bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 The BvgS PAS domain: an independent sensory perception module in the Bordetella BvgAS

2 phosphorelay

3

4 M. Ashley Sobran1 and Peggy A. Cotter1,*

5

6 1 Department of Microbiology and Immunology, University of North Carolina Chapel Hill, Chapel

7 Hill, NC 27599

8

9 *Corresponding author: [email protected]

10

11 Keywords: Bordetella, two-component system, PAS domain, BvgAS, PlrSR, respiratory infection

12 Running Title: The BvgS PAS domain is an independent signal perception domain

13

14

15

16

17

18

19

20

21

22 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

23

24 Abstract

25

26 To detect and respond to the diverse environments they encounter, bacteria often use two-

27 component regulatory systems (TCSs) to coordinate essential cellular processes required for

28 survival. In pathogenic Bordetella species, the BvgAS TCS regulates expression of hundreds of

29 genes, including those encoding all known protein virulence factors, and its kinase activity is

30 essential for respiratory infection. Maintenance of BvgS kinase activity in the lower respiratory

31 tract (LRT) depends on the function of another TCS, PlrSR. While the periplasmic venus fly-trap

32 domains of BvgS have been implicated in responding to so-called modulating signals in vitro

33 (nicotinic acid and MgSO4), a role for the cytoplasmic Per-Arnt-Sim (PAS) domain in signal

34 perception has not previously been demonstrated. By comparing B. bronchiseptica strains with

35 mutations in the PAS domain-encoding region of bvgS with wild-type bacteria in vitro and in

36 vivo, we found that although the PAS domain is not required to sense modulating signals in

37 vitro, it is required for the inactivation of BvgS that occurs in the absence of PlrS in the LRT of

38 mice, suggesting that the BvgS PAS domain functions as an independent signal perception

39 domain. Our data also indicate that the BvgS PAS domain is important for controlling absolute

40 levels of BvgS kinase activity and the efficiency of the response to modulating signals in vitro.

41 Our results indicate that BvgS is capable of integrating sensory inputs from both the periplasm

42 and the cytoplasm to control precise gene expression patterns in diverse environmental

43 conditions.

44 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

45 Importance

46

47 Despite high rates of vaccination, Pertussis, a severe, highly contagious respiratory disease,

48 caused by the bacterium Bordetella pertussis, has reemerged as a significant health threat. In

49 Bordetella pertussis and the closely related species, Bordetella bronchiseptica, activity of the

50 BvgAS two-component regulatory system is critical for colonization of the human respiratory

51 tract and other mammalian hosts, respectively. Here we show that the cytoplasmic PAS domain

52 of BvgS can function as an independent signal perception domain that is capable of integrating

53 environmental signals that influence overall BvgS activity. Our work is significant as it reveals a

54 critical, yet previously unrecognized role, for the PAS domain in the BvgAS phosphorelay and

55 provides a greater understanding of virulence regulation in Bordetella.

56

57 Introduction

58

59 Pertussis, also known as whooping cough, is a severe, highly contagious, reemerging respiratory

60 disease (1). It is caused by the bacterium Bordetella pertussis, which infects only humans and

61 appears to be capable of surviving only transiently outside the human respiratory tract (2). By

62 contrast, the closely related species Bordetella bronchiseptica infects nearly all mammals and

63 can survive long periods of time in ex vivo environments, including those with scarce nutrients

64 such as filtered pond water and buffered saline (3, 4). Despite these differences, B. pertussis

65 and B. bronchiseptica produce a nearly identical set of virulence factors that are regulated at

66 the level of transcription by functionally interchangeable BvgAS Two-Component Regulatory bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

67 Systems (TCS) (5). By activating and repressing at least four classes of genes, BvgAS controls

68 transition of the bacteria between three different phenotypic phases: the Bvg+ phase, which is

69 necessary for virulence in mammals (6), the Bvg– phase, which, in B. bronchiseptica, is required

70 for survival under nutrient-limiting conditions (4) and for interactions with non-mammalian

71 hosts (7), and the Bvgi phase, which has been hypothesized to be important for bacterial

72 transmission (8).

73

74 The BvgAS system is composed of a hybrid sensor kinase, BvgS, and a response regulator, BvgA

75 (9). BvgS contains tandem, N-terminal, periplasmic Venus Flytrap (VFT) domains, followed by a

76 transmembrane domain, a cytoplasmic Per/Arnt/Sim (PAS) domain, a histidine kinase (HK)

77 domain, a receiver (REC) domain, and a C-terminal, histidine phosphotransfer (Hpt) domain

78 (Fig. 1A). BvgA contains an N-terminal REC domain and a C-terminal helix-turn-helix (HTH) DNA

79 binding domain. When Bordetella are grown on Bordet-Gengou (BG) blood agar or in Stainer-

80 Scholte (SS) broth at 37C, BvgS autophosphorylates, and, via a His-Asp-His-Asp phosphorelay,

81 trans-phosphorylates BvgA. BvgA~P forms homodimers capable of binding DNA and activating

82 or repressing gene transcription (10). Under these conditions, BvgA~P levels are high and the

83 bacteria express the Bvg+ phase, which is characterized by transcription of all known virulence-

84 activated genes (vags), including those encoding critical virulence factors (11). When Bordetella

85 are grown at 25°C or at 37°C in the presence of millimolar concentrations of nicotinic acid (NA)

86 or magnesium sulfate (MgSO4), the BvgS phosphorelay is reversed and BvgS dephosphorylates

87 BvgA, resulting in the Bvg– phase. The Bvg– phase is characterized by the expression of genes

88 that are repressed by BvgA~P (termed virulence-repressed genes (vrgs)) and lack of expression bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

89 of vags (11). In B. bronchiseptica, flagellar regulon genes, including the flagellin-encoding gene

90 flaA, are vrgs (12). When Bordetella transition between the Bvg– phase to the Bvg+ phase or

91 when the bacteria are cultured in the presence of low concentrations of NA or MgSO4, BvgA~P

92 levels in the cell are low, resulting in the Bvgi phase, which is characterized by repression of the

93 vrgs, maximal expression of the intermediate phase gene bipA, and expression of vags with

94 promoters containing high affinity BvgA~P binding sites (such as bvgAS itself and fhaB, encoding

95 the adhesin filamentous hemagglutinin (FHA)) but not vags with promoters containing low

96 affinity BvgA~P binding sites (such as the ptx operon, which encodes pertussis toxin and its

97 export system) (10, 13, 14). The transition from Bvg+ phase to Bvg– phase is termed phenotypic

98 modulation.

99

100 Although the BvgAS phosphorelay and transcriptional control of the BvgAS regulon are well

101 characterized, the natural signals that impact BvgS activity and the mechanism by which these

102 signals are perceived and integrated by this system are unknown. Of BvgS’s three putative

103 sensory perception domains, the periplasmic VFTs have been studied most. Structure-function

104 analyses support a model in which VFT2 (the domain closest to the membrane) is in a closed

105 conformation in the absence of modulating signals, and that this conformation renders BvgS

106 active (15, 16). Hence the ‘default’ state for BvgS is one in which it is an active kinase. Binding of

107 NA to VFT2 alters its conformation and this change is apparently transduced across the

108 membrane, causing BvgS to reverse the phosphorelay and dephosphorylate BvgA, resulting in a

109 shift to the Bvg– phase (17).

110 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

111 The role of the PAS domain in regulating BvgS activity is less understood. PAS domains are

112 abundant in prokaryotic signal transduction proteins and have a conserved structure, which

113 includes a core binding pocket composed of a central, five-stranded antiparallel beta sheet and

114 flanking alpha helices (18). Despite their conserved structure, PAS domains can perceive a

115 multitude of stimuli, either by directly binding a signal molecule or by sensing with the

116 assistance of a co-factor, in addition to promoting protein-protein interactions that mediate

117 signal transduction (19–22). The PAS domain of BvgS contains two perception motifs that are

118 critical for PAS domain function in other regulatory proteins; a quinone binding motif [aliphatic-

119 (X)3-H-(X)2,3-(L/T/S)] and a highly conserved cysteine residue (C607) (23, 24). In vitro studies

120 with the cytoplasmic portion of BvgS from B. pertussis showed that kinase activity was

121 decreased in the presence of an oxidized ubiquinone analog and that the histidine in the

122 quinone binding motif (H643) contributed to this effect (23). Jacob-Dubuisson and colleagues

123 hypothesized that H643 may be involved in binding heme, but were unable to detect heme, or

124 other cofactors, bound to the BvgS PAS domain (25). Their characterization of an H643A mutant

125 of B. pertussis showed no defect in activation of the ptx promoter under aerobic conditions,

126 but, instead, a defect in responding to NA and MgSO4. This group also investigated the

127 importance of the conserved cysteine (C607) and found that a C607A mutant was similarly

128 defective only in its ability to modulate in response to NA (25). Approximately one-third of BvgS

129 homologs contain coiled-coil domains instead of PAS domains. Jacob-Dubuisson and colleagues

130 showed that a B. pertussis mutant in which the BvgS PAS domain was replaced with the coiled-

131 coil domain from the BvgS homolog of Enterobacter cloacae was similar to wild-type bacteria in

132 its ability to activate ptx expression and to respond to NA (26). Taking these and other bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

133 observations together, this group concluded that the role of the BvgS PAS domain is primarily

134 to transduce signals from the VFTs to the kinase domain and that it may not function as a

135 sensory perception domain to sense signals on its own.

136

137 Our laboratory has focused primarily on studying B. bronchiseptica because its broad host range

138 allows us to study the role of Bordetella virulence factors and their regulation using natural host

139 animal models (27, 28). We were motivated to investigate the role of the BvgS PAS domain

140 during respiratory infection by our recent observation that maintenance of BvgS kinase activity

141 in the lower respiratory tract (LRT) requires another TCS called PlrSR (29). PlrSR is a member of

142 the NtrYX family of TCS, members of which are involved in sensing and responding to changes

143 in redox status and anaerobiosis (30, 31). We hypothesized that, in the LRT, PlrSR controls

144 expression of genes that contribute to electron transport-coupled oxidative phosphorylation

145 and that deletion of plrS results in an altered redox state at the cytoplasmic membrane that is

146 sensed by BvgS, perhaps via the PAS domain, resulting in its inactivation (i.e., converting it from

147 a kinase to a phosphatase). In this study, we constructed B. bronchiseptica mutants with amino

148 acid substitutions in the BvgS PAS domain and characterized them in vitro and in vivo to test

149 this hypothesis.

150

151 Results

152

153 Mutants used in this study

154 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

155 To investigate the role of the BvgS PAS domain during infection, and to facilitate comparison of

156 our results with those obtained with B. pertussis, we constructed mutant derivatives of B.

157 bronchiseptica strain RB50 containing the same mutations as those that were present in the B.

158 pertussis BPSM strains studied by Jacob-Dubuisson and colleagues (25, 26). Specifically, we

159 constructed strains producing BvgS proteins in which H643 and C607 were replaced by alanine,

160 and a strain in which the amino acids corresponding to the PAS domain were replaced with the

161 analogous 47 amino acid (aa) region from the BvgS homolog of Enterobacter cloacae (Fig.

162 1B&C). We also constructed a strain in which BvgS C607 was replaced with the isosteric aa

163 serine. As controls, we included the Bvg+ and Bvg– phase-locked strains RB53 (aka BvgSc) and

164 RB54 (aka ∆bvgS), respectively.

165

166 Contribution of the PAS domain in regulating BvgS activity in vitro

167

168 To assess BvgS activity and sensitivity to modulating chemicals in our PAS domain mutants, we

169 used two different gfp reporters; one containing the fhaB promoter (PfhaB), which contains high-

170 affinity BvgA~P binding sites and therefore requires only very low levels of BvgA~P for

171 activation, and one containing the B. pertussis ptxA promoter (PptxA), which contains low-affinity

172 BvgA~P binding sites and requires high levels of BvgA~P for activation (10, 13, 32, 33). The

173 lower-affinity BvgA~P binding sites allow the PptxA-gfp fusion to discern a broader range of

174 BvgA~P levels (and hence BvgS kinase activities) than the PfhaB-gfp fusion. As expected, both

175 promoters were activated in wild-type bacteria (WT), but not in the ∆bvgS mutant, when grown

+ 176 in SS broth at 37°C (Bvg phase conditions). PfhaB-gfp expression was similar in the strain bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

c 177 containing the bvgS-C3 mutation (BvgS ) to that in WT, while PptxA-gfp expression was much

178 higher in BvgSc than in WT (Fig. 2A&B), indicating that BvgS kinase activity is greater in BvgSc

179 than in WT when grown in SS at 37°C. When the bacteria were grown in medium containing 4

– 180 mM NA or 50 mM MgSO4 (Bvg phase conditions), PfhaB-gfp and PptxA-gfp expression was very

181 low in the wild-type and ∆bvgS strains and remained high in BvgSc, demonstrating the

182 requirement of BvgS for activation of these promoters and the insensitivity of BvgS to NA and

C 183 MgSO4 in BvgS (Fig. 2A&B).

184

185 The “PASless” mutant (BvgSEc, the strain in which the BvgS PAS domain was replaced with the

186 analogous sequence from the BvgS homolog in E. cloacae) activated PfhaB-gfp to the same level

+ 187 as WT and PptxA-gfp to a greater level than WT when grown under Bvg phase conditions (Fig.

188 2A), suggesting that the PASless BvgS is a more efficient kinase than wild-type BvgS (Fig. 2B).

189 Expression of PptxA-gfp and PfhaB-gfp was significantly decreased when BvgSEc was cultured in the

190 presence of NA or MgSO4 compared to growth in the absence of these chemical modulators

191 (Fig. 2A&B), indicating that the PASless BvgS can sense MgSO4 and NA.

192

C 193 The H643A strain activated PfhaB and PptxA similarly to BvgS and BvgSEc when cultured in SS

194 medium at 37C, indicating that the H643A substitution does not disrupt BvgS’s ability to

195 function as a kinase, and indeed makes it a more efficient kinase, in vitro (Fig. 2A&B). When

196 grown in medium containing NA or MgSO4, PfhaB-gfp and PptxA-gfp expression was significantly

197 reduced in the H643A mutant compared to growth in the absence of NA and MgSO4, indicating

198 that the H643A substitution does not affect BvgS’s ability to detect these chemicals. bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

199

200 Substitution of C607 with alanine or serine affected BvgS activity differently, despite both

201 substitutions ablating the reactivity of the cysteine. The C607A mutation resulted in increased

+ 202 BvgS kinase activity (increased PptxA-gfp expression) under Bvg phase conditions (Fig. 2A&B). By

203 contrast, the C607S substitution resulted in decreased kinase activity (decreased PptxA-gfp

204 expression) under Bvg+ phase conditions. Neither mutation affected BvgS’s ability to respond to

205 NA or MgSO4 compared to WT (Fig. 2A&B).

206

207 In all of the BvgS PAS domain mutants, BvgS remained responsive to MgSO4 and NA, indicating

208 that the PAS domain is not required for sensing MgSO4 or NA. However, the level of BvgS

209 activity in the presence of NA and MgSO4 differed dramatically between strains, suggesting that

210 the PAS domain of BvgS does play a role in transducing signals sensed by the VFTs to the

211 cytoplasmic HK, REC and HPT domains, which is consistent with previous reports (25).

212

213 Characterization of pGFLIP-PflaA and pBam reporters in BvgS PAS domain mutants in vitro

214

215 To investigate the role of the PAS domain in vivo, we sought to measure bacterial burden and

216 BvgS activity in the respiratory tract. To measure BvgS activity in vivo, we used two reporters

217 that we have used previously (29, 34, 35). pGFLIP-PflaA is a recombinase-based reporter that

218 indicates whether the bacteria have expressed the flaA gene, and hence the Bvg– phase, at any

219 time during the course of the experiment. Expression of flaA results in a conversion of the

220 bacteria from GFP+ and Kmr to GFP– and Kms (34). By contrast, the pBam reporter provides an bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

221 indication of whether the bacteria are in the Bvg– phase at the time of plating on BG-blood

222 agar. For wild-type B. bronchiseptica RB50, approximately 1 – 5% of bacteria containing the

223 pBam plasmid will form large colonies (LCPs, for large colony phenotype) when plated and

224 grown under Bvg+ phase conditions after growth under Bvg– phase conditions (35). Although

225 this reporter vastly under-estimates the number of bacteria that are in the Bvg– phase at the

226 time of sampling, it has proven useful in discerning differences in BvgAS activity during growth

227 in vitro and in the lower respiratory tract (LRT) (29, 35).

228

229 We first characterized the use of pGFLIP-PflaA and pBam in our PAS domain mutants in vitro.

230 When grown overnight in SS broth at 37°C (Bvg+ phase conditions), all strains maintained GFP

231 fluorescence and formed few LCPs, indicating the bacteria did not modulate under these

232 conditions (Fig. 3A-D). Consistent with our previous results (29), when grown with either 4 mM

– 233 NA or 50 mM MgSO4 (Bvg phase conditions), nearly 100% of WT lost GFP fluorescence and ~2-

234 10% formed LCPs, indicating that a majority of the cells had modulated in response to these

– 235 chemicals and were in the Bvg phase at the time of plating (Fig. 3 A-D). For the BvgSEc, C607A,

236 and C607S strains, a majority of bacteria recovered after growth with NA and MgSO4 had also

237 lost GFP fluorescence, indicating that they had modulated to the Bvg– phase like WT. By

238 contrast, very few GFP– colonies were recovered for the H643A mutant, demonstrating the

– 239 inability of this strain to modulate to the Bvg phase in response to NA and MgSO4. Taken with

240 the PfhaB-gfp and PptxA-gfp fusion data, these results indicate that although BvgS with the H643A

241 substitution is able to sense NA and MgSO4, it does not dephosphorylate BvgA~P as completely

242 as wild-type BvgS such that PflaA is expressed. bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

243

244 In addition to indicating whether the bacteria are phenotypically Bvg– phase at the time of

245 plating, the pBam reporter provides an indication of how quickly BvgS transitions from

246 phosphatase activity to kinase activity, because LCPs are formed by delayed or defective

247 positive autoregulation of the bvgAS promoter (35). Although the differences were not

248 statistically significant, the proportion of LCPs for the BvgSEc, H643A, and C607A mutants after

249 growth in MgSO4, and for the BvgSEc and H643A mutants after growth in NA,

250 were lower than for WT, consistent with BvgS in these strains having increased kinase activity

251 (and hence increased positive autoregulation) compared to wild-type BvgS, as suggested by the

252 PptxA-gfp data (Fig. 2B). By contrast, the C607S mutant formed dramatically more LCPs than WT

253 bacteria after growth in NA, consistent with this BvgS protein having decreased kinase activity

254 compared to wild-type BvgS, which is also consistent with the PptxA-gfp data. Altogether, these

255 results indicate that the pGFLIP-PflaA and pBam reporters should be reliable indicators of BvgS

256 activity in our PAS domain mutants in vivo.

257

258 The PAS domain is not required for BvgS kinase activity in the LRT in PlrSWT bacteria

259

260 One hypothesis for how PlrSR influences BvgS activity when the bacteria are in the LRT is that

261 the BvgS PAS domain senses an activating signal that requires PlrSR and is needed for

262 maintenance of BvgS kinase activity (Fig. S1A). This signal would be absent in a ∆plrS mutant

263 leading to BvgS inactivation (Fig. S1B). If true, we predicted that PASless BvgS in BvgSEc would

264 be unable to sense the PlrSR-dependent activating signal and hence BvgS would fail to function bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

265 as a kinase in bacteria with wild-type PlrSR (as well as in ∆plrS bacteria). To test this hypothesis,

266 we inoculated mice intranasally with WT and BvgSEc containing pGFLIP-PflaA and pBam and

267 determined the number of colony forming units (CFU), GFP fluorescence, and LCP formation of

268 bacteria recovered from the nasal cavity (NC) and right lung lobes at 0, 1, and 3 days post-

269 inoculation. The BvgSEc mutant colonized the NC and lungs similar to WT and showed no

270 evidence of modulation (no conversion to GFP– and few, if any, LCPs) (Fig. 4A-C), indicating that

271 PASless BvgS had maintained kinase activity throughout the infection. Therefore, either our

272 initial hypothesis is incorrect or the PASless BvgS protein no longer requires a PlrSR-dependent

273 activating signal to initiate kinase activity in the LRT.

274

275

276 The C607S substitution hinders BvgS kinase activity and bacterial colonization in the lower

277 respiratory tract (LRT) of mice

278

279 We also inoculated mice intranasally with WT, H643A, C607A, and C607S containing pGFLIP-PflaA

280 and pBam to determine if the conserved histidine or cysteine residues in the PAS domain

281 contribute to maintenance of BvgS kinase activity in vivo. While the H643A and C607A mutants

282 colonized the NC and lungs similar to WT and showed no evidence of modulation (no

283 conversion to GFP– and few, if any, LCPs) (Fig. 4A-C), the C607S mutant was significantly

284 defective for colonization in the lungs and a small proportion of the C607S bacteria recovered

285 from the NC and lungs had modulated, as indicated by the loss of GFP and development of LCPs

286 (Fig. 4B&C). These data are consistent with the in vitro data suggesting that the H643A and bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

287 C607A mutants have increased BvgS kinase activity while the C607S mutant has decreased BvgS

288 kinase activity when grown under Bvg+ phase conditions. If these proteins function in vivo as

289 they do in vitro, our data suggest that decreased BvgS kinase activity, but not increased BvgS

290 kinase activity, is detrimental to the ability of the bacteria to colonize the LRT, which is

291 consistent with previous studies that demonstrate the necessity of the Bvg+ phase for

292 mammalian respiratory tract infection (6, 36).

293

294 A role for the BvgS PAS domain is revealed in the absence of PlrS in the LRT

295

296 A second hypothesis for how PlrSR influences BvgS activity when the bacteria are in the LRT is

297 that PlrSR prevents the generation of a signal that would inhibit BvgS kinase activity (Fig. S1C).

298 In a ∆plrS mutant, this inactivating signal would accumulate resulting in loss of BvgS kinase

299 activity (Fig. S1D). If true, and if that signal is sensed by the BvgS PAS domain, then PASless

300 BvgS, and potentially some of the PAS domain point mutants, would be insensitive to the

301 inactivating signal and remain active in a ∆plrS mutant. To test this hypothesis, we deleted plrS

302 in the BvgS PAS domain mutants and used the pGFLIP-PflaA and pBam reporters to evaluate BvgS

303 activity in these strains in vivo. Similar to and as previously shown for the ∆plrS mutant (29), all

304 strains containing the PAS domain mutations in addition to the ∆plrS mutation were severely

305 defective for survival in the LRT, but colonized the NC similarly to WT (Fig. 5A). Similar to the

306 parental ∆plrS strain, the C607A ∆plrS and C607S ∆plrS double mutants showed evidence of

307 modulation in the LRT, with a large percentage of GFP– bacteria recovered from the lungs at

308 days 1 and 3 post-inoculation (Fig. 5B). A small proportion of bacteria of these same strains also bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

309 modulated in the NC. These data suggest that C607 is not involved in the modulation of BvgS

310 that occurs in the LRT in the absence of PlrS. Interestingly, despite the large number of GFP–

311 CFU recovered for the C607A ∆plrS mutant, very few LCPs were generated (Fig. 5C), suggesting

312 that, in this strain, BvgS functions more efficiently as a kinase than wild-type BvgS once

313 modulatory pressure is removed, consistent with the in vitro data.

314

315 For the BvgSEc ∆plrS and H643A ∆plrS mutants, and in contrast to the ∆plrS mutant, very few

316 GFP– colonies and few, if any, LCPs were recovered from the LRT (Fig. 5B&C), indicating that

317 BvgS remained active in these strains, supporting the hypothesis that the PAS domain senses a

318 negative signal that accumulates in the absence of PlrS in the LRT. In vitro, both of these mutant

319 BvgS proteins were able to sense MgSO4 and NA, however, only the BvgSEc mutant was capable

– 320 of modulating to the Bvg phase sufficiently to activate PflaA in response to both chemicals. The

321 fact that the BvgSEc ∆plrS mutant did not activate PflaA in vivo indicates that the modulating

322 signal sensed by BvgS in the ∆plrS mutant background in the LRT is different, or is sensed

323 differently, than NA or MgSO4 in vitro. These results suggest that the PAS domain, rather than

324 the VFTs, is responsible for sensing the modulating signal that occurs in the LRT in the absence

325 of PlrS. These results are the first to implicate the PAS domain of BvgS as an independent

326 perception domain capable of influencing the activity of BvgS and support the hypothesis that

327 the BvgS PAS domain may sense an inactivating signal that is regulated by PlrSR.

328

329 Discussion

330 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

331 The fact that BvgS contains three putative signal perception domains and therefore has the

332 potential to integrate multiple distinct environmental signals has been known for 20 years (21,

333 37, 38). To date, however, roles only for the VFT domains have been identified (15–17). Studies

334 investigating the purpose of the PAS domain have suggested a role only in transducing signals

335 sensed by the VFTs to the cytoplasmic signaling domains (25, 26). Our data support this role for

336 the PAS domain. However, by investigating BvgS signaling during mammalian respiratory tract

337 colonization, we have discovered that the PAS domain can function as a signal perception

338 domain that is capable of influencing BvgS activity independently from the VFTs.

339

340 Our results cannot definitively distinguish whether the PAS domain senses an activating signal

341 that is dependent on PlrSR (Fig. S1A) or senses an inhibitory signal that is produced in the

342 absence of PlrSR (Fig. S1C). Regardless, our data support an important role for the PAS domain

343 in sensing a signal(s) independent of the VFT domains. Our in vitro transcriptional analyses,

344 showed that both wild-type BvgS and PASless BvgS can sense NA and MgSO4, presumably via

345 the VFT domains, and in response they sufficiently modulate the bacteria to the Bvg– phase. In

346 vivo, wild-type BvgS was also sensitive to a PlrS-dependent signal in the LRT (it modulated in the

347 ∆plrS strain), however, PASless BvgS was insensitive to this signal, indicating that this signal 1) is

348 sensed differently than NA or MgSO4 (not via the VFTs) and 2) specifically requires the PAS

349 domain for perception.

350

351 Similar to the PASless BvgS protein, BvgS with the H643A substitution was also insensitive to

352 the PlrS-dependent signal in vivo; the H643A strain did not activate the PflaA promoter in the bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

353 absence of PlrS in the LRT. Although this BvgS protein was capable of sensing NA and MgSO4 in

354 vitro, it did not modulate sufficiently to de-repress PflaA under these conditions, and therefore,

355 we cannot rule out that this strain did not modulate to some degree in vivo while in the LRT.

356

357 PlrSR is a member of the NtrYX family of TCS (39). Although some NtrYX family members are

358 involved in nitrogen homeostasis, others, such as those in Brucella abortus and Neisseria

359 gonorrhoeae, control genes involved in bacterial respiration (such as those encoding high

360 affinity cytochrome oxidases) in response to low oxygen (30, 40). We hypothesize that PlrSR

361 may similarly control the expression of genes that contribute to electron transport-coupled

362 oxidative phosphorylation. Absence of appropriate cytochromes oxidases when the bacteria are

363 in the LRT would likely alter the redox status of the electron transport chain components, such

364 as ubiquinone, and would also likely prevent bacterial growth. Bock and Gross showed that

365 BvgS kinase activity is inhibited by oxidized ubiquinone in vitro and that the H643 residue,

366 which resides within a putative quinone binding motif in the BvgS PAS domain, contributes to

367 BvgS’s sensitivity to this redox-active molecule (23). We are currently investigating whether

368 PlrSR controls the expression of the multiple operons encoding high and low affinity

369 cytochrome oxidases and, if so, their role in controlling BvgS activity in the LRT.

370

371 Although our data suggest that the conserved cysteine residue (C607) does not appear to

372 contribute to the sensing capacity of the BvgS PAS domain (at least not while the bacteria are in

373 the LRT), they do indicate that this residue is important for maintaining BvgS kinase activity. In

374 vitro, substitution of alanine for C607 enhanced BvgS kinase activity, while the C607S bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

375 substitution significantly decreased BvgS kinase activity. It was surprising that the C607A and

376 C607S mutations had different effects on BvgS kinase activity. Although both substitutions

377 abrogate the reactivity of the conserved cysteine, alanine and serine differ significantly by size

378 and reactivity of their R-group. Alanine is a relatively small amino acid and nonreactive

379 compared to cysteine and serine. Serine is capable of forming hydrogen bonds within a protein

380 and contains a similarly sized R-group compared to cysteine. It is possible, based on our data,

381 that when BvgS is in an active kinase confirmation C607 is modified in a way that reduces the

382 size of its side chain and that maintenance of a bulky R-group within the PAS domain, as would

383 occur with a serine amino acid, is disruptive and inhibits BvgS activation. This disruption would

384 not occur with the C607A substitution and we find that this mutation only enhances BvgS

385 kinase activity. Another possibility is that formation of aberrant hydrogen bonds by the non-

386 native serine residue may destabilize the protein conformation that promotes BvgS kinase

387 activity and would, therefore, disrupt protein function. To fully understand the impact these

388 amino acid substitutions have on BvgS, additional biochemical characterization of these

389 modified BvgS proteins is required.

390

391 Overall, our results are generally consistent with those of Jacob-Dubuisson and colleagues that

392 demonstrated a role for the BvgS PAS domain in regulating efficient BvgS signal transduction

393 (25). These results are not surprising as BvgS from B. bronchiseptica and B. pertussis are

394 functionally interchangeable and the PAS domain is highly conserved between species (only

395 differing by 2 amino acid residues) (5, 41). We did, however, identify some nuanced differences

396 in BvgS activity between B. bronchiseptica and B. pertussis. When assessing BvgS kinase activity, bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

397 our results indicated that BvgS in B. bronchiseptica fails to fully activate when the bacteria are

398 grown under laboratory conditions. We showed that replacement of some amino acids in the

399 PAS domain enhanced BvgS kinase activity compared to WT, indicating that the PAS domain

400 influences BvgS kinase activity in B. bronchiseptica. This enhancement of kinase activity was not

401 observed in B. pertussis with similar substitutions (25, 26), suggesting that BvgS may already be

402 maximally active in vitro and/or the PAS domain does not influence BvgS activation in B.

403 pertussis. These data could explain why some strains of B. pertussis appear less sensitive to

404 modulating signals in vitro (NA and MgSO4) compared to most B. bronchiseptica isolates (5). If

405 BvgS from B. pertussis has higher kinase activity in vitro, it is likely that more modulatory signal

406 would be required to inhibit BvgS kinase activity sufficiently to noticeably decrease vag

407 expression. We also observed that modification of the BvgS PAS domain had different effects

408 on BvgS’s responsiveness to NA and MgSO4 in each species. In B. bronchiseptica, BvgS with the

409 H643A substitution remained responsive to NA and MgSO4, while, in B. pertussis, this

410 substitution rendered BvgS insensitive to both chemicals (25). Similarly, the C607A substitution

411 made BvgS from B. pertussis insensitive to modulation by NA, however, BvgS responsiveness

412 was unchanged by this substitution in B. bronchiseptica (25). As the VFTs are implicated in

413 sensing NA and MgSO4, the differences we observed are likely due to the significant variability

414 of amino acid sequence encoding the BvgS VFT domains in B. bronchiseptica and B. pertussis (5,

415 41). While our data corroborate a role for the PAS domain in facilitating the transduction of

416 signals though the BvgS protein, our results also highlight that it is important to consider how

417 the VFT domains may influence the function of the PAS domain and how the synergy of all

418 perception domains contribution to the ultimate output of BvgS activity. bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

419

420 The VFT domains and the PAS domain serve as independent points of signal integration within

421 BvgS. It is well known that the modulatory signals NA and MgSO4 require the VFT domains for

422 perception in vitro and based on our working model, the PAS domain is perhaps sensitive to the

423 redox state of ubiquinone. The natural signals sensed by BvgS, however, remain unknown. By

424 having multiple separate signaling domains, BvgS has the potential to control bacterial gene

425 expression appropriately in many environments. For example, in the mammalian LRT BvgS

426 kinase activity is essential for bacterial survival, therefore, it is likely that both the VFT domains

427 and the PAS domain are in an ‘On’ or active conformation, thus permitting BvgS activity. In a

428 ∆plrS mutant in the LRT, however, BvgS is inactive and the PAS domain is likely in an ‘Off’ or

429 inactive conformation based on our results. This environment, although artificial, potentially

430 mimics a naturally occurring environment in which the PAS domain can integrate a signal that

431 overrides an activating input sensed by the VFT domains, indicating the possibility that the VFT

432 domains and the PAS domain could obtain various combinations of conformations, rendering

433 BvgS on or off depending on which domain’s influence dominates within the specific

434 environment encountered by the bacteria. Interestingly, not all BvgS homologs have PAS

435 domains, suggesting that bacteria with simplified BvgS proteins perhaps do not encounter such

436 a diverse range of environments and, therefore, do not require BvgS to integrate multiple

437 signals (26). The fact that a PAS domain in BvgS has been maintained in Bordetella is indicative

438 of the critical role this domain plays in the regulation of this important TCS.

439

440 Experimental Procedures bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

441

442 Bacterial Strains, Plasmids, and Growth Conditions

443

444 All bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli strains

445 were grown using Luria-Bertani (LB) broth or agar at 37C. B. bronchiseptica strains were grown

446 on Bordet-Gengou (BG) agar supplemented with 7.5% defibrinated sheep blood (Hemostat,

447 Cat#: DSB1) or in Stainer-Scholte (SS) broth supplemented with SS supplement (REF) at 37°C

448 (42). As needed, media was supplemented with ampicillin (Amp; 100 g/mL; Sigma, Cat#:

449 A0166), streptomycin (Sm; 20 g/mL; Gold Biotechnology, Cat#: S-150-50), gentamicin (Gm; 30

450 g/mL; Gold Biotechnology, Cat#: G-400-25), kanamycin (Km; E.coli at 50 g/mL or B.

451 bronchiseptica at 125 g/mL; Gold Biotechnology, Cat#: K-120-25), or diaminopimelic acid

452 (DAP; 300 g/mL; VWR, Cat#: AAB22391-06).

453

454 Construction of B. bronchiseptica BvgS PAS Domain Mutant Strains

455

456 DNA manipulation and cloning were performed according to standard methods (43). All

457 restriction enzymes and DNA ligase were obtained from New England BioLabs. High-fidelity PFU

458 Ultra II DNA polymerase (Agilent, Cat#: 600670) was used for the construction of all allelic

459 exchange vectors used in this study. GoTaq DNA polymerase (Promega, Cat#: M3001) was used

460 for screening purposes during plasmid and strain construction. All enzymes were used

461 according to the manufacturer’s instructions. The primers used in this study are listed in Table bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

462 2. All primers were synthesized by Eton Biosciences and purified using a standard desalting

463 method.

464

465 To create the BvgSEc mutation, primers MAB85-MAB90 were used to generate an approximately

466 1.2 kb DNA fragment, which contained a 141 bp sequence encoding an α-helical linker from the

467 BvgS homolog in Enterobacter cloacae (GI: 915610361), flanked by homologous sequences (500

468 bp each) corresponding to the regions 5’ and 3’ to the PAS domain of the B.b. bvgS gene.

469 Individual point mutations in bvgS (H643A, C607A, and C607S) were generated by constructing

470 allelic exchange plasmids containing an approximately 1.0 kb DNA fragment of the bvgS

471 sequence, centered on the PAS domain, with exact homology to bvgS except for the desired

472 substitution and a unique restriction enzyme digestion site, used for screening. Primers used to

473 generate the most 5’ and 3’ ends of the DNA fragments for each mutant were designed to

474 include a BamHI restriction site at the 5’ end and a SacI restriction at the 3’ end, which were

475 used to clone the DNA fragments into the allelic exchange plasmid pEG7S (4). Standard

476 methods were then used to deliver this plasmid to RB50 and ∆plrS strains by conjugation using

477 RHO3 cells. Double homologous recombination events that introduced the desired mutation

478 into the chromosome were selected for using 15% sucrose and screened for by restoration of

479 BvgS activity indicated by the presence of hemolysis on BG-blood agar or through restriction

480 digestion using the unique restriction enzyme site introduced for point mutants. Sanger

481 sequencing was used to verify each mutation in bvgS. The pGFLIP-PflaA and pBam reporters

482 were introduced into each mutant strain, using tri-parental mating with conjugative RHO3 or by bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

483 homologous recombination, respectively. Introduction of each reporter on to the chromosome

484 was verified using PCR.

485

486 Construction of PptxA-gfp and PfhaB-gfp Transcriptional Reporters

487

488 To construct a GFP transcriptional reporter for PptxA, the ptxA promoter (position -5 to -220

489 upstream of ptxA coding sequence) was amplified by PCR from B. pertussis Tohama I. To

490 construct a GFP transcriptional reporter for PfhaB, the fhaB promoter (position -5 to -254

491 upstream of fhaB coding sequence) was amplified by PCR from B. bronchiseptica RB50. PCR

492 products were then digested and cloned into the pUCgfpMAB plasmid using EcoRI and HindIII

493 restriction sites. The resulting plasmids were subsequently sequenced to verify that each

494 promoter was correctly fused 5’ to the S12 RBS and gfp. Each reporter plasmid was then

C 495 introduced into WT, BvgS , ∆bvgS, BvgSEc, H643A, C607A, and C607S strains using tri-parental

496 mating with conjugative RHO3. A strain of RB50 was also engineered to contain the

497 pUCgfpMAB plasmid without a promoter driving gfp expression as a control

498 (RB50::pUCgfpMAB-EV). Reporter delivery to the Tn7 site was confirmed by PCR in each strain.

499

500 Transcription of PptxA-gfp and PfhaB-gfp Reporters In Vitro

501

+ 502 To analyze expression of the PptxA-gfp and PfhaB-gfp transcriptional reporters under Bvg phase

503 conditions, strains were grown on BG-blood agar supplemented with kanamycin at 37 °C until

504 colonies formed. Approximately 3-4 colonies were used to inoculate liquid SS media containing bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

505 streptomycin and kanamycin, with or without 4mM nicotinic acid (NA) or 50mM MgSO4,

506 depending on the experimental conditions tested. All cultures were incubated overnight for ~18

507 h at 37 °C on a roller. Each overnight culture was then sub-cultured to an OD of 0.3 per milliliter

508 into fresh liquid SS media containing streptomycin and kanamycin, with or without 4mM

509 nicotinic acid (NA) or 50mM MgSO4, depending on the experimental conditions tested.

510 Subcultures were incubated for ~4 h at 37°C while rotating. Subsequently, one-hundred

511 microliters of each culture were transferred to a well of a 96-well, flat bottom, polystyrene

512 plate (Greiner Bio-One CellStar, Cat#: 655180), and the OD600 and RFU at excitation A485 and

513 emission A535 (to detect GFP) were measured on a Perkin-Elmer Victor3 1420 Multi-label plate

514 reader. Total GFP expression was normalized by initial subtraction of background fluorescence,

515 as determined from the RB50::pUCgfpMAB-EV strain, and normalization to the OD600 for each

516 sample. Cultures were prepared and tested in duplicate and three experimental replicates were

517 completed. Normalized GFP/OD600 values were averaged across all replicates.

518

519 Characterization of pGFLIP-PflaA and pBam Reporters In Vitro

520

521 Strains containing pGFLIP-PflaA and pBam reporters were grown overnight in SS medium with

522 streptomycin and gentamycin, with and without 50 mM MgSO4 or 4 mM nicotinic acid. After

523 incubation at 37 °C for 24 h on a rotating incubator, an aliquot of cells from each experimental

524 condition was diluted to a concentration of 3 x 103 cells/mL with 1X PBS and plated on BG-

525 blood agar in duplicate. Plates were incubated for 72 h at 37 °C. White light and GFP

526 fluorescence images of each plate were captured using a G:Box Imaging System (Syngene). bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

527 Total cfu, GFP positive colonies, and LCP colonies were enumerated using the colony analysis

528 protocol of the GeneTools Software package (Syngene). The total number of GFP negative

529 colonies was calculated by subtracting the number of GFP positive colonies from the total

530 number of colonies isolated. The percent GFP negative colonies and percent LCP colonies were

531 calculated based on the total number of colonies isolated. Each in vitro experimental condition

532 was tested twice.

533

534 Evaluation of B. bronchiseptica Modulation In Vivo

535

536 Bacteria were cultured overnight at 37 °C in SS media with streptomycin, kanamycin, and

537 gentamicin to maintain bacteria in the Bvg+ phase (42). Six-week-old female BALB/c mice from

538 Jackson Laboratories were inoculated intranasally with 7.5 × 104 cfu of B. bronchiseptica strains

539 (containing the pGFLIP-PflaA and pBam reporters) in 50 μL of 1 X DPBS. At each time point

540 reported, right lung lobes and nasal cavity tissue were harvested and homogenized in 1 X DPBS

541 using a mini-beadbeater. Dilutions of the tissue homogenates were then plated on BG-blood

542 agar and grown at 37C for 72 h. White light (to enumerate total cfu and LCPs) and blue light (to

543 enumerate GFP+ colonies) images of each plate were captured using a G:Box Imaging System

544 (Syngene). Total cfu, GFP positive colonies, and LCP colonies were enumerated using the colony

545 analysis protocol of the GeneTools Software package (Syngene). The total number of GFP

546 negative colonies was calculated by subtracting the number of GFP positive colonies from the

547 total number of colonies isolated. The percent GFP negative colonies and percent LCP colonies bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

548 were calculated by determining the ratio of GFP negative colonies or LCP colonies to the total

549 number of colonies isolated.

550

551 Ethics Statement

552

553 This study was carried out in strict accordance with the recommendations in the Guide for the

554 Care and Use of Laboratory Animals of the National Institutes of Health (44). All animals were

555 properly anesthetized for inoculations, monitored regularly, and euthanized when moribund.

556 Efforts were made to minimize suffering. Our protocol was approved by the University of North

557 Carolina Institutional Animal Care and Use Committee (Protocol ID: 16-246).

558

559 Statistical Analysis

560

561 Statistical analysis was performed using Prism 8.0.1 software from GraphPad Software.

562 Statistical significance was determined using a Two-Way ANOVA and Tukey’s multiple

563 comparisons test (gfp transcriptional assays) or a Kruskal-Wallis test and Dunn’s multiple

564 comparisons test (in vitro and in vivo modulation assay using pGFLIP-PflaA and pBam). P < 0.05

565 was considered significant. Figures were generated using Adobe Illustrator CS6 (Adobe

566 Systems).

567

568 Acknowledgments

569 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

570 We thank members of the Cotter lab for critical discussions and technical assistance. We thank

571 Françoise Jacob-Dubuisson for providing the plasmid used in making the BvgSEc strain. This work

572 was supported by NIH grant R01 R01AI129541 to P.A.C.

573

574 Author Contributions

575

576 M.A.B. and P.A.C. designed research; M.A.B. performed research; M.A.B. and P.A.C. analyzed

577 data; and M.A.B. and P.A.C. wrote the paper.

578

579 The authors declare no conflict of interest.

580

581 References

582

583 1. Cherry JD. 2012. Epidemic Pertussis in 2012 — The Resurgence of a Vaccine-Preventable

584 Disease. New Engl J Medicine 367:785–787.

585

586 2. Kilgore PE, Salim AM, Zervos MJ, Schmitt H-J. 2016. Pertussis: Microbiology, Disease,

587 Treatment, and Prevention. Clin Microbiol Rev 29:449–486.

588

589 3. Goodnow R. 1980. Biology of Bordetella bronchiseptica. Microbiol Rev 44:722–38.

590

591 4. Cotter PA, Miller JF. 1997. A mutation in the Bordetella bronchiseptica bvgS gene results in bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

592 reduced virulence and increased resistance to starvation, and identifies a new class of Bvg‐

593 regulated antigens. Mol Microbiol 24:671–685.

594

595 5. de Tejada G, Miller JF, Cotter PA. 1996. Comparative analysis of the virulence control systems

596 of Bordetella pertussis and Bordetella bronchiseptica. Mol Microbiol 22:895–908.

597

598 6. Cotter P, Miller J. 1994. BvgAS-mediated signal transduction: analysis of phase-locked

599 regulatory mutants of Bordetella bronchiseptica in a rabbit model. Infect Immun 62:3381–90.

600

601 7. Taylor-Mulneix DL, Bendor L, Linz B, Rivera I, Ryman VE, Dewan KK, Wagner SM, Wilson EF,

602 Hilburger LJ, Cuff LE, West CM, Harvill ET. 2017. Bordetella bronchiseptica exploits the complex

603 life cycle of Dictyostelium discoideum as an amplifying transmission vector. Plos Biol

604 15:e2000420.

605

606 8. Vergara-Irigaray N, Chávarri-Martínez A, Rodríguez-Cuesta J, Miller JF, Cotter PA, de Tejada G.

607 2005. Evaluation of the Role of the Bvg Intermediate Phase in Bordetella pertussis during

608 Experimental Respiratory Infection. Infect Immun 73:748–760.

609

610 9. Uhl M, Miller J. 1994. Autophosphorylation and phosphotransfer in the Bordetella pertussis

611 BvgAS signal transduction cascade. Proc National Acad Sci 91:1163–1167.

612

613 10. Jones AM, Boucher PE, Williams CL, Stibitz S, Cotter PA. 2005. Role of BvgA phosphorylation bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

614 and DNA binding affinity in control of Bvg‐mediated phenotypic phase transition in Bordetella

615 pertussis. Mol Microbiol 58:700–713.

616

617 11. Moon K, Bonocora RP, Kim DD, Chen Q, Wade JT, Stibitz S, Hinton DM. 2017. The BvgAS

618 Regulon of Bordetella pertussis. Mbio 8:e01526-17.

619

620 12. Akerley B, Monack D, Falkow S, Miller J. 1992. The bvgAS locus negatively controls motility

621 and synthesis of flagella in Bordetella bronchiseptica. J Bacteriol 174:980–990.

622

623 13. Williams CL, Boucher PE, Stibitz S, Cotter PA. 2005. BvgA functions as both an activator and

624 a repressor to control Bvgi phase expression of bipA in Bordetella pertussis. Mol Microbiol

625 56:175–188.

626

627 14. Zu T, Manetti R, Rappuoli R, Scarlato V. 1996. Differential binding of BvgA to two classes of

628 virulence genes of Bordetella pertussis directs promoter selectivity by RNA polymerase. Mol

629 Microbiol 21:557–565.

630

631 15. Dupré E, Herrou J, Lensink MF, Wintjens R, Vagin A, Lebedev A, Crosson S, Villeret V, Locht

632 C, Antoine R, Jacob-Dubuisson F. 2015. Virulence Regulation with Venus Flytrap Domains:

633 Structure and Function of the Periplasmic Moiety of the Sensor-Kinase BvgS. Plos Pathog

634 11:e1004700.

635 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

636 16. Herrou J, Bompard C, Wintjens R, Dupré E, Willery E, Villeret V, Locht C, Antoine R, Jacob-

637 Dubuisson F. 2010. Periplasmic domain of the sensor-kinase BvgS reveals a new paradigm for

638 the Venus flytrap mechanism. Proc National Acad Sci 107:17351–17355.

639

640 17. Dupré E, Lesne E, Guérin J, Lensink MF, Verger A, de Ruyck J, Brysbaert G, Vezin H, Locht C,

641 Antoine R, Jacob-Dubuisson F. 2015. Signal Transduction by BvgS Sensor Kinase BINDING OF

642 MODULATOR NICOTINATE AFFECTS THE CONFORMATION AND DYNAMICS OF THE ENTIRE

643 PERIPLASMIC MOIETY. J Biol Chem 290:23307–23319.

644

645 18. Möglich A, Ayers RA, Moffat K. 2009. Structure and Signaling Mechanism of Per-ARNT-Sim

646 Domains. Structure 17:1282–1294.

647

648 19. Henry JT, Crosson S. 2011. Ligand-Binding PAS Domains in a Genomic, Cellular, and

649 Structural Context. Annu Rev Microbiol 65:261–286.

650

651 20. Gilles-Gonzalez M-A, Gonzalez G. 2004. Signal transduction by heme-containing PAS-domain

652 proteins. J Appl Physiology Bethesda Md 1985 96:774–83.

653

654 21. Taylor B, Zhulin I. 1999. PAS domains: internal sensors of oxygen, redox potential, and light.

655 Microbiol Mol Biology Rev Mmbr 63:479–506.

656

657 22. Gu Y-Z, Hogenesch JB, Bradfield CA. 2000. The PAS Superfamily: Sensors of Environmental bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

658 and Developmental Signals. Annu Rev Pharmacol 40:519–561.

659

660 23. Bock A, Gross R. 2002. The unorthodox histidine kinases BvgS and EvgS are responsive to

661 the oxidation status of a quinone electron carrier. Eur J Biochem 269:3479–3484.

662

663 24. Antelmann H, Helmann JD. 2011. Thiol-Based Redox Switches and Gene Regulation.

664 Antioxid Redox Sign 14:1049–1063.

665

666 25. Dupré E, Wohlkonig A, Herrou J, Locht C, Jacob-Dubuisson F, Antoine R. 2013.

667 Characterization of the PAS domain in the sensor-kinase BvgS: mechanical role in signal

668 transmission. Bmc Microbiol 13:172.

669

670 26. Lesne E, Krammer E-M, Dupre E, Locht C, Lensink M, Antoine R, Jacob-Dubuisson F. 2016.

671 Balance between Coiled-Coil Stability and Dynamics Regulates Activity of BvgS Sensor Kinase in

672 Bordetella. Mbio 7:e02089-15.

673

674 27. Julio SM, Inatsuka CS, Mazar J, Dieterich C, Relman DA, Cotter PA. 2009. Natural‐host animal

675 models indicate functional interchangeability between the filamentous haemagglutinins of

676 Bordetella pertussis and Bordetella bronchiseptica and reveal a role for the mature C‐terminal

677 domain, but not the RGD motif, during infection. Mol Microbiol 71:1574–1590.

678

679 28. Cotter P, Yuk M, Mattoo S, Akerley B, Boschwitz J, Relman D, Miller J. 1998. Filamentous bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

680 hemagglutinin of Bordetella bronchiseptica is required for efficient establishment of tracheal

681 colonization. Infect Immun 66:5921–9.

682

683 29. Bone AM, Wilk AJ, Perault AI, Marlatt SA, Scheller EV, Anthouard R, Chen Q, Stibitz S, Cotter

684 PA, Julio SM. 2017. Bordetella PlrSR regulatory system controls BvgAS activity and virulence in

685 the lower respiratory tract. Proc National Acad Sci 114:E1519–E1527.

686

687 30. Atack JM, khanta YN, Djoko KY, Welch JP, Hasri NH, Steichen CT, Hoven RN, Grimmond SM,

688 Othman D, Kappler U, Apicella MA, Jennings MP, Edwards JL, McEwan AG. 2013.

689 Characterization of an ntrX Mutant of Neisseria gonorrhoeae Reveals a Response Regulator

690 That Controls Expression of Respiratory Enzymes in Oxidase-Positive Proteobacteria. J Bacteriol

691 195:2632–2641.

692

693 31. del Carrica M, Fernandez I, Sieira R, Paris G, Goldbaum F. 2013. The two‐component

694 systems PrrBA and NtrYX co‐ordinately regulate the adaptation of Brucella abortus to an

695 oxygen‐limited environment. Mol Microbiol 88:222–233.

696

697 32. Decker KB, James TD, Stibitz S, Hinton DM. 2012. The Bordetella pertussis model of

698 exquisite gene control by the global transcription factor BvgA. Microbiology+ 158:1665–1676.

699

700 33. Kinnear SM, Marques RR, Carbonetti NH. 2001. Differential Regulation of Bvg-Activated

701 Virulence Factors Plays a Role in Bordetella pertussisPathogenicity. Infect Immun 69:1983– bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

702 1993.

703

704 34. Byrd MS, Mason E, Henderson MW, Scheller EV, Cotter PA. 2013. An Improved

705 Recombination-Based In Vivo Expression Technology-Like Reporter System Reveals Differential

706 cyaA Gene Activation in Bordetella Species. Infect Immun 81:1295–1305.

707

708 35. Mason E, Henderson MW, Scheller EV, Byrd MS, Cotter PA. 2013. Evidence for phenotypic

709 bistability resulting from transcriptional interference of bvgAS in Bordetella bronchiseptica. Mol

710 Microbiol 90:716–733.

711

712 36. Akerley BJ, Cotter PA, Miller JF. 1995. Ectopic expression of the flagellar regulon alters

713 development of the bordetella-host interaction. Cell 80:611–620.

714

715 37. icó, Miller J, Roy C, Stibitz S, Monack D, Falkow S, Gross R, Rappuoli R. 1989. Sequences

716 required for expression of Bordetella pertussis virulence factors share homology with

717 prokaryotic signal transduction proteins. Proc National Acad Sci 86:6671–6675.

718

719 38. Arico B, Scarlato V, Monack D, Falkow S, Rappuoli R. 1991. Structural and genetic analysis of

720 the bvg locus in Bordetella species. Mol Microbiol 5:2481–2491.

721

722 39. Kaut CS, Duncan MD, Kim J, Maclaren JJ, Cochran KT, Julio SM. 2011. A Novel Sensor Kinase

723 Is Required for Bordetella bronchiseptica To Colonize the Lower Respiratory Tract. Infect bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

724 Immun 79:3216–3228.

725

726 40. de Bagüés M, Loisel-Meyer S, Liautard J-P, Jubier-Maurin V. 2007. Different Roles of the Two

727 High-Oxygen-Affinity Terminal Oxidases of Brucella suis: Cytochrome c Oxidase, but Not

728 Ubiquinol Oxidase, Is Required for Persistence in Mice▿. Infect Immun 75:531–535.

729

730 41. Herrou J, Debrie A-S, Willery E, Renaud-Mongénie G, Locht C, Mooi F, Jacob-Dubuisson F,

731 Antoine R. 2009. Molecular Evolution of the Two-Component System BvgAS Involved in

732 Virulence Regulation in Bordetella. Plos One 4:e6996.

733

734 42. Hulbert RR, Cotter PA. 2009. Current Protocols in Microbiology. Curr Protoc Microbiol

735 Chapter 4:4B.1.1-4B.1.9.

736

737 43. Joseph Sambrook, David W. Russell. 2001. Molecular cloning: a laboratory manual. 3rd Ed,

738 Volume 1-3, ISBN: 9780879695774.

739

740 44. Committee on Care and Use of Laboratory Animals. 1996. Guide for the Care and Use

741 of Laboratory Animals (NIH, Bethesda), DHHS Publ No (NIH) 85-23.

742

743 45. Anderson MS, Garcia EC, Cotter PA. 2012. The Burkholderia bcpAIOB genes define

744 unique classes of two-partner secretion and contact dependent growth inhibition

745 systems. PLoS Genet 8(8):e1002877. bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

746

747 Figure Legends

748

749 Figure 1. The sensor kinase, BvgS, has a unique PAS domain, which may play a role in signal

750 perception

751 (A) Schematic of the two-component phosphor-relay system BvgAS. BvgS is an unorthodox His-

752 Asp-His sensor kinase, comprised of tandem, periplasmic, venus fly-trap (VFT) domains

753 followed by a transmembrane (TM) domain, a cytoplasmic Per/Arnt/Sim (PAS) domain, a

754 histidine kinase (HK) domain, a receiver (REC) domain, and a histidine phosphotransfer (Hpt)

755 domain. BvgA is a classical DNA-binding response regulator, comprised of a receiver (REC)

756 domain and a helix-turn-helix (HTH) DNA-binding domain. When active, BvgS acts as a kinase

757 and phosphorylates BvgA. The phosphorylated form of BvgA (BvgA~P) can bind DNA and

758 activate or repress gene expression. When inactive, BvgS dephosphorylates and inactivates

759 BvgA. (B) Amino acid sequence of the BvgS PAS domain (highlighted in yellow) and flanking

760 domains (TM domain highlighted in grey and the HK domain highlighted in green). Red arrows

761 highlight the highly conserved cysteine (C607) and histidine (His643) residues of the BvgS PAS

762 domain that were targeted for mutagenesis. Grey numbers correspond to amino acid position

763 in BvgS. (C) Schematic of the amino acid sequence of the alpha-helical linker (underlined in red)

764 that was used to replace the BvgS PAS domain in the strain BvgSEc. The amino acid sequence of

765 the N-terminal TM domain is highlighted in grey and the C-terminal HK domain is highlighted in

766 green. Grey number corresponds to amino acid position in BvgS.

767 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

768

769 Figure 2. In B. bronchiseptica, the PAS domain influences BvgS kinase activity and sensitivity

770 to chemical modulation in vitro

C 771 Transcriptional analysis of PptxA and PfhaB activity in wild-type (WT), BvgS , ∆bvgS, BvgSEc, H643A,

772 C607A, and C607S strains of B. bronchiseptica using the gfp based reporters. All strains were

773 cultured in Stainer Scholte (SS) broth (Bvg+ growth conditions, black bars), in SS with 4mM

– – 774 nicotinic acid (Bvg growth conditions, blue bars), or in SS with 50mM MgSO4 (Bvg growth

775 conditions, orange bars) at 37 °C with agitation. (A) Activity of the high-affinity BvgA~P PfhaB

776 promoter was determined using the PfhaB-gfp reporter and (B) activity of the low-affinity

777 BvgA~P PptxA promoter was determined using the PptxA-gfp reporter. Colored bars represent the

778 mean normalized GFP fluorescence per cell (GFP/OD600) calculated from triplicate experiments.

779 Error bars represent the standard deviation of normalized GFP/OD600 for each data set.

780 Statistical significance (Two-Way ANOVA and Tukey’s multiple comparisons test, P < 0.05) of

781 values for each strain compared to WT under each growth condition is indicated with * (P <

782 0.05) and statistically significance comparisons of values for a single strain cultured in each

783 growth condition is indicated with a specific P value.

784

785 Figure 3. The PAS domain is important for BvgS inactivation in response to known modulatory

786 chemicals in vitro

787 In vitro analysis of wild-type (WT), BvgSEc, H643A, C607A, and C607S modulation in response to

788 the chemical signals nicotinic acid (NA) and magnesium sulfate (MgSO4). All strains contain the

+ 789 plasmid reporters pGFLIP-PflaA and pBam. Strains were cultured under Bvg (SS broth) and bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

– 790 under two different Bvg phase growth conditions (SS broth with 4 mM NA or 50 mM MgSO4) at

791 37 °C for ∼18 h with agitation and subsequently plated on BG-blood solid agar to enumerate the

792 percentage of GFP− colonies (A and C) and LCPs (B and D). NA was used as the modulatory

793 stimulus in (A) and (B). MgSO4 was used as the modulatory stimulus in (C) and (D). GFP

794 negativity indicates bacterial modulation and loss of BvgS kinase activity. LCPs indicate bacteria

795 present in the Bvg– phase when plated. Statistical significance (Kruskal-Wallis test and Dunn’s

796 multiple comparisons test, P < 0.05) is indicated with * (P < 0.05), ** (P < 0.01), and **** (P <

797 0.0001).

798

799 Figure 4. A conserved cysteine residue in the PAS domain is important for maintaining BvgS

800 kinase activity during respiratory tract colonization of mice

801 (A) Colonization of the nasal cavity (top) and right lung (bottom) tissue in female BALB/cJ mice

802 by wild type (WT; black circles), BvgSEc (blue squares), H643A (green hexagons), C607A (yellow

803 diamonds), and C607S (purple triangles) strains on days 0, 1, and 3 post-inoculation. All strains

4 804 contain the plasmid reporters, pGFLIP-PflaA and pBam. Mice were inoculated with 7.5 x10 cfu

805 via the external nares. Each symbol represents a single animal, with the mean colonization

806 depicted as short horizontal bars. (B) Percentage of GFP negative (% GFP–) bacteria recovered

807 from the nasal cavity (top) and right lung (bottom). GFP negativity indicates bacterial

808 modulation and loss of BvgS kinase activity. (C) Percentage of large colony phenotype (LCP)

809 producing bacteria recovered from the nasal cavity (top) and right lung (bottom). LCPs indicate

810 bacteria present in the Bvg– phase when recovered. Statistical significance (Kruskal-Wallis test bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

811 and Dunn’s multiple comparisons test, P < 0.05) is indicated with * (P < 0.05), ** (P < 0.01), and

812 **** (P < 0.0001).

813

814 Figure 5. The PAS domain is required for BvgS modulation in response to PlrSR dysfunction

815 during LRT colonization

816 (A) Colonization of the nasal cavity (top) and right lung (bottom) tissue in female BALB/cJ mice

817 by wild type (WT; black circles), ∆plrS (red triangles), ∆plrS-BvgSEc (blue squares), ∆plrS-H643A

818 (green hexagons), ∆plrS-C607A (yellow diamonds), and ∆plrS-C607S (purple triangles) strains on

819 days 0, 1, and 3 post-inoculation. All strains contain the plasmid reporters, pGFLIP-PflaA and

820 pBam. Mice were inoculated with 7.5 x104 cfu via the external nares. Each symbol represents a

821 single animal, with the mean colonization depicted as short horizontal bars. Homogenate from

822 each organ was used to determine bacterial burden shown in graph A and in vivo bacterial

823 modulation shown in graphs B and C. (B) Percentage of GFP negative (% GFP–) bacteria

824 recovered from the nasal cavity (top) and right lung (bottom). GFP negativity indicates bacterial

825 modulation and loss of BvgS kinase activity. (C) Percentage of large colony phenotype (LCP)

826 producing bacteria recovered from the nasal cavity (top) and right lung (bottom). LCPs indicate

827 bacteria present in the Bvg– phase when recovered. Statistical significance (Kruskal-Wallis test

828 and Dunn’s multiple comparisons test, P < 0.05) is indicated with * (P < 0.05), ** (P < 0.01), and

829 *** (P < 0.001).

830

831 Supplemental Figure 1. The PAS domain of BvgS may sense an activating or inactivating signal

832 in the LRT. bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

833 (A) Schematic depicting how the PAS domain may be required to sense a PlrS-dependent

834 activating signal that is required to maintain BvgS kinase activity in the LRT. Genes encoding a

835 protein(s) required for the generation of an activating signal are regulated by the PlrSR TCS. The

836 BvgS PAS domain is required for sensing this activating this signal and BvgS can function as a

837 kinase. (B) Without PlrS, generation of the activating signal does not occur and BvgS would be

838 off. (C) Schematic depicting how the PAS domain may sense a signal that inactivates BvgS in the

839 LRT. Normally, the activity of a protein(s) encoded by a gene(s) regulated by the PlrSR system

840 would inhibit the generation of an inactivating signal. (D) However, without PlrS this

841 inactivating signal can accumulate and inhibit BvgS kinase activity via the PAS domain of the

842 protein.

843 bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

B. TM Domain A. Om 559- ... WIVYL RRQIRQRKRAERALNDQLEF

VFT1 VFT1 VFT2 VFT2 PAS Domain 584- MRVLIDGT PNPIYVRDKEGRMLLCND Cm TM TM C607 PAS PAS BvgS 610- AYLDTFGVTADAVLGKTIPEANVVGDP P Phosphatase HK HK H Kinase Activity and 637- H O ALAREMHEFLLTRMSAEREPRFEDRD 2 P D Rec Rec D P Phosphotransfer Activity H643 P H Hpt Hpt H P 663- VTLHGRTRHVYQWTVPYGDSLGELKG

P Rec Rec P BvgA HTH HTH 689- IIGGWIDIT ERAELLRELHDAKESADAA

HK Domain 717- NR AKTTFLA ... vrg vag

C. TM Domain 559- ... WIVYL RRQIVRKAALRRQLNEQLTQ

LREMLTERESLLAELSEAKDRAEDS

HK Domain NR AKTTFLA ...

Figure 1. The sensor kinase, BvgS, has a unique PAS domain, which may play a role in signal perception

(A) Schematic of the two-component phosphor-relay system BvgAS. BvgS is an unorthodox His-Asp-His sensor kinase, comprised of tandem, periplasmic, venus fly-trap (VFT) domains followed by a transmembrane (TM) domain, a cytoplasmic Per/Arnt/Sim (PAS) domain, a histidine kinase (HK) domain, a receiver (REC) domain, and a histidine phosphotransfer (Hpt) domain. BvgA is a classical DNA-binding response regulator, comprised of a receiver (REC) domain and a helix-turn-helix (HTH) DNA-binding domain. When active, BvgS acts as a kinase and phosphorylates BvgA. The phosphorylated form of BvgA (BvgA~P) can bind DNA and activate or repress gene expression. When inactive, BvgS dephosphorylates and inactivates BvgA. (B) Amino acid sequence of the BvgS PAS domain (highlighted in yellow) and flanking domains (TM domain highlighted in grey and the HK domain highlighted in green). Red arrows highlight the highly conserved cysteine (C607) and histidine (His643) residues of the BvgS PAS domain that were targeted for mutagenesis. Grey numbers corre- spond to amino acid position in BvgS. (C) Schematic of the amino acid sequence of the alpha-helical linker (underlined in red) that was used to replace the BvgS PAS domain in the strain BvgSEc. The amino acid sequence of the N-terminal TM domain is highlighted in grey and the C-terminal HK domain is highlighted in green. Grey number corresponds to amino acid position in BvgS. bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

A. SS Broth SS + Nicotinic Acid PfhaB-gfp : SS + MgSO4 20 x 105 P = 0.0004 P < 0.0001

P < 0.0001 * P < 0.0001 P = 0.0004 P < 0.0001 P < 0.0001 P < 0.0001 P < 0.0001 x 5 P < 0.0001 P < 0.0001 15 10 * * P < 0.0001 * * * * 10 x 105 *

5 x 105 Normalized GFP/OD600 Normalized * 0 nd nd nd C WT BvgS ∆bvgS BvgEc H643A C607A C607S B. PptxA-gfp : 5 x 105 P < 0.0001 P < 0.0001

* P < 0.0001 P < 0.0001 * * P < 0.0001

* P < 0.0001 5

x *

4 10 *

P < 0.0001 x 5 3 10 P < 0.0001

2 x 105 * P < 0.0001

* P < 0.0001 5 *

1 x 10 * Normalized GFP/OD600 Normalized * 0 C WT BvgS ∆bvgS BvgEc H643A C607A C607S

Figure 2. In B. bronchiseptica, the PAS domain influences BvgS kinase activity and sensitivity to chemical modulation in vitro

C Transcriptional analysis of PptxA and PfhaB activity in wild-type (WT), BvgS , ∆bvgS, BvgSEc, H643A, C607A, and C607S strains of B. bronchiseptica using the gfp based reporters. All strains were cultured in Stainer Scholte (SS) broth (Bvg+ growth conditions, black bars), in – SS with 4mM nicotinic acid (Bvg growth conditions, blue bars), or in SS with 50mM MgSO4 (Bvg– growth conditions, orange bars) at 37 °C with agitation. (A) Activity of the high-affinity BvgA~P PfhaB promoter was determined using the PfhaB-gfp reporter and (B) activity of the low-affinity BvgA~P PptxA promoter was determined using the PptxA-gfp reporter. Colored bars represent the mean normalized GFP fluorescence per cell (GFP/OD600) calculated from triplicate experiments. Error bars represent the standard deviation of normalized GFP/OD600 for each data set. Statistical significance (Two-Way ANOVA and Tukey’s multi- ple comparisons test, P < 0.05) of values for each strain compared to WT under each growth condition is indicated with * (P < 0.05) and statistically significance comparisons of values for a single strain cultured in each growth condition is indicated with a specific P value. bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

A. B. ** ** 100 60

80 50 40 60

% LCP 30 40 20

% GFP Negative% GFP 20 10 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0.3% 0 0 Nicotinic Acid: – +–––– + + + + Nicotinic Acid: – +–––– + + + +

WT BvgSEc H643A C607A C607S WT BvgSEc H643A C607A C607S

C. D. * **** 100 10

80 8

60 6

% LCP 40 4

% GFP Negative% GFP 20 2

0.2% 0% 0% 0.2% 0.2% 0.8% 0% 0% 0% 0% 0 0 MgSO4: – +–––– + + + + MgSO4: – +–––– + + + + WT BvgSEc H643A C607A C607S WT BvgSEc H643A C607A C607S

Figure 3. The PAS domain is important for BvgS inactivation in response to known modulatory chemicals in vitro

In vitro analysis of wild-type (WT), BvgSEc, H643A, C607A, and C607S modulation in response to the chemical signals nicotinic acid (NA) and magnesium sulfate (MgSO4). All strains contain the plasmid reporters pGFLIP-PflaA and pBam. Strains were cultured under Bvg+ (SS broth) and under two different Bvg– phase growth conditions (SS broth with 4 mM NA or 50 mM MgSO4) 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 C) and LCPs (B and D). NA was used as the modulatory stimulus in (A) and (B). MgSO4 was used as the modulatory stimulus in (C) and (D). GFP negativity indicates bacterial modulation and loss of BvgS kinase activity. LCPs indicate bacteria present in the Bvg– phase when plated. Statistical significance (Kruskal-Wallis test and Dunn’s multiple comparisons test, P < 0.05) is indicated with * (P < 0.05), ** (P < 0.01), and **** (P < 0.0001). bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

A. Day 0 Day 1 Day 3 7 6 5 4 3 2 1 Log10 CFU/ Nasal Cavity 7 * **** 6 5 4 3 2 1 Log10 CFU/ Right Lung Ec Ec Ec WT WT WT BvgS BvgS BvgS H643A C607A H643A C607A H643A C607A C607S C607S C607S

B. C. Day 0 Day 1 Day 3 Day 0 Day 1 Day 3 10 100 * * * * ** * 80 8

60 6 / Nasal Cavity – 40 4

2 20 / Nasal Cavity % LCP % GFP % GFP 0 0 10 100 * * * 80 8

60 6 / Right Lung

– 40 4

20 / Right Lung % LCP 2 % GFP % GFP 0 0 Ec Ec Ec Ec Ec Ec WT WT WT WT WT WT BvgS BvgS H643A C607A BvgS H643A C607A BvgS H643A C607A C607S C607S C607S BvgS H643A C607A H643A C607A BvgS H643A C607A C607S C607S C607S

Figure 4. A conserved cysteine residue in the PAS domain is important for maintaining BvgS kinase activity during respiratory tract colonization of mice

(A) Colonization of the nasal cavity (top) and right lung (bottom) tissue in female BALB/cJ mice by wild type (WT; black circles), BvgSEc (blue squares), H643A (green hexagons), C607A (yellow diamonds), and C607S (purple triangles) strains on days 0, 1, and 3 post-in- oculation. All strains contain the plasmid reporters, pGFLIP-PflaA and pBam. Mice were inocu- lated with 7.5 x104 cfu via the external nares. Each symbol represents a single animal, with the mean colonization depicted as short horizontal bars. (B) Percentage of GFP negative (% GFP–) bacteria recovered from the nasal cavity (top) and right lung (bottom). GFP negativity indicates bacterial modulation and loss of BvgS kinase activity. (C) Percentage of large colony phenotype (LCP) producing bacteria recovered from the nasal cavity (top) and right lung (bottom). LCPs indicate bacteria present in the Bvg– phase when recovered. Statistical significance (Kruskal-Wallis test and Dunn’s multiple comparisons test, P < 0.05) is indicated with * (P < 0.05), ** (P < 0.01), and **** (P < 0.0001). bioRxiv preprint doi: https://doi.org/10.1101/614891; this version posted April 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

A. Day 0 Day 1 Day 3 7 6 5 4 3 2 1 Log10 CFU/ Nasal Cavity

7 6 * 5 4 3 2 1 Log10 CFU/ Right Lung Ec Ec Ec WT WT WT ∆ plrS ∆ plrS ∆ plrS H643A C607A C607S H643A C607A C607S H643A C607A C607S BvgS BvgS BvgS

∆plrS ∆plrS ∆plrS B. C. Day 0 Day 1 Day 3 Day 0 Day 1 Day 3 100 10 ** * 80 * 8

60 6 / Nasal Cavity – 40 4

20 / Nasal Cavity % LCP 2 % GFP % GFP

0 0 * * * ** ** *** * ** ** 100 *** 100 * ** 80 80 *

60 60 / Right Lung – 40 40

20 / Right Lung % LCP 20 % GFP % GFP

0 0 Ec Ec Ec Ec Ec Ec WT WT WT WT WT WT ∆ plrS ∆ plrS ∆ plrS ∆ plrS ∆ plrS ∆ plrS BvgS BvgS H643A C607A C607S H643A C607A C607S H643A C607A C607S BvgS BvgS BvgS H643A C607A C607S H643A C607A C607S H643A C607A C607S BvgS

∆plrS ∆plrS ∆plrS ∆plrS ∆plrS ∆plrS

Figure 5. The PAS domain is required for BvgS modulation in response to PlrSR dys- function during LRT colonization

(A) Colonization of the nasal cavity (top) and right lung (bottom) tissue in female BALB/cJ mice by wild type (WT; black circles), ∆plrS (red triangles), ∆plrS-BvgSEc (blue squares), ∆ plrS-H643A (green hexagons), ∆plrS-C607A (yellow diamonds), and ∆plrS-C607S (purple triangles) strains on days 0, 1, and 3 post-inoculation. All strains contain the plasmid report- 4 ers, pGFLIP-PflaA and pBam. Mice were inoculated with 7.5 x10 cfu via the external nares. Each symbol represents a single animal, with the mean colonization depicted as short hori- zontal bars. Homogenate from each organ was used to determine bacterial burden shown in graph A and in vivo bacterial modulation shown in graphs B and C. (B) Percentage of GFP negative (% GFP–) bacteria recovered from the nasal cavity (top) and right lung (bottom). GFP negativity indicates bacterial modulation and loss of BvgS kinase activity. (C) Percent- age of large colony phenotype (LCP) producing bacteria recovered from the nasal cavity (top) and right lung (bottom). LCPs indicate bacteria present in the Bvg– phase when recov- ered. Statistical significance (Kruskal-Wallis test and Dunn’s multiple comparisons test, P < 0.05) is indicated with * (P < 0.05), ** (P < 0.01), and *** (P < 0.001).