Bordetella PlrSR regulatory system controls BvgAS activity and virulence in the lower respiratory tract

Significance

Bordetella spp. includes Bordetella pertussis, the causal agent of whooping cough. The Bordetella virulence gene (BvgAS) two-component regulatory system (TCS) is considered the “master virulence regulator” in Bordetella, as it controls expression of all known virulence factor-encoding genes. We show here that another TCS, PlrSR, is required for BvgAS activity in the lower respiratory tract (LRT) and for virulence even when BvgAS is rendered constitutively active, suggesting that it controls critical functions for bacterial survival in the LRT independently of BvgAS. Our data introduce a new layer of complexity to a paradigm of Bordetella virulence control that has held for more than 30 y, and they indicate the existence of previously unknown bacterial factors that may serve as vaccine components and therapeutic targets.

Abstract

Bacterial pathogens coordinate virulence using two-component regulatory systems (TCS). The Bordetella virulence gene (BvgAS) phosphorelay-type TCS controls expression of all known protein virulence factor-encoding genes and is considered the “master virulence regulator” in Bordetella pertussis, the causal agent of pertussis, and related organisms, including the broad host range pathogen Bordetella bronchiseptica. We recently discovered an additional sensor kinase, PlrS [for persistence in the lower respiratory tract (LRT) sensor], which is required for B. bronchiseptica persistence in the LRT. Here, we show that PlrS is required for BvgAS to become and remain fully active in mouse lungs but not the nasal cavity, demonstrating that PlrS coordinates virulence specifically in the LRT. PlrS is required for LRT persistence even when BvgAS is rendered constitutively active, suggesting the presence of BvgAS-independent, PlrS-dependent virulence factors that are critical for bacterial survival in the LRT. We show that PlrS is also required for persistence of the human pathogen B. pertussis in the murine LRT and we provide evidence that PlrS most likely functions via the putative cognate response regulator PlrR. These data support a model in which PlrS senses conditions present in the LRT and activates PlrR, which controls expression of genes required for the maintenance of BvgAS activity and for essential BvgAS-independent functions. In addition to providing a major advance in our understanding of virulence regulation in Bordetella, which has served as a paradigm for several decades, these results indicate the existence of previously unknown virulence factors that may serve as new vaccine components and therapeutic or diagnostic targets.

Caused by the human-specific, Gram-negative bacterium Bordetella pertussis, whooping cough (aka pertussis) is reemerging in the United States and other developed countries, despite high vaccine coverage (1, 2). Increased incidence in recent years coincides with the switch to acellular vaccines, which induce immunity that is less durable than that induced by whole-cell vaccines or by infection with B. pertussis (3–5). Although closely related to B. pertussis, Bordetella bronchiseptica infects nearly all mammals and typically causes more chronic, long-term respiratory infections (6). Despite these differences, B. pertussis and B. bronchiseptica produce a nearly identical set of virulence factors that includes adhesins, such as filamentous hemagglutinin (FHA) and fimbriae (FIM), and toxins, such as adenylate cyclase toxin (ACT), and a type III secretion system (T3SS) (2).

The Bordetella virulence gene (BvgAS) phosphorelay-type two-component regulatory system (TCS) is considered the “master virulence control system” in Bordetella. BvgAS differentially regulates (either directly or indirectly) hundreds of genes and at least three distinct phenotypic phases (7, 8). The Bvg⁺ phase occurs when the bacteria are grown at 37 °C in Stainer–Scholte (SS) broth or on Bordet–Gengou (BG) blood agar and correlates with BvgAS activity. The Bvg⁺ phase is characterized by expression of all currently known protein virulence factor-encoding genes (referred to collectively as vags) and lack of expression of BvgAS-repressed genes (called vrgs), which includes those encoding flagella in B. bronchiseptica. The Bvg^– phase occurs when the bacteria are grown at ≤26 °C or when millimolar concentrations of MgSO₄ or nicotinic acid are added to the growth medium (referred to as “modulating conditions”). The Bvg^– phase is characterized by expression of vrg loci and lack of expression of vags. The Bvg-intermediate (Bvgⁱ) phase occurs at intermediate temperatures or in the presence of low concentrations of MgSO₄ or nicotinic acid (9). It is characterized by expression of vags that contain high-affinity BvgA binding sites at their promoters (such as fhaB, encoding FHA, fimBCD, encoding the FIM biogenesis proteins, and bvgAS itself), lack of expression of vags with low-affinity BvgA binding sites (such as cyaABDE, encoding ACT and ptxA-E, encoding pertussis toxin), lack of expression of vrgs, and maximal expression of bipA, which encodes an outer-membrane protein of unknown function (9–11). Although BvgAS activity is altered by temperature, MgSO₄, and nicotinic acid in vitro, the true signals it senses in nature are unknown.

B. pertussis and B. bronchiseptica strains containing loss-of-function mutations in bvgAS are avirulent, whereas strains containing mutations that render BvgAS active—even under modulating conditions in vitro—are indistinguishable from WT bacteria in their ability to cause respiratory tract infections (12–14). Characterization of antibody responses following infection and analyses using recombination-based reporters of gene expression indicate that modulation to the Bvg^– phase does not occur during infection (12, 15, 16). Studies with strains that produce Bvg^– phase factors ectopically in the Bvg⁺ phase have demonstrated the importance of BvgAS-mediated repression of gene expression in vivo (14, 17). Taken together, these results have been interpreted to indicate that the Bvg⁺ phase is both necessary and sufficient for the development of respiratory infection by Bordetella. Moreover, these data, together with the fact that the phenotypic profile of WT bacteria grown at 37 °C in SS medium or on BG agar (Bvg⁺ phase conditions) is identical to that of mutants containing bvgS mutations that render BvgS insensitive to modulating conditions, have led to the conclusion that these in vitro growth conditions mimic, at least to some extent, those experienced by the bacteria in the respiratory tract.

In a survey of putative TCSs in B. bronchiseptica, we discovered a gene (BB0264) predicted to encode an NtrY-like sensor kinase that is essential for B. bronchiseptica to colonize the trachea of rats following low-dose, low-volume intranasal inoculation and to persist in the lungs of mice following high-dose, large-volume intranasal inoculation (18). We named BB0264 plrS, for persistence in the lower respiratory tract (LRT) sensor. Subsequently, Hester et al. reported the identification of a CO₂ responsive regulon in Bordetella, and hypothesized that CO₂ gradients may be sensed by the bacteria as a cue indicating their localization in the LRT (19). Here, we show that plrS is required for increased virulence factor production in response to CO₂, suggesting a functional link between PlrS and BvgAS. We investigated this link, showed that plrS is required for BvgAS to be fully active in the LRT, and showed that plrS is required for persistence in the LRT even when BvgAS is constitutively active, indicating that although BvgAS is necessary in vivo, it is not sufficient. Moreover, our data suggest the existence of genes that are expressed only in the LRT that encode previously unknown virulence factors, which may serve as new therapeutic or diagnostic targets and vaccine components.

Results

PlrS Is Required for Enhanced BvgAS-Dependent Virulence-Associated Phenotypes in Response to Elevated CO₂ Concentrations.

BvgAS-regulated virulence-associated phenotypes include adherence to epithelial cells and macrophages, which is mediated by FHA (20, 21), hemolysis on blood-containing agar, which is mediated by ACT (22), and toxicity to various eukaryotic cell types in culture, which is mediated, at least in part, by the T3SS (23). All of these phenotypes were enhanced when WT B. bronchiseptica was grown at 37 °C in 5% CO₂ compared with growth at 37 °C in ambient air (Fig. 1 and Fig. S1) (19). These virulence-associated phenotypes did not increase in response to 5% CO₂ in the ∆plrS mutant, indicating that PlrS is required for this effect. The effect was not a result of acidification of the growth medium because acidification alone, without increased CO₂, did not result in increased adherence (Fig. S2). These results suggest the possibility of a functional interaction between PlrS and BvgAS.

Fig. 1.

PlrS is required for Bvg-dependent phenotypes in 5% CO₂ growth conditions. (A) Bacterial adherence to L2 lung epithelial cells. WT and ΔplrS bacteria were cultured in ambient (37 °C) or 5% CO₂ (37 °C + 5% CO₂) conditions before inoculation at an MOI of 150 onto an L2 monolayer. The number of adherent bacteria per L2 cell was enumerated by determining the average number of bacteria and L2 cells visible from four visual fields of the microscope. Error bars represent SD of the mean. (B) Hemolysis on blood agar media produced from WT, ΔplrS, and RB515 (an ACT-null mutant) colonies under the indicated growth condition. (Magnification, 1,000×.) (C) Cytotoxicity of macrophage-like J774 cells induced by WT, ΔplrS, ΔbscN (containing an insertional disruption in the gene encoding the ATPase of the T3SS), and ΔbvgS (containing a deletion in bvgS) strains, following bacterial growth under the indicated condition. Percent cytotoxicity was calculated as the ratio of lysed J774 cells resulting from bacterial inoculation to fully lysed J774 cells, and error bars represent SD. Statistical significance for A (t test with Welch’s correction) and C (ANOVA) is indicated as *P < 0.05, **P < 0.01, and ****P < 0.0001.

Fig. S1.

Bacterial adherence to macrophage-like murine J774 cells. WT and ΔplrS strains were cultured under ambient (37 °C) or 5% CO₂ (37 °C + 5% CO₂) conditions and then inoculated at an MOI of 200 onto a monolayer of J774 cells. The number of adherent bacteria per J774 cell was enumerated by determining the average number of bacteria and J774 cells visible from four visual fields of the microscope, and error bars represent SD of the mean. Statistical significance (t test with Welch’s correction) is indicated as ***P < 0.001.

Fig. S2.

Increased adherence of WT B. bronchiseptica to L2 lung epithelial cells does not result from bacteria responding to lowered pH of the growth media. WT B. bronchiseptica was cultured in SS media in ambient (no CO₂) or 5% CO₂ (+CO₂) conditions, or in ambient conditions after the media had been adjusted to pH 6.8 (pH 6.8). The pH of SS media (7.6) will decrease to 6.8 when media without bacteria is incubated under 5% CO₂ conditions. Adherence to L2 cells was enumerated by determining the average number of bacteria and L2 cells visible from four visual fields of the microscope, and error bars represent SD of the mean. Statistical significance (t test with Welch’s correction) is indicated as ***P < 0.001.

PlrS Likely Affects BvgAS-Dependent Phenotypes via PlrR.

A putative response regulator, which we are naming PlrR, is encoded immediately 3′ to plrS. Multiple attempts using various approaches to disrupt or delete plrR (BB0265) were unsuccessful, suggesting plrR is essential for cell viability under the growth conditions tested. As an alternate approach to determine if PlrS and PlrR function as a cognate TCS, we delivered a plrR allele (plrR^D52E) encoding a PlrR protein in which the predicted site of phosphorylation, Asp52, was replaced with Glu, a predicted phosphomimetic (24), to the attTn7 site in WT, ∆plrS, and ∆bvgS (which harbors a deletion in bvgS) B. bronchiseptica. When grown in 5% CO₂, the level of adherence of the ∆plrS strain expressing plrR^D52E to L2 cells was as high as that of WT B. bronchiseptica (and WT-expressing plrR^D52E), demonstrating that the plrR^D52E allele could complement a ∆plrS mutation and, therefore, that PlrS likely affects BvgAS-dependent virulence-associated phenotypes via PlrR (Fig. 2). Increased adherence in the WT and ∆plrS strains expressing plrR^D52E in bacteria grown in ambient air is also consistent with the D52E substitution functioning as a phosphomimetic, rendering PlrR constitutively active (Fig. 2). Lack of adherence by the ∆bvgS strain with and without the plrR^D52E allele in the presence or absence of 5% CO₂ confirms that bvgS is required for adherence to host cells and that PlrR^D52E does not induce a BvgAS-independent adherence activity (Fig. 2). These data provide evidence (but do not prove) that PlrS and PlrR function as a canonical TCS, and that increased adherence in response to CO₂ is mediated by PlrS via PlrR. However, these data do not rule out the possibility that BvgA and PlrR interact synergistically to affect expression of some genes or that PlrS directly interacts with BvgS, while also affecting the phosphorylation state of PlrR.

Fig. 2.

Overexpression of a phosphomimetic plrR allele rescues the ΔplrS L2 cell adherence defect. L2 lung epithelial cell adherence displayed by WT, ΔplrS, WT-plrR^D52E (a strain isogenic to WT that produces PlrR with the phosphomimetic D52E substitution), ΔplrS-plrR^D52E (ΔplrS that produces PlrR with the phosphomimetic D52E substitution), ∆bvgS, and ΔbvgS-plrR^D52E (ΔbvgS that produces PlrR with the phosphomimetic D52E substitution) following bacterial growth under the indicated condition. Error bars represent SD of the mean. Statistical significance (t test with Welch’s correction) is indicated as ****P < 0.0001.

Positive autoregulation of the bvgAS operon is well-documented (25–28). Nonetheless, to determine if PlrS affects BvgAS activity by controlling bvgAS transcription, we introduced a P_bvgA-gfp transcriptional reporter into the attTn7 site of WT and ∆plrS B. bronchiseptica, and measured GFP activity in bacteria grown at 37 °C with and without 5% CO₂. Under all conditions tested the levels of GFP—and therefore the expression of bvgAS—was not significantly different between WT and the ∆plrS strains (Fig. S3). GFP levels in the ∆bvgS strain reflect activity from the bvgA P2 promoter, which is expressed at a low level under Bvg^– phase conditions (providing the cell with a low amount of BvgAS so that it can respond, via positive autoregulation, when Bvg⁺ phase conditions are encountered) (29). This low-level expression of P_bvgA was not affected by the ∆plrS mutation. Taken together, these data suggest that PlrS, via PlrR, exerts its effects on BvgAS activity posttranscriptionally.

Fig. S3.

PlrS does not transcriptionally regulate the expression of bvgAS in vitro. WT, ∆plrS, ∆bvgS, and ∆bvgS∆plrS with the transcriptional reporter PbvgAS-gfp (the bvgAS promoter driving expression of gfp) were cultured in SS media under Bvg⁺ phase growth conditions at 37 °C with or without 5% CO₂ for ∼18 h with agitation. For each strain, OD₆₀₀ and GFP (excitation A485, emission A535) were measured and GFP expression was normalized to OD₆₀₀. The experiment was performed twice in duplicate. GFP/OD₆₀₀ values were averaged across all replicates. No statistically significant differences in expression (t test with Welch’s correction) were observed between the WT and ∆plrS strains or between ∆bvgS and ∆bvgS∆plrS strains under either growth condition.

B. bronchiseptica Lacking plrS Modulate to the Bvg^– Phase Within the LRT.

Based on the functional link between PlrS and BvgAS (Fig. 1) and the similarly rapid clearance of B. bronchiseptica ∆plrS and ∆bvgS mutants from the LRT of mice (13, 18), we hypothesized that PlrS may influence BvgAS activity within the host. To test this hypothesis, we used two reporter systems developed in our laboratory. pGFLIP contains a recombinase-based reporter system, similar to recombinase-based in vivo technology (15). Expression of flp, encoding FLP recombinase, results in excision of gfp and nptII (encoding kanamycin resistance, Km^r) genes located between FLP recombinase target (FRT) sites and conversion of the bacteria from GFP⁺ and Km^r to GFP^– and Km^s. In pGFLIP-flaA, the Bvg^– phase-specific P_flaA promoter is located 5′ to flp and conversion to GFP^– (and Km^s) indicates that the bacteria have expressed the Bvg^– phase at some point during the experiment.

In contrast to pGFLIP, which reports on the history of the bacteria, the pBAM plasmid reports on the status of the bacteria at the time of plating (29). The pBAM plasmid integrates within the bvgAS promoter region and causes the P2 promoter to be expressed at a lower-than-normal level such that, when bacteria are growing under Bvg^– phase conditions, the amount of BvgAS in a small proportion of cells in the population (∼5%) is below the threshold required for positive autoregulation, “trapping” these cells in the Bvg^– phase (29). When WT B. bronchiseptica containing the pBAM plasmid are grown under Bvg^– phase conditions and then plated onto BG-blood agar and incubated at 37 °C (Bvg⁺ phase conditions), the trapped bacteria yield colonies that are larger, flatter, and less hemolytic than colonies formed by Bvg⁺ phase bacteria because these colonies contain ∼5% phenotypically Bvg^– phase bacteria (as the colony grows, ∼95% of the daughter cells produce enough BvgAS to convert to the Bvg+ phase, whereas ∼5% remain phenotypically Bvg^– phase). We previously referred to these colonies, which are easily scored by simple visual inspection, as large colony variants (29). However, because these colonies result from phenotypic bistability and not a genetic change, it is more appropriate to refer to them as large colony phenotypes (LCPs), which we will do henceforth. Although all LCPs result from a founder bacterium that was Bvg^– phase at the time of plating, ∼95% of bacteria that contain pBAM and are Bvg^– phase at the time of plating form typical Bvg⁺ phase-appearing (non-LCP) colonies (29). Hence, although the presence of LCPs indicates that bacteria have modulated to the Bvg^– phase, it vastly underestimates the number of bacteria that are Bvg^– phase at the time of plating.

When grown at 37 °C without addition of MgSO₄ to the medium (Bvg⁺ phase conditions), WT and ∆plrS strains containing both reporter systems maintained GFP fluorescence and no LCPs formed (Fig. S4), indicating that neither strain modulated to the Bvg^– phase. When switched from Bvg⁺ phase growth conditions to Bvg^– phase growth conditions, both strains lost GFP fluorescence and a small proportion of LCPs formed after 24 h (Fig. S4). Both reporters, therefore, can accurately report that the bacteria modulated to the Bvg^– phase and ∆plrS mutants modulate to a similar extent as WT bacteria when grown in vitro.

Fig. S4.

In vitro analysis of WT and ΔplrS modulation using the plasmid reporters pGFLIP-flaA and pBam. WT and ΔplrS containing pGFLIP-flaA and pBam were cultured under Bvg⁺ and Bvg^– phase growth conditions (SS broth without and with 50 mM MgSO_4, respectively) at 37 °C for ∼18 h with agitation and subsequently plated on BG-blood solid agar to enumerate the percentage of GFP⁻ colonies (A) and LCPs (B). GFP negativity indicates bacterial modulation and LCPs indicate bacteria present in the Bvg^– phase when plated. For both GFP⁻ colonies and LCPs, the mean percentages between WT and ∆plrS strains were not statistically significant under either Bvg⁺ or Bvg^– growth conditions (t test with Welch’s correction).

Following intranasal inoculation of BALB/cJ mice, the numbers of cfu recovered from the nasal cavities and lungs of WT and ΔplrS strains containing the reporters were similar to cfu recovered of the parental strains (without reporters), indicating that the reporters do not influence virulence (Fig. 3A). The proportion of GFP^– and LCP cfu recovered from the nasal cavities was extremely low for both strains, indicating that the bacteria did not modulate to the Bvg^– phase at this site (Fig. 3 B and C). The proportion of GFP^– and LCP cfu recovered from the lungs of mice inoculated with WT bacteria was also extremely low. In contrast, 40–80% of the cfu recovered from the lungs of mice inoculated with the ∆plrS strain had lost GFP fluorescence and 10–80% formed LCPs by day 1 postinoculation (Fig. 3 B and C), indicating that a majority of these bacteria had modulated to the Bvg^– phase and that a majority of these bacteria were in the Bvg^– phase at the time of recovery from the lungs. These findings demonstrate that in strains lacking plrS, the BvgAS phosphorelay fails to remain active specifically within the LRT. Moreover, the fact that 10–80% of the bacteria recovered from the lungs formed LCPs (i.e., much more than 5%) indicates that the physiology of ∆plrS bacteria in the LRT is substantially different from the physiology of ∆plrS bacteria that have modulated to the Bvg^– phase in vitro by chemical modulators such as MgSO₄ (Discussion).

Fig. 3.

Without PlrS, B. bronchiseptica modulates to the Bvg^– phase in the LRT. (A) Colonization of the nasal cavity (Upper) and right lung (Lower) for WT (filled circles) and ΔplrS (red triangles) on days 0, 1, and 3 postinoculation. Both strains contain the plasmid reporters pGFLIP-P_flaA and pBam. Female BALB/cJ mice were inoculated with 7.5 × 10⁴ cfu via the external nares. Each symbol represents a single animal, with the mean colonization depicted as short horizontal bars. Homogenate from each organ was assessed to determine in vivo bacterial modulation shown in graphs in B and C. (B) Percentage of GFP^– bacteria recovered from the nasal cavity and right lung for WT and ΔplrS containing pGFLIP-P_flaA and pBam. GFP negativity indicates bacterial modulation. (C) Percentage of LCP-producing bacteria recovered from the nasal cavity and right lung for WT and ΔplrS containing pGFLIP-P_flaA and pBam. LCPs indicate bacteria present in the Bvg^– phase when plated. (D) Colonization of the nasal cavity (Upper) and right lung (Lower) for WT (filled circles) and ΔplrS (red triangles) containing the plasmid reporter, pGFLIP-P_ptxA. Homogenate from each organ was assessed to determine BvgAS activation in vivo, shown in graph in E. (E) Percentage of GFP^– bacteria recovered from the nasal cavity (Upper) and right lung (Lower) for WT and ΔplrS containing pGFLIP-P_ptxA. GFP negativity indicates BvgAS activation. Statistical significance (Mann–Whitney) is indicated as *P < 0.05 and **P < 0.01.

B. bronchiseptica Lacking plrS Fail to Activate BvgAS Within the LRT.

To determine if BvgAS can transition from an inactive (Bvg^– phase) to an active (Bvg⁺ phase) state in the absence of PlrS activity within the LRT, we inoculated mice with WT and ∆plrS mutants containing a pGFLIP reporter in which the Bvg⁺ phase-specific ptxA promoter from B. pertussis was cloned 5′ to flp. Both WT and ∆plrS strains activated the ptxA promoter when switched from Bvg^– phase to Bvg⁺ phase growth conditions in vitro (Fig. S5). Similar to what has been shown in B. bronchiseptica and B. pertussis previously (15, 16), 100% of the bacteria recovered from both the nasal cavity and the lungs of mice inoculated with Bvg^– phase WT bacteria activated P_ptxA, and therefore transitioned to a Bvg⁺ phase within 24 h postinoculation (Fig. 3 D and E). In contrast, although all of the cfu recovered from the nasal cavities of mice inoculated with the ∆plrS strain were GFP^– by 24 h postinoculation, only ∼80% of those recovered from the lungs were GFP^– (Fig. 3 D and E). Although, based on this reporter, only a seemingly small proportion of the ∆plrS mutants failed to switch to the Bvg⁺ phase in vivo, it is important to note that the numbers of cfu of the ∆plrS mutant recovered from the lungs at days 1 and 3 postinoculation were ∼0.5 and 2 logs lower than the numbers initially present (day 0). It is impossible to determine if the bacteria cleared from the lungs at these time points had lost GFP. However, if they remained GFP⁺ (indicating that they did not switch to the Bvg⁺ phase), then the proportion of ∆plrS bacteria that had transitioned to the Bvg⁺ phase in the LRT would in fact be far less than 1%. Taken together, our data indicate that full activation and maintenance of BvgAS activity in the LRT requires PlrS.

Fig. S5.

In vitro analysis of WT and ΔplrS phase transition using the plasmid reporter pGFLIP-ptxA. WT and ΔplrS bacteria containing pGFLIP-ptxA were cultured under Bvg⁺ and Bvg^– phase growth conditions (SS broth without and with 50 mM MgSO_4, respectively) at 37 °C for 8 h with agitation. At various time points culture samples were taken and subsequently plated on BG-blood solid agar to enumerate the percentage of GFP⁻ colonies over time.

PlrS Is Required for B. bronchiseptica Persistence in the LRT, Independent of Its Effects on BvgAS Activity.

Lack of production of BvgAS-dependent virulence factors could be the reason that plrS mutants are cleared rapidly from the LRT. To test this hypothesis, we introduced the bvgS-C3 mutation, which encodes a BvgS protein that is active even under modulating conditions in vitro (12), into the plrS mutant, and compared this ∆plrS-bvgS^c strain with the bvgS^c mutant in vitro and in vivo. Like the bvgS^c strain, the ∆plrS-bvgS^c strain formed small, domed, hemolytic colonies characteristic of the Bvg⁺ phase on BG-blood agar containing 50 mM MgSO₄ (i.e., Bvg^– phase conditions), indicating that BvgAS was constitutively active in the absence of plrS in vitro. In vivo, the ∆plrS-bvgS^c strain colonized and persisted in the nasal cavity similarly to the bvgS^c and ΔplrS strains (Fig. S6). However, the ∆plrS-bvgS^c strain was cleared from the lungs as rapidly as the ∆plrS mutant (Fig. S6), indicating that the ∆plrS mutation is epistatic to the bvgS-C3 mutation with regard to persistence in the LRT.

Fig. S6.

A constitutively-active bvgS mutation does not prevent rapid clearance of a ΔplrS strain in the LRT. Colonization of the nasal cavity and right lung for bvgS^c (black squares), ΔplrS (red triangles), and ΔplrS-bvgS^c (blue squares) strains on days 0 and 3 postinoculation. Female BALB/c mice were inoculated with 7.5 × 10⁴ cfu via the external nares. Each symbol represents a single animal, with the mean colonization depicted as short horizontal bars. Statistical significance (Mann–Whitney) is indicated as **P < 0.01.

To investigate BvgAS activity in the ∆plrS-bvgS^c strain, we used our pGFLIP-flaA and pBAM reporters. None of the bvgS^c or ∆plrS-bvgS^c colonies containing these reporters were GFP^– or displayed the LCP phenotype after growth in medium containing 50 mM MgSO₄ (Fig. S7), indicating their insensitivity to modulating conditions in vitro. Numbers of cfu of each strain recovered from the nasal cavities and lungs of BALB/cJ mice were similar to those of the strains lacking the pGFLIP-flaA and pBAM reporters (Fig. 4A). Additionally, very few, if any, GFP^– or LCP colonies were recovered from the nasal cavity for any strain, indicating that no bacteria had modulated to the Bvg^– phase in the nasal cavity (Fig. 4 B and C). As expected, a significant proportion of GFP^– colonies and LCPs were recovered from the lungs of mice inoculated with the ∆plrS mutant on days 1 and 3, consistent with our previous results (Fig. 4 B and C). In contrast, no GFP^– or LCP colonies were recovered from the lungs of mice inoculated with the bvgS^c or ∆plrS-bvgS^c mutants (Fig. 4 B and C), indicating that neither strain modulated to the Bvg^– phase in the LRT. Therefore, modulation to the Bvg^– phase and lack of BvgAS-activated virulence factors is not the only reason that ∆plrS mutants fail to persist in the LRT. These data indicate that PlrS is required for bacterial persistence in the LRT, even when BvgAS remains active, likely because PlrR (presumably phosphorylated PlrR, PlrR∼P) activates expression of one or more genes encoding proteins that are required in this environment or because PlrR∼P represses expression of one or more genes that encode proteins that are detrimental to survival in this environment.

Fig. S7.

In vitro modulation of constitutively-active bvgS strains using the plasmid reporters pGFLIP-flaA and pBam. The strains bvgS^c, ΔplrS, and ΔplrS-bvgS^c containing pGFLIP-flaA and pBam were cultured under Bvg⁺ and Bvg^– phase growth conditions (SS broth without and with 50 mM MgSO_4, respectively) at 37 °C for ∼18 h with agitation and subsequently plated on BG-blood solid agar to enumerate the percentage of GFP⁻ colonies (A) and LCPs (B). GFP negativity indicates bacterial modulation and LCPs indicate bacteria present in the Bvg^– phase when plated. Statistical significance (t test with Welch’s correction) is indicated as *P < 0.05.

Fig. 4.

PlrS is required for persistence of B. bronchiseptica in the LRT independent of BvgAS activity. (A) Colonization of the nasal cavity (Upper) and right lung (Lower) B. bronchiseptica containing constitutively active BvgS (bvgS^c; black squares), ΔplrS (red triangles), and ΔplrS-bvgS^c (blue squares) on days 0, 1, and 3 postinoculation. All strains contain the plasmid reporters pGFLIP-P_flaA and pBam. Female BALB/c mice were inoculated with 7.5 × 10⁴ cfu via the external nares. Each symbol represents a single animal, with the mean colonization depicted as short horizontal bars. Homogenate from each organ was assessed to determine in vivo bacterial modulation shown in graphs in B and C. (B) Percentage of GFP^– bacteria recovered from the nasal cavity and right lung for bvgS^c, ΔplrS, and ΔplrS-bvgS^c containing pGFLIP-flaA and pBam. GFP negativity indicates bacterial modulation. (C) Percentage of LCP producing bacteria recovered from the nasal cavity and right lung for bvgS^c, ΔplrS, and ΔplrS-bvgS^c containing pGFLIP-P_flaA and pBam. LCPs indicate bacteria present in the Bvg^– phase when plated. Statistical significance (Mann–Whitney) is indicated as *P < 0.05 and **P < 0.01.

PlrS Is Required for Survival and Persistence of B. pertussis in the LRT.

The plrSR locus is highly conserved (≥99% identical) among all strains of the classic bordetellae (B. pertussis, B. bronchiseptica, and Bordatella parapertussis). We constructed a derivative of B. pertussis strain BP536 with a large in-frame deletion mutation in plrS and compared it with WT BP536 in BALB/cJ mice. Both strains colonized the nasal cavity at similar levels, and were cleared from this site by day 3 postinoculation (Fig. 5). However, whereas ∼10³ cfu of BP536 were recovered from the lungs on days 1 and 3 postinoculation, significantly fewer BP536ΔplrS cfu were recovered from this site at these time points (Fig. 5). These data indicate that similar to B. bronchiseptica, plrS is required for the survival and persistence of B. pertussis specifically in the LRT.

Fig. 5.

PlrS is required for the persistence of B. pertussis in the LRT. Colonization of the nasal cavity (Upper) and right lung (Lower) for B. pertussis strains BP536 (filled circles) and BP536ΔplrS (red diamonds) on days 0, 1, and 3 postinoculation. Female BALB/c mice were inoculated with 1.0 × 10⁵ cfu via the external nares. Each symbol represents a single animal, with the mean colonization depicted as short horizontal bars. Statistical significance (Mann–Whitney) is indicated as **P < 0.01.

Discussion

BvgAS has been considered the master virulence regulator in Bordetella since its identification in 1983 (30) and demonstration of its penetrance by subsequent mutagenesis and genome-wide analyses (8, 31, 32). Our new data indicate, however, that in the LRT, BvgAS activity depends on PlrS, likely via the activity of PlrR. Moreover, PlrS(R) is required for bacterial survival in the LRT even when BvgS is rendered constitutively active, strongly suggesting that one or more PlrSR-dependent, BvgAS-independent genes (or genes that require both PlrSR and BvgAS) is required for bacterial survival at this site. PlrS(R) is therefore at least as important for Bordetella virulence as BvgAS.

Why is PlrS(R) required for bacterial persistence in the LRT? PlrSR belongs to the NtrYX family of TCS. NtrY family proteins, including PlrS, are predicted to contain three transmembrane domains at their N termini, followed by a periplasmically located PhoP-DcuS-CitA (PDC) domain, another transmembrane domain, then cytoplasmically located HAMP, Per-Arnt-Sim (PAS), and HisKA-type histidine kinase domains (Fig. 6.). NtrY of Brucella abortus has been shown to bind heme via its PAS domain and to function as a redox sensor, becoming active as a kinase in response to anaerobiosis (33). NtrX family response regulators contain N-terminal receiver and C-terminal DNA binding domains. In Neisseria gonorrheae and Rhodobacter capsulatus, the ntrX genes are required for induction of high-affinity cytochrome oxidases, which are required for bacterial growth under low oxygen conditions (34). Our data indicate that PlrS is required specifically in the LRT, an environment that is low in oxygen and high in CO₂. Although we have been unable to delete plrR in vitro, our experiments using PlrR^D52E indicate that PlrS affects BvgAS-dependent phenotypes in vitro via PlrR (providing evidence, but not proving, that PlrS and PlrR function as a TCS). It seems likely that PlrS functions through PlrR in vivo as well. By analogy with the few other NtrYX family members that have been studied so far, we hypothesize that PlrS senses low oxygen (and perhaps increased CO₂) in the LRT, phosphorylates PlrR, and that PlrR∼P activates expression of one or more of the high-affinity cytochrome oxidase-encoding loci present in B. bronchiseptica (and B. pertussis) (35), allowing the bacteria to respire in this environment (Fig. 6). We note, however, that NtrY family members, including PlrS, contain HisKA-type DHp domains with ExxN motifs that suggest that these proteins possess both kinase and phosphatase activities (36–38). Although our data suggest that PlrS may function to phosphorylate PlrR in vitro, especially in 5% CO₂, the contributions of its predicted kinase and phosphatase activities in vivo cannot be predicted from our current data.

Fig. 6.

PlrSR and BvgAS work in tandem to regulate gene expression required for Bordetella persistence in the LRT. Model of putative PlrSR and BvgAS regulatory networks in Bordetella. The sensor kinase PlrS may be sensitive to CO₂ or oxygen tension, directly or indirectly, within the LRT environment, transmitting a signal via autophosphorylation and phosphate transfer to its putative response regulator PlrR. PlrSR activates (pags) or represses (prgs), a set of genes independent of BvgAS that are required for survival of the bacteria specifically within the LRT. PlrSR may also regulate expression of a gene encoding a protein required for sustained activity of BvgAS in the LRT. If BvgS receives such a signal the sensor kinase is capable of phosphorelay transmission, activating transcription of essential virulence factors (vags) and repressing transcription of Bvg^– phase genes (vrgs).

Why is PlrS(R) required for BvgAS activity in the LRT? Positive autoregulation by BvgAS has been well-characterized (25–28, 39) and we showed here that bvgAS expression in both ambient air and 5% CO₂ is not dependent on plrS, by evaluating activity of the bvgAS promoter from a single-copy chromosomal reporter (to avoid issues that arise when multicopy plasmids are used) while also leaving the native bvg locus intact, thereby preserving the positive autoregulation capability of BvgAS during the experiment. Our data demonstrate that although bvgAS autoregulation (the only known mechanism that affects bvgAS expression) remains intact, plrS does not influence bvgAS expression. Moreover, our experiments with the bvgS^c mutant further indicate that bvgAS expression is not dependent on PlrS(R) in vivo because BvgAS in the ∆plrS-bvgS^c strain—which contains a single nucleotide substitution in bvgS, rendering the BvgS protein constitutively active—was active in the LRT. If bvgAS transcription required PlrSR, the ∆plrS mutation would be epistatic to the bvgS-C3 mutation with regard to BvgAS activity. Our data therefore rule out the possibility that PlrS(R) controls BvgAS activity by controlling bvgAS transcription. We hypothesize that instead, PlrSR controls expression of one or more genes that encode proteins required for BvgS activity specifically in the LRT. BvgS contains three predicted signal-sensing domains: two periplasmically located Venus flytrap domains (VFT) and a cytoplasmically located PAS domain (40, 41). Although the VFT appear to convert BvgS into a phosphatase that dephosphorylates BvgA in response to modulating conditions (i.e., MgSO₄ or nicotinic acid) (42–44), a role for the PAS domain has not been established. However, biochemical analyses using the cytoplasmic portion of BvgS suggested that the redox state of ubiquinone could affect BvgS kinase activity (45). PlrSR-dependent production of high-affinity cytochrome oxidases would allow electron transport-coupled oxidative phosphorylation to occur in the LRT, which is required for ATP production (and hence cell viability), and for preventing the accumulation of reduced ubiquinone, which could inactivate BvgS. This model is consistent with that proposed for the ArcB protein of Escherichia coli, which is activated by low-oxygen conditions (46, 47). If true, this model would predict that a role for the PAS domain in sensing signals in vitro may have been missed because the activity of low-affinity (high-efficiency) cytochrome oxidases present under all of the conditions tested would keep reduced ubiquinone levels at a minimum. These low-affinity cytochrome oxidases are also likely present and sufficient for respiration in the upper respiratory tract, making BvgS activity and bacterial survival at this site independent of PlrS.

Why was the proportion of LCPs recovered from the lungs of mice inoculated with the ∆plrS strain so much higher than the proportion obtained after switching the bacteria from Bvg^– phase conditions to Bvg⁺ phase conditions in vitro? Our previous studies showed that the reason that ∼5% of bacteria containing the pBAM plasmid integrated at the bvgAS promoter region form LCPs when shifted from Bvg^– phase conditions to Bvg⁺ phase conditions in vitro, is that activity of the P2 promoter in this strain is decreased such that ∼5% of the bacteria have BvgAS levels below the threshold required to respond to Bvg⁺ phase conditions when they are encountered. It appeared, therefore, that the maximum proportion of LCPs that would form from a population of pBAM-containing bacteria in which 100% of the bacteria had modulated to the Bvg^– phase was about 5%. We were surprised, therefore, to find that the proportion of LCPs formed after recovery of the ∆plrS mutant from the lungs of mice was up to 80%. Given that the activity of the P2 promoter is already extremely low, it seems unlikely that P2 promoter activity in the ∆plrS mutant is substantially lower in the LRT than under Bvg^– phase conditions in vitro. A more likely possibility is that the physiology of the ∆plrS mutant in the LRT is substantially different from the physiology of the ∆plrS mutant growing under Bvg^– phase conditions in vitro (and from WT bacteria in the LRT, which do not modulate to the Bvg^– phase), and that this altered physiology prevents or delays BvgS from reactivating upon exposure to Bvg⁺ phase conditions in vitro. If PlrS(R) controls expression of cytochrome oxidases in the LRT, and if BvgS is sensitive to the redox state of ubiquinone, then BvgS would be inactive in a plrS mutant in the LRT and would not become active again until cytochrome oxidase activity reached levels sufficient to reoxidize ubiquinone pools, which, in the case of a plrS mutant, could require expression of genes encoding low-affinity cytochrome oxidases, translation of proteins, and assembly of enzymatic complexes in the membrane after the bacteria are shifted to Bvg⁺ phase conditions in ambient air. This delay could account for the substantially higher proportions of LCPs formed by the plrS mutant recovered from the LRT compared with the proportion formed after growth in vitro.

Challenging the long-held paradigm that the Bvg⁺ phase is sufficient for respiratory infection and evidence that virulence factors that require PlrS(R) independently of (or codependently with) BvgAS exist is important from a translational medicine perspective. B. bronchiseptica and B. pertussis are sufficiently closely related to be considered members of the same species (35), and several vags have been shown to be functionally interchangeable (48–51). As in B. bronchiseptica, a B. pertussis plrS mutant was defective in persistence in the LRT. Our model for the role of PlrSR during infection based on studies with B. bronchiseptica is therefore likely to apply to B. pertussis as well. In addition to mounting evidence that acellular and whole-cell vaccines provide suboptimal protection that is less durable than that induced by infection with B. pertussis, recent data from studies with baboons indicate that although both vaccines protect against disease, neither protects against colonization (52). Moreover, baboons vaccinated with an acellular vaccine and then challenged with B. pertussis were able to transmit the disease to naïve animals, even though they exhibited no signs of illness (52). The fact that infection induces a mixed Th1/Th17-type immune response and acellular vaccines induce a strong Th2-type immune response (4, 53–56) suggests that reformulation of acellular vaccines with adjuvants to induce a Th1/Th17 response is a reasonable approach to controlling the reemergence of pertussis. However, the phenomenon of epitope-linked suppression would likely render such vaccines ineffective in individuals previously vaccinated with acellular vaccines and perhaps also in those vaccinated with whole-cell vaccines, which are prepared from in vitro grown bacteria. Inclusion of antigens not present in previous vaccines would avoid this problem. The existence of virulence factors that are produced only in the host during infection, as suggested by our data, is significant from this translational medicine perspective. Antigens that are produced only during infection could also be important diagnostically as a means to distinguish individuals who have been vaccinated from those who have been infected. Finally, PlrSR itself, and factors it controls, may be exploitable as therapeutic targets.

Materials and Methods

In Vitro Adherence Assays.

Bacterial adherence to rat lung epithelial (L2) cells was evaluated as described previously (21, 48). Bacteria were cultured in either ambient air or 5% CO₂ at 37° to an OD₆₀₀ of ∼1.0. Bacteria were added to a monolayer of L2 cells (∼80% confluency) at a multiplicity of infection (MOI) of 150. Adherence was visualized by Giemsa staining and light microscopy at 1,000× magnification using a Zeiss Axiostar microscope, and quantified by counting the number of adherent bacteria and total L2 cells in at least four microscopic fields.

Macrophage Cytotoxicity Assays.

J774 macrophage-like cells were grown to ∼60% confluency (∼1.5 × 10⁴ cells per well) in a 96-well microtiter dish. Bacteria were cultured in either ambient air or 5% CO₂ to an OD₆₀₀ of ∼1.0, and were then added to J774 cells at an MOI of 150. The plate was spun at 1,200 × g and then allowed to incubate in a tissue culture incubator for 3 h. Macrophage cytotoxicity was quantified by measuring lactate dehydrogenase release using Promega’s Cytotox96 Nonreactive Cytotoxicity Assay kit and a MultiskanEX plate reader (ThermoFisher Scientific) according to the manufacturer’s instructions.

Bacterial Colonization of the Mouse Respiratory Tract.

Six-week-old female BALB/cJ mice from Jackson Laboratories were inoculated intranasally with 7.5 × 10⁴ cfu B. bronchiseptica or 1.0 × 10⁵ cfu B. pertussis in 50 μL of PBS. For all time points, right lung lobes and nasal cavities were harvested, tissues were homogenized, and the number of cfu was determined by plating dilutions of tissue homogenates on BG blood agar.

Evaluation of B. bronchiseptica Phenotypic Phase Transition in Vivo.

Six-week-old female BALB/cJ mice (Jackson Laboratories) were inoculated intranasally with 7.5 × 10⁴ cfu of B. bronchiseptica pGFLIP strains in 50 μL of PBS. For experiments using the pGFLIP-PptxA reporter, bacteria were cultured in media containing 50 mM MgSO₄ to maintain the bacteria in the Bvg^– phase before mouse inoculation (15). For infections using the pGFLIP-PflaA or pBam systems, bacteria were cultured at 37 °C in media without the addition of MgSO₄ to maintain bacteria in the Bvg⁺ phase before mouse inoculation (15, 29). On days 0, 1, and 3 postinoculation, right lungs were harvested, homogenized in PBS, and plated in duplicate on BG-blood agar. For strains containing the pGFLIP-PptxA system, homogenization, dilution, and plating were carried out in the presence of 50 mM MgSO₄. The percent GFP^– colonies and percent LCP colonies were calculated by determining the ratio of GFP^– or LCP colonies to the total number of colonies isolated. Table S1 for strains and plasmids used in this study.

Table S1.

Bacterial strains and plasmids used in this study

Bacterial strain or plasmid	Description	Source
E. coli strains
DH5α	Molecular cloning strain	Gibco BRL
RHO3	Conjugation strain; Kms Δasd ΔaphA, DAP auxotroph	(58)
B. bronchiseptica strains
WT	WT B. bronchiseptica strain RB50; Smr	(12)
ΔplrS	RB50 containing an in-frame deletion in plrS removing the codons for amino acids 5–198; Smr	(18)
RB515	RB50ΔcyaA	(51)
bvgSC	RB50 containing a R570H point mutation in bvgS; Smr	(12)
ΔplrS-bvgSC	RB50 containing a R570H point mutation in bvgS and a deletion in plrS removing the codons for amino acids 5–198; Smr	Present study
ΔbscN	RB50::pSJ23; RB50 containing a plasmid insertion mutation in bscN (BB1628); Gmr	Present study
B. pertussis strains
BP536	WT B. pertussis strain; Smr	(59)
BP536ΔplrS	B. pertussis strain containing an in-frame deletion in plrS from residues 9 thru 768; Smr	Present study
Plasmids
pMD11	pSS4245-based allelic exchange vector containing a 0.8-kb DNA fragment deleting the codons for amino acids 5–198 of plrS	Present study
pGFLIP-PflaA	pUC18-based vector with PS12-gfp and nptII flanked by FRT sequences and flp recombinase driven by the RB50 flaA promoter; Apr Kmr*	(15)
pGFLIP-PptxA	pUC18-based vector with PS12-gfp and nptII flanked by FRT sequences and flp recombinase driven by the BPSM ptxA promoter; Apr, Kmr*	(15)
pTNS3	Tn7 transposase expression vector containing tnsABCD; Apr	(58)
pBam	Plasmid that integrates into the Bordetella chromosome at the bvgAS locus and allows for generation of LCPs; Apr, Gmr	(29)
pEG7	Suicide plasmid for B. bronchiseptica; Gmr	(9)
pSJ23	pEG7-based suicide plasmid containing a B. bronchiseptica DNA fragment from bscN (BB1628); Gmr	Present study

^*Km^r only under promoter-inactive conditions.

Ethics Statement.

This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (57). Our protocol was approved by the University of North Carolina Institutional Animal Care and Use Committee (Protocol ID: 13-238). All animals were properly anesthetized for inoculations, monitored regularly, and killed when moribund. Efforts were made to minimize suffering.

Statistical Analysis.

Statistical analysis was performed using Prism 6.0 software from GraphPad Software. Statistical significance was determined using unpaired Student’s t test with Welch’s correction, a Mann–Whitney test, or ANOVA test. Figures were generated using Adobe Illustrator CS6 (Adobe Systems). P < 0.05 was considered significant.

SI Materials and Methods

Growth Media and Bacterial Strains.

Escherichia coli was grown in Luria broth (LB) or on LB agar at 37 °C. WT Bordetella bronchiseptica RB50 and Bordetella pertussis BP536 (and their mutant derivatives) were grown at 37 °C on BG agar (Becton Dickinson Microbiology Systems) supplemented with 7.5% or 15% (vol/vol) defibrinated sheep blood, respectively (Hardy Diagnostics, Colorado Serum Co., and Hemostat) or in SS broth with SS supplement; 100 mg/mL (2,6-O-dimethyl) cyclodextrin (Heptakis) was added to B. pertussis cultures. When necessary, media were supplemented with streptomycin (Sm; 20 μg/mL), gentamicin (Gm; 30 μg/mL), kanamycin (E. coli, 50 μg/mL or B. bronchiseptica, 125 μg/mL), or diaminopimelic acid (DAP; 300 μg/mL). To visualize hemolysis on solid media, bacteria were grown in ambient or 5% CO₂ conditions for 2.5 d on BG blood agar plates and colonies were photographed with a Canon EOS 40D digital camera in the presence of both overhead lighting and backlighting.

Construction of Cloning Plasmids and Bacterial Strains.

Descriptions of each strain are given in Table S1. The strain ΔplrS-bvgS^c, containing the bvgS-R570H mutation and an in-frame deletion mutation in the plrS gene, was created using plasmid pMD11 (18) in an allelic exchange method that has been previously described (50). A B. bronchiseptica strain with a disruption in the bscN gene (BB1628, which encodes the ATPase for the T3SS), was created by cloning a 0.34-kb internal portion of bscN into the suicide vector pEG7, producing plasmid pSJ43, which was then integrated into the native bscN locus in the chromosome. The disruption was confirmed by PCR.

A B. pertussis 536 strain containing an in-frame deletion in the plrS (BP0571) gene was constructed by a pSS4245-based allelic exchange method such that all but the first eight and last two codons of BP0571 were deleted. Two 600-bp DNA fragments flanking the region to be deleted were generated by PCR, and to facilitate the allelic exchange, a spectinomycin-resistant (Spc^R) synthetic DNA cassette was inserted between the two PCR products to generate plasmid pQC2053. This plasmid was conjugated into BP536, to generate a ∆BP0571::spc strain, QC3729. To replace the Spc^R-marked BP0571 deletion with an unmarked, in-frame deletion, plasmid pQC2027, which had been constructed by removing the Spc^R DNA cassette from pQC2053, was conjugated into strain QC3729, and an in-frame BP0571 deletion that had replaced the BP0571::spc allele was identified by screening exconjugates for a spectinomycin-sensitive phenotype following allelic exchange. One of these was chosen as BP536ΔplrS and confirmed by PCR.

Derivatives of B. bronchiseptica that contained the pGFLIP-P_flaA and pGFLIP-P_ptx reporters were constructed using methods previously described (15). The pBam reporter system was delivered to the bacterial chromosome via homologous recombination (29).

Bacterial Adherence to J774 Cells and in Response to Being Cultured in Media with Decreased pH.

Adherence to J774A.1 (J774) murine macrophage-like cells was performed essentially as described previously (48). In this study, bacteria were cultured in ambient or 5% CO₂ conditions to an OD₆₀₀ of ∼1.0 and added to J774 cells (at ∼60% confluency) at an MOI of 200. Adherent bacteria were labeled using sera that had been collected from an RB50-infected rat, and then recognized with an anti-rat Cy3-conjugated secondary antibody. Adherence was quantified via fluorescence microscopy by counting the number of adherent bacteria and total J774 cells in at least four independent microscopic fields at 400× magnification. To determine the role of pH in bacterial adherence to L2 cells, the pH of SS media was adjusted to 6.8, filter-sterilized, and used to culture RB50 under ambient conditions. Adherence to L2 cells was performed as described in Materials and Methods.

Evaluation of B. bronchiseptica bvgAS Transcription in Vitro.

A transcriptional P_bvgA-gfp reporter was created by PCR-amplifying the bvgAS promoter (P_bvgA, 350 bp upstream of bvgA) using the forward primer 5′-GATCCCGGGGTAGTCTGGATAA-3′ and the reverse primer 5′-GATAAGCTTGATAAGAAGAAT-3′. The PCR product was digested and cloned into the XmaI and HindIII restriction sites of the miniTn7-kan-gfp plasmid (58). The resulting pUC-P_bvgA-gfp plasmid was mated into WT, ∆plrS, ∆bvgS, and ∆bvgS∆plrS along with pTNS3, which encodes the transposase, allowing for the reporter to integrate into the attTn7 site of the B. bronchiseptica chromosome. The resulting strains were confirmed by PCR.

To determine the level of transcription at the bvgAS promoter, reporter strains were grown on BG agar with appropriate antibiotics until colonies formed. Colonies were resuspended in 1× PBS, and this suspension was used to inoculate duplicate tubes grown overnight for ∼18 h in SS media with streptomycin and kanamycin either at 37 °C on a roller or at 37 °C shaking in the presence of 5% CO₂. One-hundred microliters of each culture was transferred to a well of a 96-well plate, and the OD₆₀₀ and GFP (excitation A485, emission A535) were measured on a Perkin-Elmer Victor³ 1420 Multilabel plate reader, and GFP expression was normalized to OD₆₀₀. The experiment was performed on 2 separate days in duplicate, and values were averaged across all replicates.

Evaluation of B. bronchiseptica Promoter Activity in Vitro.

Strains containing pGFLIP-PptxA were grown overnight in the presence of 50 mM MgSO₄. Cells were washed twice in 1× PBS with 50 mM MgSO₄ to prevent the premature transition to the Bvg⁺ phase and were then inoculated into fresh SS medium without MgSO₄ at a concentration of 1× 10⁹ cfu/mL, and incubated at 37 °C for 8 h. At each time point, an aliquot of cells was removed, diluted with 1× PBS containing 50 mM MgSO₄, and plated on BG-blood agar also containing 50 mM MgSO₄. Plates were incubated for 48 h at 37 °C and total cfu were enumerated. The percent GFP⁻ colonies was determined for each plate as described in Materials and Methods.

Strains containing pGFLIP-P_flaA and pBam were grown overnight in SS medium with and without 50 mM MgSO₄. After incubation at 37 °C for 24 h, an aliquot of cells from each experimental condition was diluted with 1× PBS and plated on BG-blood agar. Plates were incubated for 48 h at 37 °C and total cfu were enumerated. The percent GFP⁻ colonies was determined for each plate as described in Materials and Methods.