Bacterial LomR Induces the Vibriophage VP882 VqmA-Directed Quorum-Sensing Lysogeny-Lysis Transition

The bacterial cell-cell communication process called quorum sensing enables groups of bacteria to synchronously alter behavior in response to changes in cell population density. Quorum sensing relies on the production, release, accumulation, and detection of extracellular signal molecules called autoinducers. Here, we investigate a mechanism employed by a vibriophage to surveil host quorum sensing and tune its lysogeny-lysis decision to host cell density. The phage possesses a gene called vqmAPhage encoding a quorum-sensing receptor homologous to vibrio VqmA. Both VqmA receptors can detect the host bacteria-produced autoinducer called DPO. DPO-bound VqmAPhage launches the phage lysis process. We discover that the bacterial host produces an inducer of the VqmAPhage-directed quorum-sensing lysogeny-lysis transition. Production of the inducer appears to be widespread among bacteria. A screen of the Escherichia coli Keio collection for mutants impaired for inducer production revealed lomR, located in a prophage, and encoding a poorly understood protein. In the E. coli screening strain, lomR is interrupted by DNA encoding an insertion element. The 3’ domain of this LomR protein is sufficient to induce VqmAPhage-directed lysis. Alanine-scanning mutagenesis showed that substitution at either of two key residues abrogates inducer activity. Full-length LomR is similar to the outer membrane porin OmpX in E. coli and Vibrio parahaemolyticus O3:K6, and OmpT in Vibrio cholerae C6706, and indeed, OmpX and OmpT can induce VqmAPhage-directed activity. Possibly, development of the LomR, OmpX, or OmpT proteins as tools to direct phage lysis of host cells could be used to control bacteria in medical or industrial settings. ABSTRACT IMPORTANCE Bacteria communicate with chemical signal molecules using a process called quorum sensing. Quorum sensing allows bacteria to track their cell numbers and orchestrate collective behaviors. Recently, we discovered that a virus that infects and kills bacteria “eavesdrops” on its host’s quorum-sensing process. Specifically, the virus monitors host cell growth by detecting the accumulation of host quorum-sensing signal molecules. In response to the garnered quorum-sensing information, the virus kills the host bacterial cells when the bacterial population has reached a high cell density. This strategy presumably enhances transmission of viruses to new host cells. Here, we discover and characterize three closely-related bacterial host-produced proteins called LomR, OmpX, and OmpT that are capable of inducing the viral quorum-sensing-mediated killing program. Development of this class of inducer proteins as tools to drive “on demand” virus-mediated lysis of pathogenic host bacterial cells could be used to control bacteria in medical or industrial settings.

employed by a vibriophage to surveil host quorum sensing and tune its lysogeny-lysis 23 decision to host cell density. The phage possesses a gene called vqmAPhage encoding a 24 quorum-sensing receptor homologous to vibrio VqmA. Both VqmA receptors can detect 25 the host bacteria-produced autoinducer called DPO. DPO-bound VqmAPhage launches the 26 phage lysis process. We discover that the bacterial host produces an inducer of the 27 VqmAPhage-directed quorum-sensing lysogeny-lysis transition. Production of the inducer 28 appears to be widespread among bacteria. A screen of the Escherichia coli Keio 29 collection for mutants impaired for inducer production revealed lomR, located in a 30 prophage, and encoding a poorly understood protein. In the E. coli screening strain, lomR 31 is interrupted by DNA encoding an insertion element. The 3' domain of this LomR protein 32 is sufficient to induce VqmAPhage-directed lysis. Alanine-scanning mutagenesis showed 33 that substitution at either of two key residues abrogates inducer activity. Full-length LomR 34 is similar to the outer membrane porin OmpX in E. coli and Vibrio parahaemolyticus 35 O3:K6, and OmpT in Vibrio cholerae C6706, and indeed, OmpX and OmpT can induce 36 VqmAPhage-directed activity. Possibly, development of the LomR, OmpX, or OmpT 37 proteins as tools to direct phage lysis of host cells could be used to control bacteria in 38 medical or industrial settings. 39 presumably enhances transmission of viruses to new host cells. Here, we discover and 48 characterize three closely-related bacterial host-produced proteins called LomR, OmpX,49 and OmpT that are capable of inducing the viral quorum-sensing-mediated killing 50 program. Development of this class of inducer proteins as tools to drive "on demand" 51 virus-mediated lysis of pathogenic host bacterial cells could be used to control bacteria in 52 medical or industrial settings. 53

INTRODUCTION 54
The bacterial cell-cell communication process called quorum sensing (QS) 55 enables groups of bacteria to synchronously alter behavior in response to changes in cell 56 population density. QS depends on the production, accumulation, and detection of 57 extracellular signal molecules called autoinducers (AIs) (1, 2). Vibrios are model bacteria 58 used for QS studies. One vibrio QS system is composed of the cytoplasmic AI receptor 59 VqmA and its target, the gene encoding the regulatory small RNA called VqmR ( Figure  60 1). VqmA responds to 3,5-dimethylpyrazin-2-ol (DPO), an AI produced from threonine 61 and alanine. The Tdh (threonine dehydrogenase) enzyme is required for DPO 62 biosynthesis. The DPO-VqmA-VqmR system controls target genes including those 63 required for biofilm formation and virulence (3)(4)(5). 64 A VqmA-type QS system was recently identified in the Myoviridae virus VP882, a 65 non-integrating temperate phage whose natural host is Vibrio parahaemolyticus. Phage 66 VP882 infects other vibrios in laboratory settings, including Vibrio cholerae, suggesting 67 that a variety of other vibrios could be hosts in nature. The phage VP882 VqmA-type AI 68 receptor is encoded by a gene that we call vqmAPhage. VqmAPhage "eavesdrops" on vibrio 69 QS (6, 7). Specifically, VqmAPhage binds to host bacteria-produced DPO. Liganded 70 VqmAPhage launches the phage lysis program via production of Qtip, an antirepressor of 71 lysis ( Figure 1 and (6, 8)). In addition to inducing qtip expression, VqmAPhage can activate 72 host vqmR expression (Figure 1), suggesting that the VP882 phage regulates its own 73 lysogeny-lysis decision and host QS behaviors via VqmAPhage-directed QS. Other phages 74 also harbor genes encoding putative ligand-binding transcription factors adjacent to 75 genes encoding companion Qtip-type antirepressors (6,9,10). These findings suggest 76 that it could be common for phages to control their lysogeny-lysis decisions by tuning into 77 host-produced sensory cues using a Qtip-like lysis de-repression mechanism, and 78 perhaps also alter host biology. 79 In a previous study, cloned vqmAPhage that had been reintroduced into the V. 80 parahaemolyticus pandemic O3:K6 strain 882, the original host strain from which phage 81 VP882 was isolated, showed that in the absence of DPO, only a basal level of cell lysis 82 occurred, while increased lysis occurred when synthetic DPO was provided (6). This 83 result indicated that DPO drives VqmAPhage activity. Indeed, when vqmAPhage was 84 eliminated by transposon mutagenesis, DPO-dependent lysis was also eliminated. One 85 caveat of this earlier investigation was that, to facilitate study, vqmAPhage was artificially 86 induced due to its otherwise low level of expression during lysogeny. Nonetheless, 87 following synthetic induction of phage VP882 vqmAPhage expression, a positive feedback 88 mechanism increased vqmAPhage transcription (6). Presumably, this regulatory 89 arrangement enlarges the pool of available VqmAPhage enabling enhanced sensitivity to 90 DPO and, in turn, a more efficient launch of the lysis program. The natural inducer of the 91 VqmAPhage-driven lysogeny to lysis switch that initiates the feedback loop was not 92 identified. 93 Here, we report a class of natural inducers of VqmAPhage activity that are produced 94 by V. parahaemolyticus, V. cholerae, and other bacteria, including Escherichia coli. Using 95 the E. coli Keio collection as a screening tool, we identified a mutant defective in 96 production of the inducer, implicating lomR. The E. coli lomR gene in the Keio collection 97 is interrupted by an IS5Y insertion element (insH5) encoding a transposon (11). Our 98 assessment of which region of lomR'insH5'lomR is required for inducer activity shows 99

A component present in host cell lysates induces VqmAPhage-directed lytic pathway 114
We investigated whether a bacterial component could be the inducing factor that 115 launches the phage VP882 VqmAPhage-directed lysogeny-lysis switch. To test this 116 possibility, we engineered two E. coli Top10 reporter strains that harbor gp69-lux 117 transcriptional fusions. gp69 transcription is activated by  Gp69 is required for phage VP882 VqmAPhage-directed host cell lysis. In addition to gp69-119 lux, the E. coli reporter strains carried either a phage VP882 with a Tn5 insertion in an 120 intergenic region (called control phage VP882) or phage VP882 with a Tn5 insertion in 121 vqmAPhage (called phage VP882 vqmAPhage::Tn5). The E. coli reporter strains naturally 122 make DPO, ensuring DPO is available to bind to VqmAPhage. The rationale is to test 123 whether bacteria produce an activity capable of inducing the gp69-lux reporter and, if so, 124 whether the component requires VqmAPhage to do so. 125 In Figure 2A, the first two bars for each reporter strain depict controls and 126 demonstrate how the assay system functions. Administration of LB medium shows the 127 basal levels of lux expression that occur in response to endogenously-produced DPO. 128 Administration of LB supplemented with DPO activates expression of the reporter 129 construct, however, only in the strain harboring control phage VP882, demonstrating that 130 endogenously-produced DPO does not saturate the system, and moreover, that DPO-131 driven induction of gp69-lux requires the presence of vqmAPhage. 132 To test whether a bacteria-produced component induces VqmAPhage-directed 133 activity, and to capture any possible inducer irrespective of type of molecule, we prepared 134 and assessed crude lysates from V. parahaemolyticus O3:K6 (hereafter named O3:K6), 135 the natural host of phage VP882. Figure 2A shows that the preparation possesses an 136 activity that modestly induces gp69-lux expression only in the E. coli reporter strain 137 carrying control phage VP882. Although the data did not show statistical significance, we 138 reproducibly observed low levels of induction from O3:K6 lysates. Likewise, lysate from 139 O3:K6 harboring phage VP882 modestly, but reproducibly, induced gp69-lux activity. The 140 fact that lysate from both phage VP882 infected and non-infected O3:K6 displayed some 141 activity reinforces the notion that the inducer is not produced by the phage, but, rather, is 142 a bacterial product. Consistent with this idea, lysates from V. cholerae C6706 (hereafter 143 named C6706) as well as strains of E. coli from our collection (BW25113, Top10, and 144 DH5a) also induced the reporter, albeit to different extents, with the strongest induction 145 by the BW25113 lysate. Together, these data suggest that the inducer is not restricted to 146 vibrios. These data do not allow us to distinguish whether the inducer is identical in the 147 different bacterial genera. We return to this point in a later section. Since lysates from E. 148 coli BW25113 exhibit the strongest activation of gp69-lux expression among the bacteria 149 we tested we use E. coli BW25113 as the source of the inducer activity in most of the 150 remainder of this work. We refer to E. coli BW25113 as BW25113 from here forward. 151 We wondered whether the inducer was a small molecule, DNA, or a protein, as 152 this broad characterization would guide us in its identification. To garner preliminary 153 information, lysate from BW25113 was treated as follows: heat or Proteinase K to 154 denature proteins, Benzonase to degrade DNA, and passage through a 10 kD molecular 155 weight cut-off (MWCO) filter to crudely separate molecules by size ( Figure 2B). Heat and 156 Proteinase K treatment eliminated the inducing activity, whereas the activity was resistant 157 to Benzonase. Filtration demonstrated that the factor has a MW > 10 kD. Together, these 158 data indicate that the factor is likely a protein. From here forward we will call this host-159 produced inducer of the phage VqmAPhage pathway the "inducer protein". 160

161
Screening of the Keio collection reveals LomR as a candidate protein that induces the 162 phage VP882 QS-directed pathway 163 BW25113, which makes significant inducer activity (Figure 2A), is the parent strain 164 for an ordered library of E. coli null mutants (the Keio collection (13)), providing us a 165 means to identify the inducer protein. Specifically, we screened lysates prepared from the 166 Keio collection strains for those lacking the ability to induce our E. coli Top 10 gp69-lux 167 reporter. We identified several BW25113 mutants that appeared defective in production 168 of the inducer protein. To verify these results, we reconstructed the mutations in the 169 candidate genes in BW25113. Only lysates from BW25113 lacking lomR elicited reduced 170 gp69-lux reporter expression compared to lysates from BW25113 (Supplementary Figure  171 1). Complementation with lomR on a plasmid restored inducer production to the BW25113 172 ΔlomR mutant ( Figure 3A). Moreover, high level expression of lomR resulted in activity in 173 lysates exceeding that naturally made by BW25113 (Supplementary Figure 2) Figure  182 3). Phage LomR homologs resemble OMPs belonging to the Gram-negative porin 183 superfamily (11). The functions of phage LomR proteins are poorly understood. What is 184 known is that they are immunogenic porins proposed to assist in small molecule transport, 185 colonization, and adhesion, and they may play roles in cell lysis or in altering permeability 186 of phage infected cells (16,17). We tested whether lomR l could complement BW25113 187 ΔlomR and restore inducer activity. To do this, we transformed BW25113 ΔlomR with 188 pBAD-lomR l , and assayed lysates made from this strain in our activity assay. There was 189 indeed activity ( Figure 3B), suggesting that LomR l can also function as an inducer. here IS5Vc), a homolog of insH5, that is nearly identical in sequence to the insH5 insertion 206 in lomR BW25113 (76.7% similarity in nucleotide sequence, with 81.9% identity and 90.8% 207 similarity in amino acid sequence) so we also considered the possibility that IS5Vc could 208 play a role in inducer protein production. To examine this possibility, we cloned IS5Vc into 209 pBAD. We transformed each of these constructs into BW25113 ΔlomR, prepared lysates, 210 and assayed them in our reporter strains. Neither lomR B5 , insH5, nor IS5Vc restored 211 activity. Expression of lomR BW25113 , lomR B3 and lomR B5-3 complemented the ΔlomR defect 212 ( Figure 3B). Therefore, only the 3' domain of LomR is required to restore inducing activity 213 to BW25113 ΔlomR. Given that lomR BW25113 contains an apparently inactivating insertion, 214 we wondered how we identified lomR BW25113 in our Keio collection screen. Analysis of the 215 DNA sequence in this region shows that in lomR BW25113 , downstream of the insH5 216 insertion, is DNA encoding a start codon that lies in frame with the DNA encoding LomR B3 217 (Supplementary Figure 3). Presumably, this start codon is employed to produce LomR B3 218 in BW25113. Indeed, qRT-PCR confirmed that lomR B5 , insH5, and lomR B3 are expressed 219 in BW25113 ( Figure 3C-D) with insH5 being the most highly expressed and lomR B5 and 220 lomR B3 expressed at similar levels. We could not amplify full-length lomR BW25113 region 221 suggesting no transcription of the full region occurs due to disruption by insH5 ( Figure 3C-222 D). We obtained similar results in E. coli Top10 ( Figure 3C-D), the strain we used in our 223 reporter assays that also exhibits inducing activity ( Figure 2A) and has an insH5 element 224 inserted in lomR. 225 As above, we carried out qRT-PCR with the BW25113 ΔlomR strain. 226 Unexpectedly, the strain produced lomR B5 and insH5 transcripts, but it did not produce 227 detectable lomR B3 transcript (Supplementary Figure 4). The BW25113 ΔlomR strain used 228 in our study was reconstructed from the original Keio mutant by P1 phage transduction of 229 the lomR::kan region into BW25113 followed by elimination of the kanamycin resistance 230 marker (see Methods). Examination of the primer sequences used by Baba et. al (13) to 231 construct the ΔlomR strain in the Keio collection revealed that their design deleted the 3' 232 portion of the lomR BW25113 locus leaving the upstream sequence (i.e., the region that we 233 call lomR B5 ) and a majority of the insH5 sequence intact. Thus, we conclude that the 234 original Keio mutant and the BW25113 ΔlomR strain we made and used in our study lacks 235 only lomR B3 . These results are consistent with our finding that, in this locus, expression 236 of only the lomR B3 gene segment is required for inducer activity. 237 In E. coli K12 strains that encode intact LomR proteins and/or that have LomR l , 238 these proteins possess N-terminal signal sequences that presumably drive localization to 239 the outer membrane, consistent with their being OMPs. However, LomR B3 in BW25113 240 does not have a signal sequence and lysates prepared from strains in which we introduce 241 DNA encoding only LomR B3 exhibit activity ( Figure 3B). Thus, localization of LomR B3 to 242 the outer membrane appears dispensable for activity. To verify this notion, we focused 243 on LomR l , which we have shown can substitute for LomR B3 ( Figure 3B). We engineered 244 arabinose-inducible lomR l with its native signal sequence removed (called lomR l-SS ). 245 Lysates prepared from BW25113 ΔlomR carrying lomR l-SS activated the reporter ( Figure  246 3B). With respect to LomR B3 , we fused it to Halo, introduced it into BW25113, and showed 247 by confocal microscopy that LomR B3 -Halo is indeed diffuse in the cell ( Figure 3E). 248 Together, these findings suggest that LomR B3 inducer activity does not depend on 249 transport to and insertion into the outer membrane. All of these results are consistent with 250 our crude lysate preparations containing both membrane and cytoplasmic contents (see 251

Methods). 252
To garner additional proof that activation of the VqmAPhage-directed lysogeny to 253 lysis transition can be induced by LomR B3 , we purified this domain and added it 254 exogenously to the E. coli Top10 gp69-lux reporter strains. Purified LomR B3 was sufficient 255 for induction of activity ( Figure 3F). Consistent with this finding, pre-treating purified 256 LomR B3 with Proteinase K eliminated inducer activity ( Figure 3F). As a control, we also 257 supplied ubiquitin, a protein similar in size to LomR B3 , to the reporter strain. Ubiquitin did not induce gp69-lux ( Figure 3F). To pinpoint the amino acid (AA) residues in LomR B3 259 required to confer inducer activity, we used alanine scanning mutagenesis. LomR B3 is 86 260 AAs in length. Initially, we mutated every set of three contiguous AAs to alanines, 261 revealing the A7-P8-D9 stretch as implicated in inducer activity ( OmpL, OmpN, OmpT, OmpW, and OmpX. We individually deleted the genes encoding 273 each OMP. Figure 4D shows that lysate from only the BW25113 strain lacking OmpX lost 274 the capability to activate gp69-lux. Complementation with cloned ompX BW25113 rescued 275 inducer activity ( Figure 4E). These results suggest that the vqmAPhage pathway can be 276 activated by LomR B3 and/or its homolog OmpX BW25113 , but not generally by OMP proteins. 277 We do not understand why ompX BW25113 was not revealed in our original screen of the 278 Keio collection. 279 Not surprisingly, lysate from the BW25113 double ΔlomR ΔompX mutant 280 possessed no inducer activity, consistent with the inability of lysates prepared from the 281 single ΔlomR and single ΔompX mutant strains to induce the reporter ( Figure 4F). 282 Together, the results with the three mutants suggest that either OmpX BW25113 or LomR B3 283 is sufficient for inducer production. Because their activities are not additive, an epistatic 284 relationship could exist between LomR B3 and OmpX BW25113 in which one of the proteins 285 controls the other's production or activity. To examine this possibility, we used qRT-PCR 286 to measure transcription of lomR B3 and ompX BW25113 in both single mutants. 287 Supplementary Figure 6 shows that deletion of lomR B3 or deletion of ompX BW25113 did not 288 affect transcription of, respectively, ompX BW25113 or lomR B3 . Thus, we do not find evidence 289 for an epistatic connection. Our data do not eliminate the possibility that the LomR B3 and 290 OmpX BW25113 proteins may interact with and alter one another's activity. 291 V. parahaemolyticus and V. cholerae do not possess lomR genes, however, they 292 do harbor ompX-type genes. The C6706 ompX homolog is called ompT ( Figure 4A-C). 293 Thus, OmpX in O3:K6 (OmpX Vp ) and OmpT in C6706 (OmpT Vc ) could act as inducers of 294 the phage VP882 lysis cascade. To test this possibility, we deleted ompT Vc from C6706. 295 O3:K6 is not amenable to genetic manipulation so we could not test function in this strain. 296 Figure 4G shows that lysate prepared from C6706 ΔompT does not exhibit inducer activity 297 in contrast to lysate prepared from WT C6706, confirming that OmpT Vc can also act as 298 an inducer of VqmAPhage-directed activity. 299 300 LomR and OmpX can induce lysis in phage VP882-infected V. parahaemolyticus, the 301 natural host for phage VP882. 302 Our above studies relied on a heterologous reporter as a proxy for VqmAPhage-303 directed lysis. To test whether purified LomR B3 and purified OmpX BW25113 are capable of 304 inducing lysis in vivo, we assessed gp69-lux expression, and simultaneously, we 305 measured cell lysis by tracking optical density of O3:K6 carrying control phage VP882. 306 As a control, we administered mitomycin C (MMC), a known inducer of lysis (6, 7). MMC 307 activated gp69-lux and drove cell lysis ( Figure 5A and 5B, respectively). Exogenous 308 addition of purified LomR B3 or purified OmpX BW25113 to O3:K6 caused equally strong 309 activation of the gp69-lux fusion ( Figure 5A) and drove moderate cell lysis when phage 310 VP882 was present ( Figure 5B). Protease pretreatment of LomR B3 and OmpX BW25113 311 eliminated lysis capability. In the opposite vein, addition of ubiquitin did not induce 312 reporter expression or lysis ( Figure 5A and 5B, respectively). 313 All of the preceding assays relied on exogenous addition of the LomR B3 (or 314 OmpX BW25113 ) inducer to cells strongly suggesting that LomR B3 acts from the extracellular 315 environment. As a preliminary test of whether the VqmAPhage-directed lysis pathway can 316 be induced by LomR B3 that is present in the cytoplasm, we overexpressed lomR B3 on 317 pBAD in BW25113 ΔlomR carrying either control phage VP882 or phage VP882 318 vqmAPhage::Tn5 and examined whether lysis occurred. Supplementary Figure 7 shows no 319 difference in cell growth following lomR B3 overexpression. Consistent with this finding, we 320 note that E. coli Top10, the strain we use for our reporter assay, expresses lomR B3 at a 321 level similar to that expressed by BW25113 ( Figure 3D), yet, cell lysis does not occur. 322 This observation further suggests that physiological levels of LomR B3 in the cytoplasm 323 are insufficient for activation of the VqmAPhage pathway and LomR B3 must be provided 324 from the external environment. Thus, to date, we have no evidence that LomR can 325 function from the cytoplasm. 326

DISCUSSION 328
The DPO-VqmA QS AI-receptor coordinates group behaviors in vibrios ( Figure 1 OmpC is superior to full-length OmpC at activating the DegS protease to launch the 341 envelope stress response. The difference in activity stems from stronger binding of 342 truncated OmpC than full-length OmpC to DegS (18). Nonetheless, full-length LomR and 343 OmpX proteins localize to the bacterial outer membrane (18). However, it cannot be that 344 translocation to the membrane is required for induction of the VqmAPhage-directed lysis 345 pathway. We say this because, at least in BW25113, the LomR protein is naturally 346 truncated and possesses no signal sequence. Likewise, lysates prepared from cells 347 producing Lambda LomR lacking its signal sequence induced the VqmAPhage-directed 348 lysis pathway ( Figure 3B). These observations are consistent with our finding that LomR 349 (truncated and full-length) and OmpX induce VqmAPhage activity from the extracellular 350 environment (Figures 2, 3, and 5). Possibly, inducer protein residing in the outer 351 membrane (in the case of full-length proteins) or residing in the cytoplasm (in the case of 352 truncated versions and prior to transport to the outer membrane) are released as a 353 consequence of ordinary (i.e., non-phage-mediated) cell lysis and provide the source of 354 extracellular inducer in nature. We do not yet understand the mechanism underlying 355 LomR B3 /OmpX Vp /OmpT Vc activation of the phage lysis program. Possibilities include 356 increased vqmAPhage expression, increased VqmAPhage production, increased VqmAPhage 357 activity, or increased DPO production. We are currently exploring the underlying 358 mechanism. 359 The phage Lambda LomR porin protein has long been considered an accessory 360 protein because it is not essential for phage replication under laboratory conditions (21). 361 Rather, Lambda LomR is thought to function in lysogenic conversion. LomR proteins that 362 are encoded on cryptic phages residing in bacterial genomes are proposed to provide 363 adaptative advantages to the host bacteria, particularly during pathogenesis (21, 22). For 364 example, in Brucella species, LomR increases mammalian host cell adhesion, invasion, 365 and evasion of mammalian host defenses (23). In some disease-causing E. coli strains, 366 the LomR protein (in these cases full-length LomR, not truncated LomR) produced from 367 the genome is implicated as a virulence factor that triggers bacterial attachment to human 368 buccal epithelial cells, a step required for infection (21). Regarding origins, phage Lambda 369 LomR is presumed to be the ancestor of a five-member family of virulence-related outer 370 membrane proteins in Gram-negative bacteria, including Ail (24), Omp21 (25), PagC (26), 371 Rck (27), and OmpX (28). All of these proteins share highly conserved residues in their 372 b-barrel cores (29). Our results suggest that perhaps phage VP882 exploits this family of 373 Lambda-derived LomR homologs to synchronize its lysogeny to lysis transitions with high 374 host cell density. Such a strategy could improve phage VP882 dissemination. 375 We suggest a model in which a class of bacterial host factors regulate this phage 376 pathway, which is surprising given that induction drives host cell killing by the phage. V. 377 parahaemolyticus containing phage VP882 exists in nature, and in our laboratory 378 analyses, the cells do not die at high cell density. Thus, a mechanism must operate to 379 restrict VqmAPhage activity and enable lysogeny. Perhaps, in V. parahaemolyticus, the 380 requirement for OmpX-mediated induction of VqmAPhage activity provides this mechanism. 381 In the case of lysogens, mechanisms need to exist to tamp down OmpX production and/or 382 OmpX export to enable later VqmAPhage function when high host cell density is achieved 383 and lysis is warranted. How these different phage lifestyles are maintained, how 384 transitions between them occur, and the precise role(s) OmpX plays in phage lifestyle 385 decision making events await future analyses. 386 Known inducers of phage lysis consist almost exclusively of synthetic (i.e., 387 laboratory-used) DNA-damaging agents such as MMC that do not exist in authentic 388 environments in which bacteria and phages co-reside. Our discoveries suggest that in V. 389 parahaemolyticus, OmpX could function as a natural regulator that dictates the timing of 390 the phage lysis program. We speculate that under particular conditions, OmpX 391 accumulates in the extracellular environment as cell density increases, perhaps in step 392 with accumulation of the DPO AI, and together or sequentially, DPO and OmpX induce 393 the phage to switch from lysogeny to lysis. More certain is that the QS regulatory arm of 394 this phage's lysogeny-lysis transition can be governed by two bacteria-derived products, 395 DPO and OmpX, suggesting that discovering how non-traditional stimuli affect lysogeny-396 lysis pathways could improve our general understanding of how phages make decisions 397 in real-world environments. Further studies are required to determine if LomR and its 398 homologs, such as OmpX and OmpT, can induce lysis in natural systems. If so, one can 399 imagine them being developed as tools to drive phage lysis "on demand," to control 400 pathogenic bacteria in medical or industrial contexts. 401

Bacterial Strains 404
Strains used in this study are listed in Table S1A in Tables S1B and S1C, respectively. Plasmids used in this study were validated by 414 sequencing (Genewiz) and are listed in Table S1D Overnight cultures of BW25113 and Top10 were diluted 1:1000 and grown to OD600 = 2.0 463 prior to harvest. Transcripts were stabilized and preserved with RNA-Protect reagent 464 (Qiagen) as recommended by the manufacturer. RNA was extracted and processed for 465 qRT-PCR as described (33). In the cases in which no transcripts were detected in control 466 reactions lacking reverse transcriptase, for calculation purposes, the Ct value was 467 artificially set to 40, which is the maximum number of cycles used in the PCR reaction 468 and thus represents the limit of detection for our sample set. 469

Confocal Microscopy 470
Live-cell staining of BW25113 carrying HALO fusions was performed using a previously 471 described HALO-TMR protocol described (6)  Query sequence: To identify bacterial genes encoding LomR family proteins, lomR B3 was 480 used as the query. The lomR B3 sequence is highly homologous to the 3'-most 190 481 nucleotides of lomR l ( Figure 4A). The amino acid sequence encoded by lomR B3 was used 482 as the query for protein sequence homology analyses. 483 Search set: 21,076 bacterial strains with fully assembled genome sequences were 484 scanned. Their genomic DNA sequences were downloaded from the GenBank database 485 (35). 486 Similarity scoring: For DNA sequence alignments using the SW algorithm, identical 487 nucleotides were assigned a score of +1, and non-identical nucleotides or gaps were 488 assigned a score of -2. For protein sequence alignments, the standard substitution matrix 489 BLOSUM62 was used to compute similarity scores (36). The p-values associated with 490 the similarity scores that the SW algorithm yielded were computed by applying a shuffling 491 algorithm to the subject sequences. A p-value of < 10 -6 was used to identify homologous 492 sequences. Sequence homologs were annotated by PROKKA (37) and were verified to 493 encode LomR family or OmpX family proteins. 494

Protein structural analyses 495
HHpred (MPI Bioinformatics Toolkit) was used to identify bacterial proteins possessing 496 structural similarity to the query proteins (38). The top ~8 to 10 hits from the search were 497 used as templates to predict the structure of the query protein using MODELLER software 498 (39). Visualization and spatial alignments of protein structures were performed using 499 PyMOL (Schrödinger, version 2.4.2). 500

Quantification and Statistical Analysis 501
Data are presented as the mean ± standard errors of the means. The number of 502 independent biological replicates for each experiment is indicated in the figure legends. 503 No blinding or randomization was used in these experiments. All bioluminescence data 504 were recorded, analyzed, and plotted using Microsoft Excel and GraphPad Prism 9. 505 Imaging data were collected, processed, and prepared using LASX and Fiji. 506

Data Availability 507
All experimental data that support the findings of this study are available from the 508 corresponding author upon request. 509

ACKNOWLEDGEMENTS 511
We thank members of the Bassler laboratory for insightful discussions. This        were grown in medium containing 0.2% arabinose prior to lysis. Assay as in Figure 2A. Data were normalized to expression of the gyrB housekeeping gene. Fold-changes are 767 relative to expression in BW25113. Data are represented as mean ± SD with n=3 768 biological and n=2 technical replicates. Statistical significance was calculated using one-769 way ANOVA Tukey's multiple comparisons test between BW25113 and BW25113 ΔlomR 770 or BW25113 and BW25113 ΔompX. 771