Distinct transcriptional profiles of Leptospira borgpetersenii serovar Hardjo strains JB197 and HB203 cultured at different temperatures

Background Leptospirosis is a zoonotic, bacterial disease, posing significant health risks to humans, livestock, and companion animals around the world. Symptoms range from asymptomatic to multi-organ failure in severe cases. Complex species-specific interactions exist between animal hosts and the infecting species, serovar, and strain of pathogen. Leptospira borgpetersenii serovar Hardjo strains HB203 and JB197 have a high level of genetic homology but cause different clinical presentation in the hamster model of infection; HB203 colonizes the kidney and presents with chronic shedding while JB197 causes severe organ failure and mortality. This study examines the transcriptome of L. borgpetersenii and characterizes differential gene expression profiles of strains HB203 and JB197 cultured at temperatures during routine laboratory conditions (29°C) and encountered during host infection (37°C). Methodology/Principal findings L. borgpetersenii serovar Hardjo strains JB197 and HB203 were isolated from the kidneys of experimentally infected hamsters and maintained at 29°C and 37°C. RNAseq revealed distinct gene expression profiles; 440 genes were differentially expressed (DE) between JB197 and HB203 at 29°C, and 179 genes were DE between strains at 37°C. Comparison of JB197 cultured at 29°C and 37°C identified 135 DE genes while 41 genes were DE in HB203 with those same culture conditions. The consistent differential expression of ligB, which encodes the outer membrane virulence factor LigB, was validated by immunoblotting and 2D-DIGE. Differential expression of lipopolysaccharide was also observed between JB197 and HB203. Conclusions/Significance Investigation of the L. borgpetersenii JB197 and HB203 transcriptome provides unique insight into the mechanistic differences between acute and chronic disease. Characterizing the nuances of strain to strain differences and investigating the environmental sensitivity of Leptospira to temperature is critical to the development and progress of leptospirosis prevention and treatment technologies, and is an important consideration when serovars are selected and propagated for use as bacterin vaccines as well as for the identification of novel therapeutic targets.

Introduction Global human leptospirosis incidence approaches an estimated 1.03 million annual cases, of which, almost 60,000 are estimated to be fatal [1]. Leptospira are shed in the urine of infected individuals, most commonly by reservoir hosts, and can be transmitted directly from animal to human or indirectly picked up from the environment. Disease presentation can vary from asymptomatic in reservoir hosts, to flu-like symptoms, and multi-organ failure in severe cases in incidental hosts [2]. L. interrogans is the leading cause of human infection but leptospirosis can affect most vertebrates including all major domestic livestock. In cattle, acute leptospirosis infections present with reproductive symptoms including abortion, embryonic loss, preterm calving, or low health calves [3,4], which can result in devastating animal health and economic damages. A recent study found that 7.2% of beef cattle from an abattoir in the central United States were actively shedding L. borgpetersenii [5].
Including pathogenic and saprophytic species, there are 64 species of Leptospira identified and hundreds of serovars [6]. The presentation of disease severity and reservoir host status is highly specific between host species and the serovar/species of the infecting Leptospira. For instance, while rats have been identified as reservoir hosts of leptospirosis since the early 1900s [7], most prominently of serovars Icterohaemorrhagiae and Copenhageni, cattle are recognized as the true reservoir hosts of serovar Hardjo [4,8]. A specific species of Leptospira may represent numerous serovars, and a single serovar can include numerous species, for example, serovar Hardjo can be represented by both species L. borgpetersenii (type Hardjo bovis) and L. interrogans (subtype Hardjo prajitno). Bacterin vaccines for leptospirosis lack cross protection between serogroups and have only weak cross protection across serovars [9,10]. Thus, most vaccines are composed of numerous serovars, using those most common in the designated region and target species. In cattle, vaccination for serovar Hardjo is highly sensitive to Leptospira species, where vaccination by L. borgpetersenii and L. interrogans may not offer cross species protection [11][12][13][14]. In fact, for serovar Hardjo, species specific interactions do not stop at the serovar or species level. JB197 and HB203 are two different strains of species L. borgpetersenii serovar Hardjo both isolated from cattle at slaughter; HB203 was isolated from 10 year old dairy cow in Kansas, USA and JB197 was isolated from a bull in Nebraska, USA [15]. Both strains are highly similar by sequence homology and indistinguishable by serovar and MAT (microscopic agglutination test) [16]. Yet HB203 inoculation in the hamster model of leptospirosis results in colonization of the kidney and the establishment of a chronic asymptomatic infection. In contrast, JB197 challenge will result in severe acute disease, with Leptospira identifiable in the blood and tissue and severe damage inflicted across numerous organs, resulting in death of the hamster [16]. This evidence strongly suggests that host interactions and environmental response at the strain level are critical to the biological control or failure to control Leptospira infections.
In the context of vaccine development, much emphasis has been placed on identifying outer membrane proteins or virulence factors that could potentially confer cross protection between species and serovars. In Gram-negative bacteria, lipopolysaccharide (LPS) is one of the best characterized pathogen-associated molecule patterns (PAMPs) and is an obvious candidate target for agglutinating antibody production. Leptospira LPS is considerably less endotoxic than other Gram-negative LPS such as Escherichia coli, however, leptospiral LPS still elicits a protective antibody response against lethal challenge [17]. The variable O-antigen region of leptospiral LPS is synthesized by genes in the rfb locus, which in L. borgpetersenii, contains 31 open reading frames [18,19]. Among other virulence candidates are the leptospiral immunoglobulin-like (Lig) proteins which have been established to have protective properties as immunogens. Expression of the Lig proteins are known to be influenced by osmolarity of their environment (increased at physiological osmolarity) and culture attenuation (decreased with serial culture passage) [20,21]. While some species such as L. interrogans carry the genes encoding both LigA and LigB, L. borgpetersenii only contains LigB [22]. Notably, while concurrent knock down of LigA and LigB in L.interrogans resulted in attenuated virulence [23], in an alternative study where the L. interrogans LigB gene was disrupted, but the LigA was still functional, virulence was not diminished and colonization of the kidney was still accomplished [24]. While this study established that in L. interrogans, LigB was not essential for virulence it is possible that LigA has redundant functions, making L. borgpetersenii an increasingly interesting model for the study of LigB.
Temperature is a major factor leptospires must navigate within the environments they encounter over their lifecycle. Leptospirosis is most common in tropical environments where frequent rainfall, warm temperatures, and soil pH create optimum environments for the survival and environmental persistence of Leptospira [25]. This can be illustrated in the context of the natural lifecycle of Leptospira, which may be shed in the urine of an infected host (leaving an host kidney similar to 37˚C) to a cooler, moist environment (soil, grass, forage, water source, etc. similar to 29˚C) and then once more be picked up by a new host where it will ultimately strive to persist again in the kidney or other tissue at basal host temperature. In the laboratory, leptospires are almost exclusively cultured at 29˚C, which potentially limits what behavior researchers are able to observe in conditions that do not best mimic a true in vivo environment. This is currently an area of progress in the leptospirosis research field as our group recently described methodology for successful primary culture of live organisms at 37˚C utilizing a new HAN media formulation [26]. For the first time, this has enabled the long term culture propagation of L. borgpetersenii directly isolated from host tissue at 29˚C or 37˚C [26].
While the L. borgpetersenii serovar Hardjo strains JB197 and HB203 produce such divergent severity of disease phenotypes, we sought to examine the transcriptomic behavior of these two strains in the context of an additional in vitro factor: temperature. In this study we establish distinct differential gene expression profiles of JB197 and HB203 each cultured at both 29˚C and 37˚C. We identified distinct profiles between strains and within strain between temperatures, emphasizing the importance of strain to strain variation and the acute sensitivity of Leptospira to its environment.

Ethics statement and bacteria
All animal experimentation was conducted in accordance with protocols as reviewed and approved by the Animal Care & Use Committee at the National Animal Disease Center, and as approved by USDA Institutional guidelines. Two groups of three golden Syrian hamsters (Mesocricetus auratus) were inoculated with L. borgpetersenii serovar Hardjo strain HB203 or JB197 as previously described [16]. In the case of hamsters infected with JB197, kidney tissue was harvested when hamsters presented with clinically severe symptoms (blood on nose, pads, or urogenital tract, weight loss, dehydration, general poor condition, etc.) at approximately five days post-infection. In the case of hamsters infected with HB203, kidney tissue was harvested at three weeks post-infection. In both cases, kidney tissue was cultured to recover strains JB197 and HB203 in HAN medium incubated at 29 and 37˚C, and as previously described [26]. Recovered isolates from a single kidney from each strain were sub-cultured into HAN liquid media at indicated temperatures and harvested by centrifugation (10,000 x g, 4˚C, 30 min) at mid-late log phase of growth (~2 x 10 8 leptospires/mL). Four biological replicates were utilized for RNA isolation and additional bacterial pellets were frozen down for additional analysis. RNA samples used for RT-qPCR. Frozen original bacterial pellets containing 2.5-5 x 10 8 leptospires were later thawed and utilized for RT-qPCR validation using the Trizol method as previously described [27]. Briefly, 1 mL Trizol (Invitrogen, CA, USA) was added to frozen pellets and pipetted slowly until thawed. Samples were vortexed and incubated for 10 minutes at room temperature at which time, 260 uL chloroform was added, samples were vigorously shaken, and incubated for another 10 minutes at room temperature. Samples were centrifuged at 12,000 x g for 10 minutes at 4˚C to achieve phase separation; the aqueous phase was collected and 660 uL isopropanol was added to facilitate nucleic acid precipitation. After 10 minutes incubation at room temperature samples were centrifuged at 12,000 x g for 10 minutes at 4˚C whereafter resulting pellets were washed with 75% ethanol. Washed pellets were dried for 30 minutes using a Speedvac. Dry pellets were resuspended in 50 uL RNAse free water and incubated at 55˚C for 10 minutes. For residual DNA removal, samples were additionally treated with the Turbo DNA free kit (Ambion, Cat. No. AM1907, TX, USA) to manufacturer's specifications. RNA was examined for quality by Aligent Bioanalzyer and quantified using the Qubit RNA system (ThermoFisher, MA, USA) according to manufacturer's directions.

Sequencing
For sequencing, RNA was submitted to Iowa State University DNA Facility for library preparation using the TruSeq stranded total RNA kit (Cat. No. RS-122-2203, Illumina, Inc., San Diego, CA, USA) and sequenced using 150 Cycle single-read on an Illumina HiSeq 3000.

Analysis & genome alignment
RNAseq reads were aligned to the JB197 reference genome (RefSeq ID: chromosome 1: NC_008510.1, chromosome 2: NC_008511.1) and/or the HB203 reference genome (RefSeq ID: chromosome 1: NZ_CP021412.1, chromosome 2: NZ_CP021413.1) using Segemehl [28] with default mapping parameters. For differential gene expression analysis between strains, reads from HB203 were mapped to the JB197. Uniquely aligned reads (reads aligned to only one location in genome) were only processed for the downstream analysis and reads were formatted graphically to visualize in the Integrated Genome Browser (IGB [29]). Read counts (number of reads that aligned to a specific gene) for each gene were quantified using in-house Perl scripts. The genomic sequences of JB197 and HB203 were compared using rapid dotter software GEPARD (http://cube.univie.ac.at/gepard) with "word-length" parameter 750 for chromosome 1 and 100 for chromosome 2.

Statistics
Differential expression (DE) analysis was completed in R (v3.6.1; https://www.r-project.org/) using the DESeq2 package (v1.29.0 [30]). Briefly, raw read counts were normalized and reproducibility of the biological replicates was examined using Spearman rank correlation. Principle Component Analysis (PCA) was performed to determine outliers in the RNAseq dataset. Genes with normalized read counts less than 10 in less than three of the four replicates within a condition were dropped from DE analysis. To identify DE genes, four independent analyses (JB197 37˚C vs. 29˚C, HB203 37˚C vs. 29˚C, 29˚C JB197 vs. HB203, and 37˚C JB197 vs. HB203) were conducted using linear regression models within DESeq2. For a gene to be considered significantly differentially expressed, it was required to have an adjusted p-value < 0.05 and have a fold change (FC) greater than or equal to three. For JB197 versus HB203 comparison, positive fold change values indicate higher expression in JB197; between temperatures, 37˚C vs. 29˚C, positive values indicate upregulation at 37˚C. PCA and heatmaps were created in R (version 3.6.1) [31]. Venn diagrams were constructed with R and Venny [32].

RT-qPCR
Select genes were chosen for additional analysis and RNAseq validation by RT-qPCR. RNA was isolated and quantified as described above, and cDNA conversion was completed using the iScript Reverse Transcription Supermix (BioRad) and manufacturer's instructions. Of the common bacterial RT-qPCR control genes that had been previously validated [33], most were differentially expressed in at least one of our primary contrasts of interest (29˚C (JB197 vs. HB203), 37˚C (JB197 vs. HB203), JB197 (37˚C vs. 29˚C), HB203 (37˚C vs. 29˚C)). As a result, two separate control genes were utilized. One control gene was secA that was not differentially expressed and was suitable as a control for 29˚C (JB197 vs. HB203) and JB197 (29˚C vs. 37˚C) contrasts. The second control gene was rho, which was not differentially expressed and was suitable as a control for 37˚C (JB197 vs. HB203) and HB203 (37˚C vs. 29˚C). RNA concentrations were normalized to the condition with the lowest amount of RNA present prior to cDNA conversion. RT-qPCR was performed in 20-μL reactions (1-μL cDNA, 400nM of each forward (F) and reverse (R) primers, and 10-μL of SYBR green PCR mix (BioRad)) under the following conditions; 50˚C for 2 minutes, 95˚C for 10 minutes, followed by 40 cycles of 95˚C for 15 seconds and 60˚C for 30 seconds. Primers used were as follows: Prior to RT-qPCR, primers were tested by PCR and had primer efficiencies calculated. For all RT-qPCR runs matched reverse transcriptase negative (RT-) and no template controls were performed for quality control. The 2 −ΔΔCT method was used for data analysis, and error bars indicated in RT-qPCR figures represent 95% confidence intervals.

and 2-D gel electrophoresis and immunoblotting
Leptospires were harvested by centrifugation (10,000 x g, 4˚C, 30 min), washed twice with PBS, and processed for one-dimensional (1-D) SDSPAGE on 12% acrylamide gels (BioRad, CA, USA) as per manufacturer's guidelines. Proteins were visualized by staining with Sypro Ruby (Invitrogen,CA, USA) and lipopolysaccharide was visualized by staining with Pro-Q Emerald 300 (Invitrogen, CA, USA) as per manufacturer's guidelines. For immunoblotting, samples were transferred by semi-dry transfer (Amersham TE77 PWR) to Immobilon-P transfer membrane (Millipore, 220 Bedford, MA, USA) and blocked overnight at 4˚C with Starting-Block (PBS) blocking buffer (Thermo Scientific, MA, USA). Membranes were individually incubated with indicated antisera diluted in blocking buffer (anti-LigA/B at 1:5,000 or anti-LipL45 at 1:2,000) followed by incubation with horseradish-peroxidase anti-rabbit immunoglobulin G conjugate diluted 1:4,000 in blocking buffer (Sigma, MO, USA). Bound conjugates were detected using Clarity Western ECL substrate (BioRad, CA) and images acquired using a Bio-Rad ChemiDoc MP imaging system. Immunoblots were also performed on unwashed leptospires as used for isolation of RNA. Representative images of results are provided.
For 2-D DIGE, fractions enriched for outer membrane (OM) proteins were prepared from strain JB197 or HB203, cultured at both 29 and 37˚C, in triplicate, using Triton X-114 as previously described [34]. OM enriched fractions were resuspended in solubilization buffer (7M Urea, 2M Thiourea, 1% ASB-14) and quantified using the RC DC protein assay (BioRad, CA, USA). Protein samples (20 μg) were labeled with Cy3 or Cy5 (GE Healthcare, IL, USA) for comparison as indicated in Table 1; for each gel, 20 μg of internal standard, comprising equal μg amounts of all 12 replicates, were labelled with Cy2. 2-D gel electrophoresis of labeled samples was performed using 24cm IPG strips, pH 3-10, as previously described [35]. Gels were scanned using the Typhoon fluorescence gel scanner (Amersham) using different bandpass filters (520 nm for Cy2, 580 nm for Cy3 and 670 nm for Cy5) to image each of the three CyDyes (GE Healthcare, IL, USA). Differential protein expression was analyzed using Same-Spots (TotalLab, UK) following the software manual. The differentially expressed protein spot of interest (corresponding to predicted mass of LigB) was excised from a master gel comprising OM enriched fraction of HB203 at 29˚C and processed for identification by mass spectrometry using the In-gel Tryptic digestion kit and C18 Spin columns (ThermoFisher Scientific, MA, USA) as per manufacturer's instructions.

LC-MSMS and database searching
Peptides from the gel spot were separated by HPLC chromatography using a Proxeon Easy-nLC (Thermo Fisher Scientific, West Palm Beach, FL, USA) connected to the mass spectrometer. The chromatography used a trapping column (Proxeon Easy-Column, 2 cm, ID 100 μm, 5um, 120A, C18) and an analytical column (Proxeon Easy-Column, 10 cm, ID 75 μm, 3um, 120A, C18). The gradient used a mobile phase A (95% H 2 O: 5% acetonitrile and 0.1% formic acid) and mobile phase B (5% H 2 O: 95% acetonitrile and 0.1% formic acid). The analytical column was connected to a PicoTip Emitter (New Objectives, Woburn, MA, USA; FS360-75-15-N-20) cut to size. The column and Emitter were attached to a LTQ OrbiTrap Velos Pro (Thermo Fisher Scientific, West Palm Beach, FL, USA) mass spectrometer using the Proxeon Nanospray Flex Ion Source. The capillary temperature was set at 275˚C and spray voltage was 2.4 kV. The mass spectrometer used a data dependent method. In MS mode the instrument was set to scan 300-2000 m/z with a resolution of 60,000 FWHM. A minimal signal of 10,000 could trigger MSMS and 10 consecutive MSMS were possible. The activation type used was CID. The normalized collision energy was set to 35 and repeat mass exclusion was set to 120 seconds.
Tandem mass spectra were analyzed using Proteome Discoverer version 2.2. All MS/MS samples were analyzed using Sequest HT (Thermo Fisher Scientific, CA, USA) assuming digestion with trypsin. Proteome Discoverer search used a fasta database downloaded from UniProt. The fasta database was generated by searching Uniprot with the criteria "Hardjo", then limiting the database using the UniRef 90% feature [36]. The Hardjo UniRef90 fasta file was generated in July 2020 and consisted of 8312 entries. The mass spectrometer.raw file was searched with a fragment ion mass tolerance of 0.6 Da, a parent ion tolerance of 10.0 PPM, and allowed 2 missed cleavages. Carbamidomethyl of cysteine was specified as fixed modifications. Deamidation of asparagine and glutamine, and oxidation of methionine were specified as variable modifications.

Different protein profiles distinguish L. borgpetersenii serovar Hardjo strain JB197 compared to strain HB203
Strains JB197 and HB203 are indistinguishable by serovar testing, but significant variation in disease presentation in the hamster model point to meaningful differences that may impact host recognition or immune escape by the pathogen. Strain JB197 and strain HB203 were cultured from the kidney of hamsters with severe acute, or persistent chronic leptospirosis respectively and as previously described [16], and culture maintained at 29 or 37˚C. Serovar Hardjo is highly fastidious but the development of a new media, designated HAN media, has enabled serovar Hardjo to be cultured directly from a mammalian host at 29 or 37˚C [26]. Comparative analysis of protein content suggests that different protein profiles are expressed by strain JB197 compared to HB203 at both 29 and 37˚C,

Comparative genome analysis of serovar Hardjo strain JB197 and strain HB203
The complete genome comprising both chromosomes of L. borgpetersenii serovar Hardjo strain JB197 and strain HB203 were compared by genome alignment, S1 Fig. Both chromosome 1 and chromosome 2 were highly similar between strains (NCBI BLAST percent identity = 99.85% for chromosome I, and 99.90% for chromosome II). As previously described for serovar Hardjo strain JB197 compared to serovar Hardjo strain L550, an inversion in the middle of chromosome I differentiates strain JB197 from HB203, S1A Fig.

The transcriptome of serovar Hardjo strain JB197 and HB203
Several factors contributed to our interest in the transcriptome of strains JB197 and HB203 at 29 and 37˚C. First, serovar Hardjo strain JB197 causes an acute lethal disease in experimentally infected hamsters compared to serovar Hardjo strain HB203 which causes a persistent renal colonization. Second, the ability to culture serovar Hardjo directly from hamster tissue in HAN media at 29 or 37˚C has only recently become possible [26], and thirdly, there was evidence of differential protein expression between and within strains under different growth temperatures at the total protein level (Fig 1). RNAseq was performed on four biological replicates of each strain and temperature condition. Mapping details including uniquely and total mapped reads, are presented in S1 Table. Principle Component Analysis (PCA) indicated that strain and temperature collectively explained 89% of total variation (PC1 = 72%, PC2 = 17%) (Fig 2), and that our RNAseq biological replicates within each condition were robust and highly reproducible (Fig 2, Spearman correlations seen in S2 Fig). While all datasets clustered closely to their respective strain and growth temperature conditions, it was clear that HB203 at 29˚C and HB203 at 37˚C clustered much more tightly to each other than JB197 at 29˚C and JB197 at 37˚C, suggesting greater variation between stains, and that HB203 exhibited less of a global difference due to temperature.
To explore the effects of strain and temperature on gene expression profiles, differential expression (DE) of four main contrasts of experimental groups were analyzed. The first

PLOS NEGLECTED TROPICAL DISEASES
contrast examined differences between the two strains cultured at the classic leptospiral isolation and cultivation temperature of 29˚C: JB197 29˚C vs. HB203 29˚C. The second contrast examined strain differences cultured at the newly achieved 37˚C temperature: JB197 37˚C vs. HB203 37˚C. Since such diverse clinical signs of disease are seen in the hamster model between the acute lethal infection caused by JB197 and the chronic presentation of HB203, the third and fourth contrasts captured the effect of temperature on each strain individually, JB197 37˚C vs. JB197 29˚C, and HB203 37˚C vs. HB203 29˚C, respectively.
To broadly visually illustrate gene DE in our contrasts of interest, heatmap profiles of the top 25 genes showing the most difference in overall expression (by Z-score) of each of these contrasts are presented , Fig 3. Differences in transcriptomic profiles are prominent in each of the major contrasts of interest. In agreement with the PCA plot, color intensity indicative of Zscore indicates stronger expression differences between JB197 and HB203 at 29˚C, Fig 3A, and The DESeq2 package was used to identify genes that were significantly differentially expressed with a minimum fold change (FC) of three and adjusted p-value < 0.05 (adj. pvalue) in each of the major contrasts of interest. For the 29˚C contrast between strain JB197 vs. HB203, 440 DE genes were identified, the fifty most significant of which are reported in Table 2, and the complete list in S2 Table. Analysis of JB197 and HB203 cultured at 37˚C identified 179 DE genes, the fifty most significant of which are reported in Table 3 and the complete list in S3 Table. Within strain JB197, 135 DE genes were identified between 37˚C and 29˚C growth conditions, the fifty most significant of which are reported in Table 4 and the full list can be viewed in S4 Table. Within HB203, 41 DE genes were identified between 37˚C and 29˚C growth conditions, all of which are reported in Table 5; the complete list of DE genes, including those not significantly DE expressed by a three-fold change minimum for this contrast, are reported in S5 Table. To identify DE genes conserved across comparisons, Venn diagrams were constructed amongst the four contrasts of interest, Fig 4. Venn diagrams were created separately for 'down' regulated ( Fig 4A) and 'up' regulated ( Fig 4B) DE genes. Surprisingly, very little overlap exists between the DE gene profiles between contrasts. No genes were identified as being shared between all contrasts in the down regulated/negative fold change diagram (Fig 4A), and the lone gene (LBJ_RS13430, currently annotated as YP_801646.1) that was identified in all four contrasts of the up regulated/positive fold change diagram (Fig 4B) encodes a hypothetical protein with no predicted function or conserved motifs. The most conserved gene groups appear between 29˚C and 37˚C contrasts, with 61 genes being down regulated in JB197 compared to HB203 at 29˚C and 37˚C (blue and green circle overlap Fig 4A), and 51 genes being up regulated in JB197 compared to HB203 in 29˚C and 37˚C (blue and green circle overlap Fig 4B). The 61 shared genes down regulated in JB197 within both temperatures include several lipoproteins (LBJ_RS19300, LIC_13355 family lipoprotein, LBJ_RS15535), LigB (LBJ_RS03415), a universal stress protein (LBJ_RS18695), and genes predicted to encode hypothetical proteins. Of the 51 genes up regulated in JB197 compared to HB203 at both temperatures, many were transposases of either the IS110 or ISF5 family, several transport proteins (MMPL family transporter LBJ_RS12280), and efflux RND transporter permease subunit (LBJ_RS03290)), as well as similar to the down regulated series, genes encoding hypothetical proteins.

Most highly expressed genes
The most highly expressed genes and annotated sRNAs by normalized read count, in both JB197 and HB203, at both 37 and 29˚C, include tmRNA, RNase P, dnaK, groEL, Hsp20, elongation factor Tu, elongation factor G, DNA-directed RNA polymerase subunits, flagellin, as well as several genes encoding outer membrane proteins including lipL32, lipL21 and lipL41.

LigB is lowly expressed by JB197 compared to HB203
Notably, the well characterized outer membrane lipoprotein LigB encoding gene was identified as highly differentially expressed between the two strains (lower in JB197 compared to HB203), whether cultured at 29˚C (adj. p-value = 8.9E-278, FC = -54.1) or 37˚C (adj. pvalue = 6.1E-10, FC = -6.1) (Tables 2 and S2). Within JB197, ligB was significantly more highly expressed at 37˚C compared to 29˚C (adj. p-value = 7.0E-7, FC = 2.3), (S4 Table). In contrast, ligB was significantly more highly expressed by HB203 at 29˚C compared to 37˚C (adj. pvalue = 6.8E-8, FC = -3.9), Table 5. Given the role of LigB as a well characterized virulence factor, as well as its reported ability to act as a protective vaccinogen of pathogenic leptospires, the expression of ligB was investigated further, Fig 5. Visualization of the RNAseq data by IGB confirms the large numbers of transcripts for ligB in HB203, at both 29 and 37˚C, compared to that of JB197 (Fig 5A) (note the difference in read count scales). Differential expression of ligB was confirmed by RT-qPCR by comparison with two different control genes. Since common bacterial control genes varied in levels of expression across contrasts, secA was used as the control gene for the 29˚C (JB197 vs. HB203) and JB197 (37˚C vs. 29˚C) contrasts (Fig 5B), and rho was used as the control gene for the 37˚C (JB197 vs. HB203) and HB203 (37˚C vs. 29˚C) contrasts (Fig 5C). In addition, immunoblotting of whole cell sonicates with antibody specific for LigB confirms that HB203 expresses much larger amounts of LigB compared to that of JB197 (Fig 5D). In agreement with RNAseq data, detection of LigB by immunoblotting is diminished in HB203 when cultured at 37 compared to 29˚C. However, in contrast to RNAseq data which indicates that expression of LigB in JB197 is increased when cultures are maintained at 37˚C compared to 29˚C, the opposite was apparent by immunoblotting and less LigB was detected in JB197 maintained at 37˚C compared to 29˚C. As an additional control to demonstrate contrasting differential expression of antigens, immunoblots were also performed with anti-LipL45, an outer membrane lipoprotein originally annotated as Qlp42 [37,38] which is increased in expression in L. interrogans cultured at 37˚C compared to 29˚C, and  identified as a gene that was significantly DE by RNAseq in HB203 at 37˚C compared to 29˚C (adj. p-value = 1.96E-58, FC = 4.5) (Fig 5D and Table 5). To further investigate and validate expression of LigB as a component of the serovar Hardjo outer membrane, fractions enriched for the outer membrane proteome of HB203 were compared to that of JB197 by 2-D DIGE; a protein spot identified as LigB by mass spectrometry was detected in six biological replicates of HB203 cultured at 29 or 37˚C in larger amounts compared to that of six biological replicates of JB197 (p-value = 2.9E-07, q = 6.3E-07, Power � 0.999, FC = 28.7) (Fig 5F).

Differential expression of lipopolysaccharide and genes within the rfb locus
LPS of pathogenic leptospires is defined according to serovar status and is considered a protective antigen against homologous challenge. Inclusion of specific serovars within bacterin vaccines are believed to mediate protection via LPS. Given the importance of individual strains of Leptospira as components of bacterin vaccines to mediate protection against other strains of the same serovar, the expression of LPS by serovar Hardjo strains JB197 and HB203 was investigated further. Interestingly, JB197 produced larger amounts of LPS compared to that of HB203 when visualized after separation by gel electrophoresis (Fig 6). Accordingly, genes within the rfb locus, as defined by GenBank Accession AF078135.1 [18], were examined more closely to identify DE genes. No genes within the rfb locus met the DE criteria of significance and fold change of three. However, when fold change criterion was lowered to two, nine genes were DE between JB197 and HB203 cultured at 29˚C ( Table 6). Surprisingly, all nine were more highly expressed in HB203 compared to JB197. At 37˚C, no genes were DE with a twofold significance threshold, although LBJ_RS06485 annotated as a methyltransferase, was DE at 1.95 fold (adj. p-value = 5.6E-09) and present in higher amounts in HB203 compared to JB197.
Closer inspection of the rfb locus with IGB identified non-coding expression in the flanking intergenic regions of select genes within the rfb loci (S5 Fig). Additional genes not contained within the rfb locus, but annotated as having a function related to LPS were identified; these included LBJ_RS07965 (annotated as encoding an LPS export ABC transporter ATP-binding protein), LBJ_RS07975 (annotated as encoding the LPS  export ABC transported periplasmic protein LptC protein), LBJ_RS08415 (annotated as encoding an O-antigen ligase family protein) and LBJ_RS09125 (annotated as encoding the LPS-assembly protein LptD). Although several of these genes met adjusted p-value thresholds for significance, none were significant by either fold change three or fold change two standard (see Table 6).

Discussion
While the genomes of L. borgpetersenii serovar Hardjo strains JB197 and HB203 are strikingly similar at the nucleotide level and the serovar identity identical (both type Hardjo bovis), the disease phenotypes produced in a hamster infection model are drastically different. This suggests that variation at the strain level of Leptospira classification plays an important role in host response and species-specific interactions between host and pathogen. Interestingly, a genomic comparison identified that L. borgpetersenii is~700 kb smaller than L. interrogans and was hypothesized to be undergoing insertion sequence mediated genome reduction, with evolution towards a strict-host-host transmission cycle and loss of gene function centered on impairment of environmental sensing and metabolite transport and utilization [22]. The characterization of the L. borgpetersenii transcriptome is an important asset in assessing conserved and unique factors of infection across strains and serovars as well as comparing evolutionary changes across divergent species of Leptospira. The unique species-specific attributes of leptospiral interactions (both at the host and pathogen level), along with their sensitive response to the environment are major contributing factors heavily associated with persistent disease transmission. Understanding the sensitivity and specificity of Leptospira at the strain level is critical to the development of the next generation of therapeutics and ultimately is required for disease control. When examining the genes most highly expressed across conditions, many are identical to the top 50 most expressed genes previously identified for L. interrogans serovar Copenhageni strain L1-130 when cultured at either 29˚C, or within dialysis membrane chambers implanted in the peritoneal cavity of rats. Among these highly expressed genes were ligB, lipL32, lipL21 and lipL41, despite the different growth environments offered by EMJH versus HAN media, as well as significant genomic differences [39]. This suggests a level of consistency of base leptospiral mechanics across pathogenic species and across vastly different environments. A base level of conserved transcriptomic activity also helps emphasize strain specific nuances that may play an important role in Leptospira and host interactions, severity of disease, and immune response and/or escape. Also of interest, this study highlights several genes encoding hypothetical proteins that are conserved across contrasts of interest. For example, LBJ_RS01915 is a gene encoding a hypothetical protein that is in the top fifty most highly expressed genes in all of the major contrasts of interest (see S2-S5 Tables). Similarly, both genes LBJ_RS11060 (HB203 B9T54_RS15165) and LBJ_RS01505 (HB203 B9T54_RS10450) encode hypothetical proteins that are significantly DE in both 29˚C and 37˚C contrasts comparing the two strains and both genes are more highly expressed in HB203 compared to JB197

PLOS NEGLECTED TROPICAL DISEASES
at both temperatures (see S2 and S3 Tables). These hypothetical proteins warrant further characterization which is now being facilitated via the use of HAN media allowing culture maintenance at 37˚C and the continued development of Leptospira mutagenesis [40,41]. This is also further evidence that while leptospirosis research has focused on pan-genome genetic typing

PLOS NEGLECTED TROPICAL DISEASES
of Leptospira species, a 'pan-transcriptome' approach may offer novel insights to serovar or species-specific behavior. The divergent phenotypes seen in the hamster model after experimental challenge and the differences in total protein composition seen in Fig 1 suggested additional investigation of strain specific profiles at the transcriptomic level could be insightful. We investigated the differential gene expression between JB197 and HB203 cultured at 37˚C and 29˚C. As illustrated by the PCA analysis (Fig 2) and supported by heatmap visualization (Fig 3), the data shows biological replicates clustered more tightly together due to strain than due to temperature. Although, HB203 at 37˚C and 29˚C cluster more tightly than JB197 at 37˚C and 29˚C, which suggested greater gene expression differences within JB197 between temperatures than in HB203. The high number of DE genes in the JB197 contrast (135 genes between 37˚C and 29˚C) compared to only 41 DE genes for the HB203 contrast (37˚C and 29˚C) further supports this observation. Interestingly the highest levels of differential gene expression are seen from the 29˚C contrast (JB197 vs. HB203), both visually by heatmap (Fig 3) and by number of DE genes (440 compared to 197 for the 37˚C contrast between strains). This is a particularly critical result for several reasons. First, this is the temperature at which all in vitro culture of Leptospira have traditionally been maintained, which has vast implications for the differential behavior of different strains, especially those being cultured for bacterin vaccine use. Second, this more closely resembles the temperature of ex-host environment natural leptospires may encounter in the wild. The characterization of these gene expression changes is important for illuminating changes leptospires go through as they travel from host to environment to host. Changes occurring at these transition points may point to targets for the development of prevention technologies or environmental clean-up and control. The high levels of DE seen in the 29˚C contrast were not seen with the strain comparisons at 37˚C, suggesting a possible maintenance or homeostasis of the pathogen at an incubated temperature similar to what a host kidney/tissue environment would look like. This is also interesting in the context of colonization of leptospires in different tissues and the interaction with the host immune system that can result in clearance of the pathogen, chronic colonization of the host kidney, or severe multi-organ infection depending on the challenging Leptospira. Further, interesting among highly DE genes was the lack of conservation of DE genes across all conditions. While the most conserved genes were found in the comparisons between strains at 29˚C and 37˚C, the Venn diagrams of all DE analyses illustrate a lack of conservation among DE genes, suggesting a high level of sensitivity and strain-specific expression (Fig 4).
While a high number of genes encoding hypothetical proteins were highly DE, so were highly characterized membrane components including LigB, LipL45, and LPS which are known to interact with the host immune response (Figs 5 and 6). A subunit vaccine from a conserved portion of LigB has already been established to provide a protective immune response against challenge in the hamster model [42] and additionally, LigA and LigB surface protein expression levels change when L. interrogans is expanded at 37˚C compared to 30˚C [43], which suggests a temperature response of interest. Interestingly, in the present study ligB was inversely differentially expressed within strain by temperature (upregulated in JB197 at 37˚C compared to 29˚C and downregulated in HB203 at 37˚C compared to 29˚C). Collectively this indicates that surface proteins critical to host immune recognition of the pathogen are highly strain specific and can be influenced by environmental temperature. LipL45 is recognized as a membrane lipoprotein which has previously been reported to have been upregulated at 37˚C compared to 30˚C [37], which is consistent for the data reported here with LipL45 upregulated at 37˚C in both JB197 and HB203 compared to 29˚C (Tables 5 and S4). This evidence that certain immunogenic surface proteins, such as LipL45, are upregulated at traditional host temperature of 37˚C, makes a strong case for testing virulence factors at the newly achievable culture temperature of 37˚C with HAN media. Also important is the consideration of culturing bacterins for potential vaccine production at 37˚C opposed to the traditional temperature of 29˚C, which may increase the expression of certain virulence factors similar to LipL45, which was successful at inducing humoral and cell mediated responses in an early DNA vaccine construct [44]. LPS is recognized as a classic facilitator of pathogen and host

PLOS NEGLECTED TROPICAL DISEASES
immune response interaction. The O-antigen variable portion of LPS, encoded by the rfb region, has also been implicated in complement and adhesion mechanisms, and has previously been shown to be differentially expressed by leptospires and state of infection (more greatly expressed by leptospires recovered from the renal tissue of chronically infected rats and weakly expressed in leptospires sourced from acutely ill guinea pig tissue) [45]. Consistent with these results, this study highlights several genes within the rfb locus that are significantly Table 6. Differential expression examination of genes within the rfb locus as well as outside the rfb locus but annotated as pertaining to LPS. Differentially expressed genes with an adjusted p-value < 0.05, and fold change minimum of two are shaded in gray. proteomics, and LPS not only has implications for the selection of specific strains utilized in companion animal and livestock species vaccine design, but also for selection of temperature at which bacterin sources should be cultured and consistently maintained. Further, investigation of unannotated transcriptomic activity highlights the probability of other regulatory elements, such as sRNAs, or other potential transcriptional regulators, which may play roles in strain specific behavior that will continue to be important to characterize. Collectively, this study establishes that gene expression is highly specific at the strain level when comparing Leptospira even of the same species and serovar. Additionally, gene expression is greatly sensitive to temperature changes similar to those a leptospire may encounter naturally during its life cycle. Characterization of the transcriptome of L. borgpetersenii serovar Hardjo strains JB197 and HB203 provides insights into factors which may correlate with acute versus chronic disease in the hamster model of infection. The expression of known protective antigens, including LPS and outer membrane proteins such as LigB, can differ significantly between strains, information which needs to be defined when selecting strains for use as bacterin vaccines in domestic animals. Finally, the transcriptome of leptospires cultured at 37˚C appears less variable than that of those cultured at 29˚C, an additional consideration in bacterin preparation; any additional protective capacity of such bacterin preparations remains to be determined.

Gene ID
Supporting information S1 Table. Mapping statistics for RNAseq data including the total number of reads sequenced as well as un-mapped, multi-mapped, and uniquely mapped reads. (XLSX) S2 Table. Differential expression of annotated genes for the 29˚C contrast (JB197 29˚C vs. HB203 29˚C). Genes that were considered significantly differentially expressed were required to have an adjusted p-value < 0.05, have a fold change (FC) greater than or equal to three, and a count minimum of at least ten normalized counts in three out of four replicates for a single condition (strain and temperature combination). Significantly DE genes are highlighted in yellow. The first sheet consists of all DE analyses sorted by adjusted p-value, and the second sheet consists of all DE genes sorted by total expression. (XLSX) S3 Table. Differential expression of annotated genes for the 37˚C contrast (JB197 37˚C vs. HB203 37˚C). Genes that were considered significantly differentially expressed were required to have an adjusted p-value < 0.05, have a fold change (FC) greater than or equal to three, and a count minimum of at least ten normalized counts in three out of four replicates for a single condition (strain and temperature combination). Significantly DE genes are highlighted in yellow. The first sheet consists of all DE analyses sorted by adjusted p-value, and the second sheet consists of all DE genes sorted by total expression. (XLSX) S4 Table. Differential expression of annotated genes for the JB197 contrast (JB197 29˚C vs. JB197 37˚C). Genes that were considered significantly differentially expressed were required to have an adjusted p-value < 0.05, have a fold change (FC) greater than or equal to three, and a count minimum of at least ten normalized counts in three out of four replicates for a single condition (strain and temperature combination). Significantly DE genes are highlighted in yellow. The first sheet consists of all DE analyses sorted by adjusted p-value, and the second sheet consists of all DE genes sorted by total expression. (XLSX) S5 Table. Differential expression of annotated genes for the HB203 contrast (HB203 29˚C vs. HB203 37˚C). Genes that were considered significantly differentially expressed were required to have an adjusted p-value < 0.05, have a fold change (FC) greater than or equal to three, and a count minimum of at least ten normalized counts in three out of four replicates for a single condition (strain and temperature combination). Significantly DE genes are highlighted in yellow. The first sheet consists of all DE analyses sorted by adjusted p-value, and the second sheet consists of all DE genes sorted by total expression.