Development of capture-based enrichment RNA-Seq to delineate RpoS-regulated genes in engorged nymphal ticks

Using comparative microarray and RNA-Seq, we previously defined the Bb RpoS regulon following temperature-shift in vitro and cultivation within DMCs (11, 15). Collectively, these studies demonstrated that mammalian host signals modulate promoter recognition by RNAP-RpoS and license RpoS-mediated repression of tick-phase genes. Notably, these studies identified a cohort of genes upregulated by RNAP-RpoS only in mammals. Given that RpoS is essential for transmission (13), we reasoned that the RpoS regulon includes genes upregulated exclusively during the nymphal blood meal. In a pilot RNA-Seq study using ribodepleted RNA from engorged nymphs infected with WT strain B31, only approximately 6,700 reads mapped to protein coding genes (0.034% of approximately 20 million total raw reads) (Supplemental Table 1; supplemental material available online with this article; https://doi.org/10.1172/JCI166710DS1), a value too low to obtain comprehensive transcriptomic data. To overcome this bottleneck, we took advantage of an enrichment strategy, designated TBDCapSeq, developed by Tokarz and colleagues (26, 27), which uses hybridization probes to ‘capture’ pathogen-specific amplicons prior to sequencing (Figure 1). Using TBDCapSeq, we compared the transcriptomes of WT and ΔrpoS Bb in fed nymphs and DMCs. Summaries of the raw and mapped data are presented in Supplemental Table 1.

Workflow for TBDCapSeq. Total RNA extracted from fed nymphs infected with either WT (green) or ΔrpoS (magenta) Bb was converted to cDNA and used as input for second-strand synthesis. Libraries were prepared using dual-indexes (blue and red). Following precapture amplification, libraries were hybridized to Bb-specific biotinylated probes. Bb-specific amplicon–probe duplexes were captured using magnetic streptavidin beads (lilac), amplified using Illumina universal primers, and sequenced on a NextSeq2000. Raw reads were mapped using EDGE-pro and analyzed for differential gene expression using DESeq2. TBDCapSeq for DMC-cultivated samples was performed using the same pipeline.

Overview of TBDCapSeq analyses

Approximately 11.3 and 15.6 million raw reads were obtained from fed nymphs infected with WT and ΔrpoS strains, respectively. Of these, approximately 30% were Bb-specific, representing an approximately 1,000-fold enrichment over conventional RNA-Seq. After post-run processing, approximately 1.6 and 1.9 million reads for protein coding genes in WT and ΔrpoS, respectively, remained. Of the 1,227 protein coding genes used for mapping, roughly 1,000 were detected at more than 10 transcripts per kilobase million (TPMs) in all 3 biological replicates (Supplemental Table 2). We obtained even more robust data for DMC-cultivated spirochetes. Of the approximately 44 million total reads obtained for WT and approximately 35 million total reads obtained for ΔrpoS DMC samples, roughly 21 and 17 million were Bb-specific, with 79 and 73% mapping to protein coding genes, respectively. Approximately 1,200 genes were detected at at least 10 TPMs in all 4 biological replicates (Supplemental Table 2). Prior microarray analyses demonstrated extensive transcriptomic remodeling as spirochetes transit between ticks and mammals (28). Along these lines, hierarchical clustering and Principal Component Analysis (PCA) plots (Figure 2) showed wide separation of WT transcriptomes in fed nymphs and mammals. The distance between WT and ΔrpoS suggests that RpoS is a major contributor to this transcriptional divergence. Indeed, DESeq2 identified 213 genes differentially regulated by RpoS in fed nymphs and/or DMCs. Of the 170 RpoS-regulated genes identified in DMCs, all but 3 (bb0228, bb0454, and bbb29/malX-2) were restored to near-WT levels by trans-complementation with rpoS expressed under its native promoter (Supplemental Table 3). To ascertain the extent of bias introduced by enrichment, we compared the RpoS DMC regulons obtained by TBDCapSeq and conventional RNA-Seq (11). Of the 98 RpoS-regulated genes identified in DMCs by conventional RNA-Seq, 89 — 55 upregulated and 34 repressed — were similarly regulated by TBDCapSeq (Supplemental Table 3). The high degree of overlap between these independent data sets minimized concerns that enrichment faithfully represents the spirochete transcriptome in a given milieu.

The contour of the Bb transcriptome varies substantially across the feeding nymphal tick and mammalian host phases of the enzootic cycle. Hierarchical clustering (A) and PCA plots (B) for WT and ΔrpoS Bb in fed nymphs (3 biological replicates per strain) and following cultivation in DMCs (4 biological replicates per strain) were generated using R Studio.

The RpoS regulon changes dramatically when LD spirochetes transits from ticks to mammals

Genome-wide comparisons of WT and ΔrpoS Bb in fed nymphs and DMCs revealed that the RpoS regulon varies substantially across the enzootic cycle (7, 13). Of note, all key components of the RpoN/RpoS pathway (bb0647/bosR, bb0763/rrp2, bb0450/rpoN and bb0771/rpoS) were expressed at comparable levels in fed nymphs and DMCs (Supplemental Table 4), arguing against fluctuations in RpoS protein levels being responsible for these differences. 4 categories of differentially expressed genes were identified: (a) core genes upregulated by RpoS in both nymphs and DMCs; (b) genes upregulated by RpoS only in nymphs; (c) genes upregulated by RpoS only in DMCs; and (d) genes repressed by RpoS in mammals. Notably, no genes were repressed by RpoS during tick feeding.

Genes upregulated by RpoS in fed nymphs and DMCs. In both fed nymphs and DMCs, 52 genes were upregulated by RpoS (hereafter designated core genes) (Supplemental Table 5). Eleven, including the RpoS-upregulated prototypes bbb19/ospC and bba24/dbpA, are known to be transcribed exclusively by RpoS (i.e., absolutely RpoS-dependent) in vitro and/or in DMCs (11, 13, 15, 29). Based on a comparison of TPM values for WT and ΔrpoS samples (Supplemental Table 2), 27 additional core genes also are considered absolutely RpoS-dependent. Twenty-two of the 38 absolutely RpoS-dependent core genes, most notably ospC, dbpA and bbi42, were transcribed at comparable levels in fed nymphs and DMCs. Thirteen, including 3 Pfam54_60 paralogs (bba65, bba66 and bba73), the OspF paralog bbo39 (erpL), and 2 Mlps (bbp28/mlpA and bbm28/mlpF), were transcribed at higher levels in fed nymphs, while 18 were higher in DMCs. The DMC-enhanced group included vlsE1, the expression site for the Vls system for antigenic variation (30), bba34/oppA5, encoding an oligopeptide substrate binding protein (11, 31), and bbk32, encoding a vascular endothelial adhesin and inhibitor of the classical complement pathway (32–34). Seven core genes, including 5 related to chemotaxis (bb0680/mcp4, bb0681/mcp5, bb0671/cheX, bb0567/cheA-1, and bb0565/cheW-2), were transcribed at appreciable levels by Δ rpoS Bb, indicating dual transcription by RpoS and RpoD. An additional 6 core genes (bb0400, bb0798, bbi42, bbj27, bbk53, and bbq03), all encoding hypothetical proteins, were dually transcribed by RpoS and RpoD only in DMCs.

Genes upregulated by RpoS only during tick transmission. (Supplemental Table 6). Forty-four genes were designated tick-only genes because they were upregulated by RpoS only in feeding nymphs and not in DMCs. Of the 44, 40 were transcribed exclusively by RpoS in fed nymphs, while the remaining 4 (bb0418/dipA, bb0637/nhaC1, bb0729/gltP, and bbh09) were dually transcribed by RpoS and RpoD with a significant contribution to their expression from the former σ factor. In contrast, in DMCs, all 44 were either transcribed exclusively by RpoD or dually transcribed, but the contribution of RpoS to their expression was not statistically significant. Thus, the σ factor selectivity for genes in this group differs between ticks and mammals, with significant upregulation by RpoS occurring only in fed nymphs (≥ 3-fold difference with q ≤ 0.05). Nine tick-only genes, including the Pfam54_60 paralogs bba64 and bbe31, are required for transmission (35–38).

Genes upregulated by RpoS only within mammals. (Supplemental Table 7). Forty genes were upregulated by RpoS only within DMCs. Unlike the tick-only genes, which were transcribed to varying extents in ticks and mammals, the vast majority of DMC-only RpoS-upregulated genes were expressed exclusively in mammals (Supplemental Table 2). Two-thirds, 67%, of the DMC-only genes appeared to be absolutely RpoS-dependent, including 17 encoded on lp28-2; the contribution of this linear plasmid to virulence has not been established (11, 39). The remaining 13 DMC-only genes, including 5 related to motility and chemotaxis (bb0273/fliR, bb0578/mcp-1, bb0669/cheA-2, and bb0670/cheW-3), were dually transcribed by RpoS and RpoD in mammals.

Genes repressed by RpoS within mammals. (Supplemental Table 8). Seventy-seven RpoS-regulated genes were expressed at significantly lower levels in WT versus ΔrpoS in DMCs and, hence, are repressed by RpoS “(≥ 3-fold difference with q ≤ 0.05). RpoS-repressed genes fell into 2 groups. The first consisted of genes that were expressed at comparable levels by WT and ΔrpoS Bb in fed nymphs but were strongly repressed by RpoS in DMCs. Twenty of these tick-phase genes, including bba15/ospA, bba16/ospB, bba62/lp6.6, bba68/BbCRASP1, and the glp operon (bb0240-0243), were shown previously to be repressed by RpoS in mammals (11, 15, 25). TBDCapSeq also identified an additional 10 RpoS-repressed genes in this group, including bb0330/oppA3, encoding an oligopeptide substrate binding protein (31), and bba69, encoding a Pfam54_60 lipoprotein (35). The remaining 47 RpoS-repressed genes were transcribed by WT Bb at comparably low levels in feeding nymphs and DMCs, but showed increased expression in the absence of RpoS in DMCs. This second category of RpoS-repressed genes included 3 closely related Pfam54_60 paralogs (bbi36, bbi38, and bbi39) (35) and bbd18, encoding a known regulator of RpoS protein levels (40, 41). Given its importance to the RpoS pathway, we confirmed the expression profile for bbd18 by qRT-PCR. bbd18 was transcribed at virtually identical low levels in fed nymphs and DMCs but was upregulated 10-fold in DMC-cultivated ΔrpoS Bb (Supplemental Figure 1). Presumably, RpoS-mediated repression of bbd18 in mammals ensured that levels of this regulatory protein remain low in mammals, when RpoS is essential.

Genes differentially regulated in feeding nymphs and/or mammals independent of RpoS

A dividend of TBDCapSeq is that it enables assessment of the RpoS-independent as well as the RpoS-dependent components of the Bb transcriptome in ticks and mammals (Supplemental Table 9). Examination of hierarchical clustering and PCA plots for ΔrpoS in fed nymphs and DMCs (Figure 2) suggested that RpoS-independent, differentially expressed genes comprise a substantial component of the WT transcriptomes in these 2 milieus. After excluding RpoS-regulated genes, 250 genes differed by more than 3-fold (q ≤ 0.05) between feeding nymphs and DMCs (Supplemental Table 9). Seventy-five genes were expressed at higher levels in fed nymphs, while 175 were higher in DMCs. Most of the RpoS-independent genes upregulated in feeding nymphs encode proteins with housekeeping functions — i.e., DNA replication, cell division, and protein translation and turnover — or functions related to nutrient acquisition and intermediary metabolism. Utilization of alternate carbon sources is critical to spirochete fitness in ticks (42, 43). Five genes (bb0166/malQ, bb0367, bb0557/ptsH-2, bb0559/crr, and bb0629/fruA-2) encode components of the phosphoenolpyruvate-dependent sugar phosphotransferase system — the spirochete’s central pathway for carbohydrate transport (42, 43) — and could be involved in uptake of alternative carbon sources. Eight are related to cell wall biosynthesis, including the chitobiose transporter bbb04-06/chbCAB, as well as bb0151/nagA, bb0201/murE, and bb0841/arcA. Increased expression of chb is particularly noteworthy given that chitobiose can be used for energy generation as well as cell wall biosynthesis (20, 44). Finally, 3 encode putative regulatory proteins — BB0355, a CarD-like transcriptional regulator required for transmission (45); BB0785/SpoVG, a tick-phase DNA/RNA-binding protein of undetermined function (46, 47); and BB0047/BpuR, a DNA/RNA-binding protein — were upregulated in feeding ticks (48) (Supplemental Table 4). Of the 75 RpoS-independent genes, 6 (bb0166/malQ, spoVG, bbb04-06/chbCAB, and bbb07) expressed at higher levels in fed nymphs are upregulated by c-di-GMP in vitro (10).

While a large majority (78%) of RpoS-independent genes upregulated in DMCs encode hypothetical proteins, 14 encode gene products related to DNA replication (bb0455, bb0552/ligA, and bb0632/recD), influx/efflux of small molecules (bb0642/potA and bb0641/potB), biosynthesis of metabolic cofactors (bb0782/nadD and bb0589/pta), purine salvage (bb0384/bmpC, bb0467, bb0524, and bbb23), and maintenance of the cell envelope (bb0304/murF, bb0586/femA, and bb0721/pgsA) (49–54). Also noteworthy, bb0733/plzA, which has a virulence-related function in mice unrelated to binding of c-di-GMP (23–25), was upregulated in DMCs compared with fed nymphs. With the exception of rrp1, which was slightly higher in mammals, all other known or putative regulatory factors (7) were expressed at comparable levels in both milieus (Supplemental Table 4).

Ligand-bound PlzA impairs RpoS-mediated repression and diminishes transcription of some RpoS-upregulated genes

Using a strain, cDGC, that constitutively synthesizes c-di-GMP in mammals, we previously demonstrated that ligand-bound PlzA acts as a ‘brake’ on RpoS-dependent gene regulation, antagonizing RpoS-mediated repression and diminishing expression of RpoS-upregulated genes (25). These data led us to propose that ligand-bound PlzA is a principal determinant of the RpoS regulon during transmission and, moreover, that transition to the mammalian host-phase RpoS regulon requires cessation of c-di-GMP synthesis. To garner support for this notion on a genome-wide scale, we performed TBDCapSeq on isogenic WT, cDGC, and cDGCΔplzA strains cultivated in DMCs (Supplemental Tables 5–8). As noted previously (25), transcripts for rpoS were unaffected by either increased c-di-GMP or loss of PlzA (Supplemental Table 4). In contrast, ligand-bound PlzA had a striking effect on the RpoS regulon. Of the 77 genes repressed by RpoS in DMCs, 57, including 26 of the 30 tick-phase genes noted above, were expressed at significantly higher levels in cDGC compared with WT (≥ 3-fold difference with q ≤ 0.05). In every case, deletion of plzA restored RpoS-mediated repression in the cDGCΔplzA strain. The modulatory effect of ligand-bound PlzA in mammals also extended to 17 RpoS-upregulated genes. Expression of 10 RpoS core genes, including ospC, dbpA, bbk32, and vlsE1, and 7 DMC-only genes decreased significantly in the cDGC strain compared with WT; all but 2 (bb0580 and bb0578/mcp-1) were absolutely RpoS-dependent (≥ 3-fold difference with q ≤ 0.05). In all but 1 case, deletion of PlzA in the cDGC strain restored RpoS-upregulation to WT levels; vlsE1, the sole outlier, was transcribed at lower levels in the cDGC strain in a PlzA-independent manner (Supplemental Table 5). Expression of vlsE1 also requires the trans-acting factor YebC (55). The negative effect of c-di-GMP on RpoS-upregulation of vlsE1 raises the possibility that YebC is c-di-GMP-regulated through some unknown mechanism. Of note, 4 tick-only RpoS-upregulated genes (bbh32, bbk01, erpA, and erpB) were transcribed at higher levels by cDGC in a PlzA-dependent manner. A question that arose from the above data was whether ligand-bound PlzA acts predominantly on genes within the RpoS regulon. As illustrated by the PCA plot and hierarchical clustering (Figure 3), synthesis of c-di-GMP in mammals appears to shift the transcriptome of cDGC toward that of ΔrpoS, while cDGCΔplzA clustered closely with WT (Figure 3A). Collectively, these data suggested that the modulatory effect of c-di-GMP on RNAP-RpoS was largely PlzA-dependent and that the influence of ligand-bound PlzA outside of the RpoS regulon was negligible.

Interplay between RpoS, BosR, and ligand-bound PlzA regulates differential gene expression in feeding nymphal ticks and mammals. PCA (A) and hierarchical clustering (B) for (i) DMC-cultivated isogenic WT (WT-BbP1781), ΔrpoS, and ΔbosRΔrpoS/irpoS with and without IPTG; (ii) DMC-cultivated isogenic WT (WT-BbP1473), cDGC, and cDGCΔplzA; and (iii) isogenic WT (WT-BbP1781) and ΔrpoS within fed nymphs.

Persistence of Bb infection in mice requires RpoS and involves RpoS-mediated repression of tick-phase genes

Using a ΔrpoS strain complemented in trans (11), we previously demonstrated that loss of the complementing plasmid placed spirochetes at a survival disadvantage for up to 20 weeks following needle inoculation, supporting a requirement for RpoS during persistent infection. These studies also suggested that RpoS-mediated repression is maintained throughout infection. To confirm the requirement for RpoS-upregulated genes and RpoS-mediated repression for persistence, we developed a ΔrpoS strain (ΔrpoS/irpoS) harboring an IPTG-inducible copy of the rpoS gene inserted into the highly stable endogenous cp26 plasmid (25). When cultivated in vitro, ΔrpoS/irpoS expressed RpoS and prototypical RpoS-upregulated gene products in an IPTG concentration–dependent manner (Supplemental Figure 2A). As previously reported (56), over-expression of rpoS (i.e., more than 50 μM IPTG) was toxic (Supplemental Figure 2B). To determine whether physiological levels of RpoS could be induced in ΔrpoS/irpoS within animals, we implanted DMCs containing ΔrpoS/irpoS into rats receiving IPTG in their drinking water. Oral administration of IPTG yielded levels of RpoS and RpoS-upregulated proteins and repression of OspA and GlpD at levels comparable to those of DMC-cultivated WT Bb (Figure 4A). By immunoblot, we also confirmed the RpoS-dependence of vlsE1 revealed by RNA-Seq (Figure 4A).

Ligand-bound PlzA and BosR modulate the RpoS regulon in a reciprocal manner within mammals. (A) Lysates from DMC-cultivated WT, ΔrpoS/irpoS, ΔbosRΔrpoS/irpoS, cDGC, and cDGCΔplzA were separated by SDS-PAGE and stained with silver or immunoblotted with antisera against FlaB, RpoS, OspA, GlpD, OspC, DbpA, BBK32, and VlsE. (B) Lysates from DMC-cultivated WT, ΔbosR/irpoS, and bosRcomp/irpoS were separated by SDS-PAGE and stained with silver. Molecular weight markers (kDa) are shown at the left of each gel. “+” and “–” indicate the presence or absence of IPTG, RpoS, BosR, and PlzA, and/or c-di-GMP synthesis by the constitutively active diguanylate cyclase in cDGC strains. A and B show representative images from 3 biological replicates per strain. Uncropped immunoblots for Figure 4A are provided in Supplemental Figure 5.

Having established that ΔrpoS/irpoS Bb host-adapts normally in rats given IPTG, we used this strain to assess the contribution of RpoS to persistence in mice. First, we confirmed the infectivity of ΔrpoS/irpoS by inoculating C3H/HeJ mice. As shown in Table 1, nearly all tissues from mice infected with either WT — which received untreated water — or ΔrpoS/irpoS — which received IPTG-treated water — were culture-positive 2-weeks after inoculation, while untreated mice infected with ΔrpoS/irpoS were culture-negative. Tilly and colleagues (57, 58) previously established that OspC is dispensable for infectivity by approximately 21 days after inoculation. To avoid an OspC-related phenotype in our persistence experiments, C3H/HeJ mice infected with ΔrpoS/irpoS were maintained on IPTG-treated water for at least 4 weeks after inoculation (Figure 5A). At the 4-week time point, IPTG was removed from half of the ΔrpoS/irpoS-infected mice, while the other half was maintained on IPTG-treated water. At 6 and 8 weeks after inoculation, WT- and ΔrpoS/irpoS-infected mice maintained on IPTG were culture positive from most tissues (Table 2). However, 2 weeks after stopping IPTG-treatment (6 weeks after inoculation), only 4 of 30 tissues from ΔrpoS/irpoS-infected mice were culture positive, with a single positive site per animal. All tissues from ΔrpoS/irpoS-infected mice were culture negative 4 weeks after discontinuation of IPTG treatment. Antibodies against OspC were detected in sera from all mice 8 weeks after inoculation (Figure 5B). Strikingly, ΔrpoS/irpoS-infected mice mounted strong anti-OspA responses after discontinuation of IPTG-treatment, whereas OspA antibodies were not detected in ΔrpoS/irpoS-infected mice continuing to receive IPTG (Figure 5B).

RpoS is required for persistence in mice. (A) Experimental design to assess the contribution of RpoS to persistence in C3H/HeJ and SCID mice (5 mice per condition, per time point). Mice infected with ΔrpoS/irpoS received IPTG-treated water (blue) 1 week before inoculation. Serology was performed 4 weeks after inoculation to confirm infection (Supplemental Figure 3). At 4 weeks, IPTG was withdrawn from half of the ΔrpoS/irpoS-infected mice, while the other half received IPTG for the remainder of the experiment. WT-infected mice received untreated water (white) throughout the experiment. At 6 and 8 weeks after inoculation (p.i.), mice were euthanized for collection of blood for serology and tissues for culture (Table 2). (B) Loss of RpoS was associated with production of antibodies against OspA. Sera from individual C3H/HeJ mice collected at 6 and 8 weeks after inoculation was assayed by immunoblot using 100 ng of recombinant OspA. Sera collected 8 weeks after infection was also assayed against 100 ng of recombinant OspC. Uncropped immunoblots for Figure 5B are provided in Supplemental Figure 6.

Complementation of ΔrpoS Bb with IPTG-inducible RpoS restores virulence in C3H/HeJ mice

RpoS is required for persistence in C3H/HeJ and SCID mice

To investigate whether antibodies were responsible for clearance of ΔrpoS/irpoS following withdrawal of IPTG, we repeated the above experiment using NOD.Cg-PrkdcSCID/J (SCID) mice. As with C3H/HeJ mice, SCID mice inoculated with ΔrpoS/irpoS and maintained on IPTG-treated water for the entire experiment, as well as mice infected with WT Bb, were culture positive 6 and 8 weeks after inoculation (Table 2). Two weeks after discontinuation of IPTG treatment, ΔrpoS/irpoS spirochetes were recovered from 9 of 30 tissue sites cultured. 4 weeks after removal of IPTG, only 3 of 30 sites from ΔrpoS/irpoS-infected mice were culture positive. Collectively, these data demonstrate that the requirement for RpoS extends beyond early infection and that RpoS-dependent factors functionally unrelated to adaptive immunity are also required to sustain infection.

BosR is essential for transcriptional as well as repressive functions of RpoS

In addition to serving as an activator for RpoN-dependent transcription of rpoS, BosR also has been proposed as a repressor for ospA and other tick-phase genes (59, 60). The latter studies, however, were conducted in vitro and failed to divorce the requirement of BosR for RpoN-dependent transcription of rpoS from its putative repressor function. We reasoned that our IPTG-inducible irpoS allele, which dissociates transcription of rpoS from the Rrp2/BosR/RpoN complex, could be used to clarify the contribution of BosR to RpoS-mediated repression. Accordingly, we inactivated bosR in ΔrpoS/irpoS, generating the strain ΔbosRΔrpoS/irpoS. During in vitro cultivation without IPTG, ΔbosRΔrpoS/irpoS expressed neither RpoS nor OspC, whereas both were expressed in a dose-dependent manner when IPTG was added to the culture medium (Supplemental Figure 2C). Surprisingly, deletion of bosR ameliorated RpoS toxicity at IPTG concentrations above 50 μM (Supplemental Figure 2D). Although ΔbosRΔrpoS/irpoS Bb cultivated in DMCs in IPTG-treated rats expressed WT levels of RpoS, we observed noticeably lower levels of OspC, DbpA, BBK32, and VlsE along with incomplete repression of OspA and GlpD; this protein profile was strikingly similar to that of DMC-cultivated cDGC (Figure 4A). Complementation of ΔbosRΔrpoS/irpoS was technically challenging due to the paucity of antibiotic-resistance markers available for selection in Bb. As an alternative, we generated a bosR/irpoS strain that retained the native rpoS gene. Like ΔbosRΔrpoS/irpoS, ΔbosR/irpoS grew normally in vitro in the presence of over 50 μM IPTG (Supplemental Figure 2E) and showed dysregulation of RNAP-RpoS function when cultivated in DMCs in IPTG-treated rats (Figure 4B). Complementation of bosR in the ΔbosR/irpoS background (bosRcomp) restored RpoS-mediated toxicity during in vitro cultivation with more than 50 μM IPTG (Supplemental Figure 2F) as well as RpoS-dependent facets of mammalian host-adaption in rats given IPTG (Figure 4B). Moreover, unlike ΔrpoS/irpoS, ΔbosRΔrpoS/irpoS was avirulent in C3H/HeJ and SCID mice treated with IPTG (Table 3), demonstrating that murine infectivity requires BosR as well as RpoS.

BosR works cooperatively with RpoS to promote virulence in mice by an RpoN-independent mechanism

We next performed RNA-Seq on DMC-cultivated ΔbosRΔrpoS/irpoS Bb with and without IPTG to determine the RpoN-independent contribution of BosR to shaping the RpoS regulon in mammals (Supplemental Table 3). Of the 92 RpoS-upregulated genes in DMCs — 52 core and 40 DMC-only — 53 required BosR for transcription, as they were not upregulated in the ΔbosRΔrpoS/irpoS strain under inducing conditions (Supplemental Table 5 and 7). Moreover, all but 2 of the remaining 39 RpoS-upregulated genes showed lower folds of regulation in the absence of BosR. For example, transcripts for ospC increased by only 14-fold following induction of RpoS in ΔbosRΔrpoS/irpoS compared with 984-fold in WT compared with ΔrpoS (Supplemental Tables 3 and 5). Indeed, the immunoblots for OspC, DbpA,and BBK32 revealed that these transcriptional differences appear to be biologically relevant at the protein level (Figure 4A). Most strikingly, 75 of 77 RpoS-repressed genes were not downregulated in ΔbosRΔrpoS/irpoS despite induction of RpoS (Supplemental Table 8). The above results indicated that RNAP-RpoS function in mammals is highly dependent on BosR. This conclusion was supported by PCA and hierarchical clustering analyses (Figure 3), which suggest similarity between the transcriptomes of ΔrpoS and ΔbosRΔrpoS/irpoS in IPTG-treated rats. In contrast, the effect of ligand-bound PlzA on RpoS-dependent gene regulation was selective, affecting only 55 of 75 BosR/RpoS-repressed genes (Supplemental Table 8) and 16 of 90 BosR/RpoS-upregulated genes in DMCs (Supplemental Tables 5 and 7).