VpsR Directly Activates Transcription of Multiple Biofilm Genes in Vibrio cholerae

Vibrio cholerae, the bacterial pathogen that is responsible for the disease cholera, uses biofilms to aid in survival, dissemination, and persistence. VpsR, which directly senses the second messenger c-di-GMP, is a major regulator of this process. Together with c-di-GMP, VpsR directly activates transcription by RNA polymerase containing σ⁷⁰ from the vpsL biofilm biogenesis promoter. Using biochemical methods, we demonstrate for the first time that VpsR/c-di-GMP directly activates σ⁷⁰-RNA polymerase at the first genes of the vps and ribomatrix operons. In this regard, it functions as either a class I or class II activator. Our results broaden the mechanism of c-di-GMP-dependent transcription activation and the specific role of VpsR in biofilm formation.

ABSTRACT

Vibrio cholerae biofilm biogenesis, which is important for survival, dissemination, and persistence, requires multiple genes in the Vibrio polysaccharides (vps) operons I and II as well as the cluster of ribomatrix (rbm) genes. Transcriptional control of these genes is a complex process that requires several activators/repressors and the ubiquitous signaling molecule, cyclic di-GMP (c-di-GMP). Previously, we demonstrated that VpsR directly activates RNA polymerase containing σ⁷⁰ (σ⁷⁰-RNAP) at the vpsL promoter (P_vpsL), which precedes the vps-II operon, in a c-di-GMP-dependent manner by stimulating formation of the transcriptionally active, open complex. Using in vitro transcription, electrophoretic mobility shift assays, and DNase I footprinting, we show here that VpsR also directly activates σ⁷⁰-RNAP transcription from other promoters within the biofilm formation cluster, including P_vpsU, at the beginning of the vps-I operon, P_rbmA, at the start of the rbm cluster, and P_rbmF, which lies upstream of the divergent rbmF and rbmE genes. In this capacity, we find that VpsR is able to behave both as a class II activator, which functions immediately adjacent/overlapping the core promoter sequence (P_vpsL and P_vpsU), and as a class I activator, which functions farther upstream (P_rbmA and P_rbmF). Because these promoters vary in VpsR-DNA binding affinity in the absence and presence of c-di-GMP, we speculate that VpsR’s mechanism of activation is dependent on both the concentration of VpsR and the level of c-di-GMP to increase transcription, resulting in finely tuned regulation.

IMPORTANCE Vibrio cholerae, the bacterial pathogen that is responsible for the disease cholera, uses biofilms to aid in survival, dissemination, and persistence. VpsR, which directly senses the second messenger c-di-GMP, is a major regulator of this process. Together with c-di-GMP, VpsR directly activates transcription by RNA polymerase containing σ⁷⁰ from the vpsL biofilm biogenesis promoter. Using biochemical methods, we demonstrate for the first time that VpsR/c-di-GMP directly activates σ⁷⁰-RNA polymerase at the first genes of the vps and ribomatrix operons. In this regard, it functions as either a class I or class II activator. Our results broaden the mechanism of c-di-GMP-dependent transcription activation and the specific role of VpsR in biofilm formation.

INTRODUCTION

Vibrio cholerae, the causative agent of the enteric disease cholera, is responsible for 3 to 5 million infections as well as 100,000 to 120,000 deaths per year (1, 2). During its life cycle, V. cholerae lives in both the planktonic and biofilm state. In the human host and the aquatic reservoir, V. cholerae forms biofilms to aid in survival, transmission, and persistence (3–7). Not only do biofilms shield the bacteria from harsh environmental conditions and stresses, but they also protect the bacteria from protozoa and bacteriophage predation as well as nutrient limitation (3, 4, 8–10). Biofilms are comprised of matrix proteins, extracellular DNA, and Vibrio polysaccharides (VPS). Over 50% of the Vibrio biofilm matrix mass is comprised of VPS (11–16), which is excreted after initial attachment to surfaces (17). The biofilm structure modulator A protein, RbmA, and the biofilm matrix protein, Bap1, are also secreted from the cell surface, promoting cell-cell adhesion and cell-VPS adherence (14, 15, 17).

Genes involved in VPS synthesis are located on V. cholerae’s larger chromosome and organized into two operons, vps-I and vps-II (12, 13). vps-I contains 12 genes, including vpsU and vpsA to vpsK, while vps-II contains six genes, vpsL to vpsQ (12, 13). These two operons are separated by 8.3 kbp that contain six ribomatrix genes, rbmA, rbmC, and rbmBDEF, which also contribute to biofilm formation (Fig. 1A) (12, 14, 15).

FIG 1.

Vibrio cholerae genomic organization of biofilm biogenesis operons and sequences of their promoters. (A) VPS synthesis genes are comprised of two vps operons, vps-I (white arrows) and vps-II (black arrows). The ribomatrix protein production gene cluster is comprised of the rbm genes (gray arrows) and bap1 (not shown). Direction of transcription is indicated by the arrows. Smaller black arrows indicate the positions of the promoters upstream of vpsU, rbmA, rbmF, and vpsL. Image is approximately to scale. (B) Sequences of promoters P_rbmA, P_rbmF, P_vpsU, and P_vpsL. Determined VpsR protection sites observed with DNase I footprinting on the template strand are italicized in red with the match to the determined consensus sequence in bold; the predominant +1 TSSs from both V. cholerae RNA and in vitro transcription RNA are enlarged and in bold with a black arrow above. (Note that the match to the consensus for the VpsR binding site at vpsU is in the opposite orientation relative to the promoter.) Other 5′ ends for rbmA and rbmF RNA observed in vivo are underlined and in bold, and the other prominent 5′ end for vpsU RNA seen in vivo is green, in bold, and underlined. Assigned −10 elements are boxed with solid lines, and the −35 region located 17 bp upstream is shown in a dotted box. A potential extended −10 promoter for vpsU transcription from the alternate 5′ end is indicated by the green bar. Information for P_vpsL was determined previously and is included here for comparison (31).

Since biofilm formation is a complex process requiring the assembly of multiple gene products, it is not surprising that this process is highly regulated with numerous signaling pathways, transcriptional repressors and activators, and regulatory sRNAs (18). Among these effectors is the ubiquitous signaling molecule, cyclic di-GMP (c-di-GMP), a keystone regulator of biofilm formation in many bacterial species (19). c-di-GMP is formed from two GTP molecules by diguanylate cyclase enzymes that contain a GGDEF domain and then degraded into pGpG or two GMP molecules by phosphodiesterase enzymes containing either an EAL or HD-GYP domain, respectively. Generally, high levels of c-di-GMP increase biofilm formation, while low levels of c-di-GMP promote motility (20).

A major mechanism by which c-di-GMP controls biofilm formation and motility is through regulation of transcription (21). Bacterial RNA polymerase (RNAP), which is responsible for transcription, is comprised of five subunits (β, β′, two αs, and ω) and a specificity factor, σ (22). While σ⁷⁰ is the housekeeping σ factor that is needed for exponential growth, alternate σs are used under different conditions or times of stress (22). During transcription initiation, σ⁷⁰-RNAP first forms an unstable short-lived complex called the closed complex (RPc) (23, 24). Kinetically favorable isomerizations then transition RPc to the stable open complex (RPo) (25, 26). Within the RPo, major conformational changes occur within RNAP, the DNA bends approximately 90°, and the transcription bubble forms from −12/−11 to approximately +5 relative to the +1 transcriptional start site (TSS) (27). This bubble allows the single-stranded template DNA to descend into the active site for the start of transcription (27). Although RNAP alone can transcribe from some promoters, yielding a basal level of expression, many promoters require activators to work by stimulating RNAP directly or by acting as antirepressors of regulators that inhibit RNAP. Among the activators that directly stimulate σ⁷⁰-RNAP, many function by a class I or a class II mechanism or a combination of the two (22, 28). In class I transcription activation, activators bind to regions upstream of the core promoter, while in class II transcription activation, activators interact directly upstream or overlapping the −35 element (22, 28). Furthermore, class I activators typically help recruit RNAP, while class II activators both recruit and isomerize RNAP (22).

In V. cholerae, the transcriptional activators VpsR and VpsT are required to activate transcription from the biofilm formation operons (21, 29–36). Though both regulators bind c-di-GMP with a dissociation constant (K_d) of 1.6 μM and 3.2 μM for VpsR and VpsT, respectively (32, 35), the binding of c-di-GMP to VpsT increases its binding to DNA (30, 34, 35, 37), while the binding of c-di-GMP to VpsR has no effect on its affinity for the vpsL promoter (P_vpsL) (18, 31). This allows VpsT to work as an antirepressor of the histone-like repressor H-NS, which represses the vps-I and vps-II operons (38, 39). In contrast, we have previously shown that VpsR is a direct activator of P_vpsL (30–32). It functions as a class II activator by interacting with a region overlapping the −35 element to subsequently initiate transcription with c-di-GMP and σ⁷⁰-RNAP (31). c-di-GMP, together with σ⁷⁰-RNAP and VpsR, is required to generate the active transcription complex at P_vspL as demonstrated in both potassium permanganate footprinting and DNase I footprinting (31). This was a surprising finding given that based on amino acid homology, VpsR is classified as an enhancer-binding protein (EBP), which typically uses the alternate σ, σ⁵⁴, and ATP hydrolysis to activate transcription. However, VpsR is an atypical EBP. Not only does VpsR lack the GAFTGA motif responsible for interacting with σ⁵⁴, but VpsR is also missing the conserved residues within the Walker B domain involved in ATP hydrolysis. In vitro transcriptions demonstrate that VpsR does not require ATP hydrolysis to initiate transcription with σ⁷⁰-RNAP and c-di-GMP (31).

In addition to genes directly involved in biofilm formation in V. cholerae, VpsR, together with c-di-GMP, activates the expression of many other genes that are thought to be adaptive in a biofilm lifestyle. These genes include the transcriptional regulators vpsT, aphA, and tfoY as well as epsC, the first gene of the type II secretion system operon (32, 36, 40, 41). Although VpsR directly binds to the DNA upstream of these genes, the mechanism by which it activates their expression has not been elucidated.

In this study, we report that VpsR also directly activates P_rbmA, P_rbmF, and P_vpsU, making it a primary, direct regulator of promoters throughout the V. cholerae biofilm biogenesis operons. As we found with P_vpsL, this transcription requires σ⁷⁰-RNAP and c-di-GMP. Analogous to P_vpsL, VpsR binding at P_vpsU overlaps the −35 element, while the VpsR binding sites at P_rbmA and P_rbmF are located farther upstream from the core promoter. Though binding affinities vary in the presence and absence of c-di-GMP at each promoter, specific DNA-protein interactions do not change in the presence of the signaling molecule. Our findings demonstrate that VpsR is an adaptable, direct regulator, functioning as both a class I and class II transcription activator, to induce multiple promoters of genes essential for biofilm formation in V. cholerae.

RESULTS

VpsR binds the promoter regions of rbmA, rbmF, and vpsU with or without c-di-GMP.

Previously, VpsR has been shown to bind upstream of the aphA, epsC, vpsL, and vpsT genes in electrophoretic mobility shift assays (EMSAs) (18, 31, 32, 36, 40, 41), and in silico sequence analysis using MEME (multiple expected maximization for motif elicitation) has predicted other VpsR binding sites upstream of the following genes: vpsR, VCA0075, rbmA, vpsA, vpsU, bap1, cdgC, and rbmC (18). Using DNase I footprinting, two VpsR binding sites were also determined upstream of vpsL at positions −31 to −53 (proximal site) and −297 to −336 (distal site) relative to the vpsL promoter, P_vpsL (30). In previous work, we demonstrated that VpsR binding to the proximal site directly activates transcription in vitro from P_vpsL, while the presence of the distal site has no effect on P_vpsL transcription in the in vitro system (31). Consequently, given the location of the distal site just upstream of the divergent gene rbmF (Fig. 1), we speculated that this site might be needed for VpsR activation of rbmF transcription.

To extend this work, we investigated the possible binding of VpsR to the rbmA, vpsU, and rbmF sites using EMSAs (Fig. 2). At all three promoters, we observed the formation of multiple complexes, with larger complexes forming as more VpsR was added. This behavior can be indicative of protein-protein interactions or the ability of multiple VpsR proteins to bind to the DNA fragment. As DNase I footprints (detailed below) indicated only a single site of VpsR binding, it seems likely that these larger species represent protein-protein binding or aggregation. Consequently, it is important to note that we determined the apparent dissociation constant [K_d(app)] as the amount of VpsR needed to shift 50% of the free DNA. Thus, the K_d(app) was calculated by quantifying the amount of free DNA that has been lost rather than the amount that has been shifted.

FIG 2.

VpsR binds P_vpsU, P_rbmA, and P_rbmF with and without c-di-GMP. Representative EMSAs of ³²P-labeled DNA harboring −130 to +120 of P_vpsU (A), −223 to +84 of P_rbmA (B), or −228 to +112 of P_rbmF (C) relative to the +1 TSS with increasing VpsR concentrations from 0 to 2.5 μM (lanes 1 to 5; lanes 6 to 10) either in the absence of c-di-GMP (lanes 1 to 5) or in the presence of 50 μM c-di-GMP (lanes 6 to 10). Gray arrows indicate free DNA, while black arrows represent shifted protein-DNA complexes. Samples were incubated at room temperature for 10 min prior to electrophoresis on 5% TBE polyacrylamide gels. Apparent DNA-binding dissociation constants K_d(app) (bottom) were determined as the concentration of VpsR needed to retard 50% of the free DNA. Values from at least three EMSAs were analyzed using one-way analysis of variance (ANOVA) with Tukey’s HSD post hoc analysis (ns, not significant; *, P < 0.05).

As we had observed previously for the vpsL proximal site (31), the presence of c-di-GMP does not significantly affect VpsR binding at the vpsU site (Fig. 2A). However, for the rbmA and rbmF promoters, the presence of c-di-GMP improves VpsR binding affinity modestly (50%) and significantly (∼5-fold), respectively (Fig. 2B and C). Relative to the EMSAs for rbmA and vpsU promoters, the binding of VpsR to the rbmF promoter appeared to generate complexes that were less discrete, resulting in a smearing pattern. No kinetic data can be determined from our EMSAs, but we speculate that at this specific promoter, VpsR may be forming a less discrete complex due to a faster off rate.

Primer extension analyses define the 5′ ends of rbmA, rbmF, and vpsU transcripts in vivo.

Previously, differential RNA-sequencing (dRNA-seq) has been used to determine TSSs throughout the V. cholerae genome under low and high cell density (42). The TSSs for rbmA, rbmF, and vpsU were identified as the T, A, and G that are 110 bp upstream, 60 bp downstream, and 52 bp upstream from the translation start site, respectively (42). For vpsU, an additional TSS was identified as the A that is 315 bp upstream of the translation start site.

To compare and determine the 5′ ends of rbmA, rbmF, and vpsU in vivo transcripts, we performed primer extension analyses of RNA isolated from wild type (WT) or ΔvpsR grown in the presence or absence of c-di-GMP. We obtained major primer extension products corresponding to the A 18 bp upstream of the TTG start codon of rbmA (Fig. 3A, lane 3), to the A 45 bp upstream of the ATG start codon of rbmF (Fig. 3B, lane 3), and to the T 53 bp upstream of the ATG start codon of vpsU (Fig. 3C, lane 3). These products were not seen using RNA isolated from the ΔvpsR mutant or under unaltered c-di-GMP levels (Fig. 3, lanes 1, 2, and 4), indicating that their presence is dependent on the presence of both VpsR and c-di-GMP. As detailed below, further in vitro analyses were consistent with assigning these ends as TSSs for the promoters for rbmA (P_rbmA), rbmF (P_rbmF), and vpsU (P_vpsU). In each case, analysis of the DNA sequence upstream of the determined TSS indicated the presence of a suitable −10 element (consensus, ⁻¹²TAtaaT⁻⁷ [capital letters indicate a greater level of consensus]) or extended −10 element for P_rbmF (consensus, ⁻¹⁵TGnTAtaaT⁻⁷) at a reasonable distance from the TSS (Fig. 1B, boxes). Not surprisingly, given their need for a transcriptional activator, none of the promoters contained a strong match to consensus within the −35 region. All lacked the very highly conserved “TT” within the ⁻³⁵TTGaca⁻³⁰ sequence. (Consensus sequences represent matches to the derived E. coli σ⁷⁰-RNAP consensus [25]; −10 and −35 element sequences derived from chromatin immunoprecipitation sequencing [ChIP-seq] analyses of 497 σ⁷⁰-binding sites in V. cholerae [43] are nearly identical [⁻¹²TAnAAT⁻⁷ and ⁻³⁵TTGnaa⁻³⁰, respectively].) Though our primer extensions identified TSSs that are different from those determined from dRNA-seq (42), we are confident in our analyses, as our in vivo TSSs are consistent with our in vitro TSSs (below). We suggest that the differences observed for rbmF and vpsU could be due to RNA processing in vivo.

FIG 3.

Determination of +1 TSS in vivo and in vitro for P_rbmA (A), P_rbmF (B), and P_vpsU (C). Primer extension analyses were performed with ³²P-labeled primer that hybridized 84, 112, and 120 bp downstream of the rbmA, rbmF, and vpsU TSSs, respectively. RNA was isolated from V. cholerae (lanes 1 to 4) or in vitro transcriptions (lanes 5 to 8). GATC lanes are sequencing lanes performed using the end-labeled DNA. Black arrows designate the major 5′ ends seen in vivo and in vitro; the gray arrow designates an alternate 5′ end observed for vpsU both in vivo and in vitro. The assigned TSS for each gene is shown in Fig. 1B.

Another prominent vpsU 5′ end dependent on both VpsR and c-di-GMP was also observed in vivo, 10 bp upstream of the P_vpsU start. This upstream TSS is preceded by a good match to an extended −10 promoter sequence (designated by the green rectangle in Fig. 1B), suggesting that another P_vpsU promoter may also be active in cells.

In the presence of c-di-GMP, VpsR activates transcription from P_rbmA, P_rbmF, and P_vpsU in vitro.

A previous microarray study demonstrated that rbmA, rbmF, and vpsU were downregulated 7.6-fold, 7.3-fold, and 10.6-fold, respectively, in a V. cholerae vpsR deletion strain compared to the rugose parent strain, indicating in vivo regulation by VpsR (44). However, such regulation could be direct or indirect. Direct regulation by VpsR and c-di-GMP at these genes has not been demonstrated (29). To determine if VpsR and c-di-GMP directly activate transcription of rbmA, rbmF, and vpsU, we constructed supercoiled in vitro transcription templates containing P_rbmA, P_rbmF, and P_vpsU located upstream of the rrnBP1 terminator. In vitro transcription analyses indicated that in the presence of both VpsR and c-di-GMP, a unique transcript was significantly elevated for each template (Fig. 4, lanes 4). The activated transcription from P_rbmA, P_rbmF, and P_vpsU was ∼3-fold, 4-fold, and 17-fold over the basal level obtained with RNAP alone (lanes 1) or RNAP with either 25 μM c-di-GMP (lanes 2) or VpsR (lanes 3). Since H-NS represses basal level transcription in vivo (39), we predict these fold increases to be lower in vitro due to the absence of H-NS. This prediction is further supported by our observation that there is no detectable basal level of transcription from these promoters in vivo as observed in the primer extension analyses (Fig. 3).

FIG 4.

VpsR and c-di-GMP activate transcription from P_rbmA, P_rbmF, and P_vpsU in vitro. Representative images of single round in vitro transcription from supercoiled plasmid templates containing P_rbmA, P_rbmF, and P_vpsU with RNAP alone (lanes 1), RNAP and 25 μM c-di-GMP (lanes 2), RNAP and VpsR (lanes 3), and RNAP and VpsR and 25 μM c-di-GMP (lanes 4). The activation fold change (average from 3 reactions) for RNAP, VpsR, and 25 μM c-di-GMP relative to basal level expression of RNAP alone is indicated below the image.

Besides P_rbmA, P_rbmF, and P_vpsU, we also investigated whether VpsR and c-di-GMP could activate σ⁷⁰-RNAP transcription at other putative VpsR-regulated promoters. Previous studies using EMSAs have shown VpsR binding sites upstream of the TSSs at P_aphA, P_vpsT, and P_epsC (32, 36, 41). However, for these three promoters, we were unable to detect transcription above the basal level with VpsR and c-di-GMP (see Fig. S1 in the supplemental material).

To compare the 5′ ends of the in vitro RNAs generated from the P_rbmA, P_rbmF, and P_vpsU templates to the ends observed in vivo, we again used primer extension analyses. In each case, we obtained major products that were identical to those found in vivo (Fig. 3, lanes 8, black arrows). In addition, a minor product observed using vpsU in vitro RNA corresponded to the alternate TSS seen in vivo (Fig. 3C, lanes 3 and 5 to 8, gray arrow). The difference in intensity of this band when using in vitro RNA versus what was observed using in vivo RNA suggests that perhaps different conditions or other factors may contribute to its activity in vivo, allowing for more regulation at this promoter.

To determine the position of VpsR binding relative to the TSS, we performed DNase I footprints. We found that VpsR protected rbmA from −83 to −57, rbmF from −100 to −66, and vpsU from −52 to −31 relative to the TSS of each promoter (Fig. 1B and Fig. 5A). Although the presence of c-di-GMP improved the DNA binding affinity of VpsR to P_rbmA and P_rbmF (Fig. 2), the DNase I footprints did not reveal any differences in the protein-DNA contacts in the presence or absence of c-di-GMP. Analysis of the VpsR protection sites relative to the TSS for each promoter indicated that for P_rbmA and P_rbmF, the sites occur >50 bp upstream of the TSS, while for P_vpsU, the VpsR binding site is closer to the core promoter, like P_vpsL (Fig. 1B) (31).

FIG 5.

DNase I footprinting of VpsR with and without c-di-GMP at P_rbmA, P_rbmF, and P_vpsU. (A) DNase I footprints of the template strand were obtained using DNA alone or DNA incubated with VpsR without or with 50 μM c-di-GMP as indicated. VpsR protection is indicated to the right of each image with a dotted line. (B) Sequences of determined VpsR protection sites at P_rbmA, P_rbmF, P_vpsU, and P_vpsL aligned to indicate sequence similarity. The consensus sequence underneath was determined from the presence of a base in at least 3 out of the 4 binding sites. A previously reported Logo sequence (30) is shown for comparison.

Alignment of the protection sites for these four promoters revealed a conserved sequence of 5’ tGtcTtanaaTTGA 3′ (Fig. 5B), in which the lowercase letters correspond to a 3 out of 4 match at a particular base among the sites, and the capital letters correspond to a perfect match. We assigned this sequence as the VpsR binding site for these promoters. While this 5′→3′ sequence is found on the nontemplate strand for P_rbmA, P_rbmF, and P_vpsL, it occurs on the template strand for P_vpsU and thus in the opposite direction relative to the promoter. Comparison of the consensus to a previously determined Logo, compiled from both DNase I footprinting at P_vpsL and in silico data (30), indicated a sequence similarity (Fig. 5B). However, the Logo sequence match begins two bases downstream of the consensus sequence determined from our footprints and extends two bases farther downstream. Because of this, the Logo implies an imperfect palindrome of tcTcAnnnnTgAga, whereas our consensus lacks such a palindrome. A clearer elucidation of the specific bases required for VpsR binding awaits the determination of more binding sites and biochemical/structural analyses.

To determine the protein-DNA contacts within the activated transcription complex at these promoters, we performed DNase I footprints of the RNAP/VpsR/c-di-GMP complexes. To ensure that we were observing footprints of the activated transcription complex, we challenged complexes with heparin to eliminate unstable complexes, treated complexes with DNase I, and then isolated the stable transcription complexes from EMSAs prior to purifying the DNA. This procedure enabled us to also remove all uncut background DNA.

In the presence of both VpsR and c-di-GMP, both P_rbmF and P_vpsU generated footprints expected for an activated transcription complex. At P_rbmF, the footprint extended from approximately −85 to +22 with hypersensitivity sites at −36 and −55, consistent with the VpsR binding site centered at −78.5 together with RNAP binding at the core promoter sequence (Fig. 6). At P_vpsU, protection extended from −57 to +15, which included the VpsR binding site centered at −40.5 and protection by RNAP downstream; hypersensitivity sites were observed at −26 and −61 (Fig. 6). At either of these promoters, we also observed protection in the presence of RNAP alone (lane 2), consistent with basal transcription that was observed from these promoters in the absence of VpsR. However, within these regions, the patterns obtained by RNAP alone (lane 2) differed from those obtained in the presence of VpsR/c-di-GMP (lane 3). In particular, the strong hypersensitivity sites observed in the presence of VpsR were not seen with RNAP alone. These results indicate that the activated complex contains a different protein-DNA architecture than what is formed by RNAP alone. Unlike P_rbmF and P_vpsU, we were unable to obtain clear DNase I footprints for VpsR/c-di-GMP and RNAP at P_rbmA despite multiple attempts using either template-labeled or non-template-labeled DNA. This result suggests that the activated transcription complex at P_rbmA may be unusually unstable.

FIG 6.

DNase I footprints (template strand) of complexes containing RNAP alone or RNAP with VpsR and c-di-GMP at P_rbmF and P_vpsU. GA represents G+A ladder. Lanes 1 contain DNA only, lanes 2 contain RNAP, and lanes 3 contain RNAP, VpsR, and 25 μM c-di-GMP. Protection sites of RNAP alone are indicated by a black line to the right of each image. Protection sites of the activated transcription complexes (RNAP, VpsR/c-di-GMP, and DNA) are indicated by a gray line, and hypersensitivity sites are represented by arrows, both to the right of each image. VpsR binding sites (Fig. 1B and Fig. 5B) are indicated by the dashed lines to the right of each image.

DISCUSSION

Not surprisingly, regulation of biofilm genes is complex, requiring multiple transcriptional activators (VpsR, VpsT, and AphA) and repressors (H-NS and HapR). The activators VpsR and VpsT function in conjunction with the small signaling molecule c-di-GMP, whose level increases when biofilm formation is warranted. VpsR induces expression of vpsT, and both transcription factors jointly regulate downstream genes that are necessary for generating the biofilm matrix in a feed-forward network (32). It is thought that VpsT functions through anti-repression of H-NS. Not only do EMSAs show that both H-NS and VpsT/c-di-GMP bind rbmA and vpsU promoters as well as the intergenic region between vpsL and rbmF, but chromatin immunoprecipitation (ChIP) assays have demonstrated that deletion of vpsT results in higher H-NS promoter occupancy, while overexpression of c-di-GMP leads to a decrease in H-NS promoter occupancy at these promoters (18, 38, 39, 45). In addition to VpsR and VpsT, AphA, the master regulator of virulence expression, also upregulates vpsT and vps-I (32). Providing additional repression is HapR, which inhibits both aphA expression and biofilm formation under high-cell density conditions (33, 46). Further regulation is provided by the small molecules guanosine 3′-diphosphate 5′-triphosphate and guanosine 3′,5′-bis (diphosphate) [(p)ppGpp] and cAMP (through the cAMP receptor protein, CRP) (47, 48).

As the master regulator of the c-di-GMP-regulated transcriptional network, VpsR binds c-di-GMP with a K_d of 1.6 μM using unknown residues. Alignment of VpsR with other c-di-GMP binding proteins has not revealed a conserved c-di-GMP binding motif (49), and the crystal structure of VpsR remains unsolved. Recently, we have shown that VpsR directly activates σ⁷⁰-RNAP transcription from P_vpsL, the promoter at the head of the vps-II operon (31). In the work here, we demonstrate that VpsR directly activates σ⁷⁰-RNAP transcription from the other promoters within the biofilm formation cluster, including P_vpsU, at the beginning of the vps-I operon, P_rbmA, at the start of the rbm cluster, and P_rbmF, which lies upstream of the divergent rbmF and rbmE genes. Thus, despite its sequence similarity to EBPs that work with σ⁵⁴-RNAP, VpsR functions to directly activate σ⁷⁰-RNAP transcription for biofilm biogenesis regulation.

A variety of different mechanisms are used to direct RNAP to specific promoters to activate transcription in response to environmental cues and growth signals. Activators of σ⁷⁰-RNAP can be generally categorized into different classes based on promoter binding location and RNAP contacts. In class I activation, the activator binds upstream of RNAP (typically around −60) and interacts with one or both of the C-terminal domains of the α subunit (αCTD) (22). In class II activation, the activator binds adjacent to or overlapping the −35 element, typically interacting with the C-terminal region of σ⁷⁰ (region 4) and/or the N-terminal domain of α (αNTD) (22). Furthermore, some activators, such as the Bordetella pertussis response regulator BvgA at P_fhaB, use a combination class I/class II, contacting both σ⁷⁰ region 4 as well as the αCTDs (50). Another set of bacterial regulators, such as MerR, alter the promoter DNA conformation to shorten the suboptimal distance between the −35 and −10 elements, allowing RNAP to bind (51). As shown here, VpsR can be classified as both a class I and a class II activator (Fig. 7). At P_vpsL and P_vpsU, VpsR functions as a class II activator by binding immediately adjacent/overlapping the core promoter sequence, and at P_rbmA and P_rbmF, VpsR functions as a class I activator by binding to sequences farther upstream (22, 28).

FIG 7.

VpsR uses at least three different mechanisms to activate transcription. With σ⁷⁰-RNAP, VpsR and c-di-GMP (cdG) together activate P_rbmA and P_rbmF using a class I mechanism and activate P_vpsL and P_vpsU using a class II mechanism. In vivo ChIP-seq analyses suggest a third mechanism: VpsR is an antirepressor of H-NS at P_vpsT (38). Promoters, which we tested but did not activate transcription in vitro under our conditions, are labeled with an asterisk. Though direct VpsR binding has been demonstrated for P_vpsT, P_tfoY, P_epsC, and P_aphA in EMSAs and or DNase I footprinting (30, 32, 36, 40, 41), transcription mechanisms for these four promoters have not been recapitulated and determined in vitro.

The ability of a regulator to function as both a class I and a class II activator is not unusual and is also seen with Escherichia coli CRP. At the class I lac promoter, CRP binds upstream of the core promoter at a site centered at −61.5, and transcription activation is mediated by protein-protein interactions between a surfaced-exposed α-turn of CRP and one of the αCTDs (52). This increases the affinity of RNAP for promoter DNA, activating transcription through polymerase recruitment. At the class II galP1 promoter, CRP binds to a site centered at −41.5, overlapping the −35 element, and transcription activation is mediated by three protein-protein interactions between CRP and the αCTD, αNTD, and σ⁷⁰ region 4 portions of RNAP (52). These interactions both recruit RNAP and facilitate the transition from RPc to RPo (52).

We have shown that the presence of c-di-GMP affects the affinity of VpsR for the DNA differently depending on the promoter. While the presence of c-di-GMP increases the affinity of VpsR for its site at the class I rbmA and rbmF promoters (Fig. 2), c-di-GMP does not alter VpsR affinity for its binding site at the class II vpsU and vpsL promoters (Fig. 2 and reference 31, respectively). Based on what is known about CRP as a class I and class II activator, we speculate that VpsR’s mechanism of transcription initiation might work similarly, but in a manner that involves c-di-GMP (Fig. 7). In this scenario, we hypothesize that at the class I promoters, VpsR uses c-di-GMP to enhance DNA binding and then recruit RNAP, whereas at the class II promoters, VpsR uses c-di-GMP to affect its interaction with RNAP, stimulating RPo formation. However, as of now, the contacts between VpsR and RNAP as well as the precise mechanisms are still unknown.

Besides its involvement in biofilm regulation, VpsR also directly binds to several other genes, including the transcription factor genes aphA, vpsT, and tfoY and the promoter upstream of the type II secretion system operon (32, 36, 40, 41). Although EMSAs and/or DNase I footprinting have demonstrated the binding of VpsR to sites upstream of these promoters (32, 36, 40, 41), we did not observe activation at aphA, vpsT, and epsC with σ⁷⁰-RNAP under our in vitro transcription conditions. These results suggest that VpsR may have additional mechanisms for activating these genes (Fig. 7). In addition, previous ChIP-seq studies of H-NS occupancy in V. cholerae have indicated that VpsR functions as an antirepressor of H-NS (either directly or indirectly) at the vpsT promoter (Fig. 7) (38). Future studies, such as ChIP assays to globally identify VpsR binding sites and reconstruction of vpsT transcription with VpsR and H-NS in vitro, will help determine the mechanism by which VpsR regulates other promoters.

Finally, previous work also indicated that the intergenic region between the divergently transcribed rbmF and vpsL genes contains two VpsR binding sites and suggested that the distal VpsR binding site (relative to the +1 TSS of vpsL) functions as an H-NS antirepressor site for VpsR (31). While this may still be true, it is now also clear that this site is directly involved in activating the divergent rbmF promoter. These two VpsR binding sites are ∼240 bp apart. Our previous work has shown that VpsR forms a dimer in the presence and absence of c-di-GMP in solution (31), leading to the speculation that in the presence of DNA containing both the vpsL and the rbmF binding sites, VpsR might be able to form higher-order structures and generate DNA looping as has been seen for the E. coli regulator GalR (53). Such DNA looping would further prevent H-NS binding and repression while simultaneously bolstering transcription.

In summary, our data support a complex model of vps and rbm regulation. While VpsT/c-di-GMP is important in releasing the DNA from negative regulation by H-NS, VpsR/c-di-GMP is important for positive regulation with σ⁷⁰-RNAP. Because the various affected promoters vary in their VpsR-DNA binding affinities in the absence and presence of c-di-GMP (Fig. 2) (31), we speculate that the relative activity of each promoter will be different depending on the concentration of VpsR and the level of c-di-GMP. In addition to regulating biofilm matrix genes, VpsR and VpsT concomitantly control, in a c-di-GMP manner, several other phenotypes, including type II secretion, DNA repair, and catalase production, suggesting these transcription factors induce an adaptive program for a sessile lifestyle (41, 54, 55). This two-tiered type of regulation should allow for fine-tuning of gene expression, promoting the exquisite control needed for this sophisticated process.

MATERIALS AND METHODS

DNA.

In vitro transcription templates, pMLH04 containing the vpsT promoter from −188 to +154 relative to the +1 transcriptional start site (TSS), pMLH05 containing the aphA promoter from −127 to +154 relative to the TSS, pMLH06 containing the vpsL promoter from −97 to +213 relative to the TSS, pMLH40 containing the rbmA promoter from −223 to +119 relative to the TSS, pMLH41 containing the rbmF promoter from −228 to +145 relative to the TSS, pMLH42 containing the vpsU promoter from −130 to +155 relative to the TSS, and pMLH43 containing the epsC promoter from −100 to +188 relative to the TSS were cloned into the EcoRI and HindIII restriction enzyme sites of pRLG770 (56). This position places the inserts upstream of the rrnBP1 terminator (56). Inserts were first obtained as PCR products which had been amplified with primers (Table 1) from V. cholerae genomic DNA (BH1514) (Table 2) using Pfu turbo polymerase (Stratagene). Vectors were digested with the appropriate restriction enzymes, ligations were performed using standard Gibson techniques (57), and resulting products were transformed into DH5α. Fragments used for EMSAs and DNase I footprinting were obtained as PCR products using Pfu turbo polymerase and upstream and downstream PCR primers (Table 1) annealing to V. cholerae regions listed above. To radiolabel the DNA, template primer was treated with T4 polynucleotide kinase (Affymetrix) in the presence of [γ-³²P] ATP prior to PCR. Radiolabeled PCR products were purified as described (58). pMLH11 is a pET-28b(+) derivative (Novagen) that contains vpsR cloned within the NdeI and XhoI restriction sites constructed as previously described (31).

TABLE 1.

Plasmids and primers used in this study

Plasmid or primer	Description	Reference no. or sequence(s)
pCMW75	IPTG-inducible V. harveyi diguanylate cyclase gene qrgB	33
pCMW98	IPTG-inducible V. harveyi diguanylate cyclase gene qrgB mutant (GG→AA)	33
pMLH04	vpsT promoter −188 to +154 relative to +1 TSS cloned into pRLG770	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, ATTAAGCAACTTGGCTTATATGTG; reverse, GGGTCAGGTGGGACCCAAGCTT, CAAGCCAGAGCTCAGAAAATGGTGT
pMLH05	aphA promoter −127 to +154 relative to +1 TSS cloned into pRLG770	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, CTAAAAGGTCACAACTTTGTGGCC; reverse, GGGTCAGGTGGGACCCAAGCTT, GTGCTTCTAATGGCACTCATATCAAC
pMLH06	vpsL promoter −97 to +213 relative to +1 TSS cloned into pRLG770	31
pMLH11	vpsR cloned into pET28b(+)	31
pMLH17	vpsR cloned into pHERD20T	31
pMLH40	rbmA promoter −223 to +119 relative to +1 TSS cloned into pRLG770	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, CTATGCTTGGTTATTGCTCAATGAA; reverse, GGGTCAGGTGGGACCCAAGCTT, AATCCACTTCCGCATAAGAAGCCG
pMLH41	rbmF promoter −228 to +145 relative to +1 TSS cloned into pRLG770	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, CAAATAAAACCTTGAGACTCACTTATTAAC; reverse, GGGTCAGGTGGGACCCAAGCTT, GATCAAATGAAGCGGTAGTG
pMLH42	vpsU promoter −130 to +155 relative to +1 TSS cloned into pRLG770	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, GATTCAAATCGATGGCTTTTAATAC; reverse, GGGTCAGGTGGGACCCAAGCTT, GCGTTTTTGCCGAATCTTATCGCG
pMLH43	epsC promoter −100 to +188 relative to +1 TSS cloned into pRLG770	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, TCATGCTTAACAATGGTGTTTGCAG; reverse, GGGTCAGGTGGGACCCAAGCTT, GTAGCGTTAATCCTTCGCTGATCGG
Reverse primer for primer extension products of rbmA	84 nucleotides upstream of +1 TSS and 66 nucleotides upstream of +1 GTG translation start site of rbmA	CAATGCCAAGCATGAGGCCA
Reverse primer for primer extension products of rbmF	112 nucleotides upstream of +1 TSS and 67 nucleotides upstream of +1 ATG translation start site of rbmF	CTGCGAATGCATTCATGCTAGG
Reverse primer for primer extension products of vpsU	120 nucleotides upstream of +1 TSS and 67 nucleotides upstream of +1 ATG translation start site of vpsU	CTGCCATTGGCGAACGACA
Radiolabeled DNA for EMSA and DNase I footprinting	rbmA promoter −223 to +84	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, CTATGCTTGGTTATTGCTCAATGAA; reverse, CAATGCCAAGCATGAGGCCA
Radiolabeled DNA for EMSA and DNase I footprinting	rbmF promoter −228 to +112	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, CAAATAAAACCTTGAGACTCACTTATTAAC; reverse, CTGCCATTGGCGAACGACA
Radiolabeled DNA for EMSA and DNase I footprinting	vpsU promoter −130 to +120	Forward, GAGGCCCTTTCGTCTTCAAGAATTC, GATTCAAATCGATGGCTTTTAATAC; reverse, CTGCCATTGGCGAACGACA

TABLE 2.

Strains used in this study

Strain	Description	Source or reference no.
DH5α	E. coli strain used for cloning	NEBa
BL21(DE3)	E. coli strain used for protein purification	NEB
Rosetta 2(DE3)/pLysS	E. coli strain used for protein purification	NEB
BH1514	V. cholerae El Tor strain C6707str2 (WT)	59
WN310	ΔvpsL ΔvpsR	32

^aNEB, New England BioLabs.

Strains and growth conditions.

E. coli DH5α was used for cloning, and BL21(DE3) or Rosetta 2(DE3)/pLysS (New England Biolabs) were used for protein purification. The V. cholerae strains used in this study were derived from El Tor biotype C6707str2 (59). Growth conditions for primer extensions and protein purifications are listed below under sections labeled primer extensions and proteins, respectively.

Proteins.

E. coli RNAP core was purchased from New England Biolabs. E. coli σ⁷⁰ was purified as previously described (60). VpsR protein was isolated from Rosetta 2(DE3)/pLysS (Novagen) containing pMLH11 and grown and purified as previously described (31). Protein concentrations were determined by comparison of known amounts of core RNAP after SDS-PAGE and gel staining with colloidal Coomassie blue (Invitrogen).

In vitro transcriptions.

Single round in vitro transcriptions were assembled first in 5 μl containing 0.02 pmol of supercoiled template, 3.0 pmol of VpsR, 0 or 25 μM c-di-GMP, reconstituted RNAP (0.2 pmol of σ⁷⁰ plus 0.05 pmol of core), and transcription buffer (40 mM Tris-acetate [pH 7.9], 150 mM potassium glutamate, 4 mM magnesium acetate, 0.1 mM EDTA [pH 7.0], 0.01 mM dithiothreitol [DTT], and 100 μg/ml bovine serum albumin [BSA]). Samples were incubated at 37°C for 10 min prior to the addition of 1 μl ribonucleoside triphosphates (rNTPs) containing 1.43 mM ATP, GTP, CTP, and 36 μM [α-³²P] UTP at 5 × 10⁴ dpm/pmol and 500 ng heparin. After incubation for 10 min at 37°C, reactions were collected on dry ice, load solution (15 μl; 9.4 mM EDTA [pH 7], 0.9% bromophenol blue, and 0.9% xylene cyanol dissolved in deionized formamide) was added, and aliquots were electrophoresed on 4% (wt/vol) polyacrylamide, 7 M urea denaturing gels for 2,500 V/h in 0.5× Tris-borate-EDTA (TBE) buffer. After electrophoresis, gels were exposed to X-ray films and films were scanned using a GS-800 calibrated densitometer and analyzed with Quantity One software (Bio-Rad).

Primer extensions.

Primer extension products generated from RNA isolated from in vitro transcriptions were obtained as described above and processed according to the manufacturer’s instructions (Promega) and as described previously (31). Briefly, 5 μl of the in vitro transcription reaction (made using nonradioactive UTP) was added to 6 μl of a primer mixture containing 2× avian myeloblastosis virus (AMV) primer extension buffer, AMV reverse transcriptase, and 2 pmol of ³²P-labeled primer (rbmA primer, which anneals 84 bp downstream of the TSS of rbmA, rbmF primer, which anneals 112 bp downstream of the TSS of rbmF, or vpsU primer, which anneals 120 bp downstream of the TSS of vpsU). Samples were electrophoresed on 8% (wt/vol) polyacrylamide, 7 M urea denaturing gels for 5,000 V/h in 0.5× TBE. Imaging, densitometry, and quantification were performed as described above.

In vivo V. cholerae RNA was obtained from strain WN310 (59) containing pMLH17 (31) and either pCMW75 or pCMW98. pCMW75 contains an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible Vibrio harveyi diguanylate cyclase enzyme, QrgB, used to synthesize elevated intracellular concentrations of c-di-GMP, while pCMW98 contains an inactive version of the enzyme used as a negative control to generate unaltered c-di-GMP concentrations (33). Cells were grown and harvested, and RNA was extracted as previously described (31). Primer extension products were performed as described previously (31).

EMSAs.

VpsR-DNA complexes and transcription complexes with RNAP were formed in transcription buffer using 0.05 pmol of DNA as previously described (31). To ensure observation of specific complexes, a 1–μl solution of competitor containing 1 mg/ml of poly(dI-dC) or 500 μg/ml heparin was added to VpsR-DNA complexes (20 μl) during binding or to transcription complexes (10 μl) after binding, respectively. After addition of competitor, VpsR-DNA complexes were loaded onto a 5% (wt/vol) nondenaturing polyacrylamide gel already running at 100 V/h in 1× TBE buffer and subsequently electrophoresed for 1.5 h at 100 V/h. Transcription complexes were loaded onto a 4% (wt/vol) nondenaturing polyacrylamide gel already running at 100 V/h in 1× TBE buffer and electrophoresed for 3 h at 380 V/h. After autoradiography, films were scanned as described above. K_d(app) values were calculated as the concentration of VpsR needed to shift 50% of the free DNA.

DNase I footprinting.

Solutions in transcription buffer were assembled as described above for EMSAs using 0.04 μM DNA and, as indicated, 1.4 μM VpsR, 50 μM c-di-GMP, and/or 0.16 μM reconstituted RNAP. After incubation at 37°C for 10 min, a 1-μl solution of 1 mg/ml of poly(dI-dC) or 500 μg/ml heparin was added to VpsR-DNA complexes and transcription complexes with RNAP, respectively, for 15 s at 37°C. To initiate the cleavage reaction, 0.33 U of DNase I was added to the VpsR-DNA complexes, and 0.375 U of DNase I was added to the transcription complexes with RNAP. Solutions were incubated for an additional 30 s at 37°C, immediately loaded onto a 4% (wt/vol) nondenaturing, polyacrylamide gel already running at 100 V/h in 1× TBE buffer, and electrophoresed for 3 h at 380 V/h. After autoradiography, the protein/DNA complexes were excised, and extracted DNA was electrophoresed on denaturing gels as described (31, 61).