Hybrid histidine kinase BinK represses Vibrio fischeri biofilm signaling at multiple developmental stages

The symbiosis between the Hawaiian bobtail squid, Euprymna scolopes, and its exclusive light-organ symbiont, Vibrio fischeri, provides a natural system in which to study host-microbe specificity and gene regulation during the establishment of a mutually-beneficial symbiosis. Colonization of the host relies on bacterial biofilm-like aggregation in the squid mucus field. Symbiotic biofilm formation is controlled by a two-component signaling (TCS) system consisting of regulators RscS-SypF-SypG, which together direct transcription of the Syp symbiotic polysaccharide. TCS systems are broadly important for bacteria to sense environmental cues and then direct changes in behavior. Previously, we identified hybrid histidine kinase BinK as a strong negative regulator of V. fischeri biofilm regulation, and here we further explore the function of BinK. To inhibit biofilm formation, BinK requires the predicted phosphorylation sites in both the histidine kinase (H362) and receiver (D794) domains. Furthermore, we show that RscS is not essential for host colonization when binK is deleted from strain ES114, and imaging of aggregate size revealed no benefit to the presence of RscS in a background lacking BinK. Strains lacking RscS still suffered in competition. Finally, we show that BinK functions to inhibit biofilm gene expression in the light organ crypts, providing evidence for biofilm gene regulation at later stages of host colonization. Overall, this study provides direct evidence for opposing activities of RscS and BinK and yields novel insights into biofilm regulation during the maturation of a beneficial symbiosis. IMPORTANCE Bacteria are often in a biofilm state, and transitions between planktonic and biofilm lifestyles are important for pathogenic, beneficial, and environmental microbes. The critical nature of biofilm formation during Vibrio fischeri colonization of the Hawaiian bobtail squid light organ provides an opportunity to study development of this process in vivo using a combination of genetic and imaging approaches. The current work refines the signaling circuitry of the biofilm pathway in V. fischeri, provides evidence that biofilm regulatory changes occur in the host, and identifies BinK as one of the regulators of that process. This study provides information about how bacteria regulate biofilm gene expression in an intact animal host.

organ provides an opportunity to study development of this process in vivo using a combination 59 of genetic and imaging approaches. The current work refines the signaling circuitry of the biofilm RscS-overexpressing allele (termed, rscS*) that approximates in vivo biofilm phenotypes to 113 study BinK function, and with those approaches we demonstrated that deletion of binK can lead 114 to wrinkled colony formation, higher transcription of the syp locus, and higher production of the 115 Syp exopolysaccharide (22). Overexpressing sypG is epistatic to inhibitory signaling from 116 overexpressing binK (22). The ΔbinK strain is significantly derepressed for biofilm formation and 117 in that background, calcium stimulates colony biofilm formation without the need for induction by 118 rscS* alleles (24). The calcium induction system led to the discovery of biofilm regulator HahK, 119 which influences biofilm development in response to host nitric oxide (24,25). 120 Given the prominence of BinK as a strong negative regulator of biofilm across various V. 121 fischeri natural isolates, here we pursued multiple questions regarding its function during 122 symbiosis. First, while BinK has the predicted structure of a hybrid histidine kinase, we asked 123 whether it requires its putative phosphorylation sites for function. Second, given the strong 124 phenotypes of BinK, we asked whether canonical squid isolate ES114 would be capable of 125 colonizing squid in the absence of the positive regulator RscS if the negative regulator BinK was 126 also removed. We found that RscS was dispensable for colonization in the absence of BinK,127 and this result enabled us to investigate the relative role of each protein during symbiotic 128 colonization. Third, using direct imaging of a fluorescent syp transcriptional reporter, we asked 129 whether symbiosis polysaccharide gene expression is regulated in the host by comparing 130 We identified BinK as an orphan hybrid histidine kinase that inhibits biofilm formation 138 and is a negative regulator of squid colonization (22). Protein  To assess the function of individual alleles, we conducted colony biofilm assays. We 147 started with a ΔbinK strain and then introduced the binK gene (including 300 bp of upstream 148 and downstream sequence) into the neutral chromosomal attTn7 site (Figure 2A). This 149 approach enabled us to test the wild-type and mutant alleles in comparable isogenic 150 backgrounds. The strain background also has an allele to induce biofilm formation under culture 151 conditions through the overexpression of RscS. This allele, termed rscS* is carried on the 152 chromosome at the native rscS locus (Figure 2A) (32). In this background, deletion of binK 153 results in a wrinkled colony morphology when grown at 28 o C, while a strain with a functional 154 binK has a smooth colony morphology (22). To test whether the His362 and/or Asp794 are 155 required for BinK function, we constructed H362Q and D794A mutations, which have been 156 shown to mimic the unphosphorylated state when similarly introduced into homologous domains 157 (16,33). In an otherwise ΔbinK background, mutation in either individual residue or in both 158 residues in the same protein resulted in a non-functional BinK ( Figure 2B). We next asked 159 whether a predicted phosphomimetic allele of the REC domain D794E is functional. This allele 160 was constructed and unable to complement the lack of BinK in the rscS* biofilm induction model 161 ( Figure 2B). Using western blot analysis with a polyclonal antibody raised against a BinK 162 cytoplasmic epitope, we demonstrated that the mutant proteins are expressed to levels 7 comparable to that of the wild-type BinK ( Figure 2C). Together, these results provide genetic 164 evidence that phosphorylation of BinK residues is required for its function. Given that we did not 165 observe complementation with either the D794A that is predicted to be non-phosphorylatable or 166 with the putative phosphomimetic D794E allele, we expect that phosphoryl groups at this 167 residue are transferred to (or from) a downstream signaling partner, and that this signaling is 168 required for BinK to inhibit biofilm formation. 169 170

BinK merodiploid analysis reveals a dominant negative phenotype for the D794A allele 171
Our data above demonstrated that the H362Q and D794A alleles were individually 172 nonfunctional. Given that histidine kinases typically operate as homodimers, we next inquired 173 whether the BinK(H362Q) or BinK(D794A) had any effect when expressed in a cell that also 174 expressed wild-type BinK protein. We continued to express the test alleles from the attTn7 site, 175 but now did so in a strain expressing wild-type binK from the native locus. In the wrinkled colony 176 assay, the nonfunctional H362Q allele was recessive to the wild-type allele as expected, but the 177 D794A allele exhibited a dominant negative phenotype, displaying a lack of biofilm inhibition 178 even in the presence of the wild-type allele ( Figure 2B). A similar dominance was observed for 179 the D794A allele even if the H362Q mutation was introduced on the same polypeptide, 180 expressed from the attTn7 site ( Figure 2B). While either the H362Q or the D794A mutation 181 disable BinK's biofilm inhibitory function, the latter allele additionally interferes with the ability of 182 a wild-type copy of BinK to inhibit biofilm formation. 183 Our data above demonstrated that the putative phosphomimetic D794E allele of BinK 184 was similarly nonfunctional as the D794A allele when expressed as the only BinK allele in the 185 cell. Merodiploid analysis revealed, however, that the D794E allele did not interfere with BinK 186 activity in trans ( Figure 2B). This was the case whether it was the only mutation in binK or in the 187 context of a double binK(H362Q, D794E) allele ( Figure 2B) Figure 1C). This region of the H-box (i.e., the conserved 197 phosphoryl group-binding His and surrounding region) is important for coordinating the 198 phosphotransfer reactions in two-component proteins (35). In BinK we constructed E363A, 199 T366Q, and T366A mutations that are predicted to eliminate kinase activity, reduce 200 phosphatase activity, and eliminate phosphatase activity (with possible effects on autokinase 201 activity), respectively (35)(36)(37)(38)(39)(40)(41). Colony biofilm assays revealed nonfunctional BinK in each case 202 ( Figure S1). This result further supports a role for phosphotransfer through BinK in its functional 203 formation in a ΔsypE sypF2 strain (18). We introduced the same amino acid change into 218 chromosomal sypG within the ΔsypE sypF2 genetic background. We proceeded to determine 219 that overexpression of BinK does not reduce the colony biofilm produced in the ΔsypE sypF2 220 sypG(D53E) background ( Figure S2). Therefore, this experiment supports and extends our 221 previous work and provides strong evidence that BinK acts upstream of SypG in the control of 222 syp transcription and biofilm development. We proceeded then to examine the relative impacts 223 of RscS and BinK on colonization phenotypes in vivo. 224

RscS is not required for aggregation or squid colonization in a strain lacking BinK 226
In strain ES114, the positive regulator RscS and the negative regulator BinK both exert 227 strong impacts on colonization and both act through the response regulator SypG. We therefore 228 considered models in which RscS and BinK exert opposing influence on syp gene transcription 229 during squid colonization. Deletion of rscS alone leads to severe in vivo biofilm and colonization 230 defects, whereas deletion of binK alone leads to enhanced biofilm production and improved 231 colonization in a competitive assay (10, 22, 43). Given these opposing phenotypes, we asked 232 whether removal of both regulators--RscS and BinK--would enable the bacteria to colonize the 233 host. An examination of experiments in diverse V. fischeri strains provides some insight into this 234 question. Strains can improve their ability to colonize squid in the laboratory by mutation of binK, 235 and this includes strain MJ11 that naturally lacks RscS (14, 22 To test these models, we conducted single-strain colonization assays. As has been 244 shown previously, a ΔrscS mutant exhibits a significant colonization defect (Figure 3) (43). The 245 ΔbinK mutant is known to exhibit a competitive advantage over wild-type (22), yet in single-246 strain colonization displays similar bacterial yields and luminescence (Figure 3). The ΔbinK 247 ΔrscS double mutant strain was able to colonize up to levels indistinguishable from the wild-type 248 strain (Figure 3). This result therefore supports the model that in ES114 BinK and RscS 249 antagonize each others' activity, and that in the absence of the negative regulator BinK, the 250 positive regulator RscS is no longer required for single-strain colonization. 251 The above result was surprising in that we identified a condition in which RscS, 252 discovered twenty years ago as a strong colonization factor (43), was no longer required for 253 squid colonization in strain ES114. This prompted us to ask whether the key symbiotic behavior 254 regulated by RscS--in vivo aggregate formation--occurs in the ΔbinK ΔrscS background. For this 255 experiment, we introduced a plasmid that constitutively expresses GFP into the colonizing 256 strains from Figure 3, and we asked whether these strains form biofilm aggregates in the squid 257 mucus field. Direct visualization of the bacterial cells revealed the presence of biofilm 258 aggregates in the host for the ΔbinK ΔrscS cells ( Figure 4A). Notably, the size of these 259 aggregates were comparable to those formed by ΔbinK single mutant cells, which were larger 260 than the aggregates formed by wild-type V. fischeri (Figure 4A,B). This result therefore reveals 261 that RscS is not required for aggregate formation in a background lacking BinK, and RscS does 262 not contribute to aggregate size in this background. 263 The above results prompted us to ask whether RscS performs any detectable function in 264 a strain lacking BinK. We employed a sensitive competition assay to ask whether strains lacking 265 RscS exhibit a defect upon co-inoculation. In a competitive colonization assay in which the 266 ΔbinK ΔrscS strain was co-inoculated with a LacZ-expressing ΔbinK single mutant, the strain 267 lacking RscS exhibited an approximately 100-fold defect in the competition (Figure 5).

BinK represses syp transcription in the squid crypts 286
We next examined expression of the sypA'-gfp + transcriptional reporter in V. fischeri 287 cells that had aggregated in the host mucus (3-4 hpi). The results in Figure 7A,C reveal 288 indistinguishable overall levels of GFP activity in WT, ΔbinK, and ΔbinK ΔrscS cells in the 289 aggregates of each animal examined. We note some limitations to these data. Given that cells 290 spend a short period of time in the aggregate stage--on the order of 1-2 h--it is unclear whether 291 the reporter is revealing the steady state transcription levels from the aggregate or whether this information integrates time prior to the aggregation stage (growth in liquid media and in 293 seawater). Nonetheless, we present these data for two reasons. First, any physiological 294 response that uses transcription to regulate symbiosis would be subject to similar constraints. 295 Second, the similarity of the data points provides a useful control for the data in the crypts that 296 will be described below. Even with these caveats, we can conclude that the absence of RscS 297 does not diminish the median level of sypA transcription in the ΔbinK background. We note that 298 there was more heterogeneous sypA'-gfp + activity across the aggregate in wild type compared 299 to samples that lacked BinK ( Figure 7A). 300 We proceeded to conduct a similar analysis of the transcriptional reporter in the light 301 organ crypts (48 hpi). At this point, we observed a notable difference between the ΔbinK strain 302 and wild type, with substantially elevated sypA transcription in the cells lacking BinK ( Figure  303 7B,D). The absence of RscS did not affect the sypA reporter. From this imaging, we conclude 304 that a normal function of BinK is to repress syp transcription in the crypts. 305

DISCUSSION 307
By studying the V. fischeri-squid symbiosis model, we have refined our understanding of 308 how bacterial biofilm signaling is regulated at the initiation of bacterial colonization. This study 309 provides evidence that BinK acts as a hybrid histidine kinase, defines novel biofilm regulation in 310 the host, describes a role for BinK in that regulation, and reveals that the key ES114 311 colonization factor RscS is dispensable in the absence of BinK. These major conclusions are 312 discussed in detail below. 313 BinK acts as a hybrid histidine kinase. In our previous study, we identified BinK as a 314 putative hybrid histidine kinase based on its predicted domain structure containing CA, DHp, 315 and REC domains (Figure 1) (22). Furthermore, experimental evolution studies to improve 316 squid colonization of other V. fischeri strains revealed spontaneous binK point mutations in its 317 roles for these domains in BinK function (14). In this work, we used a combination of targeted 319 mutants with a newly-developed anti-BinK peptide antibody to demonstrate that mutagenesis of 320 predicted phosphorylation sites creates BinK proteins that are nonfunctional. 321 In most cases, histidine kinases dimerize, and our results provide genetic support that 322 this occurs in the case of BinK. In particular, our finding that a BinK protein containing a REC 323 domain that is locked as non-phosphorylated with the D794A mutation is not only nonfunctional, 324 but is dominant negative as it interferes with signaling of a wild-type binK allele expressed in the 325 same cell.  Figures 3, 5). There are three major phylogenetic groups of V. fischeri (15). Relevant for this 358 work, the ancestral Group C strains include Mediteranean squid symbionts such as strain SR5. 359 These strains encode functional BinK, but do not encode RscS, and it is unclear how they can 360 colonize squid without the biofilm-promoting activity from RscS (15). Group C strains also 361 include fish symbionts such as MJ11, which cannot colonize squid unless they gain RscS or 362 lose BinK (13, 14). Derived from this group is Group B, which includes strain ES114, which is 363 the focus of the present study. Squid symbionts in Group B typically encode both RscS and 364 BinK, and mutation of RscS leads to an inability to colonize the squid (13, 15). In this study, it 365 was found that when binK is deleted, strain ES114 no longer requires rscS for colonization 366 (Figure 3). This result mirrors a previous finding in Group C, which found that strains lacking 367 RscS (due to evolution) and also lacking BinK (due to directed mutation) could colonize squid 368 (14). Given the diversity of biofilm regulation across V. fischeri, this result was not expected and 369 highlights conserved aspects of regulation that are shared across much of the species (15). It therefore seems that one of the main functions of RscS is to antagonize BinK's negative 371 regulation of biofilm: without the negative regulator, the positive regulator is no longer absolutely 372 required for host colonization. From the evolutionary tree, we know that binK--which is found 373 throughout the species--predates the horizontal gene transfer event that enabled acquisition of 374 rscS, which is only present in a derived group of V. fischeri (13, 15). Therefore, our work raises 375 the question of whether there are other factors that antagonize BinK activity in strains that 376 colonize squid independent of RscS (e.g., the Group C Mediterranean squid symbionts). It 377 seems likely that such activity would be sufficient to enable colonization, given that mutations in 378 binK facilitate colonization by strains that are otherwise unable to colonize well (14).  (14, 15, 22). It seems likely that while the absence of BinK is beneficial 397 to enter the squid host, the absence of the regulator (and subsequent inappropriate biofilm 398 formation at later stages) may be detrimental to the daily homeostasis that is maintained long-399 term in the squid. We also examined sypA'-gfp + reporter activity in biofilm aggregates. The 400 average expression level within aggregates was similar for wild type or ΔbinK mutant cells, but 401 we observed greater heterogeneity in the expression in wild type. This suggests that in the 402 presence of BinK, there is more variability in biofilm expression, and this is worthy of further 403 study. 404 This work provides an exciting view into how biofilm gene expression is regulated in 405 vivo. We know that within a few hours, planktonic bacteria transition to a biofilm state in the host 406 mucus (3, 48). Despite a number of biofilm regulators being identified, how this process is 407 controlled at the host interface is not well understood. Our results provide evidence that this 408 regulation is dynamic over the course of colonization, as evidenced by the BinK-dependent 409 repression that occurs specifically in the crypts at 48 hpi but is not evident in the aggregates at 410 3-4 hpi. We propose that the interaction of BinK with host-derived compounds may lead to 411 downregulation of biofilm genes as bacteria transition from the biofilm aggregates during 412 initiation to cells in the crypts during the persistence stage. Finally, we observed similar sizes of 413 in vivo biofilm aggregates in the host in ΔbinK and ΔbinK ΔrscS strains, arguing that there is no 414 requirement for stimulation of biofilm through RscS to initiate a productive symbiosis. We 415 therefore posit that a key regulatory mechanism to control the planktonic-to-biofilm transition is 416 host inhibition of BinK. 417 In summary, this work provides novel insight into the function of hybrid histidine kinase 418 BinK, its relationship to RscS is regulating symbiotic biofilm formation, and the temporal control 419 of symbiotic biofilm gene expression. Future work will continue to examine the signaling 420 architecture downstream of BinK and host-derived molecules that may regulate BinK activity. 421

MATERIALS AND METHODS 423
Bacterial Strains, Plasmids, and Media 424 V. fischeri and Escherichia coli strains used in this study are listed in Table 1. Plasmids used in 425 this study are listed in Table 2 The previously generated pTn7-binK plasmid, which uses a mini-Tn7 delivery vector backbone 446 (pEVS107), was purified and used as a template. Point mutations to the binK sequence on the 447 plasmid were designed using the NEBaseChanger tool and constructed with the Q5 site-directed mutagenesis kit (New England BioLabs, Inc.). The constructed plasmid was 449 transformed into either electrocompetent or chemically competent DH5α λpir E. coli. The entire 450 binK gene on the plasmid construct was sequenced (using pEVS107 F and R primers and binK 451 sequencing primers). BinK alleles generated in this manner were then introduced into V. fischeri inoculation (water was changed at 24 h post-inoculation), at which point they were euthanized 572 by storage at −80°C. Each squid was homogenized and plated on LBS-Xgal, and the blue/white 573 colony ratios were used to score these competitions as described previously (23, 55). 574

Data analysis and graphing 576
Data analysis was conducted using Python, including the pandas library. For fluorescence of 577 colonies, aggregates, and crypts, the mean GFP and mCherry for the region of interest and a 578 nearby background region was acquired using Zen Blue Software. The background for each 579 channel was subtracted from the region of interest. To normalize GFP to plasmid copy number, 580 GFP was divided by mCherry. This resulted in the reported mean GFP/mCherry reading for 581 each individual colony, aggregate, or crypt space. Graphpad Prism was used to construct 582 graphs and perform statistical analyses.     Bacteria were inoculated into FSIO containing squid and allowed to colonize for 3 hours. Squid 787 were washed and then maintained for two days to allow establishment of the symbiosis. Shown