Transcriptional pathways across colony biofilm models in the symbiont Vibrio fischeri

ABSTRACT Beneficial microbial symbionts that are horizontally acquired by their animal hosts undergo a lifestyle transition from free-living in the environment to associating with host tissues. In the model symbiosis between the Hawaiian bobtail squid and its microbial symbiont Vibrio fischeri, one mechanism used to make this transition during host colonization is the formation of biofilm-like aggregates in host mucosa. Previous work identified factors that are sufficient to induce V. fischeri biofilm formation, yet much remains unknown regarding the breadth of target genes induced by these factors. Here, we probed two widely used in vitro models of biofilm formation to identify novel regulatory pathways in the squid symbiont V. fischeri ES114. We discovered a shared set of 232 genes that demonstrated similar patterns in expression in both models. These genes comprise multiple exopolysaccharide loci that are upregulated and flagellar motility genes that are downregulated, with a consistent decrease in measured swimming motility. Furthermore, we identified genes regulated downstream of the key sensor kinase RscS that are induced independent of the response regulator SypG. Our data suggest that transcriptional regulator VpsR plays a strong role in expression of at least a subset of these genes. Overall, this study adds to our understanding of the genes involved in V. fischeri biofilm regulation while revealing new regulatory pathways branching from previously characterized signaling networks. IMPORTANCE The V. fischeri-squid system provides an opportunity to study biofilm development both in the animal host and in culture-based biofilm models that capture key aspects of in vivo signaling. In this work, we report the results of the transcriptomic profiling of two V. fischeri biofilm models followed by phenotypic validation and examination of novel signaling pathway architecture. Remarkable consistency between the models provides a strong basis for future studies using either approach or both. A subset of the factors identified by the approaches were validated in the work, and the body of transcriptomic data provides a number of leads for future studies in culture and during animal colonization.

I n animal-microbe mutualisms, symbionts can provide nutritional, developmental, and/or defensive benefits to the host, and in turn, the symbionts may gain various benefits from association with the host (1)(2)(3)(4)(5)(6).During horizontal transmission, hosts must reacquire their symbionts each generation from environmental symbiont populations (7)(8)(9)(10)(11)(12)(13)(14).Unfortunately, the understanding of how specific microbes make this transition from environment to host is often hindered by the complexity of animal microbiomes (15,16).Symbioses with limited symbiont diversity are therefore valuable as models to identify mechanisms of colonization and transmission.The binary mutualism between the nocturnal Hawaiian bobtail squid (Euprymna scolopes) and the bioluminescent

Definition of a core biofilm regulon for V. fischeri ES114
V. fischeri wrinkled colony biofilms are a close proxy for the aggregates that form in vivo during squid host colonization (21,22).To begin to understand the genes that are induced under these conditions, we asked what patterns of gene expression were shared in two common wrinkled colony biofilm models: (i) genetic induction with an rscS overexpression allele (rscS*) in an otherwise wild-type background compared to wild-type V. fischeri ES114 and (ii) chemical induction with 10 mM CaCl 2 in a ΔbinK background (ΔbinK-Ca 2+ ) compared to ΔbinK alone (33,36) (Fig. 1B).We selected these comparisons to explicitly compare biofilm induced to the closest uninduced control conditions for each.To query gene expression in wrinkled colony biofilms, we compared the global transcriptome in the induced condition with the uninduced condition in each FIG 1 Vibrio fischeri symbiosis polysaccharide regulation.(A) V. fischeri induces biofilm formation through a phosphorelay initiated at the hybrid sensor kinase RscS, flowing through the hybrid sensor kinase SypF to the response regulator SypG.Growth on calcium media induces a secondary pathway also flowing through SypF.In opposition to this pathway, the sensor kinase BinK inhibits the activation of SypG, preventing biofilm formation.(B) V. fischeri ES114 has a smooth colony morphology when grown on Luria-Bertani salt (LBS) agar.Overexpression of the sensor kinase RscS (rscS*) or growth of a ΔbinK mutant on LBS 10 mM Ca 2+ agar permits the formation of "wrinkled" colonies with a rugose morphology due to production of the Syp exopolysaccharide.
case.Spots of overnight culture were spotted onto fresh agar and incubated at 25°C to form wrinkled colony biofilms.After 48 hours, each spot was scraped, and RNA was isolated for differential expression analysis.Throughout the study, we used thresholds of twofold induction/repression and false discovery rate (FDR) of 0.05.

Effects on exopolysaccharide gene transcription
In both biofilm models, we detected a large number of genes that were differentially expressed, with over 200 genes upregulated and over 200 genes downregulated upon biofilm induction (Fig. 2A).Comparing the results from the two models revealed 123 genes that were significantly upregulated and 109 genes that were significantly  (28).For the one gene which had no significant changes in expression (sypG, rscS* vs wild type), NS is indicated above that column.downregulated in both cases (Fig. 2B).The most highly upregulated genes included those in the Syp locus, which are required for aggregation in the host and are known to be regulated downstream of RscS and ΔbinK-Ca 2+ signaling (33) (Fig. 2C).Examination of the data revealed a remarkably consistent response for the syp genes across both models.The syp locus is organized into four operons, each of which has a σ 54 promoter and binding sites for the enhancer-binding protein SypG to facilitate transcription by Eσ 54  (28).We noted a trend in which the first gene in each operon was the most highly induced, with sypA being the highest induced gene across the entire locus with over 97-fold induction in the rscS* background and 49-fold induction in the ΔbinK-Ca 2+ background (Fig. 2D).Proceeding along each operon, the fold induction generally diminished for subsequent genes.Again, responses under the two biofilm models were highly correlated.The three SypG-regulated biofilm maturation (matrix) proteins and three biofilm-associated lipoproteins are similarly regulated by both models (Fig. 2C) (38).These data therefore illustrate that our RNA-seq analysis from agar-grown colony spots captures relevant transcriptional responses and that integration of both models yields valuable signals.In the remainder of the study, we refer to genes that show consistent regulation under both models as the "core biofilm regulon." One unexpected result was that sypG had the lowest biofilm induction among the 18 syp genes.Given the importance of SypG to biofilm regulation as the σ 54 -dependent activator for transcription of the syp promoters, we were curious to investigate the relative transcript levels of sypG within a sample, especially in the uninduced conditions.We conducted an analysis of gene expression within each condition using transcripts per kilobase million reads (TPMs) and focused on the genes within the syp locus (39).This analysis revealed that sypF and sypG had the highest TPM values in the uninduced wild-type sample (Fig. S1).These results argue that sypF and sypG have higher basal transcriptional levels in non-biofilm-induced conditions and suggest that they may be regulated in a distinct manner from the other genes across the locus.
Beyond the syp locus, V. fischeri encodes two additional exopolysaccharide loci (26,40).The products of the bcs locus produce cellulose, which permits adhesion to surfaces and may regulate the transcription of the syp locus (26,41).In both biofilm models, we observed moderate upregulation of bcs genes (Fig. S2A), with bcsQ (VF_A0885) and bcsE (VF_A0886) exhibiting the highest fold changes, fitting the syp pattern of the first gene in each respective operon being the most highly upregulated.The other exopolysaccharide locus spans VF_0157-VF_0180 and is regulated by the quorum sensing regulator LitR (40).To date, the function of this locus remains unclear, yet it does appear to enhance phage infection by the V. fischeri phage HNL01 (6).In contrast to the upregulation of the bcs and syp loci within the core biofilm regulon, we noted that a majority of genes in the VF_0157-VF_0180 EPS locus were downregulated modestly in both models (Fig. S2B).

Effects on other biofilm-related loci of note
Among the most highly upregulated genes in the core biofilm regulon, we noted a predicted curli amyloid fiber biosynthetic locus (Fig. 3A).In other bacteria, CsgA subunits polymerize into curli amyloid fibers on the exterior of the cell in a process nucleated by CsgB (42,43).Curli fibers mediate aggregation and biofilm formation in different species, and the operon structure in V. fischeri mimics that of other organisms including predicted export genes (43)(44)(45).The V. fischeri CsgA and CsgB proteins are larger than their Escherichia coli homologs, and each shares 32.0% similarity to the characterized curli proteins from E. coli K-12 (46,47).Notably, V. fischeri curli proteins are larger than their E. coli homologs, with CsgA and CsgB predicted to be 315 and 190 amino acids (2.1× and 1.25× in length, respectively).We generated a strain lacking the csgBA genes (ΔcsgBA), eliminating both the major and minor curlin subunits, and queried whether interruption of the curli genes reduced host colonization levels.We observed no significant difference in final colonization levels between wild type and the ΔcsgBA mutant (Fig. 3B), suggesting that the curli locus is not a critical colonization factor in this context.We further tested whether the absence of curli would impact biofilm aggregate formation in the host.However, we did not observe any significant reductions in the size of biofilm aggregates in the ΔcsgBA mutant compared to wild type, suggesting that the curli locus is not required in host biofilm aggregates (Fig. 3C).Outside of the squid host, we tested whether the curli amyloid fiber locus impacted in vitro models of biofilm formation by introducing the ΔcsgBA mutation into a RscS overexpression background (rscS*).We noted no significant defects in wrinkled colony formation in the rscS* ΔcsgBA mutant at both early (24 hours) and late (48 hours) timepoints (Fig. 3D).Additionally, the ΔcsgBA mutant did not have decreased binding of Congo Red dye, in contrast to what is observed for csgBA deletions in other organisms (48).Given the linkage between curli production and cellulose EPS in other organisms (49,50), we proceeded to test whether deletion of both csgBA and the cellulose EPS synthase bcsA would impact biofilm formation.In the wrinkled colony biofilm model, the strain lacking both curli and bcsA demonstrated no significant differences in morphology compared to a parental biofilm-induced strain, and it exhibited Congo Red binding equivalent to a bcsA single mutant.Therefore, while it is intriguing that the curli system is induced concomitant with biofilm formation, we did not detect a functional requirement for the curli genes.Given our results and the lack of Congo Red binding attributable to the curli gene products, we are unable to determine whether curli fibers are produced or play a role during symbiotic colonization.

Effects on flagellar motility
Outside of exopolysaccharide regulation, the core biofilm regulon had a substantial downregulation of motility genes.We observed that 32 of the 43 genes required for swimming motility in V. fischeri (51) trended toward reduced expression in both biofilm models.Of these genes, five met our significance threshold: cheB/VF_1830, cheY/VF_1833, fliN/VF_1844, fliM/VF_1845, and flrB/VF_1855 (Fig. 4A; Table S1).We therefore asked whether the rscS* or ΔbinK backgrounds had a deficit in swimming motility by examining migration in tryptone broth salt (TBS) soft agar swim plates (Fig. 4B).Compared to wild-type controls, either rscS* or the ΔbinK genetic backgrounds alone were sufficient to reduce swimming migration on soft agar by at least 15%, demonstrating that upregula tion of the biofilm pathway leads to a diminution of swimming motility (Fig. 4C).

Effects on cyclic di-GMP
In V. fischeri and in many bacteria, increased cyclic di-GMP (c-di-GMP) levels positively impact the production of the cellulose exopolysaccharide and negatively regulate swimming motility, both effects noted in our study (41,52).Within the transcriptomics data set, we asked which diguanylate cyclases (DGCs) and phosphodiesterases (PDEs) were differentially regulated under these conditions (Fig. S3A; Table S2).Both biofilm models (rscS* and ΔbinK-Ca 2+ ) upregulated two DGCs (VF_A0152 and casA/VF_1639), two PDEs (VF_A0526 and binA/VF_A1038), and one predicted bifunctional DGC/PDE (VF_0985).In contrast, both models downregulated four DGCs (VF_1350, VF_A0398, mifB/ VF_A0959, and VF_A1012), one PDE (VF_A0551) and two DGCs/PDEs (VF_0094 and VF_A0244).We next directly tested production of c-di-GMP using a plasmid-based fluorescent reporter (pFY4535) (53).This reporter has constitutive expression of the AmCyan fluorescent protein, with the expression of TurboRFP under the control of c-di-GMP-binding riboswitches.Higher levels of c-di-GMP therefore increase the Tur boRFP:AmCyan ratio.Compared to wild-type V. fischeri and control strains known for either high c-di-GMP levels (Δ6PDE) or low c-di-GMP levels (Δ7DGC) (41,52), we noted that rscS* induced a mild but significant increase in c-di-GMP levels in colony biofilms grown on standard Luria-Bertani salt (LBS) media, while no such induction was observed in the ΔbinK background (Fig. S3B).Calcium supplementation (10 mM) of LBS media greatly increased c-di-GMP levels in both rscS* and the ΔbinK colony biofilms to equiva lent levels.This result may be explained by the shared upregulation of the calciuminduced DGC casA in both models, as overexpression of CasA has previously been linked to increased c-di-GMP synthesis in calcium-supplemented liquid cultures (Fig. S3C) (54).Overall, both models of biofilm formation appear to be capable of increasing the global levels of c-di-GMP.

RscS overexpression and calcium supplementation induce regulatory programs unique to each model
Despite the contributions of both biofilm models to the shared biofilm regulon, we also noted targets that appeared to be regulated by either model alone.For the RscS overexpression model, we detected 61 genes with increased expression and 68 genes with reduced expression specific to this model (Table S3).Notable targets within the upregulated group included genes encoding the regulator HbtR/VF_A0473 and its chaperone HbtC/VF_A0474, alongside 15 genes encoded on the natural plasmid of V. fischeri ES114, pES100.Of the 68 downregulated genes, we observed multiple genes associated with flagellar motility, including fliA/VF_1834, flagellar motor protein genes motA2/VF_A0186 and motB2/VF_A0187, and the chemotaxis factors cheV2/VF_A0802 and VF_2042.Compared to the RscS overexpression model, the ΔbinK-Ca 2+ model affected fewer targets, with 38 upregulated genes and 53 downregulated genes specific to this model (Table S3).The upregulated group contained multiple iron and copper assimila tion loci and the quorum sensing autoinducer C8-HSL synthase AinS.Downregulated genes in this model included multiple flagellar genes including flaA/VF_1866, as well as predicted regulatory factors including the cold-shock DNA-binding transcriptional regulator cspG/VF_A1094.

Novel members of the core biofilm regulon are induced independently of the response regulator SypG
The dominant mode of signaling from sensor kinase RscS is through hybrid sensor kinase SypF to response regulator SypG (27).We therefore expected that removal of SypG in the rscS* overexpression background would eliminate most of the effect observed in that model, based on previous publications and our own colony biofilm assays (Fig. 5A) (25,27,55).While the syp, bmp, and bal genes did not exhibit induction in the absence of SypG, many of the remaining rscS*-induced genes continued to be differentially expressed without the downstream regulator (Fig. 5B).A total of 111 genes were significantly upregulated and 60 genes were significantly downregulated, despite the attenuation of the rscS* ΔsypG mutant in forming wrinkled colony biofilms.These results suggest that RscS maintained a regulatory program independent of SypG signaling (Fig. 5B).Visualization of these data revealed a clear bifurcation showing this unexpected SypG-independent effect within the rscS* model (Fig. 5C; Table S4).

The SypG-independent signaling pathway regulates bcs locus transcription through the transcriptional regulator VpsR independent of SypF
The detection of SypG-independent RscS regulation suggested that RscS was not limited to being simply a regulator of the syp locus.However, the factors involved in transmitting RscS signaling to alternative pathways remained unclear.One possibility was that RscS might branch a new signaling pathway through its partner sensor kinase SypF.SypF has precedent as a sensor kinase with multiple downstream response regulators, affecting both SypE and SypG within the Syp phosphorelay (27,29).Additionally, an increased activity allele of SypF (SypF1) was discovered to increase production of the cellulose exopolysaccharide (26,54).This activity required the gene VF_0454, which encodes a homolog of the Vibrio cholerae transcriptional regulator VpsR (26,54).As we observed robust induction of the cellulose exopolysaccharide (bcs) locus in the rscS* ΔsypG background (vs wild type) and in the core biofilm regulon, we hypothesized that SypFmediated VpsR signaling could explain the SypG-independent signaling pathway.To test this hypothesis, we asked whether transcription of SypG-independent gene bcsE was SypF and VpsR dependent.Using fluorescence microscopy, we measured reporter expression in wrinkled colony biofilms grown on LBS or LBS-Ca 2+ agar.While removal of SypF or SypG exhibited mild effects (up to 18%), only removal of VpsR eliminated the induction of the bcsE'-gfp + reporter (Fig. 6A and B).Testing another SypG-independent promoter, for VF_0208, revealed roles for SypF, SypG, and VpsR, yet again VpsR exhibited the strongest impact on the promoter's activity (Fig. 6C and D).Together, these data suggest that VpsR plays an important role in the regulation of genes in the rscS* model.

DISCUSSION
In this work, we profiled two separate models of biofilm formation in the model symbiont V. fischeri ES114.This analysis revealed remarkable similarities between transcriptional responses to the two models, which facilitated identification of a core transcriptional program.In addition to known factors that were among the highest expressed in both models (e.g., syp locus genes), we identified novel regulatory targets that included well-annotated genes (e.g., motility), loosely annotated genes (e.g., csg and VF_0157-VF_0180 EPS), as well as many hypothetical and unannotated genes.This work generated testable hypotheses, and we pursued multiple examples of those within the study.We validated that in both models, biofilm formation resulted in diminished swimming motility, consistent with behavior in other organisms in which biofilm and motility are discordantly regulated (56,57).
Our study provided insights into the quantitative nature of induction dynamics across biofilm operons.We observed the syp locus had dramatically higher induction for genes located earlier in each of the four operons, as seen for sypA, sypI, sypM, and sypP (Fig. 2D).This result suggests that sypA provides especially high dynamic range to use for reporting on core biofilm responses in strain ES114, consistent with previous studies that have applied sypA′-gfp + and sypA′-lacZ + transcriptional reporters (27,31,34,35,41).In uninduced conditions, we found that the genes encoding the central regulators of the Syp phosphorelay, sypF and sypG, had the highest abundance as measured by TPM among the syp gene transcripts.This result suggested to us that these transcripts may be regulated in a fashion uncoupled from the preceding genes.We speculate that there is a separate regulatory mechanism that enables a higher baseline of sypF and sypG transcripts without full syp locus induction, and that allows levels of SypF and SypG to accumulate to be able to respond to physiological induction of the pathway.A separate, SypG-independent promoter upstream of sypF would be consistent with the pattern that we observed, and such a promoter was in fact proposed in early models of the syp locus (22,26).In a recent study, induction of sypF-H in response to the biofilm stimulatory vitamin para-aminobenzoic acid was observed in a pattern distinct from the rest of the locus (32), supporting a separate regulatory mechanism that begins at sypF.Among the most highly expressed genes in our biofilm models, we noted the presence of a predicted curli amyloid fiber biosynthetic locus.In other organisms, curli fibers play critical roles in biofilm formation, aiding in cell-surface attachment and cell-cell aggregation (45,58,59).However, we noted that this locus was dispensable during host colonization and had no observable effect in culture biofilm models or in vivo biofilm aggregates.Previous literature has indicated that curli functionality is commonly aided by the presence of cellulose exopolysaccharide production, which is typically regulated in tandem with curli (50,58,60).In our transcriptional data, we noted that the V. fischeri cellulose (bcs) and curli loci appeared to be coregulated and therefore generated a double mutant in both pathways, in the case that cellulose and curli had redundant roles in biofilm formation.However, this double mutant failed to impact wrinkled colony formation, suggesting that the functional role of curli may not have been captured in our experiments.Considering that curli do not appear to be required for host colonization, it is possible that curli may be important for V. fischeri survival in the marine environment in its free-living state.One possible benefit of curli production in this context may arise from the ability of curli fibers to protect producing cells from bacteriophage predation, especially within biofilms (61).Given the abundance of bacteriophage in the marine environment (62), curli production may therefore prove to be vital in the natural context but dispensable in the laboratory environment.Outside of phage defense, the basic nature of curli as adhesins may also aid V. fischeri in adhering to biotic and abiotic surfaces in the water column, a common behavior observed in many Vibrio species (63)(64)(65).Further study is needed to identify the importance of curli to V. fischeri biology, especially in the context of biofilm formation.
The large number of genes induced in the rscS* ΔsypG strain was surprising, given that the strain does not induce biofilm formation in culture.This result suggested to us that there is a substantial novel output to the RscS signaling pathway, and in the two promoters we examined, we found that VpsR was required for activity in both cases.Mutants in vpsR exhibit a competitive defect in squid colonization and a defect in cellulose regulation in culture (26,66).Despite characterized VpsR regulation via SypF, in the rscS* model, we observed a substantial role for VpsR even when there was only a modest role for SypF (Fig. 6A and B).We note that additional work on VpsR has been conducted in V. cholerae, which has a protein that is 66% identical to the V. fischeri ortholog.There, it has been shown to be regulated by phosphate and not by phosphor ylation as would be expected of a putative response regulator (67).Furthermore, in V. cholerae, VpsR regulates the Vibrio polysaccharide locus (vps), which is absent in V. fischeri, as similarly the bcs locus is absent in V. cholerae (26).V. cholerae VpsR binding sites have been determined ( 68), but we were unable to locate similar sites upstream of genes in our data set, suggesting that V. fischeri VpsR likely binds to a different consensus sequence.We also note that in Vibrio vulnificus, the VpsR ortholog BrpR regulates the brp EPS gene cluster in concert with BrpT, providing additional diversity to VpsR regulation across the Vibrionaceae (69).Overall, therefore, the role of VpsR in V. fischeri specifically and vibrios broadly requires further clarification.
Overall, this work combines two models to identify novel aspects of biofilm regula tion, reveal patterns of gene expression across regulated loci, and uncover new factors that are coregulated with the V. fischeri symbiotic biofilm program.Our results can help to guide selection of an appropriate model given a question under consideration.The overall similarity observed suggests that most effects that have been described in the heavily studied rscS* model will carry over to the ∆binK-Ca 2+ model, which is helpful, given that some phylogenetic groups of V. fischeri do not encode RscS (if researchers do not wish to overexpress a heterologous protein in those strains) (23,24).Given that ∆binK-Ca 2+ model substantially induced c-di-GMP levels, whereas the rscS* model does not, this is another consideration that may guide choice of model.Finally, we noted specific regulatory groups for which the two models differed from the core biofilm response.Future work can leverage these baseline data on the core response and on model-specific effects to examine temporal regulation of biofilm formation, biofilm dispersal, and regulation of biofilm behaviors during host colonization.

Bacterial strains, plasmids, and media
V. fischeri and E. coli strains studied in this work are listed in Table 1.Plasmids generated or used in this work can be found in Table 2. Unless otherwise specified, V. fischeri strains were grown at 25°C in LBS medium (per liter: 25 g of Difco Miller LB broth [BD], 10 g of NaCl, 50 mL of 1 M Tris buffer [pH 7.5], and 10 mM CaCl 2 [when noted]) or TBS medium (per liter: 10 g of Gibco Bacto tryptone, 20 g of NaCl, 50 mL of 1 M Tris buffer (pH 7.5), and 3 g agar).E. coli strains used for cloning and conjugation were grown at 37°C in Luria-Bertani medium (LB; per liter: 25 g of Difco Miller LB broth [BD]) or brain heart infusion medium (BHI; per liter: 37 g brain heart infusion powder [BD]).When needed, antibiotics were supplemented at the following concentrations: kanamycin (Kan), 100 µg/mL for V. fischeri and 50 µg/mL for E. coli; chloramphenicol (Cam), 5 µg/mL for V. fischeri and 25 µg/mL for E. coli; erythromycin (Erm), 5 µg/mL for V. fischeri and 150 μg/mL for E. coli; gentamicin (Gent), 2.5 µg/mL for V. fischeri and 5 µg/mL for E. coli.When needed for specific strains, thymidine was added at 0.3 mM for E. coli.Solidified media were prepared with an agar concentration of 1.5% unless specified otherwise.For Congo Red agar, 40 µg/mL Congo Red and 15 µg/mL Coomassie blue were added to LBS.Plasmids were conjugated from E. coli strains into V. fischeri using standard techniques (70).

DNA synthesis and sequencing
Primers used in this work are provided in Table 3 and were synthesized by Integrated DNA Technologies (Coralville, IA).PCR products of gene deletions and plasmid modifications were validated using Sanger sequencing at Functional Biosciences (Madison, WI).Sequencing results were analyzed using Benchling.All Sanger sequencing products were generated with Q5 High-Fidelity polymerase (NEB), Phusion Hot Start Flex DNA

RNA isolation from V. fischeri wrinkled colonies
Strains MJM1100, MJM1198, MJM2251, and MJM3259 were streaked on LBS agar and grown overnight at 25°C (30,36,71,72).Overnight cultures were prepared in triplicate from independent colonies of each strain in LBS media, and grown overnight at 25°C with rotation.From each overnight culture, 500 µL was spotted onto LBS agar, or LBS-Ca 2+ agar, and spots were allowed to grow for 48 hours at 25°C.After 48 hours, each spot was collected into a 1:2 mix of LBS and RNAprotect Bacteria reagent (Qiagen), incubated at RT for 5 minutes, followed by vortexing at maximum speed on a Vortex Genie-2 (Scientific Industries), followed by centrifugation for 10 minutes at 5,000 x g at room temperature.Supernatants were removed, and cell pellets were stored at −80°C.RNA extraction was performed using an RNeasy PowerBiofilm kit (Qiagen), as per manufacturer instructions.Purified RNA was stored at −80°C, followed by a further on-column DNAse treatment using RNAse-free DNAse (Qiagen) followed by purification with the RNeasy MinElute kit (Qiagen).Samples were then transferred to the University of Wisconsin -Madison Biotechnology center.

Library preparation, sequencing, and analysis of RNA samples
Samples were quality-tested using both NanoDrop and Agilent Bioanalyzer measure ments.Ribosomal RNA depletion with additional probes recommended by Illumina (Table 4), stranded library preparation (Illumina Ribo-Zero Plus rRNA Depletion w/ Stranded Total RNA), and paired-end 2 × 150 sequencing was conducted at the UW-Madison Biotechnology center on an Illumina NovaSeq 6000.Table S5 shows quality control data for these samples.The resulting reads were processed by FASTP v0.20.1 (80), mapped with BWA-MEM v0.7.17 (81), read counts per gene were enumerated with HTSeq v0.13.5 (82) to V. fischeri ES114 chromosome 1, 2, and the natural plasmid pES100 (CP000020.2,CP000021.2,and CP000022.1),and differential expression analysis was conducted with EdgeR v3.32.1 (83).Follow-up analysis of the data set was performed in RStudio v1.3.959.Volcano plots of data were produced through plotting the −log 10 (FDR) on the graph y-axis and the log 2 (fold change) on the x-axis, from the EdgeR results.Overlay graphs of multiple differential expression analyses were generated by plotting the log 2 (fold change) for each gene in the stated differential expression analyses on either the xor y-axis.Transcripts per kilobase million (TPM) values for all genes were obtained from the raw read counts for all genes from HT-Seq output (39).The read counts for each gene were then divided by each gene's respective length in kilobase pairs to generate reads per kilobase (RPK).These values were summed for each of the replicates and divided by 1,000,000 to generate the scaling factor for each replicate.The RPK for each gene was then divided by its respective replicate scaling factor to generate per gene TPM values.TPM values for all conditions can be found in Table S6.

Construction of pTM267-based transcriptional reporters (pJVG11 and pJVG25)
Reporter plasmids were created through insertion of gene specific promoter elements into the established transcriptional reporter plasmid pTM267 (78).In short, ~500 bp of upstream DNA preceding the start codon for bcsE and VF_0208 were amplified using Q5 DNA Polymerase (NEB) using primers JVG_BC1 and JVG_BC2 (bcsE) and JVG_BM1 and JVG_BM2 (VF_0208).Each primer set appended XbaI and XmaI restriction sites to promoter amplicons.Inserts and the pTM267 backbone were then digested with XbaI and XmaI (NEB), gel purified using the QIAquick gel extraction kit (QIAgen), and further concentrated using a DNA Clean and Concentrator kit (Zymo Research).Purified inserts and pTM267 backbone were then ligated together using overnight ligation at 16°C with T4 DNA ligase (NEB).Ligation products were then transformed into chemically competent E. coli DH5ɑ λpir, and chloramphenicol resistant colonies were selected from plates.Potential candidates were streak purified, and used in colony PCR using primers JVG_AA5 and JVG_AA6 to amplify across the inserted element.Candidates showing the proper size amplicons were miniprepped and either sent for Sanger sequencing using primers JVG_AA5 and JVG_AA6 (pJVG11-bcsE) (Functional Biosciences) or for full plasmid sequencing (pJVG25-VF_0208) (Plasmidsaurus).

Construction of gene::bar tagged gene deletions
The creation of Δgene::bar mutants followed previously published protocols (84).In short, 1 kb regions flanking the gene(s) of interest were amplified from MJM1100 genomic DNA using Phusion Hot Start Flex polymerase (Thermo Fisher), and each flanking element shared terminal homology with an erythromycin resistance cassette amplified from plasmid pHB01 using primers HB42 and HB154.Splicing by over lap extension PCR (SOE-PCR) was then used to join the flanking regions onto the Erm R cassette (84).This joined fragment was used as mutagenic DNA in a natural transformation into V. fischeri strains containing the pLostfoX or pLostfoX-Kan plasmid (73).Recombinants were selected for erythromycin resistance, streak purified, and PCR assays were used to identify candidates with insertions of the erythromycin resistance cassette in the place of targeted genes.For candidates passing PCR screens, the gene deletion and adjacent upstream and downstream regions were PCR amplified, and amplicons were sent for Sanger sequencing using primers HB08, HB09, HB42, and HB146.
Candidates which successfully replaced the target gene with the erythromycin cassette were then saved as Δgene::erm-bar strains.In order to eliminate potential influences of the erythromycin cassette on neighboring genes, the flippase-encoding plasmid pKV496 was conjugated into Δgene::erm-bar strains via tri-parental mating with strains MJM534 and MJM3478 (76).Following this conjugation, candidates were validated for loss of the erythromycin cassette using PCR with primers flanking the deletion site, and amplicons were sent for Sanger sequencing for final validation.Candidates passing the deletion check were then saved as Δgene::bar strains.

Natural transformation of V. fischeri with genomic DNA
When necessary, plasmids pLostfoX or pLostfoX-Kan were conjugated into V. fischeri strains to induce competence to transfer alleles between strains (73,79).Natural transformation was conducted as described previously (84), in which strains carrying pLostfoX or pLostfoX-Kan were grown from glycerol stocks in LBS with their respective antibiotics.After overnight growth in LBS, a 1/100 subculture was performed into Tris minimal media (TMM) containing the antibiotic of note, with further growth overnight.
The following morning, the TMM overnight culture was subcultured 1/20 into fresh TMM and allowed to grow to OD600 0.2.This culture was then separated into 500 µL aliquots, and 2.4 µg genomic DNA was then provided to the aliquots as needed, and the cell/DNA mixtures were gently vortexed before resting at room temperature for 30 minutes.Following room temperature incubation, each aliquot was provided with 1 mL of LBS, and allowed to recover overnight at 25°C.Recovery cultures were then plated on LBS agar containing respective antibiotics, and candidates of note were streak purified for downstream confirmation.

Construction of Tn7 transposon-based chromosomal complementations of sypF, sypG, and vpsR
Chromosomal complementations of Δgene::bar mutations were conducted through provisioning of the gene of interest on a Tn7 transposon.For the ΔvpsR::bar strain, vpsR ± ~ 350 bp was amplified from MJM1100 gDNA using primers RYI643 and RYI644, and the Tn7 transposon vector pEVS107 was linearized using primers RYI446 and RYI447 with Q5 DNA polymerase (NEB) (75).Using the NEB HiFi DNA assembly kit, the vpsR insert and pEVS107 backbone were joined, and the products of this reaction were transformed into E. coli DH5ɑ λpir, and erythromycin resistant colonies were selected.Candidates were streak purified, plasmids were isolated using the QIAprep Spin Miniprep kit (Qiagen) and submitted to Plasmidsaurus for full plasmid sequencing, yielding plasmid pKK01.
For the ΔsypF and ΔsypG complementations, the ORFs for each gene were amplified with JVG_BK7 and JVG_BK8 (sypF) and JVG_BK11 and JVG_BK12 (sypG) with Q5 DNA polymerase (NEB).A variant of pEVS107 containing the constitutively active promoter for the nrdR gene (pKMB036) was linearized using primers containing homology for sypF (JVG_BK5 and JVG_BK6) and sypG (JVG_BK9 and JVG_BK10), and the plasmid backbone and inserts were joined by the NEB HiFi DNA assembly kit.Candidates were selected as erythromycin resistant colonies, and streak purified.Successful assemblies were confirmed using primers JVG_BK13 and JVG_BK14 across the insertion site using colony PCR, plasmids were isolated using the QIAprep Spin Miniprep kit (Qiagen), and purified plasmids were sent to Plasmidsaurus for full plasmid sequencing, yielding plasmids pJVG22 and pJVG23.

Construction of rscS* allelic exchange plasmid (pJVG21)
Plasmid pEVS79 (70) modified with Erm R replacing Cam R (pEVS79-Erm, Matthew Hauserman) was amplified using primers JVG_BJ1 and JVG_BJ2, and the intergenic region containing the rscS* allele was amplified from strain MJM1198 using primers JVG_BJ3 and JVG_BJ4.Products of the JVG_BJ1 and JVG_BJ2 reaction were digested prior to assembly using Dpn1 (NEB).The NEBuilder HiFi DNA Assembly kit (NEB) was then used to assemble JVG_BJ1/2 and JVG_BJ3/4, products of this reaction were then transformed into E. coli DH5α using heat shock, and transformants were selected on BHI-Erm150 media.Candidates were screened using PCR with primers JVG_BJ5 and JVG_BJ6, passing candidates were then submitted to Plasmidsaurus (Eugene, OR) for whole-plasmid sequencing yielding plasmid pJVG21.

Wrinkled colony assays
Bacterial strains were grown in LBS medium for approximately 17 hours, and 8 µL of each overnight culture was then spotted onto LBS or LBS-Ca 2+ plates, and incubated at 25°C.
At 24 and 48 hours post-spotting, plates were imaged using a Leica M60 microscope and Leica DFC295 camera.

Plasmid-based expression assay
Bacterial strains were grown in LBS-Cam medium for approximately 17 hours at 25°C, and 8 µL of each overnight culture was then spotted onto LBS-Cam and LBS-Cam-Ca 2+ agar, followed by incubation at 25°C.After 24 hours, colony spots were imaged on a Zeiss Axio Zoom v.16 large-field 561 fluorescent stereo microscope, and fluorescence levels were measured within colony spots using the Zen Blue software.Expression of the reporter (eGFP fluorescence, 488 nm excitation, 509 nm emission, and 300 ms expo sure time) was normalized to a constitutive locus on the pTM267 backbone (mCherry fluorescence, 558 nm excitation, 583 nm emission, and 700 ms exposure time) as eGFP fluorescence/mCherry fluorescence.Data analysis was conducted in GraphPad Prism using an ordinary one-way analysis of variance (ANOVA) with multiple comparisons.

Soft agar swimming motility assay
Bacterial strains were streaked on TBS agar 48 hours prior to the assay and incubated at 25°C.After 24 hours, four single colonies were cultured into TBS medium and grown at 25°C for 17 hours with rotation.Ten microliters of each overnight culture was then spotted onto TBS motility media (3 g/L agar) in technical triplicate and allowed to incubate at 25°C.After 5 hours, plates were imaged using a Nikon D810 digital cam era, and ImageJ was used to quantify the Feret diameter of each migration spot as the migration distance.Each biological replicate was produced through averaging the technical replicates.Data analysis was conducted in GraphPad Prism using an ordinary one-way ANOVA with multiple comparisons.This assay was conducted on two separate days for a total of eight biological replicates.

Congo Red assay
Bacterial strains were grown in LBS medium for 17 hours, and 8 µL of each overnight culture was spotted on LBS Congo Red media.Plates were incubated overnight at 25°C for 24 hours, after which colony spots were transferred to printer paper, air dried, and imaged as TIFF files (76).

Cyclic di-GMP reporter assay
Bacterial strains were grown in LBS-Gent at 25°C for 16 hours, and 8 µL of each overnight culture was spotted in technical duplicate on LBS-Gent and LBS-Gent-Ca 2+ plates.After 24 hours, colony spots were imaged on a Zeiss Axio Zoom v.16 large-field 561 fluorescent stereo microscope, and fluorescence levels were measured within colony spots using the Zen Blue software.Expression of the reporter (TurboRFP fluorescence, 553 nm excitation, 573 nm emission, and 150 ms exposure time) was normalized to a constitutive locus on the pFY4535 backbone (AmCyan1 fluorescence, 467 nm excitation, 496 nm emission, and 150 ms exposure time) as TurboRFP fluorescence/AmCyan1 fluorescence.Data analysis was conducted in GraphPad Prism using an ordinary one-way ANOVA with multiple comparisons.

Curli mutant squid single-strain colonization assay
Lab-reared E. scolopes hatchlings were colonized using a previously published protocol (85).In short, 3-5 × 10 3 CFUs of the wild-type strain MJM1100 or the ΔcsgBA mutant strain MJM4185 were provided to squid hatchlings, followed by a 3 hour colonization period.After 48 hours, hatchlings were measured for luminescence using a Promega GloMax 20/20 luminometer and were euthanized.Homogenization of squid tissues was carried out as in the protocol, and colonization levels were enumerated from CFU per light organ plated on LBS agar.This experiment was conducted in biological triplicate.
A Mann-Whitney test was then used to determine statistical significance between the colonization levels of MJM4185 and wild type in GraphPad Prism.

Curli mutant squid aggregation assay
Squid aggregation experiments were performed as described previously (86).In short, cultures of MJM1107 and MJM4276 were grown overnight at 25°C in LBS-Kan (77).The following morning, 10 5 -10 6 CFU/mL of each strain was added to bowls of E. scolopes hatchlings and allowed to colonize for 3 hours.Following this colonization period, hatchlings were anesthetized in 2% ethanol/FSIO for 5 minutes before fixation in 4% paraformaldehyde at 4°C for 48 hours.Following fixation, hatchlings were washed four times with 1× marine phosphate-buffered saline (50 mM sodium phosphate, 0.45 M NaCl [pH 7.4]) and dissected, and imaged with a Zeiss Axio Zoom v.16 large-field 561 fluorescent stereo microscope.Aggregate sizes were determined using the aggregate area tool, and statistical analysis was conducted using the Mann-Whitney test on GraphPad Prism.

FIG 2
FIG 2 Vibrio fischeri induces a core transcriptomic response during biofilm formation.(A) Volcano plots of the rscS* vs wild type and ΔbinK-Ca 2+ vs ΔbinK biofilm models.Green and red boxes indicate those genes with log 2 (fold change) ≥1 and ≤−1, respectively, with significant differential expression (FDR < 0.05).(B) Overlay of log 2 (fold change) values for all genes from both the rscS* vs wild type and ΔbinK-Ca 2+ vs ΔbinK biofilm models.Green and red boxes are analogous to those in panel A. (C) Overlay of annotations onto the multi-comparison overlay shown in panel B; dot legend is provided below.(D) Syp locus differential expression.Log 2 (fold change) for each of the 18 genes in the syp locus is provided from the two differential expression analyses.Operon indications are provided from previous analyses of SypG-bound promoters (28).For the one gene

FIG 3 A
FIG 3 A curli locus is highly induced by both biofilm models.(A) Overlay graph of Fig. 2C with specific curli genes (red), labeled by name.(B) Single-strain colonization data for wild type and a ΔcsgBA curli mutant.Wild type and ΔcsgBA were inoculated into filter sterilized Instant Ocean of two separate cohorts of squid hatchlings and allowed to colonize for 3 hours.Hatchlings were then washed and allowed to grow for 2 days, after which squid juveniles were euthanized.Each dot represents the average CFU per light organ of individual squid, and horizontal bars represent median CFU per light organ with 95% confidence intervals.The dashed line indicates the limit of detection.Statistical significance was calculated using a Mann-Whitney test.(C) Biofilm aggregate size during host colonization.Wild type and ΔcsgBA strains carrying the constitutive GFP plasmid pVSV102 were allowed to colonize squid hatchlings for 3 hours, at which point each squid was euthanized and fixed in 4% paraformaldehyde.Dissection of each squid then revealed the biofilm aggregates forming on the light organ surface, which were imaged with a Zeiss Axio Zoom fluorescence microscope.Each point represents the area of one aggregate.Squid which did not have any aggregates have points at 0. Horizontal bars represent median aggregate area size with 95% confidence interval.Statistical significance was calculated using a Mann-Whitney test.(D) Wrinkled colony and Congo Red binding assays.Strains were spotted on LBS media and imaged.ns, not significant.

FIG 4
FIG 4 Biofilm-induced conditions downregulate flagellar motility genes and have reduced motility.(A) Both biofilm-induced models significantly downregulate genes required for flagellar biosynthesis from Brennan et al. (51).(B) Strains were cultured in TBS media, spotted onto TBS soft-agar plates, and incubated at 25°C for 5 hours.Shown is a representative plate of n = 8 biological replicates, imaged with a Nikon D810 camera.(C) Migration distance determined using ImageJ.Statistical significance was calculated using an ordinary one-way analysis of variance using Dunnett's multiple comparisons test in GraphPad Prism.Each dot is the product of n = 3 technical replicates.Horizontal lines represent the mean with standard error of the mean.****P < 0.0001.

FIG 5
FIG 5 Deletion of the response regulator gene sypG reveals a signaling bifurcation downstream of RscS.(A) Wrinkled colony assay.(B) Volcano plot of wild type vs rscS* ΔsypG differential expression analysis.Green and red boxes demarcate genes with log 2 (fold change) ≥1 and ≤−1, respectively, with significant differential expression (FDR < 0.05).The numbers of genes in each box are listed in the figure.(C) Overlay of log 2 (fold change) values for all genes in both the rscS* vs wild type and the rscS* ΔsypG vs wild type models.

FIG 6
FIG6 RscS uses VpsR to regulate SypG-independent targets.(A) Expression data for the bcsE'-gfp + reporter for strains spotted on LBS media.Complementation backgrounds are indicated by the presence of +vpsR, +sypF, and +sypG; all complementa tions were single-copy insertions at the V. fischeri Tn7 transposon site.Fluorescence was measured within colony spots with n = 6 biological replicates, each the product of n = 2 technical replicates using a Zeiss Axio Zoom fluorescence microscope.GFP fluorescence was normalized to mCherry fluorescence to determine the expression of the reporter in each strain.Statistical analysis of each strain's reporter expression was conducted in GraphPad Prism, using an ordinary one-way analysis of variance, with Dunnett's multiple comparison test.**P ≤ 0.001, ****P < 0.0001.(B) Expression data for the bcsE'-gfp + reporter for strains spotted on LBS-Ca 2+ media, collected and analyzed as in panel A. (C) Expression data for the VF_0208'-gfp + reporter for strains spotted on LBS media, collected and analyzed as in panel A. (D) Expression data for the VF_0208'-gfp + reporter for strains spotted on LBS-Ca 2+ media, collected and analyzed as in panel A. ns, not significant.

TABLE 1
Bacterial strains used in this study(Continued)

TABLE 2
Plasmid list

TABLE 4
Supplemental rRNA depletion probes in pool vfischeri_RZP_sup_pool

rscS* into ΔsypF, ΔvpsR, ΔbcsA, and ΔcsgBA ΔbcsA
Erm plates to select for plasmid integration into the host genome.Candidates were streaked on LBS-Erm to purity, and individual colonies were patched on LBS-Erm, LBS-Cam, and LBS.Candidates showing Erm R , Cam R , and growth on LBS were then cultured in LBS-Erm and saved as single recombinants.Double recombinants were then generated as follows.Single recombinants were first cultured overnight in LBS; overnight cultures were then diluted by 1/100 into fresh LBS and allowed to grow until OD600 reached 0.2.At this point, a further subculture was prepared by repeating the 1/100 dilution into LBS and allowing the culture to grow once again to OD600 0.2.These subcultures were then plated on LBS media and allowed to grow overnight at 25°C.Resulting colonies were patched on LBS-Erm, LBS-Cam, and LBS and grown overnight at 25°C.Patches showing Erm S dropped the plasmid backbone, yet retained the Cam R of the rscS* allele were then cultured and saved as candidate double recombinants.