A disulfide constrains the ToxR periplasmic domain structure, altering its interactions with ToxS and bile-salts

ToxR is a transmembrane transcription factor that, together with its integral membrane periplasmic binding partner ToxS, is conserved across the Vibrionaceae family. In some pathogenic Vibrios, including V. parahaemolyticus and V. cholerae, ToxR is required for bile resistance and virulence, and ToxR is fully activated and protected from degradation by ToxS. ToxS achieves this in part by ensuring formation of an intra-chain disulfide bond in the C-terminal periplasmic domain of ToxR (dbToxRp). In this study, biochemical analysis showed dbToxRp to have a higher affinity for the ToxS periplasmic domain than the non-disulfide bonded conformation. Analysis of our dbToxRp crystal structure showed this is due to disulfide bond stabilization. Furthermore, dbToxRp is structurally homologous to the V. parahaemolyticus VtrA periplasmic domain. These results highlight the critical structural role of disulfide bond in ToxR and along with VtrA define a domain fold involved in environmental sensing conserved across the Vibrionaceae family.


Abstract:
ToxR is a transmembrane transcription factor that, together with its integral membrane periplasmic binding partner ToxS, is conserved across the Vibrio family. In some pathogenic Vibrios, including V. parahaemolyticus and V. cholerae, ToxR is required for bile resistance and virulence, and ToxR is fully activated and protected from degradation by ToxS. ToxS achieves this in part by ensuring formation of an intra-chain disulfide bond in the C-terminal periplasmic domain of ToxR (dbToxRp). In this study, biochemical analysis showed dbToxRp to have a higher affinity for the ToxS periplasmic domain than the non-disulfide bonded conformation. Analysis of our dbToxRp crystal structure showed this is due to disulfide bond stabilization. Furthermore, dbToxRp is structurally homologous to the V. parahaemolyticus VtrA periplasmic domain. These results highlight the critical structural role of disulfide bond in ToxR and along with VtrA define a domain fold involved in environmental sensing conserved across the Vibrio family.

Introduction:
ToxR is the founding member of a group of transmembrane transcription factors (Schlundt et al., 2017) and is conserved throughout the gram-negative Vibrionaceae family of marine bacteria (Reich and Schoolnik, 1994;Welch and Bartlett, 1998;Provenzano et al., 2000;Wang et al., 2002). Several Vibrio species are dependent on ToxR for bile resistance (Provenzano et al., 2000), and a few pathogenic species have co-opted ToxR for virulence induction (Miller and Mekalanos, 1984;Hubbard et al., 2016). Vibrio parahaemolyticus ToxR is important for colonization and virulence in mouse models (Hubbard et al., 2016), where it induces type III secretion system expression through activating expression of VtrA and VtrB (Hubbard et al., 2016). ToxS, an integral membrane periplasmic binding partner of ToxR, is also important for V. parahaemolyticus colonization in a mouse model (Hubbard et al., 2016). In V. cholerae, ToxR is required for human colonization (Herrington et al., 1988), most likely through induction of bile resistance (Provenzano et al., 2000) and, in conjunction with TcpPH, virulence gene expression (Hase and Mekalanos, 1998).
ToxR regulates many genes in response to stimuli, including ompT and ompU (Bina et al., 2003;Ante et al., 2015). ToxR directly inhibits ompT expression (C C Li et al., 2000) and activates ompU transcription (Crawford et al., 1998), leading to a change in the outer-membrane porin composition (Miller and Mekalanos, 1988). ToxR is thought to activate virulence by augmenting TcpP activity, leading to toxT transcription (Hase and Mekalanos, 1998;Krukonis et al., 2000;Krukonis and DiRita, 2003) and subsequent expression of cholera toxin and the toxin co-regulated pilus (Champion et al., 1997;Krukonis et al., 2000). Given the importance of ToxR and ToxS for both bile resistance and pathogenesis in the Vibrio family (Herrington et al., 1988;Hubbard et al., 2016), understanding how they interact with each other and respond to the environment is critical for understanding the disease processes and providing insight into how related proteins function in other pathogenic bacteria (Yang and Isberg, 1997;Peng Li et al., 2016).
Previously, experiments in minimal media have shown V. cholerae ToxR activity can be regulated by the amino acid mixture of N, R, E and, S (NRES) (Mey et al., 2012), or bile salts (Mey et al., 2012;Midgett et al., 2017). The NRES mixture led to an increase in ToxR and subsequent OmpU expression (Mey et al., 2012;Mey et al., 2015). In contrast, bile salts stimulated ToxR activity without increasing protein amounts (Mey et al., 2015;Midgett et al., 2017). This demonstrates ToxR activity can be regulated by two different mechanisms, one dependent on the Var/Csr system to increase ToxR mRNA and protein expression (Mey et al., 2015), and a second to mobilize existing ToxR to become transcriptionally competent (Mey et al., 2015;Midgett et al., 2017). We hypothesize this potentially occurs via bile salts interacting with the ToxR periplasmic domain.
ToxS, an integral membrane protein with a periplasmic domain, augments ToxR activity (Miller et al., 1989) and protects ToxR from degradation in conditions of stationary growth and alkaline pH (Almagro-Moreno et al., 2015;Lembke et al., 2018).
Furthermore, ToxS is known to affect the formation of a disulfide bond in the ToxR periplasmic domain (Ottemann and Mekalanos, 1996;Fengler et al., 2012).
Investigations into the contribution of cysteines C236 and C293 on V. cholerae ToxR function showed single cysteine mutants were nonfunctional (Fan et al., 2014), whereas a double mutant had decreased function in porin regulation, but no defects in virulence gene expression (Fengler et al., 2012). Moreover, the double mutant was shown to be more susceptible to degradation than wild-type ToxR, even in the stabilizing presence of ToxS (Lembke et al., 2018). In wild type ToxR, these cysteines are known to form either an intra-chain disulfide bond or a disulfide linked homodimer, which is present when toxS is deleted in V. cholerae (Ottemann and Mekalanos, 1995;Fengler et al., 2012).
Since ΔtoxS strains have defects in ToxR activity (Mey et al., 2012;Fengler et al., 2012;Midgett et al., 2017), it follows that the intra-chain disulfide bonded, monomeric form of the ToxR periplasmic domain is the physiological active conformation.
While the intra-chain disulfide bonded conformation of ToxR is certainly the predominant form (Ottemann and Mekalanos, 1996;Fengler et al., 2012), ToxR lacking a disulfide bond has been hypothesized to be present in vivo based on two prior studies. First, the periplasmic domain of ToxR and TcpP can form a disulfide linked heterodimer (Fan et al., 2014), which can only occur if the two cysteines in ToxR are free to form a bond. Second, bile salts induce disulfide stress by inhibiting the activity of DsbA (Xue et al., 2016), the chaperone responsible for inducing disulfide bond formation (Landeta et al., 2018). Prior work in our laboratory showed the non-disulfide bonded ToxR periplasmic domain was destabilized by bile salts, however, the salts increased binding to the ToxS periplasmic domain (ToxSp). This led us to hypothesize the ToxR periplasmic domain has evolved to become destabilized by bile salts, leading to stronger ToxS binding and increased activity of ToxR (Midgett et al., 2017).
To characterize and visualize the differences between the disulfide bonded and nondisulfide bonded forms of ToxR we expressed and purified the disulfide bonded conformer of the ToxR periplasmic domains (dbToxRp) from V. cholerae and V. vulnificus for biochemical and structural studies. The dbToxRp was able to bind ToxSp, and was destabilized by bile salts, although, unlike the non-disulfide bonded ToxR periplasmic domain (ToxRp), bile salts did not increase the binding of dbToxRp to ToxSp. The structure of dbToxRp showed the domain is globular composed of 2 αhelices, linked via the disulfide bond, and a β-sheet composed of 5 β-strands.
Furthermore, we found that dbToxRp is structurally homologous to the periplasmic domain of VtrA (Peng Li et al., 2016). A transmembrane transcription factor that shares similarities with ToxR as both have an OmpR DNA binding domain and a periplasmic domain that is involved in bile sensing (Kodama et al., 2010;Peng Li et al., 2016).
Overall, these results demonstrate the important role of disulfide bond formation in interaction with ToxS, represent a first step in understanding the structure function relationship for virulence induction and bile resistance, and pave the way for a better understanding of the biochemical nature of activation of ToxR and related proteins.

Results:
ToxR containing a disulfide bond is folded and active: The V. cholerae disulfide ToxR periplasmic domain (Vc-dbToxRp) construct was confirmed to be structurally and functionally similar to the previously analyzed ToxRp (Midgett et al., 2017)   bonded conformer was destabilized in a manner similar to the non-disulfide bonded form as previously described (Midgett et al., 2017). Finally, a pull down showed the Vc-dbToxRp interacted with the CBDI-ToxSp (Fig. 1c). Taken together, these tests confirm that Vc-dbToxRp is structurally and functionally similar to Vc-ToxRp.

dbToxRp shows increased binding to ToxSp, unaffected by bile salts:
Previous work in our laboratory revealed, somewhat unexpectedly, that the bile salt CDC increased the interaction between the periplasmic domains of ToxR and ToxS (Midgett et al., 2017). To test if CDC also increases the affinity of Vc-dbToxRp to CBDI-ToxSp we performed pull downs with ToxRp and Vc-dbToxRp in the presence and absence of CDC. Similar to previous results, CDC increased the amount of Vc-ToxRp pulled down by CBDI-ToxSp (fold change of 4.0 ± 2.0). However, CDC had a negligible effect on the amount of Vc-dbToxRp pulled down (fold change of 1.6 ± .5), indicating Vc-dbToxRp binding to ToxS is unaffected by bile salts (Fig. 2a). To compare the amounts of the ToxR periplasmic domains pulled down in the experiments we ran selected pull downs on the same gel to calculate relative amounts. Interestingly, the presence of a disulfide bond increased the interaction between Vc-dbToxRp and CBDI-ToxSp. The results clearly showed Vc-dbToxRp with CDC was pulled down more than Vc-ToxRp in in the presence or absence of CDC. As when the amount of Vc-dbToxRp pulled down was set to one, Vc-ToxRp was pulled down .22 ± .11 in comparison to Vc-dbToxRp, and Vc-ToxRp with CDC was pulled down .41 ± .05 relative to Vc-dbToxRp ( Fig. 2b). These results suggest the presence of a disulfide bond in the ToxR periplasmic domain significantly increases the affinity of ToxR for ToxS.

Crystal structures reveal the structural role of the ToxR periplasmic domain disulfide bond:
The crystal structure of V. vulnificus intra-chain disulfide ToxR periplasmic domain (Vv-dbToxRp) was solved by molecular replacement using a SeMet labeled structure to 1.25 Å, revealing an α/β fold with a 5 stranded β-sheet with one side packing against two α-helices (Fig. 3a). The two helices are connected by a disulfide bond between Cys232 and Cys289. The structure is well ordered, and electron density for the disulfide bond is clearly visible in a composite omit map of the native structure  conformational flexibility (Fig 3b last panel). We hypothesize a major role of the disulfide bond is to stabilize the structure, and importantly the ToxS binding region, by constraining this flexible loop as well as the relatively short helix α2 which follows it.

VtrA and ToxR periplasmic domains are structural and functional homologues:
Using the Vv-dbToxRp structure, we ran a DALI search (Holm and Laakso, 2016) to look for structural homologues. Not surprisingly given the prevalence of α/β folds in proteins, DALI identified a number of functionally unrelated proteins with structural similarity. Interestingly, the periplasmic domain of VtrA from V. parahaemolyticus (Peng Li et al., 2016) was ranked 118 th by the DALI server (Table S1). Despite having little sequence identity and an RMSD of 3.0-3.3 Å it is clearly a structural homolog (Fig. 3c).
Not only do both domains have the same topology and overall structural motif, but both are C-terminal periplasmic domains of transmembrane transcription factors involved in environmental sensing (Welch and Bartlett, 1998;Mey et al., 2012;Peng Li et al., 2016;Midgett et al., 2017). The main difference between the two structures is the position of the last β-strand and α-helix. In VtrA the last helix is longer, pushing the last β-strand further away from the core structure. Also, as the periplasmic domain of VtrA does not contain cysteines, there can be no stabilizing disulfide bond. Based on the clear structural and functional similarities, we propose the periplasmic domains of ToxR and VtrA represent a common domain utilized in sensing the extra-cellular environment.

dbToxRp lacks a dimerization interface:
The periplasmic domain of ToxR has been proposed to mediate dimerization, which is thought to be essential for activity. Prior studies replacing the periplasmic domain of ToxR with proteins known to dimerize, resulted in active constructs (DiRita and Mekalanos, 1991;Ottemann and Mekalanos, 1995;Kolmar et al., 1995;Dziejman et al., 1999). Therefore, it was somewhat surprising that the Vv-dbToxRp structure did not form a crystallographic dimer, nor did it illuminate a biologically relevant dimer interface. Close examination of the structure revealed two crystal contacts that could be potential dimerization interfaces (Fig. S1). The first involves the loop between β4 and β5, which makes a contact with the loop between β2 and α1 as well as the loop between β3 and β4 on a neighboring molecule. However, given that there are few hydrogen bonding or charge-charge interactions between the two protein molecules, and that dimerization in this area would position the N-termini on opposite sides of the dimer, an interface here would essentially sterically preclude both proteins being membrane anchored. Another possible interaction is between β1 and the β2-α1 loop on an adjacent monomer. While this would position both N-termini toward the membrane, the interaction is mediated by a sulfate ion and there are a few interactions between the two monomers. Analysis using PISA (Krissinel and Henrick, 2007) also corroborates that both of these interfaces are most likely biologically irrelevant. Therefore, we conclude the ToxR periplasmic domain is most likely monomeric and any dimerization is mediated by other factors, for example, via interaction with ToxS.

Discussion:
Here we have biochemically characterized and solved the structure of the ToxR periplasmic domain with an intra-chain disulfide bond. This conformation was found to be similar to the non-disulfide conformation as confirmed by NMR, indicating both constructs consist of a stable core fold. In addition, the disulfide conformer is destabilized by bile salts, similar to previously published results (Midgett et al., 2017). However, in contrast to ToxRp (Midgett et al., 2017), bile salts had no effect on the binding of the dbToxRp conformer to ToxSp.  ToxS periplasmic domain using the VtrAC structure, and how ToxR and ToxS interact to induce virulence as well as bile resistance. a. The left panel shows the VtrAC (VtrA orange, VtrC in gray) structure with arrows pointing to the interfaces between the proteins. Arrow 1 points to the VtrA β5-strand that extends the VtrC β-sheet, and arrow 2 points to VtrA βsandwich interaction with VtrC. In the right panel the Vv-dbToxRp structure (blue) was aligned with VtrA in the VtrAC structure, VtrC is depicted in gray. The numbered arrows point to the β5 strand (1) and the potential β-sheet interface (2). b. A model of the potential interactions between ToxR, ToxS, and bile salts. The non-disulfide bonded form of ToxR interacts weakly with ToxS (1). However, bile salts increase the interaction of the non-disulfide bonded ToxR to ToxS leading to ToxR activation (2). In addition, ToxS favors the formation of the intra-chain periplasmic disulfide bond conformation increasing the affinity between ToxR and ToxS (3). The complex of the periplasmic disulfide bonded ToxR with ToxS in the presence of bile salts increases ToxR activity by an unknown mechanism (4).
interact based on what is known about the VtrA system. Previously, the VtrA periplasmic domain was crystallized with its binding partner the VtrC periplasmic domain. The structure showed the proteins interact in two ways, via parallel beta strand hydrogen bonds between the β5 strand of VtrA sheet and the most N-terminal β-strand of VtrC beta-barrel (arrow 1 in Fig 4a left panel), as well as β-sandwich like interaction between the VtrA β-sheet and the C-termnal region of the VtrC β-barrel (arrow two in  Li et al., 2016). Based on this, we predict ToxS is a structural homolog of VtrC and the ToxR periplasmic domain will bind ToxS in a similar manner ( Fig. 4a right panel). Despite these structural similarities there are differences between these two regulatory pairs. First, VtrA requires VtrC for expression (Peng Li et al., 2016) whereas ToxR can be expressed by itself (Ottemann and Mekalanos, 1996;Mey et al., 2012;Fengler et al., 2012;Midgett et al., 2017). Second, ToxR depends on the formation of an intra-chain disulfide bond to properly fold while the VtrA periplasmic domain is cysteine free. It would be interesting to determine if the ToxR periplasmic domain could be stabilized without a disulfide bond and what effect that would have on ToxR function.
By combining our results with insight gleaned from the VtrAC structure, we can begin to explain how ToxR and ToxS might interact. At least in vitro, ToxR can exist in three forms, a monomer with an intra-chain disulfide bond, a disulfide-free monomer, and a (likely non-physiological) disulfide bonded homodimer. Disulfide formation in the ToxR monomer apparently constrains the β5-α2 loop and the adjacent α2 helix. This, based on structural homology with VtrAC, stabilizes the binding interface between ToxR and ToxS. Such an increase in stability is consistent with observations regarding the double cysteine mutant degradation in Lembke et al. (Lembke et al., 2018). Interestingly it has been observed that ToxS enhances ToxR intra-chain disulfide bond formation (Ottemann and Mekalanos, 1996). How this occurs in the context of the above results is unexplored.
The structure of the dbToxRp provides a potential explanation to why breaking the disulfide bond interferes with ToxS binding and how bile salts can modulate the interaction. In the absence of intra-chain disulfide bond formation, ToxR is destabilized and it is unlikely the short α2 helix would form. Additionally, α2 might assume an alternate conformation in which it blocks interaction with ToxS. Given bile salts increase the ToxRp-ToxSp interaction suggests bile salts disrupt whatever alternate structure α2 forms in the absence of the disulfide bond, thereby allowing ToxS binding. In dbToxRp, the loop and α2 are already in a conformation that stabilizes the interface, rather than blocking it, and bile salts have no effect on ToxS binding. This leaves open the question of how ToxS augments ToxR activity. ToxR is thought to be active as a dimer and its periplasmic domain has been presumed to mediate dimerization (DiRita and Mekalanos, 1991;Ottemann and Mekalanos, 1995;Kolmar et al., 1995;Dziejman et al., 1999). However, our structural analysis failed to find a dimer interface, arguing for other determinants of dimerization. ToxS is the most likely candidate for this, and we propose that disulfide bond dependent interaction of ToxR with ToxS results in formation of a dimer of ToxR and ToxS dimers, forming a complex competent for ToxR activation. Future structural studies aimed at determining the structures of ToxSp as well as that of the complex between dbToxRp and ToxSp should clarify if this is the mechanism for dimerization.
Alternately, given ToxR binds DNA, DNA could very well provide the impetus for ToxR oligomerization. In such a model ToxR binds DNA, and weak protein-protein interactions could promote dimerization or higher order oligomers, as seen in experiments with VtrA (Okada et al., 2017). Yet another possibility is DNA organizes ToxR in the membrane without specific interactions between domains. Determining which might be the case will require reconstruction of full-length ToxR in a membrane and in the presence of DNA.
In summary, our results suggest a model in which ToxR is a monomer in the membrane. The non-disulfide bonded conformation has a low affinity for ToxS, and the interaction between the two proteins increases in the presence of bile salts, leading to increased ToxR activity. When ToxR contains an intra-chain disulfide bond, its affinity for ToxS is increased and unaffected by bile salts, leaving open the question of how bile salts activate ToxR (Fig. 4b).

Cloning of ToxR periplasmic domains:
The sequences of the ToxR periplasmic domains from V. vulnificus, V. parahaemolyticus, V. fischeri, V. harveyi, and Photobacterium profundum were identified by using the ToxR periplasmic sequence from V. cholerae (T199-E294). The six ToxR periplasmic domains were codon optimized and synthesized, from 5' to 3', with a NcoI site, a 6xHis N-terminal tag followed by the coding sequence, a BamHI site, all flanked by primer sites to amplify the constructs. The constructs were PCR amplified, cut with the appropriate restriction enzymes, and the digested constructs were purified using a PCR clean up kit (Qiagen). The pET16b plasmid was digested with the same restriction enzymes, treated with CIP (NEB), and purified using a PCR clean up kit (Qiagen). The inserts were ligated into the plasmid using the Quick Ligase (NEB) and the reaction was used to transform DH5α's. Colonies from the transformation were subjected to colony PCR to determine if the plasmids contained an insert of the appropriate size. Then selected colonies were cultured overnight for mini-preps following manufacture instructions (Qiagen). The resulting plasmids were sequence verified using a primer for the T7 promoter.

Expression and purification:
The plasmids with the different ToxR periplasmic domains were transformed into Shuffle T7 Express cells. The strains were double selected to produce stable expression strains (Sivashanmugam et al., 2009). After double selection the strains were checked for production of soluble protein, by performing a scaled down purification. The ToxR periplasmic domain from V. vulnificus was found to produce the greatest amount of soluble protein. Production cultures of V. vulnificus and V. cholerae ToxR periplasmic domains were started by picking a colony from a freshly streaked plate incubated overnight at 37 ºC, and inoculating 2ml of ZYP-0.8G media (Studier, 2005) with 200 µg/ml of carbenicillin, and incubated overnight at 30 ºC. The next morning the culture was used to inoculate Terrific Broth Modified (Fisher Scientific) media supplemented with 2 mM MgSO4 and 200 µg/ml carbenicillin at 1:250, then grown to and OD600 of 1-1.2. The cultures were centrifuged at 600 xg for 10 minutes, at 25 ºC, with the brake turned off. The cells were resuspended in an equal volume of M9 media with 100 µg/ml carbenicillin and grown for 1 h at 37 ºC. The cultures were induced with 1 mM IPTG and incubated overnight at 25 ºC with loose covers to allow gas exchange.
For SeMet labeling the frozen culture was restreaked onto a plate containing 200 µg/ml of carbenicillin and incubated overnight at 37 ºC. In the morning a starter culture from a single colony in ZYP-0.8G media with 200 µg/ml of carbenicillin incubated at 37 ºC. In the evening the culture was used to inoculate cultures at 1:12.5 ratio of M9 media with 100 µg/ml carbenicillin. The culture was grown overnight at 30 ºC. The next morning the culture was used to inoculate M9 media at a ratio of 1:20. The culture was incubated at 37 ºC till an OD600 of .5. Then 25 mg/L of lysine, phenylalanine, and threonine; 12.5 mg/L of isoleucine, leucine, and valine; and 15 mg/L of selenomethionine were added to the culture. The culture was grown for 15 minutes then induced with 1 mM IPTG and incubated at 25 ºC with a loose cover allowing air flow overnight.
For 15 N and 13 C labeling the cultures were started from a single fresh colony in ZYP-0.8G media with 200 µg/ml of carbenicillin overnight at 30 ºC. The cultures were used to inoculate 1:250 TB media and incubated at 37 ºC till an OD600 of 2. The cultures were centrifuged at 600 xg at 25 ºC for 20 minutes with the brake turned off. The supernatant was discarded and the cells were resuspended in an equal volume of M9 media with 100 µg/ml carbenicillin. The resuspended cells were added to M9 media with 100 µg/ml carbenicillin, 3 g/L 15N-NH4Cl, and 10 g/L of U-13C-glucose. The cultures were incubated at 37 ºC for 1 h then induced with 1 mM IPTG and incubated at 25 ºC overnight.
To purify the protein the cells were harvested by centrifugation at 4500 xg, at 4 ºC. To obtain the non-disulfide bonded form of the ToxR periplasmic domain the thiopropyl column was washed with 2 CV's of wash buffer, then with 3 CV's of wash buffer with 100 mM DTT. Two more CV's of wash buffer with DTT were added to the column and incubated overnight at 16 ºC. The following morning the protein was eluted and concentrated to about 2 ml for gel filtration as above.

Differential Scanning Fluorometry:
Differential scanning fluorometry was performed to assess the stability of the V.
cholerae dbToxRp in the presence and absence of additives as described (Midgett et al., 2017). STATA15 was used to analyze the data. The results are reported from three independent experiments as mean ± standard deviation.

Pull downs:
Pull downs were performed as described (Midgett et al., 2017). Briefly, the chitin binding domain-intein tagged ToxS periplasmic domain (CBDI-ToxSp) was captured on chitin beads from a lysate. After washing the beads purified ToxR periplasmic domain with and without the disulfide bond were added to the tubes with and without 2 mM sodium chenodeoxycholate. After the final wash SDS sample buffer was added to the beads and the tubes were boiled. Samples were diluted and run on a gel with the results quantified and statistics determined using STATA15. The results are from three independent experiments.
Native Vv-dbToxRp was crystallized by adding in a 1:1 ratio 4 mg/ml of Vv-dbToxRp to 2.0 M ammonium sulfate, and .1 M sodium cacodylate pH 6.3 in a sitting drop.
Crystals were cryo-protected by dragging the crystal through 75% paratone N and 25% paraffin oil. Data was collected at NSLS2 FMX beam line. The data was processed using XDS (Kabsch, 2010) COOT (Emsley and Cowtan, 2004). Chimera was used for structure visualization and analysis (Pettersen et al., 2004).
Acknowledgments: Funding was provided by NIAID R21-AI140740 and from the BioMt COBRE P20-GM113132. Sequencing was performed by the Molecular Biology Shared Resource at Dartmouth. We thank Dr. Karen Skorupski for her thoughtful comments about the manuscript. We also acknowledge the beam line staff at NSLS2-FMX; Martin Fuchs, Babak Andi, and Wuxian Shi for helping with data collection.
Authors contributions: FJK and CRM conceived the project and wrote the paper.
CRM also performed some of the experiments and solved the X-ray structures. RAS performed experiments as directed, primarily purifying and crystallizing the periplasmic domain for X-ray crystallography. MP performed the NMR experiments, wrote the methods for the NMR experiments, and prepared the HSQC overlay for inclusion in the figures.

Additional Information:
The authors declare no competing interest. All strains, plasmids, and protocols are available upon request. The coordinates for both structures have been deposited in the PDB: SeMet-Vv-dbToxRp, 6uue; native Vv-dbToxRp, 6utc.