Homologs of the bat genes are present in each family of the Spirochaetes (Additional file 1: Figure S1), although not in all species. In contrast to the 5 genes present in B. fragilis, L. biflexa contains 3 bat genes and the pathogenic leptospires contain 4 2 7 8 9 . However, the batB and batC genes are fused in L. biflexa, which does not appear to be the case for the pathogenic species, and explains the discrepancy in gene number. Fusions of bat coding regions also appear to have occurred in Borrelia burgdorferi and Spirochaeta thermophila (Additional file 1: Figure S1) and were also reported for F. tularensis type A strain Schu S4 5 .

Analysis of BatA and BatB sequences identified motifs predicted to mediate protein-protein interactions, (Figure 1). The presence of these motifs suggest that Bat proteins may interact with each other to form a large protein complex. All three proteins are predicted to contain multiple trans-membrane helices, also predicted for the B. fragilis homologs, and BatD possesses a predicted signal sequence for export, suggesting that these proteins may associate with either the inner or outer membrane of L. biflexa.

The L. biflexa bat genes are located within a contiguous stretch of 11 genes on chromosome II that are transcriptionally oriented in the same direction (Figure 2A). Two different mutations were engineered using allelic replacement with the kanamycin-resistance cassette to delete either batA alone or batABD together; flanking genes were left intact. Three mutant clones from each transformation were shown to have lost the corresponding bat loci by Southern blot analysis of genomic DNA (Figure 2B). PCR analysis also confirmed the presence of the antibiotic-resistance gene (kan) and flanking genes, but bat loci were absent, as expected (data not shown). A single transformant of each type was randomly chosen for further characterization.

Transcript levels of bat genes and other ORFs were assessed by qRT-PCR with RNA from wild-type (WT), ΔbatA and ΔbatABD strains cultured in vitro (Figure 3). Transcript from the bat genes is present in the WT strain but undetectable in the ΔbatABD mutant, as expected. In the ΔbatA mutant strain, only the batA transcript is undetectable, but transcripts from the downstream ORFs, including batB and batD, were detected. Although the arrangement of the 11 genes suggest they may be co-transcribed in an operon, the deletion of the bat genes does not eliminate transcript from the downstream ORFs and we hypothesize that each gene has an independent promoter. Interestingly, even ORFs immediately downstream of the deleted genes had observable levels of transcript, even though their promoter regions were most likely located in the deleted sequences. However, the levels of transcript from the downstream genes were significantly lower in the mutant strains compared to transcript levels in the WT: htpG transcript levels were 3.7-fold lower in the ΔbatABD strain, and batB transcript levels were >12-fold lower in the ΔbatA mutant.

The signal sequence of BatD suggests a periplasmic or membrane-associated location for at least one member of this protein family. We therefore examined whether the absence of Bat proteins affected cellular shape or structure. L. biflexa morphology was assessed by scanning and transmission electron microscopy, including negative stains and freeze-substitution fixation to retain a more native state of the cells. As shown in representative images in Figure 4A, no morphological or ultrastructural differences were observed between the WT and mutant strains by any of these analyses.

Growth rates of WT, ΔbatA, and ΔbatABD strains were compared during in vitro cultivation in EMJH liquid medium and also for colony formation on solid EMJH medium. No significant differences in growth rate were observed when cultured in liquid medium, regardless of whether the cultures were aerated or static (Figure 4B). Colony morphology and rate of formation were similar among all strains (data not shown).

As the mutant strains did not display an obvious growth defect compared to WT, we assessed the growth dynamics of both parent and mutant when cultured together in the same medium (Figure 4C). WT and Δbat-ABD strains were co-inoculated into the same cultures (performed in triplicate) and assessed daily to determine if population ratios changed over time. As shown in Figure 4C, relative proportions of each strain did not change significantly over time and this was statistically confirmed by two-way Analysis of Variance (ANOVA) with the Bonferroni post-test. Therefore, the Bat proteins do not significantly affect L. biflexa growth, either in pure culture or when the mutant is mixed with an equal density of WT cells.

Previous researchers speculated that Bat proteins might provide a mechanism for coping with oxidative stress 2 4 14 . Therefore, we compared the resistance of WT and ΔbatABD strains to various concentrations of hydrogen peroxide and a more stable organic peroxide (tert-Butyl hydroperoxide), and to superoxide. We utilized the Δbat-ABD mutant in this comparison as we hypothesized that it would have a similar or greater phenotype than the single gene deletion in the ΔbatA strain. Both the WT and the ΔbatABD strain exhibited comparable levels of susceptibi-lity to all ROS tested, with greater than 90% killing when exposed to 10 μM concentrations of H₂O₂, but resistant to 1 μM (Figure 5A). Similarly, when L. biflexa strains were exposed to paraquat, a redox-cycling compound that generates superoxide, WT and mutant strains displayed similar susceptibility to paraquat concentrations (Figure 5B).

Bacteria such as E. coli and Salmonella typhimurium exhibit an inducible response to oxidative agents 15 16 . When activated by exposure to sublethal levels of oxidizing agents, this stress response allows some bacteria to induce enzymes that allow the cell to survive otherwise lethal levels of oxidants. As the Bat proteins did not aid in resistance to oxidative stress, we next tested whether their function may relate to sensing oxidizing agents and inducing a specific stress response in Leptospira. Mid-to-late log phase cultures were incubated in sublethal concentrations of either H₂O₂ (1 μM) or paraquat (0.5 μM) to potentially induce an oxidative stress response. Cultures were then subjected to various concentrations of ROS that included normally lethal levels, further incubated, and viable bacteria enumerated (Figure 6). Surprisingly, both pretreated and untreated cells were sensitive to similar concentrations of ROS, indicating that L. biflexa does not exhibit an inducible response to either H₂O₂ or superoxide. The ΔbatABD mutant strain was likewise treated but did not show any differences from the WT with either pretreatment (data not shown).

Differential in-gel electrophoresis (DIGE) of WT versus ΔbatABD mutant protein samples was used to identify changes in protein levels between strains (Figure 7). Protein samples were separated into membrane-associated and soluble fractions. No differences were observed in the soluble fractions (data not shown) but in the membrane-associated comparison, a single protein was observed to vary between samples (Figure 7, white oval). The region encompassing the protein was excised from the gel, trypsin-digested and identified by mass spectrometry as HtpG (40 peptides detected, 61% coverage), which is encoded by the gene immediately downstream of batD (Figure 2A). The HtpG protein appeared as several closely migrating spots, with the main mass of protein indicated in Figure 7. Protein levels were higher in the WT compared to the ΔbatABD strain, and differences for each spot ranged from 2.7-fold for the minor spot to greater than 4-fold higher for the main protein spot. This difference in HtpG protein levels approximately corresponds to the difference observed in transcript levels by qRT-PCR between WT and ΔbatABD strains (Figure 3). The Bat proteins were not identified by this approach. Bat protein levels may be relatively low and the fold change between mutant and WT may not be significant enough to be detected by the conditions tested here. For example, transcript levels of htpG in the WT strain are more than 10-fold higher than any of the bat transcripts (Figure 3).

L. biflexa serovar Patoc strain Patoc I (kindly provided by Dr. Dave Haake and Dr. Marija Pinne) was cultured at 30°C with shaking at 150 RPM in EMJH medium (Fisher Scientific, Pittsburgh, PA) 41 42 . Plating medium included 1.2% wt/vol Noble agar (final concentration) (Fisher Scientific) and plates were incubated at 30°C, inverted and sealed with parafilm. Kanamycin, when needed, was added to the medium at a final concentration of 20 μg/mL. Escherichia coli strains TOP10 (Invitrogen, Carlsbad, CA) or NEB5α (New England Biolabs, Ipswich, MA) were used for all plasmid manipulations.

Transformation of L. biflexa followed the protocol of Louvel and Picardeau 43 . L. biflexa deletion mutants were constructed by allelic exchange with the kanamycin-resistance marker driven by the borrelial flgB promoter 13 . Proof-reading polymerases Vent (New England BioLabs) or the Expand Long Template PCR System (Roche Applied Science, Indianapolis, IN) were used for fragment amplification according to the manufacturer’s recommendations and the fidelity of amplification was confirmed by double-stranded sequencing. Primers used for plasmid construction are shown in Table 1. The region encompassing the batABD locus and surrounding sequences was PCR-amplified using primers Lb.htpG.F and Lb.II0014.RC, yielding a 6,113 bp fragment that was then cloned into pCR-XL-TOPO (Invitrogen). Inverse PCR was used to delete the batABD genes using primers batKO.F.NheI and batAKO.RC.NheI, which incorporated NheI restriction enzyme sites for self-ligation of the resulting product. NheI restriction enzyme sites were also incorporated onto the kanamycin–resistance cassette by PCR amplification using primers Pflg.NheI.F and Tkan.NheI.RC. Both the pTopoXL::ΔbatABD and the flgB _P -kan cassette were digested with NheI and ligated together to create the allelic exchange vector pΔABD1-kn. A similar strategy was used to create the allelic exchange construct for batA (pΔbatA-kn) using primers batB.seq1.F and Lb.II0013/14.PCR1.RC to amplify a 2,565 bp fragment containing batA. Inverse PCR with primers batAKO.F.NheI and batAKO.RC.NheI were used to delete the coding region of batA and engineer the restriction enzyme sites needed to insert the kanamycin-resistance cassette. The deletions of the respective bat genes in the mutant strains of L. biflexa were confirmed by Southern blot analysis of total genomic DNA digested with the restriction enzymes NdeI and PstI, as previously described 44 45 . Primers used for probe amplification are listed in Table 1.

^a Restriction enzyme sequences designated in lower case letters.

^b TaqMan probes were labeled at the 5^′-end with FAM (6-carboxyfluorescein) and at the 3^′-end with TAMRA (6-carboxytetramethylrhodamine).

Total RNA was isolated from 10 mL cultures of exponentially growing L. biflexa cells using TRIzol reagent (Invitrogen). Cells were pelleted at 7,000 RPMs in 15 mL Falcon 2059 tubes and the pellet resuspended in 5.0 mL TRIzol. After incubation at room temperature for 2.5 min with vigorous shaking, 1 mL of chloroform was added, mixed and incubated for a further 2.5 min. The suspension was centrifuged again and the aqueous phase removed to a new Falcon tube and the RNA precipitated by addition of 5 mL isopropanol. Following a 10 minute incubation (room temperature), RNA was pelleted, washed in 75% ethanol and dissolved in 100 μL of RNase-free water. DNA was removed by treating with Turbo DNase (Ambion, INC. Austin, TX) following the manufacturer’s recommendations.

RNA was converted to cDNA using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA); reaction mixtures consisted of 1 μg RNA and were converted to cDNA per the manufacturer’s recommendations. The cDNA samples were diluted 1:20 with water and 2 μL used for subsequent quantitative PCR (qPCR) reactions. All samples were analyzed in triplicate. TaqMan Universal PCR Master Mix kit (Applied Biosystems) and PCR conditions were as previously described 46 . L. biflexa genomic DNA ranging from 10⁶ cells to 10 cells (in 10-fold serial dilutions) was used to generate a standard curve using the Ct values from the flaB primer/probe set. This standard curve was then used to interpolate the number of transcript copies from the Ct values generated from gene-specific primer/probe sets. The resulting transcript levels were then normalized to 10⁴ copies of flaB transcript. Negative controls lacking reverse transcriptase were included to demonstrate that all genomic DNA had been degraded and did not contribute to the signal.

Bacterial suspensions from cultures grown in EMJH media were prepared for scanning electron microscopy (SEM) essentially as described previously 47 . Samples were lightly sputtered with iridium and examined on a model SU-8000 scanning electron microscope operated at 2 kV (Hitachi High Technologies America, Pleasanton, CA). Images were digitized using the on-board frame card according to the manufacturer’s specifications.

For transmission electron microscopy (TEM), bacteria were prepared as described previously for imaging by microwave-assisted processing 48 . Grids were examined using a model H-7500 transmission electron microscope, operated at 80 kV (Hitachi). Digital images were captured and recorded using a model HR100 camera system (Advanced Microscopy Techniques, Danvers, MA).

Growth rate comparisons were performed in quadruplicate. Five mL cultures were inoculated at 10⁵ cells/mL from a starter culture grown to between 5 × 10⁸ to 1 × 10⁹ cells/mL, as determined by counting with Petroff-Hauser counting chambers. All cultures were incubated at 30°C; aerated cultures were shaken at 150 RPM. Cell densities were measured by optical density at 420 nm in a spectrophotometer.

Co-growth comparisons of wild-type and mutant strains were similarly tested with each strain inoculated at 10⁵ cells/mL in the same culture (for a combined concentration of 2 × 10⁵ cells/mL). Aliquots were removed daily from triplicate cultures, counted and diluted appropriately for colony formation on non-selective EMJH agar plates. PCR was performed on 24–30 colonies per plate to enumerate wild-type and mutant cells by amplifying a fragment of batB (wild-type) and the kanamycin-selectable marker (mutant).

Oxidative stress assays were also performed similarly. Peroxide-treated cultures were first diluted to 10³ cells/mL and peroxides were then added at specified concentrations and incubated for approximately 2 ½ hours, after which 100 μL samples were removed from each culture and spread on EMJH agar plates. After 4–6 days of incubation at 30°C, plates were removed and colony counts used to calculate viable cells. A similar strategy was followed for assessing whether an oxidative stress response could be induced in L. biflexa; quadruplicate cultures of 10³ cells/mL were exposed to a sublethal level of H₂O₂ (1μM) for 3 hrs with aeration, followed by the addition of specified concentrations of peroxide and a further incubation for 3 hours. Aliquots of 100 μL were removed and spread on EMJH agar plates to determine viable cell counts. Superoxide sensitivity was determined by diluting triplicate cultures to 5 × 10⁶ cells/mL and exposing to various concentrations of the superoxide-generating molecule paraquat (Sigma-Aldrich, St. Louis, MO) with incubation for 24 hrs. Cell viability was determined by counting motile cells using a Petroff-Hauser chamber with darkfield microscopy. To determine if L. biflexa produces an oxidative stress response to superoxide, triplicate cultures of 5 × 10⁶ cells/mL were pre-exposed to 0.5 μM paraquat for 2.5 hrs followed by addition of specific concentrations of paraquat. Cultures were further incubated for 24 hrs and cell viability assessed as described above.

L. biflexa isolates were grown to a cell density of ~1 × 10⁹ cells/ml and harvested by centrifugation (10,000 × g, 10 min, 23°C). Cell pellets were rinsed in PBS and lysed in PBS supplemented with 1 X Complete Protease Inhibitor (Roche Applied Science) by 3 passes through a French pressure cell (16,000 lb/in²). Cell lysates were further fractionated into soluble and membrane-associated proteins by ultracentrifugation (100,000 × g 1 h, 4°C). The membrane-associated protein pellet was rinsed with PBS and suspended in PBS+PI with the aid of a glass tissue homogenizer (Kontes Glass Co.,Vineland, NJ). Protein concentrations were determined by a modified Lowry protein assay with bovine serum albumin as a standard.

For DIGE analysis of membrane-associated proteins, 50 ug of L. biflexa wild-type or the ΔbatABD isolate was labeled with either 400 pmol Cy3 or Cy5 (CyDye minimal dye labeling kit, GE Healthcare) for 30 min on ice. As an internal control, a mixture of 25ug of the wild-type and 25 ug of the ΔbatABD samples were labeled with Cy2 for 30 min on ice. All labeling reactions were performed in DIGE labeling solutions consisting of 7 M Urea, 2M Thiourea, and 4% CHAPS in 10 mM Tris (pH 8.5). The labeling reaction was quenched by adding 1 ul of 10 mM lysine and incubating for 10 min on ice. To ensure that observed differences were not due to artifacts from preferential dye binding to proteins, several coupled samples were labeled by dye switching. Labeled proteins were stored at −20°C in the dark until isoelectric focusing.

Cy-dye labeled samples for comparison were mixed and DTT and IPGphore 3–10 buffer were added at final concentrations of 100 mM and 1.0%, respectively. The volume of each set was brought to 350 ul with isoelectric focusing solution C₄TT 49 and applied to 18 cm 3–10 non-linear IPG strips (GE Healthcare). Strips were focused using the following parameters: 12 hr rehydration, 500 V for 1 hr, 1000 V for 1 hr, 1500 V for 1 hr, 4000 V for 1 hr, and 8000V for 60,000 Vhr. Once focusing was complete, strips were stored at −80°C until equilibrated and separated in the second dimension by standard SDS-PAGE using 8-16% gradient gels (Jule, Inc., Milford, CT) 49 .

After separation of DIGE-labeled strips by SDS-PAGE, gels were scanned in the glass plates using a three laser Typhoon 9400 variable mode imager (GE Healthcare, Piscataway, NJ) at 200 microns. Differences in protein spots were quantified using DeCyder 2-D Differential Analysis Software v7.2. Protein spots of interest were excised and processed for mass spectrometry as previously described 49 . Dried peptides were sent to the Protein Chemistry section of the NIAID Research Technologies Branch, NIH for identification as described below.

The recovered peptides were re-suspended in 5 ul of Solvent A (0.1% formic acid, 2% acetonitrile, and 97.9% water). Prior to mass spectrometry analysis, the re-suspended peptides were chromatographed directly on column, without trap clean-up. The bound peptides were separated at 500 nl/min generating 80–120 Bar pressure, using an AQ C18 reverse phase media (3 u particle size and 200 u pore) packed in a pulled tip, nano-chromatography column (0.100 mm ID × 150 mm L) from Precision Capillary Columns, San Clemente, CA. The chromatography was performed in-line with an LTQ-Velos Orbitrap mass spectrometer (ThermoFisher Scientific, West Palm Beach, FL) and the mobile phase consisted of a linear gradient prepared from solvent A and solvent B (0.1% formic acid, 2% water, and 97.9% acetonitrile) at room temperature. Nano LC-MS (LC-MS/MS) was performed with a ProXeon Easy-nLC II multi-dimensional liquid chromatograph and temperature controlled Ion Max Nanospray source (ThermoFisher Scientific) in-line with the LTQ-Velos Orbitrap mass spectrometer.

Mass calibration was performed as needed with the positive ion Cal Mix prepared as described by Thermo-Scientific and monitored by routine analysis of a 10 femtomole stock sample of BSA digest. Typical acceptable results for this analysis would yield a 2800 – 3300 Mascot score, 75 – 85% coverage and 0 - +/−4 ppm error when submitted to the Mascot server using Proteome Discoverer 1.3 using the Swiss Prot-Trembl data base.

Computer controlled data dependent automated switching to MS/MS by Xcalibur 2.1 software was used for data acquisition and provided the peptide sequence information. Data processing and databank searching were performed with PD 1.3 and Mascot software (Matrix Science, Beachwood, OH).