TonB-Dependent Heme Iron Acquisition in the Tsetse Fly Symbiont Sodalis glossinidius

Abstract

Sodalis glossinidius is an intra- and extracellular symbiont of the tsetse fly (Glossina sp.), which feeds exclusively on vertebrate blood. S. glossinidius resides in a wide variety of tsetse tissues and may encounter environments that differ dramatically in iron content. The Sodalis chromosome encodes a putative TonB-dependent outer membrane heme transporter (HemR) and a putative periplasmic/inner membrane ABC heme permease system (HemTUV). Because these gene products mediate iron acquisition processes by other enteric bacteria, we characterized their regulation and physiological role in the Sodalis/tsetse system. Our results show that the hemR and tonB genes are expressed by S. glossinidius in the tsetse fly. Furthermore, transcription of hemR in Sodalis is repressed in a high-iron environment by the iron-responsive transcriptional regulator Fur. Expression of the S. glossinidius hemR and hemTUV genes in an Escherichia coli strain unable to use heme as an iron source stimulated growth in the presence of heme or hemoglobin as the sole iron source. This stimulation was dependent on the presence of either the E. coli or Sodalis tonB gene. Sodalis tonB and hemR mutant strains were defective in their ability to colonize the gut of tsetse flies that lacked endogenous symbionts, while wild-type S. glossinidius proliferated in this same environment. Finally, we show that the Sodalis HemR protein is localized to the bacterial membrane and appears to bind hemin. Collectively, this study provides strong evidence that TonB-dependent, HemR-mediated iron acquisition is important for the maintenance of symbiont homeostasis in the tsetse fly, and it provides evidence for the expression of bacterial high-affinity iron acquisition genes in insect symbionts.

INTRODUCTION

All insects house endogenous microorganisms that mediate critical aspects of their host's physiology. Tsetse flies (Diptera: Glossinidae) harbor three phylogenetically distinct endosymbiotic bacteria, including parasitic Wolbachia spp., obligate Wigglesworthia sp., and commensal Sodalis glossinidius, that are maternally transmitted during the fly's unique viviparous mode of reproduction (reviewed in references 1 and 2). Sodalis glossinidius, a Gram-negative gammaproteobacterium in the family Enterobacteriaceae, resides both intra- and extracellularly in a wide range of tsetse tissues, including the midgut, muscle, fat body, milk gland (a gland that supplies nutrients to developing intrauterine larvae), and salivary glands. Genomic data indicate that genes whose products facilitate a free-living lifestyle, such as those involved in the biosynthesis and transport of nutrients, are retained in Sodalis. However, the high number of pseudogenes (972) present in the S. glossinidius genome suggests that this bacterium is actively transitioning from a free-living to symbiotic lifestyle (3). Sodalis can be cultured outside its tsetse host, suggesting that it has not undergone a high degree of reductive evolution characteristic of ancient obligate bacterial symbioses. The ability to culture this bacterium in vitro makes it a model organism amenable for studying the physiological mechanisms that underlie insect symbioses.

Iron is an essential nutrient for almost all bacteria, including Sodalis. Many bacteria, especially those that live in association with animal hosts, are equipped with several high-affinity iron uptake systems, the expression levels of which are tightly regulated (4). When bacterial iron levels are sufficiently high, expression of high-affinity iron acquisition genes is often negatively regulated by transcriptional repressors (5). The prototypical example of this is the Escherichia coli iron regulatory protein Fur (6–8). When iron levels are high, Fe²⁺ binds to the Fur protein and Fe-Fur binds to DNA sequences known as Fur boxes in the promoters of iron-repressed genes. This binding results in the repression of transcription of these genes. When iron is limited, this repression is relieved and the genes are transcribed.

In hematophagous insects, such as tsetse flies, heme and/or hemoglobin serve as a possible source of iron for bacterial symbionts. Iron acquisition systems in Gram-negative bacteria such as Sodalis enable the transport of iron, or iron-containing molecules, through the outer membrane via compound-specific receptors (9). In most of these cases, the energy for transport is provided by the TonB/ExbB/ExbD complex present in the inner membrane and which energetically links the proton motive force in the inner membrane to the outer membrane (10, 11). Receptors that require this system for energy are known as TonB-dependent receptors. Specific contacts are made between the C-terminal domain of TonB and motifs at the N-terminal domain of the TonB-dependent receptors (12–14). Sodalis produces several putative iron acquisition systems, including a previously characterized plasmid-carried (pSG1) mechanism that synthesizes and transports the iron-capturing siderophore achromobactin (15). Additionally, three Sodalis genes (hemT, hemU, and hemV) are homologous to well-characterized high-affinity ABC transport systems that transport heme through the inner membrane in other bacteria (15–20), and Sodalis contains a putative gene (hemR) that may encode an outer membrane heme receptor. Finally, the tonB gene is present in Sodalis.

Sodalis lives in the midgut of hematophagous tsetse flies, where heme would be available after a blood meal. Furthermore, genomic erosion has failed to eliminate the hemR and hemTUV genes in Sodalis. Thus, the purpose of this study was to investigate the function and regulation of this putative TonB-dependent heme uptake system.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. E. coli strains were grown in Luria-Bertani broth (LB) or on Luria-Bertani agar (L agar) plates. Liquid cultures were incubated at 37°C with 200 rpm aeration. Sodalis glossinidius was grown at 25°C and 10% CO₂ on brain heart infusion (BHI) agar with 10% horse blood (BHIB; Haemostat Laboratories, Dixon, CA). Liquid cultures of Sodalis in BHI were started in petri dishes at optical densities at 600 nm (OD₆₀₀) of 0.02 to 0.08 and incubated without aeration. Specific growth conditions for each experiment are indicated in figure legends. To create iron-limited conditions, the iron chelator ethylene diamino-o-dihyroxyphenyl acetic acid (EDDA) was added to growth medium as indicated in the figure legends (21). To create iron-replete conditions, 40 μM FeSO₄ was added to growth medium.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Characteristics	Reference
E. coli strains
DH5α	endA1 hsdR17 supE44 thi-1 recA1 gyrA relA1 Δ(lacZYA-argF)U169 deoR [Φ80dlacΔ(lacZ)M15]	45
W3110	F− IN(rrnD-rrnE)1	46
BL21(DE3)	F− ompT hsdSB(rB− mB−) gal dcm (DE3)	Novagen
ARM110	entF::cam	27
ARM113	entF::cam tonB::kan	27
UR45	entF::cam ΔdppB726::kan	This study
HB1017	entF::Tn5 derivative of HB101; Kmr	47
JW3512-2	F− Δ(araD-araB)567 ΔlacZ4787(::rrnB-3) λ− ΔdppB726::kan rph-1 Δ(rhaD-rhaB)568 hsdR514	28
S. glossinidius strains
SOD	S. glossindius from G. moristans moristans	S. Aksoy
SOD/pAR1219	Parent strain for Sodalis mutagenesis	16
SOD/pKM208	Parent strain for Sodalis mutagenesis	This study
URSOD6	fur183::kan	16
URSOD22	hemR1239::kan	This study
URSOD23	ΔtonB::cam	This study
Plasmids
pBR322	Cloning vector; choloramphenicol and tetracycline resistance	48
pAR1219	T7 polymerase under control of lac UV5 promoter for inducing intron mutagenesis	49
pBluescript KS	High-copy-no. cloning vector; carbenicillin resistance	50
pET-22b	Expression vector; carbenicillin resistance	Novagen
pGTLN3	Promoterless and ATG-less GFP vector; carbenicillin resistance	51
pWKS30	Low-copy-no. cloning vector; carbenicillin resistance	52
pRJ32	hemR promoter-gfp translational fusion on pGTLN3; carbenicillin resistance	This study
pTA1	Sodalis hemR in pWKS30; carbenicillin resistance	This study
pTA2	Sodalis hemR and hemTUV in pWKS30; carbenicillin resistance	This study
pTA4	Sodalis hemR gene in pET-22b; carbenicillin resistance	This study
pTPW1	Sodalis hemR gene in pACYC184; tetracycline resistance	This study
pWF1	Sodalis tonB gene in pWKS30; carbenicillin resistance	This study
pWF8	E. coli hemR gene in pWKS30; carbenicillin resistance	This study

Antibiotics were used for E. coli at the following concentrations: carbenicillin (CARB) at 125 μg/ml, ampicillin (AMP) at 50 μg/ml, chloramphenicol (CAM) at 30 μg/ml, kanamycin (KAN) at 50 μg/ml, and tetracycline (TET) at 12.5 μg/ml. Antibiotics were used for Sodalis at the following concentrations: CARB at 125 μg/ml, AMP at 20 μg/ml, CAM at 3 μg/ml, KAN at 25 μg/ml, and TET at 3 μg/ml.

Insect maintenance.

G. morsitans morsitans was maintained in Yale's insectary at 24°C with 50 to 55% relative humidity. All flies received defibrinated bovine blood (Hemostat Laboratories) every 48 h through an artificial membrane feeding system (22). To generate aposymbiotic flies, pregnant, wild-type female tsetse were fed a diet supplemented with tetracycline (20 μg/ml of blood) to clear all endogenous microbes and yeast extract (1%, wt/vol) to rescue the sterile phenotype associated with the absence of obligate Wigglesworthia sp. (23). DNA made from offspring of tetracycline-treated mothers was subjected to PCR analysis with eubacterial 16S rRNA primers to confirm the absence of endogenous bacteria (data not shown). Tsetse progeny that lacked all symbiotic bacteria throughout immature development and adulthood are here referred to as aposymbiotic (G. morsitans morsitans^Apo).

Plasmid constructions.

Plasmids were isolated using the QIAprep Spin miniprep kit (Qiagen, Valencia, CA). DNA fragments and enzyme reaction mixtures were purified with the QIAquick gel extraction kit (Qiagen). All ligations were done with T4 DNA ligase (Promega) and transformed into E. coli DH5α. The sequences of the PCR primers for cloning are listed in Table 2, and all PCRs for cloning were done using PfuTurbo (Agilent, Santa Clara, CA). To construct pRJ32, a DNA fragment containing the hemR promoter and translational start sites was PCR amplified from Sodalis with primers UR313 and UR314, digested with BamHI/EcoRV, and ligated to pLR44 digested with BamHI/EcoRV. To construct pAB6, a DNA fragment containing hemTUV was PCR amplified from Sodalis with primers UR179 and UR180, digested with XbaI/XhoI, and ligated to pWKS30 digested with XbaI/XhoI. To construct pTA1 and pTA2, respectively, a DNA fragment containing hemR was PCR amplified from Sodalis with primers UR323 and UR324, digested with XhoI/KpnI, and ligated to pWKS30 or pAB6 digested with XhoI/KpnI. To construct pTPW1, a DNA fragment containing the hemR gene was PCR amplified from pTA1 with primers UR461 and UR462, digested with PstI/SmaI, and ligated to pBR322 digested with PstI/SspI. To construct pTA4, a DNA fragment containing the hemR coding sequence (located between genes SG1505 and SG1506 in the Sodalis genome) was PCR amplified from Sodalis with primers UR415 and UR416, digested with HindIII/NdeI, and ligated to pET-22b digested with HinDIII/NdeI. To construct pWF1 and pWF8, DNA fragments containing either Sodalis tonB or E. coli tonB were amplified from Sodalis or E. coli W3110, respectively, using primers UR340 and UR341 or UR394 and UR395, respectively, digested with BamHI/XbaI, and ligated to pWKS30 digested with the BamHI/XbaI.

TABLE 2.

Primers used in this study

Primer	Sequence	DNA amplified
SodrplB1F	5′-TGCTGGAAACTCTCAGCAAAT	Sodalis rplB
SodrplB1R	5′-CTCCAGACGTTCTACCACTGC
M13F	5′-GTAAAACGACGGCCAGT	Genes in pWKS30
M13R	5′-GGAAACAGCTATGACCATG
SodtonB1F	5′-CACCGTGAAATCAAACCACAG	Sodalis tonB internal
SodtonB1R	5′-CATTGACGTCGAACATCACAC
UR179	5′-GGTCTAGACTGAAGGGGAAAAGCGATAAC	Sodalis hemTUV
UR180	5′-GGCTCGAGATATGCCGTAGAAGGCTGGTC
UR311	5′-GATACGCTACGCGCACCCCC	Sodalis hemR internal
UR312	5′-CAGACCCGGCGTTGGACCAC
UR313	5′-CCCCGGATCCTAAACCACTTTATTTCTGCCAG	Sodalis hemR promoter + ATG
UR314	5′-CATAGTCATTAAGCAATCTCC
UR323	5′-CGGGGTACCTTTTATCTGT CGTTTCATCGGC	Sodalis hemR
UR324	5′-CCCGCTCGAGCCCTCGCTTAACGTCTACTCC
UR340	5′-CGGGATCCTAGAATAATTTCACGCCACGC	Sodalis tonB
UR341	5′-TGCTCTAGAGGTGGGAGATGTCATTTGTTG
UR340	5′-CGGGATCCTAGAATAATTTCACGCCACGC	Sodalis tonB for allelic exchange
UR387	5′-TGCTCTAGACAAGGTCGAAAAAGTCGCCC
UR388	5′-GCGCGGTTACCCGTTACGCACCACCCCGTCA	cam
UR389	5′-GCGCGGTTACCCAGGCGTAGCACCAGGCGTT
UR394	5′-CGGGATCCATGTCGTGCCGAAGGGTCACC	E. coli tonB
UR395	5′-TGCTCTAGATGCCCACGGTCGCGTAGTGA
UR415	5′-TTGCCATATGACTATGCATCACAGGGG	Sodalis hemR coding sequence for overexpression
UR416	5′-GGCCAAGCTTGAAGTCCACGTTAAGCCCTAAC
UR461	5′-CGGGCTGCAGTTTTATCTGTCGTTTCATCGGC	Sodalis hemR from pTA1
UR462	5′-TAAACCCGGGCCCTCGCTTAACGTCTACTCC

Construction of Sodalis mutants.

The hemR mutant (URSOD22) was constructed using the Targetron intron mutagenesis kit (Sigma-Aldridge, St. Louis, MO) as described previously (16). Briefly, the group II intron on pACD4K-C-loxP was altered according to the manufacturer's instructions to contain a hemR-targeting site located 1,239 bp 3′ of the hemR start codon. The altered intron plasmid (pTA3) was electroporated into Sodalis Sod/pAR1219 (24). The electroporation mixture was transferred to 5 ml of BHI and incubated overnight. After overnight incubation, the samples were centrifuged (13,000 × g for 1 min), and the pellets were resuspended in 3 ml BHI containing chloramphenicol and 1% glucose and incubated overnight. Intron expression was induced with 500 μM isopropyl-β-d-thiogalactopyranoside (IPTG) for 1 h. The IPTG-treated sample was centrifuged (13,000 × g for 1 min), and the pellet was resuspended in 1 ml BHI broth containing 1% glucose and incubated for 1 h. The sample was centrifuged (13,000 × g for 1 min), and the pellet was resuspended in 400 μl BHI broth and spread on four BHIB agar containing kanamycin. Single colonies were restreaked after approximately 1 week on BHIB agar plates containing kanamycin. Insertion of the intron into the Sodalis hemR gene and elimination of the wild-type gene were confirmed by PCR analysis using Sodalis primers UR323 and UR324, which flank hemR. Complementation of the Sodalis mutants was not performed, because we were unable to cure Sodalis of the intron mutagenesis or alleleic exchange plasmids, which interfere with complementation plasmids.

The Sodalis tonB mutant (URSOD23) was constructed by allelic exchange. The tonB gene and approximately 1,000 bp of flanking DNA sequence were amplified from Sodalis by PCR using primers URSOD340 and URSOD387. This PCR product was digested with BamHI and XbaI and ligated into pWKS30 cut with BamHI and XbaI to construct pWF4. The tonB gene in pWF4 was interrupted with the chloramphenicol resistance gene (cam), which was obtained by amplification from pACYC184 with primers UR388 and UR389. The cam PCR product was digested with BstEII and ligated into pWF4 cut with BstEII to construct pWF4:cam. The tonB::cam fragment was amplified by PCR with primers M13F and M3R and electroporated into Sodalis containing the plasmid pKM208 (25), which harbors the phage lambda Red recombinase genes under the control of an inducible promoter. Recombinase expression was induced with 1 mM IPTG for 60 min, followed by a 15-min heat shock at 42°C. Transformants were selected on BHIB containing chloramphenicol and FeSO₄. PCR using primers that flanked tonB was employed to verify the deletion of the wild-type tonB allele.

Construction of E. coli strains.

UR45 was constructed by P1 transduction (26) of the entF::cam mutation from ARM110 (27) into JW3512-2 (dppB::kan) (28). entF encodes a protein required for siderophore synthesis, and dppB encodes a dipeptide permease (which can also facilitate heme transport). PCR using primers that flanked entF was employed to verify the P1 transductants.

Analysis of gene expression by semiquantitative RT-PCR.

For in vitro expression, Sodalis cultures were grown in BHI from a starting OD₆₀₀ of 0.02 for 3 days and then subcultured at a final OD₆₀₀ of 0.1 into BHI either supplemented with 40 μM FeSO₄ or EDDA (8 μg/ml and 16 μg/ml) and grown overnight. RNA was isolated from cultures by using the RNeasy minikit procedure (Qiagen), and the isolated RNA was treated again with DNase I (Qiagen). cDNA was then made from 200 ng total RNA using SuperScript III and random hexamer reverse transcription (RT) primers (Invitrogen, Carlsbad, CA). For in vivo expression, total RNA was isolated from adult tsetse fly guts (n = 3) using TRIzol (Ambion) and then treated with RNase-free Turbo DNase I (Ambion). cDNA was generated from 130 ng total RNA by using SuperScript III and random hexamers (Invitrogen). Semiquantitative PCR was performed on the cDNA samples with GoTaq polymerase (Promega, Madison, WI) and primers UR311 and UR312 for hemR and primers SodtonB1F and SodtonB1R for tonB (primer sequences are listed in Table 2). Primers SodrplB1F and SodrplB1R, which amplify the constitutively expressed gene rplB, were used as a control to ensure equal amounts of cDNA in each sample. The annealing temperature for all PCRs was 55°C. PCR samples were taken prior to saturation of the PCR for agarose gel electrophoresis. Amplified DNA was visualized with ethidium bromide staining on an agarose gel.

Tsetse per os colonization assay.

One thousand CFU of Sod/pAR1219, Sod/pKM208, URSOD22, or URSOD23 were added to 1 ml of heat-inactivated (HI) bovine blood and provided to G. morsitans morsitans^Apo flies (n = 25 per group) through an artificial membrane system. Following per os inoculation with bacteria, flies were maintained on HI blood every 48 h. Gut tissues were microscopically dissected at 1, 5, and 10 days postinoculation (p.i.) with bacteria. Guts from each group (n = 5) per time point were dissected, homogenized in 0.85% NaCl, serially diluted, and plated on BHIB supplemented with antibiotics (29). CFU per plate were manually counted.

Bioassays to detect utilization of potential iron sources.

E. coli strains were seeded (10⁶ bacteria/ml) into L agar containing EDDA and any antibiotic required to maintain plasmids as indicated in the figure legends. Five microliters of each iron source (10 mM FeSO₄, 5 μM hemin, 50 μM hemin, or 1 μM hemoglobin) was spotted directly onto the agar. The size of the diameter of growth stimulation was measured after 24 h of incubation at 37°C.

Purification of membrane proteins.

Cultures of BL21(DE3)/pET-22b or BL21(DE3)/pTA4 were grown to mid-log phase in LB broth containing carbenicillin. IPTG was added to a final concentration of 5 mM to induce hemR expression on pTA4 for ∼24 h. Harvested cells (50 ml; ∼5 × 10¹⁰ bacteria) were centrifuged at 7,800 × g for 15 min and resuspended in 7 ml of 10 mM sodium phosphate buffer (pH 7.0). The bacteria were lysed by sonication with 1-s pulses for 45 s, and a Complete EDTA-free protease inhibitor cocktail pellet (Boehringer Mannheim) was added. Unbroken cells were removed by centrifugation at 7,800 × g for 15 min at 4°C. The supernatant was then centrifuged at 100,000 × g for 60 min at 4°C to collect the total membrane pellet (30). The insoluble pellet was resuspended in 2.7 ml of water and refrigerated overnight. The solution was then centrifuged at 100,000 × g for 45 min at 4°C. The remaining total membrane pellet was resuspended in 150 μl water.

Hemin binding assays.

For heme-dependent DMB (3,3 dimethoxybenzidine dihydrochloride; also called o-dianisidine) staining, 1-ml cultures of BL21(DE3)/pET-22b or BL21(DE3)/pTA4 were used for starting material. Cultures were grown to mid-log phase in L broth containing carbenicillin, and IPTG was added to a final concentration of 5 mM to induce hemR expression on pTA4 for ∼24 h. The cultures were then centrifuged at 13,000 × g, and the bacterial pellets were resuspended in 100 μl 1× Laemmli buffer (Bio-Rad Laboratories, Hercules, CA) lacking β-mercaptoethanol. The samples were not heated before loading on a gel. Proteins were separated by SDS-PAGE, incubated with hemin, and then stained with DMB using a modified procedure from Schulz et al. (31). Briefly, the gels were rinsed three times in water and then incubated for 1 h in 100 μM hemin. The gels were washed three times in water to remove unbound hemin and then incubated with 20 mg/ml DMB in a 50 mM sodium citrate–0.7% H₂O₂ solution for 3 min to detect hemin-bound proteins. The gels were destained in water overnight.

Hemoproteins were also detected using luminol-based chemiluminescence to detect the intrinsic peroxidase activity of hemoproteins (32). Aliquots of 12.5 mg of purified membrane proteins were resuspended in modified 1× Laemmli buffer (lacking β-mercaptoethanol and with lithium dodecyl sulfate [LDS] replacing SDS) and separated by LDS-PAGE (33) on duplicated gels. The samples were not heated before loading on the gel. One gel was stained with BioSafe Coomassie brilliant blue G-250 (Bio-Rad Laboratories). The second gel was washed four times with Tris-buffered saline, and peroxidase activity of the proteins was detected using the Clarity Western enhanced chemiluminescence substrate reagent (Bio-Rad Laboratories) per the manufacturer's instructions.

Identification of HemR from PAGE gels.

Protein identification of HemR from PAGE gels was performed by Kristina Nelson at the Chemical and Proteomic Mass Spectrometry Core Facility at Virginia Commonwealth University.

RESULTS

Identification of the Sodalis hemR gene.

A BLASTX search of the Sodalis genome (S. glossinidius strain Morsitans [accession number NC_007712]) identified a region (nucleotides 2524979 to 2527363; reverse complement) that encodes a putative protein homologous to putative TonB-dependent outer membrane heme receptors in other organisms. We designated this gene hemR. The deduced amino acid sequence of the S. glossinidius strain Morsitans hemR was 91% similar to a putative protein from the S. glossinidius HS strain, which was recently isolated from a human clinical infection (34). Additionally, the S. glossinidius strain Morsitans HemR exhibits 76 to 81% amino acid similarity to other uncharacterized proteins annotated as putative TonB-dependent outer membrane heme receptors in Klebsiella spp., Pectobacterium sp., Dickeya sp., and Erwinia spp. in the NCBI BLAST databases (see Fig. S1 in the supplemental material). A CLUSTALW analysis comparing Sodalis HemR with TonB-dependent outer membrane heme receptors that have been functionally characterized in other bacteria, including Pseudomonas fluorescens, Vibrio vulnificus, Neisseria meningitidis, Yersinia enterocolitica, and Shigella dysenteriae, showed similarities in the 35 to 42% range (see Fig. S1).

The Sodalis hemR gene has functional transcription and translation signals.

The Sodalis hemR gene had been annotated as a pseudogene in the original genome sequencing projects (Euginio Belda, personal communication) (3). Since the hemR transcription and translation profile had not been experimentally verified, we cloned the putative Sodalis hemR promoter and potential start codons into a green fluorescent protein (GFP) translational fusion vector (pGTLN3), which lacked a promoter, ribosome binding site, and translational start codon to drive GFP expression. The E. coli strain containing this new plasmid, pRJ32, was tested for GFP expression driven by the Sodalis hemR upstream regulatory sequences by using flow cytometry. Expression of GFP was 10 times greater in DH5α/pRJ32 (39 RFU [relative fluorescence units]) than in the control, DH5α/pGTLN3 (4 RFU) (data not shown). Thus, the sequences upstream of the Sodalis hemR gene are capable of directing transcription and translation.

Expression of Sodalis hemR is repressed in a high-iron environment by the iron-responsive transcriptional regulator Fur.

The expression of Sodalis siderophore synthesis genes is regulated by iron levels (15). To determine whether such regulation occurs with hemR, we used semiquantitative RT-PCR to measure expression levels of Sodalis hemR in cells grown under high- or low-iron conditions by using an iron chelator, EDDA. RNAs isolated from Sodalis grown under both conditions were subjected to RT-PCR analysis using hemR-specific PCR primers. Results showed that hemR was expressed at higher levels in Sodalis grown in low-iron versus iron-replete conditions (Fig. 1A). To confirm that the increased expression of hemR in EDDA-containing medium was due to chelation of iron, iron was added back to the EDDA-treated medium, and hemR expression was measured. In this case, hemR expression was reduced (Fig. 1B).

FIG 1.

Expression of hemR is iron repressed in Sodalis. For semiquantitative RT-PCR analysis of hemR expression, cDNAs were generated from wild-type Sodalis cultured under high- or low-iron conditions by using random hexamers. The cDNA was amplified using primers specific to hemR or rplB (constitutive control). (A) Sodalis culture medium was supplemented with either 40 μM FeSO₄ (Fe), 8 μg/ml of the iron chelator EDDA (8E), or 16 μg/ml of EDDA (16E). (B) Sodalis culture medium was supplemented with either 40 μM FeSO₄ (Fe), 16 μg/ml of EDDA (E), or 16 μg/ml EDDA plus 45 μM FeSO₄ (E+F). Five microliters of each PCR was loaded onto a 1% agarose gel, and the gel was stained with ethidium bromide. The experiment was performed in triplicate, and the results of a representative experiment are shown.

Iron repression is frequently mediated by Fur in Gammaproteobacteria (35). Thus, to investigate whether Sodalis hemR transcription is also regulated by Fur, the Sodalis Fur mutant URSOD6 and parental strain SOD/pAR1219 were grown under high- and low-iron conditions. Subsequently, RT-PCR was performed to measure hemR expression in both bacterial strains. SOD/pAR1219, which produces functional Fur, exhibited the repression of hemR under iron-replete conditions and increased expression under low-iron conditions (Fig. 2). Conversely, in the Fur mutant (URSOD6) grown under the same iron-replete conditions, hemR expression was aberrantly high, suggesting transcription was no longer repressed in the Fur mutant (Fig. 2). Taken together, these data suggest that expression of the Sodalis hemR gene is repressed in a high-iron environment by the iron-responsive transcriptional regulator Fur.

FIG 2.

Expression of hemR is iron repressed via Fur in Sodalis. For semiquantitative RT-PCR analysis of hemR expression in a fur mutant (URSOD6) and parent control Sodalis (Sod/pAR1219), cDNA was isolated from each strain cultured in BHI supplemented with either 40 μM FeSO₄ (Fe) or 16 μg/ml of the iron chelator EDDA (E). The cDNA was amplified using primers specific to hemR or rplB (constitutive control). Five microliters of each PCR was loaded onto a 1% agarose gel, and the gel was stained with ethidium bromide. The experiment was performed in triplicate, and results of a representative experiment are shown.

Expression of Sodalis iron acquisition genes in tsetse flies.

The Sodalis genome encodes TonB, which is predicted to provide the energy for high-affinity iron acquisition from heme and heme-containing molecules, based on TonB function in other bacteria. To examine whether hemR and tonB are expressed when Sodalis resides within tsetse flies, we performed RT-PCR on total RNA from samples of pooled adult fly guts. Production of bacterial cDNA from total RNA was verified by PCR amplification of constitutively expressed, Sodalis-specific rplB. Furthermore, amplification of tonB- and hemR-specific sequences from the cDNA demonstrated that these genes are expressed by Sodalis in tsetse flies (Fig. 3).

FIG 3.

hemR and tonB are expressed by Sodalis that resides within tsetse flies. For semiquantitative RT-PCR analysis of hemR and tonB expression in wild-type Sodalis resident within the tsetse gut, cDNA was isolated from two pools (A and B) each from three adult tsetse fly guts by using random hexamers. The cDNAs were amplified by PCR with either hemR, tonB, or rplB (constitutive control) primers. Five microliters of each RT-PCR product was run on a 2% agarose gel, and the gel was stained with ethidium bromide. +, Sodalis chromosomal DNA positive control; -, no-DNA negative control.

Sodalis hemR and tonB mutants have growth defects in tsetse flies.

Expression of tonB and hemR in Sodalis within tsetse flies suggested a potential role for these genes in the endogenous population of Sodalis in its tsetse host. Therefore, we tested whether hemR and tonB products contributed to Sodalis colonization dynamics in the tsetse gut. To do so, we inoculated four groups of G. morsitans morsitans^Apo tsetse flies per os (10³ CFU/ml of blood) with either Sodalis strain URSOD22 (hemR mutant), URSOD23 (tonB mutant), Sod/pAR1219 (wild-type hemR parent of URSOD22), or Sod/pKM208 (wild-type tonB parent of URSOD23). Subsequently, we monitored bacterial density within tsetse guts at 1, 5, and 10 days postinoculation. G. morsitans morsitans^Apo tsetse flies were used for these colonization experiments to eliminate competition between endogenous Sodalis and experimentally introduced mutant strains. We discovered that the tsetse gut contained on average 10³-fold more Sod/pAR1219 or Sod/pKM208 at the 10-day time point than at the 1-day time point. Conversely, neither mutant strain was able to proliferate in this environment (Fig. 4). In fact, no viable URSOD23 bacteria were recovered from the tsetse gut at the 5- and 10-day time points. These findings suggest that Sodalis uses the TonB-dependent HemR receptor to take up heme from the tsetse midgut environment in order to proliferate in this niche.

FIG 4.

HemR and TonB are required for Sodalis colonization of the tsetse fly gut. Four distinct groups of G. morsitans morsitans^Apo tsetse flies were inoculated per os with 10³ CFU of either SOD/pAR1219 (wild type for hemR and tonB; parent of URSOD22), SOD/pKM208 (wild type for hemR and tonB; parent of URSOD23), URSOD22 (hemR mutant), or URSOD23 (tonB mutant). At 1, 5, and 10 days postinoculation (dpi), Sodalis-colonized G. morsitans morsitans^Apo flies were sacrificed and their gut contents were plated to determine bacterial density. At 5 and 10 days p.i., no tonB mutants were recovered from the flies. Each symbol represents the gut contents of one fly. Statistical significance was determined by using the Mann-Whitney test (P < 0.5 for each mutant-to-parent comparison at 5 and 10 days p.i.).

Sodalis HemR enables use of hemin and hemoglobin as iron sources.

To test whether Sodalis hemR encodes a functional heme transporter, we expressed either Sodalis hemR (on plasmid pTA1) or hemR and the inner membrane permease operon hemTUV (on plasmid pTA2) in E. coli (strain UR45). UR45 naturally lacks an outer membrane heme transporter and has a mutation in the siderophore synthesis gene entF; thus, the strain cannot grow in iron-restricted environments. We performed bioassays on E. coli strains UR45/pTA1 and UR45/pTA2 to determine whether exogenous hemin or hemoglobin could stimulate growth of these bacteria. We found that E. coli UR45 was able to proliferate when expressing either Sodalis hemR or hemR hemTUV. Conversely, growth of E. coli UR45/pWKS30, which is a control strain unable to transport heme, was not stimulated by hemin or hemoglobin (Table 3). Taken together, these data suggest that the Sodalis HemR is sufficient for transport of heme for iron acquisition in the E. coli heterologous host strain.

TABLE 3.

Sodalis HemR enables use of hemin and hemoglobin as iron sources in E. coli strain UR45

Plasmid (gene) carried by UR45	Mean diam (mm) with indicated iron sourcea
	FeSO4 (10 mM)	H2O	Hemin		Hb (1 μM)
			5 μM	50 μM
pWKS30	10 ± 1	0 ± 0	0 ± 0	0 ± 0	0 ± 0
pTA1 (hemR)	11 ± 1	0 ± 0	9 ± 1	12 ± 1	8 ± 1
pTA2 (hemR and hemTUV)	10 ± 0	0 ± 0	8 ± 2	15 ± 1	9 ± 1

^aE. coli UR45 (entF::cam dppB::kan) containing the indicated plasmid was seeded (10⁶ bacteria/ml) onto L agar plates containing 100 μg/ml EDDA and 125 μg/ml carbenicillin. Five microliters of each indicated iron source was spotted directly onto the agar. The size of the diameter of growth stimulation around the spotted compound was measured after 24 h of incubation at 37°C. The data shown are the means ± standard deviations of three experiments. Hb, hemoglobin.

Sodalis HemR heme transport is TonB dependent.

To determine whether HemR-dependent iron acquisition required TonB, we tested the ability of hemR to stimulate HemR-mediated heme-dependent growth in an E. coli tonB mutant. E. coli mutant strain ARM113, which lacks both an outer membrane heme receptor and TonB, was transformed with a plasmid (pTPW1) that harbors the Sodalis hemR. If TonB were required for HemR-dependent iron acquisition, then E. coli ARM113/pTPW1 would fail to grow in the presence of hemin. However, growth of E. coli ARM113/pTPW1 should be restored via the expression of either Sodalis or E. coli tonB (carried on plasmids pWF1 and pWF8, respectively). The presence of hemin or hemoglobin stimulated growth of E. coli ARM113/pTPW1 that expressed either Sodalis or E. coli tonB. Conversely, an E. coli control strain (ARM113/pTPW1 harboring plasmid pWKS30) failed to grow in the presence of hemin or hemoglobin (Table 4). Taken together, these data suggest that the HemR-dependent iron acquisition requires TonB.

TABLE 4.

Sodalis HemR-dependent iron acquisition requires TonB in E. coli

Plasmid carried by ARM113/pTPW1 (hemR)	Source of TonB	Mean diam (mm) with indicated iron sourcea
		FeSO4 (10 mM)	H2O	Hemin (50 μM)	Hb (1 μM)
pWKS30	None	5 ± 0.5	0	0	0
pWF1	Sodalis	5 ± 0	0	12 ± 0.96	21 ± 1.5
pWF8	E. coli	5 ± 0	0	13 ± 1.7	20 ± 1.7

^aE. coli ARM113 (entF::cam tonB::kan) containing the indicated plasmids was seeded (10⁶ bacteria/ml) onto L agar plates containing 150 μg/ml EDDA, 30 μg/ml tetracycline, and 125 μg/ml carbenicillin. pTPW1 carries the Sodalis hemR gene. Five microliters of each indicated iron source was spotted directly onto the agar. The size of the diameter of growth stimulation around the spotted compound was measured after 24 h of incubation at 37°C. The data shown are from a representative experiment. Hb, hemoglobin.

Localization of Sodalis HemR to the bacterial membrane and characterization of hemin binding.

Bioinformatic analysis suggested that Sodalis HemR contains a membrane localization motif. To determine if this protein localizes to the bacterial membrane, we compared the membrane protein profiles of E. coli strains that either do or do not express the Sodalis hemR gene. An SDS-PAGE gel of membrane-extracted proteins revealed the presence of an ∼88-kDa band in E. coli strains that express Sodalis hemR (carried on plasmid pTA4), but not in E. coli carrying the vector control pET-22b (Fig. 5). This band was confirmed to be Sodalis HemR via mass spectrometry. To begin a preliminary examination of the potential physical interaction between HemR and hemin, we overexpressed the Sodalis HemR protein in E. coli [E. coli BL21(DE3)/pTA4] and measured hemin binding using two different techniques. First, we measured the ability of cell lysates to bind hemin by using a colorless compound (DMB) that turns reddish brown in the presence of heme (31). An ∼88-kDa band was detected by DMB staining in cell lysates from E. coli BL21(DE3)/pTA4. Conversely, lysates from an E. coli strain that does not produce HemR did not exhibit this same band (Fig. 6). Second, we treated PAGE-separated membrane proteins with a luminol-based chemiluminescent substrate as a means of detecting the intrinsic peroxidase activity of hemoproteins, such as HemR. As with the DMB staining, an ∼88-kDa band was detected by chemiluminescence in membrane-bound protein fractions of E. coli BL21(DE3)/pTA4 (Fig. 7). Taken together, these data suggest that ectopically expressed Sodalis hemR encodes a protein that is capable of binding hemin.

FIG 5.

Sodalis HemR is localized to bacterial membranes when overexpressed in E. coli. Cellular fractions from E. coli BL21(DE3) that harbored either pTA4 (hemR) or pET-22b (control plasmid) were run on an SDS-PAGE gel and stained with Coomassie brilliant blue. HemR is indicated by the asterisk. The experiment was performed in triplicate, and results of a representative experiment are shown.

FIG 6.

Detection of Sodalis HemR by heme-dependent DMB staining. Whole-cell lysates from E. coli BL21(DE3) that harbored either pTA4 (hemR) or pET-22b (control plasmid) were fractionated by SDS-PAGE. The gel was subsequently stained with DMB to detect hemin-bound proteins. HemR is indicated by the asterisk. The experiment was performed in duplicate, and results of a representative experiment are shown.

FIG 7.

Detection of Sodalis HemR by using luminol-based chemiluminescence. Membrane proteins, isolated from E. coli BL21(DE3) that harbored either pTA4 (hemR) or pET-22b (control plasmid), were fractionated by LDS-PAGE on two gels. (Left) One gel was stained with Coomassie brilliant blue; (right) the other gel was incubated with a luminol-based chemiluminescence substrate to detect the intrinsic peroxidase activity in heme-containing proteins. HemR bands are indicated by asterisks. The experiment was performed in duplicate, and results of a representative experiment are shown.

DISCUSSION

This study was undertaken to investigate the function and regulation of a putative TonB-dependent heme iron uptake system in the tsetse fly commensal symbiont Sodalis glossinidius. Sodalis lives in the midgut of hematophagous tsetse flies, where heme is available in abundance after a blood meal. Additionally, genome erosion has failed to eliminate the hemR and hemTUV genes in Sodalis. Based on these characteristics, we hypothesized that Sodalis HemR would be functional and would contribute to growth of Sodalis within its host's gut. Results from our functional and genetic analyses support this model.

We found that the hemR gene is required for normal Sodalis growth and survival in the midgut of tsetse flies, as Sodalis that lacks hemR cannot proliferate within its host. In fact, we found that the number of viable Sodalis hemR mutant cells decreased by 5 days postinfection; this suggested that not only do hemR mutants fail to proliferate, but also that the mutants are dying in the tsetse fly. Because these mutants are not metabolically active, they may be unable to survive within the immunologically hostile environment of the tsetse gut (36). Since Sodalis HemR was able to mediate bacterial growth using heme or hemoglobin as the sole iron source in our heterologous E. coli host system, a possible reason that the Sodalis hemR mutant was unable to proliferate in the tsetse gut is that it could not import heme to use as an iron source for essential physiological processes. Because tsetse flies are hematophagous insects, heme found in hemoglobin is likely to be the major iron source for Sodalis in the fly midgut. Interestingly, Sodalis mutants that fail to make the achromobactin siderophore, which is this bacterium's other major iron acquisition system, are still able to proliferate in the tsetse gut, likely because HemR is still present (15).

Our results demonstrate that Sodalis HemR-dependent heme acquisition is TonB dependent, as bacteria that express hemR, but not tonB, were unable to use heme as an iron source. The fact that both the E. coli and Sodalis tonB genes restored Sodalis hemR-dependent growth to E. coli that lacked tonB suggests that these two TonB homologues remain functionally conserved despite the fact that these bacteria diverged millions of years ago. The Sodalis and E. coli TonB proteins are 46% identical at the amino acid level. The C-terminal domains in these homologues (which interact with outer membrane receptors) are more highly conserved than the N-terminal domains, and residues that are highly conserved among other TonB homologues are also conserved in Sodalis TonB.

Sodalis lacking tonB was even more impaired in growth and survival inside the tsetse fly than the hemR mutants, suggesting that there may be additional roles for TonB in the tsetse gut. A bioinformatic search revealed five more potential TonB-dependent transporters in Sodalis, based on conserved Pfam domains (L. J. Runyen-Janecky, unpublished data). The tonB mutation may be pleiotropic, and TonB-dependent functions in addition to heme transport could be required for growth and proliferation of Sodalis in the gut. Recent work has suggested possible substrates for TonB-dependent transport that are not related to iron acquisition (37). Some of these potential substrates include nickel, sucrose, cobalt, and chito-oligosaccharides (37). A predicted TonB-dependent transporter specific for N-acetylglucosamine (NAG) has been identified in Shewanella oneidensis (38). This is especially interesting since Sodalis can use NAG as a carbon source (39). Perhaps the Sodalis tonB mutant is unable to use the available NAG for growth in tsetse flies. TonB-dependent transport may be important in other invertebrate symbionts besides Sodalis and thus was retained during evolution. An examination of positive selection in endosymbionts such as Verminephrobacter sp. and Betaproteobacteria sp., which form evolutionarily ancient associations with earthworms, indicates that TonB-dependent iron uptake has been maintained (40).

The closest matches to the deduced amino acid sequence of HemR (76 to 81% amino acid similarity) were from putative TonB-dependent outer membrane heme receptors in Klebsiella sp., Pectobacterium sp., Dickeya sp., and Erwinia sp., all which are known to be associated with plants (although Klebsiella can also cause human infections). In addition, the genomic region that encodes hemR contains remnants of phage and transposase genes. These data suggest that the hemR gene (and the adjacent gene hemS, which encodes a putative heme oxidase) might have been acquired from plant-associated bacteria during evolution. The fact that the deduced amino acid sequence of Sodalis hemR was 91% similar to a putative protein from the Sodalis HS1 strain (34) suggests that hemR was in the HS1 ancestral strain. Interestingly, this HS1 isolate was obtained from a human clinical infection acquired from a puncture wound from a crabapple tree branch. The ancestral HS strain may have acquired this gene (via a horizontal transfer event) from a resident plant-associated bacterium. Alternatively, the ancestral HS1 strain, and thus Sodalis, may have evolved from a plant-associated bacterium.

In addition to using iron in heme as an iron source, some bacterial heme auxotrophs require the entire heme molecule as heme source. Examples of such bacteria include the human pathogens Bartonella quintana and Bartonella henselae, which are transmitted via hematophagous arthropod vectors. Neither of these bacteria contains heme biosynthesis genes. Instead, they scavenge heme found in their host's blood meal (41, 42). Sodalis has retained genes that encode both heme transporters and complete heme biosynthesis pathways, suggesting that environments other than the tsetse gut might necessitate the biosynthesis of heme in tissues that are low in heme content. Additionally, it is possible that just the iron from the heme molecule is transported into Sodalis. Interestingly, Sodalis exhibits a broad tissue tropism in its tsetse host. This bacterium can be found in the tsetse's hemolymph, salivary glands, milk glands, reproductive tract components (uterus and spermathecae), and fat body (43), all tissues where heme is likely less abundant than in the fly's midgut. Future work will identify whether these heme biosynthesis genes are functional and whether the entire heme molecule is transported by HemR.

Because excess iron is toxic, most free-living bacteria precisely regulate intracellular iron levels by repressing iron acquisition systems in iron-replete environments. Our hemR expression data indicate that such regulation still occurs in Sodalis and is regulated by the iron-responsive transcriptional repressor protein Fur. This regulation, coupled with the multiple functional iron acquisition systems retained despite ongoing genomic erosion of other regulatory pathways and genes in Sodalis, suggests that this bacterium may experience variations in iron levels and sources depending on the environmental conditions in which it resides. For example, iron levels may fluctuate between feeding, and/or iron levels and sources may be different in distinct tsetse tissues (both intra- and extracellular) where Sodalis resides.

In summary, the data presented here provide support to the idea that the iron-regulated, TonB-dependent, HemR-mediated iron acquisition system is critical for the normal physiology of Sodalis glossinidius within the complex tsetse host. Because Sodalis affects tsetse physiology and vector competence (44), a complete understanding of symbiont iron acquisition is important for a complete understanding of the host-symbiont interaction.