INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Pathogenic bacteria control the expression of virulence genes in response to environmental signals. One well-characterized virulence gene regulatory system is the ToxRST system of the gram-negative bacterium Vibrio cholerae (10). ToxR is an integral inner membrane protein with a cytoplasmic DNA binding domain homologous to that of the OmpR family of transcriptional activators (29, 30). ToxR interacts with another inner membrane protein, ToxS (26), to activate transcription at three different virulence gene promoters: ctxAB, which encodes cholera toxin (27); ompU, which encodes an outer membrane protein (42); and toxT, which encodes a cytoplasmic AraC-like transcriptional activator (16). ToxR-regulated genes appear to be modulated by extracellular signals such as pH, osmolarity, chemoattractant amino acids, temperature, and oxygen tension (12, 28). The gene encoding ToxR was first identified by a genetic screen of a V. cholerae plasmid library for genes that activate the transcription of a ctx-lacZ fusion in Escherichia coli (27). ToxS was found in a subsequent screen to dramatically enhance ToxR-mediated transcriptional activation (26). The toxS gene is located downstream of toxR in an operon. Together, ToxR and ToxS positively control the expression of toxT (9). ToxT, in turn, activates at least eight different virulence gene promoters, including the ctxAB and tcpA promoters, as part of a virulence gene regulatory cascade (10).

The exact mechanism by which ToxRS activates the expression of toxT, ctxAB, and the virulence cascade is not known. When toxR is overexpressed, it can activate transcription in the absence of toxS. In contrast, when toxR is expressed from its own promoter, toxS is required for activation (26). The function of ToxS appears to be to lower the effective concentration of ToxR that is required for activation (31), but the mechanism by which this is accomplished is not understood.

Several approaches have been used to investigate the nature of ToxRS in the activation complex and the multimeric state of ToxR within this complex. Chemical cross-linking experiments suggest that ToxR can form dimers with itself as well as with ToxS (33). Further, a toxR allele encoding the E51K change is a dominant-negative mutation, suggesting that this defective ToxR variant is capable of forming complexes with ToxR and of inhibiting the activation complex (31). Experiments with fusion proteins also suggest that ToxR is a dimer or multimer in the activation complex. Fusion proteins designed to dimerize ToxR in the periplasm (i.e., proteins in which the GCN4 leucine zipper dimerization domain is substituted for the ToxR periplasmic domain) can activate transcription (19, 32). Replacement of the ToxR transmembrane domain with the glycophorin A transmembrane dimerization motif resulted in a protein that is active for transcriptional activation, and otherwise isogenic constructs carrying mutant glycophorin A transmembrane segments that fail to dimerize in vitro did not support activation (21). A ToxR-Bla periplasmic fusion protein that was designed to be a transmembrane monomer could activate transcription in one study (32), but a similar construct could not activate transcription in another study (19). These differences among various Bla fusion constructs may reflect variations with respect to their susceptibility to proteolysis or potential to multimerize in the periplasm (13). Overall, the results with various toxR mutants and fusion constructs, together with the fact that ToxR interacts with a large region of the ctx promoter that spans positions 40 through 80 (35), suggest that the ToxR activation complex is some sort of an oligomeric structure.

To get a more detailed picture of how ToxRS activates virulence genes, a two-layered genetic system was created to isolate positive control mutations in toxR and toxS that are specifically defective for transcriptional activation but that still retain ToxR-DNA binding activity. The first component utilized localized mutagenesis and the challenge phage system to select for alleles of toxRS that retained DNA binding activity. A simultaneous colorimetric screen based on the differential expression of a ctx-lacZ fusion was then used to identify those clones carrying toxRS alleles defective for activation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, phages, plasmids, and oligonucleotides. Bacteria, phages, plasmids, and oligonucleotides are listed in Table 1. Bacterial strains were stocked at 70°C in 25% glycerol. Antibiotics were used at the following concentrations: 100 µg/ml for ampicillin, 25 µg/ml for kanamycin, and 10 µg/ml for chloramphenicol. MacConkey agar (Difco) was supplemented with 2% lactose.

General and region-directed mutagenesis of pTSK and the method of mutant isolation. Plasmid pTSK was mutagenized by transformation of plasmid DNA into Salmonella typhimurium TSM215 (mutD) and TSM216 (mutS). Transformants were grown overnight at 37°C in Luria-Bertani (LB) broth with chloramphenicol, and plasmid DNA was prepared (23). Region-directed mutagenesis of a 488-bp region encoding 170 amino acids of the periplasmic region of toxR and the N-terminal 35 amino acids of toxS was performed by amplifying pTSK DNA with the PCR primers toxR-FR2 and toxR-CL under standard conditions for amplification (7). The PCR products were phenol extracted, ethanol precipitated, and digested with BstXI and XhoI. The enzymes were removed by a second phenol extraction and ethanol precipitation. Plasmid pTSK DNA was digested with BstXI and BamHI, and the 6.9-kb band containing the vector, the 5' end of toxR, and the 3' end of toxS was isolated by agarose gel electrophoresis and with GeneClean (Bio 101) and ligated to the PCR products. Ligated DNA was transformed into E. coli X90 by electroporation (23). Approximately 10⁵ colonies were collected and pooled, and mutagenized plasmid DNA was isolated.

-Galactosidase assays. -Galactosidase assays were performed in a microtiter format (2). Cells were grown overnight at 37°C in LB medium supplemented with appropriate antibiotics, diluted 1:31 in 96-well microtiter dishes, and incubated for 3 h at 37°C. Cell density was determined by A₆₅₀. Host cells were lysed by incubation with a high-titer lysate of bacteriophage T4 in -galactosidase buffer for 30 min at 37°C (4). Kinetic enzyme assays were performed at 25°C by measuring A₄₂₀ over time with a computer-controlled microtiter plate reader and SOFTmax software (Molecular Devices). Activities were calculated with the following equation: activity = 1,000bA₆₅₀¹f¹, where b is the slope of the linear least-squares fit to the plot of A₄₂₀ versus time (in minutes) and f (0.2) is the fraction of cells added to the total volume of lysate. A₄₂₀ was determined every 2 min for an hour. Values are the averages of results from at least four independent assays.

Challenge phage assays. Overnight cultures of strain MS1868 carrying pTSK and mutant derivatives (pACYC184 derivatives) were diluted 1:50 into 3 ml of LB medium supplemented with chloramphenicol at 10 µg/ml and grown for 3 h at 37°C to an A₆₀₀ of 0.1. MS1868/F' lacI^q/pJAM3, MS1868/F' lacI^q/pJAM3TMHistag, and MS1868/F' lacI^q/pTacterm (pBR322 derivatives) were grown similarly in LB medium with ampicillin and induced with 100 µM IPTG (isopropyl--D-thiogalactopyranoside) for 0.5 h. Phage P22 ctx8 (0.1 ml) at 10¹⁰ phage per ml was mixed in a 96-well microtiter dish with 0.1 ml of cells. After a 20-min incubation at room temperature, six fivefold serial dilutions were made for each infection and samples from each of the dilutions were plated as a series of 5-µl spots on LB agar containing chloramphenicol and kanamycin for pACYC184 derivatives and on LB agar containing ampicillin, 100 µM IPTG, and kanamycin for pBR322 derivatives. The plates were incubated at 37°C for 48 h. The number of cells surviving infection was then determined by counting the Kan^r colonies growing in spots with less than 20 colonies. Uninfected input cells were diluted by a factor of 10⁴, and 10-µl aliquots of cells were plated on LB agar with appropriate antibiotics to select for the plasmid-containing cells. The number of uninfected cells was determined after 16 h of growth. The fraction lysogeny was calculated as the number of kanamycin-resistant lysogens divided by the number of uninfected input cells. Values are the averages of results from at least three independent assays. Spot challenge phage assays were performed as described above, but infected cells were plated as 5-µl spots from four 10-fold serial dilutions.

Immunoblot assays. Salmonella strains were grown as described above for the challenge phage assays. Cells (2 ml) were harvested by centrifugation and resuspended in 0.2 ml of 2× sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis sample buffer and boiled for 5 min, and proteins were resolved on a 16.5% total acrylamide-3% bisacrylamide gel (40). Proteins were transferred to nitrocellulose by electrophoresis and probed with a rabbit anti-ToxR antibody (31). The blot was then probed with a secondary goat anti-rabbit immunoglobulin G antibody conjugated to horseradish peroxidase (Organon Teknika) and detected with an enhanced chemiluminescence detection kit (ECL; Amersham) in combination with autoradiography.

Construction of plasmids and phages and site-directed mutagenesis. The toxS3 (L33S) allele and a toxS deletion mutation (toxS) were constructed by the same site-directed mutagenesis procedure. The L33S change carried by toxS3 was constructed by PCR by amplifying pTSK DNA with the primers L33S and toxR-CL (Table 1). The PCR product was digested with BstXI and XhoI. Plasmid pTSK was digested with BstXI and XhoI, and the 6.9-kb vector containing the fragment was recovered by agarose gel electrophoresis and with GeneClean (Bio 101). The 488-bp PCR product was ligated to the vector fragment, and DNA was transformed into X90. The toxS allele was constructed by PCR amplification of pTSK with the primers toxS and toxRC-L and by replacing the BstXI-XhoI fragment with the PCR product that introduces the deletion within toxS. The interruption of the ToxR coding sequence at the transmembrane domain by the st11 His₆ tag (25) (the toxRTMHistag mutation) was constructed by amplifying pTSK DNA with the primers Histag and toxR-CL. The PCR product was digested with SalI and BstXI, ligated to gel-purified vector DNA (pJAM3 digested with SalI and BstXI), and transformed into X90/F' lacI^q. The st11 sequence was introduced to stabilize the protein against proteolysis as well as to aid in purification of a cytoplasmic form of ToxR on an immobilized metal affinity column. pCTX7 was constructed by cleaving pPC30 DNA with BamHI and XhoI. The PCR primers toxampBR and OmntL were used to amplify DNA from a single plaque of P22 ctx7 (35). PCR fragments were digested with EcoRI and BamHI and ligated into EcoRI- and BamHI-digested pPC30 DNA. Phage (ctx-lacZ)7 was constructed with a plasmid by phage cross between pCTX7 and RS45 (41). A single-copy lysogen of (ctx-lacZ)7 was made in X90 to construct strain JDP169 and confirmed as a single-copy lysogen by PCR (36).

Purification and N-terminal sequencing of ToxRTMHistag. JDP169/pJAM3toxRTMHistag was grown overnight in 50 ml of LB with antibiotics at 37°C. Cells (25 ml) were harvested by centrifugation (3,000 × g for 5 min), resuspended in 2 ml of sonication buffer (50 mM NaH₂PO₄ [pH 8.0], 10 mM Tris-HCl, 100 mM NaCl) and sonicated for three 30-s intervals on ice. The lysate was cleared of insoluble debris by centrifugation (12,000 × g for 15 min), and 2 ml of the supernatant was loaded onto a Talonspin metal affinity spin column (Clontech). The column was washed three times with sonication buffer, and the protein was eluted by addition of 1 ml of 100 mM EDTA (pH 8.0). The protein (50 ng) was resolved on an SDS-12% polyacrylamide gel and transferred to a polyvinylidene difluoride membrane by electroblotting. The membrane was stained with Coomassie brilliant blue (0.02%), and the slice containing the band corresponding to the product of the toxRTMHistag mutant (ToxRTMHistag) was isolated and sequenced by standard procedures with an ABI 476 protein sequencer at the Dartmouth Molecular Biology Core Facility.

Membrane localization experiments. Localization of ToxR and ToxR107 to the bacterial membrane was determined by immunoblot assay of cellular fractions (39). S. typhimurium cultures (50 ml) were grown in LB medium containing chloramphenicol to the middle-log phase of growth (optical density at 600 nm = 1.0). All subsequent steps were performed on ice or at 4°C. Cells were pelleted by centrifugation (5,000 × g), the pellets were resuspended in 0.5 ml of 200 mM Tris-HCl (pH 8.0), and 0.5 ml of sucrose buffer (50 mM Tris-HCl [pH 8.0], 1 M sucrose) was added. Cells were diluted with 1 ml of H₂O, and 10 µl of 0.5 M EDTA and 10 µl of 10-mg/ml lysozyme were added. After a 30-min incubation, MgSO₄ was added to 20 mM. Spheroplasts were pelleted at 5,000 × g for 10 min. The spheroplasts were resuspended in 5 ml of 50 mM Tris-HCl (pH 8.0) and sonicated three times for 15 s at 50% duty to lyse the cells. Unbroken cells were pelleted at 5,000 × g for 10 min, and the supernatants were saved as the total fraction. The total fractions (2 ml) were centrifuged at 230,000 × g for 15 min, and the supernatants were removed (soluble fractions). The membrane pellets were resuspended in 0.4 ml of 50 mM Tris-HCl (pH 8.0) (membrane fractions). Samples were diluted in 2× SDS-polyacrylamide gel electrophoresis loading buffer and boiled for 5 min, and 200 µl each of the total and soluble fractions and 40 µl of the membrane fraction were loaded onto an SDS-16.5% polyacrylamide gel. The proteins were transferred to nitrocellulose and subjected to immunoblot analysis with an anti-ToxR antibody.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Rationale for the genetic selection and screen. To study the mechanism of transcriptional activation by ToxRS, a genetic system was designed to isolate toxR and toxS positive control alleles that encode proteins that retain ToxR-DNA binding activity but are defective for transcriptional activation (Fig. 1). In the first step, the ToxR challenge phage was used to select for mutants that still retained ToxR-DNA binding activity (3, 35). In the second step, a genetic screen based on differential expression of a ctx-lacZ fusion was used to identify mutants defective for ToxRS-mediated activation of transcription (Fig. 1).

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Details of the two-layered selection and screen for positive control mutations in toxRS that retain ToxR-DNA binding activity but that are defective for transcriptional activation (toxR* and toxS* alleles). The system utilizes the challenge phage selection for ToxR-DNA binding in combination with a genetic screen based on loss of ToxRS-dependent activation of a ctx-lacZ promoter fusion. (A to C) Steps utilized for the isolation of mutants; (D and E) molecular basis for the selection and screen. (A) Plasmid pTSK served as the source of toxRS as well as the target for mutagenesis. Plasmid DNA was mutagenized by passage through mutator strains of S. typhimurium (mutD or mutS) or by PCR-mediated region-directed mutagenesis of toxRS. (B) Mutagenized plasmid DNA was introduced into JDP152 carrying a plasmid-borne (pCTX7) ctx-lacZ reporter. Wild-type toxRS activates transcription of the ctx-lacZ fusion, whereas toxRS alleles that are defective for transcriptional activation do not. Pooled bacterial clones were grown to exponential phase in LB medium, infected with challenge phage P22 ctx8, and plated on MacConkey lactose indicator media supplemented with appropriate antibiotics, including kanamycin. Because the decision between lytic and lysogenic development of P22 ctx8 is dependent on ToxR-DNA binding, only toxRS alleles that retain DNA binding activity survive as kanamycin-resistant lysogens. (C) Among the survivors, a subset carried toxR* or toxS* alleles that were defective for transcriptional activation and grew as white-to-pink colonies on MacConkey lactose media. Filled circles symbolize colonies carrying toxR+ and toxS+, and open circles symbolize colonies carrying toxR* or toxS* alleles. Clones carrying the toxR* or toxS* alleles were collected for further analysis. Molecular interactions in the two-layered selection and screen for toxR* or toxS* mutants are shown (D and E). Wild-type ToxRS protein interacts with pCTX7 to activate the transcription of the ctx-lacZ fusion (D), while ToxR simultaneously represses the transcription of the antirepressor gene (ant) carried by the P22 ctx8 challenge phage. The protein encoded by a toxR* or toxS* allele (symbolized by an X in panel E) disrupts the transcriptional activation of ctx-lacZ by ToxRS, but the P22 ctx8 challenge phage is channeled into lysogenic development by ToxR-mediated repression of P22 ant.

4

6

Isolation of toxR and toxS positive control mutations. Plasmid pTSK (toxR⁺S⁺) was passaged through Salmonella mutator strains carrying either a mutD or mutS mutation and purified. Mutagenized plasmid DNA was then transformed into the challenge phage host JDP152 carrying the ctx-lacZ fusion and plated. Transformants were pooled, grown until they reached exponential phase, and infected with P22 ctx8. The infected cells were diluted, so approximately 10³ cells were plated on MacConkey selection plates after challenge phage infection. White and pink lysogens which arose at a frequency of 10³ were selected from three different pools of mutagenized pTSK DNA for each mutator strain.

Mapping the toxR mutations by DNA sequencing. All of the mutD- and mutS-induced mutations occurred in toxR. Surprisingly, many of the mutations mapped to the region encoding the periplasmic domain of ToxR (toxR102, toxR104, and toxR106) (Fig. 2; Table 2). Two of these alleles contained frameshift mutations: toxR102 (E276fs, where fs means frameshift) and toxR106 (K239fs). A deletion occurred within a poly(dG · C) tract in toxR102 and a poly(dA · T) tract in toxR106. A stop codon was introduced into the region encoding the periplasmic domain of toxR104 (W229 [amber]). Three mutations were isolated in the region encoding the ToxR cytoplasmic domain. Two mutations, toxR101 (M98fs) and toxR103 (S93fs), produced proteins that bound to DNA very weakly in the spot challenge phage assay (Fig. 2). A third mutation, toxR105, resulted in the E82K change in the predicted DNA binding and transcriptional activation domain (31).

View larger version (25K): [in this window] [in a new window]   FIG. 2.   Schematic representation of toxR and toxS alleles that are defective for transcriptional activation but that still retain ToxR-DNA binding activity. The wild-type ToxR and ToxS proteins are shown at the top. The putative ToxR-DNA binding motif (DNA) is shown as an open box, the transmembrane domain (TM) is shown as a dark line, and the periplasmic domain (P) is shown as a filled box. ToxS has a transmembrane domain close to the N terminus. Frameshift peptides are signified by a hatched box. The length of the toxR or toxS coding sequence is shown with the length of the frameshift peptide (if any) in parentheses. Mutations toxS2 and toxS3 carry the change L33S. Allele toxS2 was isolated by region-directed PCR-mediated mutagenesis and carried a silent secondary mutation in the nontranslated region between toxR and toxS. To confirm that the L33S change was solely responsible for the phenotype, toxS3 (L33S) was constructed by site-directed mutagenesis. The toxS and cytoplasmic variant ToxRTMHistag constructions are described in Materials and Methods. The ability of each derivative to activate the transcription of ctx-lacZ is shown. A + indicates >90% activity, and a  indicates activity below 25% of that of the wild type. DNA binding activity as measured by the challenge phage assay with phage P22 ctx8 is indicated. A + signifies full DNA binding activity (fraction survival of >5 × 103), and a  indicates fraction survival of <1 × 104.

PCR mutagenesis of the region encoding the periplasmic domain of ToxR and the N terminus of ToxS. To study the potential role of the ToxR periplasmic domain and ToxS in activation, mutations were generated by two-layered selection and screen by PCR-mediated mutagenesis (7) of the region encoding the periplasmic domain of ToxR and the N terminus of ToxS. Four of six mutations mapped to a region encoding the periplasmic domain of ToxR (Fig. 2; Table 2): toxR107 (N200fs), toxR108 (P271fs), toxR109 (L280Oc), and toxR110 (G247Op). Surprisingly, no missense mutations were isolated in toxR. We isolated two toxS mutations that impaired the ability of ToxR to activate transcription but that left ToxR-DNA binding activity intact (toxS1 and toxS2) (Fig. 2; Table 2). Allele toxS1 encodes the first 36 amino acids of ToxS as well as a frameshift peptide of 5 amino acids. Allele toxS2 encodes a single amino acid substitution, L33S, as well as a base change in the untranslated region between toxR and toxS. Allele toxS3 (L33S) was constructed by site-directed mutagenesis to separate the L33S change from the spacer region mutation carried by toxS2. The L33S change was shown to inhibit activation but not ToxR-DNA binding.

Activation of toxin gene transcription by toxR and toxS positive control alleles. To quantify transcriptional activation by the toxR and toxS alleles, -galactosidase assays were performed on E. coli JDP169 [(ctx-lacZ)7] carrying one of the different pTSK derivatives (Fig. 3) (2, 4). As expected, all of the toxR and toxS alleles were found to drastically impair transcriptional activation. ToxR by itself (toxS) activates transcription to a level 20-fold above basal expression of the ctx-lacZ fusion. The addition of toxS enhanced ctx transcription by another fivefold (toxR⁺ toxS⁺). The toxR and toxS mutations decreased activation by greater than 10-fold.

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Activation of ctx-lacZ in E. coli by toxR or toxS positive control alleles. Plasmid pTSK or pTSK variants were the source of toxRS expression in these assays. Wild-type ToxRS activates transcription from the V. cholerae O395 ctxAB promoter carried on a lysogenic lambda phage, (ctx-lacZ)7, in E. coli JDP169. -Galactosidase activity is shown in Barrick units for each mutant allele. Plasmid JDP169/pACYC184 is included as a toxRS-negative control.

Quantitation of ToxR DNA binding in toxR and toxS mutant strains. To test the ability of ToxR to bind to DNA, quantitative challenge phage assays were performed on S. typhimurium MS1868 carrying the toxR and toxS derivatives of pTSK (Fig. 4). The fraction survival can be used to compare the relative DNA binding activities encoded by each toxR and toxS allele. Most of the toxR and toxS alleles expressed strong ToxR-DNA binding activity. Only mutants toxR101 (M98fs) and toxR103 (S93fs) containing truncations in the cytoplasmic domain failed to interact strongly with the ctx promoter fragment carried by P22 ctx8. All of the toxR mutations that truncate the periplasmic domain up to the transmembrane domain, the toxS1 and toxS3 (L33S) alleles, and the toxR105 allele with the cytoplasmic amino acid substitution (E82K) appear to encode ToxR molecules that bind to DNA tightly, as measured by the challenge phage assay.

View larger version (30K): [in this window] [in a new window]   FIG. 4.   Challenge phage assay results showing ToxR-DNA binding activity produced by toxR or toxS positive control alleles. Salmonella strain MS1868 carrying pTSK (toxR+ toxS+) or each mutant allele was grown to middle-log phase in LB media at 37°C and infected with P22 ctx8 carrying the ToxR binding site of the ctx promoter from V. cholerae 569B. Challenge phage assays are described in Materials and Methods. ToxR acts as the repressor of P22 lytic development in these assays. Fraction survival is the number of cells surviving infection divided by the number of input cells. Values were averaged from the results of three independent experiments and differed by less than twofold.

Role of ToxS in DNA binding and transcriptional activation by ToxR. To test if toxS is required for ToxR-DNA binding in the challenge phage system and in ToxR-mediated transcriptional activation, a toxS mutant was constructed. As shown in Fig. 3, ToxR activates transcription poorly in the absence of toxS, but challenge phage assays with P22 ctx8 show that ToxS protein is not necessary for ToxR-DNA binding when ToxR is expressed from a pTSK derivative carrying a toxS mutation (Fig. 4).

Steady-state level of ToxR in Salmonella strains expressing toxR and toxS positive control alleles. To test if the steady-state level of ToxR protein was altered by toxR and toxS mutations, immunoblot assays with an anti-ToxR antibody were performed (Fig. 5). MS1868/pTSK (toxR⁺S⁺) produced a strong signal corresponding to full-length ToxR protein. Surprisingly, the toxR mutants carrying periplasmic truncations appeared to be unstable and proteolytically processed to a population of smaller forms (Fig. 5A, lanes 3 to 9). One proteolytic form that is shared by all ToxR periplasmic truncation derivatives is a 22-kDa intermediate. The size of the ToxR region that includes the N terminus and the transmembrane domain can be predicted to be 22 kDa. Mutations toxR101 and toxR103 resulted in truncations prior to the transmembrane domain. These were not detectable by immunoblot analysis. Production of ToxR protein by toxR or toxS alleles that do not carry a truncated toxR coding sequence are shown in Fig. 5B. The level of full-length ToxR protein produced by the toxR105 (E82K) mutant was similar to that of the wild type, and the level produced by the toxS1 as well as the toxS mutant was slightly lower. In contrast, the toxS3 (L33S) mutation reduced the amount of full-length ToxR severely, but some proteolytic products, including the 22-kDa intermediate, were detectable. Thus, the toxS3 (L33S) mutation has a phenotype that is similar to a periplasmic toxR truncation in that the proteolyzed ToxR protein is still able to interact with DNA (Fig. 4) but is not able to activate transcription (Fig. 3).

View larger version (31K): [in this window] [in a new window]   FIG. 5.   Immunoblot analysis of the steady-state level of ToxR protein produced by pTSK derivatives expressing toxR and toxS positive control alleles. (A) Immunoblot analysis of ToxR truncations produced in vivo in S. typhimurium MS1868. Whole-cell extracts were prepared for each mutant, and proteins were separated on a denaturing SDS-polyacrylamide gel. Cells were grown under conditions identical to those of the challenge phage assays. The proteins were transferred to nitrocellulose, and an anti-ToxR antibody was used to detect cross-reactive species. Schematic representations of each variant ToxR protein are shown above each lane (Fig. 2). Lanes: 1, toxR toxS negative (pACYC184); 2, toxR+ toxS+ (pTSK); 3, toxR109 (L280Oc); 4, toxR102 (E275fs); 5, toxR108 (P271fs); 6, toxR110 (G247Op); 7, toxR106 (K239fs); 8, toxR104 (W229Am); 9, toxR107 (N200fs); 10, toxR101 (M98fs); 11, toxR103 (S93fs). (B) Immunoblot analysis of toxR* or toxS* alleles that carry the full-length toxR coding sequence was performed as described for panel A. Lanes: 1, toxR+ toxS+ (MS1868/pTSK); 2, toxR105 (E82K); 3, toxS1 (E37fs); 4, toxS3 (L33S); 5, toxS. See the legend to Fig. 2 for an explanation of symbols and abbreviations.

Effect of toxR and toxS mutations on toxR⁺. Multimeric protein complexes are often sensitive to coexpression of a defective form of one of the components. An inactive component (encoded by a dominant, negative mutation) can poison the activity of the remaining components, rendering the complex inactive. To test if any toxR* or toxS* allele was dominant and negative to ToxR function, each pTSK derivative carrying a variant toxRS operon was transformed into JDP169/pVM7 [(ctx-lacZ)7 toxR⁺]. Plasmid pVM7 expresses toxR constitutively from the strong tet promoter. The -galactosidase activities of these strains revealed that most of the alleles were not dominant to toxR⁺ function in this assay (Fig. 6). A notable exception to this was the toxS3 (L33S) mutation, which is strongly dominant with respect to ToxR transcriptional activation.

View larger version (25K): [in this window] [in a new window]   FIG. 6.   Effect of toxR or toxS positive control mutants on the transcription of (ctx-lacZ) fusion strains in the presence of a second plasmid source of toxR+. To test if any of the toxR and toxS mutations are dominant to the activity of toxR+ carried by pVM7, each pTSK mutant toxRS operon was introduced into JDP169/pVM7 [H(ctx-lacZ) toxR+]. -Galactosidase activity is shown. JDP169/pACYC184/pBR322 bearing DNA encoding no activators produces 13 Barrick U of -galactosidase activity (data not shown).

Correlation of membrane localization of ToxR and DNA binding activity. The toxR positive control allele encoding the shortest periplasmic C-terminal deletion that displays full DNA binding activity, toxR107 (N200fs), carries only 1 amino acid of the wild-type coding sequence downstream of the transmembrane domain of toxR and a 9-amino-acid frameshift peptide. It is readily detectable as a single band on an immunoblot assay (Fig. 5A), is localized to the membrane fraction (Fig. 7), and interacts with DNA tightly in the challenge phage assay (Fig. 4). ToxR101 (M98fs) and ToxR103 (S93fs) are cytoplasmic derivatives that interact with DNA poorly (Fig. 4) and are not detectable by immunoblot assay (Fig. 5A). To test if a stable cytoplasmic ToxR derivative can bind to DNA or activate transcription, we constructed ToxRTMHistag (25). The st11 His₆ sequence (H₆KNQHE), which is an engineered C-terminal extension previously shown to protect the P22 Arc repressor from intracellular proteolytic degradation, was used to replace the transmembrane domain of ToxR and to create a 22-kDa cytoplasmic derivative that is stabilized against proteolysis (ToxRTMHistag). In these experiments, toxR and toxRTMHistag were expressed from the strong tac promoter of plasmid pTacterm. Figure 8 shows that the level of protein expressed from Salmonella strain MS1868/F' lacI^q/pJAM3toxRTMHistag under IPTG induction is comparable to that of the ToxR-overproducing host MS1868/F' lacI^q/pJAM3 (toxR⁺). When these hosts were infected with the challenge phage P22 ctx8, the toxR⁺ host formed lysogens at a high frequency (Table 3) whereas a host expressing the toxRTMHistag mutation formed lysogens at a low frequency. Data from this experiment show that ToxR⁺ interacts with the ctx promoter but that ToxRTMHistag does not. Similarly, when toxR is overexpressed from pJAM, ToxR can activate transcription of the ctx-lacZ fusion in the absence of ToxS, whereas overexpression of ToxRTMHistag does not activate transcription (Table 3).

View larger version (93K): [in this window] [in a new window]   FIG. 7.   Subcellular localization of ToxR and periplasmic truncation derivative ToxR107. S. typhimurium extracts were fractionated and subjected to immunoblot analysis with an anti-ToxR antibody as described in Materials and Methods. ToxR, MS1868/pTSK (toxR+S+); ToxR107, MS1868/pTSK (toxR107). Lanes: T, total protein; S, soluble protein; M, membrane fraction; C, total protein fraction from the control (MS1868/pACYC184). (A) Immunoblot analysis; (B) Coomassie brilliant blue staining of gel.

View larger version (22K): [in this window] [in a new window]   FIG. 8.   Results of immunoblot assay showing the steady-state level of ToxR and ToxRTMHistag under overexpressing conditions. MS1868/F' lacIq/pTacterm, MS1868/F' lacIq/pJAM3 (toxR+), and MS1868/F' lacIq/pJAM3toxRHistag were grown and induced with IPTG as described in Materials and Methods for challenge phage assays. Whole-cell extracts were resolved on an SDS-polyacrylamide gel and transferred to nitrocellulose. Membranes were probed with an anti-ToxR antibody. Schematic representations of ToxR and ToxRTMHistag are described in the legend to Fig. 2. Lanes: 1, vector control; 2, toxR+; 3, toxRTMHistag.

View this table: [in this window] [in a new window]   TABLE 3.   Transcriptional activation and DNA binding by toxR+ and the toxRTMHistag mutation when toxR was overexpresseda

N-terminal sequencing of the cytoplasmic ToxRTMHistag product. There are two potential methionine initiator codons of toxR. The ToxRTMHistag product provided a source for N-terminal sequence analysis. The product was purified with an immobilized cobalt metal affinity column and subjected to protein microsequence analysis to determine the ToxR start site (see Materials and Methods). N-terminal microsequence analysis showed that the first amino acid is actually serine 11, suggesting that the true start site may be the methionine residue at position +10 with respect to the current numbering system. Consistent with this possibility, bacterial species related to V. cholerae, Vibrio parahaemolyticus, and Vibrio fischeri encode toxRS and sequence alignments show that these ToxR proteins start at +10 relative to the position of V. cholerae ToxR (38). Determination of the true toxR translational start site will require mutagenesis of the two potential ATG start codons.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The ToxR regulon of V. cholerae is one of the best-characterized systems that control bacterial virulence gene expression. However, the details of how the ToxRS complex activates transcription are not fully understood. An interaction between membrane-bound ToxR and ToxS is thought to occur in the periplasm (11). In this model, ToxR is the DNA binding component and ToxS enhances the ability of ToxR to bind to DNA and activate transcription. In addition to the proposed interactions between ToxR and ToxS, several studies suggest that ToxR is a dimer in the activation complex (19, 21, 31-33).

To gain further insight into activation by ToxRS, we have isolated toxR and toxS mutations in the context of a plasmid-borne toxRS operon that impair the ability of the ToxR protein to activate transcription but leave ToxR-DNA binding unaltered (Fig. 2). The toxR positive control mutations reveal that the periplasmic domain of ToxR has a central role in transcriptional activation but that it is dispensable for DNA binding. Any deletion within the ToxR C-terminal periplasmic domain was defective for activation but not DNA binding, as measured in vivo by the challenge phage system. The in vivo steady-state level of ToxR molecules carrying C-terminal deletions was altered. For example, ToxR109, which is missing the last 14 C-terminal amino acids, is proteolytically processed in vivo, indicating that an intact C terminus of ToxR is important for protection against proteolysis (Fig. 5). Presumably, the truncated forms of ToxR are unfolded and subject to proteolysis by periplasmic proteases (18). Challenge phage (Fig. 4) and immunoblot (Fig. 5) assays suggest that the smaller fragments still retain DNA binding activity. Presumably, the population of processed molecules bind to DNA strongly in the challenge phage assay but they cannot be assembled into the activation complex.

In addition to toxR mutations that encode C-terminal truncations, null mutations in toxS (i.e., toxS) have a defect in ToxR-dependent activation. Challenge phage assays indicate that ToxS is not required for strong DNA binding. Although the toxS mutation impairs transcriptional activation, the steady-state level of ToxR protein expressed from pTSK is similar to the level of the toxR⁺ toxS⁺ background, indicating that ToxS does not modulate the steady-state level or stability of ToxR (Fig. 5). The toxS mutation, or any other toxS null mutation, can be considered to be a toxS positive control mutation because ToxS is required for transcriptional activation but not ToxR-DNA binding. Another example of a toxS positive control allele is toxS1, which encodes a C-terminal truncation and an intact N-terminal transmembrane domain. This mutation shows that an intact ToxS periplasmic domain is required for transcriptional activation by ToxR. Thus, the toxS1 allele has a phenotype similar to that of the toxS mutation and ToxR is maintained at a steady-state level similar to that produced by the toxS⁺ strain. However, the toxS3 allele encoding the L33S change located within the ToxS periplasmic domain appears to destabilize the structure of ToxR, as determined by proteolysis in vivo, providing further genetic evidence that ToxR and ToxS directly interact in the periplasm (Fig. 5B). The proteolyzed form of ToxR generated in the presence of toxS3 (L33S) still retains full DNA binding activity and behaves similarly to the toxR mutations that truncate the periplasmic domain. Therefore, mutations in the periplasmic domain of both ToxR and ToxS destroy ToxR-mediated activation but DNA binding is not altered. The ToxS3 (L33S) protein appears to unfold the periplasmic domain of ToxR, rendering it susceptible to periplasmic proteases. Interestingly, the toxS3 (L33S) allele is epistatic to activation by two different plasmid sources of toxR⁺ (Fig. 6). Plasmid pVM7 (toxR⁺) expresses ToxR at a high-enough level to activate ctx transcription in the absence of ToxS (see lanes pACYC184 and toxS). This activation of ctx expression by overexpression of toxR from pVM7 can be decreased by coexpression of the toxS3 mutation from pTSK. Thus, it appears as though ToxS3 (L33S) has an activity that can alter the conformation of the periplasmic domain of ToxR. This finding implies that ToxS⁺ may possess a chaperone-like activity (43) that modulates the conformation of the ToxR periplasmic domain so that it can be assembled into the activation complex.

One toxR missense mutation that disrupts ToxRS activation of gene expression, but does not affect DNA binding, was identified in a region corresponding to the ToxR cytoplasmic DNA binding domain. This mutation, toxR105 (E82K), behaves like a classic positive control mutation (15) and can be predicted to disrupt the interaction between RNA polymerase and the ToxR-DNA binding domain. Unlike the toxR mutations that alter the periplasmic domain, the toxR105 product is not processed in the periplasm (Fig. 5B), suggesting that the periplasmic domain is normally folded and can interact with ToxS but that this interaction is not sufficient to activate transcription. The DNA binding motif of ToxR is homologous to that of OmpR of E. coli. OmpR is a member of the response regulator family of bacterial two-component regulatory systems, binds to DNA, and regulates transcription in response to phosphorylation by the sensor kinase EnvZ (37). Although ToxR is a transmembrane protein and does not appear to be phosphorylated like other response regulators, the DNA binding domain probably adopts a fold that is similar to that of OmpR (31). Two groups have recently published X-ray crystal structures of the DNA binding domain of OmpR (20, 24). The structure of the DNA binding domain of OmpR is similar to those of helix-turn-helix DNA binding proteins but is actually a helix-loop-helix protein (20, 24, 30). The E82K mutation carried by toxR105 is found within the predicted loop region of the ToxR-DNA binding motif. Similarly, three ompR mutations that abolish transcriptional activation (E193K, A196V, and E198K) are found within the loop region (37). The loop region appears to be involved in contacts between OmpR and the  subunit of RNA polymerase (30). These parallels suggest that ToxR may bind to DNA and activate transcription in a manner that is similar to that of phosphorylated OmpR (17).

An additional aspect of ToxR function that we were able to assess using the mutants isolated in this study is the question of whether membrane localization is required for either DNA binding or activation of transcription. The results from previous work in this area have been inconsistent. Using hybrid ToxR proteins capable of dimerizing, Ottemann and Mekalanos have demonstrated a membrane requirement for activation of transcription from the ctx promoter (32), whereas Kolmar et al. have suggested that the cytoplasmic form of a very similar hybrid molecule can activate transcription (19). In the present study, we found that cytoplasmically localized ToxR proteins expressed by deletion mutants carrying lesions removing a portion of the gene encoding both the transmembrane and periplasmic domains of ToxR (toxR101 and toxR103) fail to interact with DNA strongly (Fig. 4). ToxR101 and ToxR103 proteins cannot be detected by immunoblot assay (Fig. 5). ToxRTMHistag was constructed to create a cytoplasmic variant that is stabilized against proteolysis by the st11 sequence (25) (Fig. 8). ToxRTMHistag is stable and yet fails to bind to DNA or activate transcription (Table 3). ToxR107 protein, which is missing almost the entire periplasmic domain, binds to DNA strongly but fails to activate transcription. Because ToxR107 contains an intact membrane-spanning domain, it is membrane associated (Fig. 7). We interpret this to mean that ToxR must be membrane bound in order to interact with ctx promoter DNA and activate transcription.

The picture where the ToxR periplasmic domain can spontaneously fold in the absence of ToxS but where it cannot activate transcription emerges. This form of ToxR interacts with DNA in the challenge phage assay, but only after the addition of ToxS can ToxR be assembled into the activation complex. This complex may be an oligomeric assembly of ToxR within the membrane, and we propose that ToxR assembles on the DNA in a manner similar to that of OmpR (17).

In our model for ToxRS-mediated transcriptional activation, ToxR molecules are located within the cytoplasmic membrane and possibly at the pole opposite the flagella of a V. cholerae cell (19). Newly synthesized ToxR molecules are exported to the cytoplasmic membrane. In the periplasm, the periplasmic domains are assembled and properly folded into higher-order complexes by membrane-bound ToxS. Once these complexes are assembled by ToxS, then transcriptional activation can occur. The toxR alleles encoding truncations that map between the C terminus and the transmembrane domain cannot form these higher-order structures because they are degraded or missing periplasmic sequences that are required for the interaction with ToxS. ToxR protein is not degraded in a toxR⁺ toxS background, but it is not assembled or folded into the proper activation complex. Presumably, overexpression of ToxR in the absence of ToxS can assemble ToxR into the activation complex since toxS is not required for activation when toxR is highly expressed from multicopy plasmids.

Our model for ToxRS-mediated activation of transcription varies little from what has been proposed for OmpR. Mutations in the cholera toxin promoter that impair ToxRS-mediated binding map to a large region (approximately 80 to 40), suggesting that ToxR binds in a cooperative manner (35). Footprints of OmpR on the ompC promoter also protect a large region spanning approximately positions 100 to 40 (22). We propose that ToxS protein assembles ToxR dimers (or monomers) into higher-order complexes. These higher-order complexes, as with OmpR (17), NtrC (44), and other response regulators, activate transcription. The multimerization of OmpR is driven by EnvZ-mediated phosphorylation of OmpR. We propose that multimerization of ToxR is driven by ToxS and that this multimerization occurs within the inner membrane.