Transcriptional Response of Leptospira interrogans to Iron Limitation and Characterization of a PerR Homolog

Culture conditions. L. interrogans sv Manilae was grown in EMJH medium (33) at 30°C. For the la1857 mutant, M776, kanamycin at a final concentration of 50 μg/ml was added. To study the effects of iron limitation, triplicate 25-ml cultures of the parent and la1857 mutant were grown to mid-log phase (4 × 10⁸ to 6 × 10⁸ cells/ml), and then the bacteria were pelleted (4,000 × g, 15 min, room temperature), washed in 10 ml of EMJH base (Difco), and resuspended in 1 ml EMJH base. The culture was then transferred to 100 ml EMJH medium with or without preincubation with 40 μM 2,2′-dipyridyl (Sigma) and grown to mid-log phase before harvesting of bacteria for RNA isolation.

RNA purification.Leptospires were grown to a density of 2.5 × 10⁸ to 6.5 × 10⁸ cells/ml before harvesting of bacteria for RNA purification. The cell count was determined as described previously (2). Cultures were harvested and RNA was isolated from leptospires as described previously (41) and then further treated with a Turbo DNA-free kit (Ambion), according to the manufacturer's instructions. To confirm lack of contamination with genomic DNA, 0.5 μg of RNA was used as template in a PCR using primers toward la0615, encoding a conserved hypothetical protein in serovars Lai and Manilae (Table 1).

Microarray construction, hybridization, and analysis.The L. interrogans microarray was printed as described previously (41). Labeled cDNA probes were synthesized from 2 μg of total RNA using a 3DNA Array 900MPX expression array detection kit (Genisphere) and then hybridized with the microarray slides, as described previously (41). Microarray hybridizations were scanned using a GMS 418 array scanner (Genetic MicroSystems). The Cy3 and Cy5 images were aligned and then overlaid with a grid using ImaGene, version 5.1 (Biodiscovery), to allow accurate gene identification and quantification of fluorescence intensity. Spots were examined manually, and poor spots with very low signals or inconsistent morphologies were flagged for elimination from the analysis.

Three independent RNA samples (biological replicates) from parent and mutant strains of L. interrogans serovar Manilae grown with or without 2,2′-dipyridyl were compared. The orientation of Cy3 to Cy5 labeling was the same for replicates 1 and 3, while replicate 2 was a dye swap. Raw data from the comparisons were analyzed using the web-based program BioArray Software Environment (BASE) (62). Raw data from replicate arrays (3 for each comparison) were combined and used for further analyses, as described previously (41). Genes were considered to be differentially expressed if they were at least 2-fold up- or downregulated with a P value of <0.05.

Validation of microarray data by real-time RT-PCR.Real-time reverse transcription-PCR (RT-PCR) was performed as described previously (41). The primers used are shown in Table 1. Known concentrations of L. interrogans serovar Manilae genomic DNA were used to construct a gene-specific standard curve so that the concentration of template in each reaction could be determined. The gene encoding flagellum subunit B, flaB, was used as the normalizer for all reactions. Melting curve analysis confirmed that all PCRs amplified a single product.

Statistics for category comparisons.Fisher's exact test was performed for pairwise comparisons of the frequencies of various groupings of leptospiral genes affected by iron limitation (Fig. 1).

• Open in new tab
• Download powerpoint

FIG. 1.

Percentages of genes which were up- or downregulated under iron limitation and across the L. interrogans sv Lai genome in each COG category. The COG functional categories are as follows: information storage and processing (11% of coding sequences in L. interrogans serovar Lai genome) (includes J, translation; K, transcription; L, replication, recombination, and repair); cellular processes and signaling (20% of coding sequences in the serovar Lai genome) (includes D, cell cycle control, cell division, and chromosome partitioning; V, defense mechanisms; T, signal transduction mechanisms; M, cell wall, membrane, or envelope biogenesis; N, cell motility; U, intracellular trafficking, secretion, and vesicular transport; O, posttranslational modification, protein turnover, and chaperones); metabolism (19% of coding sequences in the serovar Lai genome) (includes C, energy production and conversion; G, carbohydrate transport and metabolism; E, amino acid transport and metabolism; F, nucleotide transport and metabolism; H, coenzyme transport and metabolism; I, lipid transport and metabolism; P, inorganic ion transport and metabolism; Q, secondary metabolite biosynthesis, transport, and catabolism); poorly characterized (50% of coding sequences in the serovar Lai genome) (includes R, general function prediction only; S, function unknown; and −, not in COGs). The asterisk indicates that genes predicted to encode inorganic ion transport and metabolic proteins were overrepresented among genes upregulated by low-iron conditions (Fisher's exact test, P < 10⁻⁵).

Phylogenetic analysis.Sequence alignment and phylogenetic analysis were performed using the software Geneious Basic, version 4.7.5, to compare similarity between each of the putative L. interrogans Fur homologs with other Fur, Zur, and PerR proteins.

Oxidative stress assays.Three biological replicates of mutant and wild-type cultures were tested in triplicate for sensitivity to hydrogen peroxide and cumene hydroperoxide. Hydrogen peroxide and cumene hydroperoxide were serially diluted using EMJH medium in a 96-well plate from 6.87 mM to 6 μM and 5 mM to 3 μM, respectively, in a 100-μl final volume. The mutant or wild-type cultures (5 × 10⁶ cells in 100 μl EMJH medium) were added to the wells, and the plates were incubated overnight at 30°C. The minimum bactericidal concentration (MBC) was then determined by dark-field microscopy.

Microarray data accession number.The data discussed in this publication have been deposited in the NCBI Gene Expression Omnibus database and are accessible through GEO series accession number GSE20422 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20422 ).

Effects of iron limitation on transcription.We identified 49 and 43 genes which were downregulated and upregulated in response to low iron, respectively (Table 2 ). While the microarray data were generated using L. interrogans sv Manilae, for gene nomenclature we have used locus tags from L. interrogans sv Lai (LA numbers), except where the gene is unique to sv Copenhageni (LIC numbers) and is present in our draft sequence of the sv Manilae genome. Therefore, while some genes have been predicted to be part of a transcriptional unit in sv Lai and/or sv Copenhageni, it is unknown whether sv Manilae has a similar genome arrangement.

To determine if any groups of genes were overrepresented, the differentially expressed genes were sorted into functional categories on the basis of clusters of orthologous groups (COG) (66) (Fig. 1). Consistent with the high proportion of uncharacterized genes (∼50% of coding sequences in the L. interrogans genomes), the majority of differentially expressed genes encode proteins of hypothetical or uncharacterized function (56% and 53% of down- and upregulated genes, respectively, in COG categories R, S and no COG designation) (Fig. 1).

Growth under low-iron conditions did not affect expression of genes in COG categories F (nucleotide transport and metabolism), N (cell motility), O (posttranslational modification, protein turnover, and chaperones), or U (intracellular trafficking, secretion, and vesicular transport), suggesting little change to these cellular processes (Fig. 1). There were no upregulated genes in COG categories D (cell cycle control, cell division, and chromosome partitioning), M (cell membrane biogenesis), and Q (secondary metabolites biosynthesis, transport, and catabolism); but 16.7% of downregulated genes were in these categories (versus 7.1% across the genome; P = 0.012), possibly to divert cellular resources toward expression of essential proteins which are involved in iron acquisition and active transport. Genes assigned COG category P (inorganic ion transport and metabolism) were overrepresented in genes which were upregulated compared with the genome-wide frequency (18.6% and 2.4%, respectively; P < 10⁻⁵), a finding consistent with COG category P, including proteins predicted to be involved in transport of iron.

L. interrogans genes involved in iron transport and utilization.Expression of the gene encoding LA0572, predicted to be involved in transport of ferrienterobactin (55), and expression of the gene encoding LA2242, a desferrioxamine receptor homolog, were upregulated (Table 2), similar to what was found in L. biflexa (44). Expression of la2242 is also induced by transition from low to physiological osmolarity conditions (48) and exposure to serum (57).

FecA is a surface protein which mediates transport of ferric citrate across the outer membrane (72). The gene encoding LA3468, a FecA homolog, was induced 2.2-fold in L. interrogans under low-iron conditions. In E. coli, the ferric iron transport genes (fecABCDE) are also upregulated under low-iron conditions (5). However, the ABC transport proteins involved in ferric citrate uptake by FecA in L. interrogans are unknown, since no other genes of the fec operon have been identified and the genes encoding the two-component signal transduction genes, fecIR (67), are also absent.

LB191 (or HbpA) from L. interrogans sv Lai is a hemin binding protein which is expressed in cultures grown under low-iron conditions (6). Our study confirmed upregulation of lb191 transcription under iron limitation. Genes encoding a heme oxygenase (LB186) and a putative heme permease (LB187) were also upregulated. The genes appear to be arranged in an operon and together with lb191 may be part of a heme acquisition and utilization locus. Since heme is likely to be the main source of iron in the host, upregulation of these genes was not surprising, especially given that heme oxygenase is required by L. interrogans for survival in the hamster model of infection (52). Genes encoding LB186 and LB187 were also upregulated in response to an increase in osmolarity (48) and exposure to serum (57), consistent with the importance of these genes for survival in the host.

Genes encoding predicted ferrous iron transport proteins, FeoA and FeoB (LA2578 and LA2579, respectively), were not differentially expressed under low-iron conditions, in contrast to the downregulation observed in L. biflexa (44). This difference may be due to differences in metabolism or iron uptake mechanisms between pathogenic and saprophytic leptospires. Iron transporter proteins require an energy transduction complex, consisting of TonB, ExbB, and ExbD, for uptake of iron (3). There are two predicted TonB or TonB-like proteins encoded in the L. interrogans genome, five ExbD-related biopolymer transport proteins, and four ExbB- or TolQ-like transport proteins grouped in four distinct loci. TolQ is structurally similar to ExbB and can substitute for ExbB activity (10, 21). None of these was differentially expressed, although la3246 (encoding a predicted ExbD-related biopolymer transport protein) was slightly upregulated (1.53-fold) under low-iron conditions. The constitutive expression of these genes suggests that the proteins are involved in transport of substances other than iron or their expression may respond to different iron sources. The gene bfr, encoding a putative bacterioferritin (LA2690), was 3.1-fold downregulated, consistent with the reduced need for iron storage proteins and other nonessential iron-containing proteins in order to release iron for crucial cellular processes. There was no change in transcription of genes encoding the other predicted bacterioferritins: Dps (LA3598), a DNA-binding ferritin-like protein, or LA4021, a predicted bacterioferritin-associated ferredoxin. In E. coli, Fur indirectly represses expression of bacterioferritin under iron-limiting conditions via derepression of RyhB, a small noncoding RNA which acts by hybridizing to the transcript and inhibiting translation (46). RyhB also negatively regulates a number of genes encoding iron-binding and metabolic proteins and thus, along with Fur, plays a major role in iron regulation in E. coli (47). Downregulation of bacterioferritin (LA2690) and other genes in L. interrogans may be regulated by a similar mechanism. However, there have been no small RNA species characterized in Leptospira spp. to date.

In some bacteria, Fur autoregulates its own transcription in response to iron levels (20, 30). This occurs via RyhB in E. coli (69). While transcription of the genes encoding Fur family proteins was not altered in L. interrogans under low-iron conditions (see Table S1 in the supplemental material), it is possible that their expression may also be regulated posttranscriptionally by small RNA species. Expression of the la1857 ortholog in L. biflexa (LEPBIa2461) was likewise unaffected by the iron concentration, but the other three fur-like genes were downregulated at least 10-fold under iron limitation conditions (44). The reasons for this discrepancy are unknown but further suggest different regulatory mechanisms in pathogenic and saprophytic Leptospira.

Effect of iron limitation on membrane proteins and putative virulence factors.Various leptospiral outer membrane proteins are downregulated by iron, including LipL36 (LA0492), pL50 (LA3469), and pL24 (LA0412), a putative electron transfer flavoprotein (17). LA3469 (LruB) plays a role in equine recurrent uveitis and has an IrpA domain which is associated with iron metabolism regulation (70). Our results showed a 4.6-fold upregulation of la3469 with iron limitation. Also, in contrast with the findings of the study of Cullen et al. (17), we detected no change in transcription of la0412 or lipL36. The apparent discrepancy may be due to the differences in 2,2′-dipyridyl concentrations and the period of growth under iron-limiting conditions. However, we have observed previously that transcription levels do not necessarily reflect protein abundance (41, 42). LipL36, for example, is downregulated in vivo and during growth at 37°C (17, 28, 53), but our studies showed that lipL36 transcription was not temperature responsive, suggesting posttranscriptional regulation (41).

LA0695 (also known as Lsa24 [7], LenA [64], or LfhA [71]) is expressed during mammalian infection and can bind to the complement regulatory protein factor H (71), laminin, collagen IV, and fibronectin (7), suggesting roles in resistance to complement-mediated killing and adhesion to host cells. LA0695 is one of five L. interrogans proteins (LenA-LenF) that are structurally similar to mammalian endostatins (64). Previous studies showed that the transcript of la0695 was slightly upregulated by an increase in osmolarity (1.55-fold) (48), whereas there was no change in response to temperature (41). However, for unknown reasons, there was poor hybridization and therefore a low signal intensity of la0695 spots on the microarray slides, resulting in inconclusive data for la0695 transcripts. Given that LA0695 is likely to play a role in virulence, we selected this gene for real-time RT-PCR analysis and found that la0695 was upregulated 2.8-fold. Genes encoding two other endostatin-like proteins, lenB (LA3103) and lenD (LA1433), were slightly upregulated (1.6- and 1.8-fold, respectively) by iron limitation, whereas lenC, lenE, and lenF (la0563, la4324, and la4073, respectively) were not differentially expressed (see Table S1 in the supplemental material). lenD is also upregulated under conditions of increased temperature and osmolarity (41, 48). LenB has factor H-binding ability, while both LenB and LenD can bind to laminin (64). The Len proteins appear to provide functional redundancy, whereby expression of different proteins performing similar roles may be regulated by different mechanisms and/or environmental signals, as suggested by results of this and previous microarray studies (41, 48). Functional redundancy and differential regulation suggest that the individual Len proteins may play other as yet undefined roles in different stages of pathogenesis.

Using iTRAQ and 2DGE analyses, Eshghi et al. (22) identified five proteins in L. interrogans sv Copenhageni which were upregulated under in vivo-like conditions (iron limitation combined with the presence of serum): the essential virulence factor Loa22 (61), a putative glycosyl hydrolase (LIC13050/LA0505), a putative coagulase (LIC13166/LA3961), a putative catalase (KatE or LIC12032/LA1859), and a TolC-like outer membrane protein (LIC12575/LA1100). However, the corresponding genes were not differentially expressed in our study, which examined iron-limiting conditions alone.

We identified 20 genes which encode proteins with no similarity to proteins found in other bacteria with increased transcription under low-iron conditions. Of these, 16 were absent from the genome of the saprophyte L. biflexa (Table 2) (58) and may encode unique leptospiral virulence factors.

Genes involved in iron acquisition are transcriptionally regulated by iron availability through the action of Fur in many species of bacteria. Louvel et al. (45) were not able to identify putative Fur boxes in the promoter regions of genes predicted to be involved in iron uptake. Likewise, we also were unable to identify a clear consensus Fur box for genes upregulated by low iron.

Genes with altered expression in the la1857 mutant.The genome of L. interrogans encodes four predicted Fur homologs, LA1857, LA2887, LA3094, and LB183; we have constructed a transposon insertion mutation in la1857 (51) with the insertion 62 bp into the 438-bp gene. Global transcription of the M776 mutant strain was compared with that of the wild-type serovar Manilae parent, both of which were grown under iron-replete conditions. If LA1857 does indeed function as a ferric uptake regulatory protein, we expected that its inactivation would have the same effect as growth under iron limitation. However, our results showed that this was not the case. Genes differentially expressed in the LA1857 mutant did not correspond to the iron-regulated genes in the wild-type strain, with only two downregulated and one upregulated gene being common to both data sets. Therefore, LA1857 does not play a significant role in regulation of iron homeostasis. Repeated attempts to complement the mutation were unsuccessful, but given the >1-kb intergenic region between la1857 and the next gene, there are unlikely to be downstream polarity effects.

We identified 11 and 20 genes which were down- and upregulated, respectively, in the mutant strain (Table 3). The microarray data were validated by real-time RT-PCR analysis of 13 genes, yielding a correlation coefficient (R²) of 0.7638 (data not shown). In the la1857 mutant, expression of LA1857 was negligible (Table 3) and likely represents nonspecific background fluorescence.

The Fur family of metalloregulatory proteins, Fur, Zur (zinc uptake regulator), and PerR (a peroxide stress regulator), are structurally similar and share high sequence similarity (29). Multiple Fur homologs have been described in other bacterial species; Campylobacter jejuni has two (Fur and PerR) and B. subtilis has three (Fur, Zur, and PerR) (11, 25, 68). In B. subtilis, while the paralogs are structurally similar, each performs distinct functions with very little or no regulon overlap (24). Likewise, the four predicted Fur homologs in L. interrogans may play distinct regulatory roles. B. burgdorferi does not require iron for growth, yet it possesses a Fur homolog, BosR, with 54.3% similarity to the B. subtilis PerR (59), which regulates dps, hemA, katA, and mrgA, all of which are involved in responses to oxidative stress and metal limitation (11, 14). In C. jejuni, PerR regulates the katA and ahpC genes (68). Therefore, on the basis of our microarray data (see below), LA1857 appears to be functionally more similar to PerR and is likely to function as a derepressor of genes encoding proteins involved in the oxidative stress response.

Catalase (KatA) and alkyl hydroperoxide reductase or peroxiredoxin (AhpC) are important bacterial enzymes in defense against oxidative stress. The katE gene, encoding a putative catalase, was upregulated 2.4-fold in M776. However, expression of ahpC (LA2809) was not altered. A gene encoding a putative cytochrome c peroxidase (mauG-2 or la0666) was also upregulated in the mutant. Cytochrome c peroxidase plays a role in protection against peroxide killing in Neisseria gonorrhoeae (63). Resistance to oxidative stress is likely to be important in the early stages of infection, allowing pathogenic leptospires to persist in phagocytes and disseminate (18). Genes in the heme biosynthesis pathway (hemCD, hemB, and hemL) were also upregulated more than 2-fold in the mutant strain. The first gene in the putative heme synthesis operon, hemA, was slightly upregulated at 1.8-fold (P < 0.001) (see Table S1 in the supplemental material).

Phylogenetic analysis showed that LA1857 and LA2887 are indeed more similar to the B. subtilis PerR protein and BosR from B. burgdorferi, while LB183 and LA3094 are more similar to the E. coli Zur and Fur proteins, respectively (Fig. 2). B. burgdorferi BosR regulates the oxidative stress response and interfaces with the RpoS cascade and subsequent expression of known virulence factors; inactivation of BosR resulted in loss of infectivity in mice (31, 32, 56). However, the M776 mutant strain was not attenuated for virulence in the hamster model of infection (51), a finding that is consistent with derepression of oxidative stress genes potentially regulated by LA1857 and that indicates different roles for LA1857 and BosR.

• Open in new tab
• Download powerpoint

FIG. 2.

Phylogenetic analysis of L. interrogans Fur homologs and other Fur family proteins generated by neighborhood joining with 100 bootstrap replicates and rooted at the midpoint. Branch support values are shown as percentages.

In B. subtilis, the Fe(II)-bound form of PerR is highly sensitive to metal-catalyzed oxidation by bound ferrous ions, resulting in Fe release and loss of DNA binding activity, leading to derepression of oxidative stress genes (38). Accordingly, we also observed that expression of katE was further increased by 4.3-fold in the mutant versus wild type when both were grown under low-iron conditions (see Table S1 in the supplemental material). In many bacteria, expression of KatA and AhpC is regulated by OxyR in response to oxidative stress (15). OxyR is an activator of peroxide stress genes and is not a metalloprotein. Instead, hydrogen peroxide catalyzes disulfide bond formation in OxyR, allowing it to bind to and induce expression of peroxide stress genes (73). Since genes encoding OxyR and SoxRS are absent from L. interrogans genomes (54, 60), we hypothesize that LA1857 is a PerR-like regulator and a functional analog of OxyR, although they clearly have different modes of action. C. jejuni also lacks oxyR, and the peroxide stress response is regulated by PerR (68). However, unlike C. jejuni, katE expression in L. interrogans was not iron responsive (see Table S1 in the supplemental material), and given that there was little overlap in the low-iron and mutant microarray data sets, regulation of responses to iron limitation and peroxide stress appear to be distinct. Since LA1857 is a Fur-like protein, divalent metal ions would be required to activate DNA binding activity. A number of residues (H37, D85, H91, H93, and D104) have been shown to be important for divalent metal binding in the B. subtilis PerR (38). An alignment of LA1857 and the B. subtilis PerR showed that the metal binding residues are also present in LA1857 (Fig. 3).

• Open in new tab
• Download powerpoint

FIG. 3.

Alignment of amino acid sequences of LA1857 and B. subtilis PerR showing identical residues (*), conserved residues (:), and semiconserved residues (.). Residues shown to be important for divalent metal binding (Fe²⁺ or Mn²⁺) in B. subtilis PerR are also present in LA1857 (boxed).

While inactivation of la1857 significantly increased the expression of katE under iron-replete conditions, expression of the la1858 gene immediately upstream of katE was not altered in the mutant. The la1858 and katE genes in serovar Manilae overlap by 1 nucleotide, suggesting transcriptional coupling. However, analysis of the sequence upstream of katE revealed a putative PerR binding box with 13/15 matches to the consensus sequence (24), consistent with our hypothesis that LA1857 is a PerR homolog and regulator of katE.

Inactivation of PerR in C. jejuni results in hyperresistance to peroxide stress (68). Since mutation of la1857 resulted in derepression of katE and mauG-2 (Table 3), to determine whether this also results in increased resistance to oxidative stress, sensitivity to hydrogen peroxide between the mutant and wild-type strains was compared. The mutant was 8-fold more resistant to hydrogen peroxide; the MBC for the la1857 mutant was 430 μM, while the wild type was killed at 54 μM (Fig. 4). This finding is consistent with the higher levels of katE expression seen in the mutant, and therefore, LA1857 does indeed play a role in the oxidative stress response. However, we did not observe any difference in resistance to cumene hydroperoxide (Fig. 4). In Streptococcus pyogenes, PerR also regulates the peroxide stress response, but unlike C. jejuni and B. subtilis, the S. pyogenes PerR does not regulate ahpC (34). Glutathione peroxidase (GpoA) and AhpC are not required in the induced response to peroxide stress in S. pyogenes but do contribute to resistance against cumene hydroperoxide. Likewise, in L. interrogans PerR did not regulate ahpC (LA2809), which was not differentially expressed in the mutant versus the wild type. In addition, genes encoding other AhpC homologs (LA0734, LA0862, LA2312, LA2494, LA2809, LA3442, and LB117) and two GpoA homologs (LA1007 and LA4299) were also not differentially expressed in the LA1857 mutant, which may explain the lack of difference in resistance to cumene hydroperoxide between the LA1857 mutant and the wild type. The expression behavior of the ahpC and gpoA homologs in wild-type L. interrogans grown under peroxide stress conditions awaits further study.

• Open in new tab
• Download powerpoint

FIG. 4.

Minimum bactericidal concentrations of hydrogen peroxide and cumene hydroperoxide for the LA1857 mutant (Mut) and wild type (WT). Three biological replicates were tested in triplicate, yielding identical results. The LA1857 mutant was 8-fold more resistant to killing by hydrogen peroxide, whereas there was no difference in resistance to cumene hydroperoxide.

It would be interesting to examine the roles of the other Fur homologs in L. interrogans. However, directed mutagenesis is extremely inefficient in pathogenic leptospires, with only two targeted mutations in a pathogenic species of Leptospira being reported to date (16, 39). Ongoing screening of our transposon library has not yet yielded mutations of the other Fur homologs. It is thus not possible to inactivate each of the homologs individually and/or in combination and study the subsequent effect on transcription patterns. This is the first report of peroxide stress regulation by a Fur-like protein in L. interrogans. However, since there are four predicted Fur homologs in L. interrogans svs Lai and Copenhageni, we suggest that each plays a distinct role in iron, metal, or stress responses.