Exit from Competence for Genetic Transformation in Streptococcus pneumoniae Is Regulated at Multiple Levels

Development of natural competence in S. pneumoniae entails coordinated expression of two sets of genes. Early gene expression depends on ComE, a response regulator activated by the pheromone CSP (Competence-Stimulating-Peptide). Subsequently, an early gene product (the alternative sigma factor ComX) activates expression of late genes, establishing the competent state. Expression of both sets of genes is transient, rapidly shut off by a mechanism that depends on the late gene, dprA. It has been thought that the rapid shutoff of late gene expression is the combined result of auto-inhibition of ComE and the instability of ComX. However, this explanation seems incomplete, because of evidence for ComX-dependent repressor(s) that might also be important for shutting off the response to CSP and identifying dprA as such a gene. We screened individual late gene mutants to investigate further the roles of ComX-dependent genes in competence termination. A ΔdprA mutant displayed a prolonged late gene expression pattern, whereas mutants lacking cbpD, cibABC, cglEFG, coiA, ssbB, celAB, cclA, cglABCD, cflAB, or radA, exhibited a wild-type temporal expression pattern. Thus, no other gene than dprA was found to be involved in shutoff. DprA limits the amounts of ComX and another early gene product, ComW, by restriction of early gene expression rather than by promoting proteolysis. To ask if DprA also affects late gene expression, we decoupled late gene expression from early gene regulation. Because DprA did not limit ComX activity under these conditions, we also conclude that ComX activity is limited by another mechanism not involving DprA.


Introduction
Streptococcus pneumoniae, a gram positive bacterium inhabiting the human upper respiratory tract, is an opportunistic pathogen which causes many kinds of infection, such as pneumonia, meningitis and otitis. One of the most remarkable features of this microbe is its ability of natural genetic transformation, or competence, discovery of which led to the identification of DNA as the genetic material [1]. In laboratory cultures of S. pneumoniae, an outburst of competence occurs during the mid-log phase, emerging and shutting off rapidly, within about 30 minutes. Recent research has uncovered several aspects of the regulatory process by which virtually all cells of an unsynchronized pneumococcal culture shift, in concert, from an incompetent state to a fully competent state [2]. It starts with the synthesis of a polypeptide ComC, which is exported through a cell membrane transporter ComAB, and processed to a 17-amino-acid peptide, named CSP (competence stimulating peptide). Acting as a pheromone, CSP binds ComD and activates the two component system: histine kinase ComD and response regulator ComE. Phosphorylated ComE is thought to act as a transcriptional activator, recognizing an imperfect direct repeat in front of the promoters of several genes, termed early genes. Among these are comAB and comCDE, which act to produce more CSP, and comX and comW, whose products together turn on expression of the set of genes termed late genes. ComX acts as an alternative sigma factor recognizing a promoter sequence termed the "cinbox'' or ''combox". Finally, some late genes up-regulated by ComX participate in a variety of functions related to transformation, including DNA uptake and processing, recombination, fratricide, and immunity to fratricide, while others have roles that remain to be discovered, but are not required for transformation. While it is clear that the auto-catalytic peptide pheromone CSP serves to coordinate development of competence among the cells of a culture, the signals or stresses that trigger the developmental cycle are only beginning to be discovered [3][4][5].
During the period of maximal competence, also termed the X state [3], transcription is dominated by an excess of the alternative sigma factor, ComX, which is otherwise entirely absent from the cell. To escape from this state, key connections in the circuit must be interrupted decisively. A dramatic temporal pattern of mRNA accumulation and loss in response to an acute dose of the CSP signal attests to the coordination and strength of these effects [6][7][8]. A brief period of early gene expression is followed by a brief period of late gene expression and a somewhat longer period of competence reflecting the activities of the accumulated late gene products.
The lag in expression of the late genes is explained by the role of the early gene product ComX as an alternative sigma factor driving expression of late genes from specific non-canonical promoters. However, as rapidly as this chain of responses to CSP is established, its effects are nearly as rapidly reversed [6,7,9,10], leading to the remarkably transient nature of competence. Early gene mRNA quickly accumulates, increasing at least 100-fold between 2 and 10 min after a sudden increase in CSP level, but these messages then disappear just as quickly, leaving less than 10% of maximal levels in the cell by 15 min. As no specifically targeted anti-mRNA mechanism is known, this suggests that ComE-directed transcription stops abruptly by 10 min. While the ComE protein itself is stable [11], a prime candidate for the cause of this change is a change in the amount of phosphorylated ComE (ComEP). Transcripts of late genes follow a delayed but similar temporal course, peaking at 13 min and largely disappearing by 17 min, implying as well that ComX activity is largely dissipated by 13 min. Although labile, ComX protein is present well past 20 min, suggesting that some inhibition of ComX activity occurs before it physically decays. A further indication of an antagonist of ComX activity was recently provided: while mutation of ClpP stabilized ComX, it failed to prolong its transcriptional activity, detected either as prolonged transformation or as transcription of late genes [12]. Thus, there appear to be at least two targets of competence shutoff mechanisms, the activity of ComEP (which might be inactivated by a phosphatase) and the activity of ComX.
Martin et al [13] reported that the ComE R120S mutant exhibits a much delayed shutoff of competence, with transformation continuing past 90 min, much longer than the 20-min period of competence in wild type cultures. This behavior indicates that ComE activity is an important determinant of the exit from competence and that cells in a competent culture can support an extended period of transformation. Indeed, Claverys and Håvarstein [2] proposed that ComE itself acts to shut off its own activity when accumulated to a high enough level. Thus, it has been thought that exit from competence might be a two-step process involving, first, dephosphorylation or other inactivation of the stable regulator, ComE, followed by degradation of the unstable proteins, ComX and ComW [2,14]. More recently the case for the importance of the second, proteolytic, step was clouded when it was observed that competence still shuts off in a strain where ComX and ComW are stabilized by inactivation of cognate proteases [12]. Furthermore, competence induced by ectopic expression of comX and comW, without participation of other early genes, also follows a course leading to rapid shutoff [15]. These results together indicate that there is at least one factor shutting off competence by targeting ComX directly, independent of proteolysis of ComX and independent of any regulatory effects on expression of early genes.
Additional clues to the mechanism by which competence is terminated are provided by the phenotypes of comX and dprA mutants. While comX mutants do not transform, or become competent by the criterion of expression of late genes, they do respond to CSP, with over-expression of ComX [9,16] and other early genes [7]. The temporal pattern of this response is remarkable: it is not rapidly reversed, as in wild type; instead the induced early gene expression continues for generations [16]. This simple result immediately suggests that one or more late gene products may be important for the reversal of the CSP response, although another logical possibility is that ComX itself acts as such a repressor. A remarkably similar regulatory phenotype was described for mutants defective in the late gene, dprA. Bergé reported that the dprA mutation blocked transformation (later explained by roles of DprA in stabilization or recombination of donor DNA fragments), but also that it causes an exacerbated (,60 min) growth arrest upon CSP treatment, while simulta-neously permitting continued expression of the comCDE early competence operon [17].
To search more broadly for late gene(s) implicated in reversal of the response to CSP, Peterson et al [7] examined the competence kinetics of many mutants defective in genes that were induced in competent cells but not required for transformation (green genes in Figure 1). None of the transformable mutant strains tested displayed an extended period of competence. Mirouze et al. subsequently demonstrated that it was possible to complement the regulatory defect of a comX mutant in escape from the CSP response, restoring transient expression of the comCDE operon, by ectopic comX-independent expression of dprA, under control of an early class promoter [18]. They proposed that the relevant activity of DprA might be either to promote de-phosphorylation of ComE or to block phospho-transfer to ComE by ComD, either of which would be expected to cause a broad effect on all early genes in addition to the effect they observed on the comCDE operon.
Although DprA thus seems necessary for shutoff of the (early) comCDE operon and sufficient for ComX-independent shutoff of comCDE transcription, its regulatory target remains uncertain, and it is unknown whether additional late genes might contribute as well to exit from the X state. To identify possible additional pieces of this puzzle, we looked further among the late genes for products that act to promote the exit from competence. Although the late genes outlined in red in Fig. 1 (cclA, celA, cflA, cglA, coiA, radA, recA, and ssbB) are required for transformation and have critical roles in DNA transport, processing, or recombination, they had not been examined for possible roles in competence termination, perhaps because of the defective transformation in such mutants. This impediment could be circumvented with an indirect method, simply by monitoring shutoff of competence gene expression, instead of decay of transformability per se as an indicator of exit from the competent state. Since the late gene expression pattern is itself a reliable indicator of competence kinetics [17,19] and late gene products are the effectors of transformation, the expression pattern of a late gene revealed by a simple LacZ assay in a ComX + background will best reflect competence development and persistence, even in non-transformable mutants. Using this strategy, with a lacZ transcriptional reporter at the ssbB late gene, we examined mutants defective in additional late genes that are required for transformation. The mutant lacking dprA displayed, as expected, a prolonged late gene expression pattern, whereas mutants lacking cbpD, cibABC, cglEFG, coiA, ssbB, celAB, cclA, cglABCD, cflAB, or radA, exhibited the wild-type pattern of rapid extinction of late gene expression. We further report that closer examination of the effects of DprA on competence regulation revealed that DprA interacts with ComE to control early gene expression but does not affect proteolysis of labile early gene products, while a second, dprA-independent and non-proteolytic, mechanism targets ComX activity during escape from the X-state.

Materials and Methods
Bacterial Strains, Oligonucleotides, Plasmids, and Culture Media AllpneumococcalstrainsusedinthisstudyaredescribedinTable1. The transformation-proficient strains are the derivatives of the Rx1 strain [20] which carries a mis-sense mutation in cps3D [21]. CP2000 carries a deletion of all capsular polysaccharide synthesis genes between aliA and dexB, respectively. CPM7 is a derivative of CP1250 with the lacZ reporter plasmid pEVP3 inserted in the ssbB locus, with a truncated ssbB upstream of the insert and an intact ssbB downstream. All the strains were grown in the casein hydrolysate/tryptone medium, CAT [16,22]. Antibiotics were used at the following selective concentrations: erythromycin (Em), 0.05 mg/ml; streptomycin (Sm), 100 mg/ml; kanamycin (Kan), 200 mg/ml; and tetracycline (Tet), 0.25 mg/ml. CSP1 [23] was obtained as a custom peptide from Mimitopes LLC (Raleigh). Synthetic oligonucleotide primers, purchased from Eurofins MWG Operon (Huntsville), are listed in Table 2. E. coli strains were grown to late exponential phase in LB media [24], and stored with 20% sterile glycerol. For use, the frozen stocks were streaked onto LB agar plates with appropriate antibiotics, and a single colony was then inoculated into fresh medium. Ampicillin was used 0.1 mg/ml for selection of E. coli. Yeast strains s were grown in YPD or SD synthetic media [25], and stored with sterile glycerol at 20% final concentration. For use, frozen stocks were streaked onto YPD agar plate, and a single colony was then inoculated into fresh YPD or synthetic selective media. 1 mM or 3 mM 3AT (3-Amino-1,2,4-triazole) was used to select against histidine production. Plasmids used in this study are listed in Table 3.
Transformation. For S. pneumoniae, cultures grown in CAT to OD550 = 0.03 were exposed to DNA (0.1 mg/ml) for 45 min after the treatment with CSP (0.25 mg/ml), CaCl 2 (0.5 mM), and bovine serum album (0.04%). Transformants were selected in CAT solidified with 1.5% agar containing appropriate antibiotics. For E. coli, 60 ml desalted, electro-transformable cells in 10% glycerol were mixed with 3 ml plasmid DNA (,100 ng) and electroporated with voltage 2.0 kV, resistance 200 V and capacitance 25 mF. After recovering in 1 ml LB broth medium at 37uC for one hour, the cell culture was spun down and plated. For yeast, 1 ml OD600.5 haploid cells were spun down and resuspended in 0.5 ml PEGLET buffer (40% PEG, 0.1 M LiAC, 10 mM EDTA, 0.1 M Tris-HCl, pH 7.5). About 100 mg ssDNA (salmon sperm DNA, Sigma) plus 0.5,1 mg plasmid DNA were added to the resuspended cells. The cell culture was seated on the bench at room temperature overnight. Next day, the cells were spun down again and plated.
Creation of strain CP2000. To create the capsule-less strain CP2000 by use of the Janus replacement cassette [26], CP1250 was transformed with genomic DNA from strain R6J (R6S but cps::kan-rpsL + ) [27], using kanamycin for selection. One transformant with the correct structure was retained as strain CP1999. CP1999 was transformed with a donor DNA created by the ligation of two BamHI digested PCR fragments. The first fragment was created by amplifying CP1500 genomic DNA with primers TTM01 and TTM02; the second fragment was created by amplifying the same genomic DNA with primers DAM823 and DAM827. A Sm R transformant with the correct structure, removing genes between dexB and aliA, was retained as CP2000, after verification by PCR and sequencing of the new junction/ deletion.

Construction of parental strains CP2108 and
CP2125. The parent strain CP2108 was made by the crosses among three strains: CP2000, CPM7, and CP1415. CP2000 (Dcps), was transformed with CPM7 genomic DNA; a Cm R transformant with pEVP3 inserted at ssbB and exhibiting CSPdependent b-galactosidase activity was named CP2107. This new strain was further transformed with CP1415 genomic DNA, to incorporate a disrupted comA gene. The disrupted comA gene in an Em R transformant was confirmed by PCR and the strain was named CP2108. Then CP2108 was transformed with CP1359 (DclpP::PcTet) genomic DNA to make CP2125.
Creation of ectopic comXcomW strain, CP1896. To create strain CP1896 (aga::comX::comW::PcKan), two fragments were amplified from CP1372 DNA: one contained part of aga and a complete comX (primers DAM786 and PL82); the second contained a Kn R cassette and part of rafE (primers DAM791 and Internal deletion mutants studied in this paper are indicated by blue ovals. C represents the cinbox promoter, TACGAATA. Fill colors symbolize the functions of proteins, as indicated by the key on the right. Bent arrow and lollipop stand for constitutive promoter and terminator respectively. Red borders indicate genes required for transformation; green borders indicate genes not required for transformation. Black borders indicate genes whose importance for transformation has not been directly determined. Genes described by numbers (ORF numbers in TIGR4) are those whose functions in competence are unknown. The ORF numbers for named late genes are: cinA(sp1941), cbpD(sp2201), cibA(sp0125), coiA(sp0978), dprA(sp1266), ssbB(sp1908), cclA(sp1808), cglA(sp2053), celA(sp0054), cflA(sp2208). The late gene clusters and associated cinbox are as in Peterson et al [9], but without cases of apparent read-through transcription identified in [34]. doi:10.1371/journal.pone.0064197.g001 DAM792). A third fragment containing comW was amplified from CP1500 using DAM794 and DAM790. After digestion by BamHI (Fermentas), and/or by EcoRI (Fermentas), the three fragments were purified, ligated, and used directly as donor for transforming strain CP1250 as described above. One Kn R transformant was retained as strain CP1896 after sequencing the insert, which exactly matched the predicted sequence. Preparation of late gene mutants. Late gene mutants used in this paper were obtained by transforming the parent strains (CP2108 and CP2125) with either the genomic DNA of strains containing the corresponding late gene disruption (DcbpD, DcibABC, DcglEFG, DcoiA, DdprA, and DradA), or with donor DNA synthesized by molecular cloning (DP c cinA, DcelAB, DcclA, DcglABCD, DcflAB, and DssbB). For synthesizing donor DNA, two pairs of primers were used to amplify the upstream and downstream sequences individually. Then, the two sequences were ligated with a Kan R cassette to make up the donor DNA for transformation.
Induction of competence using CSP. Cultures of pneumococcus were started by inoculating 200 ml complete CAT medium plus 10 mM HCl with 1/100 volume of a frozen stock of cells (OD 550 0.1). At the first visible turbidity during growth at 37uC, 10 ml was transferred to a 18 mm by 150 mm tube for monitoring optical density. When the culture reached an OD 550 of about 0.06, it was induced to competence with CaCl 2 (to 0.5 mM), BSA (to 0.002%), and CSP (to 250 ng/ml). Samples were taken periodically from the culture for various analyses as described in subsequent sections.
Induction of competence using raffinose. Cultures of pneumococcus were started by inoculating 200 ml complete CAT medium plus 10 mM HCl with 1/100 volume of a frozen stock of cells (OD 550 0.1). At the first visible turbidity during growth at 37uC, 10 ml was transferred to a 18 mm by 150 mm tube for monitoring optical density. When the culture reached an OD 550 of about 0.05, the culture was transferred to 30uC. When the culture reached an OD 550 of about 0.1, it was induced to competence with CaCl 2 (to 0.5 mM), BSA (to 0.002%), and raffinose (to 0.1% w/v). Samples were taken periodically from the culture for various analyses as described in subsequent sections.
Sampling for western analysis. For Western blot analyses 1.8-ml samples were withdrawn from the culture, chilled rapidly on dry ice, without freezing, and then kept at 4uC until harvesting by centrifugation (10,0006g, 2 min, 4uC). After each cell pellet was resuspended with 35 ml loading buffer (50 mM Tris-HCl (pH 6.8), 2% SDS, 0.1% bromophenol blue, 10% glycerol, and 100 mM dithiothreitol) and heated at 95uC for 10 min, 15 ml of the lysate was loaded into one lane of an SDS-PAGE gel.
SDS-PAGE. SDS-PAGE was done as described previously [24] using the Bio-Rad Mini-Protean II gel apparatus. Each gel was a 15-well, 1.5-mm-thick discontinuous gel composed of a 5% stacking gel and a 15% resolving gel prepared according to the manufacturer's recommendations. Protein samples were prepared by mixing with one volume of loading buffer (100 mM Tris-HCl (pH 6.8), 4% SDS, 0.2% bromophenol blue, 20% glycerol, and 200 mM dithiothreitol) and heating at 95uC for 10 minutes. The gels were run in 25 mM Tris, 250 mM glycine, 0.1% SDS, pH 8.0 at 65 V until the dye reached the resolving gel, then at 120 V until the dye reached the bottom of the gel.
Western analysis. For Western analysis, SDS-PAGE gels were run as described above, washed 3 times for 5 min with 100 ml deionized water, and equilibrated for 15 min in transfer buffer (25 mM Tris, 192 mM glycine, 10% (v/v) methanol, pH 8.0). The proteins were trans-blotted from the gel to a PVDF membrane (Immobilon P sq , Millipore) in transfer buffer for 2 hr at 36 V at 4uC. The membrane was then blocked overnight at 4uC in TBS-T (20 mM Tris-HCl (pH 7.6), 137 mM NaCl, 1% (v/v) Tween-20) with 5% nonfat dry milk. The membrane was incubated for 90 min at room temperature with primary antibody, specific for ComW or ComX, diluted by 1:3000 in 5 ml TBS-T with 1% nonfat dry milk in a small plastic pouch. After washing 3 times with 100 ml TBS-T, then incubating for 1 hr at room temperature with secondary antibody (Anti-Rabbit IgG conjugated to HRP, Amersham) diluted 1:20 000 in TBS-T with 1% nonfat dry milk and then washing 3 times with 100 ml TBS-T, the position of secondary antibody on the membrane was detected using an ECL substrate (ECL Plus, Amersham) and either Hyblot CL film (Denville Scientific) or the Alpha Imager CCD Camera (Alpha Innotech). Typical exposure times were 1 to 5 min for the film and 5 to 15 min for the CCD camera. Quantification was done by spot densitometry using AlphaEaseFC (Alpha Innotech).
Yeast two-hybrid assay. A yeast two-hybrid assay was established by inserting dprA (prey) into yeast plasmid pGBDUC2 (containing DNA binding domain of GAL4 transcription factor), and inserting comD, comE, comX, comW, and recA (baits) into yeast plasmid pACT2 (containing transcription-activating domain of GAL4), using PCR followed by restriction enzyme digestion and ligation at BamHI and XhoI, or SalI sites. The two plasmids with corresponding inserts were first transformed into E. coli (DH5a) to replicate and purified with QIAprep Miniprep kit. The purified plasmid DNA was then transformed into yeast haploids PNS468 (MATE-a) and PNS752 (MATE-a) respectively. The two transformed haploid cells with different mating types were selected in SD medium without leucine (SD-Leu) or SD medium without uracil (SD-Ura) and mated to make diploids in SD agar plate without both (SD-Leu-Ura). The diploids were collected and stored at 280uC in 20% glycerol stocks. Diploids were grown again on YPD agar plates and frog-replicated onto four test agar plates: SD-Leu-Ura; SD-Leu-Ura-His; SD-Leu-Ura-His+1 mM 3AT; SD-Leu-Ura-His+3 mM 3AT [28]. The four test plates were photographed daily during incubation at 26uC for 7 days.

Among Candidate Late Competence Genes Examined, Only dprA is Required for Normal Shutoff of Late Gene Expression in a Wild-type Background
To investigate the roles of additional late genes in competence termination, we monitored the expression pattern of a late gene transcriptional reporter, which is established as a reliable indicator of competence development [17]. For this purpose, we constructed a parent strain that contains a lacZ reporter at the late gene, ssbB and is also comA deficient, so as to avoid any potential growth deficiency that might be caused by prolonged competence in a mutant deficient in exit from competence. The late gene expression pattern in response to CSP in this reporter parent strain was verified as identical to that reported previously [7,9,16,29]. Mutations in genes with known roles in transformation were crossed into the parent strain. Included were cibABC and cbpD, which affect fratricide, and cinA, linked to the cinbox promoter required for induction of recA by CSP. The internal deletion of each gene or operon was replaced by a Kn R marker. Altogether, we examined 12 mutants, defective in a total of about 20 late genes (circled with in blue ovals in Fig. 1): DcbpD, DcibABC, DssbB, DcglEFG, DcoiA, DdprA, DcelAB, DcclA, DcglABCD, DcflAB, DradA and DPc-cinA. Each mutation's structure was confirmed by PCR, and the border of  each deletion is shown in Fig. 1. Since the transformation efficiency of these mutants was already known to be very low, we verified that transformation rates for the new strains were comparable to the values expected from the literature (Table 4).
To determine the effect of each of the 12 mutations on exit from competence, each mutant reporter strain was treated with CSP under standard conditions and sampled 0, 10, 20, 30, 40, 60, or 80 minutes later for LacZ assay. The enzyme levels, which indicate late gene expression patterns, are shown as a function of the time after CSP induction in Fig. 2. Nearly all of these mutants displayed normal (brief) patterns of late gene expression, with a burst of LacZ synthesis restricted to the period between 10 and 30 min after addition of CSP. The single exception was the dprA mutant, in which late gene expression continued beyond 60 min The pattern shown in Fig. 2 reproduces the growth defect reported by Bergé [17] for the R6 strain, and further supports his interpretation by showing directly that expression of a late gene continues for an unusually long time after exposure of a dprA mutant to CSP; that is, the cells appear to remain in an active X-state for a greatly extended period when DprA is missing.
The greatly prolonged expression of a late gene suggests a prolonged presence of ComX and thus that DprA affects not only expression of the comCDE operon, but also expression of the duplicate comX genes. Combined with the report of Mirouze [18] Table 1). B. Growth patterns monitored as culture optical density at 550 nm before and after CSP treatment. Bracket indicates period of LacZ assay shown in panel A. doi:10.1371/journal.pone.0064197.g002 that premature expression of dprA repairs the transience defect of a comX mutant, these results suggest that DprA is both necessary and sufficient for the shutoff of early gene expression. From the absence of any similar effect of the other late gene mutations on the temporal pattern of ssbB expression we conclude further that none of the other late gene products examined has a strong role in shutoff of early gene expression. Because of the uniquely strong effect of DprA on shutoff, it is of interest to determine its effects on gene expression in more detail, both to know if it is the only shutoff agent, and to identify its regulatory targets.

The Kinetics of Exit from Competence is not Altered in a ClpP Protease-deficient Background
We were concerned to be able to detect regulation of comX at multiple levels. While the effect of DprA shown above is dramatic, as expected for a factor controlling early gene expression, the change in late gene expression kinetics might be more subtle for mutation of a gene reducing ComX activity directly. Thus, for all late gene mutants tested above, except the DdprA mutant, levels of ComX and ComW would be expected to start to decline by ,30 minutes after CSP induction, simply due to the rapid dprAdependent halt to early gene expression and subsequent decay of ComX and ComW that had accumulated during the brief window of ComE activity. Therefore, any extension of late gene expression occasioned by mutation of a gene acting specifically to suppress ComX activity during this window might be modest in length and difficult to detect. To extend the window of ComX availability and thus improve the chance of detecting such a ComX-dependent shutoff gene targeting ComX itself, we decided to study the temporal pattern of late gene expression in late gene mutants in a protease-deficient background (DclpP), in which both ComX and  ComW would be stabilized. ClpP and ClpE are largely responsible for the proteolysis of ComX and in strains deficient for either of the two proteins, ComX becomes stable [30]. Similarly, ClpP and ClpC are largely responsible for the proteolysis of ComW and in strains deficient for one or the other protease subunit, ComW is stable [12].
Adopting the same strategy as in the previous section, a new parent strain, CP2125, was made from the original one by disruption of the clpP gene, to increase stability of both ComW and ComX proteins. From this new parent strain, we again obtained 12 late gene mutants: DcbpD, DcibABC, DssbB, DcglEFG, DcoiA, DdprA, DcelAB, DcclA, DcglABCD, DcflAB, DradA and DPc-cinA, and verified their structures and competence phenotypes as above. These mutants were analyzed for their late gene expression pattern after induction by CSP. In the DclpP background, we observed the same result as in the protease-proficient wild type: only the dprA mutant displayed prolonged late gene expression (with an elevated b-galactosidase activity, which may be due to the deficiency in host Clp proteases (Fig. 3). Since our survey of ,19 late genes other than dprA did not reveal any whose loss extends the X-state when the half-lives of ComX and ComW are prolonged by interruption of their proteolysis, we conclude that none of these gene products is individually responsible for suppression of ComX activity during exit from competence.

Accumulation of the Early Gene Products, ComX and ComW, is Increased and Prolonged in a dprA Mutant
The prolonged late gene expression in a DdprA mutant might in principle reflect abrogation of a direct effect of DprA on late gene expression or interruption of an effect of DprA at the level of ComE, affecting early gene expression. These possibilities could be distinguished by comparing the accumulation of the early gene products, ComX and ComW, in a dprA mutant to that in wild type. If they displayed comparable levels of ComX and ComW, it would suggest action of DprA on late gene expression; whereas higher levels of ComX and ComW in the dprA strain would indicate that DprA shuts off expression of these additional early genes. To test the possibility that DprA negatively regulates the amounts of ComX and ComW produced during competence development, we used Western blotting to compare the levels of ComX and ComW in dprA + and dprA 2 strains upon treatment with CSP. As shown in Figure 4, both ComX and ComW accumulated too much higher levels in the dprA mutant and remained at a high level even after 70 min of exposure to CSP. In contrast, in the dprA+ strain ComX and ComW reached a lower maximum at 15-20 min, and then declined below detectable levels by 50 min. We conclude that dprA mutation prolonged competence by enhancing the accumulation of the early gene products, ComX and ComW.

The Effect of DprA on the Accumulation of ComX and ComW does not Depend on ClpP
In light of the lability of ComX and ComW proteins, the difference in the CSP-induced levels of ComX and ComW between dprA 2 and dprA + strains might have two quite different explanations: 1) DprA may promote the proteolysis of ComX and ComW, acting as chaperon or adaptor that greatly stimulates the targeting of these proteins to the proteases responsible for their lability, ClpEP and ClpCP, respectively; or 2) DprA may inhibit the production of ComX and ComW, at either translational or transcriptional stages. To distinguish experimentally whether the increase in amounts of ComX and ComW caused by dprA mutation reflects a negative effect on proteolysis of the labile ComX and ComW proteins or, in parallel to the effect of DprA on expression of the comCDE operon, an enhancement of the rate of synthesis of these proteins, we compared the amounts of ComX and ComW in a DclpPdprA + mutant to the amounts in a DclpP dprA 2 mutant. As ClpP is necessary for the proteolysis of both ComX and ComW, both proteins are stable in a DclpP background and an anti-proteolytic mechanism would have no effect in that context.
Late gene expression differed radically between the two strains was ( Figure 5), consistent with the result in Figure 4. The amounts of ComX and ComW, as determined by Western blot, continued to increase for more than 50 min in the dprA clpP mutant, whereas the levels of the two proteins became constant after 15 min in the dprA+ clpP mutant strain. The continued accumulation of both proteins well after the time that transcription of early genes would stop in a wild type strain [7] shows that DprA acts to block transcription (or translation) of both genes, not to promote proteolysis of these labile early proteins. Since ComX and ComW belong to distinct operons in the early class, and DprA suppresses transcription of the early operon comCDE [17], this now shows that three early operons respond similarly to deprivation of DprA. We conclude that DprA suppresses the transcription of at least three, and likely all, early genes.

DprA Interacts with ComE
In view of the effect of DprA on early gene expression, we made a limited search to distinguish among possible molecular targets, through a yeast two-hybrid screen for protein interactions. ComD or ComE, the major players in turning on early gene expression, were the major candidates in our screen. Because DprA interacts with RecA [31], a recA insert was adopted as a positive control. Furthermore, because of results indicating that DprA does not affect late gene expression (described below), comX and comW inserts were included as negative controls. After confirmation of the constructions, pGBDUC2 and pGBDUC2-dprA were transformed into yeast haploid NSY752(a). On the other hand, pACT2, pACT2-comD, pACT2-comE, pACT2-comX, pACT2-comW and pACT2-recA were transformed into yeast haploid NSY468(a). Six diploids were obtained by mating between NSY752(a) and NSY468(a) cell lines, as indicated in Figure 6.
After incubation of the diploids in four kinds of media (SD-Ura-Leu, SD-Ura-Leu-His, SD-Ura-Leu-His+1 mM 3AT and SD-Ura-Leu-His+3 mM 3AT), all diploids grew on SD-Ura-Leu ( Figure 6A), confirming that plasmids pGBDUC2 and pACT2 (or their derivatives) were harbored respectively in the diploids. But in the plates without histidine (SD-Ura-Leu-His), only the diploids containing inserts dprA+comE and dprA+recA produced colonies. This indicates that DprA can interact with ComE as strongly as it interacts with RecA ( Figure 6B). The growth in plates without histidine but with the competitive inhibitor of the production of histidine, 3AT, further confirmed the interaction between DprA and ComE is positive and strong, perhaps even stronger than the interaction between DprA and RecA because the diploid with dprA and recA inserts was unable to grow in the media SD-Ura-Leu-His+1 mM 3AT and SD-Ura-Leu-His+3 mM 3AT after a week incubation ( Figure 6C, 6D). Further verification of the interaction between DprA and ComE was achieved with yeast two-hybrid by incorporating more negative controls, including double empty vectors and one empty vector with one vector plus insert (data not shown). Taken together, these results suggest a direct interaction between DprA and ComE, an interaction that might mediate its strong negative effect on expression of early genes.

Late Gene Expression is Independent of DprA after Ectopic Induction of comX and comW
The prolonged transcription of the late gene reporter seen in dprA mutants contrasts with the rapid extinction of late gene expression in clpP mutants despite the continued presence of high levels of the otherwise labile ComX and ComW [12], suggesting that DprA or some other late com gene product might also play an additional regulatory role by affecting late gene transcription directly, supplementing any effects on early gene expression and the amount of ComX and ComW produced.
Since ComX and ComW are the only early gene products required for high levels of late gene transcription and cells can become fully competent when comX and comW are ectopically expressed under CSP-independent regulation [15], we sought to decouple late gene expression from early gene expression by introducing both comX and comW at the aga locus, to allow their expression under regulation by the raffinose-inducible aga promoter. By inducing competence development with raffinose while by-passing expression of the other early genes, DprA's possible effect on late gene expression could thus be separated from its effect on early gene expression. The gene that codes for agalactosidase, aga, was retained in the construct, so that expression from the Aga promoter could be monitored by measuring agalactosidase activity [32], while the ssbB::lacZ reporter allowed monitoring of late gene expression by measuring b-galactosidase activity. To stabilize ComX and ComW, the new strain was also deficient in the ClpE and ClpC ATPases. The new strain, with ectopically regulated comX and comW, CP1902 (aga::comX::comW, DclpC, DclpE, ssbB::lacZ::ssbB + ), was transformed with the dprA mutation of strain CP1389, to create the isogenic dprA derivative, strain CP1932.
Since the entire CSP sensing circuit is intact in both the ectopic comX comW strain CP1902 and the isogenic dprA derivative (CP1932), the two strains could be treated with CSP to induce competence development through the activity of early genes, or treated instead with raffinose, to induce competence without expression of any early genes other than comX and comW. When treated with CSP, the two different strains responded with different expression patterns, as expected, confirming the effect of the dprA mutation on the 'early' regulatory pathway (Fig. 7). When the two strains were treated instead with raffinose, the patterns of late gene expression in the dprA + and dprA 2 strains were identical (Fig. 7), except for a somewhat broader curve compared to CSP-induced late gene expression. The broader pattern reflects a slower rise in raffinose-induced ComX and ComW expression. Similar results were obtained with the aga::comX nis::comW strain constructed by Luo (data not shown). As dprA thus had no effect on late gene expression driven by ectopically expressed ComX and ComW, we conclude that DprA does not affect late gene expression by any direct effect on the activity of the alternative sigma factor, ComX. Instead, expression of the early genes via the CSP sensing circuit is necessary to reveal an effect of DprA on late gene expression. Logically, this also suggests that the difference in expression in the CSP treated dprA 2 vs dprA + cells is due solely to DprA's effect on the regulation of early gene expression.

CSP Induces Higher Levels of ComX and ComW in a dprA 2 Background than does Raffinose in the Ectopic ComX/ComW Expression Regime
Since late gene expression induced by raffinose was independent of a functional dprA gene, we infer that there may be another regulator, which reverses late gene expression soon after it begins. We further hypothesize that the apparent failure of this inhibition in the dprA mutant induced by CSP ( Fig. 2 and Fig. 5) could be explained ifComX and ComW were produced in amounts far above those accumulated in the ectopically induced cells of Fig. 7. When ComX and ComW were produced in much higher amount, they might overwhelm the hypothetical regulator, without being affected by it.. To test this hypothesis, we compared levels of both ComX and ComW in CP1932 induced by CSP to the levels in the same strain induced by raffinose, sampling during 80 minutes after addition of the respective inducers. Two parallel SDS-gels were run, one probed with anti-ComX antibody and another probed with anti-ComW antibody. CSP-induced levels of both ComX and ComW were indeed much higher than those induced by raffinose (Figure 8), even while late gene expression, as represented by a ssbB reporter, though prolonged, was not increased in rate (Fig. 1, Fig. 7). Thus, the greatly elevated ratio of ComX to a hypothetical late gene product acting as an inhibitor could explain why late gene expression patterns in dprA mutants shut off after raffinose induction but not after induction by the native CSP pheromone (in Figure 7), and indicates the existence of another competence repressor.

Discussion
The activity of alternative sigma factors is often controlled quite strictly at multiple levels, perhaps because they can cause a global shift in gene expression, which could be especially harmful if carried out under inappropriate circumstances. In the case of the alternative sigma factor central to competence development in S. pneumoniae, at least five distinct mechanisms of regulation are already established or glimpsed. (I) Transcription of comX depends on a TCSTS that coordinates within local populations and responds to unknown elicitors. (II) Production of ComX protein depends on an early competence gene, comW. (III) Separately, activity of ComX depends on ComW. (IV) ComX is labile, targeted by the ClpEClpP ATP-dependent protease. (V) ComW is labile, targeted by the ClpCClpP ATP-dependent protease. With the observations reported here on escape from the X state, two additional mechanisms that regulate the activity of ComX or ComW can be considered, both apparently forming negative feedback loops that ensure that induction of late gene expression is self-limiting and transient. One mechanism requires DprA and targets ComE, perhaps through direct interaction between these two proteins themselves; the other is less well defined, but is independent of dprA and appears to target ComX or other determinant of transcription of late genes.
Since ComX and ComW are central players during competence development and the disappearance of ComX and ComW occurs at about the same time as the loss of transformation [7,9,12,14], it is attractive to suggest that the disappearance of ComX and ComW itself accounts for the shutoff of late gene expression [2,3]. However, the relative timing of events during response to CSP is difficult to reconcile with this simple mechanism. Specifically, late gene mRNA largely disappears even before the levels of ComX and ComW proteins begin to drop, suggesting that the activity of ComX is itself subject to some additional form of control [6,8,12,14]. As some non-proteolytic factor thus seems to play a major additional role in the shut off of ComX activity, we began to look for possible candidates among ComX-induced late genes. Peterson et al [7] screened mutations of many of the late genes for effects on the rate of exit from competence, but found none that caused a pronounced extension of the period of transformation. Transformation defective mutants were not examined for this phenotype, however, prompting us to investigate the latter class of late genes directly. dprA mutants treated with CSP displayed a prolonged period of expression of the late gene ssbB, in addition to the prolonged expression of the early gene operon comCDE previously described as a mutant phenotype for dprA by Bergé [17] and by Mirouze et al [18]. Comparison of levels of ComX and ComW in dprA mutants vs wild type in protease defective backgrounds revealed that DprA not only turns off expression of the early gene operon comCDE, but also has a parallel effect on the early genes comX and comW, thus strongly suggesting an effect on expression of all early genes. Our present results suggest that the target of DprA action within the regulators of early gene expression may be ComE itself. This would provide a direct path to permitting exit from the competent state, although less direct effects of DprA or of its complexes with RecA or with ssDNA cannot be ruled out. As a response regulator of a two component system, phosphorylated ComE accounts for turning on early gene expression. It is our speculation that DprA, via interacting with ComE, might cause de-phosphorylation of ComE, or hinder its recruitment of RNAP to promoter regions, to control early gene expression. Further studies, including distinguishing the two mechanisms and identifying interacting surfaces in both proteins, are warranted.
Recently published results of a parallel study establishing a key role for DprA in exit from competence are based on comXindependent exit from competence when dprA is converted essentially into an early gene [33]. The present study, based on the complementary approach of removal of individual late genes to reveal which is needed for comX-dependent exit from competence, strengthens this conclusion.The apparent role of DprA in terminating late gene transcription could thus in principle be either a secondary effect of its inhibition of early gene expression or could reflect an additional direct effect on late gene expression. To see if DprA also restricted late gene expression directly by an effect on ComX, a new strain was created in this study for the ectopic expression of comX and comW. This new strain can develop competence upon induction of comX and comW by raffinose treatment. In this strain, the ability to transform was transient despite continued presence of ComX and ComW (Fig. 7), providing a good background to evaluate the possibility of a direct effect of DprA on late gene expression. In a dprA mutant derivative of this strain, the pattern of late gene expression following comX and comW induction perfectly matched the pattern in the dprA + parent. This strongly suggests that DprA does not suppress ComX activity in late gene transcription, but that another gene may be responsible for limiting ComX activity to a short time window. This hypothetic regulator is unlikely to be CSP-induced because it could be induced by raffinose, but instead appears to be dependent, directly or indirectly, on induction of comX (or comW), It remains a challenge to reconcile the different patterns of late gene expression in dprA mutants under CSP and raffinose inductions. Specifically, why is late gene expression prolonged in the dprA mutant if there is a separate (late) inhibitor directly targeting ComX? We propose that DprA is not the only factor that shuts off competence, but that a second independent inhibitor accounts for the termination of late gene expression in dprA mutants in the ectopic expression system as well as for the prompt termination of late gene transcription in the WT while ComX is still present [12,14]. As this second repressor was apparently ineffective in CSP-induced cultures in the dprA mutant background, we hypothesize that ComX and ComW are accumulated to different levels with CSP and raffinose. In the DdprA mutant induced by CSP, where DprA, the factor normally curbing early gene expression, is removed, there would be a continuous supply of ComX and ComW at elevated levels, so that even if the second repressor could inactivate a normal amount of ComX or ComW, it could be overwhelmed by the unusually high amounts of these regulators. Direct comparison verified this inferred difference in the levels of ComX and ComW achieved under the two expression regimes, but the hypothetical factor responsible for the shut off of late gene expression remains unknown. While here we have ruled out ,20 late gene products, several other late genes as well as the entire class of 'delayed' genes remain untested. Since competence in S. pneumoniae both imposes a stress on the competent cell itself and creates a potential hazard to nearby cells, it should not be surprising that its initiation and termination are both controlled at multiple levels.