Transcriptional Responses of Leptospira interrogans to Host Innate Immunity: Significant Changes in Metabolism, Oxygen Tolerance, and Outer Membrane

Background Leptospira interrogans is the major causative agent of leptospirosis. Phagocytosis plays important roles in the innate immune responses to L. interrogans infection, and L. interrogans can evade the killing of phagocytes. However, little is known about the adaptation of L. interrogans during this process. Methodology/Principal Findings To better understand the interaction of pathogenic Leptospira and innate immunity, we employed microarray and comparative genomics analyzing the responses of L. interrogans to macrophage-derived cells. During this process, L. interrogans altered expressions of many genes involved in carbohydrate and lipid metabolism, energy production, signal transduction, transcription and translation, oxygen tolerance, and outer membrane proteins. Among them, the catalase gene expression was significantly up-regulated, suggesting it may contribute to resisting the oxidative pressure of the macrophages. The expressions of several major outer membrane protein (OMP) genes (e.g., ompL1, lipL32, lipL41, lipL48 and ompL47) were dramatically down-regulated (10–50 folds), consistent with previous observations that the major OMPs are differentially regulated in vivo. The persistent down-regulations of these major OMPs were validated by immunoblotting. Furthermore, to gain initial insight into the gene regulation mechanisms in L. interrogans, we re-defined the transcription factors (TFs) in the genome and identified the major OmpR TF gene (LB333) that is concurrently regulated with the major OMP genes, suggesting a potential role of LB333 in OMPs regulation. Conclusions/Significance This is the first report on global responses of pathogenic Leptospira to innate immunity, which revealed that the down-regulation of the major OMPs may be an immune evasion strategy of L. interrogans, and a putative TF may be involved in governing these down-regulations. Alterations of the leptospiral OMPs up interaction with host antigen-presenting cells (APCs) provide critical information for selection of vaccine candidates. In addition, genome-wide annotation and comparative analysis of TFs set a foundation for further studying regulatory networks in Leptospira spp.


Introduction
Leptospirosis, which is characterized by hemorrhage, diarrhea, jaundice, severe renal impairment, and aseptic meningitis, etc., has emerged as a global zoonotic infectious disease in the past decade [1]. Several pathogenic Leptospira species cause infection, which include more than 15 genospecies and 230 serovars distributed geographically. Other free-living saprophytic Leptospira species, such as Leptospira biflexa, do not infect humans and animals. The pathogenic, saprophytic Leptospira and several other intermediate species all belong to the Spirochaetes, a unique phylum in eubacteria including other pathogens, such as Borrelia burgdorferi and Treponema pallidum. Leptospira interrogans is the most prevalent pathogenic Leptospira species which survives in natural environments and animal reservoir hosts, and infects humans through abrasions in the skin or mucous membrane. The main reservoir hosts of L. interrogans are wild rodents and domestic animals, which can persistently excrete L. interrogans through urine. The shed leptospiral cells can survive in moist soil and water for a long time before infecting a new host [2]. Therefore, L. interrogans adapts to diverse natural environments and evades host immune defense during infection to maintain transmission. This makes L. interrogans an important pathogen in understanding leptospirosis.
The genome sequences of L. interrogans (strain Lai 56601 and Fiocruz L1-130), pathogenic Leptospira borgpetersenii (strain L550 and JB197), and saprophytic L. biflexa (strain Patoc I Paris/Ames) have been released in the past few years [3,4,5,6]. The genome size of L. interrogans (,4.6M) is larger than those of L. borgpetersenii (,3.9M) and L. biflexa (,3.9M), which is consistent with the evidence that L. interrogans retained more genes from the common ancestor while acquiring exogenous genes during evolution. Comparative genomics have been preformed to identify potential virulence genes in L. interrogans [7]. However, few virulence factors have been experimentally confirmed due to the lack of efficient methods for genetic manipulation of pathogenic Leptospira [8]. In addition, many of the putative functional genes are in multicopies or families with high degree of redundancy, which further hampers virulence determinants using genetic approaches and molecular Koch's postulate. For example, two major outer membrane protein genes, ligB [9] and lipL32 [10], which are highly conserved in pathogenic Leptospira and absent in non-pathogenic L. biflexa, have been inactivated in L. interrogans and verified to be dispensable for infection.
In comparison to the other pathogenic spirochetes, L. interrogans encodes more putative signal transduction and transcriptional regulation genes [11]. Several global gene expression studies have elucidated the transcriptional responses of L. interrogans to temperature, osmolarity, and host serum [12,13,14,15]. Among these factors, osmotic stress was identified as a key signal affecting the leptospiral transcriptome. However, these microarray analyses identified few genes whose expression has been shown to be differentially regulated during mammalian infection by proteomics and other approaches [16,17,18]. In particular, several major OMPs genes (e.g., lipL32, qlp42 and loa22) are differentially regulated in vivo. This is likely due to the environmental factors in vitro are not the major signals Leptospira senses during mammalian infection. Therefore, global analysis of leptospiral gene expression in animal or infection models are vital to identify differentially regulated genes relevant to pathogenesis.
Co-cultivation of pathogenic Leptospira with host immune cells is widely used as an infection model to study leptospirosis [19,20]. Although pathogenic Leptospira is not considered a typical intracellular pathogen, recent studies showed that pathogenic Leptospira can attach, invade, and induce apoptosis of mammalian macrophages, and escape host innate immunity during the early stage of infection [21,22]. In addition, our study demonstrated differential survivability of L. interrogans within murine or human macrophages, which may contribute to the different severity between the mild chronic infection in reservoir animals and the acute lethal infection in humans [23]. Rapid uptake of L. interrogans by phagocytes were also verified by the naive zebrafish embryos model, suggesting that phagocytosis may be a key defense mechanism during the early stage of infection [24]. In this study, we performed microarray analysis on leptospiral gene expression in response to innate immune cells of murine and human origin. We found a dramatic influence of L. interrogans gene expression by host macrophage interaction, including genes of the major OMPs. A bioinformatic approach was used to determine regulators responsible for differential gene expression. This approach identified a putative OmpR transcription factor, which may be involved in the regulation of major OMP genes.

Bacterial strain
L. interrogans Serovar Lai Strain Lai 56601 was obtained from the National Institute for the Control of Pharmaceutical and Biological Products, Beijing, China. For microarray hybridization purpose, a single colony was picked from the EMJH [25,26] plate (1% agar) and verified by 16S rDNA-specific and gyrB1 (DNA gyrase subunit B1 gene)-specific primer PCR and gene sequencing. The virulence of the L. interrogans isolate was restored by passage through Dunkin-Hartley ICO: DH (Poc) guinea pigs (10-12days old, weighing 120-150g each) before infection. As an in vitro control design, the isolate was cultured in liquid EMJH medium for 5 passages and named E0 sample after the EMJH medium. The culture condition of each passage was growth in 200ml liquid EMJH media at 28uC under aerobic conditions for 120 h to reach exponential growth phase. Three biological replicates (E0-1/2/3) were used for microarray purpose. Before sample collection, one volume of bacterial culture was mixed with a one-tenth volume of ice-cold phenol/EtOH stop solution [10% water-saturated phenol (pH,7.0) in ethanol] and chilled rapidly [27]. Leptospiral cells were harvested by centrifugation at 8,000 g, 4uC for 15 min. All animals were handled in strict accordance with good animal practice as defined by the relevant national and/or local animal welfare bodies, and all animal work was approved by the Animal Ethics Review Committee of Zhejiang University.

Host cell lines
Murine monocyte-macrophage-like cell line J774A.1 and human acute monocytic leukemia cell line THP-1 were obtained from American Type Culture Collection (Manassas, VA) and grown in RPMI 1640 medium (Invitrogen, Carlsbad, CA) supplemented with 10% (V/V) heat-inactivated fetal calf serum (FCS, Gibco/Invitrogen, Carlsbad, CA) with antibiotic, in a humidified 5% CO 2 atmosphere at 37uC. The suspended THP-1 cells were treated with 5 nM phorbol myristate acetate (PMA; Sigma-Aldrich St. Louis, MO) for 24h. After differentiation, the cells were washed three times with sterilized PBS buffer, and rested for 24h in new cell medium to ensure that they reverted to a resting phenotype before infection. All cells were cultured in

Author Summary
Leptospirosis is an important tropical disease around the world, particularly in humid tropical and subtropical countries. As a major pathogen of this disease, Leptospira interrogans can be shed from the urine of reservoir hosts, survive in soil and water, and infect humans through broken skin or mucous membranes. Recently, host adaptability and immune evasion of L. interrogans to host innate immunity was partially elucidated in infection or animal models. A better understanding of the molecular mechanisms of L. interrogans in response to host innate immunity is required to learn the nature of early leptospirosis. This study focused on the transcriptome of L. interrogans during host immune cells interaction. Significant changes in energy metabolism, oxygen tolerance and outer membrane protein profile were identified as potential immune evasion strategies by pathogenic Leptospira during the early stage of infection. The major outer membrane proteins (OMPs) of L. interrogans may be regulated by the major OmpR specific transcription factor (LB333). These results provide a foundation for further studying the pathogenesis of leptospirosis, as well as identifying gene regulatory networks in Leptospira spp.
Responses of L. interrogans to Macrophages www.plosntds.org 225 cm 2 tissue culture flasks (Corning. Inc., Big Flats, NY) and the cell numbers were counted using haemocytometer.

Infection models
The cultured mammalian cells were washed three times with sterilized PBS buffer to remove antibiotic, fresh media without antibiotics were added, and cultured for an additional 12 h before infection. Leptospiral cells were harvested by centrifugation at 8, 000 g, 20uC for 15 min, and washed three times with sterilized PBS buffer. The leptospiral pellets were re-suspended in 37uC RPMI 1640 medium with 10% (V/V) heat-inactivated FCS and the bacterial numbers were counted with a Petroff-Hausser counting chamber (Fisher Scientifics, Houston, Texas). Then 10 ml of leptospiral suspension (10 9 ) were added into 10 7 macrophage cells (bacteria:cell = 100:1) and incubated in 5% CO 2 at 37uC. These co-cultured L. interrogans samples were defined as J (J774A.1) and T (THP-1) samples respectively after the names of the mammalian cell lines. In order to evaluate the impact of mammalian cell culture medium on L. interrogans, RPMI 1640 medium controls [RPMI 1640 medium with 10% (V/V) heatinactivated FCS] were introduced into experiments as M (RPMI 1640) samples. That is, L. interrogans grew in RPMI 1640 medium with 10% (V/V) heat-inactivated FCS which had been deposited in 5% CO 2 at 37uC for 12 h beforehand. Three biological replicates were designed for each sample for microarray purpose. To guarantee the integrity of the total RNA, the survival of the L. interrogans samplings drawn from all above-mentioned infection models were verified by darkfield microscope analysis (4006). Then the co-cultured L. interrogans samples (J, T and M) were RNA-stabilized and collected at 45 min when L. interrogans began to attach the host cells, or at 90min when the attachment rate reached the stable level [28]. In detail, the attached leptosiral cells were gathered by washing the macrophage cells twice with sterilized PBS at the time-points of 45 min and 90 min respectively. The collections were mixed with a one-tenth volume of ice-cold stop solution and chilled immediately. Then, the mixtures were centrifuged at 1, 000 g for 5 min at 4uC to exclude the pellets of J774A.1 and THP-1 cells. The supernatants were centrifuged at 8, 000 g, 4uC for 15 min to collect the leptospiral pellets. This RNA stabilization procedure is essential for microarray analysis because of the short life time of leptospiral total RNA ( Figure 1A).

RNA purification and ds cDNA synthesis
Leptospiral total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA), then purified by RNeasy Mini Kit (QIAGEN, Hilden, Germany) with on-column DNase digestion (QIAGEN, Hilden, Germany) according to the RNeasy Mini handbook. RNA quantity and integrity was determined using the RNA 6000 Nano Laboratory-on-a-Chip kit and the Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA). For each sample, about 10 mg of total RNA was mixed with 600 ng of random hexamer primers (TaKaRa, Otsu, Japan) and denatured at 65uC for 5 min. Then the first strand cDNA was synthesized using 2 ml (400 U) SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA) according to the protocol recommended by the manufacturer. The double strand cDNA (ds cDNA) sample was synthesized using the 2nd Strand Synthesis section of the M-MLV RTase cDNA Synthesis Kit (TaKaRa, Otsu, Japan) according to the manufacturer's instructions. Following RNase H (Invitrogen, Carlsbad, CA) and RNase A (Ambion, Austin, TX) digestion for 1 h, ds cDNA sample was purified with the QIAquick PCR Purification Kit (QIAGEN, Hilden, Germany) according to the QIAquick Spin handbook.

Microarrays and hybridization
The L. interrogans Serovar Lai Strain Lai 56601-specific highdensity, photolithography-based, mono-plex DNA microarray chip was designed and produced by Roche NimbleGen, Inc. Each slide consisted of a total of 385,000 oligonucleotide probes (60-mer each probe) which covered all predicted 4,727 ORFs of the whole genome (NC_004342 and NC_004343). In average, sixteen probes were designed for each ORF, which is one of the strength of this technology. Each probe pair consisted of a sequences matched to the ORF, and another adjacent sequence harbored mismatched bases for the determination of background and cross-hybridization. Note that the original annotation for L. interrogans Serovar Lai Strain Lai 56601 included more than 900 putative small ORFs (less than 150 bp). In contrast, the homologs of these small ORFs were not included in the later genome annotation for other 5 Leptospira strains. However, we found that some of these putative ORFs had very high level of expression (Data not shown). Thus, these small ORFs were included in our microarray analysis. For each hybridization, 1 mg of ds cDNA was labeled with Cy3-9mer Primers (TriLink Biotechnologies, San Diego, CA) using the Klenow fragment (New England Biolabs, Beverly, MA) exo-extending reaction. 1.5 mg of Labeled cDNA sample was individually hybridized to the microarray using the MAUI hybridization system from Roche NimbleGen, then washed and dried according to the Roche NimbleGen standard procedure.

Data extraction and statistical analysis
The microarrays were scanned using the Axon GenePix 4000B microarray scanner at 5-mm resolution. The data were extracted using Roche NimbleGen NimbleScan TM software and an algorithm (courtesy of Y. Qiu, University of Wisconsin School of Medicine) was applied to obtain a single measurement of signal intensity for each ORF. Data were normalized using the quantile normalization method and the total signal intensity of a given ORF was converted to estimates of transcript abundance by using the robust multiarray average (RMA) procedure. For all microarrays, a P value for each ORF was calculated by a two-tailed Welch's unpaired t test comparison of the five microarray replicates for each sampling condition. The fold changes of an ORF between two relevant sampling conditions were calculated as the division value of average signal intensity of 3 biological replicates: average J, T or M sample signal intensity at a certain time-point/average E sample signal intensity. Only fold changes of at least 62 and P#0.05 were considered significant and included in this report.

Validation of microarray data by quantitative real-time RT-PCR
Primers for six randomly selected L. interrogans Serovar Lai Strain Lai 56601 genes (Table S1) were designed with Primer Premier software version 5 (Premier Biosoft International, Palo Alto, CA). New batches (in triplicate) of bacterial RNA were used in real-time RT-PCR. RT reaction mixtures contained 1 mg of total RNA, 300 ng of random hexamer primers (TaKaRa, Otsu, Japan), 0.5 ml (100 U) of Superscript III (Invitrogen, Carlsbad, CA) reverse transcriptase, and 500 mM concentrations each of dATP, dCTP, dGTP, and dTTP. After denaturation on 65uC for 5 min, the samples were incubated at 50uC for 1 h, followed by 10 min at 70uC to synthesize the first strand cDNA. 50 ng of cDNA were mixed with 12.5 ml of 26SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). Assays were performed in triplicate with the ABI PRISM model 7500 sequence detection instrument. Amplicon quantification in real-time RT-PCR was performed by comparison with gene-specific standard curves constructed form known concentrations of each purified amplicon. The melting curve analysis was also used to evaluate that the accumulation of SYBR Green-bound DNA was gene-specific.

Extraction of total proteins and outer membrane proteins
The Leptospira pellets were harvested from infection models at 1h, 2h, and 4h according to the method described previously in this report. Total leptospiral protein was extracted with Triton X-100. For each sample, the leptospiral pellet was washed twice in 0.5 ml PBS-5mM MgCl 2 and resuspended in 0.5 ml bacteria lysis buffer which was composed of 50 mM Tris-HCl (pH8.5), 2 mM EDTA, 100 mM NaCl, 0.5% Triton X-100 (Calbiochem, La Jolla, CA), 100 mg/ml lysozyme (Sigma-Aldrich St. Louis, MO) and 1ml/ml protease inhibitor PMSF (Sigma-Aldrich St. Louis, MO). The suspension sample was subjected to three cycles of freezing, thawing and tip sonication, following by centrifugation at 10,000g for 10 min to exclude the indissoluble fragments. The soluble supernatant was dialyzed by 1% SDS and tested by SDS-PAGE electrophoresis. The leptospiral OMP samples were extracted by solubilization with 1% Triton X-114 (Calbiochem, La Jolla, CA) according to the method reported previously [29].

Verification of protein changes by Western blotting
Protein concentration was estimated by BCA protein assay (Pierce, Rockford, IL). For leptospiral lipoproteins (LipL32, LipL41) and outer membrane protein A (OmpA), the OMP samples were used in Western blotting analysis. For transmembrane protein (OmpL1), flagellar components (FliH, FliI, FliY and FliN) and intracelluar protein (Mce), the total protein samples were used. Equivalent amounts of protein (1 mg) were separated by 1-D electrophoresis with 10% SDS-PAGE, and electrotransferred onto polyvinylidene difluoride membranes (PVDF membranes, Millipore, Billerica, MA) using a Trans-Blot Semi-Dry Transfer Cell (Bio-Rad, Hercules, CA). The membranes were blocked in TBST [20 mM Tris (pH 7.6), 137 mM NaCl, 0.1% Tween 20] containing 5% non-fat milk and probed with corresponding antibodies (1:2000) respectively overnight at 4uC. Then, the membranes were incubated with peroxidase-conjugated antirabbit immunoglobulin G (Jackson ImmunoResearch Laboratories Inc., PA) for 2 h at room temperature and visualized on X-ray film using enhanced chemiluminescence reagents (Millipore, Billerica, MA). The band intensities were estimated by densitometric scanning using the Gel Doc 2000 system (Bio-Rad, Redmond, WA) and Quantity One software. Data shown are from three independent experiments.

Genome annotation, cluster and pathway analysis of microarray data
To supply new functional annotation of the L. interrogans Serovar Lai Strain Lai 56601 genome, assignment of putative functions was performed by means of a combination of BLAST-based and HMM-based searches of the KEGG gene database and InterPro protein domain database. Hierarchical cluster analysis of microarray data was preformed by the Cluster 3.0 software and visualized by the TreeView software [30]. Several similarity matrices, including correlation matrices [Correlation (uncentered), Absolute Correlation (uncentered), Spearman Rank Correlation, Kendall's tau] and distance matrices (Euclidean distance and City-block distance) were used in cluster analysis to define the significantly changeable subclades. The KEGG pathway was introduced to analyze the changing magnitude in each biological pathway by gathering statistics from the significant gene expression changes. The proportion of regulated genes in the 14 leptospiral KEGG pathway was calculated and displayed by Microsoft Excel diagram.
Responses of L. interrogans to Macrophages www.plosntds.org

Discovery and evolutionary analysis of transcription factors
Candidate TFs were collected by InterProScan program (http://www.ebi.ac.uk/Tools/InterProScan/) and the Gene Ontology terms were obtained from the InterProScan results using InterPro2GO (http://www.geneontology.org/external2go/ interpro2go) [31]. Then, the DNA binding domains (DBDs) of the candidate TFs were used in further BLAST search against all Leptospira proteins to identify new TFs. All putative TFs were classified into different families according to categories of their DBDs. This strategy helped us to find new TFs that had not been defined before ( Figure 1B). Finally, phylogenetic analyses of the whole TF sequences were performed using Neighbor-Joining method of MEGA 4.0 software to reveal the evolutionary relationship of the putative TFs [32].

Annotation of previous defined hypothetical ORFs
In order to conduct genome-wide transcriptional analyses of L. interrogans, we first performed an updated annotation using BLAST and InterProScan tools with up-to-date databases. Specifically, we focused on annotating the hypothetical ORFs, which comprises about 40% of total ORFs of the genome of L. interrogans Serovar Lai Strain Lai 56601 [3]. This annotation allowed us to assign putative functions to 375 hypothetical proteins (Table S2), and identified several functionally important homologs missing in the previously annotated genome. These included a ferrous iron transporter, outer membrane lipoprotein carrier protein LolA, phospholipase C, the SecE subunit for protein translocation complex, a cell wall hydrolase, and the flagellar hook-length control protein chromogranin. In addition, several ORFs involved in host-pathogen interaction were identified, including a haem oxygenase-like protein, type-III fibronectin, hemopexin, a preventhost-death protein, caspase catalytic proteins, a nitrilase/cyanide hydratase, ricin B lectin, and cadherin-like protein.

Microarray experimental design
To study global transcriptional responses of L. interrogans upon interaction with the host innate immune system, we chose two different mammalian cell lines: murine monocyte-macrophage-like cell line J774A.1 and human acute monocytic leukemia cell line THP-1. This represented interaction with macrophage of either natural mammalian reservoir or human host. Briefly, a clone of L. interrogans Serovar Lai Strain Lai 56601 was used to infect a guinea pig and subsequently recovered from kidney tissue. Isolated infectious spirochetes were split to two samples. One sample was kept under in vitro cultivation and passage in EMJH medium for 5 passages (approximately one month) then subjected to RNA extraction and defined as in vitro cultivation sample (sample E0). The other sample was divided into three groups and co-cultured with cell lines J774A.1, THP-1, or RPMI 1640 medium with 10% heat-inactivated FCS designated as samples J, T, and M, respectively. After 45 or 90 minutes, spirochetes were harvested, subjected to RNA extraction, and cDNA synthesis ( Figure 1A). Each cDNA sample was labeled and used for a single hybridization. Average signal intensity of 3 biological replicates was calculated as a valid data for each sample (E0, J45, J90, T45, T90, M45 or M90). The transcriptional fold-changes of ORFs in J, T or M samples were calculated relative to sample E0. Analysis of differentially expressed genes were confined to those with changes $2-fold (P#0.05). The microarray raw data are available at http://cibex.nig.ac.jp/ under CIBEX accession no. CBX129.

Global transcriptomic analysis
Based on the correlation and distance similarity matrices of Cluster 3.0 software, L. interrogans genes differentially expressed upon interaction with macrophages were grouped into several clades ( Figure 2A). Clades of up-regulation and down-regulation of gene expression occurred mainly in comparison between the in vitro cultivated spirochetes and bacteria that interacted with either mouse-derived macrophage or human macrophage cell lines, not in comparison between the in vitro cultivated spirochetes and bacteria in RPMI 1640 medium controls. In other words, incubation of spirochetes with the cell culture medium alone had little influence in gene expression (M45/E0 and M90/E0). In addition, the cluster analysis showed that T45/E0, T90/E0 and J45/E0 were closely related, whereas J90/E0 was similar to M45/ E0 and M90/E0, both in correlation matrices and in distance matrices. This suggested a more transient change of gene expression profile in L. interrogans upon interaction with murine macrophages (J90/E0) than with human-derived macrophage cells (T90/E0). About 65% of down-regulated genes and 45% upregulated genes were hypothetical protein genes, which was consistent with previous transcriptomic results that these genes are often regulated in different environments [12,14].
Validation of the microarray data by qRT-PCR was shown in Figure 3. To evaluate the alterations in leptospiral biological pathways, proportions of up-regulated and down-regulated genes in KEGG pathways were calculated (Figure 4). Transcription and translation systems, carbohydrate, energy and lipid metabolism, and signal transduction systems exhibited significant downregulation patterns, while biosynthesis of secondary metabolites, membrane transport and metabolism of cofactors and vitamins showed up-regulation patterns in our infection models. In Responses of L. interrogans to Macrophages www.plosntds.org addition, several important KEGG sub-pathways were significantly altered. Most genes of the citric acid cycle (TCA cycle), flagellar assembly, oxidative phosphorylation, fatty acid metabolism and ribosome synthesis were down-regulated, while most genes of starch and sucrose metabolism, porphyrin metabolism, and twocomponent systems were up-regulated (Data not shown). Categories exhibiting significant regulation are discussed below in detail.

Energy, carbohydrate, and lipid metabolism
Most of the genes involved in this category were down-regulated upon interaction with macrophages. First, genes involved in oxidative phosphorylation were down-regulated. These changes included a significant down-regulation (3-6 folds) of a putative operon (LA0242-0244, included in down-regulated Clade 5) encoding cytochrome caa3 oxidase subunit II, cytochrome C oxidase polypeptide I and cytochrome C oxidase polypeptide III (Table S2). Previous reports showed that the leptospiral cytochrome C oxidase polypeptide III is replaced by an alternative subunit that has weak oxygen affinity, which contributes to the low rate of oxidative phosphorylation in L. interrogans [4]. The reduced expression of this operon observed suggested that L. interrogans reduced its O 2 requirement upon interaction with host cells. Furthermore, most genes of the citric acid cycle (TCA cycle) as well as a putative operon encoding the leptospiral F 0 F 1 ATP synthetase subunits (LA2780-LA2782) [4], were significantly down-regulated upon interaction with macrophages.
Leptospira can not utilize glucose and other sugars as carbon source. Instead, it degrades long chain fatty acid through betaoxidation for carbon and energy production. In fact, tween-80, not glucose, are essential ingredients of the semi-synthetic EMJH medium for in vitro growth of Leptospira. Consistent with this, recent comparative genomics analyses on six leptospiral genomes revealed that Leptospira has limited sugar transport system [3,7]. As such, L. interrogans encodes more genes of the fatty acid metabolism genes than Escherichia coli K12. There are three homologs of long-chain-fatty-acid CoA ligase (LA0106, LA2177, and LA2309), which are the rate-limiting enzymes for fatty acid degradation [3]. Our microarray data revealed that expression of LA0106 and LA2309 were down-regulated (2-8 folds) during interaction with host cells. In addition, one of glycerol utilizing genes, LA0587, encoding lactonizing lipase, was dramatically down-regulated (4-9 folds). These results are consistent with down-regulation of its oxidative metabolism pathways described above.
In addition to the general trend of gene down-regulation, few genes in this category were up-regulated upon interaction with macrophages. Genes involved in nitrogen metabolism [2-nitropropane dioxygenase (LA2727), glutamate synthase (LB286)], were up-regulated. One of the genes in methane metabolism, the catalase gene (LA1859), was also significantly up-regulated 4-7 fold persistently. Since catalase is also involved in tryptophan metabolism and oxidative resistance, it remains to be determined the exact role the catalase plays during the initial stage of leptospiral infection. In addition, three genes in starch and sucrose metabolism [CDP-glucose 4,6-dehydratase (LA1632), alpha-glucosidase II (LA2944), glucose-6-phosphate isomerase (LA3888)] and three genes in glycolysis/gluconeogenesis [phosphoglyceromutase (LA0439), dihydrolipoamide dehydrogenase (LA2115), a interrogans Strain Lai 56601 possesses the genes involved in the biosynthesis of unsaturated fatty acids pathway (KEGG pathway: lil01040), which suggests that this strain may synthesize some unsaturated fatty acids de novo. In contrary, B. burgdorferi has no unsaturated fatty acids biosynthesis genes, and instead scavenges polyunsaturated fatty acids from BSK II growth medium or hosts [40]. The putative rate-limiting enzyme in this pathway, omega-6 fatty acid desaturase gene (delta-12 desaturase gene, desA, LA0502), was dramatically up-regulated in RPMI 1640 medium controls (.10-fold). The relatively modest and late up-regulation of this gene upon interaction with macrophages may reflect the culturing conditions and micro-environments, such as elevated temperature and osmolarity [12,15]. Considering that the RPMI 1640 medium controls and host cells were all eutrophic in unsaturated fatty acids, the implication and mechanism of the upregulations of desA were still unclear.
Oxygen tolerance and DNA repair L. interrogans must evade oxidative killing mediated by host cells including macrophages. However, the L. interrogans genome has only few predicted genes involved in resistance to oxidative stress and reactive oxygen species (ROS). All four pathogenic leptospiral genomes lack homologues of fqg, nfo, nei or superoxide dismutase (sod) [3,5]. L. interrogans Strain Lai 56601 has glutathione peroxidase genes (LA1007, LA4299) and thiol peroxidase gene (LA0862), but their level of expression were very low and did not have significant change upon interaction with macrophages. Pathogenic Leptospira also have cytochrome C oxidase genes, which may be involved in protection from O 2 stress. Our microarray results showed a significant down-regulation of these genes (LA0242-0244), suggesting they may not be important for resistance to oxidative killing in our models.
Catalase is one of the proteins that plays an important role in resisting oxidative killing by phagocytes [41]. Both pathogenic and non-pathogenic L. biflexa have catalase genes in their genomes, but they are not homologs and belong to different enzyme groups: L. interrogans has a heme-containing katE homolog, whereas L. biflexa has a heme-containing dual functioning peroxidase/catalase katG homolog [42]. Our microarray result showed that expression of katE was very high and further up-regulated during interaction with host cells. However, this up-regulation appeared not to be the result of direct interaction with macrophages, but rather due to other host factors such as elevated temperature and mammalian serum, since katE gene expression was also increased in the M samples. This was consistent with previous reports that these factors can influence catalase expression in Leptospira [12,15]. Interestingly, it was reported that non-pathogenic Leptospira is more susceptible to H 2 O 2 killing in vitro [43], which suggests katE may play an important role in Leptospira infection.
In addition, the high expression and up-regulation of 2-Cys thioredoxin peroxidase gene (LA2809) indicated this gene may contribute to resisting oxidative stress. It was significantly up-  Responses of L. interrogans to Macrophages www.plosntds.org regulated in RPMI 1640 medium (4-8 folds) and moderately upregulated upon interaction with macrophages (2-3 folds). Cluster analysis revealed that this gene, the fatty acid desaturase gene and the catalase gene were assembled into a same clade, Clade 3 ( Figure 2B). However, considering that it was not regulated by elevated temperature and host serum in previous studies [12,15], the mechanism of its up-regulation may be different from those of the fatty acid desaturase and the catalase.
All six Leptospira genomes contain integrated DNA repair systems, such as the base-excision repair, the photo reactivate, and the SOS repair, etc [7]. Unexpectedly, no significant upregulation of these genes in L. interrogans was found upon interaction with macrophages. The only major change was the down-regulation of the major recombinase A gene (recA, LA2179) in the homologous recombination pathway. The implication of this change remains unclear and suggested that L. interrogans experienced limited DNA damage under our conditions.

Signal transduction, chemotaxis and motility
A distinct feature of L. interrogans relative to other spirochetal pathogens such as B. burgdorferi and T. pallidum, is that it has more two-component signal transduction systems (with more than 26 pairs of histidine kinases and response regulators). Regulation of two-component systems often occurs at the level of phosphorylation, not at the level of transcription. As expected, few obvious changes were observed in the microarray analysis for the histidine kinase and response regulator genes except that a two-component response regulator gene (LA2548) and a neighboring sensory transduction histidine kinase gene (LA2549) were moderately upregulated temporarily at the 45 min time-point. In addition, other signal transduction related genes were differentially expressed. The acetyl-CoA acetyltransferase (LA0828) and the phosphate-binding protein PstS (LB297) were moderately down-regulated, while the operon including the potassium-transporting ATPase subunit A (kdpA, LA3112) and the potassium-transporting ATPase B chain (kdpB, LA3111) were moderately up-regulated at the 45 min timepoint. The only change in the GGDEF sensory system was that the sensory box/GGDEF family protein (LA2931) was moderately upregulated (Table 1).
Overall, genes involved in motility were down-regulated in L. interrogans upon interaction with macrophages. The major flagellin genes, flaB1 (LA2017) and flaB2 (LA2019), whose products had been confirmed by proteomics [39,44], were down-regulated with a wide range (3-20 fold), suggesting reduced motility which is correlated with the down-regulation of energy generating metabolism during this process. Unlike the persistent downregulation of the Clade 1, the expression of flaB1 and flaB2 were significantly down-regulated (15-20 fold) upon interaction with murine J774A.1 cell lines only at the 45 min time-point (J45/E0), but returned to their original E0 levels at the 90 min time-point (J90/E0). However, these transient down-regulations were not observed when L. interrogans interacted with human THP-1 cell lines. The down-regulation of flaB1/2 in T samples was maintained less than 6-fold during the 90-min period. These different regulation patterns in J and T samples seemed to be related to the differences of invasive motility and immune evasion ability of L. interrogans infecting murine and human macrophage cells [23]. In addition, flagellar motor switch protein fliY gene (LA2613) was also down-regulated; this down-regulation was verified by immunoblotting analysis in this study ( Figure 5). LA2418, which encodes another abundant flagellin protein, the hook-associated protein, showed 5 to 15-fold down-regulation. Consistent with the low motility and stable attachment of L. interrogans during the early stage of infection, a putative cheY gene (LA1253) which can interact with switch complex at the flagellar motor base to alter the rotating direction [45], was down-regulated at the time-point of 90 min in both T and J samples. The chemotaxis motA protein (LA0662) was up-regulated temporarily at the 45 min time-point ( Table 1). The flaB1/2 genes are typically used as an internal control for gene regulation in many studies. The observation of significant down-regulation of flaB1 and flaB2 shown herein suggests that they are not suitable as control genes, especially for gene expression under the in vivo conditions.

LPS and O-antigen synthesis
Unlike other none-lipopolysaccharide (LPS) spirochetes, Leptospira contains an intact and conserved biosynthesis system of LPS. Leptospiral LPS can activate host cells through TLR-2, which is distinct from other bacterial LPS [1,46]. The structure of leptospiral lipid A and oligosaccharide can be modified [47,48], which may attribute to the complicated diversity of more than 200 serovars. About 80% of ORFs involved in the leptospiral LPS synthesis pathway are similar to those of E. coli, which may indicate the highly conserved evolution of this genome locus [3]. However, unlike other gram-negative bacteria in which the lipid A biosynthesis genes are in an operon, these genes are scattered across the genome in L. interrogans. The O antigen synthesis genes remains clustered in the rfb operon with 103 kb in size [7].

Sphingomyelinases and hemolysins
Leptospiral sphingomyelinases and hemolysins are hypothesized to degrade the host cell membrane and acquire nutrition [3]. Some were verified to be haemolytic enzymes or pore-forming cytotoxin [49,50]. Pathogenic L. interrogans and L. borgpetersenii has 5 and 3 sphingomyelinase genes respectively, while non-pathogenic L. biflexa has no such genes [6]. Recent data showed that the major secreted sphingomyelinase C precursor gene (sph2, LA1029) is host-inducible during infection [51,52]. Our microarray data showed that the sph2 was expressed at extraordinary high level even in spirochetes cultivated in the EMJH medium. The sph2, along with another sphingomyelinase C precursor gene, sph1 (LA1027), were further up-regulated either in the RPMI 1640 medium controls or upon interaction with macrophage cells. Upregulation of these genes were likely due to changes in osmolarity, since it was previously shown that that expression of Sph2 increases when L. interrogans was grown in a physiologically relevant osmotic concentration [52].
Regarding non-sphingomyelinase hemolysins, one of the putative hemolysin genes, which is present only in pathogenic Leptospira, tlyC (LA3937), was up-regulated upon interaction with macrophages. The hemolytic activity of TlyC remains controversial. One earlier study showed that TlyC had hemolytic activity [50]. A recent study from another group reported that TlyC had no hemolytic activity, but it was a surface protein that mediates interaction with host extracellular matrix (ECM) [53]. Neverthe-Responses of L. interrogans to Macrophages www.plosntds.org

Other pathogenesis-related genes
Several adhesion or invasion-related genes were annotated in the original L. interrogans Serovar Lai Strain Lai 56601 genome, including mce, invA, atsE and mviN [3]. The precise functions for most of these genes are largely unknown, with exception for mviN. Recent study shows that MviN (MuiJ) of E. coli is a peptidoglycan lipid II flippase [54]. In this study, mviN, as well as another putative virulence factor, mce, were up-regulated in L. interrogans up interaction with macrophages.
L. interrogans has several additional virulence-associated genes such as a collagenase gene (LA0872), a PAF acetylhydrolase gene (pafAH, LA2144) [55], a von Willebrand factor type A gene (vwa, LA0697), and a paraoxonase gene (pon, LA0399). The homologs of pafAH, vwa and pon are also present in saprophyte Leptospira [7]. These genes were largely unchanged in this model, with the except that vwa and pon were transiently up-regulated upon interaction with murine macrophages (J samples).
Iron acquisition is one of the essential survival strategies many pathogens possess for establishing infection in mammalian hosts [56]. TonB-dependent receptors are associated with iron acquisition [57] and may transport heme and hemoprotein [58]. Some of these outer membrane receptor genes (LA0572, LA2641 and LA3258) were up-regulated, while another receptor gene (LA3242) was down-regulated (Table S2) upon interaction with macrophages. The iron resources of EMJH and RPMI 1640 medium controls (with 10% heat-inactivated FCS) are somewhat different. The available iron in the EMJH medium is free Fe 2+ , whereas in the mammalian cell culture medium, iron source is Fe 3+ in transferrin (TRF, siderophilin) provided by serum. This difference may also affect leptospiral iron transportation and uptake. For example, hemO (LB186) in Clade 6, the virulence gene encoding heme oxygenase for iron acquisition from hemoglobin [59,60], was transiently up-regulated only in the RPMI 1640 medium controls. This regulation could be due to a difference in iron status among the EMJH medium, the RPMI 1640 medium controls, and the macrophage-containing cell culture medium.

Outer membrane proteins and lipoproteins
The major components on the surface of Leptospira are transmembrane OMPs and lipoproteins [39]. These OMPs are functionally and structurally important for nutrition uptake, signal transduction, cell stabilization, and immunogenicity [61]. A mutant lacking ompA-like outer membrane protein, Loa22  Responses of L. interrogans to Macrophages www.plosntds.org (LA0222), had been generated, and showed that this gene was essential for L. interrogans pathogenesis [62]. Several major OMPs had been employed for developing subunit vaccines or serological tests for leptospirosis [37,63,64,65,66,67]. Based on genome annotation and lipoprotein prediction, there are at least 150 predicted lipoprotein genes (including predicted lipoproteins on inner and outer membrane) and about 100 predicted transmembrane OMP genes defined in the genomes of L. interrogans [3,4,11,68,69]. Cluster analysis of these 255 OMPs and lipoprotein genes revealed that half of these genes were upregulated, while 40% of them were down-regulated ( Figure 6). It was interesting to note that most of these changes were due to the interaction with macrophage rather than in the RPMI 1640 medium controls. Thus, it was likely that the membrane profiles underwent a series of dramatic changes upon interaction with macrophages. Most of the up-regulated genes were moderately regulated and almost all of these genes were putative transmembrane OMPs and putative lipoproteins not previously verified by proteomics [68]. One of the well-studied up-regulated genes was lig genes. The leptospiral lig genes encode several surface Lk90-like proteins containing immunoglobulin-like repeats, including LigA, LigB and LigC [7,70,71]. The L. interrogans Serovar Lai Strain Lai 56601 has LigB (LA3778) and LigC (LA3075) [3], whereas L. interrogans serovar Copenhageni strain Fiocruz L1-130 has LigA and LigB. Expression of ligB is up-regulated during infection, and LigB has been suggested as a putative virulence factor [72]. However, recent inactivation of ligB in L. interrogans Serovar Copenhageni Strain Fiocruz L1-130 did not affect leptospiral pathogenicity [9]. It was proposed that loss of LigB was compensated by the presence of LigA [8]. In this study, both expression of ligB and ligC in L. interrogans Serovar Lai Strain Lai 56601 were up-regulated either upon interaction with macrophages or in the RPMI 1640 medium controls alone. This result was consistent with the increased expression of ligB by evaluated temperature, host serum and during infection [15,73]. Concurrent up-regulation of LigB and LigC also supported the functional compensatory hypothesis among LigA, LigB, and LigC.
It is well established that leptospiral major OMPs are differentially expressed in semi-in vivo conditions or during infection [17,18,74]. We found that the expression levels of leptospiral OMPs were different in EMJH, Korthof and other leptospiral culture mediums (Date not shown). However, factors that contribute to differential regulation of these major surface proteins were largely unknown. Previous studies indicated that the common factors, including temperature, osmolarity, iron, or host serum, did not significantly influence expression of these major OMPs [12,13,14,15,59]. Our results indicated that the interaction with host cells was an important factor in triggering differential expression of major OMPs in L. interrogans, a phenomenon not achieviable using leptospiral culture media. Since most of these surface proteins are major antigens and the macrophage is an important antigen presenting cell of the host, down-regulation of this group of proteins was likely an immune evasion mechanism of L. interrogans [75].

Identification of potential transcription factors involving in differential gene expression
The above-mentioned significant changes of gene expression, especially the dramatic down-regulations of the major OMPs, created a problem when determining which regulation systems were involved in differential gene expression. The first step to solve this problem is to identify the TFs which were directly involved in the major changes. More than forty sigma factors, anti-sigma factors, and anti-sigma factor antagonists were defined in the genome of L. interrogans Serovar Lai Strain Lai 56601. Their expression levels were unchangeable upon interaction with host cells in our microarray study (Table S2). Especially, Sigma S (RpoS), the sigma factor that plays a key role in differential gene expression in another well-studied spirochete, B. burgdorferi, is not present in the six released Leptospira genomes [76]. The anti-sigma factors can control sigma factors activity at post-translation level, and may be involved in the major OMPs regulation.
B. burgdorferi has only a few of specific TFs, while there are lots of specific TF homologs defined in the leptospiral genomes, which increased the possibility that specific TFs also played important The ORFs down-regulated at least 5-fold in J or T samples were included in this table. The mean fold values were inverted into negative reciprocal values when the fold changes were 0.5 or less. The supplementary annotations generated in this study were showed in brackets. Clade ID: the clade ID for the significantly regulated ORF defined in genome-wide cluster analysis ( Figure 2B). doi:10.1371/journal.pntd.0000857.t002 roles in the regulation of the major OMPs. The leptospiral signal transduction proteins had recently been classified by domainbased rules in MiST2 database (http://mistdb.com/) [77]. In this database, the putative TFs were classified into several catalogs, such as one-component proteins, two-component proteins, and response regulators, etc, but not systematized into specific TF families named after their original function. In this study, the DBD definitions were obtained from InterPro integrative protein signature database by InterProScan program and well annotated by InterPro2Go. The InterPro database integrates PROSITE, PRINTS, Pfam, ProDom, SMART, TIGRFAMs, PIR superfamily, SUPERFAMILY Gene3D and PANTHER databases, which guaranteed the accuracy of the definition of the functional DBD domains [31]. All putative TFs were classified into specific TF families based on the original definition of the DBD domains, which enabled us to predict the potential function of the putative TFs. In addition, the phylogenic tree for each of the TF families was constructed based on the whole TF sequences, which helped us to compare the TF homologs within specific families, and identify the specific TFs that only existed in pathogenic Leptospira, which may be associated with leptospiral pathogenisis ( Figure 1B). Overall, the total number of specific TFs of non-pathogenic L. biflexa was almost twice than that of pathogenic Leptospira species (Table S3). That is, L. biflexa had about 100 specific TFs, while L. interrogans and L. borgpetersenii had less than 50 specific TFs. In addition, L. biflexa had higher proportions of TFs (TFs/ORFs) than pathogenic Leptospira, which is consistent with its strong survivability and high growth rate [6]. Based on domain analysis, 18 specific TF families were defined in six released leptospiral genomes (Figure 7). Several TF families were not found in all leptospiral genomes. The HTH_11 family existed only in L. interrogans, and the MerR, LytTR, LysR, Crp and GntR families existed only in L. biflexa. The CopG families existed in L. borgpetersenii and L. biflexa, but was absent in L. interrogans. (The previous definitions of CopG TFs in the genome of L. interrogans Lai 56601 were not precise, and there were no CopG TFs in the genome of L. interrogans Serovar Copenhageni Strain Fiocruz L1-130.) Based on the microarray date in this study, several specific TF genes with high expression levels were identified, such as LB333 of the OmpR family, LA3094 of the Fur family, LA1447 of the LexA family, LA0900 of the MarR family and LA3531 of the ArsR family, etc, which may contribute to the major regulation in our microarray study.
Most specific TF families of non-pathogenic L. biflexa were larger than those of the pathogenic Leptospira. The only exception was that the OmpR TF family of pathogenic Leptospira species (L. interrogans and L. borgpetersenii), which was larger than that of L. biflexa. Considering that the OmpR TF was first defined as a regulator of outer membrane porin genes (ompC and ompF) in E. coli [78,79], it is possible that the leptospiral OmpR TFs were also involved in the regulation of the porins or other membrane proteins. Pathogenic Leptospira may regulate OMPs more efficiently than non-pathogenic L. biflexa. Furthermore, this OMP regulation may be related with leptospiral pathogenisis. If so, it would be consistent with the down-regulation of the major OMPs observed herein and previously [18].
The molecular phylogeny of OmpR family revealed four monophyletic origins in all six Leptospira spp. (Figure 8A). Two exceptions were that LA3108 homologs were only found in L. interrogans, and LA1919 homologs only existed in pathogenic Leptospira species (L. interrogans and L. borgpetersenii). Based on domain analysis (Figure 7), LA1919 was supposed to encode a TF with seven putative transmembrane regions, which was seldom present in prokaryotes but common in eukaryotes. Of note, LB333 was the Responses of L. interrogans to Macrophages www.plosntds.org most abundantly expressed TF gene and the only OmpR TF gene highly expressed in EMJH medium, but was down-regulated significantly upon interaction with macrophages ( Figure 8B). Furthermore, it was concomitantly down-regulated with the group of major OMP genes including ompL1, lipL32, lipL41, lipL48 and ompL47, implying LB333 contributed to the differential regulation of this group of genes.

Concluding remarks
The global trancriptomic analyses of pathogens using a bacteria-host cell interaction model provide new understandings of immune evasion and pathogenesis for many bacterial pathogens [80,81,82]. Of note, a host-adapted model using a dialysis membrane chamber (DMC) for the Lyme disease spirochete, B. burgdorferi, has been used for genome-wide analysis of transcriptome in response to host-specific signals. However, this method can not be used for L. interrogans Serovar Lai Strain Lai 56601 as it can not survive and replicate within DMC (Date not shown). In this study, we adopted macrophage-derived cell models to analyze the leptospiral transcriptomic changes upon interaction with host cells or to host-specific signals. The adaptability to host microenvironments and immune evasion mechanism of L. interrogans revealed in this study were significantly different from those of the previous in vitro studies. Although the main virulence factors of Leptospira remain largely unknown due to the high difficulty of genetic manipulation, these significant changes of L. interrogans in different microenvironments led us to recommend that further research should be performed under conditions that imitate in vivo conditions as much as possible.
One of the major observations in this study was the dramatic down-regulation of major OMPs (e. g., LipL32, OmpL1, LipL41 and LipL48) upon interaction with host cells. This was in contrast to previous microarray analyses using various growth conditions, such as varied temperature, osmolarity, or the presence of host serum, which did not observe changes in expression of these genes. The fact that these genes, including LipL32, LipL41, and LipL21, etc, have been shown differentially regulated during infection, suggests that interaction with host is a key signal for regulating expression of these genes [17,18]. Our observations in this study suggested that down-regulation of these major OMPs may be important for the immune evasion of pathogenic Leptospira, a strategy similar to B. burgdorferi [83,84]. These changes in the OMPs profile also suggested that host cells induce a substantial change in surface protein profile, which is important when considerating vaccine candidates against leptospirosis [80].
Regulation of the major OMP genes of L. interrogans appeared to be different from that of B. burgdorferi. The expression levels of the major leptospiral OMPs, such as LipL32, LipL41, and OmpL1, etc, were relatively stable and not responsive to common environmental cues, whereas the major surface lipoprotein OspC of B. burgdorferi, is induced by elevated temperature, lowered pH, higher cell density, the presence of CO 2 or host serum. Regulation of ospC is directly controlled by sigma S (RpoS), which is further controlled by transcription factors including Rrp2, RpoN (sigma N), and BosR [76,85]. L. interrogans does not have rpoS homologue in its genome. The specific TFs may play important roles in gene regulation of the major leptospiral OMPs. Co-regulation of LB333 with the major OMPs suggests that LB333 may be involved in such regulation. In addition, the comparative and evolutionary relationship of all the leptospiral specific TFs revealed in this study also facilitates further research on identifying the regulation networks in Leptospira spp.
This report focused on the common regulation of L. interrogans Strain Lai 56601 infecting the murine and human macrophage cell lines, especially the persistent down-regulation of the major OMPs. However, several differences in the transcriptomic changes of L. interrogans Strain Lai 56601 infecting the murine vs. human macrophage cell lines were also observed. Cluster analysis on the global transcriptomics (Figure 2A) showed that the J90/E0 was somewhat similar to M45/E0 and M90/E0, which indicated that most of the gene regulations in L. interrogans upon interaction with murine macrophages (J90) were not persistent changes. For example, several genes of anti-anti-sigma factor (anti-sigma factor antagonists), ArsR TF, heat shock proteins, and chaperonins in Clade 4 were differently regulated in J and T samples, which may contribute to the differential regulations upon interaction with murine vs. human macrophages both at transcription level and at translation level. In addition, the major flagellin genes, flaB1/2 (LA2017 and LA2019), were persistently down-regulated in T samples, but only transiently down-regulated in J samples. Whether this difference contributed to the different behaviors of L. interrogans infecting murine vs. human macrophages remains unclear [23].
One limitation of our study was that macrophage cell lines instead of primary macrophages were used as infection models. However, our previous study had revealed that the behaviors of L. interrogans Strain Lai 56601 in the immortalized macrophage models were same as those in the primary macrophage models [23]. In addition, the high homogeneity and culturability of cell lines guaranteed the data reliability and repeatability in the highly sensitive microarray analysis. Therefore, only macrophage cell lines were employed as infection models in this leptospiral transcriptomics research. Another limitation was that the transcriptional responses revealed by microarray, qRT-PCR and Western blotting in this study only displayed the average and general regulations of all leptospiral cells infecting host cells. In fact, there were some individual differences among the invasive leptospiral cells morphologically [23]. Further cellular proteome analysis may complement the understanding of the individual regulation of pathogenic Leptospia during infection [86].