|Home | About | Journals | Submit | Contact Us | Français|
The spirochetes of the Leptospira genus contain saprophytic and pathogenic members, the latter being responsible for leptospirosis. Despite the recent sequencing of the genome of the pathogen L. interrogans, the slow growth of these bacteria, their virulence in humans, and a lack of genetic tools make it difficult to work with these pathogens. In contrast, the development of numerous genetic tools for the saprophyte L. biflexa enables its use as a model bacterium. Leptospira spp. require iron for growth. In this work, we show that Leptospira spp. can acquire iron from different sources, including siderophores. A comparative genome analysis of iron uptake systems and their regulation in the saprophyte L. biflexa and the pathogen L. interrogans is presented in this study. Our data indicated that, for instance, L. biflexa and L. interrogans contain 8 and 12 genes, respectively, whose products share homology with proteins that have been shown to be TonB-dependent receptors. We show that some genes involved in iron uptake were differentially expressed in response to iron. In addition, we were able to disrupt several putative genes involved in iron acquisition systems or iron regulation in L. biflexa. Comparative genomics, in combination with gene inactivation, gives us significant functional information on iron homeostasis in Leptospira spp.
Leptospira belongs to the bacterial phylum of spirochetes, which has a deep branching lineage in Bacteria, as indicated by 16S rRNA analysis (42). The genus Leptospira was initially divided into two groups: the pathogenic Leptospira referred to as Leptospira interrogans sensu lato and the saprophytic Leptospira referred to as L. biflexa sensu lato (9). Saprophytic and pathogenic Leptospira spp. were first classified into serovars, with more than 220 serovars defining the pathogens. More recently, DNA-DNA hybridization studies separated Leptospira species into 17 genomospecies, including 7 pathogenic species (9).
In the past decade, leptospirosis has emerged as a widespread zoonosis, and its incidence is high in tropical countries. Leptospirosis is acquired by direct or indirect contact with the urine of infected animals such as rodents (9). Virulence mechanisms and more generally the fundamental understanding of the biology of the causative agent of leptospirosis remain largely unknown. Recently, the genome sequencing of two serovars of L. interrogans sensu stricto, the main species associated with human leptospirosis, has been achieved (38, 48). However, the lack of genetic tools in pathogenic Leptospira does not allow the full characterization of genes of interest. Only recently the first evidence of gene transfer has been demonstrated in L. interrogans by transposition of Himar1, a transposon of eukaryotic origin (13). In contrast, numerous tools for genetic manipulation of saprophytic Leptospira species have been developed in recent years (7, 27, 35, 43, 44, 51, 56). These studies enable the use of the saprophyte L. biflexa as a model spirochete. The availability of the genome sequence of the saprophyte L. biflexa (unpublished data) and its comparison with the genomes of pathogenic species give us functional information on the lifestyles of Leptospira spp. in the environment and the infected host.
Iron plays a central role in many major biological processes, such as the electron transport chains for most living cells, including Leptospira spp. However, a few organisms, such as the spirochete Borrelia burgdorferi, do not require iron for growth (46). Spirochetes possess a double-membrane structure composed of a cytoplasmic membrane that differs substantially from that of gram-negative bacteria, the periplasm, and the outer membrane (9), which may constitute a barrier for molecules that could be used as an iron source. In gram-negative bacteria, iron sources can be recognized by specific outer membrane receptors, called TonB-dependent receptors, and then transported across the inner membrane by periplasmic binding protein-dependent ABC permeases (38, 48). We recently characterized the fecA- and feoB-like genes by random transposon mutagenesis in L. biflexa (35). Genes involved in iron acquisition are usually transcriptionally regulated by the availability of iron through regulators such as the ferric uptake regulator protein Fur. Cullen et al. have shown that the expressions of some genes from L. interrogans were regulated by iron (19), and fur-like genes are present in the L. interrogans genomes (38, 48).
In this study, the analysis of the genome sequence of L. biflexa allowed us to identify putative genes involved in iron transport and regulation. We showed that some of these putative genes modulate their expression in response to iron. To study the function of these genes, we generated several mutants in L. biflexa, and their phenotypes were characterized. Since pathogenic Leptospira spp. as well as saprophytic species need to obtain iron to grow in vitro, and probably in vivo in the host, better knowledge of the iron uptake systems and their regulation is essential to understand the pathogenesis of this intriguing group of organisms.
L. biflexa serovar Patoc strain Patoc1 and L. interrogans serovar Lai strain Lai (National Reference Center for Leptospira, Paris, France) were grown at 30°C in EMJH (22, 31) medium. When necessary, kanamycin and spectinomycin were added at 40 μg/ml. Minimal iron EMJH medium was prepared by either omitting iron sulfate (normally 330 μM) or, prior to inoculation, treating normal EMJH medium with 50 μM 2,2′-dipyridyl (Sigma-Aldrich, St. Louis, MO) overnight. Media for testing the ability of Leptospira to use specific sources of iron was prepared by supplementing the dipyridyl-treated EMJH medium. The siderophores ferrichrome (final concentration, 20 μM), enterobactin (final concentrations, 10 to 100 μM), and aerobactin (final concentration, 50 μM) were purchased from EMC Microcollections GmbH (Germany). Desferrioxamine (final concentration, 10 μM), also called ferrioxamine B, refers to deferoxamine mesylate (Sigma-Aldrich, St. Louis, MO). Bovine hemin (final concentration, 10 μM) and lactoferrin (final concentration, 10 μM) were obtained from the Sigma-Aldrich Company (St. Louis, Mo). Iron citrate and iron chloride were used at 100 μM.
Genomic DNA of Leptospira was isolated as previously described (44). Plasmid DNA was purified using the Plasmid Miniprep kit (QIAGEN GmbH, Hilden, Germany). Total RNA was isolated by using RNAwiz (Ambion Inc.) and treated with DNaseI. For transcription studies, RNA was isolated from exponential-phase cultures of L. biflexa grown in EMJH or in dipyridyl-treated EMJH. The absence of DNA contamination was confirmed by PCR. RNA concentration was measured by spectrophotometry at 260 nm. Reverse transcription-PCR (RT-PCR) of RNA was performed using conditions recommended by the manufacturer (SuperScript One Step RT-PCR with Platinum Taq; Invitrogen) with primer pairs (primer nucleotide sequences are available on request) corresponding to selected open reading frames as previously described (12). The semiquantitative determination of transcript levels by RT-PCR was performed with 1, 10, and 100 ng of total RNA from L. biflexa. The amplified products were analyzed by agarose gel electrophoresis. The assays were performed in triplicate. For real-time quantitative reverse transcription-PCR (qRT-PCR), RNA (800 to 1,500 ng) was reverse transcribed using random primers and reverse transcriptase as described in the manufacturer's instructions (Roche Diagnostics, GmbH, Mannheim, Germany). The cDNA was used as template for gene-specific primer pairs with a LightCycler FastStart DNA MasterPLUS SYBR Green (Roche Diagnostics) in a LightCycler apparatus (Roche Diagnostics). The thermal cycling conditions were as follows: 10 min at 95°C, followed by 45 cycles of 10 s at 95°C, 5 s at 60°C, and 13 s at 72°C, then 1 cycle of 5 s at 95°C, 15 s at 65°C, and 30 s at 40°C. Data were analyzed using RelQuant (Roche Diagnostics). In all cases, transcript levels were determined in duplicate and at least two independent RNA samples were used for each condition tested. The relative expression of target genes was normalized to the level of 23S rRNA as an endogenous control.
To mediate allelic exchange, a pGEM7Z-f+ (Promega) derivative plasmid was used for the construction of plasmids containing insertional inactivated genes. The process was as follows: PCR primers for the amplification of the kanamycin or spectinomycin resistance cassette and the left and right arms of the target gene were designed, and in each instance a restriction endonuclease site was introduced at each end of each PCR product. The resulting three PCR products were digested with the appropriate restriction endonucleases and ligated into the pGEM7Z-f+ derivative plasmid. The plasmid constructs delivering the inactivated allele were formed by insertion of a resistance cassette between the right and left arms (~0.5 kb in length) of the target gene; this introduces a partial gene deletion. The plasmids, which are not replicative in Leptospira spp., were then subjected to UV irradiation and used to deliver the inactivated alleles in L. biflexa as previously described (44). Kanamycin- or spectinomycin-resistant colonies were picked and tested for the insertion of the resistance cassette in the target gene by PCR as previously described (44). Random insertion mutagenesis using Himar1 was carried out in medium with or without hemin as previously described (35). We used RT-PCR assays to check that the mutations did not prevent transcription of genes downstream of the inactivated genes.
The sequenced strain, L. biflexa serovar Patoc strain Patoc1, was initially isolated from stream water (3) and maintained in the collection of the National Reference Center of Leptospira (Institut Pasteur, Paris, France). L. biflexa genomic DNA was randomly sheared by nebulization (hydroShear; GeneMachines) to short (1.5 to 2.5 kb) and long (35 to 45 kb) DNA fragments, and the insert DNAs were end repaired and ligated into a derivative of plasmid pGEM7-Zf+ (Promega) and fosmid pCC1FOS (Epicenter, Madison, WI), respectively. Sequencing reactions were performed, from both ends of DNA template, using an ABI PRISM BigDye Terminator cycle sequencing ready reaction kit and run on a 3700 or a 3730 xl Genetic Analyzer (Applied Biosystems) at the Genomics Platform (Pasteur Genopole Île-de-France). Phred (23), Phrap (26), and in-house software (cover and coverparse; unpublished data) were used for genome assembly. The complete genome sequence was obtained from 58,663 end sequences (giving >8× coverage). Annotation was done using MaGe (http://www.genoscope.cns.fr/agc/mage/) (59), which allows graphic visualization of the L. biflexa annotations enhanced by a synchronized representation of synteny groups in other genomes chosen for comparisons. Coding sequences (CDSs) likely to encode proteins were predicted with the AMIGene (10) and MICheck (18) software. Putative orthologs showed at least 30% identity and a minimum ratio of 0.8 to the length of the smallest protein. Each predicted gene was assigned a unique identifier prefixed with “LEPBIa” for the large chromosome, “LEPBIb” for the small chromosome, and “pLEPBI” for the 74-kb plasmid. TMHMM, version 2.0 (34), and PRED-TMBB (4) were used to identify putative transmembrane domains and β-barrel domains, respectively. Deduced amino acid sequences were also analyzed using the databases of Pfam protein families (6), membrane transport systems (http://www.membranetransport.org/) (47), and ABC systems (ABSCISSE v3.0 database; http://www.pasteur.fr/recherche/unites/pmtg/abc/database.iphtml) (11). The complete genomic sequence of L. biflexa serovar Patoc strain Patoc1 analyzed in this work will be published in another study.
Iron is essential for the growth of both saprophytic and pathogenic Leptospira spp. (24). In biological systems, iron is typically complexed to other molecules, and free iron is virtually absent. Leptospira spp. are usually cultivated in a standard albumin Tween 80 medium, the EMJH medium, containing iron sulfate as an iron source. To test for other iron sources, iron sulfate was omitted during the preparation of the EMJH medium. We find that whereas L. biflexa was not able to grow in iron sulfate-free EMJH, L. interrogans retained a wild-type growth. It is postulated that iron traces in iron sulfate-free EMJH were sufficient for the observed growth of L. interrogans. Further experiments were therefore done in iron-depleted EMJH that had been preincubated with 2,2′-dipyridyl. In these conditions, growth tests demonstrated that both L. biflexa and L. interrogans were able to use iron chloride, iron sulfate, and iron citrate. The most abundant source of iron in the host is heme and heme-containing proteins (62). Leptospira spp. were able to use exogenous hemin/hemoglobin as an iron source. Together with our previous study on an L. biflexa hemH mutant (27), this suggests that L. biflexa uses heme/hemoglobin as an iron as well as a heme source. Lactoferrin is another host protein that binds iron with high affinity, but neither L. biflexa nor L. interrogans was able to use lactoferrin as an iron source. Siderophores are high-affinity ferric chelators which are generally low-molecular-weight compounds synthesized and secreted by microorganisms in response to iron restriction. The secretion of siderophores by L. interrogans could explain its ability to grow in iron-depleted medium. However, analysis of the Leptospira genomes did not allow the identification of any genes that encoded proteins related to proteins involved in siderophore synthesis or siderophore secretion. In addition, L. biflexa failed to grow in the supernatant derived from an iron-depleted culture of L. interrogans. This result suggests that L. interrogans does not produce siderophore or that this siderophore is not utilized by L. biflexa. Leptospira spp. do not appear to synthesize siderophores; however, they could use exogenous siderophores of other microorganisms as an iron source. Among the hydroxamate-type siderophores, aerobactin and ferrichrome were used by both L. biflexa and L. interrogans, while desferrioxamine was only used by L. biflexa. Desferrioxamine is produced by the gram-positive Streptomyces pilosus (37) and is utilized by some gram-negative bacteria, such as Yersinia enterocolitica (8) or Vibrio vulnificus (65), and aerobactin is synthesized by Shigella spp. as well as some Escherichia coli clinical isolates, while the siderophore ferrichrome is synthesized by various fungal species. The catechol siderophore enterobactin is also produced by enterobacteria but was not utilized as an iron source by Leptospira spp.
The in vitro utilization of exogenous siderophores suggests that Leptospira spp. encounter the corresponding siderophores in their environment. While it is not surprising that Leptospira would use exogenous siderophores as an expedient way to acquire iron, it is not clear why saprophytes have the ability to use hemin and hemoglobin as an iron source.
Recently, the complete genome sequences of L. interrogans serovar Lai (48) and L. interrogans serovar Copenhageni (38) have been determined. The two genomes exhibit 95% identity at the nucleotide level and a further 99% identity for predicted protein-coding genes that are orthologs. The original annotations found that L. interrogans serovar Lai has nearly 1,000 more genes (4,727 open reading frames versus 3,658 CDSs) than L. interrogans serovar Copenhageni (38). However, this discrepancy may not reflect the reality and rather may be due to the annotation criteria used by the two genome projects (58). Using the MICheck software (18) on the two genomes, we predicted a total number of 3,798 and 3,651 CDSs for L. interrogans serovar Lai and L. interrogans serovar Copenhageni, respectively. Since all the homologous gene products of serovars Lai and Copenhageni share more than 95% identity, the term L. interrogans will therefore refer to both of the serovars. The complete genome sequence of the saprophyte L. biflexa serovar Patoc strain Patoc1 is also available and consists of approximately 3,790 predicted coding genes (M. Picardeau et al., unpublished data).
Iron participates as a metabolic cofactor in a variety of biochemical processes involving electron transfer. Genes encoding proteins that typically require iron as a cofactor, such as cytochromes, catalase, and tricarboxylic acid enzymes, were detected in both L. biflexa and L. interrogans. In contrast, a putative superoxide dismutase gene (LEPBIa0027) was only detected in L. biflexa. As previously shown, Leptospira spp. possess a complete heme biosynthesis pathway, including a ferrochelatase (27). A putative heme oxygenase was also found in the genomes of L. biflexa (LEPBIa0669) and L. interrogans (LB186/LIC20148). We have used comparative genomics to identify putative genes involved in iron acquisition systems and iron regulation in the genomes of L. biflexa and L. interrogans. We used the BLAST program and queried all L. biflexa protein sequences exhibiting significant similarities with iron transporters and iron regulators against the L. interrogans databases and vice versa. Genes encoding putative hemolysins, hemolysin excretion apparatus, and iron storage proteins were also included in this study (Table (Table11).
We have shown that Leptospira spp. were able to utilize various iron sources such as siderophores, but the uptake system mechanisms are not known. The acquisition of molecules by bacteria often relies on the active transport through dedicated outer membrane receptors. In gram-negative bacteria, transport of iron sources is mediated by outer membrane receptors, also called TonB-dependent receptors, which utilize the energy produced by the inner membrane complex of TonB, ExbB, and ExbD.
Bacteria often possess multiple TonB-dependent receptors, each providing the bacterium with specificity for different iron sources. For example, E. coli has one set of TonB, ExbB, and ExbD as well as eight TonB-dependent receptors (29). Other gram-negative organisms may have two or more TonB-ExbB-ExbD-like systems with distinct specificities for the TonB-dependent receptors (36). Analysis of the predicted proteins encoded by the L. biflexa and L. interrogans genomes revealed five putative ExbB and ExbD proteins and three putative TonB proteins. These genes are grouped in five distinct loci, and their arrangements are similar in both L. biflexa and L. interrogans. Consistent with an inner membrane localization, the putative TonB and ExbD proteins possess a single transmembrane segment (TMS) near the N-terminal end, and the putative ExbB proteins have three TMS (Table (Table1).1). Further analysis of the genomes reveals a total of 8 L. biflexa and 12 L. interrogans genes encoding proteins related to TonB-dependent receptors (Table (Table1),1), and it is notable that only one gene encoding a protein related to a TonB-dependent receptor (LEPBIa3017/LB279/LIC20214) is genetically linked to an ExbB-ExbD-TonB system. Despite low primary sequence similarity, the structure of TonB-dependent receptors is usually well-conserved in bacteria; all share a 22-stranded transmembrane β-barrel that forms a pore, which is plugged from the periplasm by a globular N-terminal domain (67). The 19 putative TonB-dependent receptors of L. biflexa and L. interrogans are 684 to 991 amino acids in length. Protein analysis using the Pfam database showed that they all exhibit a putative plug domain (E value from e-03 for LA3102 to e-24 for LA3021) and a TonB-dependent receptor domain (E value from e-03 for LEPBIa3354 to e-31 for LEPBIa3432). Further analysis predicts that these proteins are hydrophobic and have a putative large β-barrel domain typical of a TonB-dependent receptor. None of the putative TonB-dependent receptors of Leptospira possesses an N-terminal extension that could interact with anti-sigma factor, as in the E. coli fecA-fecIR system (15).
TonB-dependent receptors usually have high affinity and specificity for their substrates. In E. coli, the well-characterized TonB-dependent receptors FecA, FhuA, FepA, and BtuB are required for the active transport of ferric citrate, ferrichrome, enterobactin, and vitamin B12, respectively. However, substrate specificity is difficult to identify by sequence analysis of the receptors.
The construction or identification of mutants in each of the genes encoding TonB-dependent receptors and subsequent characterization of substrate specificity of the mutants is the preferred approach to the characterization of these receptors. Such an approach would not have been possible until the recent development of some key genetic manipulation techniques for Leptospira (35). By random transposon mutagenesis in L. biflexa, we recently identified mutants with insertions in a gene (LEPBIa1883) encoding a protein that shares homology with FecA (35). By screening a total of 6,000 L. biflexa transposon mutants for hemin auxotrophy, 11 out of 14 auxotrophic mutants exhibited the transposon inserted in LEPBIa1883 at distinct locations (data not shown). These strains were impaired in their ability to use iron citrate, iron chloride, iron sulfate, and aerobactin as an iron source (Table (Table2).2). Interestingly, aerobactin-like siderophores are derived from citrate (39), and therefore aerobactin and iron citrate share a similar structure that is evidently recognized by the same receptor. The gene product of LEPBIa1883 (642 amino acids in length) has 48% identity (E value, e-229) with the L. interrogans LA3468/LIC10714 product (650 amino acids in length). The reciprocal best BLAST hit test indicates that these proteins are orthologous and strongly suggests that they share the same function.
Besides fecA, we have attempted to disrupt the other putative genes encoding TonB-dependent receptors by allelic exchange in L. biflexa. Gene inactivation of pLEPBI0018, LEPBIa3432, LEPBIa3362, and LEPBIa3354 resulted in a wild-type phenotype in iron-depleted medium supplemented with different iron sources (Table (Table2).2). The lack of a phenotype in these mutants could be due to functional redundancy with another iron uptake system.
Disruption of LEPBIa2760 resulted in a mutant that was impaired in its ability to use desferrioxamine as an iron source (Table (Table2).2). Introduction of a recombinant plasmid harboring the LEPBIa2760 locus restored the ability of the mutant to use desferrioxamine (data not shown). Desferrioxamine is also utilized via a TonB-dependent receptor, called FoxA, in Yersinia enterocolitica (8). These results are evidence that LEPBIa2760 encodes the receptor protein for ferrioxamines in L. biflexa.
Finally, we failed to obtain double-crossover events in LEPBIa0500 and LEPBIa3017. This may indicate that these genes are essential for the survival of L. biflexa. Leptospira spp. have an absolute requirement for vitamin B12, which is usually transported via TonB-like systems in other gram-negative bacteria. Since vitamin B12 is a cofactor for enzymes of major biological processes, inactivation of its receptor should result in nonviable mutants. Amino acid comparisons of TonB-dependent receptors of heme, hemoglobin, siderophores, and vitamin B12 revealed a highly conserved domain containing the FRAP and NPNL amino acid box (14, 53). This conserved domain was found in some TonB-dependent receptors from Leptospira, including LEPBIa0500 (Table (Table1;1; Fig. Fig.11).
It is important to note that the current understanding about the mechanism of TonB-dependent transport across the outer membrane comes from studies in E. coli. Further studies are required to understand the precise physiological role of the TonB-dependent receptors of Leptospira spp. in iron uptake.
The TonB systems release Fe(III)-containing complexes into the periplasm, and these complexes are transported across the inner membrane via ATP-binding cassette (ABC) transport systems. In bacteria, the passive diffusion of Fe(II) through the outer membrane can represent a second source of iron; active transport of ferrous irons across the cytoplasmic membrane is normally distinct from the transport of ferric iron. In E. coli, uptake of Fe(II) across the cytoplasmic membrane is performed by the energy-driven high-affinity transporter FeoAB (32) and by the proton-dependent MntH transporter from the NRAMP (natural resistance-associated macrophage proteins) family (17).
The NRAMP family of membrane metal transporters was originally identified in eukaryotes and then in prokaryotes. Phylogenetic analyses of the NRAMP family of proteins suggested horizontal transfer from eukaryotes to bacteria (17). Characterized eukaryotic NRAMP proteins transport divalent cations such as iron, manganese, and zinc through the cytoplasmic membrane. In bacteria, these proteins have been most extensively characterized in enterobacteria where they act as manganese transporters, hence the designation MntH proteins. The Mycobacterium tuberculosis MntH also transports significant amounts of both iron and zinc (1). L. biflexa LEPBIa0902 displays a small but significant similarity to MntH proteins (Table (Table1).1). This protein contains 11 putative TMS, which is consistent with the predicted topology model of MntH proteins (28). However, most of the conserved residues found essential for the function of E. coli MntH (28), i.e., Asp-34, Glu-102, Asp-109, Glu-112, and Asp-238, were not found in LEPBIa0902. Disruption of LEPBIa0902 resulted in a wild-type phenotype in iron-depleted medium supplemented with different iron sources (Table (Table2),2), manganese, or zinc (data not shown). In conclusion, we do not have evidence on the role of LEPBIa0902 as a member of the NRAMP family.
In an earlier study, random transposon mutagenesis allowed us to identify an feoB-like gene in L. biflexa (35). By using the same methodology, we also identified an L. biflexa feoA mutant (Table (Table2).2). FeoA, a protein of approximately 75 amino acids in length, and FeoB were shown to be the only ferrous ion transport systems in both E. coli and Legionella pneumophila (32, 49). FeoB proteins are cytoplasmic membrane proteins that have two main regions: a hydrophilic N-terminal domain with GTPase activity and a hydrophobic C-terminal region with 7 to 12 transmembrane-spanning α-helices (FeoB in L. biflexa has 8 predicted TMS). L. biflexa feoA and feoB mutants were impaired in the uptake of ferrous/ferric iron, iron citrate, and the siderophores desferrioxamine, aerobactin, and ferrichrome (Table (Table2).2). This suggests that iron is released from the siderophores in the periplasm and then transported into the cytoplasm via FeoAB. Previous studies have demonstrated that FeoB was important for the infectivity of pathogenic bacteria such as Legionella pneumophila, Helicobacter pylori, and Shigella flexneri (49, 50, 60). The L. interrogans feoA and feoB genes (Table (Table1)1) may also have a major role in the acquisition of ferrous iron while in the host.
Besides the NRAMP-type and the FeoAB-type iron transporters, four distinct families of ABC transporters related to iron uptake are known (33). Their components can mediate the transfer of ferric iron, siderophores, heme, and vitamin B12 into the cytosol of prokaryotes. The ABC transporters are typically composed of periplasmic binding proteins, one or two identical or homologous permeases, and one or two ATPases located on the inner surface of the cytoplasmic membrane and supplying the system with energy. The ATPases are the most conserved modules among the ABC transporters, and the permeases are characterized by their overall hydrophobicity. Only a few ABC transport systems of the iron/metals type has been described in spirochetes (21, 30). By similarity searching against a database of characterized ABC transporters (ABSCISSE v3.0 database), the L. biflexa locus containing genes LEPBIa1264 to LEPBIa1266 is predicted to encode a MET (metallic cation uptake) family ABC transporter (11). No orthologous locus of this putative ABC transporter was detected in L. interrogans (Table (Table1).1). Inactivation of the gene encoding the ATP-binding protein, LEPBIa1265, had pleiotropic effects on growth in the presence of an Fe(III) source but not with Fe(II), i.e., iron sulfate (Table (Table22).
The characterization of an L. biflexa hemH mutant suggested that Leptospira can transport the entire heme molecule into the cell (27). Transport of heme across the cytoplasmic membrane is usually mediated via ABC transporters (62). In L. biflexa, but not in L. interrogans, four genes encoding a putative ABC transport system (genes pLEPBI0012 to pLEPBI0015; Fig. Fig.2)2) showed significant similarity to the corresponding proteins encoded by a well-defined hemin uptake system operon, hmuSTUV. Similar operons have been described in numerous pathogenic bacteria (57). Interestingly, pLEPBI0018, encoding a putative TonB-dependent receptor, is located immediately downstream of the putative hmuSTUV operon in the L. biflexa circular plasmid p74 (Fig. (Fig.2).2). In Yersinia pestis, the gene encoding the hemin receptor HmuR is also linked to the operon hmuSTUV. This genetic organization suggests that L. biflexa pLEPBI0018 encodes the hemin receptor. However, hemin uptake was not affected in the pLEPBI0018 mutant (Table (Table2).2). In addition, the alignment of putative leptospiral TonB-dependent receptor sequences (Fig. (Fig.1)1) reveals that a His residue, thought to be essential for heme binding in the HemR receptor from Y. enterocolitica (14), was not found in pLEPBI0018 and was only in the protein encoded by LA3149 (Fig. (Fig.11).
The ABC transporter proteins of the L. biflexa hmu locus exhibit similarities to Y. pestis HmuS, HmuT, HmuU, and HmuV proteins (Table (Table1).1). It was previously proposed that HmuS was a heme oxygenase (55), hence the designation of heme-degrading protein. However, the inactivation of either Y. pestis hmuS or Shigella dysenteriae shuS showed that these mutants were still able to use hemin and all hemoproteins as an iron source (57, 66). In addition, no heme oxygenase activity was detected for the HmuS homolog protein, ShuS, of S. dysenteriae (63). Similarly, an L. biflexa mutant of the hmuS-like gene grew like the wild-type strain when hemin was provided as the sole source of iron (Table (Table2).2). It has to be noted that the L. biflexa genome also contains another putative gene, LEPBIa0669, encoding a protein with 62% and 55% similarity to the heme oxygenase from Synechocystis sp. strain PCC6803 and human, respectively. Another hypothesis is that HmuS is involved in heme storage and/or the oxidative stress. Two types of iron storage proteins are also found in the genomes of Leptospira spp.: the heme-containing bacterioferritin and the Dps protein (Table (Table1).1). In E. coli, Dps does not have a strict function in iron storage, and it also protects DNA against the combined action of ferrous iron and hydrogen peroxide in the production of a hydroxy radical. In Y. pestis, HmuT is proposed to be a periplasmic binding protein which specifically binds heme and acts as a receptor for the active uptake of heme into the cytoplasm. The HmuU and HmuV proteins have been proposed to comprise the cytoplasmic membrane permease and ATPase, respectively. Transcriptional analysis by RT-PCR with a set of primers for the genes pLEPBI0012, pLEPBI0013, pLEPBI0014, and pLEPBI0015 revealed that these genes form an operon (data not shown). Inactivation of the putative hmuV (pLEPBI0013) by allelic exchange had no effect on hemin utilization (Table (Table2).2). This hemin uptake system was also found not to be essential for hemin utilization in Vibrio cholerae and Bradyrhizobium japonicum (40, 41). This might be explained by the presence of alternative mechanisms or low-affinity systems for transporting hemin across the cytoplasmic membrane.
Extracellular bacteria such as L. interrogans could release heme and hemoglobin from host red blood cells by the secretion of hemolysins, constituting a mechanism by which bacteria can gain access to ready sources of iron in the host. Once released, hemin may be rapidly bound to host proteins but may also be directly transported by bacteria or by binding of hemin or hemin complexes to TonB-dependent receptors (62). The genomes of Leptospira spp. contain several genes encoding putative hemolysins, even in the saprophyte L. biflexa (Table (Table1).1). There are at least five L. biflexa and eight L. interrogans genes that encode products that exhibit similarities to hemolysins (Table (Table1).1). In a recent study, Zhang et al. (68) demonstrated that the recombinant proteins encoded by the hemolysin genes of L. interrogans have hemolytic activities in E. coli (Table (Table1).1). In addition, a putative hemolysin secretion system similar to the E. coli-hemolysin (HlyA) secretion system (16) was identified in L. biflexa (LEPBIa0357-LEPBIa0360), but no orthologous system was found in L. interrogans (Table (Table1).1). The putative L. biflexa hemolysin secretion system comprises HlyB-, HlyD-, and TolC-related proteins. Feasibly, L. interrogans could use an alternative TolC-based secretion system, given the presence of a gene encoding a TolC-related protein (LA3927/LIC13135). Other genes encoding putative proteases may also be involved in degrading heme-containing compounds like in Porphyromonas gingivalis (54).
Bacteria typically regulate their metabolism in response to iron availability. In E. coli, most of the genes involved in iron acquisition are transcriptionally regulated by the ferric uptake regulator protein Fur (29). The expression of seven L. biflexa genes of interest, including four fur-related genes, the ferrous iron transporter gene (feoB), and two TonB-dependent receptors involved in iron transport (fecA and tbr3), were analyzed by quantitative reverse transcription-PCR (qRT-PCR). Our results were supported by the semiquantitative RT-PCR method (data not shown). No significant change in the relative expression of fur1 (LEPBIa2461) was determined in response to iron availability. Iron depletion led to a threefold increase in transcript levels of tbr3 (LEPBIa2760) encoding the receptor for desferrioxamine, whereas the other genes (fecA, feoB, fur2, fur3, and fur4) showed more than a 10-fold decrease in expression (Fig. (Fig.3).3). Interestingly, fecA and tbr3 have different transcription profiles. Under iron limitation, the expression of tbr3 is induced to utilize siderophores, which are high-affinity iron-chelating molecules. Conversely, the expression of fecA, which encodes a receptor for a relatively large number of iron sources (i.e., iron citrate, iron sulfate, iron chloride, and aerobactin), is more important in the presence of higher iron concentrations (Fig. (Fig.3).3). The expression of these genes involved in iron uptake is therefore sensitive to iron level, but expression is likely to be regulated by a complex regulatory mechanism in which Fur may not be the exclusive regulatory protein.
In the genomes of both L. biflexa and L. interrogans, four fur-like genes were identified (Table (Table1).1). On chelation of iron from the growth medium, the level of transcripts from the L. biflexa fur-like genes LEPBIa2152 (fur4), LEPBIa2330 (fur2), and LEPBIa2849 (fur3) decreased at least 10-fold, while the expression of LEPBIa2461 (fur1) was independent of iron concentration (Fig. (Fig.3).3). Among the Fur family of proteins, there are three types of proteins (Fur, Zur, and PerR) sharing high sequence identity. In E. coli, Fur regulates iron uptake and siderophore biosynthesis, Zur regulates zinc uptake systems, and PerR regulates oxidative stress response genes (29). We should find Fur, Zur, and PerR proteins in the genomes of Leptospira spp. Indeed, zinc and iron are important nutrients for Leptospira, and high concentrations of these metals are usually toxic for bacteria. A leptospiral PerR protein may also play an important role when Leptospira spp. encounter environmental oxidative stresses. All Fur family proteins share an N-terminal DNA-binding domain with a helix-turn-helix motif and conserved metal-binding sites (45). PerR regulators typically have two separate CXXC motifs in the C terminus (29). Such motifs were found in the leptospiral LB183, LEPBIa2152, and LEPBIa2849 Fur-like proteins. LEPBIa2849 also presents significant similarities with an oxidative stress regulator from Borrelia burgdorferi (52). In addition to iron-binding sites, E. coli Fur possesses one zinc-binding site composed of Cys93 and Cys96, which are also perfectly conserved in Zur proteins (25). Analysis of the Fur-like proteins from Leptospira spp. indicates that LEPBIa2461/LA1857 on one hand and LA2887 on the other hand do not contain this motif and therefore may not function as Fur or Zur proteins (Fig. (Fig.4).4). Since it is not possible to distinguish between Zur and Fur regulators on the basis of sequence, we have attempted to disrupt each gene by allelic exchange in L. biflexa to determine their biological role. Inactivation of LEPBIa2152 and LEPBIa2461 resulted in a wild-type phenotype in EMJH liquid medium (Table (Table2)2) as well as a wild-type peroxide sensitivity (data not shown). The lack of obvious phenotype in the LEPBIa2461 and LEPBIa2152 mutants could be due to functional redundancy with another member of the Fur family. The fur-like genes LEPBIa2330 and LEPBIa2849 showed different expression under iron-replete and -deplete conditions, and in the absence of allelic exchange mutants they appeared essential, as is the case in some other fur genes (20).
Fur-binding sites, known as the Fur boxes, were originally identified as a 19-bp inverted repeat sequence in the promoter region of iron-regulated genes. A 150-bp region preceding the start codon of putative iron-regulated genes from L. biflexa and L. interrogans such as fecA, feoAB, and fur-like genes were analyzed for the presence of putative Fur-binding sites (5). Although these genes are likely members of the Fur regulon, we were not able to identify a leptospiral Fur box. The DtxR protein family is another family of iron regulators that were first found in gram-positive organisms with a high-GC content and also in the spirochete Treponema pallidum (30). No DtxR-like proteins were detected in the genomes of Leptospira spp. Despite the large number of putative extracytoplasmic function sigma factors in both L. biflexa and L. interrogans, no FecI-related proteins, which are referred to as iron starvation sigma factors (15, 61), were identified. This suggests the absence of this type of signal transduction cascade in Leptospira. A high concentration of extracellular iron, which can be toxic for bacteria, can be detected by two-component systems (64). At least 47 potential response regulator genes were identified in the L. interrogans genome (2). This indicates that Leptospira spp. have developed a vast array of detection systems that enable them to respond to environmental signals, one of which could be iron.
The possession of specialized iron transport systems for the saprophyte L. biflexa and the pathogen L. interrogans may reflect the various iron sources they may encounter in their diverse habitats. A more detailed comparative genomic analysis should provide clues to the lifestyle of Leptospira in the environment and in the infected host, increasing our understanding of the transition from environmental bacterium to major human and animal pathogen (unpublished data).
Based on our findings, a model for iron uptake in Leptospira can reasonably be proposed (Fig. (Fig.5).5). The analysis of the genome of the pathogen L. interrogans has allowed the identification of 12 putative TonB-dependent receptors, while L. biflexa possesses 8 putative TonB-dependent receptors. This difference suggests that pathogenic species are able to use a wider range of iron sources than saprophytes. Alternatively, the pathogens may also present redundancy in their genome content. As in gram-negative bacteria, periplasmic binding proteins may shuttle iron-containing complexes from TonB-dependent receptors to cytoplasmic membrane ABC transporters that in turn deliver them in the cytoplasm. The pathogen L. interrogans may obtain iron from heme by secreting hemolysins to lyse red blood cells and thereby making the iron available for uptake. Surprisingly, L. biflexa, a nonpathogenic species, has putative hemin uptake and hemolysin secretion systems. Notably, the L. biflexa genome does not contain orthologs of the L. interrogans sphingomyelinases, which may be involved in the typical vascular damage seen in acute leptospirosis. Leptospira spp. also possess uptake systems that use siderophores produced by other bacteria or fungi. This ability to utilize xenosiderophores broadens the available sources of iron and allows the bacterium to occupy extended ecological niches. Bacterial iron homeostasis is best understood in E. coli, a bacterium phylogenetically distant from Leptospira. This study is a first step towards understanding iron acquisition systems in Leptospira. This study highlights the importance of mutagenesis tools, such as random transposon mutagenesis systems, to the molecular analysis of Leptospira.
This work was supported by Pasteur Genopole Île-de-France (Genomic Platform of the Institut Pasteur) and the French Ministry of Research ACI IMPBio (Genoscope). S.B. was supported by a grant from the Scientific Research FIL (Parma, Italy).
We thank Dieter Bulach for critical reading of the manuscript.
Published ahead of print on 15 September 2006.
†Supplemental material for this article may be found at http://jb.asm.org/.