YbtP and YbtQ: two ABC transporters required for iron uptake in Yersinia pestis


Robert D. Perry. E-mail rperry@pop.uky.edu; Tel. (+1) 606 323 6341; Fax (+1) 606 257 8994.


Yersinia pestis, the causative agent of plague, makes a siderophore termed yersiniabactin (Ybt), which it uses to obtain iron during growth at 37°C. The genes required for the synthesis and utilization of Ybt are located within a large, unstable region of the Y. pestis chromosome called the pgm locus. Within the pgm locus, just upstream of a gene (ybtA) that regulates expression of the Ybt receptor and biosynthetic genes, is an operon consisting of 4 genes —ybtP, ybtQ, ybtX and ybtS. Transcription of the ybtPQXS operon is repressed by Fur and activated by YbtA. The product of ybtX is predicted to be an exceedingly hydrophobic cytoplasmic membrane protein that does not appear to contribute any vital function to Ybt biosynthesis or utilization in vitro. ybtP and ybtQ encode putative members of the traffic ATPase/ABC transporter family. YbtP and YbtQ are structurally unique among the subfamily of ABC transporters associated with iron transport, in that they both contain an amino-terminal membrane-spanning domain and a carboxy-terminal ATPase. Cells with mutations in ybtP or ybtQ still produced Ybt but were impaired in their ability to grow at 37°C under iron-deficient conditions, indicating that YbtP and YbtQ are needed for iron uptake. In addition, a ybtP mutant showed reduced iron accumulation and was avirulent in mice by a subcutaneous route of infection that mimics flea transmission of bubonic plague.


Yersinia pestis, the causative agent of bubonic plague, can use a variety of compounds as iron sources (Sikkema and Brubaker, 1989; Staggs and Perry, 1991). In addition to a haem uptake system that allows the bacterium to use a number of haemoproteins as sources of iron (Hornung et al., 1996), Y. pestis possesses an ABC transport system for inorganic iron (Bearden et al., 1998) and produces at least one siderophore termed yersiniabactin (Ybt) (Wake et al., 1975; Fetherston et al., 1995), which is used to obtain iron during growth at 37°C. Additional iron transport system(s) appear to play a role in the growth of Y. pestis at lower temperatures, as ybt mutants grow at 30°C under stringent iron starvation conditions (Sikkema and Brubaker, 1987; Lucier et al., 1996; Bearden et al., 1998).

Ybt was first purified from Yersinia enterocolitica and subsequently isolated from Y. pestis. Structurally, Ybt is a phenolate-thiazole siderophore that is related to pyochelin and anguibactin, iron-binding compounds produced by Pseudomonas aeruginosa and Vibrio anguillarum respectively (Cox et al., 1981; Jalal et al., 1989; Haag et al., 1993; Drechsel et al., 1995; Chambers et al., 1996; Gehring et al., 1998a; R. Perry et al., submitted). In Y. pestis, Ybt probably plays an important role in establishing an infection at the site of a flea bite, as mutants that are unable to produce or transport Ybt are avirulent in mice by peripheral routes of infection, but fully virulent when injected intravenously (Une and Brubaker, 1984; Bearden et al., 1997; S. W. Bearden and R. Perry, submitted).

Some of the genes needed for the biosynthesis and utilization of Ybt are clustered in a pathogenicity island that is part of an unstable region of the Y. pestis chromosome, the pgm locus (Fetherston et al., 1992; Lucier and Brubaker, 1992; Fetherston and Perry, 1994; Carniel et al., 1996; Buchrieser et al., 1998; Gehring et al., 1998a). Five Y. pestis gene products that are probably involved in Ybt synthesis have been identified. These include two high-molecular-weight proteins, HMWP1 and 2, a 30 kDa putative thioesterase, YbtT, a 56 kDa homologue to EntE of Escherichia coli, YbtE, and a 48 kDa protein probably involved in salicylate synthesis, YbtS. Mutations in the gene encoding HMWP2 (irp2 ), YbtS or YbtE abolish siderophore production (Bearden et al., 1997; Gehring et al., 1998a). YbtE adenylates salicylate and transfers this activated compound to HMWP2 (Gehring et al., 1998b). HMWP2 is homologous to a family of proteins involved in the non-ribosomal synthesis of peptides, whereas HMWP1 has domains in common with non-ribosomal peptide synthetases and polyketide synthases (Guilvout et al., 1993; Gehring et al., 1998a,b). Thus, Ybt is probably assembled from activated components bound to HMWP1 and 2, using a similar mechanism to that used in the biosynthesis of certain siderophores, bioactive peptides and antibiotics (Kleinkauf and Von Döhren, 1996; Gehring et al., 1998a,b).

Uptake of Ybt is mediated by initial contact with an outer membrane receptor, Psn, in Y. pestis (FyuA in Y. enterocolitica), which also serves as the receptor for the bacteriocin, pesticin (Kutyrev et al., 1992; Rakin et al., 1994; Fetherston et al., 1995). As in a number of other systems, transport of Ybt across the outer membrane is TonB dependent (Heesemann et al., 1993). Expression of the receptor and biosynthetic genes are negatively regulated by iron, through the actions of Fur (Carniel et al., 1992; Staggs et al., 1994). In the presence of iron, Fur binds to specific sites in the promoter region of iron-regulated genes and represses their transcription (Bagg and Neilands, 1987; de Lorenzo et al., 1988). Maximum expression of psn and the Ybt biosynthetic genes also requires YbtA, an araC homologue, that probably uses Ybt as an inducer (Fetherston et al., 1996). Once across the outer membrane, the fate of Ybt is unknown. In other iron transport systems, the iron moiety or iron–siderophore complex is bound by a periplasmic-binding protein and presumably shuttled to an inner membrane permease. Passage through the inner membrane occurs through an ABC transporter consisting of the inner membrane permease associated with an ATP-binding protein (Ames et al., 1992; Higgins, 1992; Braun, 1997; Mietzner et al., 1997; Linton and Higgins, 1998). Here, we identify two genes, ybtP and ybtQ, that are probably involved in Ybt-mediated iron transport through the inner membrane.


Analysis of ybtP, ybtQ, ybtX and ybtS sequences

We have shown previously that transcription of the Ybt biosynthetic genes and receptor are controlled by an AraC-type activator termed YbtA (Fetherston et al., 1996). Nucleotide sequencing of the region upstream of ybtA identified another open reading frame (ORF) starting 166 bp upstream of the start codon for ybtA and transcribed in the opposite direction (Fig. 1). Sequence analysis revealed what appeared to be an operon consisting of four genes designated ybtP, ybtQ, ybtX and ybtS, encompassing 6363 bp (Fig. 1). YbtP and Q each contain 600 amino acid residues and show the strongest homology to members of the ABC transporter family. ybtP is predicted to encode a protein of 66 289 Da with a pI of 10.12, whereas ybtQ presumably encodes a 66 408 Da protein with a pI of 6.75. Both proteins contain an amino-terminal hydrophobic region with six or seven possible transmembrane helices in YbtP and five to seven predicted transmembrane segments in YbtQ. An ABC transporter signature motif is located in the carboxy-terminal portion of both proteins (Ames et al., 1992; Higgins, 1992; Linton and Higgins, 1998) (Fig. 2). The 5′ end of ybtQ overlaps the 3′ end of ybtP by 14 bases.

Figure 1.

. Genetic organization of the ybtPQXS and ybtA operons. Relevant restriction enzyme sites used to construct mutations are indicated. Removal of the NruI fragment resulted in an in frame deletion in ybtP (ΔybtP2064 ). A KasI digest generated an in frame fusion between ybtQ and ybtX (ΔybtQX2066 ). Digestion with NaeI created an in frame deletion in ybtX (ΔybtX2067 ). An ampicillin gene cassette was inserted into the SmaI site in ybtP to generate a second ybtP mutation (ΔybtP2065 ). Numbers between ybtP and ybtA indicate the 166 bp intergenic region. All other numbers correspond to the location of stop codons.

Figure 2.

. Amino acid sequence comparisons between CydD–YbtP (A) and YbtP–YbtQ (B). Identical, highly conserved and semi-conserved amino acids are identified by ‘*’, ‘:’ and ‘.’ respectively. The bracketed and bold sequences indicate the ABC transporter domains and numbers denote amino acid number in each protein. The four boxed regions are conserved motifs (Walker A, ABC signature, Walker B and an unnamed fourth motif), as defined by Linton and Higgins (1998), within the ABC transporter domain.

A database search using the amino-terminal portion of YbtP revealed around 46% similarity between YbtP and a hypothetical ABC transporter (RV1348; GenBank accession no. Z75555) from Mycobacterium tuberculosis as well as CydD from E. coli. An alignment between YbtP and CydD is shown in 2Fig. 2A. CydD is required for the synthesis of cytochromes in E. coli and is thought to function as a heterodimer with CydC in the export of an unknown substrate (Poole et al., 1993; Cook et al., 1997). A conserved seven amino acid sequence (VLTFVLR) present in both YbtP and CydD does not appear to be a common motif. The amino-terminal half of YbtQ shows the highest similarity to YbtP (Fig. 2B) and is 46.6% similar to another hypothetical ABC transporter (RV1349) from M. tuberculosis that is in the same operon as the M. tuberculosis RV1348 gene that has similarities to YbtP. YbtQ also has limited homology to a multidrug resistance-like ATP-binding protein (Mdl) in E. coli. The highest degree of similarity among YbtP, YbtQ and CydD is within the C-terminal ABC domain (Fig. 2). Similarities to the amino-terminal portions of YbtP and YbtQ lie both within and between potential transmembrane domains.

ybtX probably encodes a very hydrophobic protein of 44 781 Da with a pI of 10.32 and anywhere from 9 to 11 transmembrane segments. There are two possible initiating methionine codons for YbtX, separated by 105 bp. Based on the proximity to the stop codon for ybtQ, the molecular mass and pI were calculated assuming that the downstream ATG is the probable start of the YbtX coding region. Database searches revealed only limited similarity between YbtX and AmpG from E. coli, which included regions inside and outside of the membrane-spanning domains.

The putative product (47 990 Da, pI of 5.54) of the last gene in the operon, ybtS, is similar to anthranilate synthases and isochorismate synthases. The function of ybtS in Ybt biosynthesis is presented in detail elsewhere (Gehring et al., 1998a) and will not be discussed further in this paper.

Expression from the ybtPQXS promoter

The promoter region for the ybtPQXS operon is shown in Fig. 3. Two repeat sequences that may correspond to YbtA binding sites are located upstream of a putative −35 element. A potential Fur binding site (FBS), which matches 14 bp of a 19 bp E. coli consensus FBS (Braun and Hanke, 1991) is located immediately upstream of a probable −10 region.

Figure 3.

. Nucleotide sequence of the region between ybtA and ybtP. Putative YbtA binding sites are underlined and labelled (YbtA BS). Bold letters denote potential Fur binding sites. Sequences resembling E. coli consensus ribosome binding sites (RBS), −35 and −10 regions are indicated.

Polymerase chain reaction (PCR) was used to isolate the 166 bp fragment corresponding to the putative promoter region for the ybtPQXS operon (Fig. 3). This fragment was cloned in front of the lacZ gene in the single-copy vector pEU730 (Froehlich et al., 1994), and the resulting construct (pEUYbtP) was electroporated into various Y. pestis strains. The β-galactosidase activity of these strains grown in a deferrated, defined medium, PMH, in the presence or absence of iron is presented in Table 1. In KIM6+, which contains all of the genes needed for Ybt synthesis and utilization, expression of lacZ from the ybtPQXS promoter is repressed 55-fold when the cells are grown in the presence of 10 μM iron. This repression does not occur in KIM6-2030+, a Y. pestis strain containing a kanamycin insert in the fur gene (Table 1), indicating that the operon is Fur regulated. We have shown previously that the Ybt receptor (Psn in Y. pestis) as well as the biosynthetic genes are regulated by an AraC-like protein called YbtA (Fetherston et al., 1996). Likewise, YbtA appears to regulate expression of the ybtPQXS operon, as evidenced by the > 95-fold reduction in the activity of the ybtP::lacZ reporter construct in KIM6-2055 (ybtA::kan2055 ), as well as in KIM6 cells grown under iron-deficient conditions (Table 1). KIM6 is missing the entire 102 kb pgm locus, including the receptor, biosynthetic genes and ybtA. Ybt siderophore also appears to be required for maximum expression of the ybtPQXS operon, as the activity of ybtP::lacZ in a Ybt biosynthetic mutant (KIM6-2046.1) grown in the absence of iron is 22-fold lower than in KIM6+. The activity of pEUYbtP in KIM6-2067 (ΔybtX ) was essentially the same as in KIM6+ (data not shown), suggesting that ybtX does not regulate expression of the ybtP operon.

Table 1. . β-Galactosidase activity of Y. pestis strains grown to mid-log phase at 37°C in PMH. a. Enzyme activity is expressed in Miller units (Miller, 1992).Thumbnail image of

Effects of ybtP, ybtQ and ybtX mutations

To determine if ybtP, ybtQ and ybtX play a role in iron transport, we generated several Y. pestis strains with mutations in these genes. Two strains contain mutations in ybtP : one an in frame deletion (KIM6-2064) and a second strain, KIM6-2065, with an ampicillin gene cassette inserted near the end of the ybtP coding region (Fig. 1). KIM6-2066 contains a deletion that removes most of ybtQ as well as the 5′ end of ybtX and fuses the first 16 amino acids of YbtQ to the last 356 amino acids of YbtX. An in frame deletion removes a major portion of ybtX in KIM6-2067 (Fig. 1).

The effect of these mutations on the growth of KIM6-2065, KIM6-2066 and KIM6-2067 in iron-deficient PMH at 37°C is shown in Fig. 4. Similar results were obtained for both ybtP mutant strains; only the data for KIM6-2065 are shown. KIM6-2065 (ybtP::amp) and KIM6-2066 (ΔybtQX ) exhibited growth defects when cultured at 37°C under iron-deficient conditions (Fig. 4). However, the ybtX mutation did not adversely affect the iron-deficient growth of Y. pestis KIM6-2067 (Fig. 4B), suggesting that ybtX is not required for growth under iron-deficient conditions at 37°C. As the ybtP mutants as well as the ybtQX mutant were impaired in their ability to grow in iron-deficient PMH at 37°C, then both ybtP and ybtQ are probably essential components of the Ybt iron transport system. A plasmid carrying ybtPQ and 37 bp of ybtX was able to complement the growth defect in KIM6-2066 (Fig. 4B). Curiously, the ybtP::amp mutation could be complemented by a plasmid encoding intact ybtP and 440 bp of ybtQ, suggesting that either the amp insertion into ybtP is not a polar mutation or a cryptic promoter drives expression of ybtQ in KIM6-2065 (Fig. 4A).

Figure 4.

. Iron-deficient growth of Y. pestis strains. All strains were grown in iron-deficient PMH at 37°C. Culture optical density at 620 nm was monitored as an indication of growth. A. KIM6+ (Ybt+; •); KIM6-2065 (ybtP::amp2065 ; ▵); KIM6-2065(pYbtP) (ybtP::amp2065/ybtP+; ▴). B. KIM6+ (•); KIM6-2066 (ΔybtQX2066 ; ▵), KIM6-2066(pYbtPQ) (ybtQ + ΔybtX ; ▴), KIM6-2067 (in frame ΔybtX2067 ; □).

For reasons that are unclear, the ybtX mutant consistently grew slightly better than its Ybt+ parent under iron-deficient conditions (Fig. 4B). Reporter gene studies in the mutant did not show overexpression from the ybtP promoter, and titring of Ybt supernatant levels did not indicate higher levels of siderophore expression by the mutant compared with the parental strain (data not shown). Thus, overexpression of the Ybt system does not appear to be the explanation for this modest growth effect.

YbtP and YbtQ resemble the subfamily of ABC transporters involved in product export, in that they have amino-terminal membrane-spanning domains and a carboxy-terminal ATPase (Fath and Kolter, 1993). Furthermore, neither protein has any of the motifs found on the permeases associated with other iron transport systems. Therefore, it was conceivable that YbtP and YbtQ were involved in the export of Ybt. Mutants in siderophore exporters might be expected to exhibit growth defects under iron-deficient conditions as well as defects in siderophore secretion and iron transport. Therefore, we used a biological assay to test for siderophore production by KIM6-2065 and KIM6-2066. Y. pestis strains with mutations in Ybt biosynthetic genes are unable to grow on iron-restricted PMH plates (PMH-S) at 37°C unless supplied with an exogenous source of the siderophore. Both KIM6-2065 and KIM6-2066 were able to support the growth of a Ybt biosynthetic mutant, KIM6-2046.1, on PMH-S plates. However, only KIM6-2067 (ΔybtX ), but not KIM6-2065 or KIM6-2066, could be cross-fed by KIM6+, a strain that produces Ybt. These results indicate that the ybtP::amp and ΔybtQX strains produce and secrete Ybt but are unable to internalize it.

The effect of the ybtP::amp mutation on iron transport was analysed by measuring the uptake of 55Fe. KIM6+ and KIM6-2065 were grown for about eight generations in iron-deficient PMH. Cultures in mid-log phase were incubated with or without the protonophore, carbonyl cyanide m-chlorophenyl-hydrazone (CCCP), in PMH-containing 55FeCl3. As seen in Fig. 5, cells bearing a mutation in ybtP were completely impaired in their ability to transport iron actively, performing about the same as cultures treated with CCCP.

Figure 5.

. Uptake of 55Fe by Y. pestis KIM strains. Transport of 55Fe by KIM6+ (♦, ▪) and KIM6-2065(ybtP::amp2065 ) (•, ▴) cells was assessed at 37°C in iron-deficient PMH in the presence (▪, ▴) or absence (♦, •) of 100 μM CCCP. All strains were pregrown at 37°C in PMH under iron-deficient conditions.

To determine the importance of ybtP to the pathogenesis of plague, we determined the LD50 value of a slightly attenuated strain of Y. pestis, KIM5-2053.11+, carrying the ybtP::amp mutation. Groups of five NIH/Swiss Webster mice were injected subcutaneously with bacterial doses ranging from approximately 3 to > 104 colony-forming units. The LD50 value for this ybtP mutant strain is greater than the highest bacterial dose used, 7.6 × 104. In contrast, the LD50 of the parent strain was 115 bacterial cells.


In this study, we have identified four new genes, ybtP, ybtQ, ybtX and ybtS, that are part of the region in the Y. pestis chromosome dedicated to the synthesis and utilization of the siderophore, yersiniabactin. Expression of the ybtPQXS operon is negatively regulated by iron, through the action of Fur, and requires YbtA. Previous work has shown that YbtA may function in concert with Ybt to activate transcription of the Ybt receptor (Psn in Y. pestis) as well as the Ybt biosynthetic genes, while repressing its own transcription (Fetherston et al., 1996) Likewise, maximal expression of the ybtPQXS operon requires both Ybt and YbtA.

Both ybtP and ybtQ are needed to sustain the growth of Y. pestis at 37°C under iron-deficient conditions. However, ybtX mutants did not exhibit any iron-deficient growth defects at 37°C. Thus, ybtX appears to be dispensable for growth under iron-restrictive conditions, at least in vitro. Cross-feeding studies indicated that strains with mutations in ybtP, Q or X continue to secrete Ybt, but only the ybtX mutant strain was able to use exogenous Ybt. Hence, ybtP and ybtQ appear to be required for the uptake of iron from Ybt. The limited growth of YbtP and YbtQ mutants under iron-deficient conditions (Fig. 4) is probably caused by the presence of other iron transport systems in Y. pestis (Perry and Fetherston, 1997; Bearden et al., 1998).

KIM6-2065 cells were clearly defective in iron transport (Fig. 5), even though these cells presumably still make YbtQ. This suggests that both YbtP and YbtQ are needed for iron uptake. The level of iron transport in the ybtP::amp strain was essentially the same as that achieved by CCCP-poisoned cells, despite the presence of other energy-dependent iron transport systems that are probably still functional (Bearden et al., 1998). Presumably, Ybt siderophore, which continues to be produced by KIM6-2065, binds the excess iron in the media making it unavailable for transport by these alternative systems.

The role of YbtX, if any, in Ybt synthesis and utilization is unclear. YbtX is predicted to be a very hydrophobic protein with limited homology to AmpG, a putative permease involved in recycling cell wall murein components (Lindquist et al., 1993; Jacobs et al., 1994). Mutations in ybtX had no effect on the iron-deficient growth of the respective strains. In this regard, YbtX is reminiscent of P43, a hydrophobic protein encoded by sequences within the enterobactin gene cluster (Chenault and Earhart, 1991; Shea and McIntosh, 1991). Inserts in the P43 coding region likewise had no effect on enterobactin production or the ability of strains to grow under iron-deficient conditions. Whether proteins such as YbtX and P43 have a role during in vivo growth that is masked or unnecessary in vitro remains to be elucidated.

In all iron transport systems studied so far in Gram-negative bacteria, transport across the cytoplasmic membrane requires a periplasmic-binding protein as well as an ABC transporter. The entire region involved in Ybt synthesis and utilization has been sequenced in Y. pestis, and no obvious candidate for the periplasmic-binding protein has been detected (Gehring et al., 1998a). There are only three proteins with unknown functions within the Ybt region, YbtU, YbtT and YbtX. We have shown that YbtX is not required for iron transport in vitro and, thus, is unlikely to function as the periplasmic-binding protein for the Ybt system. ybtU is located downstream of irp1 in the irp2 operon and is predicted to encode an approximately 41 kDa protein, but lacks any apparent signal sequence. ybtT is located downstream of ybtU and contains a thioesterase-like domain (Gehring et al., 1998a). In addition, none of these proteins contains motifs found on other periplasmic-binding proteins involved in iron transport (Tam and Saier, 1993; Braun et al., 1998). Either the Ybt system does not use a classic periplasmic-binding protein, or it is encoded elsewhere in the genome.

YbtP and YbtQ are members of the ABC transporter family and probably function as a heterodimer in the transport of ferri-Ybt. Both YbtP and YbtQ contain the permease and ATP-binding domain fused into a single protein, which makes them structurally unique among the group of ABC transporters associated with iron transport. All other iron transport systems identified to date have the permease and ATPase contained in separate proteins. Given the fused nature of YbtP and YbtQ, it is not surprising that they are missing the EAA region thought to be involved in mediating interactions between the permease and ATP-binding protein of other ABC transporters (Dassa and Hofnung, 1985; Mourez et al., 1997). However, it is interesting that YbtP and YbtQ also lack motifs present on the ABC transporter subfamily of iron transport permeases, including those that transport free iron as well as iron–siderophore complexes (Chenault and Earhart, 1991; Shea and McIntosh, 1991; Saurin and Dassa, 1994; Saurin et al., 1994; Braun, 1997; Mietzner et al., 1997). Thus, YbtP and YbtQ may represent a new group of iron transport permeases.

Experimental procedures

Bacterial strains and media

All the Y. pestis strains used in this study were derived from KIM6+ (Sikkema and Brubaker, 1987; Fetherston et al., 1992). KIM6+ possesses all of the known Y. pestis virulence determinants except for pCD1, a 70.5 kb plasmid encoding the low-calcium response stimulon (Straley and Bowmer, 1986). KIM6 has spontaneously deleted the 102 kb pgm locus, which contains the genes for haemin storage as well as Ybt biosynthesis and utilization (Fetherston et al., 1992; Fetherston and Perry, 1994). KIM6-2030+ contains a kanamycin resistance gene (kan) inserted into fur (Staggs et al., 1994), whereas KIM6-2046.1 has a kan insertion in irp2, the first gene in the Ybt biosynthetic operon (Fetherston et al., 1995; Gehring et al., 1998a). The construction of KIM5-2053.1+, a Y. pestis strain carrying a deletion of some of the genes necessary for the expression of the pH6 antigen, has been described by Bearden et al. (1997). For virulence studies in mice, a derivative of pCD1 with a Mu dI1 insert in yopJ (pCD1::Mu dI1–73; Straley and Bowmer, 1986) was introduced into KIM5-2053.1+ by electroporation generating KIM5-2053.11+ (Bearden et al., 1997). Plasmids and suicide vectors were maintained in E. coli DH5α and DH5α(λpir ) respectively.

Stock cultures of Y. pestis were streaked onto Congo red (CR) plates (Surgalla and Beesley, 1969) to verify their pigmentation phenotype before inoculation of tryptose blood agar (TBA) slants or heart infusion broth (HIB). Bacteria were grown in the defined medium, PMH, rendered iron deficient by treatment with Chelex 100 (Staggs and Perry, 1991). PMH-S plates were derived from deferrated PMH by the addition of 0.5 mM NaCO3, 0.01 mM MnCl2, 4 mM CaCl2 and 1% agarose. E. coli strains were grown in either Luria broth (LB) or Terrific broth (Tartof and Hobbs, 1987). Where appropriate, antibiotics were added to the following concentrations: kanamycin and streptomycin at 50 μg ml−1; spectinomycin and ampicillin at 100 μg ml−1.

Construction of Y. pestis mutants

All Y. pestis mutant strains were generated by allelic exchange using mutated DNA fragments cloned into suicide vectors carrying the sacB gene and an R6K origin of replication. Two strains with mutations in ybtP were created, KIM6-2064 and KIM6-2065. Construction of KIM6-2064 began with an NruI digest of pSDR498.13, which removes a 465 bp fragment in ybtP (see Fig. 1) and generated pYBTP1. A 2.7 kb Sal I–EcoRV fragment containing the deletion was transferred from pYBTP1 to the suicide vector pSUC1 to yield pYBTP1.1. The deletion was then introduced into KIM6+ by allelic exchange. Briefly, pYBTP1.1 was electroporated into Y. pestis, and colonies containing the integrated plasmid were selected on TBA plates containing ampicillin. Individual Apr colonies were grown overnight in HIB in the absence of antibiotics, and aliquots were plated on CR agar containing 5% sucrose. Sucrose-resistant (Sucr) colonies lacking the 465 bp NruI fragment were identified by colony blot hybridization. Southern blot hybridization was used to confirm the allelic exchange event. To produce KIM6-2065, an ampicillin gene cassette was ligated into the unique SmaI site of ybtP (see Fig. 1) present in a 3.1 kb XhoI–EcoRV fragment cloned from pSDR498.12 into the suicide vector pKNG101. After electroporation of this construct (pKNG498Ap) into KIM6+, the cells were allowed to recover for 1 h and then incubated overnight at 37°C with ampicillin. Genomic DNA from Apr, Sucr colonies was analysed by Southern blot hybridization to identify strains containing an ampicillin gene insert in ybtP.

The 1.958 kb deletion in KIM6-2066 that removes most of ybtQ and part of ybtX was made by digesting pSDR498.17 with KasI to yield pSDR498.17D. The AscI fragment from pSDR498.17D that contains the deletion was used to replace the wild-type AscI fragment in pSDR498.12, generating pSDR498.12D. Cloning a 1.8 kb NruI–Sal I fragment from pSDR498.12D into pSUC1 produced pCVDYbtQ, the suicide vector used for allelic exchange. This deletion should create a fusion protein containing the first 16 amino acids of YbtQ and the carboxy-terminal 356 amino acids of YbtX.

KIM6-2067 contains a 789 bp in frame deletion in ybtX produced from an NaeI digest of pSDR498.18 to form pSDR498.18D. A 1.6 kb EcoRV–SphI fragment was cloned from pSDR498.18D into pSUC1 to create the suicide vector pCVDYbtX. To generate KIM6-2066 and KIM6-2067, the respective suicide vectors were electroporated into KIM6+, and co-integrants were selected on TBA plates containing 50 μg ml−1 carbenicillin. The merodiploid strains were grown overnight in HIB, and aliquots of the overnight cultures were spread on CR plates containing 5% sucrose. Sucr colonies were screened by colony blot hybridization to identify those lacking the appropriate DNA fragments. For KIM6-2066 and KIM6-2067, Southern blot hybridization and PCR analysis, respectively, were used to confirm that the wild-type DNA had been replaced by the mutated fragment.

Plasmids, sequencing and recombinant DNA techniques

All the plasmids used in this study are listed in Table 2. The suicide vector, pSUC1, was modified from pCVD442 and lacks the 1.9 kb mob region. In addition, the R6K origin of replication was moved to an NdeI site downstream of the sacB gene, and a unique BamHI restriction site was engineered into the multicloning site.

Plasmids were purified by alkaline lysis (Birnboim and Doly, 1979) and further purified by polyethylene glycol precipitation (Humphreys et al., 1975). A standard CaCl2 procedure was used to introduce plasmids into E. coli (Sambrook et al., 1989). Y. pestis cells were transformed by electroporation, as described previously (Fetherston et al., 1995). Plasmid DNA was sequenced by the dideoxynucleotide chain termination method (Sanger et al., 1977) using Sequenase version 2.0 (Amersham Pharmacia Biotech), [35S]-dATP (New England Nuclear/Dupont) and 7-deaza-dGTP. Samples were electrophoresed at 70 W on 6% polyacrylamide gels containing Tris borate–EDTA buffer and 8.3 M urea. Fixed, dried gels were exposed to Kodak BioMax MR film at room temperature. Synthetic oligonucleotide primers purchased from Integrated DNA Technologies were used to extend the sequence. The intelligenetics suite series of programs was used to assemble the sequence. Homology searches were performed using blast (Altschul et al., 1990; 1997; Gish and States, 1993). Potential transmembrane domains were detected using das (Cserzo et al., 1997) and Tmpred. Alignments were performed using clustalw (Thompson et al., 1994).

Nucleotide sequence accession number

The ybtPQXS genes were sequenced in our laboratory and independently in a collaborative project with Dr Fred Blattner's research group. Both sequences were identical. The ybtPQXS sequence has been deposited in GenBank (accession number AF091251). AF091251 also contains the sequence of the entire Ybt region and pathogenicity island (Gehring et al., 1998a).

Construction of a ybtP reporter plasmid and β-galactosidase assays

The 166 bp region between the start codons for ybtA and ybtP was amplified from pSDR498.12 by PCR using primers ybtP1 (5′-GGCGCGCCGACCTGGTTATCTCCCTG-3′) and ybtP2 (5′-GGGGTACCGGGAGTAACTGAATTTCC-3′). Reactions containing 0.2 mM dNTPs and 0.2 μM primers consisted of 20 s at 94°C, 30 s at 52°C and 30 s at 72°C for 25 cycles followed by a single cycle at 72°C for 7 min. The products were extracted with phenol–chloroform (1:1) and ethanol precipitated. PCR fragments digested with Asp718 and AscI were ligated into the Asp718/AscI sites of the single-copy lacZ reporter plasmid pEU730 (Froehlich et al., 1994). A clone, pEUYbtP, containing the ybtPQXS promoter driving lacZ expression was identified by sequencing and electroporated into various Y. pestis strains. Lysates were prepared from bacterial strains grown in PMH in the presence or absence of iron through two transfers for a total of approximately six generations. The β-galactosidase activity, using ONPG as a substrate, was measured spectrophotometrically and expressed in Miller units (Miller, 1992).

Iron transport studies

Bacterial cells were grown in deferrated PMH through two transfers for a total of about six generations. When the culture reached an OD620 of ≈0.3 during the second transfer, an aliquot was removed for transport studies. A portion of the culture was incubated with 100 μM CCCP for 10 min at 37°C before the addition of 0.2 μCi ml−155FeCl3. Samples (0.5 ml) were collected at various times after the addition of radioactive iron and filtered through a 0.45 μm filter presoaked in PMH containing 20 μM FeCl3. The filters were washed twice with PMH and counted in scintillation fluid in a Beckman LS3801 scintillation counter using the setting for 35S. Uninoculated controls showed no retention of the isotope by the filters. Samples (25 μl) were counted directly to determine the total number of counts present in the transport assay. The data are expressed in terms of percentage transport of radioactive iron for a culture normalized to an OD620 of 0.4.

Feeding studies

Bacterial cultures grown in deferrated PMH for about six generations at 37°C were streaked across from each other on PMH-S plates and incubated at 37°C. Y. pestis strains that do not produce Ybt are unable to grow on PMH-S plates at 37°C, but they can be cross-fed by Ybt-producing strains.

Virulence testing

To generate a strain for virulence testing in mice, the ybtP::amp mutation was first crossed into KIM5-2053.1+ by allelic exchange followed by introduction of pCD1::Mu dI1–73 to yield KIM5-2065.1. All work was performed in a BL3 facility. KIM5-2065.1 and KIM5-2053.11+ were grown at 26°C in deferrated PMH containing 50 μM haemin through two transfers for a total of six or seven generations. Cells were harvested at an OD620 of ≈0.4, pelleted and resuspended in mouse isotonic PBS (149 mM NaCl, 16 mM Na2HPO4, 4 mM NaH2PO4). Five- to seven-week-old female NIH/Swiss Webster mice were injected subcutaneously with 0.1 ml of 10-fold serial dilutions of the bacterial suspension. Five mice were used for each bacterial dose. The number of cfus inoculated was determined by plating serial dilutions of each dose in duplicate on TBA–kanamycin plates. The mice were examined daily for a period of 3 weeks. LD50s were calculated according to the method of Reed and Muench (1938).


This research was supported by Public Health Services Grant AI33481 from the National Institutes of Health. We would like to thank Scott Bearden and Jessica Shah for their help with some experiments.