normally induced (Tsolis et al., 1996). The majority of S. typhimurium that are recovered from the small intestine are associated with Peyer’s patches (Hohmann et al., 1978); these bacteria are the progeny of those cells that initially invaded the intestinal epithelium. The fact that the sit null mutant is attenuated in the small intestine suggests that SitABCD is important for growth in all tissues after invasion of the epithelium. This is not inconsistent with the fact that the sitB IVET fusion strain cannot initiate an intestinal infection, but rather, suggests that induction takes place after invasion.

S. typhimurium synthesizes the siderophore enterobactin for iron(III) uptake (Luckey et al., 1972). Two studies have shown that enterobactin biosynthetic mutants are not significantly attenuated in a mouse model of infection (Benjamin et al., 1985; Tsolis et al., 1996; but also see Yancey et al., 1979), although they are affected in their ability to grow in normal mouse sera in vitro (Tsolis et al., 1996). S. typhimurium can also utilize siderophores, such as ferrichrome (Luckey et al., 1972) and ferrioxamines (Earhart, 1996), that are produced by other organisms. Uptake of ferrioxamines is mediated by the outer membrane protein foxA. Kingsley et al. (1999) have recently shown that foxA mutants are dramatically attenuated, both orally and i.p., in BALB/c mice. However, ferrioxamine uptake, like other known siderophore transport, is largely dependent on the energy-transducing protein, tonB (Kingsley et al., 1999). Using LD50 assays, Tsolis et al. (1996) reported that tonB feo double mutants were largely unaffected after i.p. infection. Thus, mutating foxA has a significantly greater affect than mutating tonB, which is required for foxA-mediated iron transport. This suggests, as noted by Kingsley et al. (1999), that the severe defect observed in the foxA mutant is not a result of simple iron deficiency.

Our data indicate that a mutation in the sit locus confers a virulence phenotype in the presence of functionally redundant systems, suggesting that SitABCD may play a major role in iron acquisition during infection. However, the primary iron source in the host is not clear. As noted, S. typhimurium are adept at using a large variety of metabolites and siderophores produced by a number of organisms for iron uptake (Luckey et al., 1972; Earhart, 1996; Reissbrodt et al., 1997). Perhaps a siderophore or siderophore-like substance is the primary source of iron in the host. SitABCD, however, resembles a group of ABC transporters that have been proposed to transport iron from the periplasm across the cytoplasmic membrane in a siderophore-independent fashion (Zhou et al., 1999). An example of this group is the fbpABC transporter of Neisseria gonorrhoeae. Human transferrin is bound to the gonoccocal outer membrane protein, Tbp1 (Cornelissen et al., 1992). Iron is released into the periplasm where it is bound by the periplasmic binding protein, FbpB and subsequently transported into the cytoplasm (Adhikari et al., 1996). SitABCD may function in an analogous manner, although a putative outer membrane receptor and the appropriate ligand have not been identified. We are further characterizing the sit system to determine its substrate specificity and TonB dependence in an effort to understand the role of iron acquisition during the various stages of disease.

Experimental procedures

Bacterial culture growth conditions

All S. typhimurium strains used in this study are isogenic derivatives of strain ATCC 14028 (Table 3). Bacteria were cultured aerobically at 37°C in LB broth or on LB agar plates unless otherwise stated. LB and E minimal medium were prepared as described (Maloy et al., 1996). Antibiotics when required were included at the following concentrations: kanamycin (Kn), 50 µg ml−1; chloramphenicol (Cm), 20 µg ml−1; tetracycline (Tc), 25 µg ml−1; ampicillin (Ap), 50 µg ml−1. Strains containing ΔpurA were grown in rich media supplemented to 27 µg ml−1 adenine and 16 µg ml−1 thiamine (AdB1). Lactose MacConkey agar was used to monitor the lac gene expression.

Table 3. Bacterial strains and plasmids.
Strains or plasmidsGenotypeaReference/Sourceb
  • a

    . All strains are isogenic derivatives of 14028.

  • b

    . This study unless otherwise noted.

S. typhimurium
 14028Wild typeATCC
 JS107 zjg8101::kan Mann and Slauch (1997)
 JS130 zjg8103::pir 
 JS140ΔpurA3141::kan pIVET1::sitB′ 
 JS141 zjg8101::kan pIVET1::sitB′ 
 JS142 sitA100::MudCm 
 JS143JS141 phoP102::Tn10dCm 
 JS144JS141 phoQ-24 zce850::Tn10dTc 
 JS145ΔpurA3141::kan Constitutive pIVET1 fusion 
 JS146JS141 fur-1 zbg8107::MudCm 
 JS147JS141 hilA339::kan 
 JS148JS141 sitA100::MudCm 
 pAJ1156pGP704::‘sitA sitB′ 
 pWKS30pSC101 ori, Apr Wang and Kushner (1991)

IVET selection in tissue culture cells

The IVET library (T. L. Stanley and J. M. Slauch, unpublished data) contains random 4–7 kb S. typhimurium chromosomal fragments cloned into the pIVET1 vector (Mahan et al., 1993a; Slauch et al., 1997). This plasmid library has been integrated into the chromosome of the ΔpurA strain, JS120, by homologous recombination. Thus, each fusion-containing strain is merodiploid, where one copy of a promoter drives the fusion while the second copy controls the wild-type gene of interest. The integrated library was grown in LB AdB1 under invasion conditions (Lee and Falkow, 1990). Invasion assays were performed essentially as described (Lee et al., 1992). Briefly, ≈ 105 hepatocytes (Hepa 1–6; ATCC) were grown in DMEM (Dulbecco’s modified Eagle medium, Sigma) containing 10% FBS (fetal bovine serum, Sigma) in individual wells of 24-well tissue culture plates. Approximately 1 × 107 colony-forming units (cfu) of the IVET library was then mixed with the eukaryotic cells and incubated for 30 min at 37°C in 5% CO2 to allow invasion of bacteria. The cells were then washed with PBS (phosphate-buffered saline). DMEM containing the antibiotic gentamicin (Gm) at 100 µg ml−1 was added and the cells were incubated for an hour. The cells were washed with PBS, and DMEM containing Gm at 6 µg ml−1 was added and the cells were incubated for 24 h. The wells were washed with PBS to remove the Gm and the eukaryotic cells were lysed with 1% Triton X-100. The bacterial cells were recovered and enumerated by plating serial dilutions on lactose MacConkey agar containing Ap Kn AdB1. The remainder of the post selection library was grown in LB Ap AdB1 and the selection was repeated as described above. After two rounds of selection, 50 colonies that had a Lac phenotype were chosen for further analysis.

DNA sequence analysis

The pIVET::sitB′ fusion plasmid was recovered via P22 transduction (Mahan et al., 1993b) into JS130, a Pi-producing derivative of S. typhimurium strain ATCC 14028. Sequencing was carried out by the W. M. Keck Center for Comparative and Functional Genomics at the University of Illinois, Urbana-Champaign, IL, using a primer to the 5′-end of the purA gene sequence of pIVET1. Sequence analyses were carried out using the GCG package (Wisconsin-Madison).

Isolation of a sit insertion mutation

The plasmid pAJ1156 contains a 500 bp PCR fragment corresponding to the region immediately upstream of the original pIVET::sitB′ fusion joint in the Pi-dependent plasmid, pGP704 (Miller and Mekalanos, 1988). Random MudCm insertions were isolated in the Pi+ strain JS130 containing pAJ1156 as previously described (Hughes and Roth, 1988). Cmr colonies were pooled and plasmid DNA was purified and used to transform JS130 for Apr and Cmr, thus identifying plasmids containing MudCm insertions. The Apr Cmr transformants were pooled and a P22 lysate was made and used to transduce the Pi strain JS107 selecting Cmr and scoring Aps. These Cmr, Aps transductants are the result of MudCm insertions in the 500 bp chromosomal fragment of the plasmid that have been subsequently recombined into the chromosome, facilitated by P22 plasmid transduction (Mann and Slauch, 1997). The location of the resulting sitA100::MudCm was determined by DNA sequence analysis.

Cloning of the sit operon

The sit operon was cloned starting with the sitA100::MudCm strain. The restriction map of SPI1 (Penheiter et al., 1997) indicated that there were no HindIII sites in the region presumed to contain sitABCD. Chromosomal DNA extraction was performed using the Qiagen protocol (Qiagen Inc) and digested to completion with HindIII. DNA fragments corresponding to ≈ 15 kb were isolated and cloned into the pSC101 derivative, pWKS30 (Wang and Kushner, 1991). The sitABCD clone was identified by selecting Cmr. The corresponding sit+ plasmid, pAJ107, was obtained by recombination via P22 transduction to remove the MudCm insertion (Mann and Slauch, 1997). The DNA sequence of the sitABCD operon was recently reported by Zhou et al. (1999) and is entered in GenBank under accession number AF128999.

Virulence studies and competition assays

Six to 10-week-old BALB/c or C3H/HeN mice (Harlan) were inoculated either orally or i.p. with 0.2 ml of a bacterial suspension diluted in sterile 0.2 M sodium phosphate buffer, pH 8.0 (oral) or sterile 0.15 M NaCl (i.p.). After 4–7 days, the mice were sacrificed by CO2 asphyxiation and the small intestine (oral), spleen (oral and i.p.) and/or one lobe of the liver (i.p.) were harvested. These organs were homogenized, and serial dilutions were plated on the appropriate media to determine cfu per organ. In all competition assays, the inoculum consisted of a 1:1 mix of two bacterial strains. The actual cfu and relative percentage represented by each strain was determined by direct plating of the inoculum. The Student’s t-test was used for all statistical analyses.

In vitro growth under iron-limiting conditions

Overnight LB cultures of both wild-type and sitA100::MudCm strains were washed and resuspended in an equal volume of 0.15 M NaCl (for LB assays) or E minimal medium containing 0.4% glucose and 0.1% casamino acids. A 1:500 dilution of each suspension was made in either LB or E minimal medium containing 0.4% glucose and 0.1% casamino acids respectively. Aliquots (100 µl) of these dilutions were used to inoculate 96-well microtitre plates such that the media and final concentration of iron chelator is as indicated in Fig. 2. Microtitre plates were incubated at 37°C in a Jitterbug microplate shaker (Boekell Industries). After 24 h, the OD600 was read using a 7520 microplate reader (Cambridge Technology). Two independent colonies of each strain were tested.

β-Galactosidase assays

All assays were performed in a Pur+ background to ensure that there was no selection for expression of the sitB IVET fusion. Expression of the IVET fusion was determined following overnight growth under various environmental conditions. Previous experiments had shown that growth phase has no affect on the induction of this locus. β-Galactosidase assays in microtitre plates were performed as described previously (Slauch and Silhavy, 1991).


We thank R. Edwards and M. Kim for critically reading the manuscript; C. Lee for generously providing us with the hilA339::Kn allele and for helpful discussions; J. Foster for the fur-1 allele; D. Essex-Sorlie for her invaluable help with statistical analyses; R. Perry for sharing data prior to publication; and students at the Cold Spring Harbor Laboratory Advanced Bacterial Genetics Course, who carried out the low-iron IVET selection. This study was supported by NIH grant AI37530 and ACS Junior Faculty Research Award JFRA-633.