INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Nearly all living organisms require trace amounts of iron. For pathogens, the iron- and heme-chelating proteins of mammalian hosts are barriers to iron acquisition that must be overcome. A number of iron and hemoprotein transport systems from a variety of pathogens have been characterized and have demonstrated roles in the infectious process (9, 11, 22, 23, 51, 53).

Yersinia pestis, the etiologic agent of bubonic and pneumonic plague, has three partially characterized iron transport systems. The hemoprotein uptake (Hmu) system of Y. pestis allows the bacterium to use hemin, hemoglobin, haptoglobin-hemoglobin, myoglobin, heme-hemopexin, and heme-albumin as iron sources (25, 50). A siderophore-dependent system (Ybt) synthesizes yersiniabactin, ABC transport components, and a regulatory protein that are all encoded within a high-pathogenicity island that is present in highly pathogenic strains of Y. pestis, Yersinia pseudotuberculosis, Yersinia enterocolitica, and several types of Escherichia coli pathogens. In Y. pestis, the high-pathogenicity island lies within the pgm locus, a 102-kb chromosomal region subject to high-frequency deletion (10, 18, 20, 24, 28). The Ybt system is essential for iron acquisition during the early stages of plague (4, 5, 16). The YfeABCD system of Y. pestis belongs to an ABC family of bacterial cation transporters and transports both iron and manganese. It plays a role in iron acquisition during the later stages of plague (5, 6). Studies with iron chelators suggest that Y. pestis possesses an iron transport system that functions at 26 to 30°C but not at 37°C. This putative 26°C iron transport system is independent of the Ybt and Yfe transport systems (5, 29).

In this study, we describe the identification, cloning, and initial characterization of a Y. pestis ABC transporter called YfuABC. It has high homology to iron transporters in Y. enterocolitica (YfuABC) (40) and Serratia marcescens (SfuABC) (2). The Y. pestis system is iron and Fur regulated and enhanced growth of a siderophore-deficient strain of E. coli under iron-chelated conditions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and media. The bacterial strains and plasmids used in this study are listed in Table 1. Y. pestis strains missing the low-calcium-response (Lcr) virulence plasmid pCD1 are completely avirulent (34) and were used in all physiological and genetic experiments. All strains were stored in phosphate-buffered glycerol at 20°C. Y. pestis strains were grown in heart infusion broth (HIB), in Luria broth (LB), or on tryptose blood agar base (TBA). The pigmentation (Pgm) phenotype of strains was confirmed on Congo red plates (48). For growth under iron-restricted conditions, a colony from a Congo red plate was inoculated onto TBA slants and incubated for 24 to 48 h at 26 or 37°C. Cells on the TBA slant were suspended in PMH2 deferrated by Chelex 100 extraction (46), diluted to an optical density at 620 nm (OD₆₂₀) of 0.1 in deferrated PMH2 broth, and grown aerobically for ~8 h before inoculation of a second transfer containing fresh deferrated PMH2. PMH2 is a modified, defined medium derived from PMH (46); PMH2 contains 50 µM PIPES (piperazine-N,N'-bis[2-ethansulfonic acid]) in place of 50 µM HEPES. This change in buffers reduces acidification of the medium due to bacterial growth (data not shown). For iron transport assays only, bacterial cells were grown and assayed in PMH. Growth of the cultures was monitored by determining the OD₆₂₀ with a Genesys5 spectrophotometer (Spectronic Instruments, Inc.) at regular intervals. For some studies, PMH2 was supplemented with either 2,2'-dipyridyl (DIP) or ethylenediamine-di(o-hydroxyphenyl acetic acid) (EDDA) to chelate residual iron in the medium. Contaminating iron in EDDA was extracted as previously described (38). E. coli strains were grown in either LB, nutrient broth with 85.6 mM NaCl (NB), or Tris-glucose-thymidine medium without FeCl₃ (TG-Fe) (43). All glassware used in these studies was cleaned in a chromic-sulfuric acid solution (46.3 g of potassium dichromate per liter of ~11.5 M sulfuric acid) or Scotclean (Owl Scientific, Inc.) to remove contaminating iron and then rinsed copiously with deionized water. All reagents and media were made with deionized water after passage through a Nanopure cartridge system (Barnstead). When appropriate, media included antibiotics at the following concentrations: ampicillin, 100 µg/ml; tetracycline, 12.5 µg/ml; kanamycin, 50 µg/ml; and chloramphenicol, 30 µg/ml.

Recombinant DNA techniques and plasmids. Plasmids were isolated by alkaline lysis (7, 26) and transformed into various E. coli strains by standard calcium chloride transformation (3) or electroporated into Y. pestis (17). Bacterial genomic DNA was isolated by a method utilizing lysozyme-sodium dodecyl sulfate (SDS)-proteinase K (3) and further purified by phenol and chloroform extractions (3). DNA restriction endonucleases, T4 DNA ligase, and calf intestinal alkaline phosphatase were used according to the manufacturer's specifications.

-Galactosidase assays. Y. pestis KIM6+, KIM6, and KIM6-2030 cells containing pYFU4 (yfu::lacZ) were harvested during exponential growth from second-transfer cultures in PMH2 broth containing either no added iron source or 10 µM FeCl₃. -Galactosidase activities from whole-cell lysates of these cultures were measured as previously described (30). Since Y. pestis is naturally -galactosidase negative in this assay, the activity obtained from strains carrying reporter plasmids correlates directly with promoter activity of the lacZ fusion reporter (21, 46).

Construction of Y. pestis mutants. A deletion encompassing an upstream open reading frame (ORF), the yfuABC promoter, and most of the yfuA gene was made by deleting 1,808- and 198-bp BamHI fragments (Fig. 1) from pYFU2 to generate pYFU3. The mutated region was then cloned into the suicide vector pKNG101 (27). The resulting recombinant plasmid, pYFU3.1 (Table 1), was introduced separately into Y. pestis KIM6+ (Ybt⁺ Yfe⁺), KIM6 (Ybt Yfe⁺), KIM6-2031.1+ (Ybt⁺ Yfe), and KIM6-2031.1 (Ybt Yfe) by allelic exchange. Y. pestis merodiploid strains were selected on TBA plates containing 50 µg of streptomycin/ml. Subsequent screening of these strains for exchange of the mutant alleles for wild-type alleles was accomplished by selection for sucrose resistance as described previously (4). To confirm that the deletion mutation had been exchanged, the yfu region of each strain was amplified by PCR using primer YFUP1 (5'-ACTGCCATACTGCCATCG-3') and YFUP2 (5'-ACTCAGTGCAGCCTGTGC-3'). Reactions containing 250 µM deoxynucleoside triphosphates, 1.5 mM MgCl₂, and 1 µM primers were performed for 10 min at 94°C, 45 s at 94°C, 30 s at 50°C, and 30 s at 72°C for 25 cycles followed by a single cycle at 72°C for 5 min. These primers used should amplify a 2,486-bp product in the yfu⁺ strains and a 480-bp product in the yfu strains. Products of the expected sizes were observed in all yfu⁺ and yfu strains; both products were observed in all merodiploid strains (data not shown).

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Genetic organization of the Y. pestis yfuABC operon and similarities to selected iron ABC transport systems. (A) Relevant genes and restriction sites of Y. pestis DNA in pYFU1. Base pair numbering is shown on the top line. A putative Fur binding sequence (FBS) is 95 bp from the start of yfuA. For YfuA, the unprocessed and processed molecular masses (MW) and pIs are given. Y. pestis DNA present in pYFU2 and pYFU3 are indicated by the lines shown. (B) Percent identity/similarity to each of four iron ABC transport systems is shown below the corresponding Y. pestis gene.

Iron transport assays. Y. pestis KIM6 (Ybt Yfe⁺ Yfu⁺), KIM6-2031.1 (Ybt Yfe Yfu⁺) and KIM6-2082.1 (Ybt Yfe Yfu) cells were acclimated to iron-deficient growth at 37°C or 26°C in PMH medium for five to six generations, and 0.1 µCi of ⁵⁵FeCl₃/ml was added to exponentially growing Y. pestis cells. Samples of 0.5 ml were withdrawn at regular intervals for 40 min, collected by vacuum filtration through 0.45-µm-pore-size GN-6 nitrocellulose membranes (Gelman Sciences), and washed twice with PMH medium. Unfiltered samples determined the total radioisotope content of cultures in each experiment. Sample filters were suspended in Bio-Safe II counting cocktail (Research Products International), and the counts per minute of each sample was measured in a Beckman LS3801 liquid scintillation spectrometer with a counting window of 0 to 1,000 keV. To demonstrate energy-dependent uptake and correct for nonspecific binding, cells were poisoned metabolically with 100 µM carbonyl cyanide m-chlorophenylhydrazone (CCCP) 10 min before the addition of isotope.

Protein analysis. In vitro transcription-translation of plasmid-encoded proteins was performed with an E. coli S30 cell extract system (Promega Corp.). Proteins were radiolabeled with ³⁵S-labeled amino acids (DuPont NEN Research Products) according to the manufacturer's recommendations, and equal amounts of trichloroacetic acid-precipitable counts were resolved by SDS-polyacrylamide gel electrophoresis (PAGE). Dried gels were exposed to Kodak BioMax MR film at room temperature. Homology searches of protein databases were performed with BLAST version 2.1 (1). Alignments were performed using CLUSTAL W (49). Molecular masses and pIs were determined using DNAMAN 4.16 (Lynnon Biosoft). Signal sequence cleavage sites were determined using Signal P (32).

Virulence testing. Previously, we have used plasmid pCDl::MudI1734-73 (yopJ::MudII1734; Km^r) to transform Lcr strains for virulence testing (4, 5, 16). Although a previous study with a YopJ mutant did not show a significant effect on virulence in mice injected intravenously (47), in Y. pseudotuberculosis YopJ is required for inducing apoptosis in macrophages (31). To construct a marked pCD1 with no mutations in expressed genes, a 9,978-bp BglII fragment from pCD1, containing a portion of the pseudogene yadA'/'yadA, was cloned into pBGL2 to generate pBGCD3. A 3,674-bp BglII-SmalI piece was isolated from pBGCD3 and inserted into BamHI and EcoRV-digested pWSK129 to yield pWSKYadA. Using a HindIII site within the polylinker of pWSK129, a 1,858-bp Scal-HindIII fragment was excised from pWSKYadA and cloned into the EcoRV-HindIII sites of pACYC184 to create pACYCYadA. An approximately 1,200-bp EcoRV fragment containing an ampicillin gene cassette (bla) was excised from pRL494e and inserted into the unique EcoRV site within the 'yadA gene of pACYCYadA. This plasmid was named pACYCYadAp. To create the suicide vector, pKNGYadAp, an approximately 3.6-kb XbaI-SalI piece from pACYCYadAp was inserted into the corresponding sites in pKNG101.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Sequence analysis. BLAST searches and analyses of the Y. pestis CO92 genome sequence database at the Sanger Centre (Yersinia pestis CO92 genomic sequence database [ftp://ftp.sanger.ac.uk/pub/pathogenic/yp/YP.dbs]) and the Y. pestis KIM10+ genome sequence database at the University of Wisconsin (UW) Genome Project (http://magpie.genome.wise.edu/ cgi-bin/Authenticate.cgi/uwgp_blast.html) with the deduced amino acid sequence of YfuA from Y. enterocolitica identified a potentially functional yfuABC operon in both plague biotypes. Y. pestis yfuABC appear to be in a single operon with a Fur binding sequence in the promoter region (Fig. 1). The Y. pestis Yfu system is a member of the cluster of bacterial ABC iron transport systems (39; http://www-biology.ucsd.edu/~msaier/transport/titlepage.html) that include Sfu of S. marcescens, Hit of Haemophilus influenzae, Yfu of Y. enterocolitica, and Afu of Actinobacillus pleuropneumoniae (2, 8, 13, 40). Thus, YfuA likely acts as the periplasmic binding protein (PBP) which passes on the substrate to a dimer of YfuB, the inner membrane (IM) permease. Translocation across the IM is probably energized via ATP hydrolysis by YfuC, the ATP-binding protein or hydrolase. The Y. pestis KIM10+ system is most homologous to that in Y. enterocolitica, showing identities of 83.7% for YfuA, 87.3% for YfuB, and 76.8% for YfuC. Percents similarities/identities to Sfu of S. marcescens are nearly as high as those to Yfu of Y. enterocolitica, while those to Hit of H. influenzae and Afu of A. pleuropneumoniae (8, 13) are lower (Fig. 1). The higher degree of similarity among the enteric organisms is shown in Fig. 2, an alignment of Y. pestis YfuA with the PBPs from Y. enterocolitica, S. marcescens, Neisseria meningitidis, and H. influenzae.

View larger version (86K): [in this window] [in a new window]   FIG. 2.   CLUSTAL W amino acid sequence alignments of the PBPs from S. marcescens (SfuASm), Y. enterocolitica (YfuAYe), Y. pestis (YfuAYp), N. meningitidis (FbpANm), and H. influenzae (HitAHi). Identical amino acids are shown in black boxes, while similar amino acids are shown in gray boxes. The consensus line (Con.) shows identical (capital letters) and similar (dots) amino acids. Locations of the yfuA primers from the Y. enterocolitica sequence used in reference 6 are shown by arrows.

Cloning and expression of Y. pestis yfuABC genes. A PCR product generated from primers within the yfuA gene (see Materials and Methods) was used as a probe to screen our Sau3AI library of Y. pestis KIM6+ genomic DNA (35). Subcloning of pYFU238 to yield pYFU2 (Table 1) and restriction site mapping indicated that the library clone was missing ~500 bp of the 3' end of yfuC. To recover a full-length operon, a 9.8-kb XhoI-PstI Y. pestis chromosomal DNA fragment encoding yfuABC was cloned into pBluescript II KS+, generating pYFU1. In vitro transcription-translation of pYFU1 and pYFU2 followed by SDS-PAGE analysis of the products identified four polypeptides that could correspond to YfuA, YfuC, and the upstream ORF (Fig. 1 and 3). The larger YfuB was not detected (Fig. 3); IM permeases are often not detected by SDS-PAGE due to their hydrophobicity. Similar analysis of pYFU3.1, which contains the deletion encompassing the upstream ORF, the yfu promoter, and yfuA (Table 1), suggests that this deletion eliminates expression of all yfu genes as expected (Fig. 3).

View larger version (67K): [in this window] [in a new window]   FIG. 3.   Autoradiogram of plasmid-encoded proteins labeled with 35S-amino acids by in vitro transcription-translation and separated by SDS-PAGE. Molecular weight markers (MW) and their corresponding masses in kilodaltons are shown. Four relevant proteins are identified by arrows.

View this table: [in this window] [in a new window]   TABLE 2.   -Galactosidase activities of Y. pestis strains containing pYFU4

Iron-deficient growth of E. coli 1017. To determine if Y. pestis yfuABC genes enhanced growth of E. coli 1017 (an enterobactin-deficient mutant) under iron-chelated conditions, we transformed this strain with pYFU5 (Yfu⁺), pYFU6 (yfuAB⁺ yfuC'), or pBR322 (the moderate-copy-number vector for both recombinant plasmids) (Table 1). Transformed 1017 cells were grown overnight in NB at 37°C and then diluted into NB containing 50 µM DIP for growth analysis (Fig. 4). Growth of 1017 and growth of 1017(pYFU6) cells in this iron-chelated medium were nearly identical to each other and significantly inhibited compared to 1017(pYFU5). For unknown reasons, growth of 1017(pBR322) was inhibited compared to 1017 without any recombinant plasmid. These results suggest that an intact yfuABC operon enhanced growth of 1017 by acquiring iron from the chelated medium and that YfuC is essential for this function. The truncated YfuC' product is either nonfunctional or unstable. However, in defined TG-Fe with and without EDDA supplementation to 10, 25, or 50 µM, growth enhancement of 1017(pYFU5) compared to 1017 and 1017(pBR322) was not observed (data not shown).

View larger version (19K): [in this window] [in a new window]   FIG. 4.   Growth of E. coli 1017 and derivatives in NB medium supplemented with 50 µM DIP at 37°C. pBR322 is the vector for plasmids pYFU5 and pYFU6. pYFU5 encodes an intact yfuABC operon, while pYFU6 is yfuAB+ yfuC'.

Iron-deficient growth of Y. pestis yfu mutants. Previously, iron acquisition defects of Yfe mutants were masked by the more efficient Ybt-siderophore-dependent system (5). Consequently, we constructed single, double, and triple iron transport mutants to analyze the role of Yfu in iron uptake. The triple mutant (KIM6-2082.1; Ybt Yfe Yfu) grew as well at 26°C on solidified PMH2 containing 60 or 80 µM DIP as its parental strain (KIM6-2031.1; Ybt Yfe Yfu⁺) (data not shown). Thus, the Yfu system is not the undefined 26°C iron transport system described by Lucier et al. (29).

yfuA2082

View larger version (23K): [in this window] [in a new window]   FIG. 5.   Growth of Y. pestis strains KIM6 (Ybt Yfe+ Yfu+), KIM6-2082 (Ybt Yfe+ Yfu), KIM6-2031.1 (YbtYfe Yfu+), and KIM6-2082.1 (Ybt Yfe Yfu) in deferrated PMH2 with (+Fe) and without (Fe) FeCl3 supplementation to 10 µM. Cultures were incubated at 37°C (A) or 26°C (B).

View larger version (23K): [in this window] [in a new window]   FIG. 6.   Growth of Y. pestis strains at 37°C in deferrated PMH2 with increasing concentrations of the iron chelator DIP. (A) KIM6-2031.1 (Ybt Yfe Yfu+); (B) KIM6-2082.1 (Ybt Yfe Yfu).

Iron uptake of Y. pestis yfu mutants. We next compared the abilities of Yfu⁺ and Yfu strains to accumulate ⁵⁵FeCl₃. For these studies, KIM6 (Ybt Yfe⁺ Yfu⁺), KIM6-2082 (Ybt Yfe⁺ Yfu), KIM6-2031.1 (Ybt Yfe Yfu⁺), and KIM6-2082.1 (Ybt Yfe Yfu) were grown in PMH at 26 and 37°C; ⁵⁵FeCl₃ and, where appropriate, the energy poison CCCP were added to growing cell cultures. Energy-dependent iron uptake was observed in all four strains (Fig. 7). Levels of iron accumulation by KIM6 and KIM6-2082 were nearly identical at both temperatures. The Yfe strain KIM6-2031.1 accumulated less iron than its KIM6 parent, as previously demonstrated (5). However, the triple iron transport mutant KIM6-2082.1 (Ybt Yfe Yfu) was as effective as its KIM6-2031.1 (Ybt Yfe Yfu⁺) parent in iron accumulation at both temperatures.

View larger version (26K): [in this window] [in a new window]   FIG. 7.   Uptake of 55FeCl3 by Y. pestis strains KIM6 (Ybt Yfe+ Yfu+), KIM6-2082 (Ybt Yfe+ Yfu), KIM6-2031.1 (Ybt Yfe Yfu+), and KIM6-2082.1 (Ybt Yfe Yfu). Where indicated (closed symbols), cells were metabolically poisoned by addition of 100 µM CCCP 10 min prior to addition of isotope. These data are from a single assay but are representative of three independent experiments.

LD₅₀ studies in mice. To determine the contribution of the Y. pestis Yfu transport system to virulence in mice, we compared KIM5(pCD1Ap)+ (wild type) to KIM5-2082.3+ (Yfu). By subcutaneous injection (to mimic a flea bite), both strains were fully virulent. In two separate trials, KIM5(pCD1Ap)+ had LD₅₀s of 15 and <4.6, while KIM5-2082.3+ had an LD₅₀ of 10.8. We also tested the effect of Yfu in a Yfe background. KIM5-2031.12+ (Ybt⁺ Yfe Yfu⁺) was no more virulent than KIM5-2082.11+ (Ybt⁺ Yfe Yfu); respective LD₅₀s were 74.3 and <82 (lowest concentration tested).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Using Y. pestis genomic DNA, we previously failed to generate a PCR product using primers from the DNA sequence of Y. enterocolitica which showed high amino acid conservation to the PBPs from S. marcescens and N. gonorrhoeae (6). One region of high amino acid conservation, upon which one primer (shown in Fig. 2) was based, differs in Y. pestis YfuA compared to Y. enterocolitica YfuA and SfuA (Fig. 2) and accounts for our previous negative results. In this study, we used BLAST searches with the Y. enterocolitica YfuA sequence to identify the yfuABC operon in Y. pestis KIM10+ (UW Genome Project) and CO92 (Sanger Centre). Unlike the Ybt and Hmu/Hem iron and hemoprotein transport systems of Y. pestis and Y. enterocolitica, which are nearly identical (20, 50), YfuABC in these two organisms shows more divergence despite the high degree of similarity (81.4 to 89.9%) (this study and reference 40). Although the three enteric PBPs are more closely related to each other than to those of N. meningitidis and H. influenzae, YfuA of Y. enterocolitica and SfuA of S. marcescens are more similar to each other than to YfuA of Y. pestis (Fig. 2).

Based on the deduced amino acid sequence, the Y. pestis Yfu ABC transport system belongs to the TC 3.A.1.10 cluster of ABC iron transporters that includes Sfu of S. marcescens, Hit of H. influenzae, Yfu of Y. enterocolitica, and others (39; http://www-biology.ucsd.edu/~msaier/transport/titlepage.html). YfuA serves as the PBP, while YfuB is the IM permease and YfuC is the ATP hydrolase. Saken et al. (40) described a fourth gene, yfuD, that is downstream of yfuC in Y. enterocolitica. YfuD of Y. enterocolitica showed homology to several hypothetical proteins and modest similarity to several bacterial transporters (40). A similar gene is not present within DNA 1 kb downstream of yfuABC in Y. pestis. The best match from a BLAST search of the Y. pestis KIM10+ genome (UW Genome Project) was an ORF on a separate contig than that containing yfuABC. Using this ORF, the best match from a BLAST search of the database was a threonine efflux protein from Salmonella enterica serover Typhimurium (probability score of 5 × 10⁸⁵). The probability score for the aligned regions of this ORF and YfuD of Y. enterocolitica was 2 × 10²². We conclude that Y. pestis does not have a yfuD homolog. Although we believe that YfuD is not part of the Yfu system, it is possible that the Y. pestis Yfu system may not function efficiently due to the absence of YfuD. No outer membrane porin or receptor is associated with the Y. pestis yfuABC locus. In Y. enterocolitica and S. marcescens the Yfu and Sfu systems are not TonB dependent (40, 54), suggesting a TonB-independent receptor or porin. Our experimental results show that the Y. pestis yfuABC promoter is repressed by excess iron via the Fur regulator (Table 2). These results and in vitro transcription-translation (Fig. 3) indicate that the genes are expressed and protein products are made. The recombinant operon enhances growth of E. coli 1017 (an enterobactin-deficient mutant) in NB chelated with DIP. However, Yfu is not the putative 26°C iron transport system hypothesized from iron chelators studies (29).

It is possible that the Y. pestis Yfu system transports a cation other than iron as its primary substrate. However, complementation of an iron transport defect in E. coli by Y. pestis yfuABC, iron-repressible expression of Y. pestis yfuABC, and the degree of similarity of the Y. pestis Yfu system to ABC iron transport systems of other organisms all suggest that iron is the primary substrate. Nonetheless, the Yfu system does not appear to be a major system for iron acquisition by Y. pestis in mice or under the in vitro conditions tested in this study. No differential effect on growth of Yfu mutants in PMH2 supplemented with the iron chelator DIP or EDDA was observed. Short-term iron uptake assays did not detect any uptake due to the Yfu system in Y. pestis (Fig. 7). As measured by LD₅₀ studies, Y. pestis Yfu and Yfe Yfu mutants were no less virulent than their Yfu⁺ parental strains. Saken et al. (40) also failed to demonstrate a role for Yfu in the virulence of Y. enterocolitica in mice.

The residual growth and iron uptake of a Ybt Yfe Yfu strain of Y. pestis at 37°C suggests that an unidentified system that acquires iron even under iron-chelating conditions is functioning. Whether this system is specific for iron or iron is accumulated by a system designed for uptake of another cation or substrate remains to be determined.

The role and importance of Yfu in the survival and disease properties of plague are uncertain. Yfu might be an ancestral system that is no longer essential to the lifestyle of Y. pestis. Alternatively, Yfu could be important for survival under environmental conditions that we have not tested. Perhaps the appropriate bound iron source for Yfu uptake was not present in the media that we used. Yfu might have a role in pneumonic plague but not bubonic plague. We have previously hypothesized that different iron/hemoprotein transport systems are effective in different organ systems in mammals (5); this may extend to different rodent species. Yfu appears to be irrelevant in the mouse but may be important for iron acquisition in one or more of the many other rodent hosts sensitive to plague. Finally, the Yfu system could be important in acquiring iron during growth in the flea gut. Determination of the validity of any of these speculations awaits future testing.