|Home | About | Journals | Submit | Contact Us | Français|
The filamentous bacteriophage CTXΦ transmits the cholera toxin genes by infecting and lysogenizing its host, Vibrio cholerae. CTXΦ genes required for virion production initiate transcription from the strong PA promoter, which is dually repressed in lysogens by the phage-encoded repressor RstR and the host-encoded SOS repressor LexA. Here we identify the neighboring divergent rstR promoter, PR, and show that RstR both positively and negatively autoregulates its own expression from this promoter. LexA is absolutely required for RstR-mediated activation of PR transcription. RstR autoactivation occurs when RstR is bound to an operator site centered 60 bp upstream of the start of transcription, and the coactivator LexA is bound to a 16-bp SOS box centered at position −23.5, within the PR spacer region. Our results indicate that LexA, when bound to its single site in the CTXΦ prophage, both represses transcription from PA and coactivates transcription from the divergent PR. We propose that LexA coordinates PA and PR prophage transcription in a gene regulatory circuit. This circuit is predicted to display transient switch behavior upon induction of CTXΦ lysogens.
The horizontal transfer of CTXΦ, the lysogenic filamentous phage that carries the genes encoding cholera toxin, contributes to the emergence of new toxigenic strains of Vibrio cholerae, the cause of epidemic cholera (6, 31). For stable CTXΦ transfer to occur, the infecting single-stranded DNA CTXΦ genome must integrate site-specifically into the V. cholerae genome and establish a lysogenic program that involves the transcriptional repression of CTXΦ genes involved in bacteriophage replication and morphogenesis (17). The CTXΦ prophage expresses a 14-kDa repressor, RstR, that represses the expression of CTXΦ structural genes and provides immunity to secondary CTXΦ infection (12, 32). The CTXΦ prophage can be induced with DNA-damaging agents such as mitomycin C and UV radiation, a process that is mediated by a second repressor, the host SOS repressor LexA (25). Prophage induction does not kill the host bacterium and results in only a modest increase in CTXΦ titers, unlike the massive phage growth and host lysis that occurs upon induction of phage lambda.
The CTXΦ replication and morphogenesis genes initiate transcription from the strong PA promoter located in ig-2, an intergenic region separating the rstA replication gene from the divergently transcribed rstR gene encoding the CTXΦ repressor (12) (Fig. (Fig.1B).1B). RstR strongly represses PA transcription in CTXΦ lysogens by binding to a set of three operator sites that surround PA (13). The host SOS repressor LexA also represses PA by binding to a single SOS box centered −48.5 bp from the start of PA transcription (25) (Fig. (Fig.1B).1B). LexA, together with RecA, regulates a network of more than 30 host genes that monitors and responds to DNA damage (7). Upon DNA damage or replication fork arrest, RecA, together with single-stranded DNA and ATP, stimulates the autoproteolysis of LexA (14), leading to the transient derepression of numerous genes encoding DNA repair functions. We previously showed that LexA repression of PA occurs when DNA-bound LexA occludes a promoter UP element that overlaps the LexA binding site, thereby preventing the C-terminal domain of the α subunit of RNA polymerase (RNAP) from gaining access to this UP element (24).
How CTXΦ lysogens regulate RstR levels is not understood but seems to be critical to the functioning of the regulatory circuit governing CTXΦ lysogeny. For example, the O2 operator and the single SOS box in CTXΦ overlap, such that RstR and LexA cannot both bind to their respective sites in O2 (25). However, our finding that CTXΦ lysogens can be induced by DNA damage indicates that the O2 region is predominantly occupied by LexA. One hypothesis that could account for these observations is that the lysogenic levels of RstR are tightly regulated, such that the concentration of RstR is maintained at a level high enough to bind O1, thereby repressing CTXΦ prophage transcription, but not sufficiently high to allow binding to O2 or O3. To investigate this hypothesis, we have identified the rstR promoter PR and determined that RstR both positively and negatively autoregulates its own expression from PR. Interestingly, RstR autoactivation requires LexA binding to the SOS box that overlaps with the −35 sequence of PR. Thus, RstR and LexA, which were known to function as repressors of CTXΦ transcription from PA, also function as coactivators of rstR transcription from PR. These results delineate a CTXΦ regulatory circuit with similarities to the classic phage lambda switch but also with unique features that are better suited to a temperate filamentous bacteriophage.
Strain HK386 is a ΔlacZ derivative of the environmental V. cholerae strain 2740-80 (attRS+ CTX−) (19). The lysogen strain HK138 was constructed by transformation of HK386 with CTXKn plasmid DNA (31). Escherichia coli BW25113 [Δ(araD-araB)567 ΔlacZ4787(::rrnB-3) lacIp-4000(lacIq) λ− rph-1 Δ(rhaD-rhaB)568 hsdR514] (4) and BW27784 (BW25113 ΔaraFGH ΔaraEp::PCP18-araE) (11) were obtained from the E. coli stock center. Strain TP608 (AB1157 Δ(recC-ptr-recB-recD)::Ptac-gam-red-pae-cI822 sulA6209::tet lexA71::Tn5) was kindly provided by A. R. Poteete (20). sulA6209::tet, followed by lexA71::Tn5, was transduced into strain BW27784 by serial P1 transduction using a P1vir lysate grown with strain TP608, as previously described (18).
Oligonucleotide primers were high-pressure liquid chromatography purified by the manufacturer (Operon Technologies) and end labeled with P32-labeled γ-ATP (10 Ci/mmol) and T4 polynucleotide kinase. Total RNA was isolated from V. cholerae and E. coli using Trizol reagent, according to the manufacturer's instructions (Invitrogen). Ten micrograms of total RNA was annealed to 0.8 pmol of labeled primer and extended using SuperScript III, according to the manufacturer's directions for first-strand cDNA synthesis (Invitrogen). Reactions were terminated by addition of 0.1 M EDTA and extracted with phenol-chloroform (dilution of 24:1). Nucleic acids were ethanol precipitated, resuspended in denaturing sample buffer (99% formamide, 0.1% 1N NaOH, and 0.01% each of bromophenol blue and xylene cyanol FF), and heated briefly to 92°C. The reaction products, together with a set of dideoxy DNA sequencing reactions generated with the same radiolabeled primer, were separated on 8% DNA sequencing gels.
pBAD-RstR has been described previously (12). The lacZ transcription reporter plasmids pCB182 and pCB192 (27) were modified by cloning a 300-bp fragment carrying the rrnBT1 transcription terminator (30) into the unique SmaI site, yielding pCB182N and pCB192N. This modification prevented unregulated PA transcription from disrupting plasmid replication. pHK301 contains CTXΦ DNA from positions +5 to −115 relative to the start of rstR transcription, while pHK312 contains sequences from +96 to −160, both cloned into pCRII (Invitrogen) (13). The CTXΦ fragments were excised by digestion with SalI and HindIII and ligated to SalI- and HindIII-digested pCB192N DNA, yielding the PR-lacZ reporters pHK301N and pHK312N. The PA-lacZ reporter pHK393 was constructed by cloning the XbaI-HindIII fragment from pHK301 into reporter pCB182N. The O1-18 mutant derivatives were generated by PCR-mediated mutagenesis using a QuikChange site-directed mutagenesis kit (Stratagene). Similarly, a derivative of pHK301N carrying the m1 substitution, pHK883, was generated by site-directed mutagenesis. The mutations were confirmed by DNA sequencing. The m1 SOS box mutation was originally carried on the plasmid pMQm1. For in vitro transcription, the CTXΦ region from +5 to −115 (relative to the start of rstR transcription) carrying the m1 SOS box allele was amplified from pMQm1 (M. Quinones, unpublished data) with primers containing XhoI and HindIII restriction sites at their 5′ ends. The resulting DNA fragment was cloned into pCRII (Invitrogen), yielding pHK725.
LacZ assays were performed as described previously (18). Fresh colonies grown on LB agar plus antibiotics were used to inoculate 2 ml LB broth plus antibiotics and l-arabinose as an inducer. Cultures were incubated on a roller wheel at 37°C for 12 to 14 h and diluted 1:10 in phosphate-buffered saline prior to being assayed.
RstR activity was measured in cell extracts using a gel-shift assay similar to that used to measure CI concentrations in E. coli (5). A 250-ml sample of cultures of BW27784 carrying pBAD-RstR was grown in LB plus antibiotics and the indicated concentration of l-arabinose to late log phase (optical density at 600 nm of ~0.8). Cells were harvested, washed, concentrated to a volume of 2 to 3 ml in GS buffer (20 mM Tris-HCl [pH 7.5], 150 mM KCl, 10% glycerol), and lysed in a prechilled French pressure cell. Cell debris was removed by centrifugation, and small aliquots were stored frozen at −70°C. Samples were thawed on ice and serial diluted with a similar cell extract made from cultures of BW27784 lacking pBAD-RstR plasmid. A 120-bp DNA probe containing the high-affinity O1 operator site was generated by PCR with one primer fluorescently labeled with 5-FAM (Operon Technologies) and purified using a QIAquick PCR purification kit (Qiagen). Binding reactions were carried out in GS buffer containing 10 mM dithiothreitol and 10 μg/ml sheared salmon sperm DNA as the nonspecific binding competitor. Samples were incubated at 4°C for 30 min with gentle mixing. One microliter of dye solution (0.1% bromophenol blue) was added, and samples were loaded and run on 6% polyacrylamide and 0.5× Tris-borate-EDTA DNA retardation gels (Invitrogen). Gels were scanned using the blue laser of a Typhoon scanner (GE Biosciences), and the band intensities corresponding to the bound and unbound DNA fractions were determined using ImageQuant software. The volume of extract resulting in 50% binding of probe DNA was interpolated from the binding curves, and RstR binding activity (RU) was calculated as [(1/V50)/total protein concentration (μg/ml)] × 100, where V50 is the volume of cell extract (μl) resulting in 50% shifting of probe DNA. Protein concentrations were determined with the Coomassie Plus kit (Pierce).
Anti-His6-tagged RstR (RstR6H) sera were prepared in BALB/c mice as previously described (8) and used in Western blots at a dilution of 1:5,000. Western blots were developed with anti-mouse antibodies conjugated to horse radish peroxidase (Pierce Biotechnology).
RstR6H was purified as previously described (13). E. coli LexA was expressed from pJWL228 in E. coli T7 Express lysY/lacIq (New England Biolabs) and purified as previously described (16). The chromatography media were substituted in order to make use of a programmable fast-performance liquid chromatography system. Fraction II material (3 ml, from 500 ml of starting culture) was diluted to 0.1 M NaCl and loaded onto a HiTrap SP fast-flow column (1 ml) connected to an ÄKTApurifier fast-performance liquid chromatography system (GE Biosciences). Buffer conditions used were identical to those described for the original phosphocellulose chromatography step (16). LexA was eluted in a 20-ml gradient from 0.1 M to 1.0 M NaCl. LexA-containing fractions (2.5 ml) were pooled and dialyzed versus 1 liter buffer Q (0.02 M Tris-HCl [pH 8.0], 0.1 mM Na2+-EDTA, 10% glycerol) containing 0.2 M NaCl. Dialyzed material was diluted to 0.08 M NaCl with buffer Q and loaded onto a MonoQ 5/50 column (1 ml) preequilibrated in buffer Q plus 0.08 M NaCl. LexA was eluted using a 40-ml gradient from 0.08 M to 0.6 M NaCl. LexA eluted as a single peak early in the gradient (data not shown). LexA peak fractions were pooled, and small aliquots were stored at −70°C. This material was estimated to be 95% pure, as judged by Coomassie blue-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis. LexA concentrations were determined by UV absorption at 280 nm, using a molar extinction coefficient of 6,900 M−1 cm−1.
Single-round runoff transcription reactions were carried out using heparin sulfate, as previously described (33). Template DNAs were generated by PCR from pHK301 or pHK725 (m1) plasmid DNA using pairs of primers that anneal to flanking vector sequences. Primer sequences are available upon request. PCR products were purified using a QIAquick PCR purification kit and quantitated by UV absorption at 260 nm. Template DNA and E. coli RNAP (σ70) holoenzyme (Epicentre) were present at 2 nM and 20 nM, respectively. Five units of RNase inhibitor (Agilent Technologies) was added to all reactions. Template DNA and repressors were preincubated at 37°C for 15 min. RNAP was added, and the incubation continued for 15 min. A mix containing 2 μCi of P32-labeled α-UTP (10 mCi/mmol; Perkin Elmer), nucleoside triphosphates (10 μM each of CTP, GTP, and ATP, and 1 μM UT, final concentration) and heparin sulfate (100 μg/ml, final concentration) was added, and the incubation was continued for a further 30 min. A set of 32P-labeled RNA standards were generated using T7 RNAP, according to the manufacturer's instructions (Agilent Technologies). All reactions were terminated by ethanol precipitation. Pellets were resuspended in denaturing sample buffer and briefly heated to 95°C, and RNA products were fractionated on 6% denaturing polyacrylamide gels.
Primer extension analysis located the 5′ end of rstR mRNA at a position 23 bp upstream of the RstR start codon (Fig. (Fig.1A).1A). Similar results were obtained by S1 nuclease mapping (data not shown). The putative rstR promoter PR contains the −10 sequence TACAAT, which closely matches the consensus σ70 −10 sequence. However, PR lacks a recognizable DNA sequence that closely matches the σ70 consensus −35 sequence (28) (Fig. (Fig.1B).1B). The single LexA binding site in CTXΦ (25) is centered −23.5 bp from the start of rstR transcription, within the spacer region of PR and overlapping the presumed −35 sequence (Fig. (Fig.1B).1B). The PR promoter is closely opposed to the divergent PA promoter, which directs the transcription of CTXΦ genes required for replication and viral assembly. A significant fraction of rstR primer extension products contained 5′ extensions as long as 10 residues in length (Fig. (Fig.1A).1A). Similar 5′ extensions have been observed at promoters whose transcription, like that of rstR, initiates at a string of three A residues (34). During transcription initiation, repeated slippage by RNAP at the string of three T residues on the DNA template strand leads to the incorporation of 5′ A tracts in the nascent mRNA. A similar slippage process could produce the 5′ extensions we observe in rstR mRNA.
To investigate factors that regulate rstR transcription, we first analyzed LacZ expression from two plasmid-borne PR-lacZ fusions in V. cholerae. pHK301N contains the O1 and O2 operators but lacks O3, while pHK312N includes all three RstR operators (Fig. (Fig.1B).1B). The activity of these reporters varied markedly, depending on whether the host strain carried the integrated CTXΦ prophage. Both pHK301N and pHK312N expressed high LacZ activity in HK138, which carries tandem integrated copies of CTXΦ (1,000 and 1,250 Miller units, respectively), but markedly lower activities (50 and 200 Miller units, respectively) in HK386, an isogenic strain lacking CTXΦ. Since CTXΦ lysogens are known to express RstR (12, 31), which represses the expression of most other CTXΦ genes, we hypothesized that RstR activates its own transcription from PR. The O1 operator site, located at position −40 to −80 from the start of rstR transcription (Fig. (Fig.1B),1B), was the likely site of RstR-mediated activation of PR. We therefore constructed derivatives of pHK301N and pHK312N that contain a mutated O1 site (O1-18), such that RstR no longer bound to O1. As shown in Fig. Fig.2,2, the O1-18 mutation is a deletion/substitution mutation that lies near the center of O1 and results in a complete loss of RstR binding to O1 (Fig. (Fig.2B).2B). LacZ activities were determined for the mutant reporter plasmids in the same lysogen/nonlysogen strain pair. In both cases, LacZ activities determined in the CTXΦ lysogen strain were reduced to the basal levels observed in the nonlysogen strain (data not shown). These data are consistent with the hypothesis that RstR binding to the upstream O1 site is required for PR transcription.
Using the same set of reporter plasmids, we investigated whether PR autoactivation could be detected in E. coli by expressing RstR from the arabinose-inducible plasmid pBAD-RstR (12). With pHK301N (O1+ O2+), a strong peak of LacZ activity was observed at inducer concentrations of approximately 2 × 10−5 percent arabinose (Fig. (Fig.3A).3A). The peak of PR activation ranged between 15- and 20-fold over basal expression levels (no arabinose). At higher inducer concentrations, a strong decrease in lacZ expression was observed. This RstR-mediated repression cannot be explained by nonspecific binding of RstR to the PR promoter region, since a lacZ reporter to a control promoter that is not regulated by RstR was not detectably repressed by RstR in analogous experiments (data not shown). Primer extension analysis of RNA extracted from the same arabinose-induced cultures showed that lacZ transcription in E. coli originated from the true PR promoter (data not shown). LacZ expression from the mutated reporter pHK301N(O1-18) could not be activated by RstR expressed from pBAD-RstR (Fig. (Fig.3A),3A), indicating that PR transcription requires RstR binding to O1 positioned −40 to −80 from the start of rstR transcription.
We were able to relate the changes in PR activity to changes in the cellular concentration of RstR by directly measuring RstR activity in extracts of arabinose-induced cultures of BW27784 (pBAD-RstR) using a gel-shift assay and by Western blotting (Fig. (Fig.3B).3B). The peak of PR transcription was observed at approximately 2 × 10−5 percent arabinose, while strong repression was observed at approximately 100-fold-higher inducer concentrations (Fig. (Fig.3A).3A). Over that range, RstR activity increased approximately 14-fold (10.7 to 142 RU). In a parallel DNase I footprinting experiment using a DNA probe carrying O1 and O2, we estimated the relative concentrations of RstR required to bind O1 rather than to bind both O1 and O2. As shown in Fig. Fig.3C,3C, an approximately eightfold-higher concentration of RstR was required to observe both O1 and O2 occupancy rather than occupancy of O1 alone. Taken together, these data support the idea that RstR, when present at low cellular concentrations, binds to the high-affinity O1 site upstream of PR and activates PR transcription, while PR repression occurs when high concentrations of RstR lead to additional RstR binding to the lower-affinity O2 site.
When we analyzed LacZ expression in E. coli from pHK312N, a reporter containing all three RstR operators, we observed a 2.5-fold increase in LacZ activity over basal levels, followed again by a decrease in LacZ activity at high inducer concentrations (Fig. (Fig.3A).3A). Our previous analysis of RstR-operator DNA interactions indicated that RstR binds cooperatively to O2 and O3 (13). Cooperativity was supported by DNase I footprint studies showing that the difference in the apparent affinity of RstR for O1 and O2 is significantly narrowed by the presence of O3 on the same DNA fragment (22). A likely explanation for the poor activation observed with pHK312N is that pBAD-directed expression of RstR is not sufficiently fine-tuned to allow for full activation of PR, which requires RstR binding to O1, without autorepression, which we propose involves RstR occupancy of O2 and O3. These data are consistent with the hypothesis that RstR autoactivation occurs when RstR is bound only to O1, and autorepression occurs when additional RstR binds to O2 and O3.
We next analyzed the role of LexA in rstR autoregulation. This global SOS regulator represses CTXΦ gene expression from PA by binding to a single SOS box centered −48.5 bp from the start of PA transcription (25) (Fig. (Fig.1B).1B). We used E. coli for these studies, as we have been unable to delete the lexA gene from V. cholerae (22). This approach is supported by previous studies showing that E. coli LexA fully substitutes for V. cholerae LexA in PA repression (25). To confirm that the test strain constructed for these experiments, BW27784 sulA6209 lexA71, lacks LexA activity, we introduced plasmid pMQLZ, which carries the known V. cholerae LexA-regulated promoter PlexA fused to the lacZ gene of E. coli (25). The resulting strain exhibited an expected sixfold increase in LacZ activity compared to that of an isogenic lexA+ strain. As shown in Fig. Fig.4A,4A, the RstR-mediated activation of PR transcription from pHK301N (O1+ O2+) was abolished by lexA71. lexA71 did not significantly alter the expression of RstR from pBAD-RstR, since a PA-lacZ reporter was efficiently repressed by RstR in the same mutant strain (Fig. (Fig.4B;4B; see below). These results suggest that RstR and LexA are both required to activate transcription from PR. To explore whether LexA mediates activation by directly binding to the single SOS box within PR, the left half-site of the LexA binding site in pHK301N was changed from CTGT to CTAT (nucleotide change underlined) (Fig. (Fig.1B1B and and5B).5B). This mutation, m1, was previously shown to abolish LexA binding to the SOS box in CTXΦ (25). This mutated PR:lacZ reporter plasmid (pHK883) could not be activated by RstR when assayed in BW27784 sulA6209 carrying a wild-type allele of lexA (data not shown), indicating that a direct interaction between LexA and the SOS box is required for RstR-mediated transcription from PR.
Using the same set of mutant E. coli strains, we also analyzed the roles of LexA and RstR in regulating the divergent PA promoter, the primary CTXΦ promoter. In the absence of RstR, LexA repressed PA transcription by approximately threefold (no arabinose) (Fig. (Fig.4B),4B), consistent with our previous studies of PA regulation (24). Regardless of the presence of LexA, RstR strongly repressed PA transcription (Fig. (Fig.4B),4B), confirming our earlier findings showing that LexA and RstR independently repress PA transcription.
Since removal of LexA from the cell could have indirect consequences on PR activity, we investigated the roles of RstR and LexA in CTXΦ gene transcription using in vitro transcription assays. Purified RstR (RstR6H) and/or E. coli LexA was incubated with DNA templates containing O1 and O2 but lacking O3. E. coli RNAP(σ70) was added, and transcription was allowed to proceed for one round using heparin sulfate. As shown in Fig. Fig.5A,5A, a transcript corresponding to the predicted 211-nucleotide (nt) PR transcript was observed only in the presence of both RstR6H and LexA. PR transcription increased with increasing RstR concentrations, then decreased at the highest RstR levels tested, similar to the profile of PR activity observed in PR-lacZ reporter assays (Fig. (Fig.3A).3A). At its peak, PR activation resulted in a 21-fold increase in transcript levels. A transcript corresponding to the predicted 94-nt PA transcript showed the expected pattern of dual repression by LexA and RstR (Fig. (Fig.5A).5A). To confirm that these transcripts originated from the PA and PR promoters, a second overlapping DNA template was generated, in which the predicted sizes of the PA and PR runoff transcripts were 175 nt and 115 nt, respectively. Using this DNA template, a transcript corresponding in size to the predicted PR transcript was again observed only in the presence of LexA and RstR, and a 175-nt RNA product corresponding to the PA transcript showed the expected pattern of dual repression by LexA and RstR (data not shown). These results demonstrate that RstR and LexA stimulate transcription from PR in a codependent manner while also repressing transcription from PA. Our results also confirm that RstR represses PR activity at high repressor concentrations.
We investigated the requirement for LexA binding to the SOS box by analyzing transcription from a DNA template containing the G-to-A substitution mutation, m1, in the left half-site of the LexA binding site within PR. m1 was previously shown to abolish LexA binding to this site without affecting RstR binding to the O2 region (25). Using m1 DNA (O1+ O2+) as a transcription template, we did not observe stimulation of the predicted 145-nt PR transcript in reactions containing RstR and LexA (Fig. (Fig.5B),5B), indicating that LexA binding to the known SOS box within PR is required for activation. As expected, PA transcription from the m1 template could still be repressed by the addition of RstR6H (Fig. (Fig.5B)5B) but was no longer repressed by the addition of LexA, confirming that the presence of m1 results in a nonfunctional SOS box. These data demonstrate that PR activation requires LexA binding to the single SOS box within PR and confirm our in vivo PR-lacZ studies, showing that LexA, as well as LexA binding to the SOS box, is required for PR activation.
We previously demonstrated a direct role for both RstR and LexA in repressing CTXΦ transcription from the PA promoter (24, 25). Here we show that RstR and LexA perform the additional role of coactivating rstR transcription from the divergent PR promoter. LexA has a relatively minor role in repressing transcription from PA, whereas LexA is essential for rstR expression from PR. Therefore, LexA and RstR function as both repressors and activators of CTXΦ gene expression. Additionally, we found that high concentrations of RstR repress PR activity, indicating that CTXΦ lysogens use both positive and negative autoregulatory mechanisms to tightly regulate repressor levels. Our findings suggest that maintenance of CTXΦ lysogeny uses a regulatory circuit requiring the interplay of host- and phage-encoded transcription factors.
RstR autoactivates its own expression from the PR promoter when bound to O1 and autorepresses its own expression when bound to O2 and O3. Several lines of evidence support this hypothesis. PR-lacZ fusions expressed high levels of LacZ in CTXΦ lysogens, and a mutation (O1-18) that abolished RstR binding to the upstream O1 operator eliminated the 20-fold increase in LacZ activity observed in CTXΦ lysogens. Also, a shortened PR-lacZ fusion (O1+ O2+) was strongly activated in E. coli when RstR was provided from the arabinose-inducible plasmid pBAD-RstR (Fig. (Fig.3A).3A). Since an O1+ O2+ reporter could be easily activated using a heterologous system, we exploited this feature to define the roles of RstR and LexA in PR transcription regulation. By varying the RstR levels using the PBAD promoter system (Fig. (Fig.3B),3B), we showed that RstR activated PR transcription at low concentrations and strongly repressed PR transcription at high concentrations. LexA was absolutely required for PR autoactivation in vivo (Fig. (Fig.4A).4A). In vitro transcription studies carried out with purified components confirmed that RstR and LexA function directly as coactivators of rstR transcription and also confirmed that RstR negatively autoregulates PR transcription (Fig. (Fig.55).
Using a heterologous transcription reporter system in E. coli, we observed only strong PR activation, with constructs lacking the O3 operator positioned downstream of PR (Fig. (Fig.1B).1B). The presence of O3 improves RstR binding to O2, probably due to cooperative interactions between neighboring DNA-bound RstR tetramers (13, 22). We propose that the weak activation of a PR-lacZ fusion carrying all three operators (Fig. (Fig.3A)3A) is due to the difficulty in modulating RstR concentrations that allow for O1 occupancy but not that of O2 and O3. However, the same full-length lacZ fusion expressed high LacZ activity in V. cholerae CTXΦ lysogen strains, where RstR levels are presumably maintained by the positive and negative autoregulatory mechanisms outlined above. Alternatively, there could be physiological differences between V. cholerae and E. coli that limit PR transcription through an unknown mechanism.
Although there have been many studies describing LexA-mediated transcription repression, there is only one previous report of LexA functioning directly as a transcription activator (29). Together with this previous report, our discovery that LexA functions as a coactivator of RstR transcription suggests that the host SOS regulon may contain a subset of genes whose transcription is also activated by LexA. In support of this hypothesis, genome-wide studies of gene expression following DNA damage identified a number of genes in E. coli whose mRNA levels rapidly decrease upon UV exposure (3).
LexA's role as a coactivator of PR transcription is unusual, as it binds to a site centered −23.5 bp from the start of rstR transcription, within the promoter spacer region and overlapping the −35 sequence (Fig. (Fig.1B).1B). In contrast, most activators bind upstream of the −35 promoter region, where they make contacts with the C-terminal domain of the α subunit, the α-subunit N-terminal domain, or the σ region 4 of RNAP and/or other transcription activators (1). Nevertheless, a small number of transcription activators bind to sites within or overlapping the target promoter spacer region. These include the phage T4-encoded transcription activator MotA (9) and the lambda CII protein (10), which make direct contacts with RNAP, whereas activators such as the E. coli MerR protein act by distorting the DNA to bring the −10 and −35 promoter elements into alignment (2). Future studies will determine whether RstR- and LexA-mediated activation of rstR transcription involves direct interactions with RNAP.
Although we favor a direct model to explain RstR and LexA activation at PR, alternative models could be envisioned in which LexA and RstR stimulate PR transcription indirectly. One such model is that LexA and RstR act by preventing promoter interference between PA and PR. In this model, RNAP, when engaged in transcription at the strong PA promoter, directly represses PR transcription due to the close proximity of the two promoters. LexA and RstR would thereby activate PR transcription indirectly by interfering with PA transcription. To investigate the possibility of promoter interference between PA and PR, we altered the −10 sequence of PA from TATTTT to TATAAC (change underlined) and analyzed the effect of this change on PA and PR activity levels. The mutation reduced PA activity by approximately 50% in the absence of LexA and RstR (2,500 to 1,200 Miller units). However, this alteration to PA did not result in an increase in basal transcription from that of PR (45 to 50 Miller units), indicating that there is no strong promoter interference in this system. Also, if promoter interference was responsible for repressing PR, one would predict that RstR bound to O1, which strongly represses PA transcription (Fig. (Fig.4B4B and and5A),5A), would suffice to activate PR transcription. However, our in vivo and in vitro results demonstrate clearly that RstR is not sufficient to activate PR. Taken together, these observations suggest that PA and PR do not intrinsically interfere with one another.
The gene network governing CTXΦ lysogeny shares certain features with the lysogenic switch of λ (21). The promoters are divergently arranged and overlap with multiple repressor operator sites. During lysogeny, the primary phage promoters are repressed by a phage-encoded repressor, and the repressors both activate and repress their own expression. However, in contrast to lambda induction, which trips a highly stable switch that commits lysogens to phage development and host lysis, our findings suggest that CTXΦ induction follows transient and reversible switch kinetics. When CTXΦ lysogens are exposed to DNA damage-inducing agents, LexA levels would fall. The loss of LexA, which leads to the partial derepression of CTXΦ transcription from PA, would also result in a rapid cessation of RstR expression from PR, as LexA is absolutely required for PR transcription activation. CTXΦ induction should cease and rstR expression should resume, once the host DNA damage is repaired and LexA recovers to normal levels. The recovery kinetics of the CTXΦ switch might vary, as the recovery of LexA levels would likely depend upon the extent of DNA damage and the rate of DNA repair (15, 26). The recruitment by CTXΦ of the host LexA repressor appears to have resulted in a transient switch circuit, one that may be particularly suited to a filamentous bacteriophage that reproduces without killing its host.
The negative autoregulation of RstR levels is a critical feature of the proposed CTXΦ switch. We previously showed that RstR and LexA compete for binding to their respective overlapping sites at O2 (25). Therefore, RstR concentrations must be maintained at levels that allow for O1 occupancy but leave O2 vacant and accessible to LexA. In cells experiencing high levels of RstR, O2 and O3 would be occupied by RstR, displacing LexA, and SOS treatments would not induce CTXΦ gene expression. Negative autoregulation ensures that RstR levels do not drift upwards unchecked. Negative autoregulation is also critically important to the proper functioning of the λ switch, as overproduction of CI due to mutations that disrupt negative autoregulation results in lysogens that cannot efficiently switch to lytic development (5).
We thank B. Davis and A. L. Sonenshein for valuable discussions and A. Poteete and J. W. Little for kindly providing strains and plasmids. We are especially indebted to A. Hochschild and lab members for advice and discussions.
This work was supported by the Howard Hughes Medical Institute and NIH grant AI-42347.
Published ahead of print on 7 August 2009.