|Home | About | Journals | Submit | Contact Us | Français|
Previous work has shown that irvR is required for the proper regulation of genetic competence and dextran-dependent aggregation due to its ability to repress the transcription regulator irvA. In this study, we determined the mechanism used to relieve the repression of irvA. We demonstrate that IrvR is a “LexA-like” protein with four conserved amino acid residues likely required for IrvR autocleavage activity. Furthermore, recombinant IrvR protein purified from Escherichia coli was competent to undergo autocleavage in vitro. Using several truncated IrvR constructs, we show that the amino acids adjacent to the autocleavage site are essential for relieving irvA repression and engaging the irvA-dependent regulatory pathway primarily through the ClpXP and ClpCP proteases. By extending the IrvR C terminus with an epitope derived from the autocleavage site, we were also able to create a constitutive Clp-dependent degradation of the full-length IrvR protein. This suggests that the derepression of irvA occurs through a two-step mechanism involving the initial autocleavage of IrvR and exposure of a proteolytic degradation sequence followed by Clp-dependent degradation of the IrvR DNA binding domain. Thus, irvA derepression is highly analogous to the genetic switch mechanism used to regulate lysogeny in bacteriophages.
Streptococcus mutans is a Gram-positive oral commensal species that is commonly recognized to be one of the primary organisms associated with the development of dental caries (cavities) (2, 4, 21, 26, 33, 35, 39). Caries is a chronic, progressive disease that occurs as a consequence of bacterial metabolism. Thus, virulence for cariogenic species such as S. mutans is directly proportional to its success at long-term persistence and proliferation within the oral biofilm. This success is critically dependent upon environmental stress tolerance abilities as well as the capability to eliminate competitor species that vie for the same ecological niche (1, 13, 23). Interestingly, genetic mutations of pleiotropic regulators in S. mutans often yield phenotypes to several or all of these persistence-related abilities. For example, various phenotypic studies of S. mutans luxS mutant strains have reported alterations in oxidative and acid stress tolerances as well as defects in biofilm formation, natural competence, and bacteriocin (mutacin I) production (18-20, 36, 38). It is currently unknown why these phenotypes tend to occur together, but our studies suggest that this could be at least partially explained by the ability of certain mutations to trigger the expression of the transcription regulator irvA.
IrvA was first identified while investigating the mutacin I-negative phenotype of the luxS mutant. While irvA expression was found to occur at a low basal level in the wild type, it was highly expressed in the luxS mutant strain (18). However, a double deletion of irvA and luxS was able to suppress the mutacin I deficiency phenotype. Similarly, several other genetic mutations were also found to induce irvA and possess mutacin I deficiencies (34). In addition, irvA induction was demonstrated to create a severe natural competence deficiency in the cell as well (25). Similar to mutacin I, a deletion of irvA could suppress the competence phenotype associated with irvA expression. This suggests that irvA can strongly influence multiple virulence-related functions but only under inducing conditions.
Recently, we reported the identification of the regulator primarily responsible for repressing irvA expression (25). A deletion of this gene, which we referred to as irvR, caused a constitutive derepression of irvA and the activation of the dextran-dependent aggregation response (DDAG). In the wild-type cell, DDAG is typically observed only under environmental stress conditions and has been demonstrated to be mediated by the dextran binding protein GbpC (3, 28, 29). Accordingly, gbpC was shown to be much more highly expressed in the irvR background but not in the irvR/irvA double mutant background. Our laboratory and others have also observed a similar increase in gbpC expression in the luxS background (18, 32). Thus, experimental evidence suggests that irvA is associated with both stress-responsive and virulence-related pathways in the cell.
Consistent with the stress hypothesis, irvR and irvA resemble the phage regulators cI and cro in their genetic organization as well as predicted amino acid sequences. In the λ phage, the CI repressor is responsible for maintaining lysogeny by repressing cro. However, when the cell encounters environmental stress, the CI repressor will undergo autocleavage followed by proteolytic degradation, which relieves the repression of cro and allows the lytic cycle to progress (27). In the current study, we aimed to determine the mechanism used to relieve the repression of irvA by IrvR. We demonstrate that the activity of IrvR is controlled in a manner highly analogous to that of the CI protein. It has the ability to undergo autocleavage, which promotes the proteolytic degradation of its DNA binding domain leading to the derepression of irvA. These data also imply that the classic genetic switch mechanism used to induce the lytic response in phage is not unique to viruses and could be a common strategy used in other organisms.
The bacterial strains and plasmids used in this study are listed in Table Table1.1. Escherichia coli cells were grown in Luria-Bertani (LB) broth or on LB agar (Difco) at 37°C. E. coli strains harboring plasmids were selected with 100 μg ml−1 ampicillin (Fluka), 100 μg ml−1 kanamycin (EMD), or 150 μg ml−1 spectinomycin (Sigma). All S. mutans strains were grown anaerobically (85% N2, 10% CO2, and 5% H2) at 37°C. S. mutans strains were cultivated in brain heart infusion (BHI) medium (Difco). For the selection of antibiotic-resistant colonies, BHI plates were supplemented with 800 μg ml−1 kanamycin, 15 μg ml−1 tetracycline (Sigma), 15 μg ml−1 erythromycin (Sigma), or 900 μg ml−1 spectinomycin.
Total RNA extraction, cDNA synthesis, and real-time PCR of S. mutans UA159 and its derivatives were performed as described previously (25). The protocol is briefly summarized as follows. Overnight cultures of S. mutans UA159 and its derivatives were diluted 1:30 in 30 ml BHI with 0.4% bovine serum albumin. Cells were harvested by centrifugation after the optical density at 600 nm (OD600) reached ~0.3. The pellets were resuspended in a mixture of 700 μl Tris-EDTA buffer (pH 8.0) and 600 μl of Tri reagent (Sigma). The suspensions were then homogenized with a FastPrep-24 system (MP Biomedicals), followed by acidic phenol-chloroform extraction and isopropyl alcohol precipitation. After treatment with RNase-free DNase (Promega), RNA was further purified over RNeasy spin columns (Qiagen). Total RNA (500 ng) was used for cDNA synthesis using SuperScript II (Invitrogen) according to the manufacturer's protocol. Real-time reverse transcription-PCR (RT-PCR) was performed with Bio-Rad iTaq SYBR green Supermix with ROX, and the 16S rRNA gene was used as the housekeeping gene reference.
Determination of genetic competence and DDAG assay were performed essentially as described previously (25). In brief, cells were grown as mentioned above to an OD600 of ~0.3. Genomic DNA containing a tetracycline marker (10 μg ml−1) was added to each reaction mixture, and the cultures were incubated for an additional 2 h. After the incubation, the cells were briefly dispersed by sonication and plated on tetracycline-containing BHI agar plates as well as on nonselective BHI plates. Successful transformation was scored based on tetracycline resistance, and the total viable cell population was determined by counting the number of colonies growing on nonselective plates. The transformation efficiency was determined by calculating the ratio of transformants to total viable cells.
For the DDAG assay, overnight cultures of S. mutans and its derivatives were diluted (1:100) into 3 ml fresh BTR-G broth. The cultures were incubated for an additional 24 h and then divided into two 1-ml portions with or without 100 μg ml−1 dextran T2000 from Leuconostoc spp. Each pair of tubes was swirled briefly, and aggregation was observed as obvious clumping and cell precipitation within 1 to 2 min of swirling the cultures.
All strains are listed in Table Table1,1, while all of the primers used for strain construction are listed in Table Table2.2. Different truncations of irvR were created to investigate its function. Sequences encoding IrvR1-122, IrvR1-191(VSA), IrvR1-191(VSD), and IrvR1-191(VDD), where the superscript numbers represent amino acids and the superscript letters represent C-terminal epitopes, were PCR amplified with Pfu Turbo DNA polymerase (Stratagene) and the primer pairs irvR-c F/irvR-c1-122 R, irvR-c F/irvR-c1-191(VSA) R, irvR-c F/irvR-c1-191(VSD) R, and irvR-c F/irvR-c1-191(VDD) R. The PCR products were digested with EcoRI/XhoI and cloned into EcoRI/SalI-digested pDL278 to give pGN02, pGN03, pGN04, and pGN05. The resulting plasmids were transformed into S. mutans UA159 to create the strains GN02c, GN03c, GN04c, and GN05c. Linearized plasmid pGN01R was next transformed into each of these strains to delete the chromosomal copy of irvR to create GN02 (IrvR1-122), GN03 [IrvR1-191(VSA)], GN04 [IrvR1-191(VSD)], and GN05 [IrvR1-191(VDD)]. A similar strategy was used for the construction of GN06 (IrvR1-122+VSA), GN07 (IrvR1-122 + 3VSA), GN08 (IrvR1-122 + 8VSA), GN09 (IrvR+VSA), GN10 (IrvR+3VSA), and GN11 (IrvR+8VSA), except that the following primer pairs were used: irvR-c F/irvR-c1-122+VSA R, irvR-c F/irvR-c1-122 + 3VSA R, irvR-c F/irvR-c1-122 + 8VSA R, irvR-c F/irvR-c+VSA R, irvR-c F/irvR-c+3VSA R, and irvR-c F/irvR-c+8VSA R. The clpB, clpC, clpE, clpL, clpX, clpP, and htrA single mutants were constructed by an overlapping-extension PCR approach. To generate the constructs, two fragments corresponding to approximately 1 kb of the upstream and downstream sequences of each gene was generated by PCR using Pfu Turbo DNA polymerase with the primer pairs clpB KO Up F/R and clpB KO Dn F/R, clpC KO Up F/R and clpC KO Dn F/R, clpE KO Up F/R and clpE KO Dn F/R, clpL KO Up F/R and clpL KO Dn F/R, clpX KO Up F/clpX Em-D Up R, clpX Em-D Dn F/clpX KO Dn R, clpP KO Up F/R and clpP KO Dn F/R, and htrA KO Up F/R and htrA KO Dn F/R. Each of the primers listed as Up R and Dn F incorporated 18 bases complementary to the erythromycin resistance cassette. The erythromycin resistance gene ermAM (16) was amplified by PCR using the primers ermAM F and ermAM R. All three PCR amplicons were purified with the Qiagen PCR purification kit and mixed in a 1:1:1 ratio. The mixture served as the template for a second PCR with the appropriate Up F/Dn R primers. The resulting PCR amplicons (BD-ermAM, CD-ermAM, ED-ermAM, LD-ermAM, XD-ermAM, PD-ermAM, and AD-ermAM, respectively) were cotransformed with linearized plasmid pGN01R into the strain GN03c to delete the chromosomal, wild-type copy of irvR as well as clpB, clpC, clpE, clpL, clpX, clpP, and htrA, all by allelic replacement. This created the strains GN03B, GN03C, GN03E, GN03L, GN03X, GN03P, and GN03A, respectively. The strains GN10P (ΔclpP/IrvR+3VSA) and GN11P (ΔclpP/IrvR+8VSA) were obtained by cotransformation of PD-ermAM and linearized pGN01R into the strains GN10c and GN11c, respectively.
For the construction of clpX/clpC, clpX/clpE, and clpX/clpL double mutants, another clpX deletion construct was employed. In this case, the clpX deletion construct was assembled using an overlapping extension PCR approach with the primers clpX KO Up F/clpX Tc-D Up R and clpX Tc-D Dn F/clpX KO Dn R to generate the amplicon XD-TetM. The primers clpX Tc-D Up R and clpX Tc-D Dn F incorporated 18 bases complementary to the tetracycline resistance cassette tetM (7). The PCR amplicons (CD-ermAM, ED-ermAM, and LD-ermAM) were transformed into GN03c to delete clpC, clpE, and clpL by allelic replacement to create GN03cC, GN03cE, and GN03cL. The strains GN03XC [ΔclpX/ΔclpC/IrvR1-191(VSA)], GN03XE [ΔclpX/ΔclpE/IrvR1-191(VSA)], and GN03XL [ΔclpX/ΔclpL/IrvR1-191(VSA)] were obtained by cotransformation of XD-TetM with linearized plasmid pGN01R into the strains GN03cC, GN03cE, and GN03cL, respectively, to delete clpX and the wild-type copy of irvR in each strain. The resulting transformants were selected sequentially with tetracycline followed by kanamycin.
We first constructed pBSra01 to be used as the template for the mutagenesis of irvR. A fragment containing the entire open reading frames of irvR and irvA as well as their shared intergenic region was PCR amplified with the primers irvRA F/irvRA R, digested with BamHI and XhoI, and ligated into the same sites of pBluescript II KS+. Site-directed mutagenesis was performed using inverse PCR. Mismatches encoding the intended point mutations were incorporated into the 5′ends of primers. Prior to PCR amplification, the primer pairs were phosphorylated with T4 polynucleotide kinase (NEB) to facilitate the religation of amplicons into circular plasmids for E. coli transformation. Inverse PCR was performed with Pfu Turbo DNA polymerase (Stratagene) and subsequently treated with DpnI to digest the parental DNA template. pBSra02 and pBSra03 were amplified using their corresponding mutagenesis primer pairs irvRA191D F/R and irvRK260A F/R. Plasmids were prepared from transformants, and the desired mutations were confirmed by DNA sequencing.
The full-length irvR was expressed in the pET44b system. Wild-type irvR and its mutant derivatives were amplified using the primer pairs pET44b-irvR F/R with pBSra01, pBSra02, and pBSra03 as templates. The addition of 5′-phosphates to pET44b-irvR F allowed the subsequent ligation. The amplicons were then digested with XhoI and ligated into PshAI/XhoI-digested pET44b. In addition, pGN03 was used as a template for PCR and cloned into pET44b using the same strategy. The resulting recombinant plasmids p44r01, p44r02, p44r03, and p44r04 were transformed into BL21(DE3) pLysS E. coli for overexpression. NusA-IrvR and its derivatives were purified according to the manufacturer's protocol (Novagen). After purification, buffer exchange was performed with TSG buffer (50 mM Tris-HCl, pH 7.5, 50 mM NaCl, and 5% glycerol) using Amicon Ultra centrifugal filter devices (molecular weight cutoff, 50 kDa; Millipore). Protein concentrations were determined with protein assay dye reagent concentrate (Bio-Rad) using bovine serum albumin (BSA) standards and stored at −80°C for future use.
For E. coli samples, 120 μg of His6-NusA-IrvR and its derivatives was incubated in 1× rEK cleavage/capture buffer with 0.5 U recombinant enterokinase (Novagen) for 2 h at 22°C. Protein (12 μg) was used for 15% SDS-PAGE separation. After electrophoresis, proteins were transferred to an Immobilon-PSQ transfer membrane (Millipore) in a Mini Trans-Blot electrophoretic transfer cell (Bio-Rad). The membrane was blocked in 40 ml phosphate-buffered saline (PBS) containing 7% (wt/vol) nonfat dry milk overnight at 4°C with agitation, washed twice with 30 ml PBS for 15 min each wash, and then incubated with PBS-diluted anti-IrvR polyclonal rabbit antiserum for 1 h. Next, two additional washes with PBS and one with TBST (20 mM Tris-HCl, 150 mM NaCl, 0.1%[vol/vol]Tween 20, pH 7.4) were performed before adding a 1:30,000 dilution of an alkaline phosphatase-conjugated goat anti-rabbit antiserum (Sigma). After being incubated 1 h, the membrane was washed twice with TBST and once with alkaline phosphatase (100 mM Tris, pH 9.5, 100 mM NaCl, and 5 mM MgCl2) buffer. Lastly, the membrane was immersed in 10 ml BCIP/NBT (Sigma) for staining. Stained membranes were air dried and imaged.
Samples from S. mutans were prepared as follows. S. mutans UA159 and its derivatives were cultivated overnight at 37°C. The overnight cultures were diluted 1:20 in BHI in a total volume of 200 ml. The cells were collected by centrifugation at an OD600 of 0.3. The pellets were then resuspended in 4× 900 μl PBS buffer (pH 7.0) containing 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM benzamidine, and 1 mM EDTA. The suspension was transferred to a 2-ml screw-cap tube containing 500 μl 0.1-mm silica beads (Biospec) and submitted to three consecutive 30-s homogenization cycles with a FastPrep-24 system (MP Biomedicals) set at a speed of 6.0 m/s. After homogenization, the solution was centrifuged for 15 min. The supernatant was concentrated, and total protein concentration was determined with protein assay dye reagent (Bio-Rad) using a BSA standard. S. mutans protein (50 μg) was resolved on 12% SDS-PAGE gels. The Western blot procedure was performed as described above.
Previously, we hypothesized that IrvR and IrvA function as a regulatory pair (25). Consistent with this hypothesis, both regulators also exhibit extensive homology (approximately 40% identity) to the CI and Cro regulators of numerous lysogenic bacteriophages, even though the genome of UA159 does not contain prophage or prophage particles. Recently, we compared IrvR and various phage CI-like repressor proteins and observed that IrvR homology typically starts approximately 15 to 16 amino acids downstream of CI translation start sites. This discrepancy prompted us to inspect the sequence annotation for IrvR (SMu1275 in Oralgen and SMU.1398 in NCBI). Surprisingly, we noticed that the annotated translation start site was devoid of any obvious ribosome binding site (RBS) in the vicinity (Fig. (Fig.1A).1A). While scanning for potential alternative start sites, we also noticed that the IrvR open reading frame actually extends an additional 13 codons upstream of the annotated start site to an alternative start site with a potentially strong RBS. In addition, 16 codons downstream of the annotated start site was another potential start site with a weaker but plausible RBS. Thus, we constructed luciferase translational fusions to each start site and measured reporter activity. Each reporter exhibited similar activity (data not shown), which suggested that the first start site was likely the beginning of the open reading frame. Next, we performed a ClustalW alignment of the first 60 amino acids of the amended IrvR sequence using the N termini of several phage repressor proteins. As shown in Fig. Fig.1B,1B, the N-terminal region of IrvR is nearly identical to several streptococcal phage repressors and highly homologous to a repressor protein from the lactococcal phage BK5-T, which further supports the upstream start codon as the likely translation start site.
The CI protein and other members of the LexA-like protein family each contain two domains, an N-terminal DNA binding domain and a C-terminal S-24 peptidase domain. While considerable sequence variability exists among LexA-like proteins, they all exhibit strict conservation of several core amino acids required for the autocleavage ability of the proteins. In the E. coli LexA protein, these are Ala84 and Gly85, which comprise the cleavage site, and Ser119 and Lys156, which comprise key peptidase catalytic residues (5, 31). To determine whether irvR might encode a similar autocleavage mechanism, we performed a multiple sequence alignment using ClustalW and the amended IrvR sequence. As shown in Fig. Fig.2,2, IrvR, the CI repressor of the streptococcal phage EJ-1, the LexA-like protein HdiR from Lactococcus lactis, and the E. coli LexA all exhibit conservation of the cleavage site and peptidase catalytic residues, despite sharing little overall homology. In addition, it is apparent that all four proteins share nearly identical spacing between the cleavage site and catalytic residues as well. From this alignment, the predicted cleavage site of IrvR would consist of Ala191 and Gly192, while Ser224 and Lys260 would form the predicted peptidase catalytic residues.
A characteristic feature of LexA, CI, and other LexA-like proteins is their ability to undergo spontaneous autocleavage in vitro when incubated in basic pH conditions (15, 30). Given the apparent conservation of the autocleavage site and catalytic residues of LexA-like proteins, we were interested to determine whether these amino acids were similarly required for IrvR autocleavage. To this end, we created several IrvR expression constructs: a wild-type IrvR, an IrvRA191D autocleavage site mutant, and an IrvRK260A catalytic site mutant. As described in Materials and Methods, we expressed each of these constructs in E. coli and incubated the proteins in pH 7.4 buffer for 2 h before performing Western blotting with α-IrvR polyclonal antibodies. As shown in Fig. Fig.3,3, in the wild-type IrvR, we could detect the full-length protein migrating at approximately 34 kDa as well as the predicted N-terminal and C-terminal cleavage products of 23 and 11 kDa, respectively. However, the predicted cleavage site mutant (IrvRA191D) and the predicted catalytic site mutant (IrvRK260A) proteins were strongly impaired in their ability to undergo autocleavage under these conditions, as no cleavage products were detectable. For comparison, we also expressed a His6-tagged IrvR truncated at the predicted cleavage site (Ala191). As expected, the His-tagged IrvR fragment migrated only slightly more slowly than the N-terminal cleavage product of the wild type, due to the presence of a His tag.
Upon autocleavage between Ala191 and Gly192, the N-terminal DNA binding domain of IrvR will expose a new C terminus ending with Val-Ser-Ala-COOH (VSA). Highly similar C-terminal epitopes can be found after the cleavage sites in the phage EJ-1 CI protein (VSA), the lactococcal LexA-like regulator HdiR (LSA) and the E. coli LexA protein (VAA) (Fig. (Fig.2).2). Furthermore, the C-terminal VAA tag has been previously identified as a degradation epitope used to target the N-terminal DNA binding domain of LexA1-84 for degradation by the ClpXP protease (24). Similarly, the ClpP protease is involved in the degradation of the DNA binding domain of HdiR (30). In order to determine whether the IrvR VSA epitope might also be required for the degradation of the IrvR DNA binding domain, we cloned the wild-type irvR and several irvR mutant genes onto the E. coli-Streptococcus shuttle vector pDL278 and compared their ability to repress irvA expression in an irvR mutant background (Fig. 4A and B). Previously, we demonstrated that the full-length irvR could be supplied in trans to complement an irvR deletion (25). This strain served as our positive control. As shown in Fig. Fig.4B,4B, IrvR1-122 was able to strongly repress irvA expression even though this protein was truncated about 70 amino acids upstream of the autocleavage site VSA epitope. This suggested that the transcription regulatory function of IrvR is encoded within the first 122 amino acids of the protein, which is consistent with the predicted DNA binding domain assignment for IrvR (Fig. (Fig.4A).4A). While the larger IrvR1-191(VSA) construct still contained the IrvR DNA binding domain, truncation at the autocleavage site VSA epitope severely inhibited the ability of this protein to repress irvA. This was also true for IrvR1-191(VSD), which contained an Ala-to-Asp mutation in the C-terminal VSA epitope. For both of these strains, irvA expression was almost 70-fold higher than that for UA159, which is nearly identical to our previous results with an irvR deletion mutant (25). However, this was not the case for IrvR1-191(VDD), which contained Asp substitutions in the last two amino acids of VSA. This strain was almost fully competent to repress irvA, which suggested that this construct likely encoded a protein with increased stability compared to IrvR1-191(VSA) or IrvR1-191(VSD). To determine whether this difference could be attributed to Clp-dependent proteolysis, we deleted the clpP protease in the IrvR1-191(VSA) strain and found that irvA gene expression was reduced approximately 100-fold compared to that of the same strain containing a wild-type copy of clpP. Taken together, these results suggest that ClpP is likely responsible for relieving the repression of irvA by degrading the N-terminal DNA binding domain of IrvR via sequence-specific determinants at the autocleavage site. Since it is known that the proteolytic function of ClpP requires the activity of an ATP-hydrolyzing recognition subunit (9), we also assayed irvA expression in the IrvR1-191(VSA) strain containing mutations in each of the predicted Clp ATPases of S. mutans. Surprisingly, none of these mutations could suppress irvA expression nor could a deletion of the protease htrA (data not shown). Previous work in both E. coli and Bacillus subtilis has shown that certain epitopes recognized by ClpX are also substrates for other Clp ATPases as well (8, 10, 40). For this reason, we suspected that a similar redundancy may exist in S. mutans. Therefore, we created a series of strains containing double mutations of clpX and clpC, clpE, or clpL and tested their ability to repress irvA in IrvR1-191(VSA). As shown in Fig. Fig.4B,4B, irvA expression in the clpX/ IrvR1-191(VSA) background is indistinguishable from that in IrvR1-191(VSA). However, each of the clpX double mutants had an increased capacity to repress irvA. While the clpX/clpL and clpX/clpE double mutants exhibited moderately enhanced repression, the combination of clpX and clpC mutations reduced irvA expression from approximately 70-fold to just 4.5-fold over the wild-type value. This suggested that in S. mutans, sequence recognition at the IrvR autocleavage site can be achieved by multiple Clp ATPases, with a particular bias for both ClpX and ClpC.
Recently, we demonstrated that a deletion of IrvR causes a loss of natural competence and the constitutive induction of dextran-dependent aggregation (DDAG) (25). Therefore, we assayed our various truncated IrvR mutant strains to determine which, if any, of them were proficient at controlling these phenotypes. First we examined genetic competence. In the IrvR1-122 and the IrvR1-191(VDD) strains, which were strongly repressive of irvA expression, the transformation efficiency was almost identical to the wild-type strain UA159 and the full-length IrvR (Fig. (Fig.5).5). In contrast, the IrvR1-191(VSA) construct exhibited transformation efficiencies below our detection limit, similar to an irvR deletion mutant (25). Surprisingly, a small number of transformants were detectable in the IrvR1-191(VSD) strain, despite the fact that it was strongly deficient in its ability to repress irvA. The clpP mutation was able to suppress the competence-negative phenotype of IrvR1-191(VSA), but the transformation efficiency remained well below that for the wild type. This is likely due to the fact that Clp-dependent proteolysis is required for competence development (14, 22). When testing DDAG phenotypes, we found that each strain mirrored the results seen with competence. Both IrvR1-122 and IrvR1-191(VDD) resembled the wild-type and full-length IrvR strains, whereas IrvR1-191(VSA) and IrvR1-191(VSD) displayed strong DDAG phenotypes (Fig. (Fig.6A).6A). It was not possible to determine whether a clpP mutation could suppress the DDAG phenotype of IrvR1-191(VSA), as the clpP mutation itself induced DDAG (Fig. (Fig.6B).6B). Even in the absence of added dextran, clpP mutant cells already tended to aggregate (14), albeit to a lesser degree. Since the role of ClpP was difficult to confirm phenotypically, we performed a Western blot assay to determine the relative abundance of the various IrvR1-191 variants in vivo. As shown in Fig. Fig.7,7, IrvR1-191(VSA) and IrvR1-191(VSD) were both undetectable, while the construct containing two point mutations in the VSA epitope [IrvR1-191(VDD)] was present. In the clpP background, IrvR1-191(VSA) was readily detectable at levels comparable to those of the IrvR1-191(VDD) strain, which suggests that ClpP is indeed required for the degradation of the N-terminal autocleavage fragment of IrvR. Given that we also saw a rescue of IrvR1-191(VSA) protein in the clpX/clpC double mutant, we conclude that the degradation likely functions through specific sequence recognition at the C-terminal end of the autocleavage fragment.
Based upon our genetic and biochemical data, it appeared that the sequence of IrvR contains an epitope that targets the N-terminal autocleavage fragment for proteolytic degradation. In LexA, the VAA epitope promotes degradation by ClpXP only when it is located at the C terminus, which explains the requirement for autocleavage to initiate the process of degradation (24). We reasoned that if this was also the case for IrvR, it might be possible to add a VSA tag to the C terminus of the full-length protein and promote constitutive degradation in the absence of autocleavage. However, as shown in Fig. Fig.8,8, adding a VSA tag did not promote the degradation of IrvR. Since VSA was apparently insufficient as a degradation epitope, we created another two strains containing additional amino acids from the autocleavage site added to the C terminus of the full-length IrvR. Strain IrvR+3VSA included the three adjacent amino acids upstream of VSA (YGEVSA), whereas IrvR+8VSA contained an additional eight amino acids upstream of VSA (ENIQIYGEVSA). Interestingly, both of these constructs were efficiently degraded but only when ClpP was present. In the clpP background, both IrvR+3VSA and IrvR+8VSA were no longer targeted for degradation. Additionally, we were able to force the degradation of the truncated IrvR1-122 construct by adding the ENIQIYGEVSA tag to its C terminus. However, unlike the full-length IrvR, in the IrvR1-122 strain, the YGEVSA tag was not sufficient to promote its degradation (data not shown). As expected, adding VSA was also insufficient to target IrvR1-122 for degradation. Thus, we found no evidence suggesting that VSA alone could function as a self-contained degradation sequence, even though it was clearly required. However, the ENIQIYGEVSA tag seemed to contain all of the required elements to target a protein for Clp-dependent proteolysis, whereas the YGEVSA tag appeared to function only in the proper context.
Previously, we noted the similarities between the IrvR/IrvA system of S. mutans and the CI/Cro system of the λ phage (25). These proteins share sequence homology and a similar genetic arrangement on the chromosome. In addition, IrvR and IrvA are probably responsive to stress as well. Here we show that IrvR controls the expression state of irvA in a manner quite analogous to that of CI. It contains all of the features conserved in LexA-like proteins that confer the ability to undergo autocleavage. Once autocleavage has occurred, the IrvR N-terminal DNA binding domain will become sensitive to sequence-dependent proteolytic degradation, resulting in irvA derepression.
Previous work on the E. coli LexA protein has demonstrated that autocleavage is necessary to expose a buried epitope responsible for targeting the N-terminal DNA binding domain for degradation by the ClpXP protease (24). For LexA, this new C terminus ends in VAA, which the authors noted was quite similar to the epitope added to the C terminus of proteins targeted for destruction via the SsrA tag (LAA). The IrvR autocleavage site between Ala191 and Gly192 also contains a similar epitope (VSA) that is likely part of a larger recognition sequence crucial for relieving irvA repression. Our phenotypic and irvA expression data with the IrvR1-122 and IrvR1-191(VDD) strains suggest that the N-terminal portion of IrvR likely remains stable after autocleavage (Fig. (Fig.44 to to6).6). This may partially explain why it is necessary to encode an embedded Clp-dependent degradation tag. For example, the protease FtsH can recognize C-terminal sequences similar to those recognized by ClpXP (12), but unlike the ClpXP protease, FtsH is unable to degrade proteins with high levels of intrinsic thermostability (11). Thus, efficient recognition by the Clp protease is probably crucial for determining the sensitivity of regulation within the IrvR/IrvA system. Indeed, just two amino acid substitutions at the C terminus of the IrvR N-terminal autocleavage fragment made the difference between the constitutive derepression of irvA and its constitutive repression. Even the single amino acid substitution in IrvR1-191(VSD) was sufficient to partially rescue the competence phenotype (Fig. (Fig.5),5), despite the fact that we were unable to detect this protein in our Western blots (Fig. (Fig.7).7). This likely attests to the sensitivity of the genetic switch mechanism used by IrvR, as subtle changes in the efficiency of its degradation may be all that is required to affect the proper functioning of the system.
Given the central role of ClpP in determining the stability of the N-terminal autocleavage fragment of IrvR, it was initially surprising that we were not able to re-create the clpP phenotype with a mutation of any of the known Clp ATPases. The Clp ATPase subunits play an essential role in the recognition and unfolding of proteins targeted for degradation in the core of the Clp proteolytic complex (9). Therefore, it seemed highly unlikely that ClpP alone could be responsible for the degradation of the IrvR N-terminal autocleavage fragment. Previous studies of other organisms have suggested that Clp ATPases possess overlapping sequence recognition requirements. For example, in E. coli, both ClpXP and ClpAP can degrade proteins bearing the 11-amino-acid SsrA tag (10). ClpX interacts with residues 9 to 11 of the tag, whereas ClpA recognizes amino acids 1, 2, and 8 to 10 (8). Also, a similar redundancy was shown with the B. subtilis RsiW anti-sigma factor, which, like IrvR, requires an initial cleavage event to expose a Clp-dependent degradation sequence. Using a mutant RsiW truncated at the cleavage site, the authors demonstrated that a clpX/clpE double mutation is required to fully reproduce the clpP phenotype (40). The specificities of the S. mutans Clp ATPases are currently uncharacterized; thus, our studies of IrvR degradation are some of the first insights into the sequence recognition determinants of its Clp ATPases. Like those of E. coli and B. subtilis, our results suggest that S. mutans Clp ATPases can have functional redundancy. We found that ClpX and ClpC are both likely to efficiently recognize the exposed C-terminal portion of IrvR following autocleavage, whereas ClpL and ClpE may possess a minor recognition ability. These limited contributions of ClpL and ClpE may also account for the slightly higher irvA expression in the clpX/clpC double mutant strain than in the clpP mutant (Fig. (Fig.4B4B).
From the data presented here, it appears that the IrvR/IrvA system shares many features found in the phage regulators CI and Cro. While there is increasing evidence that the cleavage and proteolysis paradigm employed by LexA is likely a mechanism shared by numerous regulatory systems, to the best of our knowledge, the only other characterized system in bacteria that resembles the IrvR/IrvA system is the PrtR/PrtB system of Pseudomonas aeruginosa. PrtR and PrtB are both transcription regulators that share a similar genetic arrangement as CI and Cro. In addition, PrtR functions through the derepression of PrtB in response to stress (37). It has not been determined experimentally whether PrtR undergoes autocleavage and Clp-dependent proteolysis, but sequence conservation suggests this is likely the case (17). Based upon our own searches, we have identified numerous uncharacterized IrvR/IrvA-like pairs in a variety of other Gram-positive and Gram-negative species (data not shown). Therefore, the mechanism employed to regulate irvA is apparently not an unusual feature of S. mutans; rather, it is likely just one example of a common regulatory scheme found in other bacteria.
This work was supported by an NCRR COBRE P20-RR018741-05 and an NIDCR DE018893 grant to J.M. and an NIDCR DE014757 grant to F.Q.
Published ahead of print on 28 December 2009.