ABSTRACT
Transmission of pathogenic Leptospira between mammalian hosts usually involves dissemination via soil or water contaminated by the urine of carrier animals. The ability of Leptospira to adapt to the diverse conditions found inside and outside the host is reflected in its relatively large genome size and high percentage of signal transduction genes. An exception is Leptospira borgpetersenii serovar Hardjo, which is transmitted by direct contact and appears to have lost genes necessary for survival outside the mammalian host. Invasion of host tissues by Leptospira interrogans involves a transition from a low osmolar environment outside the host to a higher physiologic osmolar environment within the host. Expression of the lipoprotein LigA and LigB adhesins is strongly induced by an upshift in osmolarity to the level found in mammalian host tissues. These data suggest that Leptospira utilizes changes in osmolarity to regulate virulence characteristics. To better understand how L. interrogans serovar Copenhageni adapts to osmolar conditions that correspond with invasion of a mammalian host, we quantified alterations in transcript levels using whole-genome microarrays. Overnight exposure in leptospiral culture medium supplemented with sodium chloride to physiologic osmolarity significantly altered the transcript levels of 6% of L. interrogans genes. Repressed genes were significantly more likely to be absent or pseudogenes in L. borgpetersenii, suggesting that osmolarity is relevant in studying the adaptation of L. interrogans to host conditions. Genes induced by physiologic osmolarity encoded a higher than expected number of proteins involved in signal transduction. Further, genes predicted to encode lipoproteins and those coregulated by temperature were overrepresented among both salt-induced and salt-repressed genes. In contrast, leptospiral homologues of hyperosmotic or general stress genes were not induced at physiologic osmolarity. These findings suggest that physiologic osmolarity is an important signal for regulation of gene expression by pathogenic leptospires during transition from ambient conditions to the host tissue environment.
Leptospirosis is a common zoonosis, especially in tropical regions of the world where contaminated water serves to facilitate transmission. Pathogenic leptospires frequently produce a nearly asymptomatic infection in maintenance host animals. After being shed in the urine, most Leptospira species are able to survive for long periods of time in moist soil and natural bodies of water despite nutrient-poor conditions (8, 11, 45). The importance of water as a vehicle for leptospiral transmission is emphasized by well-documented outbreaks of leptospirosis after urban flooding and in association with water sports (29, 39). Upon contact with a mammalian host, the bacteria disseminate hematogenously to many organs, including the central nervous system, lungs, and liver, eventually reaching the proximal renal tubules. In maintenance hosts such as the rat, infection results in long-term renal tubular colonization with limited impairment in kidney function. However, in humans and other accidental hosts, a combination of leptospiral virulence factors and host inflammatory responses results in hepatorenal failure, pulmonary hemorrhage, or death in 5 to 25% of severe cases (6, 38).
An exception to the typical leptospiral life cycle is Leptospira borgpetersenii serovar Hardjo, which is transmitted by direct contact with contaminated body fluids and fails to survive in nutrient poor environments, compared to Leptospira interrogans (7). Recent comparative genomic analysis revealed that L. borgpetersenii is undergoing genomic decay relative to L. interrogans. Most notably, many genes encoding environmental sensing and metabolite transport and utilization functions were impaired in L. borgpetersenii but functional in L. interrogans. The clinical features of leptospirosis observed in L. borgpetersenii and L. interrogans infections are similar, although the latter is usually associated with more severe infections. Thus, L. borgpetersenii appears to be selectively losing genes that encode products necessary for survival outside of the host while retaining genes required for virulence and survival in mammalian hosts.
Outer membrane proteins (OMPs) are likely to play important roles in the interaction of bacterial pathogens with the mammalian host. Three types of leptospiral OMPs have been described: transmembrane OMPs, peripheral membrane proteins, and lipoproteins (13). Of the three types of OMP genes, those encoding lipoproteins are the most abundant in leptospiral genomes (42, 49, 52), and several have been demonstrated to interact with host molecules in vitro. LigA, LigB, and Lsa24 mediate adhesion of L. interrogans to various components of the extracellular matrix (4, 9a). Lsa24 (LfhA) is also a factor H-binding protein (59). Although the functions of the other leptospiral lipoproteins remain largely unknown, several additional surface-exposed outer membrane lipoproteins that may interact with the extracellular environment have been described (14, 15, 37, 53). Some of these proteins are constitutively expressed during cultivation in artificial medium and are detected during infection of the mammalian host (5, 19, 24, 25). However, other OMPs have been shown to be differentially regulated during infection (1, 36, 46).
There is growing evidence that expression of many leptospiral genes, including those that encode OMPs, are regulated by temperature. For example, the expression of LipL36 is down-regulated when the temperature is shifted from 30°C to 37°C, consistent with the inability to detect the lipoprotein in leptospires residing in the kidneys of infected hamsters (12, 26, 41). In contrast, the peripheral membrane protein Qlp42 (P31LipL45) was up-regulated by the temperature shift (40). In two recent global analyses of the response of transcript levels to temperature, subsets of genes that are induced or repressed when L. interrogans cultivated at 20° or 30°C is shifted to temperatures found in the mammalian host were identified (33, 48).
Additional environmental factors are undoubtedly important in inducing the appropriate changes in leptospiral gene expression during host infection. For example, expression of the LigA and LigB adhesins is strongly induced when the osmolarity of the medium is raised to the level found in mammalian host tissues, suggesting that osmolarity is an important environmental signal that can alter leptospiral gene expression during the transition from the ambient environment to host tissues (36). For this reason, we used whole-genome microarrays to examine global changes in transcript levels in response to an upshift in osmolarity mediated by sodium chloride to better understand the changes that L. interrogans may undergo during the early stages of host invasion. Although previous microarray studies have examined the effects of hyperosmolarity on gene expression (2, 9), to our knowledge, the study presented here is the first to describe the global effects of physiologic osmolarity, which is more relevant to understanding the changes in gene expression that occur during infection. Our results reveal specific changes in gene transcript levels that reflect entry of leptospires from an environmental source into a mammalian host. Importantly, several genes that were induced by increased osmolarity were also increased by temperature shifts, suggesting that temperature and osmolarity work in concert to induce genes critical for successful development of mammalian infection. Moreover, genes that were repressed at physiologic osmolarity were much more likely than other genes to be missing or nonfunctional in L. borgpetersenii, suggesting that physiologic osmolarity is a key signal for the environment-to-host transition.
MATERIALS AND METHODS
Microarray construction.The leptospiral microarray was printed by the Australian Genome Research Facility, as described previously (33). In brief, oligonucleotides of 70 bases in length were designed for each of the 3,666 open reading frames (ORFs) of L. interrogans serovar Lai, strain 56601, and the 45 ORFs unique to L. interrogans serovar Copenhageni, strain Fiocruz L1-130. In addition, two oligonucleotides were designed for the 45 longest ORFs and the 46 pseudogenes found in the Lai strain. Forty-three of the 45 longest ORFs and 44 of the 46 Lai pseudogenes are intact in the Copenhageni strain. Each oligonucleotide was printed in pairs in 16 subarrays, each with a 22-column by 24-row configuration. The 16 subarrays were printed in duplicate so that each oligonucleotide was present on the slide in quadruplicate.
Culture conditions.The virulence of L. interrogans serovar Copenhageni strain Fiocruz L1-130 was maintained by passage through Golden Syrian hamsters and maintained in EMJH medium supplemented with 1% rabbit serum and 100 μg/ml of 5-fluorouracil (Sigma, St. Louis, MO) (17, 35). Cultures were grown in triplicate for 20 h at 30°C in EMJH or EMJH supplemented with 120 mM NaCl, which raised the osmolarity from ∼67 mOsM to ∼300 mOsM (calculated), which is the osmolarity in rat and human tissues (32, 56). Cultures were grown to a density of 8 × 107 to 3 × 108/ml before harvesting for RNA purification.
RNA purification and microarray hybridization. L. interrogans cultures were rapidly chilled in a dry ice-ethanol bath, and bacteria were collected by centrifugation in a Sorvall SS-34 rotor at 4°C. RNA was isolated from bacteria with TRIzol reagent (Invitrogen, Carlsbad, CA) as described previously (33). Labeled cDNA probes were synthesized from 2 μg leptospiral RNA and hybridized to the microarrays as described previously (33).
Analysis of microarray images.Microarray hybridizations were scanned with the Agilent DNA microarray scanner (Agilent Technologies, Foster City, CA) located in the UCLA Microarray Core Facility, and the fluorescence intensities of the spots from the Cy3 and Cy5 images were quantitated with ImaGene, version 5.1 (Biodiscovery, El Segundo, CA).
The comparison between L. interrogans grown at low and physiologic osmolarity had three biological replicates with a dye swap for each replicate, resulting in six arrays. Raw data from the direct comparisons between ∼66 mOsM and ∼300 mOsM were analyzed using the web-based program BioArray Software Environment (51). Spot-specific median signals were corrected for local background by subtracting spot-specific median background intensities. Spots with very low intensities (<250) were omitted. The data for each array were normalized independently using the global median ratio, which scales the intensities such that the median of the ratio between the Cy3 and Cy5 channels, with spots within 5% of the highest or lowest intensities omitted, is 1. MA plots of the normalized data revealed differences among the curves of the individual print-tip groups. Therefore, each array was further normalized using print-tip loess normalization (61), followed by scaling among the replicate arrays such that the distributions of their intensities had the same median absolute deviation. The P values associated with moderated t test, which borrows information from all other genes to obtain better estimates of the variances, were calculated using empirical Bayes (54). The P values were then adjusted for multiple testing by controlling the false discovery rate (54). Genes with an absolute relative ratio of greater than 1.5-fold and significance at P values of <0.05 were selected as differentially expressed.
Validation of microarray data by real-time RT-PCR.Eight leptospiral genes representing a broad range of differential expression were selected for real-time reverse transcriptase PCR (RT-PCR) analysis. Primer Premier software version 5 (Premier Biosoft International, Palo Alto, CA) was used to design primers for real-time RT-PCR (see Table S1 in the supplemental material). RT-PCR was conducted with new batches of RNA. RT reaction mixtures contained 2 μg of total RNA, 10 μM concentrations of random nonamer primers (Sigma), 10 U of Superscript II (Invitrogen) reverse transcriptase, and 500 μM concentrations each of dATP, dCTP, dGTP, and dTTP. Samples were incubated at 42°C for 1 h, followed by 10 min at 70°C. The synthesized cDNA was then diluted 1/200 prior to use in real-time RT-PCR. Reactions were performed in triplicate. Each 50-μl reaction mixture contained 10 μl of cDNA, 0.4 μM concentrations of each gene-specific primer (see Table S1 in the supplemental material), and 25 μl of SYBR green PCR master mix (QIAGEN, Valencia, CA). Real-time PCR was carried out with the DNA Engine Opticon continuous fluorescence detection system from MJ Research (Waltham, MA). The gene encoding a flagellar sheath protein, flaA2, was used as the normalizer for all reactions. Melting curve analysis confirmed that all RT-PCRs amplified a single product.
Statistics for category comparisons.Chi-squared analysis was performed for comparisons of frequencies of various groupings of leptospiral genes affected by physiologic osmolarity (see Fig. 2 to 4). To account for multiple P value calculations, α values were adjusted with an online calculator available at the Graphpad web site and are specified in the legends to Fig. 2 through 4.
RESULTS AND DISCUSSION
Experimental design.Identification of genes that are regulated during interaction with host tissues and the mechanisms that regulate these infection-associated genes are keys to understanding leptospiral pathogenesis. For this reason, leptospiral whole-genome microarrays were developed to enable studies of global changes in transcript levels in response to environmental conditions that may mimic conditions in vivo. In our previous microarray study, RNA extracted from L. interrogans maintained for several passages at various temperatures was analyzed (33). Therefore, we attempted to maintain L. interrogans in EMJH with sodium chloride added to physiologic osmolarity. However, although the leptospires increased in density up to 2 logs at physiologic osmolarity, they could not be passaged into fresh medium containing sodium chloride added to physiologic osmolarity. Clearly, leptospires can grow at physiologic osmolarity in vivo, but some as yet undefined differences in the EMJH medium did not allow prolonged growth in vitro (see “Metabolism” below). Hence, RNA was extracted from overnight cultures of L. interrogans incubated in EMJH and EMJH supplemented with 120 mM sodium chloride. The increases in cell density following overnight incubation were similar in the two culture media. Using thresholds of 1.5 for the change in transcript level and 0.05 for the α value, we identified 124 salt-induced and 94 salt-repressed genes (3.5% and 2.6% of all genes, respectively). The strength of this approach is reflected in the median P values of 2.2 × 10−6 and 4.0 × 10−5 for genes induced and repressed, respectively, by osmolarity, with P values adjusted for multiple comparisons (see Tables S2 and S3 in the supplemental material). It is possible that some genes differentially regulated by sodium chloride are not regulated by other ionic or nonionic solutes such as sucrose. However, it is important to note that sodium chloride accounts for most of the osmotic activity inside a mammalian host (32, 56). The 25 most strongly induced and repressed genes are shown in Tables 1 and 2, while the complete lists of salt-regulated genes are provided in Tables S2 and S3 in the supplemental material.
The 25 most strongly up-regulated L. interrogans genes at physiologic osmolarity compared with low osmolarity
The 25 most strongly down-regulated L. interrogans genes at physiologic osmolarity compared with low osmolarity
Validation of microarray data.The 87 serovar Copenhageni genes represented by two oligonucleotides corresponding to segments near the 5′ and 3′ ends of the ORFs were examined. Among these genes, 10 demonstrated differential regulation by at least 1.5-fold. The signals obtained with the 10 pairs of 5′ and 3′ oligonucleotides were plotted against each other (see Fig. S1A in the supplemental material). The signals showed a high correlation (R2 = 0.9334), suggesting that the signals obtained with leptospiral genes represented by one oligonucleotide are likely to be valid.
Eight genes were selected for real-time RT-PCR analysis to validate the microarray data. The flaA2 gene was used for normalization, since the microarray data and immunoblots (data not shown) demonstrate that its levels did not change with salt concentration. A high correlation (R2 = 0.8697) was observed between expression values obtained by microarray analysis and real-time RT-PCR (see Fig. S1B in the supplemental material).
Analysis of transcriptional changes of genes in potential operons further confirmed the validity of the microarray data. We examined the remaining salt-regulated genes using a conservative definition of an operon as a cluster of genes located fewer than 100 bp apart without intervening stem-loop structures, which may be transcriptional terminators. In total, 78 of the 218 genes differentially regulated by salt were components of 52 potential operons containing genes with similar (fractional mean deviation ≤ 0.32) fold changes in signal (see Table S4 in the supplemental material). Furthermore, 51 of the 78 differentially regulated genes (65%) were in 27 potential operons in which all genes were regulated in the same direction with P values for all genes of <0.05. An additional 24 differentially regulated genes, which were regulated less than 1.8-fold, were members of 22 operons in which the P value was >0.05 for some genes in the operon. Only three differentially regulated genes (4%) were members of potential operons in which genes were regulated in opposite directions, with P values of <0.05 in each direction. These data show that the microarray results were consistent with what is known about the structural arrangement of genes in the leptospiral genome, especially for those genes that were regulated by at least twofold by sodium chloride.
Our microarray data are also consistent with preexisting data on the regulation of leptospiral genes by environmental conditions. In a previous study, we showed that expression of ligA and ligB is strongly induced by osmolarity, whereas expression of lipL41 was unaffected (36). The microarray data show that the transcriptional signals corresponding to oligonucleotides from both the ligA/ligB identical region and the ligB unique region were strongly induced (Table 1). In contrast, the lipL41 transcript level was unaffected by salt, consistent with previous immunoblot and RT-PCR analyses. Several other genes that are expressed during infection were induced by sodium chloride. The sph2 gene, which encodes a sphingomyelinase C (Lk73.5), was one of the most strongly up-regulated genes (Table 2) (relative ratio, 6.4-fold) and is thought to be induced during infection based on differential antibody reactivity studies (1). The lipL53 gene (LIC12099), encoding a predicted lipoprotein reactive with antisera from patients with leptospirosis (22), was strongly up-regulated (relative ratio, 4.0-fold) by sodium chloride. LIC12315, a predicted lipoprotein paralog of the factor H-binding protein, LfhA (59), was slightly up-regulated by salt (see Table S2 in the supplemental material) (relative ratio, 1.8-fold), consistent with the requirement for complement resistance of leptospires disseminating via the bloodstream.
Comparison with temperature microarray data.Leptospiral genes induced by a temperature upshift from 30° to 37°C were overrepresented among genes up-regulated by sodium chloride (Fig. 1). Seventeen percent (22/124) of genes up-regulated by salt were also up-regulated by temperature, which was significantly greater (P < 0.0064) than the percentage expected based on the frequency of temperature up-regulated genes in the genome (8.6% = 299/3484) (33). Induction by temperature was observed 4.7 times more commonly than expected among the 25 most strongly salt-induced genes (40% = 10/25). The genes ligB, sph2 (Lk73.5), and crp (cyclic AMP [cAMP] regulatory protein) were among those whose expression was induced by both temperature and sodium chloride (Fig. 1; Table 1). It is not surprising that some leptospiral genes are coregulated by both temperature and sodium chloride because both of these conditions are potentially important environmental signals to the organism for the transition from outside to inside the mammalian host. In contrast, two genes strongly induced by osmotic upshift were down-regulated by temperature upshift (Fig. 1; Table 1). It is possible that the strongest expression of these two genes occurs during exit from the host in urine into a lower temperature environment.
Comparison of effects of osmolarity and temperature on transcript levels. All genes in the L. interrogans genome are plotted on a log-log graph using the relative change in transcript level resulting from overnight shifts in temperature from 30 to 37°C (x axis) and in osmolarity from that found in EMJH culture medium to that found in the mammalian host (y axis). Lines demarcate the 1.5-fold cutoff as a criterion for differential regulation. Brackets encompass the 25 most salt-up-regulated and -down-regulated genes.
Relevance of physiologic osmolarity to the environment-to-host transmission.Bacteria respond to stressful environmental conditions in part by inducing expression of proteins that enhance their survival. Many of these proteins, including those induced under osmotic stress, are conserved among bacteria. Of the 22 osmotic stress and general stress response genes of known function recognized in the leptospiral genome (see Table S5 in the supplemental material), none was induced by salt upshift. The salt-induced Sph2 sphingomyelinase could possibly release choline from host membranes, which many bacteria under osmotic stress convert to the osmoprotectant glycine betaine. However, L. interrogans lacks the betT gene that encodes the choline transporter. Moreover, the betA gene, which catalyzes the first step in the conversion of choline to glycine betaine, is down-regulated (Table 2). These results indicate that the upshift to physiologic osmolarity does not induce the osmotic or general stress response, although we cannot rule out the possibility that the leptospires are undergoing nutritional stress (see “Metabolism” below). Although it is theoretically possible that L. interrogans has a unique mode of responding to osmotic stress involving some of the numerous hypothetical proteins induced by sodium chloride, it is more likely that the genes induced by salt encode proteins with other functions that enable the bacteria to survive and express putative virulence determinants in mammalian hosts.
A search for orthologs of L. interrogans genes differentially regulated by sodium chloride in the recently sequenced L. borgpetersenii genome was revealing (7). Most cases of leptospirosis are caused by serovars of L. interrogans and L. borgpetersenii, and the clinical symptoms caused by the two species are similar, although they may differ in severity. However, their modes of transmission differ: L. interrogans is usually acquired from water sources contaminated by urine of reservoir hosts, whereas epidemiological evidence indicates that L. borgpetersenii serovar Hardjo is transmitted directly from host to host (7). Comparison of the genomes of L. interrogans and L. borgpetersenii strain JB197 revealed that 29% of the genes in L. interrogans Fiocruz L1-130 are absent or pseudogenes in L. borgpetersenii strain JB197 (Fig. 2), consistent with L. borgpetersenii undergoing a process of genome reduction due to its direct host-to-host transmission cycle (7). Based on this observation, one would expect that L. interrogans genes that are down-regulated upon host entry would be those that encode proteins that are required outside the host and that these genes would tend to be lost in L. borgpetersenii. We therefore hypothesized that correct simulation of host conditions in vitro would result in down-regulation of L. interrogans genes that are pseudogenes in L. borgpetersenii. Indeed, of the 94 L. interrogans salt-repressed genes, 47% were pseudogenes or absent in L. borgpetersenii JB197, a significantly higher frequency than the 29% that were pseudogenes or absent in the genome overall (Fig. 2) (P < 0.0064). When we focused on the 25 most strongly repressed L. interrogans genes, the percentage that were pseudogenes or absent in L. borgpetersenii was 64% (Fig. 2). In contrast, only 24% of L. interrogans salt up-regulated genes, and 36% in the top 25, were pseudogenes or absent in L. borgpetersenii (P = 0.16 and 0.45, respectively). Similar results were obtained when L. borgpetersenii strain L550 was used for comparison to L. interrogans Fiocruz L1-130. These results are consistent with the impaired ability of L. borgpetersenii to survive outside the mammalian host and strongly indicate that the leptospiral response to physiologic osmolarity is relevant to environmental adaptation.
Percentage of salt-regulated genes in L. interrogans that were absent or pseudogenes in L. borgpetersenii. The bar graph shows absent homologs or pseudogenes in L. borgpetersenii as percentages of the 25 most strongly regulated genes (25), all salt-regulated genes (SR), and all genes in the L. interrogans genome (Gm). Homologs of genes both up- and down-regulated by salt were examined in L. borgpetersenii serovar Hardjo strains JB97 and L550. Asterisks indicate types of genes that were overrepresented compared to the number expected based on the genome-wide frequency (chi-squared analysis, α = 0.0064).
Cell envelope biogenesis.Because OMPs are exposed to the external environment, alteration in their expression levels would be crucial for adaptation of L. interrogans to changing environments. One of the most striking findings of our analysis of the response of the transcriptome to physiologic osmolarity was the greater than expected number of genes encoding lipoproteins that were differentially expressed. The spirochete-specific sequence analysis algorithm, SpLip, found that lipoproteins represent 4.6% (164/3531) of the genes present in the L. interrogans genome (52). As shown in Fig. 3, 89% (11/124) of salt-induced proteins and 19.2% (18/94) of salt-repressed proteins were lipoproteins. Lipoproteins were overrepresented to an even greater extent among the most strongly induced and repressed proteins (28% and 32%, respectively; P < 0.0018). Regulation of lipoproteins by physiologic osmolarity is consistent with the notion that surface-exposed proteins are required by the organism to adapt to changing environments. Host induction of lipoproteins may be necessary for expression of virulence functions, consistent with the salt-induced LigA and LigB adhesins (36) and the factor H-binding LfhA paralog (59). Interestingly, the leptospiral genome contains two versions of the lipoprotein export pathway protein LolA (LIC12545 and LIC10359), which is involved in shuttling lipoproteins across the periplasm to the outer membrane. Since only LIC10359 was up-regulated by sodium chloride (Table 3), we speculate that different sets of lipoproteins may be targeted for export to the outer membrane by the two versions of LolA. Nonredundancy of function of paralogous export proteins is further suggested by the finding that expression of one of seven leptospiral versions of a potential TolC secretion protein (LIC11941) and one of four versions of membrane fusion protein (LIC12430) was repressed by physiologic osmolarity.
Predicted locations of proteins encoded by salt-regulated genes. Locations of genes up-regulated (A) and down-regulated (B) by sodium chloride as percentages of the 25 most strongly regulated genes, all regulated genes, and the L. interrogans genome are shown. Genes were predicted to encode proteins that are exported (EXP), lipoproteins (LIP), noncytoplasmic proteins (NC), and proteins located in the outer membrane (OM), periplasm (PER), inner membrane (IM), or cytoplasm (CYT). Asterisks indicate types of genes that were overrepresented compared to the number expected based on the genome-wide frequency. Chi-squared analysis was used to compare the numbers of genes in each category with the expected numbers based on the numbers found in the entire genome (chi-squared analysis, α = 0.0018).
L. interrogans genes of known or predicted function regulated at physiologic osmolarity compared with low osmolarity grouped by COG categoryd
Two of the most strongly induced genes were LIC12340 and LIC12339, which were up-regulated 10.8 and 6.0-fold, respectively. These genes encode members of a family of 13 exported paralogous proteins, which include LIC10639, LIC10695, LIC10778, LIC10870, LIC11358, LIC12715, LIC12778, LIC12791, LIC12963, LIC12985, and LIC12986. All members except for one are predicted to be 71.1 to 73.3 kDa in size. It is possible that the expression of individual members is regulated by different environmental signals.
Metabolism.The availability of different nutrients inside and outside the mammalian host requires changes in the metabolic capacity of leptospires establishing infection. As for most bacterial pathogens, iron is essential for the growth and survival of L. interrogans (18). Several L. interrogans genes involved in iron acquisition and storage were induced by sodium chloride (Table 3). One of 12 TonB-dependent receptors encoded by the leptospiral genome (30, 34), LIC11694, was up-regulated 2.0-fold by salt. The Leptospira biflexa ortholog of LIC11694 is required for its growth in iron-depleted EMJH medium containing desferrioxamine. Although L. interrogans cannot use desferrioxamine as a source of iron (34), it is possible that the LIC11694 TonB-dependent receptor is involved in the uptake of iron bound to another siderophore. A heme oxygenase (LIC20148), which catalyzes the first step in the breakdown of heme, and a permease that appears to be encoded by a gene in the same operon (LIC20149) were up-regulated by sodium chloride. A bacterioferritin-associated ferridoxin (LIC13209), which may function in iron storage, was also up-regulated. These findings suggest that, during infection, L. interrogans increases expression of proteins required for acquiring and storing iron, which is present in growth-limiting amounts in the host.
L. interrogans uses long-chain fatty acids (LCFA) as its sole carbon and energy source in vitro (27). LCFA uptake in E. coli by the FadL outer membrane transporter is coupled to the activation of the fatty acid by FadD, a long-chain acyl coenzyme A (CoA) synthetase in the inner membrane (16). In the cytoplasm, the long-chain acyl-CoA binds to the FadR regulatory protein and inactivates it, thereby preventing FadR from activating transcription of several genes whose products are required for fatty acid degradation. Consequently, several fab genes, whose products are required for biosynthesis of LCFA, are not transcribed when LCFA is present in the medium (16). The leptospiral fadL (LIC12524) and fadD (LIC10094) genes were down-regulated, and the fabB (LIC12317) and fabH (LIC12323) genes, encoding two isoforms of acyl carrier protein synthase, which is necessary for fatty acid biosynthesis, were up-regulated by sodium chloride (Table 3). Also consistent with the reduction of fatty acid degradation at physiological osmolarity was the reduced expression of the sucCD operon (LIC12573 and LIC12574), encoding the α and β subunits of succinyl-CoA synthetase. Hence, the transition to physiological osmolarity appears to be associated with a shift from degradation to biosynthesis of fatty acids. The biological reason for L. interrogans to repress utilization of its sole carbon source is unknown, but it may account for the inability to maintain L. interrogans at physiological osmolarity in vitro. Growth at physiologic osmolarity in vivo may require carbon sources that are absent in EMJH medium, as L. interrogans down-regulates its fatty acid catabolism pathway. Similar to what is observed with catabolite repression in Escherichia coli (55), Crp (see “Signal transduction” below) may mediate a switch in utilization from fatty acids to a different carbon source found in vivo.
Signal transduction.The signal transduction machinery must sense changes that occur in the environment and transmit those changes to genes or proteins that would allow adaptation to the changing environment. Figure 4 shows that 15% of genes up-regulated by sodium chloride are assigned to category T (signal transduction mechanisms), which is significantly greater (P < 0.0013) than the 5% of genes in the entire genome encoding category T proteins. Additionally, 7% of down-regulated genes were members of category T. The large number of differentially regulated category T genes demonstrates the important role that salt regulation has in the organism's adaptation to the drastic change in conditions when it enters a mammalian host. Most of the affected category T products fall into one of several categories: proteins that bind or metabolize cyclic nucleotides, members of two-component regulatory systems, or anti-sigma factor antagonists (Table 4).
Percentages of genes assigned to COG (clusters of orthologous groups) and categories among salt down- up-regulated genes and in the L. interrogans genome. The COG functional categories are as follows: information storage and processing (includes translation [J], transcription [K], and replication, recombination, and repair [L]), cellular processes and signaling (includes cell cycle control, cell division, and chromosome partitioning [D], defense mechanisms [V], signal transduction mechanisms [T], cell wall, membrane, or envelope biogenesis [M], cell motility [N], intracellular trafficking, secretion, and vesicular transport [U], and posttranslational modification, protein turnover, and chaperones [O]), metabolism (includes energy production and conversion [C], carbohydrate transport and metabolism [G], amino acid transport and metabolism [E], nucleotide transport and metabolism [F], coenzyme transport and metabolism [H], lipid transport and metabolism [I], inorganic ion transport and metabolism [P], and secondary metabolite biosynthesis, transport, and catabolism [Q]), and poorly characterized (includes general function only [R] and function unknown [S]). The asterisk indicates that genes predicted to encode signal transduction mechanisms were overrepresented among genes up-regulated by sodium chloride (chi-squared analysis, α = 0.0013).
L. interrogans signal transduction genes at physiologic osmolarity compared with low osmolarity
Among all known bacterial pathogens, L. interrogans harbors the greatest number of genes encoding adenylate cyclase; L. interrogans has 18 such genes, 12 of which are predicted to be in the cytoplasmic membrane (20). The large number of adenylate cyclase enzymes suggests that cAMP is an important signaling molecule in Leptospira spp., the concentration of which reflects conditions internal and external to leptospiral cells. One gene encoding cytoplasmic adenylate cyclase, LIC12327, was up-regulated 1.8-fold by sodium chloride. LIC11484, which is up-regulated 4.0-fold by salt, encodes the cAMP-binding transcriptional regulatory protein Crp. In E. coli, transcription of crp in E. coli is subject to osmoregulation in a cAMP-dependent manner (3). Crp is a key global transcriptional regulator of virulence genes in Pseudomonas aeruginosa (60) and Yersinia enterocolitica (47), with the second messenger cAMP playing a central role in both cases. We demonstrated previously that the leptospiral crp gene was up-regulated when the temperature was shifted from 30°C to 37°C (33). These observations suggest that cAMP and Crp play a central role in regulation of gene expression during the early stages of host infection. The up-regulation of signal transduction genes LIC10641 and LIC10921, which possess the EAL domain implicated in cyclic diguanylate phosphodiesterase activity (57), suggest that c-di-GMP is also a second messenger whose intracellular concentration reflects the external osmolarity.
The genome of L. interrogans encodes 48 sensors and 38 response regulators that are members of two-component regulatory systems (21, 43, 58). The large number of two-component regulatory systems reflects the ability of L. interrogans to survive in a variety of different environments (20). The expression of many two-component systems, including several that have been shown to be positively autoregulated, is under environmental regulation (10, 23, 44). Transcript levels expressing three sensors (LIC11425, LIC13289, and LIC12379) and two response regulators increased with salt upshift. One of the three sensors, LIC12379 (1.9-fold up-regulated) has the conserved CHASE3 extracellular sensory domain and is in the same operon as the response regulator LIC12380 (Table 4), which barely missed the cutoff (1.48-fold up-regulated) (see Table S4 in the supplemental material), suggesting that LIC12379 and LIC12380 form a salt-regulated two-component system. The response regulator LIC20254 (3.2-fold up-regulated) includes an OmpR-type DNA-binding domain, and LIC12504 (2.8-fold up-regulated), like LIC12380, has only the receiver domain, which transmits signals to its target by protein-protein interactions (21). Additionally, a single gene encoding a response regulator with a histidine kinase output domain (LIC12235) was down-regulated 1.7-fold. Thus, host infection may be associated with changes in the levels and activities of several two-component regulatory systems whose output domains function by a variety of mechanisms to alter the activities of key genes and proteins that allow L. interrogans to adapt to host conditions.
The functions of alternative sigma factors are often negatively regulated by anti-sigma factors that bind directly to the sigma factor and thereby prevent them from interacting with the core RNA polymerase. One category of sigma factor regulators involves “partner-switching” modules that include an anti-sigma factor with protein kinase activity, an anti-sigma factor antagonist whose activity is affected by its phosphorylation state, and a phosphatase that targets the phosphorylated anti-sigma factor antagonist. Partner-switching signaling pathways regulate expression of the type III secretion system in Bordetella (31) and biofilm formation in Staphylococcus epidermidis (28). The leptospiral genome encodes 9 anti-sigma factors and 19 anti-sigma factor antagonists. LIC11480 and LIC11426, which encode an anti-sigma factor and phosphatase, respectively, were both up-regulated 1.6-fold by sodium chloride. Four of the 19 anti-sigma factor antagonists were differentially regulated by salt. LIC11496, LIC12766, and LIC11801 were up-regulated, and LIC11525 was down-regulated, by sodium chloride. The genome of L. interrogans encodes 11 extracytoplasmic function sigma factors, which in other bacteria are activated in response to stress or injury to the bacterium outside of the cytoplasm (50). The partner-switching proteins whose levels are affected by physiologic osmolarity are likely to affect the activities of some of the extracytoplasmic function sigma factors during host invasion.
Concluding remarks.In this study, we demonstrated that exposure of L. interrogans to the osmolarity found in host tissues induced a profound shift in gene transcript levels that is likely to facilitate adaptation to a mammalian host leading to successful establishment of infection. Lipoproteins were overrepresented among genes differentially regulated by physiologic osmolarity, especially among those that were most strongly affected. Lipoproteins whose levels are up-regulated at physiologic osmolarity, such as the LigA and LigB adhesins and factor H binding LfhA paralog, may be important in the pathogenesis of L. interrogans (9a, 36, 59). We also found that a greater fraction of genes encoding signal transduction proteins were up-regulated by physiologic osmolarity than expected from their frequency in the genome. This observation is an indication of the importance of signal transduction in adaptation to conditions in the host, which differ dramatically from conditions in the environment. Furthermore, the osmotic and general stress response genes were not induced, indicating that, at the osmolarity found in host tissues, a large effort of the regulatory machinery of L. interrogans is directed toward host adaptation, not stress survival. On the other hand, a greater than expected number of genes up-regulated by sodium chloride were also up-regulated by a temperature shift from 30°C to 37°C. Remarkably, almost half of L. interrogans genes that were down-regulated by salt were pseudogenes or absent in the obligate parasite L. borgpetersenii. Taken together, these data indicate strongly that physiologic osmolarity is a key signal sensed by leptospires during the transition from the ambient environment to infection of mammalian host tissues.
ACKNOWLEDGMENTS
This work was supported by VA Medical Research Funds (to J.M. and D.A.H.), NIH grant AI-34431 (to D.A.H.), and a grant from the National Health and Medical Research Council, Canberra, Australia (to B.A.).
We thank David Bischoff (VA Greater Los Angeles Healthcare System) for assistance in using the DNA Engine Opticon system and David Scott (VA Greater Los Angeles Healthcare System) for technical support with the microarray hybridizations.
FOOTNOTES
- Received 6 October 2006.
- Returned for modification 19 November 2006.
- Accepted 11 March 2007.
- Copyright © 2007 American Society for Microbiology
