Hal Is a Bacillus anthracis Heme Acquisition Protein

Miriam A. Balderas,a Christopher L. Nobles,a Erin S. Honsa,a Embriette R. Alicki,b and Anthony W. Maressoa

Department of Molecular Virology and Microbiology, Baylor College of Medicine, Houston, Texas, USA,a and Interdepartmental Program in Cell and Molecular Biology,Baylor College of Medicine, Houston, Texas, USAb

The metal iron is a limiting nutrient for bacteria during infection. Bacillus anthracis, the causative agent of anthrax and a poten-tial weapon of bioterrorism, grows rapidly in mammalian hosts, which suggests that it efficiently attains iron during infection.Recent studies have uncovered both heme (isd) and siderophore-mediated (asb) iron transport pathways in this pathogen.Whereas deletion of the asb genes results in reduced virulence, the loss of three surface components from isd had no effect,thereby leaving open the question of what additional factors in B. anthracis are responsible for iron uptake from the most abun-dant iron source for mammals, heme. Here, we describe the first functional characterization of bas0520, a gene recently impli-cated in anthrax disease progression. bas0520 encodes a single near-iron transporter (NEAT) domain and several leucine-richrepeats. The NEAT domain binds heme, despite lacking a stabilizing tyrosine common to the NEAT superfamily of hemopro-teins. The NEAT domain also binds hemoglobin and can acquire heme from hemoglobin in solution. Finally, deletion of bas0520resulted in bacilli unable to grow efficiently on heme or hemoglobin as an iron source and yielded the most significant phenotyperelative to that for other putative heme uptake systems, a result that suggests that this protein plays a prominent role in the repli-cation of B. anthracis in hematogenous environments. Thus, we have assigned the name of Hal (heme-acquisition leucine-richrepeat protein) to BAS0520. These studies advance our understanding of heme acquisition by this dangerous pathogen and jus-tify efforts to determine the mechanistic function of this novel protein for vaccine or inhibitor development.

The extracellular concentration of free iron in mammals is keptvery low (�10�18 M) (7, 9, 10, 43) as a means to prevent the

formation of insoluble aggregates and metal-induced free radicals(51). Iron concentrations are regulated in the body by high-affin-ity iron-binding proteins such as transferrin and ferritin, whichtransport and store iron, respectively (1, 43, 51, 52). However,approximately 80% of the host iron pool is contained in heme(iron protoporphyrin IX) (8), which is mostly bound to hemoglo-bin, the major oxygen carrier protein in vertebrates (42, 47). Thereare two principal systems by which bacteria satisfy their need foriron. One mechanism is to secrete siderophores, small moleculesthat bind ferric iron through high-affinity interactions (43, 51).This mechanism is unlikely to account for iron import from hemesources owing to the strict sequestration of the iron in and the lowoff rates from heme and hemoproteins. In this regard, bacteriahave evolved secreted or cell surface proteins that bind and ac-quire either free heme or heme from host hemoglobin (43, 51).The current paradigm is that imported iron is used as a cofactorfor DNA synthesis and energy production, thereby promotingbacterial infection and replication in infected hosts (20, 51). Thus,the assimilation of iron from host heme is considered to be animportant step in the maintenance and persistence of a bacterialinfection.

Bacillus anthracis, a Gram-positive spore-forming bacteriumand the causative agent of anthrax, grows to high densities duringinfection of mammalian hosts, implying that this pathogen con-tains efficient and versatile iron-scavenging mechanisms. In 2004,Cendrowski et al. reported the identification of two siderophoresystems in this pathogen, asb (anthracis siderophore biosynthesis)and bac (Bacillus anthracis catechol) (13). Functional loss of asbbut not bac reduced the growth of B. anthracis in iron-depletedmedium and virulence in an inhalational mouse model of anthraxdisease. This defect was traced to an inability of the asb mutantspores to outgrow inside macrophages (13), leaving open the

question of how bacillus attains needed iron when it reaches bloodor host tissues. In 2006, Maresso et al. identified an eight-genesystem with similarities to the then recently described isd (iron-regulated surface determinant) locus, with one component, IsdC,being a heme-binding protein and important for bacillus growthon heme (34). However, the deletion of isdC did not attenuatevirulence in guinea pigs, which suggested that Isd may be dispens-able for anthrax disease progression (23). During a search forgenes that are surface anchored by transpeptidase sortases, Gasparet al. annotated that the B. anthracis Sterne strain contained a gene(bas0520) with an LGATG sorting signal (21). This gene is pre-dicted to encode an N-terminal near-iron transporter (NEAT)domain, several internal leucine-rich repeat (LRR) regions, and aC-terminal sortase-like cell wall anchor (Fig. 1A). Analysis of thisNEAT domain revealed sequence similarity to the NEAT domainsof isdC (21%), isdX1 (29%), and bslK (49%) and the 5 NEATdomains of isdX2 (�29%) (this work). Furthermore, a recentstudy by Carlson at al identified the Ames homolog of bas0520(gbaa0552) as being highly upregulated under conditions of ironstarvation, and deletion of the gene resulted in an approximately110-fold increase in the 50% lethal dose of B. anthracis Sterne 34F2in an inhalation model of anthrax disease (11). Hypothesizing thatBAS0520 is the missing factor needed for heme assimilation dur-ing anthrax disease, we demonstrate here that BAS0520 is a heme-binding protein that takes heme from mammalian hemoglobinand is necessary for growth on heme and hemoglobin in low-iron

Received 23 April 2012 Accepted 25 July 2012

Published ahead of print 3 August 2012

Address correspondence to Anthony W. Maresso, maresso@bcm.edu.

Copyright © 2012, American Society for Microbiology. All Rights Reserved.

doi:10.1128/JB.00685-12

October 2012 Volume 194 Number 20 Journal of Bacteriology p. 5513–5521 jb.asm.org 5513

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

environments. With these new findings, we have appropriatelyassigned the name of Hal (heme-acquisition leucine-rich repeatprotein) to BAS0520 and propose efforts to examine whether Halis suited for vaccine or drug development.

MATERIALS AND METHODSBacterial strains. Escherichia coli strains (DH5 and BL21) were used forthe cloning and amplification of hal and were grown in Luria broth (LB)supplemented with 50 �g/ml ampicillin (Fisher Scientific, Pittsburgh,PA). Wild-type B. anthracis strain Sterne (34F2) (46) was used for thegrowth studies and to generate a complete deletion in hal. For this pur-pose, hal was deleted by allelic replacement using the temperature-sensi-tive plasmid pLM4 (36). Briefly, genomic DNA of B. anthracis strainSterne 34F2 was extracted using a Wizard DNA purification kit (Promega,Madison, WI), and 1,000 bp of the 5= and 3= sequences flanking the halgene was PCR amplified with primer pairs hal-SmaI (5=-CCCGGGGCAAGGAAATGGTGACAGCAAAGTTGG-3=) and hal-KpnI (5=-GGTACCC

GAGATATGATTACATTTAG-3=) as well as hal-KpnI (5=-GGTACCGGTGTAAATATATTAGAGAAGGAGG-3=) and hal-SacI (5=-GAGCTCCCCATTCATATGAAAATGATATTTCATATG-3=), respectively. Followingligation of the two fragments at the KpnI site, the 2-kb insert was clonedbetween the SmaI/SacI sites of pLM4 and plasmid DNA was amplified inthe dam mutant E. coli strain K1077 prior to electroporation into B. an-thracis. After transformation into strain Sterne, bacilli were first grown at30°C (permissive temperature) on LB (20 �g/ml kanamycin) and thenshifted to 43°C (restrictive temperature), followed by growth at 30°C toinduce plasmid loss, thereby generating a B. anthracis �hal strain. Bacteriawere examined for kanamycin resistance by plating on LB agar, and DNAsequencing was performed to verify the presence or absence of wild-typeand mutant allele nucleic acid sequences. Isogenic deletions in bslK and alleight genes of the isd system were also constructed using this procedure.Primer pairs 5= flank (bslK-SmaI [5=-CGTTTTGACGTTATCGTTTCAG-3=] and bslK-KpnI [5=-GAGCGGAAAGCGTACTTATG-3=]) and 3= flank(bslK-KpnI [5=-GAGGCGCACAACATTCAC-3=] and bslK-SacI [5=-GTT

FIG 1 Functional annotation of Hal. (A) Hal was analyzed using a combination of PROSITE, SignalP, and KEGG. The arrowhead indicates the predicted site ofsignal peptidase cleavage. Hal is annotated as having a NEAT domain (residues 29 to 152, gray highlighted), 16 leucine-rich repeats (bold; labeled 1 through 16),and a C-terminal Gram-positive bacterium anchor (residues 1028 to 1070; underlined). (B) The NEAT domain of Hal was aligned with the eight NEAT domainsof B. anthracis, the NEAT domain of IlsA, a protein involved in iron uptake in B. cereus, and Shr from Streptococcus pyogenes (14, 39). Hal contains two regionscommon to all NEAT domains, an N-terminal 310 helix of 4 to 6 amino acids and a C-terminal � hairpin (bracketed regions). It is notable that, unlikeheme-binding NEAT domains, Hal lacks a conserved tyrosine residue (instead, there is a phenylalanine) in the hairpin that is known to stabilize the interactionwith the heme iron (hairpin, bold residues).

Balderas et al.

5514 jb.asm.org Journal of Bacteriology

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

TTTACATTACCGAAACCG-3=]) were used for bslK deletion. For the isddeletion, the primers were 5= flank (isd-SmaI [5=-GGTCTACTACTTGTGTATTTAG-3=] and isd-KpnI [5=-GCTAATTAAATAATGGGTAGAAG-3=]) and 3= flank (isd-KpnI [5=-CGCATGTACGACTAACCTTCC-3=]and isd-SacI [5=-CCATGACGATCGTCAATCCATG-3=]).

To construct the complementation strain, the full-length hal gene wasPCR amplified from the genomic DNA of B. anthracis strain Sterne 34F2using forward (5=-GATCGATCGTCGACCATTTAGATTTATATATTTTGGAGG-3=) and reverse (5=-GATCGATCGGTACCCTAGTGGTGATGGTGATGATGCCTCTCCTCCTTCTCTAATA-3=) primer pairs with en-gineered SalI and SphI restriction enzyme sites, respectively. The PCRproduct was digested and cloned into a B. anthracis and E. coli shuttlevector, pUTE657 (27), such that expression was under the control of theisopropyl-�-D-thiogalactopyranoside-inducible hyper-spac promoter,generating phal. Ligation products were transformed into E. coli DH5�and selected on 100-�g/ml ampicillin agar plates. Plasmid DNA was thentransformed into E. coli K1077 (with dam and dcm mutations) and thenelectroporated into B. anthracis �hal strain to create B. anthracis �halphal.

Protein purification. The NEAT domain (amino acids 29 to 152) ofHal (HalN) was purified as a glutathione S-transferase (GST) fusion pro-tein similar to that described for other NEAT-domain proteins of B. an-thracis (16, 19, 28, 34, 35, 48). Genomic DNA of B. anthracis strain Sterne34F2 and primer pair halN-BamHI forward (5=-GATCGATCGGATCCGAGAATATGGCTGTACAAAGTC-3=) and halN-BamHI reverse (5=-GTACGTACGGATCCTTACTACAAATGTACGCATATTCAC-3=) were usedto amplify DNA encoding the NEAT domain. DNA was then cloned intothe BamHI restriction site of pGEX2TK to create pgst-HalN, a vector withinducible expression of HalN as a GST fusion protein containing a throm-bin cleavage site. pgst-HalN was then transformed into E. coli BL21 andgrown in 100 ml of LB plus ampicillin at 37°C. Overnight cultures werethen seeded into 1.5 liters of LB with ampicillin at 37°C and rotated at 250rpm for 2 h, at which point isopropyl �-D-thiogalactopyranoside (finalconcentration, 1.5 mM; Sigma, St. Louis, MO) was added to the culture,which was incubated for an additional 2 h. Cells were next sedimented bycentrifugation at 6,000 � g for 10 min and resuspended in phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM sodium phos-phate dibasic, 2 mM potassium phosphate monobasic, pH 7.4), and thecells were lysed by French press. The lysate was then centrifuged at 14,000 � gfor 15 min, and the supernatant was sterilized with a 0.22-�m-pore sizecellulose filter. The filtrate was then subjected to affinity chromatographyusing 2 ml of glutathione-Sepharose resin (GE Healthcare, Piscataway,NJ) that was preequilibrated with 20 ml of PBS. Bound fusion proteinGST-HalN was washed with 40 ml of PBS, GST-free HalN was releasedwith the addition of 50 units of thrombin (Calbiochem, Darmstadt, Ger-many), and thrombin was removed from the HalN preparations usingaminobenzamidine resin (Sigma). Prior to determining the binding prop-erties of the NEAT domain, endogenous iron-porphyrin was removed bymethyl ethyl ketone (MEK) treatment to generate apoprotein as describedpreviously (49). The concentration of HalN was determined by the bicin-choninic acid assay (Pierce, Rockford, IL) or the Bradford assay (Bio-Rad,Hercules, CA). Typical yields using this purification method are approx-imately 4 mg/liter of E. coli culture. All protein preparations were stored at�20°C.

Heme binding to HalN. HalN (10 �M) was incubated with (0.25 to 5.0�M) or without hemin chloride in PBS, pH 7.4, for 30 min at 25°C, andincubation was followed by a spectral analysis at from 250 to 650 nm usinga DU800 spectrophotometer (Beckman-Coulter, London, United King-dom).

Heme acquisition from hemoglobin. GST-HalN (7.5 �M) was im-mobilized on 2 ml of glutathione-Sepharose resin, washed with 40 ml ofPBS, and incubated with 1 ml of bovine methemoglobin (2.5 �M, Sigma)for 30 min at 25°C. Control columns consisted of a protein-only control(incubated with PBS) and another treated with hemoglobin with no pro-tein immobilized. After 30 min, the columns were drained (hemoglobin

fraction), washed 2 times with 10 ml of PBS, and treated with 0.5 ml ofreduced glutathione (25 mM) for 5 min to elute bound GST-HalN. Awavelength scan from 250 to 650 nm was used to assess the relativeamount of heme in each sample. Proteins samples were analyzed by SDS-PAGE and silver stain (Bio-Rad silver kit).

Association with hemoglobin. The interaction between HalN withmethemoglobin was determined by surface plasmon resonance (SPR) us-ing a BIAcore 3000 biosensor (Amersham Biosciences). Holo-bovinemethemoglobin (in 50 mM Tris-HCl, pH 7.0) was covalently coupled to aCM5 sensor chip at 25°C to a density of 3,600 response units (RUs) usingamine chemistry as previously described (30, 38). HalN (50 to 200 �M) inHBS-N (0.01 M HEPES, 0.15 M NaCl, pH 7.4) was injected at 20 �l/minfor 300 s at 25°C. Binding data were attained using double referencing,where parallel injections of analyte are flowed over a CM5 surface thatlacks hemoglobin (baseline) and these readings are subtracted from theexperimental (HalN over hemoglobin) signal. Data were analyzed usingBIAevaluation (version 4.1) software (Amersham Biosciences) after fit-ting the data to a 1:1 Langmuir binding model (15, 30, 38). Each concen-tration of HalN was injected twice, and each experiment was performed intriplicate.

Growth studies. RPMI (Life Technologies, Grand Island, NY) waschelated of iron by incubation with 2.5 g of Chelex-100 per 100 ml ofmedium for 12 h at 4°C (cRPMI). B. anthracis Sterne 34F2 wild-type,�hal, �bslK, and �isd (all containing empty plasmid pUTE657) strainsand the complemented mutant (B. anthracis �hal phal) were subculturedin LB overnight for 12 h at 37°C, and the bacterial pellets were washed withcRPMI and inoculated into cRPMI for 12 h at 37°C. Ten microliters of thisculture at an optical density of 1.0 was then inoculated into 4 ml of freshcRPMI with or without hemin (1, 10 �M) or hemoglobin (10, 100 �M),and growth (absorbance 600 nm) at 37°C was monitored from 2 to 16 h.

RESULTSHal annotation and NEAT-domain purification. The gene forhal is not localized to the isd locus and thus may code for a proteinthat is functionally separate from the Isd system. To gain insightinto the function of hal, the amino acid sequence for this proteinwas analyzed using a combination of PROSITE, SignalP, andKEGG domain annotation programs (31, 40, 45) (Fig. 1A). Theanalysis revealed the presence of a putative N-terminal signal pep-tide (amino acids 1 to 21) and a proposed C-terminal Gram-pos-itive bacterium anchor (GPA) region (amino acids 1028 to 1070).These two findings, along with the bioinformatic identification ofHal as a surface protein (the gene is designated BA0552 in theAmes strain of B. anthracis) (21), suggest that Hal is secretedthrough the general secretory pathway and tethered to the surfacepeptidoglycan, most likely by covalent attachment by a sortasetranspeptidase. Furthermore, between the signal peptide and theGPA are 16 LRRs, with each repeat containing 21 amino acids.There are two clusters of these repeats; the first cluster contains 9repeats and the second cluster contains 7, with each cluster sepa-rated by 171 amino acids. The exact function of bacterial LRRs isstill unresolved, but several reports link them to protein-proteininteractions (32, 33). The NEAT domain of Hal also contains tworegions common to all NEAT domains, an N-terminal 310 helix of4 to 6 amino acids and a C-terminal � hairpin consisting of 6 to 8residues (Fig. 1B, bracketed regions). However, unlike heme-binding NEAT domains, Hal lacks a conserved tyrosine residue(instead, it possesses a phenylalanine) in the hairpin that is knownto stabilize the interaction with the heme iron (Fig. 1B, bold resi-due in hairpin region), thus drawing into question whether Hal isa functional heme-binding protein. Consistent with its observedexpression under low-iron conditions (11), a putative 19-residueFur box (12, 26) is located 112 bp upstream of the transcriptional

Function of Hal

October 2012 Volume 194 Number 20 jb.asm.org 5515

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

start site of hal. Since Fur is considered to be the primary repressorof genes involved in iron uptake, the presence of this Fur boxbefore hal (and related genes) signifies regulation by iron avail-ability (Table 1).

Hal is a heme-binding protein. We reasoned that the mostlogical way to begin to deduce the function of Hal was to assess ifit is involved in heme acquisition. As the first step in this process,we cloned the DNA encoding the putative NEAT domain of hal(HalN) into pGEX2TK to create a fusion to GST. RecombinantGST-HalN was expressed in E. coli and isolated using GST affinitychromatography. Following removal of the GST tag, pure HalNwas attained and used for subsequent functional studies (Fig. 2A).

To determine if the predicted HalN can associate with heme,the spectral properties of HalN were measured after purificationfrom E. coli. Figure 2B (0 �M heme) shows the presence of a smallpeak of absorbance at 403 nm (Soret band), a diagnostic featureassociated with the presence of iron-porphyrin in heme-bindingproteins, as described previously (16, 19, 28, 34, 35, 48). Thissuggests that some HalN copurifies with heme during its purifica-tion from E. coli. To further test heme-binding function, the co-purifying heme in the HalN preparation was removed with acidtreatment, followed by methyl ethyl ketone treatment, and theapoprotein was titrated with hemin. As observed in Fig. 2B, theaddition of hemin to HalN resulted in a clear absorbance increaseand a red shift in the Soret band compared to reaction mixturescontaining only hemin, which consistently showed a broad peak at380 nm (compare the black to the gray bars). Further titration ofhemin resulted in a gradual movement of the peak to the absor-bance maximum of the control, an indication that HalN is satu-rated and consistent free hemin now dominates the absorbancespectrum. These data suggest that the NEAT domain of Hal iscapable of binding heme, despite lacking a stabilizing tyrosinecharacteristic of all previously studied heme-binding NEAT do-mains.

Hal can acquire heme from hemoglobin. Erythrocytes ac-count for a large amount (85%) of the organismal heme (42), andhemoglobin is the most abundant iron reservoir (approximately 3g out of the 4 g total host iron pool) (8, 50); thus, circulatinghemoglobin may provide a source of heme for Hal during B. an-thracis infection. To test this hypothesis, we coupled GST-HalN toglutathione-Sepharose and incubated the complex with or with-out bovine hemoglobin. After 30 min, GST-HalN and hemoglobinfractions were separated and heme content was assessed by Soret

spectroscopy. As observed in Fig. 3A, GST-HalN incubated withhemoglobin yielded a higher Soret absorbance than GST-HalNincubated with a buffer control. Performing this experiment intriplicate and quantifying the ratio of the absorbance at 403 nm(heme) to that at 280 nm (GST-HalN) indicated that the differencein heme was approximately 3-fold between each group, suggestiveof significant heme transfer to GST-HalN (Fig. 3B). As analyzed bySDS-PAGE and silver stain analysis, this effect was not due tohemoglobin contamination of the GST-HalN fraction (Fig. 3C;compare the migration of the eluted proteins to that of the hemo-globin standard). Taken together, these data indicate the NEATdomain of Hal is sufficient in the extraction of heme from hemo-globin.

A Hal-hemoglobin interaction may facilitate heme removal.To determine if the observed heme transfer from hemoglobin to theNEAT domain of Hal is possibly mediated by a protein-protein in-teraction mechanism (active process), we assessed the potential bind-ing of HalN to hemoglobin using surface plasmon resonance spec-troscopy. Hemoglobin was coupled to a carboxymethyl surface, andincreasing amounts of recombinant apo-HalN were infused over thesurface. After fitting the data to a 1:1 binding model (30, 38), as ob-served in Fig. 3D, dose-dependent increases in the response units, ameasure of physical association, were observed for chambers con-taining HalN and hemoglobin. These results suggest that the NEATdomain of Hal directly engages hemoglobin, a process that possiblymediates heme acquisition from hemoglobin.

Loss of hal compromises growth on heme and hemoglobin.The analysis of the NEAT domain of Hal indicates that Hal is aheme-binding protein that actively acquires heme from host he-moglobin. To determine whether this protein plays a functionalrole in enhancing the growth of B. anthracis in low-iron, high-heme environments, we generated an isogenic strain with aknockout in hal (�hal) and tested this strain for growth on hemeand hemoglobin relative to that of a wild-type strain and twoadditional knockouts: one that harbors a complete deletion in alleight genes of the isd-like locus (�isd) and one that lacks the genecoding for bslK, the only other annotated NEAT protein in B.anthracis (�bslK). BslK is a surface protein that binds heme with a

TABLE 1 Fur box sequences of hal, bslK, and isda

Gene Fur boxDistance fromATG (bp)

Consensus GATAATGATAATCATTATChal GACAATGATAATGAAAATC �112bslK GACATTGATAATCATTATC �72isd locus 1 GAGAATGATAATCAATAAA �221isd locus 2 GAGAATGATAATCAATAAT �127isd locus 3 GAGAATAATTATCATTTTC �47isd locus 4 TAAAATGAGAATAATTATC �55isdG locus 1 GATAATGATTTTCATTATC �56isdG locus 2 GACAATGATAATGATTTTC �62a The similarity of each putative Fur box for hal, bslK, and isd with a consensus Fur boxsequence was determined. Deviations from the consensus sequence are shaded gray,and the distance of the box from the transcriptional start site is noted.

FIG 2 Hal is a heme-binding protein. (A) HalN was expressed in E. coli as aGST fusion protein and purified by affinity chromatography. The SDS-poly-acrylamide gel shows the fusion and GST-free (arrow) forms of HalN. (B) HalN(10 �M) from panel A was incubated with heme (1 to 10 �M; black lines), andthe absorbance from 200 to 450 nm was recorded and compared to that forheme-only controls for all concentrations used (gray lines). About 10% ofpurified HalN copurified with bound heme (see the results for 0 �M), as de-termined using the pyridine hemochrome method (3). The data are represen-tative of one of three independent determinations.

Balderas et al.

5516 jb.asm.org Journal of Bacteriology

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

very high affinity and whose exact role in B. anthracis iron homeo-stasis is not known. At low concentrations of heme (1.0 �M) andhemoglobin (10 �M), there were observable differences in theextent of growth between the wild-type and �hal strain; however,these differences were not significant (Fig. 4C and D). When theconcentrations of heme (10 �M) and hemoglobin (100 �M) wereincreased, significant differences in the growth of �hal comparedto that of wild-type B. anthracis were observed, with the �halstrain growing poorly on both iron sources (Fig. 4A and B). Eachstrain grew equally well when grown in nutrient-rich LB (Fig. 4E),and the growth defect on heme or hemoglobin was partially orfully restored, respectively, by the plasmid expression of full-length hal in the knockout strain. Together, these results suggestthat the absence of hal leads to a growth defect of B. anthracisSterne when heme or hemoglobin is the only iron source. Inter-estingly, the �hal strain showed a higher growth defect than boththe �isd and �bslK strains on heme or hemoglobin, a result thatsuggests that Hal may play a more prominent or unique role iniron uptake from heme iron than these other NEAT-containingproteins or systems. These data also highlight the importance of

further studying Hal and uncovering its full function in B. anthra-cis growth and heme acquisition. In agreement with the biochem-ical results presented in Fig. 2 to 4, these data provide the firstevidence that Hal functions as a bacillus hemoprotein that specif-ically acquires heme from host hemoglobin.

DISCUSSION

Here we report the first functional characterization of bas0520,hereby named hal (heme-acquisition leucine-rich repeat protein),a gene important for inhalational anthrax and upregulated underlow-iron conditions (11). Our findings suggest that (i) the NEATdomain of Hal (HalN) can bind heme, despite lacking a stabilizingtyrosine, (ii) HalN can acquire heme from hemoglobin, (iii) aphysical complex forms between HalN and methemoglobin, and(iv) Hal is important for growth on heme and hemoglobin, seem-ingly more so than Isd or BslK, which may explain the markeddefect in virulence (in comparison to that of wild-type infection)observed for an inhalational anthrax murine model using a Sternestrain lacking this protein (11).

It is now recognized that Gram-positive pathogenic bacteriautilize secreted or cell surface proteins containing NEAT domainsto mediate the acquisition and import of heme (24, 25, 35, 37). B.anthracis contains five genes that harbor one or more NEAT do-mains (23, 28, 34, 35, 48). Three of these (isdX1, isdX2, and isdC)are part of the Isd locus, an array of eight genes proposed to en-code a system that acquires heme from hemoglobin and deliverscaptured heme to the cell surface and across the cell membraneinto the bacterial cytosol (19, 29, 34, 35). IsdX1 and IsdX2 aresecreted proteins that extract heme from hemoglobin and deliverthe heme to cell wall-bound IsdC (23, 29, 35). These are highlyantigenic proteins, a finding that suggests an important role inheme-iron uptake during anthrax disease (22). Interestingly, atriple mutant lacking all three genes encoding the Isd proteinsIsdX1, IsdX2, and IsdC was not reduced in virulence, as assessedby a guinea pig model of infection (23). The biochemical study ofIsdX1 and IsdX2, however, has led to important insights into howbacillus NEAT-domain proteins mechanistically function, includ-ing the identification of NEAT-hemoglobin and NEAT-NEATcomplexes (for transfer) and amino acids that facilitate heme andhemoglobin association (16, 29). Comparison of the properties ofthe NEAT domains of IsdX1 and IsdX2 to those of Hal yieldsinteresting observations when considering the mechanism ofheme coordination. For example, Hal has a phenylalanine in thefifth position of the heme-binding motif (YDKEF in Hal), a five-residue stretch of amino acids common to all proteins withNEATs (Fig. 1). For heme-binding NEAT domains, the first ty-rosine in this region serves as the sixth axial ligand for the heminiron (29). The second tyrosine hydrogen bonds to the first, aninteraction that stabilizes the coordination (25, 41, 44). Both ty-rosines are essential for heme stabilization in the NEAT domain,and all the NEAT-containing proteins from B. anthracis that con-tain the second tyrosine efficiently bind heme (25, 29, 41, 44). Forexample, the second NEAT domain of IsdX2 lacks the secondtyrosine and consequently does not associate with heme (28, 29).We modeled the structure (Fig. 5) of the Hal NEAT domain usingour published structure of IsdX1 (Protein Data Bank [PDB] ac-cession number 3SZ6) as a template. The IsdX1 model demon-strated that tyrosine 112, conserved in all bacillus NEAT-contain-ing proteins and required for heme association, is the sixth axialligand (the iron-to-hydroxyl distance is 2.2 Å). Interestingly, phe-

FIG 3 The NEAT domain of Hal can acquire heme from methemoglobin. (A)GST-HalN (7.5 �M) was incubated with or without bovine methemoglobin(2.5 �M) for 30 min at 25°C, and proteins were separated by affinity chroma-tography using glutathione-Sepharose. Hb, hemoglobin. (B) Heme occupancywas calculated by taking the ratio of the Soret absorbance (heme, 403 nm) tothe protein absorbance (280 nm) for three independent experiments. (C) Tenmicroliters of each elution was applied to an SDS-polyacrylamide gel, andGST-HalN or hemoglobin was detected by silver stain. Lane M, molecular massmarkers; lane ST, standard representing an aliquot of methemoglobin to pro-vide a reference for hemoglobin’s mobility upon SDS-PAGE. (D) Apo-HalN(50, 100, 200 nM) was infused over methemoglobin coupled to a carboxy-methyl chip, and response units were measured for 300 s, followed by bufferinfusion to monitor dissociation. Each concentration of HalN was injectedtwice, and each experiment was performed in triplicate. One representativeexperiment is shown.

Function of Hal

October 2012 Volume 194 Number 20 jb.asm.org 5517

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

nylalanine 116 is within 5 Å of this tyrosine, suggesting that thearomatic ring of F116 may function in a comparable way as atyrosine aromatic ring, as found in other similar NEAT proteins(i.e., to stabilize the coordinating hydroxyl). Interestingly, there isa tyrosine on the opposite side of the heme-binding pocket in the310 lip helix (ESYAT). Ekworomadu et al. observed this region inIsdX1 to regulate heme and hemoglobin binding (16), and othershave proposed that the lip helix is important due to its locationover the heme-binding pocket (41, 44). In Hal, the tyrosine resi-due in this region is in close proximity to the heme and extendsabove the distal side of the heme iron. A bistyrosine linkage, oneon each side of the heme face, would be a novel finding for abacterial protein with a NEAT domain. The binding of heme byHal provides evidence that this rule is not strictly observed, mean-ing that the heme-iron association of this protein is likely differentthan that of other NEAT-containing proteins. The model alsoconfirmed the presence of the lip-helix tyrosine directly over theheme iron. However, the distance between the lip-helix tyrosinehydroxyl and the heme iron is approximately 5 Å, seemingly too

far to act as an iron ligand. Additional studies are required todetermine the role of this tyrosine in heme binding and transfer;however, it is a possibility that the lip-helix tyrosine interacts withthe heme porphyrin itself, strengthening the interaction betweenthe heme and NEAT.

A previous study by Tarlovsky et al. discovered and character-ized the B. anthracis protein BslK as possessing a single NEATdomain that localizes noncovalently to the surface of B. anthracis,most likely via the three S-layer homology (SLH) regions (48).BslK was shown to bind heme with high affinity and transfersheme to IsdC. The role of this protein in B. anthracis pathogenesisor heme uptake has not yet been tested. However, it is clear thatdeletion of bslK does not render B. anthracis unable to grow onheme or hemoglobin (Fig. 4). Interestingly, the �bslK strain grewbetter than the isogenic wild-type strain on hemoglobin (andheme at later time points; Fig. 4). This may be partially explainedby considering the high affinity of BslK for heme (almost no hemedissociation is observed in a 24-h period) (48). In the absence ofbslK, Hal and Isd systems have access to a larger heme pool, thus

FIG 4 B. anthracis lacking hal grows poorly on heme and hemoglobin as the only iron source. Wild-type, �hal, �hal plus pUTE657hal (designated phal), �bslK,or �isd B. anthracis strains were grown on hemin (10 �M, 1 �M) (A, C) or hemoglobin (100 �M, 10 �M) (B, D) in iron-chelated RPMI or LB (E) at 37°C, andgrowth was monitored by measuring the optical density at 600 nm (O.D.600) for 2 to 8, 10, 12, or 16 h. Each wild-type strain also contained the empty pUTE657vector as a control.

Balderas et al.

5518 jb.asm.org Journal of Bacteriology

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

possibly allowing these two systems to function more efficiently.This hypothesis is currently being tested.

Gaspar et al. searched the sequenced genome of B. anthracis forputative sortase-anchored proteins and described bas0520 as cod-ing for a protein with a proposed GPA (21). In fact, Hal containstwo of the three features that are common for cell wall-attachedproteins, including a hydrophobic and polybasic stretch of resi-dues at the C terminus. However, a conserved putative prolineobserved in the second position of the canonical sortase A penta-peptide recognition motif (LPXTG) is missing in the sequence forHal (the closest resemblance is LGATG). More recently, Carlsonet al. noted that hal is upregulated under low-iron conditions andfound the Sterne strain lacking this gene to demonstrate an ap-proximately 100-fold reduction in virulence in a mouse model ofinhalational anthrax (11). The putative GPA motif, the proposedimportance of hal in virulence, the presence of a heme-bindingNEAT domain, and the above-mentioned results all suggest thatHal may be an important mediator of iron uptake from bloodsources. Indeed, findings in this report include data that supportthis hypothesis and possibly provide a link between anthrax dis-ease progression and heme assimilation by this protein from he-moglobin.

In addition to a NEAT domain, Hal contains 16 LLR se-quences. The presence of evenly spaced repeating units of leucineallows a protein or domain to fold into a horseshoe shape with thehydrophilic residues exposed to solvent (32, 33). The internal leu-cines create a hydrophobic core that may enhance interactionswith proteins. LRRs are found in evolutionarily diverse proteinsharboring various numbers of copies of this structural motif rang-ing from 1 to 30 (32, 33). Functional attributes of LRRs include the

modulation of cell adhesion and signal transduction, especially ineukaryotic cells. However, bacterial LRRs have not been exten-sively studied. A well-noted Gram-positive bacterial LRR proteinis Listeria monocytogenes internalin (InlA), a surface protein thatbinds the host receptor E cadherin, a process that mediatesthe invasion of this pathogen into epithelial cells, which suggeststhe importance of the LRR region in the invasion process (4). TheStreptococcus pyogenes protein Slr, which also contains severalLRRs, binds to type 1 collagen, thereby showing that LRRs partic-ipate in bacterial adhesion to tissue components (5). LRR regionsare also associated with bacterial toxins or effectors (e.g., the typeIII secreted protein YopM from Yersinia pestis) (18). These studiessuggest that at least some fraction of LRR-containing proteins inbacteria promote the interaction of bacteria with host tissues, aprocess that likely contributes to pathogenesis. Interestingly, al-though LRRs may function to drive protein-protein interactions,work here suggests that Hal’s NEAT domain alone is sufficient forthe interaction with hemoglobin. This interaction may drive hemetransfer, although proof of this would require knowledge of therates of heme loss from hemoglobin and transfer to Hal. We alsocannot rule out the possibility that the LRRs strengthen this asso-ciation or allow binding to other serum hemoproteins, such as thehemoglobin-haptoglobin complex, hemopexin, or other iron-binding proteins (14, 41). Indeed, the IslA protein from Bacilluscereus, which contains a NEAT domain and several LRRs, bindsheme, hemoglobin, and ferritin (14). We are currently evaluatingthe role of the LRRs in the heme acquisition functions of Hal.

One question raised by this study is why B. anthracis wouldrequire three heme uptake systems that evolved separately (Isd,BslK, and Hal). There are two possible reasons for this when con-sidering an infection with B. anthracis. First, each system may bedifferentially expressed under certain conditions, in certain tis-sues, or at different times and thus plays a functional role at adifferent point during the infection. For example, deletion of thebiosynthetic operon that encodes the siderophore anthrachelin(petrobactin) in B. anthracis results in a loss of spore germinationin macrophages (13). A second possibility is that bacillus hasevolved multiple systems to increase the overall rate of heme ac-quisition. In this regard, each component is functionally redun-dant, but collectively, the presence of multiple systems ensuresthat the cell maximizes its heme uptake capabilities. However, wecannot rule out the possibility that there are additional unrecog-nized functions for these proteins that impact our results, espe-cially when one considers that Hal and BslK both contain unchar-acterized protein domains (LRRs and SLH, respectively). It is clearfrom the growth studies whose results are presented in Fig. 4 thatthe uptake of heme in this pathogen may be a complex processwith inputs from multiple systems. For example, in the absence ofisd, bacilli seem to grow better than the wild-type strain on highconcentrations of heme or hemoglobin. The simplest explanationis that in the absence of one system (BslK or Isd) at these concen-trations of heme or hemoglobin, less overall heme is transportedinto the cell, leading to less heme toxicity and subsequently greatergrowth. Conversely, in the absence of a system, the cells may besensing that they are iron starved and thus increase the expressionof hal, which enhances growth. We are currently devising experi-ments to differentiate between these possibilities.

The results presented here allow us to suggest a basic model forthe function of Hal (Fig. 6). Upon entry into a mammalian host, B.anthracis encounters a low-iron environment and Fur-mediated

FIG 5 Structural model of the NEAT domain of Hal. The structural model ofthe NEAT domain of Hal was created using IsdX1 as the homology template inthe SWISS-MODEL workspace (2, 8). The resulting PDB file was then manip-ulated in PyMOL (PyMOL Molecular Graphics System, version 1.5.0.1;Schrödinger, LLC). The 310 helix residues (EESYAT) are shown in cyan, andthe � hairpin containing the heme-binding sequence (YDKEFKIQ) is demon-strated in yellow. Hemin porphyrin (magenta) was modeled into the NEATdomain heme-binding pocket using Coot (2, 6, 17). Oxygen atoms are red, andnitrogen atoms are dark blue. (B) The proximal and distal sides of the heme areshown, demonstrating the interaction of heme with nearby residues of Hal.

Function of Hal

October 2012 Volume 194 Number 20 jb.asm.org 5519

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

repression of the gene is relieved. As Hal is passed through the Secsecretion system, the N terminus is recognized by one of the threeB. anthracis sortases, possibly sortase A, thereby covalently attach-ing the protein to the cell wall. Upon exposure to blood or tissues,the NEAT domain temporarily associates with released hemoglo-bin, and the heme is removed and passed to an unidentified trans-porter(s) associated with the cell membrane. The LRRs may aidthis protein-protein interaction or, alternatively, initiate bindingto additional host proteins. The import of heme promotes bacte-rial replication and, eventually, anthrax disease, presumablythrough the use of the iron in key cellular processes.

ACKNOWLEDGMENTS

We thank Preeti Zanwar for a critical reading of the manuscript.This work was supported by a grant (AI069697 to A.W.M.) from the

National Institutes of Health.

REFERENCES1. Aisen P, Enns C, Wessling-Resnick M. 2001. Chemistry and biology of

eukaryotic iron metabolism. Int. J. Biochem. Cell Biol. 33:940 –959.2. Arnold K, Bordoli L, Kopp J, Schwede T. 2006. The SWISS-MODEL

workspace: a web-based environment for protein structure homologymodelling. Bioinformatics 22:195–201.

3. Berry EA, Trumpower BL. 1987. Simultaneous determination of hemesa, b, and c from pyridine hemochrome spectra. Anal. Biochem. 161:1–15.

4. Bierne H, Sabet C, Personnic N, Cossart P. 2007. Internalins: a complexfamily of leucine-rich repeat-containing proteins in Listeria monocyto-genes. Microbes Infect. 9:1156 –1166.

5. Bober M, Mörgelin M, Olin AI, von Pawel-Rammingen U, Collin M.2011. The membrane bound LRR lipoprotein Slr, and the cell wall-anchored M1 protein from Streptococcus pyogenes both interact withtype I collagen. PLoS One 6:e20345. doi:10.1371/journal.pone.0020345.

6. Bordoli L, et al. 2009. Protein structure homology modeling usingSWISS-MODEL workspace. Nat. Protoc. 4:1–13.

7. Braun V, Killmann H. 1999. Bacterial solutions to the iron-supply prob-lem. Trends Biochem. Sci. 24:104 –109.

8. Bridges KR, Seligman PA. 1995. Disorders of iron metabolism, p 1433–1472. In Blood: principles and practice of hematology. J. B. LippincottCompany, New York, NY.

9. Bullen JJ. 1981. The significance of iron in infection. Rev. Infect. Dis.3:1127–1138.

10. Bullen JJ, Rogers HJ, Griffiths E. 1978. Role of iron in bacterial infection.Curr. Top. Microbiol. Immunol. 80:1–35.

11. Carlson PE, Jr, Carr KA, Janes BK, Anderson EC, Hanna PC. 2009.Transcriptional profiling of Bacillus anthracis Sterne (34F2) during ironstarvation. PLoS One 4:e6988. doi:10.1371/journal.pone.0006988.

12. Carpenter BM, Whitmire JM, Merrell DS. 2009. This is not your moth-er’s repressor: the complex role of fur in pathogenesis. Infect. Immun.77:2590 –2601.

13. Cendrowski S, MacArthur W, Hanna P. 2004. Bacillus anthracis requiressiderophore biosynthesis for growth in macrophages and mouse viru-lence. Mol. Microbiol. 51:407– 417.

14. Daou N, et al. 2009. IlsA, a unique surface protein of Bacillus cereusrequired for iron acquisition from heme, hemoglobin and ferritin. PLoSPathog. 5:e1000675. doi:10.1371/journal.ppat.1000675.

15. Dementieva IS, et al. 2009. Pentameric assembly of potassium channeltetramerization domain-containing protein 5. J. Mol. Biol. 387:175–191.

16. Ekworomadu MT, et al. 2012. Differential function of lip residues in themechanism and biology of an anthrax hemophore. PLoS Pathog.8:e1002559. doi:10.1371/journal.ppat.1002559.

17. Emsley P, Lohkamp B, Scott WG, Cowtan K. 2010. Features and devel-opment of Coot. Acta Crystallogr. D Biol. Crystallogr. 66:486 –501.

18. Evdokimov AG, Anderson DE, Routzahn KM, Waugh DS. 2001. Un-usual molecular architecture of the Yersinia pestis cytotoxin YopM: a leu-cine-rich repeat protein with the shortest repeating unit. J. Mol. Biol.312:807– 821.

19. Fabian M, Solomaha E, Olson JS, Maresso AW. 2009. Heme transfer tothe bacterial cell envelope occurs via a secreted hemophore in the Gram-positive pathogen Bacillus anthracis. J. Biol. Chem. 284:32138 –32146.

20. Ferreira GC. 2008. Iron metabolism: inorganic biochemistry and regula-tory mechanisms. John Wiley & Sons Inc., New York, NY.

21. Gaspar AH, et al. 2005. Bacillus anthracis sortase A (SrtA) anchorsLPXTG motif-containing surface proteins to the cell wall envelope. J. Bac-teriol. 187:4646 – 4655.

22. Gat O, et al. 2006. Search for Bacillus anthracis potential vaccine candi-dates by a functional genomic-serologic screen. Infect. Immun. 74:3987–4001.

23. Gat O, et al. 2008. Characterization of Bacillus anthracis iron-regulatedsurface determinant (Isd) proteins containing NEAT domains. Mol. Mi-crobiol. 70:983–999.

24. Grigg JC, Ukpabi G, Gaudin CFM, Murphy MEP. 2010. Structuralbiology of heme binding in the Staphylococcus aureus Isd system. J. Inorg.Biochem. 104:341–348.

25. Grigg JC, Vermeiren CL, Heinrichs DE, Murphy MEP. 2007. Haemrecognition by a Staphylococcus aureus NEAT domain. Mol. Microbiol.63:139 –149.

26. Griggs DW, Konisky J. 1989. Mechanism for iron-regulated transcrip-tion of the Escherichia coli cir gene: metal-dependent binding of fur pro-tein to the promoters. J. Bacteriol. 171:1048 –1054.

27. Hammerstrom TG, Roh JH, Nikonowicz EP, Koehler TM. 2011. Bacil-lus anthracis virulence regulator AtxA: oligomeric state, function andCO(2)-signalling. Mol. Microbiol. 82:634 – 647.

28. Honsa ES, Fabian M, Cardenas AM, Olson JS, Maresso AW. 2011. Thefive near-iron transporter (NEAT) domain anthrax hemophore, IsdX2,scavenges heme from hemoglobin and transfers heme to the surface pro-tein IsdC. J. Biol. Chem. 286:33652–33660.

29. Honsa ES, Maresso AW. 2011. Mechanisms of iron import in anthrax.Biometals 24:533–545.

30. Howell S, Kenmore M, Kirkland M, Badley RA. 1998. High-densityimmobilization of an antibody fragment to a carboxymethylated dextran-linked biosensor surface. J. Mol. Recognit. 11:200 –203.

31. Kanehisa M, Goto S. 2000. KEGG: Kyoto Encyclopedia of Genes andGenomes. Nucleic Acids Res. 28:27–30.

32. Kobe B, Deisenhofer J. 1994. The leucine-rich repeat: a versatile bindingmotif. Trends Biochem. Sci. 19:415– 421.

33. Kobe B, Kajava AV. 2001. The leucine-rich repeat as a protein recogni-tion motif. Curr. Opin. Struct. Biol. 11:725–732.

34. Maresso AW, Chapa TJ, Schneewind O. 2006. Surface protein IsdC andsortase B are required for heme-iron scavenging of Bacillus anthracis. J.Bacteriol. 188:8145– 8152.

35. Maresso AW, Garufi G, Schneewind O. 2008. Bacillus anthracis secretes

FIG 6 Model for the function of Hal. Upon exposure of B. anthracis to alow-iron environment, Hal is expressed and covalently coupled by a sortase tothe cell wall via its C-terminal GPA. Surface Hal may have its N-terminalNEAT domain exposed to the extracellular environment, positioned to inter-act with released hemoglobin. Once bound, Hal extracts the heme from he-moglobin and retains the heme through a unique heme-protein linkage, even-tually transferring the bound heme to a downstream membrane-localizedtransporter. The LRRs may aid in this process or provide for bacillus binding toother iron-containing or host tissue proteins. Either the heme or the liberatediron is then used in essential biological processes, which enhances replicationin iron-limiting environments. CW, cell wall; M, cell membrane.

Balderas et al.

5520 jb.asm.org Journal of Bacteriology

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

proteins that mediate heme acquisition from hemoglobin. PLoS Pathog.4:e1000132. doi:10.1371/journal.ppat.1000132.

36. Marraffini LA, Schneewind O. 2006. Targeting proteins to the cell wall ofsporulating Bacillus anthracis. Mol. Microbiol. 62:1402–1417.

37. Mazmanian SK, et al. 2003. Passage of heme-iron across the envelope ofStaphylococcus aureus. Science 299:906 –909.

38. Murphy M, Jason-Moller L, Bruno J. 2006. Using Biacore to measure thebinding kinetics of an antibody-antigen interaction. Curr. Protoc. ProteinSci. Chapter 19:Unit 19.14.

39. Ouattara M, et al. 2010. Shr of group A streptococcus is a new type ofcomposite NEAT protein involved in sequestering haem from methaemo-globin. Mol. Microbiol. 78:739 –756.

40. Petersen TN, Brunak S, von Heijne G, Nielsen H. 2011. SignalP 4.0:discriminating signal peptides from transmembrane regions. Nat. Meth-ods 8:785–786.

41. Pilpa RM, et al. 2006. Solution structure of the NEAT (NEAr transporter)domain from IsdH/HarA: the human hemoglobin receptor in Staphylo-coccus aureus. J. Mol. Biol. 360:435– 447.

42. Ponka P. 1999. Cell biology of heme. Am. J. Med. Sci. 318:241–256.43. Ratledge C, Dover LG. 2000. Iron metabolism in pathogenic bacteria.

Annu. Rev. Microbiol. 54:881–941.

44. Sharp KH, Schneider S, Cockayne A, Paoli M. 2007. Crystal structure ofthe heme-IsdC complex, the central conduit of the Isd iron/heme uptakesystem in Staphylococcus aureus. J. Biol. Chem. 282:10625–10631.

45. Sigrist CJA, et al. 2010. PROSITE, a protein domain database for func-tional characterization and annotation. Nucleic Acids Res. 38:D161–D166. doi:10.1093/nar/gkp885.

46. Sterne M. 1946. Avirulent anthrax vaccine. Onderstepoort J. Vet. Sci.Anim. Ind. 21:41– 43.

47. Stojiljkovic I, Perkins-Balding D. 2002. Processing of heme and heme-containing proteins by bacteria. DNA Cell Biol. 21:281–295.

48. Tarlovsky Y, et al. 2010. A Bacillus anthracis S-layer homology proteinthat binds heme and mediates heme delivery to IsdC. J. Bacteriol. 192:3503–3511.

49. Teale FW. 1959. Cleavage of the haem-protein link by acid methylethyl-ketone. Biochim. Biophys. Acta 35:543.

50. Tong Y, Guo M. 2009. Bacterial heme-transport proteins and their heme-coordination modes. Arch. Biochem. Biophys. 481:1–15.

51. Wandersman C, Delepelaire P. 2004. Bacterial iron sources: from sidero-phores to hemophores. Annu. Rev. Microbiol. 58:611– 647.

52. Weinberg ED. 2009. Iron availability and infection. Biochim. Biophys.Acta 1790:600 – 605.

Function of Hal

October 2012 Volume 194 Number 20 jb.asm.org 5521

on June 9, 2018 by guesthttp://jb.asm

.org/D

ownloaded from