ToxR activates the Vibrio cholerae virulence genes by tethering DNA to the membrane through versatile binding to multiple sites

Significance

ToxR is one of the key proteins in the regulatory cascade that leads to the production of the cholera toxin and other virulence factors when Vibrio cholerae colonizes the human intestine. Using X-ray crystallography, we show how multiple ToxR molecules bind to DNA to regulate the expression of distinct genes. Unexpectedly, ToxR recognizes diverse and extensive DNA sequences using a topological DNA readout mechanism as opposed to a specific DNA recognition sequence. By proposing an integrated model for ToxR-mediated virulence regulation, we provide crucial insights into specific molecular mechanisms causing the disease cholera. In a wider perspective, we also supply structural data on how a membrane-anchored bacterial one-component signal transduction system functions.

Abstract

ToxR, a Vibrio cholerae transmembrane one-component signal transduction factor, lies within a regulatory cascade that results in the expression of ToxT, toxin coregulated pilus, and cholera toxin. While ToxR has been extensively studied for its ability to activate or repress various genes in V. cholerae, here we present the crystal structures of the ToxR cytoplasmic domain bound to DNA at the toxT and ompU promoters. The structures confirm some predicted interactions, yet reveal other unexpected promoter interactions with implications for other potential regulatory roles for ToxR. We show that ToxR is a versatile virulence regulator that recognizes diverse and extensive, eukaryotic-like regulatory DNA sequences, that relies more on DNA structural elements than specific sequences for binding. Using this topological DNA recognition mechanism, ToxR can bind both in tandem and in a twofold inverted-repeat-driven manner. Its regulatory action is based on coordinated multiple binding to promoter regions near the transcription start site, which can remove the repressing H-NS proteins and prepares the DNA for optimal interaction with the RNA polymerase.

Cholera is an acute life-threatening diarrheal infection caused by the bacterium Vibrio cholerae. Endemic in several countries, cholera outbreaks may also occur under catastrophic events or war situations. Upon arrival of V. cholerae to the small intestine, two integral membrane proteins, ToxR and TcpP, activate the toxT promoter (1–6). This triggers a regulatory cascade resulting in secretion of cholera toxin and the expression of the toxin coregulated pilus (7–9). ToxR alone also down-regulates the expression of membrane porin OmpT and up-regulates that of OmpU, leading to protection against bile salts in the human intestinal epithelium (10–13).

A number of studies have described residues of ToxR required for DNA binding and transcription activation. These include residues predicted to be in the DNA-binding helix and wing domain for DNA binding and the α-loop for transcription, specifically (3, 5, 14, 15). While these studies have implicated several critical residues through genetic screens for ToxR activity, the precise interaction mode is unknown, and other important residues may have been overlooked. Thus, a structural analysis of ToxR bound to DNA is essential to get a comprehensive view of ToxR/DNA interactions. Furthermore, as a winged-helix transcription factor, the orientation of ToxR monomers on the promoter has certain implications for interaction with RNA polymerase and TcpP, a neighboring transcription factor on the toxT promoter essential for RNA polymerase (RNAP) stimulation on that promoter (4–6). Again, solving the structure of ToxR on the toxT promoter can provide a wealth of information on potential interactions between ToxR and RNAP or ToxR and TcpP. Previous promoter analysis studies using footprinting and gel-shift analysis indicated broad ToxR-binding sites on several promoters (1, 3). Mutant promoter studies indicated multiple binding sites for ToxR on the toxT, ompU, ompT, and ctxA promoters (16, 17). These studies utilized genetic screens or scanning transversion mutations along the various promoters. As such, it is unclear whether nucleotide substitutions affecting DNA binding and gene activation are disrupting a specific side-chain/base pair interaction between the DNA-binding helix of ToxR and the promoters or affecting a structural promoter element (like an A/T tract) that affects the shape of the DNA and the ability of winged-helix proteins to interact with it. For these various reasons, solving the structure of ToxR on DNA promoters will clarify these ambiguities on how ToxR activates (or in some cases represses) multiple promoters in V. cholerae.

One of the unique aspects of one-component systems is the integration of environmental signal sensing and DNA binding into a single transmembrane protein. For ToxR activation of toxT in conjunction with TcpP, this membrane localization is essential (18). Recent single-molecule studies following the mobility of TcpP in the inner membrane in conjunction with various toxT promoter deletions and in the presence or absence of ToxR have shown a critical role for ToxR in facilitating TcpP binding to toxT promoter DNA from the membrane as measured by different mobility states of TcpP in the presence or absence of ToxR (19). These live cell imaging studies are in agreement with previous studies showing the importance of both membrane localization and ToxR/TcpP interaction for toxT activation, the lynchpin of the V. cholerae virulence gene expression cascade (4, 18).

Our structures of the ToxR cytoplasmic domain bound to the toxT and ompU promoters shed light on many of these areas, showing that ToxR is a versatile virulence regulator that can recognize diverse 7-bp DNA sequences, which have in common a thymine in the sixth position preceded by an AT-rich region with a narrowed minor groove. These findings provide a more complete model for ToxR-mediated virulence regulation in V. cholerae and a mechanism of action that could be extended to other prokaryotic one-component signal transduction systems with transmembrane regions (20).

Materials and Methods

Protein Expression and Purification.

We used a nonpathogenic Escherichia coli strain carrying a plasmid codifying for ToxR^cyt (residues 6 to 115) fused to the C-terminal KHHHHHH sequence. The plasmid includes a kanamycin-resistance gene and the protein-coding region under control of the toxR promoter sequence. Therefore, the strain expresses ToxR^cyt with no need of induction. Cultures were grown at 37 °C until an OD_600nm of ~0.7 was reached. After that, cultures were kept at 25 °C overnight for protein expression. Cells were resuspended in 20 mM Tris–HCl pH 7.4, 250 mM NaCl, 20 mM imidazole, DNase I, and Complete Protease Inhibitor Cocktail Tablets (Roche). After 15 min of incubation at 4 °C, lysis was performed with a cell disruptor (Constant Systems Ltd.). The sample was then clarified by ultracentrifugation and purified by a three-step protocol. Briefly, the protein was filtered and loaded onto a HisTrap 5 mL HP affinity column equilibrated in buffer A (20 mM Tris–HCl pH 7.4, 250 mM NaCl, and 20 mM imidazole) and eluted in buffer B (20 mM Tris–HCl pH 8.0, 250 mM NaCl, 350 mM imidazole); the protein was then loaded onto a HiLoad 16/60 Superdex 75 size-exclusion column run with buffer C (20 mM Tris–HCl pH 7.4 and 250 mM NaCl), and finally, the sample was purified by a reverse-affinity step with the HisTrap 5 mL HP column, using buffers A and B, after incubating the protein with carboxypeptidase overnight at room temperature with gentle agitation (100 μL carboxypeptidase/1 mg ToxR^cys) to remove the C-terminal histidine tag. The protein sample was concentrated using Amicon centrifugal filters (Merck Millipore).

Protein–DNA Complex Preparation and Gel Shift Assays.

Oligonucleotides with the desired sequences and their complementary ones were ordered from Biomers and resuspended in 50 mM NaCl, 20 mM Tris pH 8, 2 mM MgCl₂ buffer (ompU promoter), or double-distilled water (toxT promoter). The sequences were ompU19 (5′-CATATCATTTTACTAACTG-3′), toxT20 (5′-CTCAAAAAACATAAAATAAC-3′), toxT40 (5′-CCAAAAAACATAAAATAACATGAGTTACTTTATGTTTTTC-3′), and toxT25 (5′-CTTTATGTTTTTCTTATGTAATACG-3′). Equimolar amounts of each couple of complementary oligonucleotides were denatured at 80 °C for 30 min and annealed overnight by gradual cooling, after turning off the bath. ToxR^cyt was slowly added to the annealed dsDNA at a 2:1 (protein:dsDNA) ratio (ToxR^cyt-ompU19) or at a 1:1.2 (protein:DNA) ratio (ToxR^cyt-toxTpro complexes). Samples were then incubated at 4 °C for 3 h. Purification was performed using a 10/300 Superdex 200 column in 20 mM Tris pH 7.6, 1 mM EDTA, and 125 mM NaCl buffer. Gel shift assays were performed running SDS-PAGE gels [30% (w/v) acrylamide/Bis, 10X TAE, 10% PSA, and 0.1% TEMED] with 0.5X TAE running buffer for 25 min at 200 V and 4 °C.

Crystallization, X-ray Data Collection, and Crystallographic Processing.

DNA-complex fractions were concentrated to ~10 mg/mL. Crystals of ToxR^cyt/DNA complexes were obtained by sitting drop and/or hanging drop vapor diffusion in the following conditions: ToxR^cyt/ompU19 (0.2 M ammonium sulfate, 0.01 M cadmium chloride, 0.1 M PIPES pH 6.5, and 10% ethylene glycol), ToxR^cyt/toxT20 (30% PEG 4000, 0.1 M sodium acetate pH 4, and 0.15 M ammonium acetate), ToxR^cyt/toxT40 (32% PEG MME 500, 0.175 M sodium chloride, and 0.1 M glycine pH 8.5), and ToxR^cyt/toxT25 (25% PEG 4000 and 0.3 M ammonium sulfate). All crystals were mounted in loops and flash-frozen in liquid nitrogen, using a cryoprotective buffer. X-ray diffraction data were collected at BL13-XALOC beamline at ALBA synchrotron in Cerdanyola del Vallès (Spain), using a Dectris Pilatus 6M detector. Data were collected at 0.9792 Å and processed using XDS (21). SCALA (22) was used for scaling and merging the data.

Structure Solution, Model Building, and Coordinate Refinement.

All MR procedures were performed with PHASER (23). For Tox^cyt-toxT20, the search model was monomer E of the E. coli PhoB DNA binding domain (PDB ID: 1GXP) (24) and a 10-bp idealized dsDNA. DNA strands were mutated and grown manually considering the oligonucleotide sequence. The other Tox^cyt/DNA structures were solved using a ToxR^cyt monomer, alone or bound to a 10-bp DNA fragment extracted from the Tox^cyt/toxT20 structure, as a search model. DNA was then mutated, and the models were refined. Atomic models were traced in Coot (25) and refined using REFMAC5 (26) and Phenix (27). All the models were validated using MolProbity (28).

Construction of toxT Promoter Mutants and β-Galactosidase Assays.

Mutation of the −65/−51 region of the toxT promoter was performed using site-directed mutagenesis of the toxT-lacZ fusion plasmid pTLI2 (1) as previously described (29). Briefly, plasmid pTLI2 was PCR amplified in the presence of mutagenic primers covering the −65 and −51 region: top primer, 5′ C​AAA​AAA​CAT​AAA​ATA​ACA​TGA​GTT​ACT​ggc​gag​gTT​TCT​TAT​GTc​Agc​CGT​CTGTAACTTGTTCTTATG 3′, bottom primer

5′ C​ATA​AGA​ACA​AGT​TAC​AGA​CGg​cTg​ACA​TAA​GAA​Acc​tcg​ccA​GTA​ACT​CAT​GTTATTTT​ATGTTTTTTG 3′. After PCR amplification, the reactions were digested with DpnI to degrade the starting plasmid, and PCR products were transformed into E. coli DH5α. pTLI2 derivatives from candidate ampicillin-resistant colonies were isolated using a Qiagen plasmid miniprep kit and sent for DNA sequencing (Genewiz). Several clones incorporating the following toxT promoter sequence change from −67 to −47 were isolated: 5′ TTATGTTTTTTTCTTATGTAATA 3′ to 5′ GGCGAGGTTTTCTTATGTCAGC 3′ (10 point mutations in bold). The mutant form of pTLI2 was introduced into V. cholerae expressing ToxR (parental strain O395) or lacking ToxR (ΔtoxR strain EK307), and β-galactosidase assays were performed comparing the wild-type pTLI2 to the −65/−51 pTLI2 mutant and the empty vector, pTL61T, as previously described (29). In addition, the previously characterized promoter mutants A(−94)C and A(−84)C were included as negative controls, as they are not recognized by ToxR (17). Analysis of two ToxR mutants affecting the DNA-recognition helix (ToxR-T77A and ToxR-Q78A) was also assessed by the β-galactosidase assay as described previously (5) after constructing the mutations in the previously described ToxR expression plasmid pSK-toxR-HA (5) using PCR mutagenesis and the primers 5′ GCAAGGTTTTGAAGTCGATGATTCCAGCTTAgCCCAAGCC 3′ top and 5′ GCATTTTGCGCAGAGTCGAAATGGCTTGGGcTAAGCTGG 3′ bottom for T77A and 5′ GTTTTGAAGTCGATGATTCCAGCTTAACCgcAGCCATTTC 3′ top and 5′ GAGCATTTTGCGCAGAGTCGAAATGGCTgcGGTTAAGCTG 3′ bottom for Q78A. Expression of ToxR was confirmed by western blot analysis using an anti-ToxR antibody (6) or anti-HA antibody (Covance).

Gel Shift Assay for toxT Mutant Promoter.

To confirm that predicted mutations in the −65/−51 region of the toxT promoter affected DNA binding, promoter gel shift analysis was performed using an FAM-labeled fluorescent wild-type toxT promoter double-stranded probe from −72 to −36, as described previously (30). Upon addition of a 25-fold excess of unlabeled (cold) double-stranded wild-type toxT promoter probe from −72 to −36 or the −65/−51 mutated unlabeled probe, the ability to compete with the FAM-labeled probe was assessed. Binding reactions were composed of increasing concentrations of the ToxR protein of interest [wild-type ToxRcyt or ToxRcyt-R84A; known to lack DNA binding activity (5)], 1 µL of 100 µg/mL salmon sperm DNA, 2 µL of 5× binding buffer (100 mM Bis–Tris pH 6.5, 500 mM NaCl, 50 mM MgCl2, 500 µg/mL BSA, and 25% glycerol), 0 or 2 µL of 25X cold probe competitors, 1 µL of the FAM-labeled promoter probe (stock at 10 µM), and the remaining volume was H₂O to reach 10 µL. Binding reactions were incubated for 30 min at room temperature, and samples were loaded onto a 9% polyacrylamide gel and run at 120 V for 45 min in the cold (4 °C).

Structure Analysis.

Protein–protein interactions were analyzed with the PISA server (31). Protein–DNA interactions were analyzed with NucPlot (32). DNA conformations of the structures were analyzed with 3DNA (33). Figures were prepared with ChimeraX (34) and PyMOL (The PyMOL Molecular Graphics System, Version 1.2r3pre, Schrödinger, LLC).

Results and Discussion

Versatile Binding of ToxR to Multiple Regions of the toxT and ompU Promoters.

Due to the extensive length of the ToxR-binding site on the toxT promoter, based on DNase I footprinting (3), we combined three overlapping crystal structures of the cytoplasmic domain of ToxR (ToxR^cyt) bound to the toxT promoter (toxTpro) DNA to assess ToxR binding to that promoter. One crystal structure was sufficient to assess ToxR^cyt binding to the ompU promoter (ompUpro) (Fig. 1A) (SI Appendix, Table S1).

Fig. 1.

Structures of ToxR^cyt bound to the toxT and ompU promoters. (A) Promoter regions covered by the partial ToxR^cyt/toxTpro structures that were used to build the ToxR^cyt/toxT composite model (Left) and by the ToxR^cyt/ompUpro19 structure (Right). The toxTpro inverted repeat sequence is highlighted in cream color. Putative TcpP-binding regions are highlighted in blue. At the bottom, surface representations of the final ToxR^cyt/toxT composite model and the ToxR^cyt/ompU model. (B) ToxR^cyt protein structure extracted from the ToxR^cyt/ompU19 complex structure. Three different orientations of its ribbon representation are shown, with the main DNA-contacting regions, α-helix and β-strand elements indicated.

The DNA-bound ToxR^cyt reported here (Fig. 1B) displays a mostly N-terminal five-stranded β-sheet (antiparallel β1-β4 and β8 parallel to β1), a central three-helix bundle (α1-α3) with a small β-strand inserted between α1 and α2 (β5), and a C-terminal hairpin (β6-β7, the wing) that also interacts with β5 forming a three-stranded antiparallel β-sheet. The segment comprising helices α2 and α3 forms a modified winged helix-turn-helix DNA-binding motif (wHTH) (35), with α2-α3 connector (α-loop or transactivation loop) replacing the turn. The structure is similar, although not identical, to the recently reported NMR model of ToxR not bound to DNA (PDB ID: 7NMB) (36) (rmsd = 2.94 Å). The main differences are located at the α-loop and at the segment connecting helix α3 with the wing (ancillary wing region) (SI Appendix, Fig. S7B). Fewer differences are observed when the crystal structure is compared with the AlphaFold model (rmsd = 0.696 Å).

Four ToxR^cyt molecules are bound to the populated regulatory region of toxTpro extending from position −45 to −97 (Fig. 1A). Although a specific consensus binding sequence is not shown along this 53-bp region, the ToxR^cyt molecules are consistently bound to the same position in distinct but overlapping structures (monomers B and D are bound to the same sequence in two of the structures). Moreover, molecules B and D are bound to an inverted repeat sequence in opposite orientations, while molecules C and D are bound in tandem. These observations confirm that the ToxR binding to the DNA is quite versatile but not indiscriminate.

While ToxR-binding sites centered at −94 and −84 were characterized previously (3, 17), ToxR-binding sites centered at −74 (molecule C), −65 (molecule B), and −51 (molecule A) were unexpected (Fig. 1A). Monomer A, the closest molecule to the transcription start, binds from T-56 to A-47. Surprisingly, this region overlaps the putative site specific for TcpP binding (29). Monomer B is bound four base pairs upstream from monomer A (from C-69 to T-60). This is an A-T enriched region, at the downstream half of an inverted repeat (highlighted in orange, Fig. 1A). This could explain the reverse orientation of monomer B with respect to monomer D (its equivalent at the other half of the inverted repeat). Upstream of monomer B are monomers C (from T-81 to T-72) and D (from A-91 to A-82), exactly 10 base pairs apart (one complete DNA turn) from each other and hence arranged in a continuous, straight-line tandem. While previous toxT promoter footprinting and mutant analyses predicted monomer D, monomer C was unanticipated (3, 17, 37). However, nucleotide A-74 was shown previously to be important for toxT activation (17). The multiple binding of ToxR^cyt molecules to this toxTpro region suggests an effective way to keep the DNA attached to the cytoplasmic membrane, thus facilitating the subsequent binding of membrane-anchored TcpP to the adjacent region, and the eventual recruitment of the RNA polymerase. The requirement for ToxR membrane localization in order to facilitate TcpP-mediated toxT activation is well established (4, 18, 19). Note that TcpP is predicted to have a much shorter linker between the transmembrane helix and the DNA-binding domain than ToxR, thus being much less flexible as it attempts to find and bind to the toxT promoter. A promoter prone to bind with so many regulatory proteins involved is typical of eukaryotic gene expression regulation, where large enhancer regions are covered by multiple transcription factors cooperating to form the so-called enhanceosomes (38), and is less commonly seen in prokaryotes (39).

At the ompU promoter, ToxR is a direct transcription activator and hence must play an additional role apart from recruiting the DNA to the membrane. We solved the structure of the ToxR^cyt bound to a 19-bp oligonucleotide (ompU19) containing the −47 to −30 ompU promoter sequence and a similar structure with less resolution using a 20-bp oligonucleotide (containing the −48 to −30 ompU promoter sequence) (SI Appendix, Table S1). In these structures, two tandem pairs of ToxR^cyt monomers are bound to the ompUpro DNA in opposite directions (Fig. 1A). Both tandems are related to each other by a rotation of 180° about a binary axis perpendicular to the DNA axis, located approximately in the middle of the structure.

ToxR^cyt–DNA Interactions Reveal a Topological DNA Recognition Mechanism.

Protein–DNA interactions throughout the toxTpro region covered by our composite model (base pairs −45 to −97) (SI Appendix, Figs. S1–S4) have subtle differences among the four bound ToxR^cyt molecules. We can, however, outline a general common binding pattern (Fig. 2), which is quite like that of the effector domain of the signal transduction transcriptional activator PhoB (24). Protein–DNA interaction occurs through five specific ToxR^cyt elements: the α3 recognition helix, bound to the major groove; the wing, constituted by the turn connecting the β6-β7 hairpin and inserted to the minor groove; the ancillary wing loop, connecting α3 and β6; the beginning of the α1 helix; and the α-loop connecting α2 and α3 (Fig. 2 A–F). Remarkably, Gln78 is the only residue that contacts the DNA bases and not the phosphate backbone. In monomer D, Gln78 interacts directly with A-85 N6 through its Oδ, while it forms at the same time two water-mediated hydrogen bonds, one between the same Oδ and A-85 N9 and the other between Nε and the O2 of the thymine complementary to A-83 (T-83) (Fig. 2G). Thr77 is also a key residue, displaying Van der Waals interactions with the DNA C5 atom of T-86 (Fig. 2G). In fact, this interacting thymine is the only conserved base in all the binding sites reported in this work (considering both toxTpro and ompUpro), as shown in the sequence logos (Fig. 3). Previous mutagenesis studies also demonstrated the importance of nucleotides A-83 and T-86 for toxT promoter activation (17). A Q78R substitution in ToxR has also been shown to disrupt toxT and ompU activation (5).

Fig. 2.

Protein–DNA interactions at the ToxR^cyt/toxTpro and ToxR^cyt/ompUpro structures. (A) Sequence of the ToxR^cyt used in this work, with the residues contacting the DNA highlighted (black background), showing the boundaries of the secondary structure elements and the name of the main DNA-contacting regions. Depiction of the residues interacting with toxT promoter DNA (side chains visible) at the (B) recognition helix region; (C) wing element; (D) ancillary wing loop; (E) beginning of helix α1; and (F) transactivation α2-α3 loop (α-loop). Zoomed regions including solvent molecules (red spheres) show the hydrogen bond (blue dashed lines) and van der Waals (green dashed lines) interactions and distances (black dashed lines) between the DNA and (G) Thr77 and Gln78 and (H) Pro101 and Lys102 of monomer D in the ToxR^cyt/toxT20 structure and (I) Thr77 and Gln78 of monomer W in the ToxR^cyt/tompU19 structure.

Fig. 3.

Binding sequence logos for all the ToxR^cyt binding sites reported in this work. Sequences used for LOGO calculations at weblogo.berkeley.edu: ToxR^cyt-A-toxT (ttatgta), ToxR^cyt-B-toxT (aaacata), ToxR^cyt-C-toxT (taacatg), ToxR^cyt-D-toxT (aaacata); ToxR^cyt-W-ompU (ttatatc), ToxR^cyt-X-ompU (tttacta), ToxR^cyt-Z-ompU (gttagta), ToxR^cyt-Y-ompU (tgatata).

At the other promoter, ompUpro, the four ToxR^cyt monomers share the main interactions with the DNA reported above for toxTpro (SI Appendix, Fig. S4). Gln78 seems to play again an important role, directly interacting with DNA bases, although this interaction is with a cytosine C-43, rather than and A or T (Fig. 2I). Thr77 seems to be critical as well, since van der Waals interactions with the 5-methyl groups of a consensus T residue (Fig. 2I) are observed in all four binding ToxR^cyt molecules. To confirm the role of Gln78 and Thr77 of ToxR in toxT and ompU activation, we constructed alanine substitution mutants of each residue and assessed their activity using toxT-lacZ and ompU-lacZ fusion reporters. The ToxR-Thr77Ala mutant was highly defective for both promoters, whereas the ToxR-Gln78Ala mutant maintained 30% activation of the toxT promoter while being completely defective for ompU activation (Fig. 4). This confirms the critical nature of the Thr77/T-86 interaction at the toxT promoter and Gln78/C-43 and Thr77/T44 interactions at the ompU promoter.

Fig. 4.

ToxR residues Thr77 and Gln78 are required for full activation of the toxT and ompU promoters. ToxR (wild-type), ToxR-T77A, ToxR-Q78A, or empty vector (pSK) was expressed in V. cholerae ΔtoxR strains harboring either a toxT-lacZ fusion (strain EK1072) or ompU-lacZ (strain EK410) (5). Strains were induced for 3 to 4 h with 100 µM IPTG to express toxR, and β-galactosidase readings were measured. Results are normalized to WT ToxR, set at 100% activation. *P < 0.05 as assessed by Student’s t test. Experiments were performed at least three times in duplicate (n = 6).

The DNA-binding consensus motif consists of an A/T tract of seven base pairs where the sixth base is always a T (Fig. 3). It suggests that the DNA conformation (rather than specific bases), plus an invariable T/Thr77 van der Waals interaction, determines the binding. When examined individually (SI Appendix, Fig. S5), the three ToxR^cyt/toxTpro structures reveal that both the minor and major grooves of the DNA are compressed at certain regions of the toxTpro. In the ToxR^cyt/toxTpro20 structure, the DNA has a B conformation and is only slightly curved (4.9°) due to the compression of the minor groove in an AT-rich region from A-92 to A-89. This results in the formation of a typical hydration spine described for AT-rich sequences (40) and a narrow minor groove (41). The minor groove narrows up to 8.9 Å and allows the suitable fitting of Lys 102 and Pro101, which enter the minor groove and display important contacts with DNA (Fig. 2C) (SI Appendix, Fig. S1) without affecting the first layer of the hydration spine (Fig. 2H). This is in keeping with previous studies that had demonstrated a critical role for Lys102 in DNA binding and a more supportive role for Pro101 for engaging the minor groove (6).The recognition helix interacts at the downstream (−86 to −83) major groove displaying most of the contacts with DNA (Fig. 2B), probably contributing to stabilize the minor groove compression. A similar conformation is found in the other ToxR^cyt/toxT and ToxR^cyt/ompU structures solved. The same topological or “indirect” DNA readout mechanism is shown by Ler, an H-NS family protein, which recognizes compressed minor grooves associated with AT-tracts and senses them using the side chain of an arginine (as Lys102 does in ToxR) (42). This supports the idea that one of the functions of ToxR is the displacement of the H-NS proteins that repress some ToxR-regulated promoters (43, 44). ToxR may accomplish that by using the same DNA readout mechanism to compete with H-NS proteins for shared binding locations.

Protein–Protein Interactions Stabilize Previously Bound ToxR^cyt Monomers.

Although protein–protein interactions between ToxR^cyt molecules (Fig. 5) are not essential for DNA binding (since a single ToxR molecule can bind DNA, Fig. 1A, toxT20 structure), they may provide cooperative binding at certain sites. At the toxTpro, monomers C and D share an interface area of 409.9 Ų. The contact regions (Fig. 5A) comprise the wing, the ancillary wing region, and the α1-β5 loop of monomer C and the β2-β3 loop, the β4-α1 loop, and the α2-α3 loop of monomer D, respectively (Fig. 5B). According to the PDBePISA server (31), the interface area formed by these interactions is not sufficient for dimer complex formation by itself; hence, the driving force for the association is likely DNA binding. Similarly, the interface areas at each ompUpro tandem pair (236.0 Ų between monomers W and X; 305.2 Ų between monomers Y and Z) are not extensive enough to ensure the dimerization of ToxR^cyt molecules. Again, both ompUpro tandem interfaces are based on hydrogen bonds involving a side chain of one monomer and a main-chain carbonyl oxygen of the other (Fig. 5 C and D) and the electrostatic contact between Lys102 of monomers X and Y and Phe69 of monomers W and Z, respectively. This last interaction, also found at the tandem arrangements in the toxTpro, is relevant as the Phe69 side chain from one monomer is fixing in place the Lys102 side chain of the neighboring monomer (Fig. 5E). Thus, after recognition of a compressed minor groove by Pro101 and Lys102, the incorporation of a second ToxR molecule may secure the first binding, introducing cooperativity to the process. However, previous studies have shown an F69A mutant of ToxR maintains 92% activation of the toxT promoter, while it only supports 35% activation of ompU (5). Perhaps activation of toxT promoter is less affected by this mutation due to the multivalent binding of ToxR monomers to that promoter.

Fig. 5.

Protein–protein interactions at the ToxR^cyt/toxTpro and ToxR^cyt/ompUpro structures. (A) Ribbon and surface representations of the ToxR^cyt monomers D and C bound to the toxTpro40 DNA. Monomers D and C are colored in dark green and light green, respectively. Top, front, and rear views are shown, and protein–protein interface areas are indicated in purple for monomer D and in gold for monomer C. (B) Detail of the main molecular contacts between ToxR^cyt monomers D and C. The color code is the same as in the previous panel. (C) Interactions between the Y (red) and Z (orange) monomers of ToxR^cyt bound to the ompUpro19 DNA. (D) Interactions between W (purple) and X (blue) monomers of ToxR^cyt bound to the ompUpro19 DNA. (E) Surface representation of the interaction between Phe69 of monomer W (purple) and Lys102 of monomer X (blue) of ToxR^cyt bound to the ompUpro19 DNA. (F) Binding of ToxR to sites centered at −65 and −51 is not required for toxT activation. Plasmids harboring the wild-type (WT) toxT promoter fused to lacZ, mutant toxT promoters, or no toxT promoter (−) were assessed for toxT activation in the presence or absence of ToxR in V. cholerae strain O395 (ToxR+) or EK307 (ToxR−). Mutant −65/−51 contained 10 nucleotide transversions in the downstream region of the toxT promoter. Mutants −94 and −84 were reported previously to fail to bind ToxR and cannot be activated in V. cholerae (17). Samples were tested four times in duplicate. *P < 0.01 relative to the wild-type toxT promoter in the presence of ToxR as assessed by Student’s t test. Levels of ToxR were confirmed with an anti-ToxR antibody (6). (G) ToxR binding to a FAM-labeled fluorescent wild-type toxT promoter probe from −72 to −36 was measured in the absence or presence of unlabeled (cold) competitor toxT promoter DNA added at a 25-fold excess. While the wild-type toxT promoter cold competitor DNA completely competed away ToxR binding to the FAM-labeled probe (lane 6), the toxT mutant probe with the −65 and −51 ToxR binding sites mutated (10 mutations in all) was unable to compete with the wild-type FAM-labeled toxT probe (lane 7). ToxR showed dose-dependent binding to the FAM-labeled toxT probe, and a ToxR mutant, ToxR-R84A, was unable to bind the toxT promoter (lane 8). R84 was shown previously to be critical for ToxR binding to the toxT promoter (5) and makes direct contact with DNA (Fig. 2B).

One of the key findings of this structural study is that ToxR can bind a wide toxTpro area from nucleotide −95 all the way to −51, overlapping the TcpP-binding site. However, the more downstream ToxR binding sites are not required for toxT activation as evidenced by the fact that the DNA sequence of the −65 and −54 toxT promoter regions can be dramatically altered and toxT activation is unchanged (Fig. 5F). This mutated promoter region fails to bind ToxR based on its inability to compete with a fluorescently labeled wild-type toxT promoter in a gel shift assay (Fig. 5G). In this experiment, we mutagenized the downstream binding sites from the partial consensus sequence of AACATAAAG (−65 bottom strand) to CCTCGCCAG and TAATA (−54 top strand) to TCAGC (avoiding nucleotides important for TcpP-binding). This heavily altered toxT promoter construct still maintained ~100% activation compared to a strain harboring a wild-type toxT-lacZ promoter reporter (Fig. 5 F and G).

Model for the ToxR^cyt/ompUpro19-RNAP Holoenzyme Complex.

To understand how the ToxR binding to the ompUpro may activate OmpU expression, we created a model of ToxR^cyt/ompUpro19 complex bound to the RNA polymerase holoenzyme (PDB ID: 1IW7) (45) using, as a guide, the PhoB-phobox-σ4-β-flap complex structure (PDB ID: 3T72) (46). This crystal structure contains two PhoB^E protomers and one RNA polymerase σ4-β-flap tip helix chimera bound to a 26-mer phoA pho box-encoding DNA. We superimposed the PhoB monomer contacting the RNA polymerase chimera to the ToxR monomer that is in the −35 ompUpro region. In the resulting model (SI Appendix, Fig. S6), the RNA polymerase σ subunit recognition helix fully enters the DNA major groove of the ompUpro. Furthermore, it interacts with the transactivation loop (47) located between α2 and α3 helices of ToxR^cyt. This may explain why Val71 in the α-loop was previously shown to be especially critical for ompU activation and less important for toxT activation, where ToxR instead facilitates TcpP interaction with RNA polymerase (5). This model is compatible with a functional transcription complex. The DNA-binding domain of ToxR is located at the N terminus of the transmembrane protein. In the ToxR-RNA polymerase model, the C termini of the ToxR monomers are facing the opposite side of the interaction surface with the polymerase. Hence, the structure does not cause steric hindrance between the different parts, even in the presence of the full-length ToxR. It is unclear whether the presence of four ToxR molecules bound to the ompUpro is necessary or it is a transitory complex to firmly attach the DNA to the membrane, and the formation of the subsequent transcription activation complex may involve fewer ToxR monomers. Further experiments are required to shed more light on the activating role of ToxR when bound to the ompUpro, especially the structural determination of a transcription initiation complex incorporating the RNA polymerase holoenzyme. To the same extent, more structural information is needed on the nature of the interactions taking place at the periplasmic space of V. cholerae, although the now available computed structure models from Alphafold DB (48, 49) for the full-length ToxR, TcpP, TcpH, and TcpS may shed some light on this question (SI Appendix, Supplementary Discussion).

Conclusions

The adaptable DNA-binding mode shown by ToxR in our structures illustrates an uncharted mechanism for the prokaryotic one-component signal transduction systems across the membrane. By ensuring multiple binding events near the transcription initiation site, ToxR overcomes the constraints of having to interact with a cytosolic target DNA while being attached to the membrane. These constraints may have led to the evolution of one-component systems to the more typical and well-characterized two-component systems including phosphotransfer domains (20).

Gathering all the data discussed above, we put forward an integrated model for ToxR-mediated virulence regulation in V. cholerae (Fig. 6). The activation of the toxTpro is extremely fine-tuned by environmental conditions and involves the existence of extensive regulatory regions that modulate the response. toxTpro is repressed by H-NS proteins (43) (Fig. 6A). Upon arrival of V. cholerae to the mucus layer covering the intestinal epithelium, bile salts enter the bacterial cell by crossing its outer membrane (Fig. 6B). ToxS, which forms a periplasmic heterodimer with ToxR, is the actual sensor of bile in the periplasm. The binding of bile salt molecules to the ToxS pocket reshapes the conformation of the ToxS-ToxR heterodimer (50). This rearrangement could induce the self-association of ToxS-ToxR heterodimers to potentially form heterotetramers. The local increase of ToxR^cyt molecules then favors a multiple, cooperative binding to the toxT regulatory region. This results in the displacement of H-NS proteins repressing transcription in the region from −102 to −45 (Fig. 6C), possibly spread from a main nucleation site far upstream (between −256 and −172) (43). DNA local conformation, rather than a highly conserved consensus sequence, determines the binding. Multiple ToxR molecules bind progressively along the toxT promoter and recruit the DNA to the membrane (Fig. 6D). This would create the suitable conditions for the action of TcpP. With the DNA strongly tethered to the membrane, TcpP can either displace the ToxR molecules bound to the promoter-proximal site centered at −51 (Fig. 6 E.i) or interact with those ToxR molecules to stabilize its own binding (Fig. 6 E.ii), in accordance with the “hand holding” (4, 29) and “cooperativity” (19) models, respectively, thus emphasizing the importance of membrane localization by ToxR (18). In fact, our structures demonstrate that multiple wHTH domains can bind to very short regions of DNA, and hence, a simultaneous binding to form a putative ToxR-TcpP heterodimer around the −50 region of the toxTpro cannot be ruled out. Once TcpP binds the DNA, it would recruit and activate the RNA polymerase and transcription of toxT would commence (Fig. 6F). The fact that the ToxR-binding sites centered at −65 and −51 can be mutated and toxT activation still occurs (Fig. 5 F and G) suggests either that ToxR binding to the sites centered at −94 (17), −84, and −74 are sufficient to facilitate TcpP binding to the toxT promoter or that occupancy of the downstream lower affinity ToxR binding sites may only occur at higher ToxR concentrations or upon recognition of other periplasmic signals, such as cyclo(Phe-Pro) (51). It is possible that under these conditions, ToxR binding to the −65 and −51 region may displace TcpP and turn off expression of the toxT promoter, reminiscent of the ompF promoter in E. coli where ompF is activated by OmpR-P at lower concentrations, but inhibited at higher OmpR-P levels (52). As, such, these low-affinity ToxR-binding sites may be more important for turning off toxT transcription when appropriate in the intestine, at the time of transmission to the next host. Previous studies have shown overexpression of ToxR decreases toxT expression compared to endogenous levels of ToxR expression (17), possibly by displacing TcpP.

Fig. 6.

Model for V. cholerae virulence cascade signal transduction system. (A) toxTpro is repressed by H-NS proteins while V. cholerae is in the aquatic environment; (B) Upon colonization of the intestinal epithelium, bile salts bind to the ToxS pocket of mandatory ToxS-ToxR heterodimer, triggering the formation of ToxS-ToxR heterotetramers; (C) The local increase of ToxR^cyt molecules enables a multiple, cooperative binding to the toxTpro regulatory region, which results in the displacement of the H-NS proteins repressing toxT; (D) Multiple ToxR molecules bind the toxTpro and recruit the DNA to the membrane; (E) Once the DNA is attached to the membrane, TcpP can bind the toxTpro either competing with ToxR (E.I) or interacting with it through a hand-hold mechanism (E.II); (F) When TcpP is properly bound to the toxTpro regulatory region, it recruits the RNA polymerase and activates toxT transcription; (G) At the ompUpro, where no TcpP molecules are involved, ToxR molecules bind near the -35 region and recruit the RNA polymerase to facilitate ompU activation.

The activation of the other ToxR-regulated promoters, where no TcpP molecules are involved, becomes much simpler since the response is not so finely modulated. ToxR molecules would bind ompUpro near the −35 region and recruit RNAP to facilitate ompU activation (Fig. 6G).

In summary, our results indicate that ToxR starts the regulation of the V. cholerae virulence promoters using a topological DNA recognition mechanism to bind to lengthy regulatory regions both in a cooperative way in tandem and in an inverted-repeat mode. This binding of multiple ToxR molecules leads to i) the removal of the repressing H-NS proteins from the promoters and ii) the recruitment of the DNA to the membrane. When this has been accomplished, ToxR may act either as an independent regulator or as a co-activator (5). ToxR may directly recruit the RNA polymerase to ompUpro to activate ompU expression but can also recruit toxTpro to TcpP and stabilize its binding to enhance TcpP-mediated RNA polymerase activation.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Atomic coordinates and crystallographic structure factors were deposited in the Protein Data Bank (PDB) under accession codes PDB 8B4B (53), PDB 8B4C (54), PDB 8B4D (55), and PDB 8B4E (56).