Independent Promoter Recognition by TcpP Precedes Cooperative Promoter Activation by TcpP and ToxR

ABSTRACT Cholera is a diarrheal disease caused by the Gram-negative bacterium Vibrio cholerae. To reach the surface of intestinal epithelial cells, proliferate, and cause disease, V. cholerae tightly regulates the production of virulence factors such as cholera toxin (ctxAB) and the toxin-coregulated pilus (tcpA-F). ToxT is directly responsible for regulating these major virulence factors while TcpP and ToxR indirectly regulate virulence factor production by stimulating toxT expression. TcpP and ToxR are membrane-localized transcription activators (MLTAs) required to activate toxT expression. To gain a deeper understanding of how MLTAs identify promoter DNA while in the membrane, we tracked the dynamics of single TcpP-PAmCherry molecules in live cells using photoactivated localization microscopy and identified heterogeneous diffusion patterns. Our results provide evidence that (i) TcpP exists in three biophysical states (fast diffusion, intermediate diffusion, and slow diffusion), (ii) TcpP transitions between these different diffusion states, (iii) TcpP molecules in the slow diffusion state are interacting with the toxT promoter, and (iv) ToxR is not essential for TcpP to localize the toxT promoter. These data refine the current model of cooperativity between TcpP and ToxR in stimulating toxT expression and demonstrate that TcpP locates the toxT promoter independently of ToxR.

facilitate development of alternative methods of treatment, thereby reducing the global burden of cholera.
Upon ingestion, typically from contaminated water or food, V. cholerae colonizes the crypts of the villi in the distal portion of the small intestine and stimulates production of virulence factors essential for disease progression, such as the toxin-coregulated pilus and cholera toxin (TCP and CtxAB, respectively) (12)(13)(14)(15)(16)(17). Transcription of tcp and ctxAB is directly activated by ToxT (18)(19)(20)(21). Expression of toxT is highly regulated and positively stimulated by ToxR and TcpP, two membrane-localized transcription activators (MLTAs), which directly bind to the toxT promoter (toxTpro), with binding sites at 2104 to 268 and 255 to 237, respectively (18,(22)(23)(24)(25)(26)(27)(28). TcpP and ToxR are bitopic membrane proteins, each containing a cytoplasmic DNA-binding domain (within the PhoB and OmpR families, respectively), a single transmembrane domain, and a periplasmic domain (29). ToxR appears to have an accessory role in toxT regulation. Evidence supporting the model that ToxR assists TcpP in toxT expression includes that (i) TcpP binds downstream of ToxR, closer than ToxR to the putative RNA polymerase binding site on toxTpro, and (ii) overexpression of TcpP results in ToxR-independent toxT transcription activation (18,24,25,28). Furthermore, we have previously measured the single-molecule dynamics of TcpP and noted that deletion of toxR decreases but does not eliminate the prevalence of TcpP-DNA binding events (30). However, it remains unclear how TcpP and ToxR identify the toxTpro from the cytoplasmic membrane.
Signal transduction pathways in prokaryotes consist of one-component and twocomponent regulatory systems that manage cellular processes in response to extracellular information such as pH, temperature, chemical gradients, and nutrients (31)(32)(33). One-component regulatory systems combine their input and output functions in a single protein. MLTAs are a unique family of one-component regulators as they function from the cytoplasmic membrane, whereas the majority (;97%) of one-component regulators are localized in the cytoplasm (31). These one-component MLTAs like TcpP and ToxR comprise a sensor domain and an output domain that are separated by a transmembrane domain. MLTAs have been experimentally characterized in other, Grampositive and Gram-negative, pathogenic bacteria and have been shown to regulate genes important for pathogenesis (such as capsule production, acid tolerance, antibiotic resistance, virulence gene regulation, and natural competence) (34)(35)(36)(37)(38)(39)(40)(41)(42)(43). Using the Microbial Signal Transduction Database (MIST), we collected candidate MLTAs from 20 bacterial species and found that the prevalence and diversity of MLTAs are much higher than previously anticipated (see Fig. S1 in the supplemental material). These data indicate that MLTAs are more common among bacteria than previously appreciated. Yet, it remains unclear how MLTAs identify a specific promoter(s) while localized to the cytoplasmic membrane. Some challenges emerge in understanding how MLTAs affect their function of activating transcription in response to external stimuli. For example, diffusion of these regulators is constrained to the cytoplasmic membrane. Additionally, the chromosome structure, which is not static, is known to influence the association of an MLTA with its target sequence (44)(45)(46)(47)(48)(49)(50)(51)(52)(53). How MLTAs locate their target sequences while bound to the membrane represents a major gap in our knowledge. Here, we investigated the subcellular single-molecule dynamics of TcpP-PAmCherry to understand how TcpP localizes to the toxTpro and to develop a general model for how MLTAs identify their DNA targets.
Our approach was to apply superresolution single-molecule tracking (SMT) in living cells. Previous work demonstrated that TcpP molecules exhibit heterogeneous diffusion patterns (30,54). Here, we expand upon this earlier work to study the effect of specific mutations, which alter TcpP binding to DNA or the potential association of TcpP with ToxR, on TcpP subcellular mobility. By tracking the movement of TcpP-PAmCherry molecules within single living V. cholerae cells, we determined the distributions of the heterogeneous motions of TcpP and detected changes in these diffusion coefficients in response to targeted genetic alterations. From these data, we identify three biophysical states (fast diffusion, intermediate diffusion, and slow diffusion), we propose a biological role corresponding to each state, and we suggest an alternative model of toxT activation where TcpP independently identifies the toxTpro prior to assistance from ToxR.

RESULTS
Single-molecule tracking of TcpP-PAmCherry is useful to study promoter identification but cannot probe regulated-intramembrane proteolysis. To investigate the dynamics of individual TcpP molecules, we generated a V. cholerae strain in which the wild-type (WT) tcpP allele is replaced with one expressing TcpP fused at its C terminus to a photoactivatable fluorescent protein, PAmCherry (tcpP-PAmCherry). Levels and activity of TcpP are controlled by a two-step proteolytic process known as regulated intramembrane proteolysis (RIP) (55)(56)(57). Under RIP-permissive conditions (defined as LB, pH 8.5, 37°C, shaking at 210 rpm) the C terminus of TcpP becomes sensitive to proteolysis by Tsp, a site-1 protease, and YaeL, a site-2 protease; this sensitivity results in the inability of the cell to activate toxT expression. Under RIP-nonpermissive conditions (defined as LB, pH 6.5, 30°C, shaking at 110 rpm), TcpP is protected from RIP by TcpH (55)(56)(57).
We investigated whether we could assess RIP dynamics using single-molecule tracking. Like wild-type TcpP, TcpP-PAmCherry was sensitive to RIP in the absence of TcpH, indicated by lower levels of TcpP-PAmCherry in tcpP-PAmCherryDtcpH relative to tcpP-PAmCherry (see Fig. S2A in the supplemental material). Second, in both tcpP-PAmCherry and tcpP-PAmCherryDtcpH a smaller species of TcpP-PAmCherry was observed, referred to as TcpP-PAm* (Fig. S2A). A similar result has been observed for native TcpP in DyaeL cells and indicates RIP (56). Complementation of tcpP-PAmCherryDtcpH with a plasmid encoding TcpH resulted in a band with the mass of native TcpP (;29 kDa) (Fig. S3). These data indicate that TcpP-PAmCherry resists RIP in a TcpH-dependent fashion similar to native TcpP. As expected, native TcpP was not detected in the absence of TcpH. These data indicate that (i) TcpP-PAmCherry is sensitive to RIP, (ii) TcpH can protect TcpP-PAmCherry from RIP, and (iii) addition of PAmCherry to the C terminus of TcpP reduces RIP of TcpP-PAmCherry relative to TcpP. These conclusions are supported by similar levels of TcpA, CtxB, and toxT expression in tcpP-PAmCherry and tcpP-PAmCherryDtcpH (54) (Fig. S2A and Fig. S4). Notwithstanding the detectable levels of TcpP-PAmCherry on immunoblots of total proteins from tcpP-PAmCherryDtcpH, we observed almost no TcpP-PAmCherry molecules in our single-molecule tracking experiments. As a result, we are unable to collect sufficient data to perform any analysis of tcpP-PAmCherryDtcpH cells. Though we cannot determine how RIP influences TcpP-PAmCherry single-molecule dynamics, fusion of PAmCherry to the C terminus of TcpP does not affect its ability to stimulate toxT expression (Fig. S4). In addition, activity of TcpP is influenced by homodimerization, mediated by a periplasmic cysteine residue (C207) (58,59). We sought to determine if addition of PAmCherry to the C terminus of TcpP promotes its ability to dimerize. To test this, we measured toxT expression in both tcpP-PAmCherry and tcpPC207S-PAmCherry cells (Fig. S5). We found that PAmCherry does not compensate for loss of C207, suggesting that it does not stimulate dimerization of TcpP-PAmCherry. These data indicate that PAmCherry does not simulate dimerization of TcpP-PAmCherry. Lastly, addition of PAmCherry to the C terminus of TcpP does not affect the growth rate of V. cholerae (Fig. S6). Therefore, TcpP-PAmCherry is an effective tool to understand how TcpP locates the toxTpro from its position in the membrane.
Baseline dynamics of TcpP-PAmCherry. Single-Molecule Analysis by Unsupervised Gibbs sampling (SMAUG) characterizes the motion of molecules based on the collection of measured displacements (steps) in their single-molecule trajectories. SMAUG estimates the biophysical descriptors of a system by embedding a Gibbs sampler in a Markov chain Monte Carlo framework. This nonparametric Bayesian analysis approach determines the most likely number of mobility states and the average diffusion coefficient of single molecules in each state, the population of each state, and the probability of transitioning between different mobility states over the course of a single trajectory (54). In our previous study, we determined that TcpP-PAmCherry molecules in V. cholerae cells transition between multiple biophysical states: fast diffusion, intermediate diffusion, and slow diffusion (54).
Here, we collected a new robust set of TcpP-PAmCherry tracking data in living V. cholerae cells (54,454 steps collected from 7,601 trajectories) to further refine our analysis and to assign biochemical mechanisms to these biophysical observations (a sample of these tracks is shown in Fig. 1b; see also Video S1 at https://doi.org/10.5281/zenodo.5222485). Consistent with our previous results, we ascertained that TcpP-PAmCherry exists in three distinct states (slow diffusion, intermediate diffusion, and fast diffusion; blue, orange, and purple, respectively, in Fig. 1c). Furthermore, we determined that TcpP-PAmCherry molecules do not freely transition between all the diffusion states: we observe that TcpP-PAmCherry molecules can transition between the fast state (purple) and the intermediate state (orange) and /s), the average probabilities of transitioning between mobility states at each step are indicated as arrows between those two circles, and the circle areas are proportional to the weight fractions. Low-significance transition probabilities less than 4% are not displayed; for instance, the probability of TcpP-PAmCherry molecules transitioning from the fast diffusion state to the slow diffusion state is 1%. Numbers above the arrows indicate the probability of transition. between the intermediate state (orange) and the slow state (blue) freely, but there is no significant probability of transitions directly from the fast diffusion state (purple) to the slow diffusion state (blue) on successive steps (Fig. 1d). Thus, the intermediate diffusion state represents a critical biochemical intermediate between the slow and fast diffusion states.
The high transition probability of TcpP-PAmCherry molecules from the intermediate diffusion state to the fast diffusion state (50%) is unexpected, as the fast diffusion state represents the smallest population of TcpP-PAmCherry molecules (9%), with a low probability (8%) of TcpP-PAmCherry molecules transitioning from the fast diffusion state back to the intermediate diffusion state (Fig. 1d). While we cannot directly determine how RIP influences the dynamics of TcpP-PAmCherry, the stark difference in the transition probabilities and the populations of TcpP-PAmCherry in the fast and intermediate diffusion states suggests that fast-diffusing TcpP-PAmCherry molecules are potentially sensitive to some form of degradation.
Given this baseline for the dynamics of TcpP-PAmCherry, we hypothesize that (i) the three diffusion states (slow, intermediate, and fast) are features of TcpP-PAmCherry molecules with three biologically distinct roles; (ii) the slow diffusion state is occupied by TcpP-PAmCherry molecules interacting with DNA, such as toxTpro; and (iii) the intermediate diffusion state is influenced by ToxR. We further explore these three hypotheses with V. cholerae mutants below.
Mutation of the toxTpro decreases the slow diffusion state occupancy. We hypothesized that the slow TcpP-PAmCherry diffusion state encompasses molecules specifically interacting with DNA at its binding site in the toxTpro. The molecular weight of chromosomal DNA (chromosome 1, 2.96 Mbp) is much higher than that of any protein.
Thus, binding of TcpP-PAmCherry to this promoter on the chromosome should result in an extremely low apparent diffusion rate. To test our hypothesis, we removed key binding sites for TcpP (255 to 237) and both ToxR and TcpP (2112 to 11) in the toxTpro, generating tcpP-PAmCherry toxTproD(255-11) and tcpP-PAmCherry toxTproD(2112-11) (Fig. 2), both of which resulted in a drastic reduction in TcpA production, similar to that of a DtcpP mutant (Fig. S2A). toxT expression was reduced in tcpP-PAmCherry toxTproD(2112-11) but not in tcpP-PAmCherry toxTproD(255-11) (Fig. S4). It is possible that the toxTproD(255-11) mutation causes TcpP-PAmCherry and ToxR to stimulate expression of a nonfunctional toxT mRNA. Regardless, loss of either region of the toxTpro results in loss of production of the TcpA virulence factor.
Relative to the wild type ( Fig. 1), deleting both the ToxR and TcpP binding sites [toxTproD(2112-11)] reduces the percentage of slow-diffusing TcpP-PAmCherry to very low levels (7%; Fig. 2b). Thus, TcpP-PAmCherry in the slow diffusion state requires toxTpro; therefore, we propose molecules in this state are bound to toxTpro. On the other hand, loss of the TcpP binding site alone [toxTproD(255-11)] reduces the percentage of slow TcpP-PAmCherry molecules only subtly (from 43% to 34%; Fig. 2d). This result is consistent with earlier observations demonstrating that association with ToxR can restore the function of TcpP variants otherwise unable to bind the toxTpro (18,24).
Furthermore, our single-step analysis of TcpP-PAmCherry in the toxTproD(2112-11) cells indicates five distinct TcpP-PAmCherry diffusion states, an increase from three states in the wild type (Fig. 2b). In particular, the percentage of TcpP-PAmCherry molecules within the intermediate state overall increased (48% to 78%), but our analysis showed that these moderate moving molecules in fact cluster into three distinct substates (yellow, light orange, and orange in Fig. 2b). These intermediate TcpP-PAmCherry diffusion substates appear when TcpP-PAmCherry is unable to associate with the toxTpro. Though large-scale changes in the chromosome structure following the promoter deletion may play a role, these intermediate TcpP-PAmCherry diffusion substates may represent true biochemical interactions that are too short-lived to precisely distinguish and identify due to our current time resolution of 40 ms/acquisition. Further investigation is required to understand the specific biological roles of these substates, but indeed as discussed below, we detect these intermediate substates in all the other mutants studied here ( Fig. 3 and 4).
ToxR promotes TcpP-PAmCherry association with the slow and fast diffusion states. ToxR is a critical regulator of toxT expression through its role supporting TcpP interaction with the toxTpro (18,24,25). Prior studies have shown that TcpP and ToxR interact in response to low oxygen concentrations, and ToxR antagonizes H-NS from the toxTpro (24,60,61). Several models for TcpP-mediated toxT transcription implicate ToxR in recruitment of TcpP molecules to the toxTpro (18, 23-25, 28, 30). Another model invokes "promoter alteration" to suggest that ToxR promotes TcpP-toxTpro interaction by displacing the histone-like protein (H-NS) and altering DNA topology rather than recruiting TcpP molecules to the toxTpro (28).

Independent Promoter Recognition by TcpP
® tcpP-PAmCherry DtoxRS with toxR did not change overall levels of TcpA (Fig. S7). Complementation of tcpP-PAmCherry DtoxRS toxTproD(255-11) with ToxR did not restore TcpA to wild-type (WT) levels (Fig. S7). These data show that TcpP-PAmCherry can stimulate toxT expression and bind to the toxTpro independent of ToxR. WT TcpP can stimulate toxT expression independent of ToxR, but only upon TcpP overexpression (18,24). Due to reduced sensitivity of TcpP-PAmCherry to RIP, we measure higher levels of TcpP-PAmCherry relative to TcpP (Fig. S2A). This observation suggests that cooperativity between ToxR and TcpP is necessary only when levels of TcpP are low (i.e., when TcpP is sensitive to RIP).
As shown in Fig. 1d, we found that TcpP-PAmCherry molecules do not freely transition between all the diffusion states: the intermediate diffusion state is an important diffusion state for TcpP-PAmCherry molecules to transition between the fast and the slow diffusion states. Since the ToxR-TcpP interaction is proposed to enable TcpP to associate with the transcription complex at toxTpro (18,24), we reasoned that ToxR is responsible for the preferred intermediate-to-slow-state transition of TcpP-PAmCherry. However, in the DtoxRS mutant (Fig. 3b) as in the wild type (Fig. 1c), only TcpP-PAmCherry molecules in the slowest of the intermediate diffusion substates were likely to transition to the slow diffusion state (orange and blue diffusion states, respectively; Fig. S8b). These transition probabilities suggest that ToxR is not responsible for the restricted transition of TcpP-PAmCherry between the slow and fast diffusion states. Furthermore, the absence of ToxR reduced the probability of TcpP-PAmCherry entering the fast diffusion state and increased the probability of TcpP-PAmCherry leaving the fast diffusion state (Fig. 1d and Fig. S8b). Taken together, these data indicated that ToxR sequesters a portion of the total TcpP-PAmCherry population away from the toxTpro. We reasoned that increased levels of ToxR might sequester TcpP molecules to an inactive state (represented by the intermediate diffusion state). To test this hypothesis, we overexpressed ToxR in a tcpP-PAmCherry background and quantified virulence factor expression (i.e., TcpA) (Fig. S9). We found that elevated ToxR levels reduced virulence factor levels in both WT and tcpP-PAmCherry cells. Furthermore, overexpression of ToxR also decreased the percentage of TcpP-PAmCherry in the slow diffusion state (17% versus 43%) and resulted in the formation of a subintermediate diffusion state, similar to tcpP-PAmCherry DtoxRS (Fig. 4b). These data suggest that elevated levels of ToxR can repress toxT expression by reducing the percentage of TcpP molecules entering the slow diffusion state.

DISCUSSION
How MLTAs find their target sequences from the membrane represents a major gap in knowledge. Here, we started to address this by investigating single-molecule dynamics of TcpP-PAmCherry. Taken together with previous work, the data presented here demonstrate that TcpP-PAmCherry molecules diffuse in at least three distinct biophysical states (fast, intermediate, and slow diffusion) but do not freely transition between all diffusion states (54). We hypothesized that each of these biochemical states have distinct biological roles. Specifically, we hypothesized that the slow diffusion state represented TcpP-PAmCherry molecules interacting with the toxTpro. To test this hypothesis, we made targeted deletions to the toxTpro and of toxRS, and we mutated the TcpP DNA binding domain (K94E). Our biophysical measurements of these mutations support the hypothesis that the slow diffusion state is occupied by TcpP-PAmCherry molecules interacting specifically with DNA at toxTpro. Additionally, we observed that TcpP-PAmCherry molecules only transition to the slow diffusion state from the intermediate diffusion state and that ToxR is not responsible for this transition specificity. These data support a modified promoter alteration model (28) in which ToxR binds to the distal region of the toxTpro to promote TcpP binding to the proximal region of the toxTpro or, in the absence of its binding site, ToxR directly interacts with TcpP to stimulate toxT expression. Our data do not suggest that ToxR directs or recruits TcpP to the toxTpro.
While ToxR is critical for TcpP to stimulate toxT expression (18,24,27), our data demonstrate that TcpP-PAmCherry can support toxT expression and virulence factor production without ToxR, which may be a consequence of the greater stability of TcpP-PAmCherry than of native TcpP (see Fig. S2A and Fig. S4 in the supplemental material). Moreover, our single-molecule imaging finds a higher percentage of the TcpP-PAmCherry molecules in the slow diffusion state in tcpP-PAmCherry DtoxRS cells than Independent Promoter Recognition by TcpP ® in tcpP-PAmCherry DtoxRS toxTproD(255-11) cells (Fig. 3). In addition, prior DNase I footprinting experiments have demonstrated that in cells lacking toxR TcpP protects a larger region of the toxTpro (2100 to 232), i.e., TcpP protects most of the ToxR binding and TcpP binding sites in the DtoxRS mutant (18). Taken together, these results indicate that (i) ToxR is not essential for TcpP to locate the toxTpro and (ii) TcpP is able to interact with the toxTpro independent of ToxR. In addition, our data show that DtoxRS reduces the percentage of DNA-bound TcpP-PAmCherry but does not decrease the probability of TcpP-PAmCherry molecules transitioning from the intermediate state to the bound state ( Fig. 3 and Fig. S8b). Despite a reduction in the percentage of DNA-bound TcpP-PAmCherry, TcpP-PAmCherry stimulates WT toxT expression independent of ToxR (Fig. S4). These data support the promoter alteration model (28) in which, rather than ToxR recruiting TcpP to the toxTpro, ToxR assists TcpP to stimulate toxT transcription once TcpP independently associates with the toxTpro. Counterintuitively, in the absence of ToxRS TcpP-PAmCherry molecules have a lower probability of exiting the slow diffusion state (Fig. S8b). Given that RIP of TcpP-PAmCherry impedes our ability to image TcpP-PAmCherry, these data suggest that TcpP-PAmCherry molecules might be sensitive to RIP while interacting with the toxTpro and that ToxRS may inhibit RIP of TcpP while interacting with the toxTpro. If this is the case, given that we are unable to image TcpP-PAmCherry molecules that are sensitive to RIP, it might explain why we observe a lower percentage of TcpP-PAmCherry molecules in the slow diffusion state and yet we observe WT toxT expression in the absence of ToxRS. However, future experiments are required to determine if ToxRS inhibits RIP of TcpP while interacting with the toxTpro.
Under certain conditions, ToxR can negatively influence toxT expression. In response to stationary-phase accumulation of the cyclic dipeptide cyclic phenylalanine-proline (cyc-Phe-Pro), ToxR stimulates production of LeuO, resulting in downregulation of the tcpP regulator aphA (62,63). Our data suggest that ToxR can also reduce toxT expression by influencing TcpP-PAmCherry single-molecule dynamics (Fig. S8b). Deletion of toxRS reduces the overall probability of TcpP-PAmCherry molecules transitioning between the intermediate and fast diffusion states (Fig. S8b). Moreover, elevated levels of ToxR reduce both the percentage of TcpP-PAmCherry in the slow diffusion state and virulence factor production ( Fig. 3f and Fig. S9), suggesting that ToxR can antagonize toxT expression by promoting transition of TcpP molecules to the fast or subintermediate diffusion states. A similar phenotype has been reported previously (18). Lastly, prior electrophoretic mobility shift assays also indicate that ToxR can sequester TcpP from the toxTpro. In DtoxRS cells TcpP is able to bind to the toxTpro 273-145 (toxTpro lacking the ToxR binding region), but not in the presence of ToxR molecules (18). It remains unclear how ToxR sequesters TcpP-PAmCherry molecules from the slow diffusion state. However, we hypothesize that ToxR promotes TcpP molecules to transition away from the slow diffusion state to prevent aberrant toxT expression. Follow-up experiments are required to test this hypothesis.
Currently, the biological roles of the intermediate diffusion states (or intermediate diffusion substates) are unclear, but the intermediate states are certainly important, as TcpP molecules transition to the toxTpro-bound state from them. There is nearly a 10fold difference in diffusion coefficients between the slow and intermediate diffusion states (0.006 mm 2 /s versus 0.044 mm 2 /s, respectively; Fig. 1c). This difference cannot be explained by dimerization or interaction of ToxR and TcpP-PAmCherry alone: the mobility of membrane-localized proteins scales linearly with the number of transmembrane helices, such that increasing the number of transmembrane helices via dimerization from one to two would reduce the diffusion coefficient only by a factor of two (64). One possibility is that TcpP-PAmCherry molecules undergo fast diffusion in less protein-dense areas of the cytoplasmic membrane relative to TcpP-PAmCherry molecules undergoing intermediate diffusion. Prior single-molecule analysis of 209 membrane-localized proteins in Bacillus subtilis revealed that only 6% of all membrane proteins imaged were homogeneously distributed throughout the cytoplasmic membrane (65). Heterogeneous distribution of membrane-localized proteins in B. subtilis suggests that similar distribution of membrane-localized proteins in V. cholerae can occur. It remains unclear why the vast majority of these membrane-localized proteins in B. subtilis have heterogeneous diffusion dynamics. One possibility is that these membranelocalized proteins have different preferences for lipid-ordered and lipid-disordered membrane domains. Prior studies have demonstrated that transmembrane domain properties (e.g., surface area, length, and posttranslational modifications) are major factors in determining lipid-ordered or lipid-disordered membrane domain preference (66). We are currently exploring if lipid-ordered and lipid-disordered membrane domains influence diffusion dynamics of TcpP molecules within the fast and intermediate diffusion states.
Alternatively, it is possible that the diffusion coefficients of TcpP-PAmCherry molecules in the intermediate state are undergoing nonspecific interactions with DNA whereas the slowest TcpP-PAmCherry molecules are specifically bound at toxTpro. Our data show that there are some slow-moving TcpP-PAmCherry molecules when major regions of the toxTpro are deleted or when key residues within the DNA binding domain of TcpP are mutated (i.e., tcpP[K94E]-PAmCherry; Fig. 2 and 4). When considering our alternative model of nonspecific DNA binding by TcpP, our data suggest two possibilities: (i) TcpP-PAmCherry molecules in the slow diffusion state represent TcpP molecules that make both specific and nonspecific interactions with DNA or (ii) TcpP-PAmCherry molecules in the slow diffusion state interact specifically with non-toxTpro DNA (i.e., TcpP regulates additional genes). Several genes appear to have altered gene expression upon deletion of tcpPH (67). However, these experiments have yet to be replicated. Thus, future experiments would be required to test these hypotheses.
These results provide deep insights that further expand the model of cooperativity between ToxR and TcpP-PAmCherry. Our data demonstrate that ToxR assists TcpP to associate with the toxTpro even in the absence of the TcpP binding site, further supporting the established model of cooperativity between TcpP and ToxR. The data also show that TcpP can locate the toxTpro, interact with the toxTpro, and stimulate toxT expression independent of ToxR. This supports the promoter alteration model in which TcpP molecules independently associate with the toxTpro while ToxR enhances this association by altering toxTpro topology to stimulate toxT transcription. In addition to independently associating with the toxTpro, these data show that ToxR promotes transition of TcpP molecules to the fast and subintermediate diffusion states, shifting the equilibrium of TcpP molecules away from the toxTpro. The mechanism by which ToxR promotes transition of TcpP molecules away from the slow diffusion state is currently unclear but will be the subject of future investigation. Given that toxT expression is highly regulated, we speculate that sequestration of TcpP molecules from the toxTpro is yet another mechanism to fine-tune toxT expression. It is probable that other MLTAs, found in both Gram-negative and Gram-positive bacteria, have similar biophysical properties (Fig. S1). Continued exploration of MLTA biophysical properties could be leveraged to develop alternative strategies to inhibit MLTAs to treat bacterial infections without exacerbating the global antibiotic resistance crisis.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Escherichia coli and V. cholerae strains used here can be found in Table S1A in the supplemental material. Unless otherwise stated, E. coli and V. cholerae cells were grown on lysogeny broth (LB) plates, or in LB at 210 rpm, at 37°C. LB was prepared according to previous descriptions (68). To stimulate virulence, V. cholerae cells were diluted from overnight cultures in LB and subcultured under virulence-inducing conditions: LB, pH 6.5, 110 rpm, 30°C; filter sterilized. Here, the LB pH was adjusted by adding HCl (1 N) to pH 6.5 (60.05), and then the medium was filter sterilized to maintain pH. Where appropriate, antibiotics and cell wall intermediates were added at the concentrations given in parentheses: streptomycin (100 mg ml 21 ), ampicillin (100 mg ml 21 ), and diaminopimelic acid (DAP) (300 mM).
Plasmid construction. Plasmid vectors were purified using the Qiagen miniprep kit. Plasmid inserts were amplified from V. cholerae genomic DNA using Phusion high-fidelity polymerase (Thermo Scientific). Splicing by overlap extension was used to combine the entire plasmid insert sequences together; see Table S1B for the primer list. Plasmid vector was digested by restriction digestion using KpnI-HiFi and XbaI (New England BioLabs) at 37°C for 2 h. After digestion, the plasmid vector and insert were added to Gibson assembly master mix (1.5 ml insert, 0.5 ml vector, 2ml master mix) (New England BioLabs) and incubated at 50°C for 1 h. Assembled plasmid was electroporated into E. coli lpir cells and recovered on LB plates with ampicillin and DAP. Independent Promoter Recognition by TcpP ® Bacterial strain construction. Strain construction follows the protocol outlined in reference 69. Briefly, E. coli lpir harboring the pKAS plasmid and the donor V. cholerae strain were incubated in LB (broth or agar) supplemented with DAP overnight at 37°C. The remaining cells were then spread on LB plates containing ampicillin or thiosulfate-citrate-bile-sucrose (TCBS) plates containing ampicillin. Counterselection for loss of the pKAS construct by V. cholerae cells was done by incubating cells in LB for 2 h and then for 2 h with 2,500mg ml 21 streptomycin (both at 37°C, 210 rpm). Twenty microliters of this culture was spread onto LB plates containing 2,500 mg ml 21 of streptomycin and incubated overnight at 37°C. Streptomycin-resistant colonies were screened for the chromosomal mutation of interest via colony PCR using Taq DNA polymerase (Thermo Fisher). Genomic DNA was purified from possible mutants and sequenced (Genewiz) to validate the exchange. Because overlapping open reading frames encode TcpP and TcpH, tcpH was cloned downstream of PAmCherry to maintain its expression, and a stop codon was introduced within the first three codons of the native tcpH coding sequence to prevent out-of-frame translation of PAmCherry.
Growth curves. V. cholerae strains were initially grown on LB plates containing streptomycin (100 mg ml 21 ) overnight at 37°C, and then an individual colony was picked and grown overnight in LB at 37°C. V. cholerae cells were diluted to an optical density at 600 nm (OD 600 ) of 0.01 from the overnight LB into a 96-well plate (Cell Pro) with 200 ml of virulence-inducing medium per well. The plate was then incubated at 30°C with shaking every 30 min before each measurement in a SPECTROstar Omega plate reader (BMG Labtech).
RT-qPCR. RNA was extracted from V. cholerae cells grown under virulence-inducing conditions. RNA was preserved by resuspending pellet cells in 1 ml TRIzol (Sigma-Aldrich) and then purified using an RNeasy kit (Qiagen). RNA was further purified with Turbo DNase treatment. RNA quantity and quality were measured via UV-visible (UV-Vis) spectrophotometry (NanoDrop ND-1000) and by detection of large and small ribosomal subunits via 2% agarose gel. RNA was then converted to cDNA using Superscript III reverse transcriptase (Thermo Scientific). Real-time quantitative PCR (RT-qPCR) was performed using 5 ng of cDNA in SYBR green master mix (Applied Biosystems). RecA was used as a housekeeping gene of reference to calculate the cycle threshold values (DDC T ) (70,74). See Table S1B for primers.
Single-molecule microscopy. V. cholerae strains were grown on LB plates containing streptomycin (100 mg ml 21 ) overnight at 37°C, and then an individual colony was picked and grown overnight in LB at 37°C. V. cholerae cells were diluted from LB under virulence-inducing conditions and grown until they reached mid-log phase. They were then washed and concentrated in M9 minimal medium with 0.4% glycerol. A 1.5-ml droplet of concentrated cells was placed onto an agarose pad (2% agarose in M9, spread and flattened on a microscope slide) and covered with a coverslip. Cells were imaged at room temperature using an Olympus IX71 inverted epifluorescence microscope with a 100Â 1.40-numerical-aperture (NA) oil-immersion objective, a 405-nm laser (Coherent Cube 405-100; 50 W/cm 2 ) for photoactivation, and a coaligned 561nm laser (Coherent-Sapphire 561-50; 210 W/cm 2 ) for fluorescence excitation. Fluorescence emission was filtered with appropriate filters and captured on a 512-by 512-pixel Photometrics Evolve electron-multiplying charge-coupled device (EMCCD) camera. To prevent higher-order excitation during photoactivation, a pair of Uniblitz shutters controlled the laser beams such that samples were exposed to only one laser at a time. During imaging, the cells were given a 40-ms dose of 405-nm light every 90 s. Images were collected continuously every 40 ms, and acquisitions lasted 5 to 7 min each.
Data analysis. Recorded single-molecule positions were detected and localized based on point spread function fitting using home-built code, SMALL-LABS (71). This program reduces biases due to background subtraction, increasing the precision of each molecule localization. Subsequent localizations of the same molecule were then connected into trajectories using the Hungarian algorithm (72,73). All trajectories from each movie for a given condition were combined and analyzed together using the Single-Molecule Analysis by Unsupervised Gibbs sampling (SMAUG) algorithm (54). This algorithm considers the collection of steps in all trajectories and uses a Bayesian statistical framework to estimate the parameters of interest: number of mobility states, diffusion coefficient, weight fraction, transition probabilities between states, and noise.
Data availability. The data presented here will be made available from the corresponding authors upon request.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.  We declare no conflicts of interest.