Identification and Characterization of the Hemophore-Dependent Heme Acquisition System of Yersinia pestis

Abstract

Yersinia pestis possesses a heme-protein acquisition system (Hmu) that allows it to utilize heme and heme-protein complexes as the sole sources of iron. Analysis of the Y. pestis CO92 genomic sequence revealed a second heme-protein acquisition gene cluster that shares homology with the hemophore-dependent heme acquisition system (Has system) of Serratia marcescens. This locus consisted of the hasR_yp receptor gene, the hasA_yp hemophore gene, and genes encoding components of the HasA_yp dedicated ABC transporter factor (hasDE_yp), as well as a tonB homologue (hasB_yp). By using a reconstituted secretion system in Escherichia coli, we showed that HasA_yp is a secreted heme-binding protein and that expression of HasA_yp is iron regulated in E. coli. The use of a transcriptional reporter fusion showed that the hasRADEB promoter is Fur regulated and has increased activity at 37°C. Hemoglobin utilization via the Has_yp system was studied with both E. coli and Y. pestis, for which has and has hmu mutant strains were used. No contribution of the Has system to heme utilization was observed in either E. coli or Y. pestis under the conditions we tested. Previously it was shown that a deletion of the Hmu system had no effect on the virulence of Y. pestis in a mouse model of bubonic plague. An Hmu⁻ Has⁻ double mutant also retained full virulence in this model of infection. This report constitutes the first attempt to investigate the contribution of the hemophore-dependent heme acquisition system in bacterial pathogenicity.

Iron plays a central role in the metabolism of most bacteria, in which the Fe²⁺/Fe³⁺ redox pair and its wide range of electron transfer capacity are well suited to many enzymatic reactions. Despite iron abundance in nature, bacteria face an acute iron supply problem due to the extremely low concentration of soluble Fe³⁺ at physiological pH under aerobic conditions. In vertebrate hosts, iron deficiency is further increased by a set of iron-withholding defenses leading to the sequestration of most intracellular iron by ferritin or hemoproteins such as hemoglobin, whereas extracellular iron is complexed by transferrin, lactoferrin, hemopexin, and haptoglobin (5, 42).

To overcome this problem, gram-negative bacteria have evolved numerous energy-dependent iron-regulated systems to acquire iron from iron- or heme-containing compounds by three distinct acquisition strategies. These include (i) the release of small molecules with a high affinity for iron called “siderophores,” (ii) the direct binding of iron-containing protein compounds to outer membrane receptors, or (iii) the extracellular secretion of heme-binding proteins called “hemophores” that subsequently transport and deliver heme to their cognate outer membrane receptor. The coexistence of more than one of the above iron-scavenging systems is relatively common in bacteria (for review, see reference 42). Iron acquisition has been shown to be critical for the survival of pathogenic bacteria during the course of infection. In some instances, the first two strategies described above have been directly related to bacterial virulence (2, 14, 27). However, it is still not known to what extent the more recently characterized hemophore-dependent heme acquisition system contributes to the virulence of the bacteria for which it has been described so far: Serratia marcescens (22), Pseudomonas aeruginosa (24, 25), and Pseudomonas fluorescens (29).

Yersinia pestis, the bacterial agent of bubonic and pneumonic plague, possesses several iron acquisition systems that may be important for survival in its various iron-deficient niches. Inorganic iron can be acquired via an iron and manganese uptake system (termed Yfe) or through the secretion of a siderophore called “yersiniabactin” (6, 15, 30a). Y. pestis can also use hemin and a wide variety of host heme-containing compounds through a heme transport system (termed “Hmu”) (17) composed of at least five components: HmuR, -S, -T, -U, and -V. Under iron-depleted conditions, HmuR, the outer membrane receptor, and HmuTUV, a putative ABC-dependent transport system, are required for the use of hemin, hemin-albumin, and myoglobin, while HmuR is the only Hmu component needed to use hemoglobin and heme-hemopexin as iron sources. HmuS (HemS in Yersinia enterocolitica) may participate in removal of iron from hemin; however, conclusive demonstration of its role has been elusive (38, 39, 41, 42). Studies performed with mice showed that although both the iron/manganese and the yersiniabactin transport systems are important for Y. pestis, a deletion of the Hmu system has no effect on virulence in mice by a subcutaneous route of infection (2, 41)

During the course of the Hmu study, an hmu deletion mutant still grew under iron-depleted conditions in the presence of high concentrations of hemoglobin, leading to the hypothesis that an Hmu-independent uptake system may be involved in hemoglobin utilization (41). One possible candidate was identified in the Y. pestis CO92 genomic sequence released by the Sanger Centre as a sequence sharing homology with the hemophore-dependent heme acquisition system (Has system) of S. marcescens (22), P. aeruginosa (24, 25), and P. fluorescens (29). HasA hemophores are small extracellular proteins secreted by an ABC-dependent pathway, a process involving their carboxy-terminal secretion signal and its recognition by a tricomponent membrane ABC transporter. HasA is a heme-binding protein that delivers heme to a dedicated outer membrane receptor, HasR. Interaction between HasA and HasR induces a 100-fold increase in the capacity of the system to bind heme compounds and broadens the spectrum of potential heme sources for HasR (22, 24, 42).

In this study, we have begun characterizing the Has system of Y. pestis and investigated its biological function. We showed that HasA_yp is a secreted heme-binding protein. Secretion, hemoglobin utilization, and induction conditions of the Has_yp system were studied with both Y. pestis and Escherichia coli. The effect of the deletion of the Y. pestis Has system in conjunction with a deletion of the Hmu system on the virulence of Y. pestis was investigated with a mouse model of infection. This report constitutes the first attempt to investigate the contribution of the hemophore-dependent heme acquisition system to bacterial pathogenicity.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The bacterial strains and plasmids used in this study are listed in Table 1. E. coli strains were cultivated in Luria broth (LB) (28). Y. pestis was grown in LB or on tryptose blood agar base (TBA). For growth under iron-restricted conditions, Y. pestis cells were grown in the defined medium PMH or PMH2 (13, 37). PMH2 is a modified version of PMH that contains 50 μM PIPES (piperazine-N, N′-bis[2-ethanesulfonic acid]) instead of HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid). This change reduces medium acidification due to bacterial growth. PMH and PMH2 were deferrated with Chelex 100 resin (Bio-Rad) prior to filter sterilization as previously described (37). Alternatively, to reduce iron available to E. coli or Y. pestis, cells were grown in LB medium containing 0.2 mM 2,2′-dipyridyl (LBD). Hemin, hemin-agarose, and bovine hemoglobin were obtained from Sigma Chemical Co. Bovine hemoglobin agar plates were prepared as described in reference 12.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant genetic markers or features	Source or reference
Straina
Y. pestis
KIM5(pCD1Ap)+	Wild type, Ampr	13
KIM6+	Lcr− (pCDI−)	30
KIM6	Lcr− Pgm− (Ybt− Δpgm)	30
KIM6-2030	Lcr− Pgm− Fur− (fur::kan), Kmr	36
KIM6-2061.1	Lcr− Pgm− HmuR−	This study
KIM6-2085	Lcr− Pgm− HmuR− Has−	This study
KIM6-2045.1	Lcr− Psn−	10
KIM6-2081.1+	Lcr− Hmu− Has−	This study
KIM5-2045.7	Lcr+ (pCDIAp) Ampr Psn−	This study
KIM5-2081.1+	Lcr+ (pCDIAp) Ampr Hmu− Has−	This study
6/69	Wild type	3
6/69c	Lcr− (pYV−)	This study
6/69c(hasA)	Lcr− HasA−	This study
6/69c (pPla−)	Lcr− pPla−	This study
6/69 (pPla−)	pPla−	This study
6/69 (HPI−)	Pgm− (Δpgm)	7
E. coli
C600	F−thr leu fhuA lacY thi supE	Laboratory collection
TG1	supE thi Δlac-proAB F traD36 proAB lacIqZΔM15	Laboratory collection
MC4100	F−araD139 ΔlacU169 relA1 rpsL150 thi mot flb5301 deoC7 ptsF25 rbsR	Laboratory collection
MC4100 fur::cat	fur	Laboratory collection
MC4100 hemA	hemA	R. Kadner
Plasmids
pACYC184	Medium copy no., Tetr Cmr	9
pSyc150	HasDEsm Cmr	22
pEU730	Single-copy-no. reporter vector, promoterless lacZ, Spr	11
pCD1Ap	bla cassette inserted into ′yadA of pCD1 from Y. pestis KIM5, Ampr	13
pLG338	Low copy number, Tetr Kmr	40
pR10K	HasRxm Kmr	12
pBGS18+	Kmr	35
pHas98	HasRADEByp 8.828-kb Sau3A Y. pestis KIM6 fragment inserted at BamHI site	This study
pHas98ΔBg12	HasByp Deletion from hasR through hasE, 5.375-kb BglII deletion of pHas98	This study
pACYCHas	HasRADEByp' Cmr, pHas98 8.5-kb EcoRI-StuI into pACYC184 EcoRI-ScaI	This study
pHasRA	HasRAyp+D′ Ampr, pHas98 4.3-kb HindIII-Asp718 into pUC19 HindIII-KpnI	This study
PHasRA2	HasRAyp+D′, Kmr, pHasRA EcoRI-HindIII insert into pBGS18	This study
pHasA	HasRAyp+ Ampr, PCR fragment into EcoRI-BamHI sites of pUC18	This study
pEUHas	Single-copy-no. hasR::lacZ reporter, 383-bp hasR promoter ligated into pEU730, Spr	This study

^aFor Y. pestis KIM strains, + designates strains with an intact pgm locus. 

Molecular biological procedures.

Standard procedures were used for cloning and analysis of DNA, PCR, electroporation, and transformation. Enzymes used to manipulate DNA were from New England Biolabs. Oligonucleotides were from Gibco BRL or Integrated DNA Technologies, Inc. Some DNA sequencing was performed by Genome Express S.A. In some instances, sequencing reactions were performed via the dideoxynucleotide chain termination method (33) by using ³⁵S-dATP (Amersham/USB), Sequenase version 2.0 (Amersham/USB), and 7-deaza-dGTP (Boehringer Mannheim Biochemicals). Samples were electrophoresed through a 6% polyacrylamide gel containing 8.3 M urea (Sigma) cast in Tris-borate-EDTA buffer (32). Dried gels were exposed at room temperature to Kodak Biomax MR film.

Sequence analysis.

Amino acid sequence comparisons were performed with the Clustal package program (16). Homology searches were performed with Blast 2.0 at http://www.ncbi.nlm.nih.gov/BLAST/unfinishedgenome.html. DNA sequence manipulations were done with DNAStrider 1.3 (26) or DNAMAN Version 4.21 (Lynnon Biosoft).

Construction of recombinant plasmids.

pHasA was created as follows. The hasA_yp gene was amplified by PCR with oligonucleotides YpHasArbs-5 (5′- CCCGAATTCTTAATAATTAAAAGGACATTATCATGAGTA-3′) and YhasAend-3 (5′-CCCAAGCTTAGCCCTAGCGGTGGTATCAATGACTCA GCC-3′) as primers and with Y. pestis 6/69 genomic DNA as a template. The amplified DNA was digested with EcoRI and HindIII and ligated into the same sites of pUC18. For the construction of pEUHas, in which the lacZ reporter gene was placed under the control of the hasR promoter region, a 383-bp region upstream of the putative initiating methionine residue for HasR was amplified from pHas98 by using PCR primers Hasp1 (5′-GATCTTTTAAACTTAATAGACTG-3′) and Hasp2 (5′-TTATGAAACTTCATCCTTAAC-3′). The products were blunt-end ligated into the PmeI site of pEU730 (11), and the resulting plasmid was designated pEUHas. All constructs were checked by PCR with specific and vector-based primers. Then each construct was sequenced.

Construction of has mutations in Y. pestis strains.

The Sanger Centre database of Y. pestis CO92 DNA was searched for has-related sequences by using the hasRADE sequence from S. marcescens. Based on the Y. pestis sequence, PCR primers HasYP1 (5′-CCGATTACAGCATCTCCT-3′) and HasYP2 (5′-GCGCATCAGACATATAACC-3′) were designed and used to produce a probe for screening a Y. pestis KIM6+ genomic DNA library. The resultant clone, pHas98, contains an approximately 8.8-kb fragment inserted into pLG338. An ∼5.4-kb BglII fragment was removed from pHas98 to create pHas98ΔBgl2. This deletion removes all of hasA and hasD as well as portions of hasR and hasE. A 2.5-kb StuI-to-HpaI fragment was cloned from pHas98ΔBgl2 into the SmaI site of pKNG101 (18) to yield pKNGhas. The mutated has alleles were introduced into Y. pestis KIM6 strains by allelic exchange as previously described (2). The genotype of all strains was confirmed by PCR with HasYP1 and HasYP2 primers, which are missing from the mutants, as well as HasD1 (5′-GATGCGCCATACAGAGCC-3′) and HasD2 (5′-GGCATTGTCGCCATCAGG-3′) primers, which flank the deleted region.

Protein analysis.

E. coli strains harboring various plasmids, S. marcescens SM365, and Y. pestis strains were grown at 28°C (Y. pestis) or 37°C (Y. pestis, S. marcescens, and E. coli) in LB, LBD, or PMH2. Cells from overnight or exponential cultures were centrifuged for 10 min at 5,000 × g at 4°C. The supernatants were concentrated by precipitation with either 10% trichloroacetic acid (inactive supernatants that do not retain HasA heme-binding activity) or 80% ammonium sulfate (active supernatants that retain HasA heme-binding activity). Inactive concentrated cell extracts were prepared by boiling whole cells in sodium dodecyl sulfate (SDS) buffer. Proteins were analyzed by SDS-polyacrylamide gel elctrophoresis (PAGE) followed by either Coomassie blue staining or Western blot analysis. Anti-HasAyp antibodies were prepared against HasA_yp protein purified on hemin-agarose, separated on SDS-polyacrylamide gel, and electroeluted. The protein was mixed with Titer Max adjuvant (Interbiotech) and injected into a rabbit to raise antibodies. The rabbit serum was used in Western blot analysis at a dilution of 1/2,000. Nondenaturating PAGE was performed as described in reference 25.

For analysis of in vitro-synthesized products, ∼5 μg of purified, circular, plasmid DNA was added to an E. coli S30 extract system (Promega Corp.) and supplemented with ³⁵S-labeled amino acids (DuPont NEN Research Products). After a 30-min incubation at 37°C, the products were acetone precipitated, resuspended in sample buffer, and electrophoresed on a 12% polyacrylamide gel containing SDS. The gels were dried and exposed overnight to Kodak BioMax MR film.

β-Galactosidase assays.

Lysates of Y. pestis strains carrying pEUHas were prepared from cells exponentially growing in PMH2 in the presence or absence of iron through two transfers for a total of approximately 6 generations. The β-galactosidase activity, with ONPG (o-nitrophenyl-β-d-galactopyranoside) used as a substrate, was measured spectrophotometrically and expressed in Miller units (28).

HasA_yp purification heme affinity chromatography.

Five hundred milliliters of E. coli C600(pSyc150, pHasA) culture grown overnight at 37°C was centrifuged for 10 min at 5,000 × g at 4°C. The supernatant was concentrated by 80% ammonium sulfate precipitation. The protein pellet was resuspended and extensively dialyzed against 10 mM Tris-HCl (pH 7.5) and then subjected to heme affinity chromatography as described in reference 25. According to the manufacturer's specifications, there is a 10- to 100-fold excess of putatively available hemin compared to the estimated number of HasA molecules used in the hemin-agarose procedure Inactive and active supernatants (for biological tests) were prepared as described in reference 25.

Role of Has in hemoglobin utilization.

Growth stimulation of an E. coli hemA mutant producing HasR_sm by exogenously supplied HasA hemophores was tested as follows. The HasR_sm-producing strain was mixed with 3 ml of top agar and poured onto LBD plates supplemented with 10⁻⁶ M hemoglobin. Five-millimeter-diameter wells were cut in the agar and filled with 50 μl of hemin-agarose-purified HasA_yp or HasA_sm protein. Growth around the wells was recorded after overnight incubation at 37°C.

For Y. pestis strains, cells were grown at 28°C in the deferrated, defined medium, PHM2, in the absence of iron through two transfers for a total of approximately 6 generations. Third-transfer cultures were inoculated to an optical density at 620 nm (OD₆₂₀) of 0.1 into PMH2 containing 50 μM ethylenediamine-N,N′-diacetic acid (EDDA) and either 0, 1.25, 2.5, 5, or 10 μM hemoglobin, assuming a molecular weight of 64,640 for hemoglobin. Growth of cells was monitored at regular intervals with a Genesys5 spectrophotometer (Spectronics Instruments, Inc.).

Dot blot HasA-binding assay.

The interaction between HasA_sm or HasA_yp with HasR_sm was examined as follows: Cultures of MC4100 fur::cat carrying pBGS18+ or pR10K were grown exponentially in LB medium at 37°C and harvested when they reached an OD₆₀₀ of 1. Cell pellets were suspended in Tris-buffered saline (TBS) buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl) to an OD₆₀₀ of 0.5. Aliquots of 50 ml were applied to nitrocellulose filters. The filters were dried for 10 min at room temperature and saturated by incubation for 1 h at 37°C in BS blocking solution (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.5% skim milk, 0.001% antifoam, 0.01% NaN₃). The filters were then incubated for 30 min at 30°C in 1 ml of 50 mM Tris-HCl (pH 7.5) containing serial dilutions of either HasA_yp or HasA_sm or containing no HasA, washed three times with BS, and then incubated for 1 h at 37°C with polyclonal anti-HasA_sm or anti-HasA_yp antibodies diluted 1:2,000 in blocking solution. After three washes in BS, the filters were incubated with an alkaline phosphatase-conjugated goat anti-rabbit serum diluted 1:7,000 in BS. After three washes in BS, antibody binding was revealed as in other immunodetection tests (23).

Virulence testing.

Y. pestis strains lacking the low-calcium-response virulence plasmid are avirulent by any route of infection. A derivative of this plasmid, containing an ampicillin resistance gene inserted into the pseudogene yadA′/′yadA (pCD1Ap) (13), was electroporated into KIM6+, KIM6–2045.1, and KIM6–2081+, generating KIM5(pCD1Ap)+, KIM5–2045.7, and KIM5–2081.1+. All strains were constructed and tested for virulence in a BL3 facility. Bacteria were grown at 26°C in PMH2 containing 50 μM hemin and ampicillin (100 μg/ml) through two transfers for a total of approximately 7 generations. Cells were harvested at an OD₆₂₀ of between 0.5 and 0.7, pelleted, and resuspended in mouse isotonic phosphate-buffered saline (PBS; 149 mM NaCl, 16 mM Na₂HPO₄, 4 mM NaH₂PO₄ [pH 7.0]). Five- to 7-week-old female NIH/Swiss Webster mice were injected subcutaneously with 0.1 ml of 10-fold serial dilutions of the bacterial suspensions. Groups of four mice were used for each bacterial dose. The number of cells injected was determined by plating serial dilutions on TBA plates containing ampicillin. Mice were monitored daily for 2 weeks. Fifty percent lethal doses (LD₅₀s) were calculated by the method of Reed and Muench (31).

RESULTS

Identification of the Y. pestis Has system.

Analysis of the Y. pestis CO92 (biotype orientalis) complete genomic sequence released by the Sanger Centre (http://www.sanger.ac.uk/Projects/Y_pestis) revealed the existence of a locus sharing homology with the hemophore-dependent heme aquisition system (Has, for heme acquisition system) of S. marcescens, P. aeruginosa, and P. fluorescens. The Y. pestis has operon contains five open reading frames (ORFs) (Fig. 1) called “hasR_yp,” “hasA_yp,” “hasD_yp,” “hasE_yp,” and “hasB_yp ” according to the nomenclature used for the homologous genes of the S. marcescens has operon (12). The predicted sizes, protein locations, and putative functions, as well as percent identities of the deduced amino acid sequences with those found in protein databases, are shown in Table 2. A comparison of the CO92 has sequences with those from Y. pestis KIM10+ (biotype mediaevalis) (http://magpie.genome.wisc.edu/cgi-bin/Authenticate.cgi/uwgp_blast.html), shows that the deduced amino acid sequences of these five ORFs are nearly identical between these two biotypes (data not shown). The deduced amino acid sequence of HasR_yp, the putative outer membrane receptor of the system, displayed a high level of identity with HasR_sm, except for a short N-terminal extension missing in HasR_yp (Fig. 2B). A well-conserved Fur binding site sharing 94% identity with the E. coli consensus sequence (GATAATGATAATCATTATC) was identified 340 bp upstream of the hasR_yp initiation codon, suggesting that has_yp genes may be iron regulated. Further upstream of hasR_yp, a gene coding for a protein 90% identical to the E. coli argino-succinate lyase ArgH was identified. The intergenic region between argH and hasR contains a stem-loop structure, suggesting that argH is not part of the hasRADEB putative operon. While this stem-loop structure may act as a transcription terminator, its localization close to the Fur binding site suggests that this structure may play a role in the Has system regulation. Finally, the deduced amino acid sequence of a protein 50.6% identical to HasF_sm, the third component of the HasA transporter, was found to be located outside the main has_yp putative operon (Fig. 1). A single ORF was found in hasF_yp locus, located 281 kb away from hasB. Thus, HasF_yp does not seem to be part of an operon and does not contain any Fur binding site located upstream of the coding sequence.

FIG. 1.

Genetic organization of the Y. pestis Has system. Genetic organization of the 8.3-kb Y. pestis Has_yp system region identified in this study. The partial sequence comparison of the Fur binding site is shown above the diagram. The stem-loop structure located 40 bp upstream of the Fur binding site is indicated.

TABLE 2.

Properties and homologies of proteins of the Y. pestis Has system

Protein	Mass (kDa)	Location or putative functiona	% Identity with system:
			Hassm	Haspa	Haspf
HasR	93.4/89.6b	OM, HasA receptor for uptake	48.0	38.0	44.0
HasA	22.2	Secreted hemophore	33.8	28.7	20.3
HasD	65.0	IM, ATP-binding protein, HasA export	55.0	56.1	56.9
HasE	48.9	IM, ABC permease, HasA export	44.2	43.0	39.1
HasB	29.3	IM TonB-like protein	25.3	NAc	NA
HasF	50.6	OM, HasA export	69.6	13.4	27.8

^aOM, outer membrane; IM, inner membrane. 

^bUnprocessed/processed. 

^cNA, not applicable. 

FIG. 2.

Clustal W amino acid sequence alignments of HasA and the N-terminal region. For both panels, identical amino acids are shown in black boxes, while similar amino acids are shown in gray boxes. The consensus line (Con.) shows identical (uppercase) and similar (lowercase) amino acids. (A) Amino acid sequence comparison of the N-terminal regions of the Y. pestis and S. marcescens HasR hemophore-dependent receptors HasR_yp and HasR _sm, respectively. (B) Amino acid sequence comparison of HasA hemophores identified in Y. pestis, S. marcescens, P. aeruginosa, and P. fluorescens (HasA_yp, HasA_sm, HasA_pa, and HasA_pf, respectively).

HasA_yp is a secreted heme-binding protein.

The homology between HasA_sm and HasA_yp (33.8% [Table 2]) suggested that, like HasA_sm, HasA_yp may be a secreted protein, although the differences between HasA_yp and other HasA hemophores (Fig. 2A), particularly in the C-terminal region corresponding to the secretion signal, raised some questions about the capacity of HasA_yp to be secreted. Previous studies showed that HasA_sm is secreted in E. coli by a hybrid transporter composed of two S. marcescens inner membrane proteins, HasD_sm and HasE_sm, and the E. coli outer membrane protein TolC, which complements HasF_sm function (22). We tested whether this transporter also secretes HasA_yp. E. coli MC4100, carrying pSYC150, which expresses hasD_sm and hasE_sm, was cotransformed with pHasA_yp, which contains hasA_yp cloned with its own ribosome binding site under Plac promoter control (Table 1). Figure 3 shows that HasA_yp was as efficiently secreted by the Has hybrid transporter as HasA_sm.

FIG. 3.

HasA_yp is secreted through the S. marcescens Has system. Samples were prepared as described in Materials and Methods and Coomassie blue stained after SDS-PAGE analysis. The equivalent of 1 ml of supernatants from cultures at an OD₆₀₀ of 0.5 was loaded in each lane. Lanes: 1, molecular size markers; 2, MC4100 hemA; 3, MC4100 hemA(pSyc150); 4, MC4100 hemA(pHasA_yp); 5, MC4100 hemA(pSyc150, pHasA_yp); 6, MC4100 hemA(pSyc150, pHasA_sm); 7, molecular size markers.

To study the biological function of HasA_yp, the culture supernatant of strain E. coli MC4100(pSyc150, pHasA_yp) grown in LB medium was precipitated with 80% ammonium sulfate, solubilized, and subjected to affinity chromatography on hemin-agarose. Unbound and bound proteins were analyzed by SDS-PAGE and Coomassie blue staining. A 19-kDa protein corresponding to HasA_yp was retained on hemin-agarose, showing that HasA_yp is a heme-binding protein (Fig. 4A). To test whether HasA_yp produced in E. coli was able to acquire heme from hemoglobin, the samples were loaded in a nondenaturing polyacrylamide gel in which hemoproteins are known to migrate without dissociating from heme. In these gels, heme causes a significant shift in the migration of the bound polypeptide. As observed in previous studies (21, 25), the electrophoretic mobility of both HasA_sm and HasA_yp was increased in the presence of heme or hemoglobin (Fig. 4B), thus demonstrating that HasA_yp can acquire heme from hemoglobin.

FIG. 4.

HasA_yp is a heme-binding protein. (A) Isolation of HasA_yp by heme affinity chromatography. Coomassie blue staining is shown. Concentrated supernatants were incubated with hemin-agarose as described in Materials and Methods. (Left) HasA_sm, NA, E. coli MC4100(pSyc150, pHasA_sm) concentrated supernatant not adsorbed (flowthrough material in excess) on hemin-agarose; A, E. coli MC4100(pSyc150, pHasA_sm) concentrated supernatant adsorbed on hemin-agarose. (Right) HasA_yp, NA, E. coli MC4100(pSyc150, pHasA_yp) concentrated supernatant not adsorbed on hemin-agarose; A, E. coli MC4100(pSyc150, pHasA_yp) concentrated supernatant adsorbed on hemin-agarose. The presence of HasA_sm and HasA_yp in NA lanes indicates that the capacity of the hemin-agarose columns was saturated by the supernatant concentrates used here. (B) Heme acquisition and formation of holoHasA_sm and holoHasA_yp after incubation of HasA_sm or HasA_yp with heme or hemoglobin. Four microliters (1 μg/ml) of HasA_sm prepared from ammonium sulfate-concentrated supernatants of E. coli MC4100(pSyc150, pHasA_sm) and 2 μl of HasA_yp (2 μg/ml) prepared from ammonium sulfate-concentrated supernatant of E. coli MC4100(pSyc150, pHasA_yp) were incubated with equal volumes of heme or hemoglobin (10⁻³ M) for 45 min at room temperature. The mixtures were loaded onto a 15% polyacrylamide gel and electrophoresed in the absence of SDS. The gel was stained with Coomassie blue. Protein bands complexed with heme (holoproteins) have a faster migration rate than uncomplexed protein (apoproteins).

Search for HasA_yp in Y. pestis supernatants.

In order to investigate HasA_yp production and secretion in Y. pestis, the culture supernatants and cell pellets of strains 6/69 and KIM6+ grown under iron-rich or iron-depleted conditions were harvested and probed with anti-HasA_yp antibodies. Iron deprivation of Y. pestis was successfully obtained, since the iron-regulated high-molecular-weight proteins (8) were present in cell extracts (data not shown). Only a very faint band cross-reacting with the HasA antibodies was detected at ca. 15 kDa, in the cell pellet of iron-replete and iron-starved bacteria (data not shown). This band was still detected in the supernatant of Y. pestis 6/69_c (has mutant), in which the chromosomal hasA_yp gene was replaced by a truncated copy by reverse genetics, suggesting that it corresponds to some cross-reacting material. The effect of temperature on hasA_yp expression was evaluated by growing the cells at 28 or 37°C. Neither temperature allowed the visualization of HasA_yp. The influence of the presence or absence of the Y. pestis plasmids pYV and pPla and of the pgm locus (containing the high-pathogenicity island [HPI]) was also investigated by using the isogenic strains Y. pestis 6/69, 6/69_c, 6/69_c (pPla⁻), 6/69(pPla⁻), and 6/69(HPI⁻), but none of the mutants produced HasA_yp. Similar negative results were obtained with Y. pestis KIM6+ (data not shown).

The reasons for the absence of HasA detection in Y. pestis may be that (i) the induction conditions used were not appropriate, (ii) HasA is unstable or rapidly degraded, (iii) its secretion rate is too low to be detected, or (iv) the Has system is not functional. Some of these hypotheses were investigated.

In Y. pestis, the hasRADEB promoter is Fur regulated and presents increased activity at 37°C.

To investigate transcription from the hasRADEB promoter region, we cloned the putative promoter region of hasRADEB upstream of lacZ in the reporter plasmid pEU730 (11). This construct, termed pEUHas, was electroporated into Y. pestis strains KIM6+, KIM6, and KIM6–2030. The cells were grown in the presence or absence of iron and assayed for β-galactosidase activity. The results indicated that expression of lacZ from this promoter fragment was iron regulated, showing an approximately three- to fivefold induction when the cells were grown in the absence of iron (Table 3). Transcription of hasR::lacZ apppears to be regulated by Fur, because the levels of β-galactosidase activity were constitutively high in the fur mutant, KIM6–2030(pEUHas). We also observed a minor but reproducible difference in the β-galactosidase values obtained with KIM6(pEUHas)+ and KIM6(pEUHas), which were slightly higher in the KIM6+ strain, suggesting that a component of the 102-kb pgm locus may participate in the expression of hasRADEB. Finally, β-galactosidase activity was six- to sevenfold higher in KIM6(pEUHas)+ cells grown at 37°C than in those grown at 26°C.

TABLE 3.

β-Galactosidase activities of Y. pestis strains containing lacZ reporters

Straina	β-Galactosidase activity (mean ± SE Miller units)a		−Fe/+Fe ratio
	+Fe	−Fe
KIM6(pEU730)+	118 ± 14	203 ± 71	1.7
KIM6(pEUHas)+	8,264 ± 1,532	39,738 ± 6,182	4.8
KIM6(pEU730)	227 ± 49	285 ± 26	1.3
KIM6(pEUHas)	5,822 ± 1,208	19,023 ± 6,597	3.3
KIM6-2030(pEU730)	200 ± 46	446 ± 62	2.2
KIM6-2030(pEUHas)	22,152 ± 4,492	28,674 ± 5,235	1.3

^aBacterial cells were grown in deferrated PMH2 at 37°C and harvested during mid-log-phase growth. Whole-cell lysates were assayed for β-galactosidase activity as described in reference 28. 

Expression of HasA_yp is iron regulated in E. coli.

To study expression from the hasRADEB_yp promoter region in E. coli, we monitored HasA_yp secretion in E. coli MC4100(pSYC150) carrying pHasRA2 containing hasR_yp and hasA_yp cloned with their own putative promoter region. As revealed by Coomassie blue staining or immunodetection with anti-HasA_yp antibodies, HasA_yp was secreted into the supernatant of E. coli MC4100(pSYC150, pHasRA2) grown in iron-depleted but not iron-rich media (Fig. 5). These results indicate that in E. coli, hasA_yp (and possibly hasR_yp) is expressed from its own promoter and that expression is iron regulated.

FIG. 5.

Iron regulation of HasA_yp secretion in E. coli. Supernatant samples were prepared as described in Materials and Methods and subjected to SDS-PAGE analysis followed either by Coomassie blue staining (A [equivalent of 10 ml of supernatant from cultures at OD₆₀₀ of 0.5 loaded in each lane]) or immunodetection with anti-HasAyp antibodies (B [equivalent of 1 ml of culture at OD₆₀₀ of 0.5 loaded in each lane]). (A and B) Lanes: 1, molecular size markers; 2, MC4100(pHasRA2) in LB medium; 3, MC4100(pHasRA2) in LB medium supplemented with 0.2 mM 2,2′-dipyridyl; 4, MC4100(pSyc150, pHasRA2) in LB medium; 5, MC4100(pSyc150, pHasRA2) in LB medium supplemented with 0.2 mM 2,2′-dipyridyl.

Study of the functionality of the Y. pestis Has secretion system in E. coli

To test the functionality of the Y. pestis HasA_yp transport system, the entire hasRADEB locus was cloned into pLG338 or pACYC184 to produce plasmids pHas98 and pACYCHas (Table 1). pHas98 and pACYCHas were introduced into E. coli MC4100, and the recombinant strains were grown in iron-rich (LB) or iron-depleted (LBD) medium and monitored for HasA_yp secretion. No secreted HasA_yp could be detected in the iron-depleted or iron-rich medium (data not shown).

In vitro transcription or translation experiments with pHas98 showed Has-specific bands at ca. 87, 25, and 22 kDa (see arrows in Fig. 6, lane 2). The same bands were observed when a clone containing only hasRAD′ (pHasRA) was used in the transcription/translation system (data not shown), but are absent from reactions containing the vector plasmid, pLG338 (Fig. 6, lane 1). The 87-kDA band likely corresponds to HasR_yp, while the 25- and 22-kDa proteins probably represent HasA_yp and a degradation product generated from HasA_yp. None of these bands was present in reactions containing pHas98ΔBglII (Fig. 6, lane 4). Instead, a new protein of ca. 41 kDa is seen, which is similar in size to the ca. 43-kDa fusion product predicted to occur as a result of the deletion. No bands corresponding to HasD, HasE, and HasB were convincingly observed. All three proteins contain at least one cysteine as well as several methionine residues and thus should have been labeled in the in vitro reactions. These results indicate that at least HasR_yp and HasA_yp are expressed from pHas98 and that the HasA_yp secretion system encoded by pHas98 or pACYCHas may be expressed but not functional when reconstituted in E. coli. The biological function of the Y. pestis Has system was further investigated as described below.

FIG. 6.

Autoradiogram of plasmid-encoded proteins labeled with ³⁵S-labeled amino acids by in vitro transcription and translation. The sizes of molecular mass standards (lane 3) are shown on the right side of the figure. The locations of HasR, HasA, and a HasA degradation product (lane 2) are indicated by the arrows on the left side of the figure. The arrow on the right points to a fusion protein (lane 4) generated by a deletion within the has operon. The plasmids used for the in vitro transcription and translation reactions were pLG338 vector (lane 1), pHas98 (lane 2), and pHas98ΔBgl2 (lane 4).

Study of the Y. pestis HasA-dependent heme acquisition system in E. coli

The biological function of HasA_sm is the delivery of heme scavenged from various heme sources to the outer membrane receptor HasR_sm. In S. marcescens, HasA_sm is required for iron acquisition from heme and hemoglobin and was shown to interact with its outer membrane receptor, HasR_sm, both in vivo and in vitro (10). To test whether HasA_yp could substitute for the biological function of HasA_sm, we performed an in vivo growth stimulation test with hemoglobin plates (see Materials and Methods). In this test, a lawn composed of HasR_sm-expressing cells [MC4100 hemA(pR10K)] is plated on LB agar containing a low concentration of hemoglobin, and a solution of HasA_sm or HasA_yp is placed in a well in the agar. A productive interaction between HasA_sm and HasR_sm leads to bacterial growth around the HasA-containing well. As expected, HasA_sm promoted bacterial growth, but this promotion effect was not observed around the well containing native HasA_yp (data not shown). Consistently, in vitro dot blot binding experiments did not show any interaction between HasA_yp and HasR_sm (data not shown). These results suggest that HasA_yp cannot substitute for the biological function of HasA_sm, probably due to its lack of functional interaction with HasR_sm.

The expression of hasA_yp and hasR_yp from plasmids pHasRA and pHasRA2, carrying hasR_yp and hasA_yp, raised the possibility that HasR_yp could reach the outer membrane and is able to interact with hemophores available at the bacterial cell surface. To test this, we performed either in vivo growth stimulation tests with hemoglobin or in vitro dot blot binding experiments with cells potentially expressing HasR_yp [MC4100 hemA(pHasRA)]. We did not see any growth stimulation or dot blot interaction with either HasA_yp or HasA_sm (data not shown). These results suggest that if HasR_yp is correctly expressed in pHasRA and interacts with Y. pestis or S. marcescens HasA hemophores, this interaction does not lead to heme uptake under the conditions tested with E. coli.

We then investigated whether such a functional interaction could be detected with the whole Y. pestis Has system reconstituted in E. coli. To determine if the Y. pestis heme acquisition system allows cells to use hemoglobin as a source of iron, we introduced pHas98 or pACYCHas (encoding hasRADEB) and pHasRA (encoding hasRA) into an E. coli hemA mutant strain (Table 1). This mutant grows aerobically if supplemented with δ-aminolevulinic acid, but not with exogenously supplied hemoglobin. Growth was tested on iron-rich or iron-depleted agar plates in the presence of hemoglobin at concentrations ranging from 10⁻⁴ to 10⁻⁷ M. No growth was observed under any of the conditions tested, while a control strain carrying the Has_sm system grew at these various hemoglobin concentrations (data not shown). These results indicate that the Y. pestis Has-dependent heme acquisition system is not functional in E. coli and are consistent with the lack of HasA_yp secretion from pHas98- and pACYCHas-harboring cells.

Growth studies of has and has hmu mutant strains of Y. pestis

In Y. pestis, the Hmu system is the primary system that enables the bacteria to use heme and heme-containing compounds as a source of iron (17, 41). However, hmu mutants can still grow in the presence of high concentrations of hemoglobin, suggesting that there may be a second, lower-affinity system that the bacteria can use to obtain iron from hemoglobin. To determine if the Has system is responsible for this activity in hmu mutant strains, we examined the growth of KIM6 (Ybt⁻ Has⁺ Hmu⁺), KIM6–2061.1 (Ybt⁻ Has⁺ Hmu⁻), and KIM6–2085 (Ybt⁻ Has⁻ Hmu⁻) at 28°C in a defined medium containing 50 μM EDDA and different concentrations of hemoglobin (Fig. 7). Similar results were obtained with cells grown at 37°C. KIM6 readily grew in this medium with as little as 1.25 μM hemoglobin (Fig. 7A). However, the growth of KIM6–2061.1 (Fig. 7B) and KIM6–2085 (Fig. 7C) was not significantly stimulated by any tested concentration of hemoglobin. Furthermore, the growth of the hmu mutant was not significantly different from that of the hmu has mutant strain with any of the tested concentrations of hemoglobin. These results suggest that the Has system does not allow the bacteria to use hemoglobin as an iron source under the conditions tested in this study.

FIG. 7.

Growth of Y. pestis KIM6 (has⁺ hmu⁺ ), KIM6–2061.1 (has⁺ hmu mutant), and KIM6–2085 (has hmu double mutant) cells at 28°C in deferrated PMH2 with hemoglobin as the sole iron source. Iron-depleted cells were transferred to PMH2 containing 50 μM EDDA alone (+) or with hemoglobin at concentrations of 1.25 (▵), 2.5 (▴), 5 (○), or 10 (●) μM. These graphs depict one of two separate experiments with similar results.

Virulence testing of Y. pestis has mutants.

Previously, we had shown that a Y. pestis strain (KIM5–2060.21+) with a mutation in the hemin utilization system (Hmu) was fully virulent in mice by a subcutaneous route of infection (41). To determine if the Y. pestis Has system contributes to virulence in mammals, we constructed a has hmu double mutant, KIM6–2081.1+. The virulence plasmid with an ampicillin gene cassette inserted into the yadA′/′yadA pseudogene (pCD1Ap) was electroporated into KIM6–2081.1+, generating KIM5–2081.1+. Y. pestis strains lacking the virulence plasmid are avirulent by any route of infection. Groups of four mice were infected subcutaneously with 10-fold dilutions of bacterial cells grown at 26°C. The LD₅₀s are listed in Table 4. The has hmu mutant was fully virulent by this route of infection, with an LD₅₀ identical to that of the wild-type strain [KIM5(pCD1Ap)+]. In contrast, a strain bearing a mutation in the receptor (Psn) for the siderophore yersiniabactin is completely avirulent. This result suggests that the heme uptake systems Hmu and Has do not play a role in the virulence of Y. pestis for mice, at least by a subcutaneous route of infection.

TABLE 4.

Virulence testing of Y. pestis derivatives in mice infected subcutaneously

Strain	Relevant characteristics	LD50 (CFU)
KIM5(pCDIAp)+	Psn+ Hmu+ Has+ Lcr+	<4.6
KIM5-2081.1+	Psn+ Hmu− Has− Lcr+	<4.2
KIM5-2045.7	Psn− Hmu+ Has+ Lcr+	>5.4 × 106

DISCUSSION

As a flea-mammal parasite, Y. pestis experiences a wide range of environmental conditions in which the ability to scavenge iron and heme contributes to its colonization potential and virulence. Accordingly, many inorganic iron or heme uptake systems in Y. pestis were characterized (Perry et al., submitted). The heme acquisition systems identified in Y. enterolitica (HemRSTUV) and Y. pestis (HmuRSTUV) (17, 38, 39, 41) involve an outer membrane TonB-dependent receptor and a periplasmic binding protein transport apparatus. The availability of Y. pestis genome sequences allowed the identification of a potential second category of heme uptake system relying on type I secretion of an extracellular heme-binding protein hemophore called “HasA” (24, 41; Perry et al., submitted). Here we characterized the Has system of Y. pestis and studied its function in E. coli and Y. pestis.

The comparison of the four Has systems identified so far in S. marcescens, P. aeruginosa, P. fluorescens, and Y. pestis showed that the genetic organization of the Y. pestis Has system is closely related to the homologous system of S. marcescens. It encodes the outer membrane receptor (HasR) and the two inner membrane components (HasD and HasE) of the ABC transporter of the extracellular hemophore HasA. As in the S. marcescens has operon, a gene showing similarities in its deduced amino acid sequence to tonB, hasB, is located downstream of hasE. HasB was shown to be a TonB-like protein specifically dedicated to the HasA-dependent heme acquisition system (29a). A similar function may be associated with the Y. pestis protein HasB. A hasB equivalent was not found associated with the two other Has systems identified so far in P. aeruginosa and P. fluorescens.

Comparisons between Y. pestis HasA_yp and previously characterized hemophores revealed that they display relatively low levels of identity (33, 28, and 20% identity with HasA_sm, HasA_pa, and HasA_pf, respectively). Amino acid sequence comparisons showed that HasA_yp contains a 15-amino-acid insertion near the C terminus compared to HasA_sm (Fig. 2). This insertion was also found in P. aeruginosa and P. fluorescens HasA hemophores.

We showed here, by using a reconstituted secretion system in E. coli, that HasA_yp exhibits secretion and heme-binding properties very similar to those of HasA_sm. However, in vivo and in vitro experiments indicated that HasA_yp does not interact with HasR_sm. This lack of interaction between heterologous pairs of hemophores and the HasR receptor was previously investigated in the case of HasA_pa and HasR_sm, and it was shown that the ability of HasA_pf to interact with HasR_sm depends on the removal of the C-terminal insertion. The additional stretch of 15 amino acids inserted at the C terminus of HasA_pf and HasA_pa was proposed to be responsible for the observed inhibition of heme delivery of HasA_pf or HasA_pa to HasR_sm (1, 24). While it may also be the case for HasA_yp (M. S. Rossi and J. M. Ghigo, unpublished results), these data further support the hypothesis that HasR-HasA couples constitute bipartite receptor entities that closely interact. These interactions may well be highly species specific, preventing eventual detrimental cross talk between hemophore-dependent systems expressed by two different bacterial species. Further characterization of functional HasR receptors may help clarify this point.

We showed that HasA_yp expressed from pHasA_yp can be secreted through a heterologous transporter in complementation experiments with the S. marcescens Has transporter. However, the introduction in E. coli of the whole Y. pestis has region led neither to HasA secretion nor to the reconstitution of a functional heme acquisition system. In particular, secretion studies did not provide evidence that the Y. pestis transporter components HasD_yp and HasE_yp are functional in E. coli. However, the lack of in vivo function in E. coli cannot be explained only by a HasA_yp secretion defect, since a heme-deficient strain expressing HasR_yp and HasA_yp was unable to acquire heme, even at a high hemoglobin concentration at which hasR-alone-dependent heme acquisition should be observed. This suggests either that HasR is not properly expressed at the cell surface or, less trivially, that other Has-specific components are missing in these heterologous experiments. These components may be required either for the proper function of the receptor or for further heme internalization.

Previous studies demonstrated that Has systems are iron regulated, in good agreement with the identification of Fur binding sites (or “Fur boxes”) in these systems (21, 25, 29). Fur boxes are found in the promoter regions of iron-regulated genes and constitute fixation sites for the DNA-binding repressor Fur, which tightly regulates (represses) these genes when associated with ferrous iron as a corepressor. In Y. pestis, such a potential Fur binding site, albeit unusually distant upstream of the start codon for hasR, was also identified. Moreover, the use of a reporter plasmid where the hasRADEB putative promoter region was placed upstream of a promoterless lacZ gene showed that the Y. pestis has operon is active and iron regulated through a Fur-dependent control mechanism. We therefore tested and probed the supernatants of various Y. pestis 6/69 and KIM6 derivatives with anti-HasA_yp antibodies by using cells grown in iron-chelated synthetic and rich media at 28 and 37°C. None of these conditions allowed the detection of an extracellular HasA_yp. Although we provided evidence that the has operon is at least partially expressed in E. coli and that the has promoter is active and iron inducible when introduced in Y. pestis, we cannot formally rule out the possibility that the observed lack of function comes from the absence of expression of some of the components of the Y. pestis Has system. Such situations in which genes potentially involved in Yersinia environmental fitness are nonfunctional in Y. pestis have been reported (34). However, neither a frameshift mutation nor an obvious deletion was seen in the genes of the hasRADEB Y. pestis putative operon.

Previous studies suggested that, in addition to the Hmu system, there may be a second, lower-affinity system that Y. pestis can use to obtain iron from hemoglobin. However, the growth study presented here suggests that the Has system does not allow the bacteria to use hemoglobin as an iron source. The results obtained with the ΔhmuP′RSTUV mutant differ from those of earlier studies in which a graded response to increasing concentrations of hemoglobin was observed (41). Previous experiments used Y. pestis strains that produced the siderophore yersiniabactin (Ybt). Thus, the previous results with the ΔhmuP′RSTUV mutant were probably due to Ybt-mediated acquisition of iron from either hemoglobin degradation products or the EDDA-Fe complex, and not to the existence of a functional low-affinity heme transport system.

The recognition that an intact functional yersiniabactin system is a prerequisite for mouse virulence in the early stages of infection emphasized the fact that iron acquisition is critical for the progression of Y. pestis infections (30a). Our study, with a has hmu double mutant, constitutes the first investigation of the contribution of a hemophore-dependent system to bacterial pathogenicity. Our results with subcutaneous infection in mice suggest that the Has heme uptake system, as shown for the Hmu system, does not play a role in Y. pestis virulence in the mouse model of bubonic plague. Further experiments will be required to determine whether other routes of infection or infection in other animal hosts rely on Hmu and/or Has heme acquisition systems.

Taken together, our results suggest that a plausible reason for the absence of a HasA-related phenotype in Y. pestis may be that induction of the Has system was not appropriate in our experimental model and under the conditions used. While our data indicate that iron starvation participates in the regulation of the Has system, other factors may have to be considered. Comparison between HasR homologues revealed that HasR_yp lacks an N-terminal extension present in HasR_sm and other hemophore-dependent receptors (Fig. 2A) (data not shown). In E. coli, the amino terminus of the FecA outer membrane receptor is involved in regulating the Fec iron-dicitrate acquisition system (4, 19). Thus, regulation of the Has_yp system may be different from those of other Has systems and could require other factors. We noted the presence of strong stem-loop structures in the hasRADEB promoter region. Moreover, we showed that this promoter displays an increased hasRADEB promoter activity at 37°C compared to 26°C. This suggests that expression of the Has system may be temperature responsive, as shown to be the case for several bacterial virulence genes (20).

This study suggests that two distinct heme acquisition systems may coexist in Y. pestis and in other Yersinia species, supporting the hypothesis that hemophore-dependent heme acquisition systems are widespread in members of the family Enterobacteriaceae. While some of these systems have now been characterized at the structural and molecular level, their role in environmental survival and growth of gram-negative bacteria still awaits conclusive experimental demonstration.