Two operons, comAB and comCDE, play a key role in the co-ordination of spontaneous competence development in cultures of Streptococcus pneumoniae. ComAB is required for export of the comC-encoded competence-stimulating peptide (CSP). Upon CSP binding, the histidine kinase ComD activates ComE, its cognate response regulator, required for autoinduction of comCDE and for induction of the late competence genes. To understand better the early control of competence development, mutants upregulating comCDE (ComCDEUP) were isolated using a comC–lacZ transcriptional fusion. Mutants were generated by polymerase chain reaction mutagenesis of the comCDE region and by in vitro transposon mutagenesis of the chromosome. Both types of ComCDEUP mutants exhibited similar phenotypes. They differed from wild type in displaying trypsin-resistant transformation, competence under acid growth conditions and expression of comCDE under microaerobiosis; increased production of CSP in the mutants could account for the various phenotypes. The ComCDEUP transposon mutations included four independent insertions in the ciaR gene, which encodes the response regulator of a two-component system previously found to affect competence, and two immediately upstream of the comAB operon. The latter two resulted in comAB overexpression, indicating that CSP export is rate limiting. Among comDE point mutations, a single amino acid change in ComD (T233I) conferred constitutive, CSP-independent competence and resulted in comAB overexpression, providing support for the hypothesis that ComE regulates comAB; a ComE mutant (R120S) exhibited altered kinetics of competence shut-off. Collectively, these data indicate that pheromone autoinduction, cross-regulation of the comAB and comCDE operons and, possibly, competence shut-off contribute to the early control of competence development in S. pneumoniae. They argue for a metabolic control of competence, mediated directly or indirectly by CiaR, and they suggest that both comAB and comCDE are potential targets for regulation.
Our understanding at the molecular level of the regulation of the development of competence for genetic transformation of Streptococcus pneumoniae has made considerable progress over recent years. It is now well established that this development is under the control of a two-component regulatory system (TCS) composed of the histidine kinase (HK) ComD and its cognate response regulator (RR) ComE (Pestova et al., 1996). The two genes are part of an operon that also includes the comC gene (Pestova et al., 1996; Cheng et al., 1997). The latter gene encodes a prepeptide that is exported and matured by a dedicated transporter, ComAB (Hui et al., 1995), into the competence-stimulating peptide (CSP), a 17-residue-long pheromone that constitutes a cell-to-cell signal (Håvarstein et al., 1995). The CSP is sensed by ComD, and the signal is most probably transmitted to ComE. ComE binds direct repeat sites adjacent to comCDE and comAB (Ween et al., 1999), which most probably accounts for upregulation of comCDE (Pestova et al., 1996; Alloing et al., 1998) and comAB (Lee and Morrison, 1999) in response to CSP. Interestingly, the strong induction of comCDE transcription only lasts for about 5 min and, while comCDE transcription is shut off, induction of late competence genes occurs (Alloing et al., 1998). The late competence genes share a non-canonical promoter consensus, TACGAATA (Campbell et al., 1998; Claverys and Martin, 1998; Pestova and Morrison, 1998), which is likely to represent the binding site for a recently identified specific sigma factor, ComX (Lee and Morrison, 1999).
The early control of competence induction is not understood as well as later steps in the process. It was proposed that CSP accumulated passively depending on basal level expression of comC, possibly by transcriptional readthrough past the terminator of the tRNAArg located immediately upstream of comCDE (Pestova et al., 1996). An alternative model would involve regulated production of pheromone, which could be temporarily increased in response to changes in environmental conditions (e.g. modifications of nutritional or physical/chemical growth conditions). The observation of an increase in the expression of comCDE in an Obl mutant background (Alloing et al., 1998) provided some credence to this hypothesis. Oligopeptide transport, which involves three oligopeptide-binding lipoproteins (Obl), is believed to play a pivotal role in sensing environmental conditions and to modulate indirectly the expression of several genes (Claverys et al., 2000a). It was also observed that a strain carrying a ciaR− mutation displayed competence under conditions inhibiting the development of competence of wild-type (wt) cells (Giammarinaro et al., 1999). The CiaRH TCS was initially identified in a screen for penicillin resistance determinants in the laboratory (Hakenbeck et al., 1999). Most recently, inactivation of ciaRH was found to result in overexpression of comCDE, and it was proposed that CiaR negatively regulates this operon (Echenique et al., 2000).
In this paper, we report the results of genetic studies to gain a better understanding of the early control of competence development, based on the isolation of mutants upregulating comCDE (ComCDEUP phenotype) obtained by polymerase chain reaction (PCR) mutagenesis of the comCDE region and by in vitro transposon mutagenesis of the chromosome.
Results and discussion
comCDE-linked mutations leading to overexpression of the operon
To construct strains harbouring a comC–lacZ transcriptional fusion to be used as a reporter of comCDE expression, we took advantage of plasmid pXF520. This non-replicating (in S. pneumoniae) plasmid is designed to fuse lacZ to comC while maintaining a fully functional copy of the comCDE operon downstream of the fusion and confers CmR (Pestova et al., 1996). Surprisingly, a mixed population of white and blue CmR colonies was observed after transformation of a wt S. pneumoniae recipient with plasmid pXF520 and plating in T− medium containing Xgal (Xgal-T−). This medium differs from T+, a medium that allows in situ colony competence, by the lack of CaCl2, NaOH and bovine serum albumin (BSA; Experimental procedures) and does not normally support competence development. Strains R331 (white colony) and R810 (blue colony) were picked as representative clones and analysed further. Upon replating in Xgal-T−, strains R331 and R810 gave uniform populations of white and blue colonies respectively. This observation strongly suggested the presence of mutation(s) carried by plasmid pXF520 that would lead to overexpression of comCDE, thus accounting for the blue colonies, whereas white colonies would correspond to recombination events leading to integration of plasmid pXF520, but not of the mutation(s) responsible for comCDE overexpression. To localize the mutation(s), we used a set of PCR fragments covering the comCDE region, amplified from plasmid pXF520 as well as from strain R810, and from strain R331 as a control, as donor in the transformation of strain R331, looking for blue colonies in Xgal-T−. The mutation was unambiguously mapped in the S. pneumoniae chromosomal insert carried by plasmid pXF520 and in the corresponding chromosomal region of strain R810. For example, a 1658-bp-long PCR fragment amplified from R810 DNA yielded 55% blue colonies compared with 0.1% for the fragment amplified from R331 DNA (data not shown). Finally, sequence analysis of the S. pneumoniae insert in plasmid pXF520 identified a single nucleotide change, G → T (UP in Fig. 1) within the terminator of the tRNAArg located upstream of comC (Fig. 1). This change is likely to destabilize the transcription terminator and to allow transcriptional readthrough of comCDE, thus explaining the overexpression of comCDE in T− medium. The mutation in pXF520 readily accounts for a previously unexplained high level of expression of comCDE in strains CPM6 and CPM10 (Table 3 in Lee and Morrison, 1999).
In the above-mentioned transformation experiments, we noticed that control transformations using comCDE PCR fragments amplified from wt (strain R331) DNA also yielded blue colonies with a frequency depending primarily on the length of the donor fragment, but also on the amplified region. For example, yields of blue colonies varied from 0.1% to 1.3%, 10.9% and 21% for 1658-, 2161-, 2465- and 3389-bp-long PCR fragments respectively (data not shown). This prompted us to isolate such colonies and to map the mutation(s) that we termed cup (for competence up; see below). Six out of eight independent mutations were localized in the comCDE region using a set of PCR fragments covering the comCDE region (data not shown). The fragments harbouring mutations were sequenced to determine base change(s) (Fig. 1). Five mutations identified in comD (cup3–6 and cup11) resulted in changes in amino acid residues located in the HK domain (Fig. 1). One in comE (cup10) changed a residue located in the middle of the last α-helix at the C-terminal extremity of the receiver domain of the RR (Lange et al., 1999). cup1 and cup2 probably corresponds to mutation(s) not genetically linked to comCDE.
Cup mutations result in competence development under non-optimal conditions
To determine whether overexpression of comCDE in T− medium also affected transformability of the mutants, spontaneous profiles of competence were established in C+Y standard competence medium and in C-maleate (Experimental procedures). Substitution of maleate for phosphate has been shown to prevent competence of wt cells (Tomasz and Hotchkiss, 1964). Contrary to the wt (Fig. 2, top), strains harbouring the cup3, cup10 (Fig. 2) and cup5 (Table 1) mutations displayed competence in C-maleate. Transformation levels varied with the strains, with the strain harbouring the cup10 mutation exhibiting the highest level.
Table 1. Effect of cup mutations on comCDE expression and on spontaneous competence development.
Competence profiles were also studied under acid growth conditions that prevent competence of wt cells. All mutant strains examined displayed competence (data not shown; Table 1). Finally, we assayed transformation of some of the mutants in the presence of 2 µg ml−1 trypsin, using the wt strain as a control. Trypsin has been shown to inhibit competence development in wt cultures (Tomasz and Hotchkiss, 1964), most probably by preventing the accumulation of CSP above a competence-inducing threshold. The tested strains (cup5 and cup6) displayed a transformation-resistant-to-trypsin (Trt) phenotype (Table 1), suggesting that ComDY312C and ComDT290A required less circulating CSP for activation than wt ComD. The possibility that these mutants could be totally CSP independent was ruled out by an additional experiment (see below).
Because the effects of cup mutations were reminiscent of those of trt1, a mutation isolated on the basis of its Trt phenotype (Lacks, 1970), we decided to investigate whether the mutation was located in the comCDE region. We first introduced the trt1 mutation into a strain harbouring the comC–lacZ fusion (R331) by transformation with chromosomal DNA of a trt1 strain kindly provided by S. Lacks (Table 2) and selection for blue colonies in Xgal-T−. DNA was extracted from a blue transformant and used to transform a wt recipient to CmR. About 60% of the CmR colonies turned blue, indicating that the trt1 mutation was linked to pXF520. DNA sequencing of the region revealed the presence of a single nucleotide change in the comD segment encoding the HK domain (Fig. 1). This result was confirmed by sequencing the corresponding segment of the initial donor trt1 strain. The transferred trt1 mutation conferred a Trt phenotype, as expected, and the ability to transform in C-maleate (Fig. 2) and under acid growth conditions (Table 1).
Table 2. Bacterial strains, plasmids and oligonucleotide primers used in this study.
Strain, plasmid or primer
. Plasmids replicating autonomously in E. coli only.
. Small letters in the oligonucleotide sequence indicate extension to introduce a BamHI restriction site (underlined).
CP1200, but comA::ermAMA (Ω202; by transformation with plasmid pXF202 DNA); EryR, SmR
ColE1 derivative, ApR, CmR; carries a 1303-bp-long minitransposon containing the inverted repeats (IRs) of the Himar1 transposon and ≈ 100 bp of Himar1 transposon sequences flanking the cat CmR gene (magellan2)
gaatggatcCAATTTTTTTATAATTTTTTTAATCTG; matches downstream of aad9 in plasmid pR350 (small letters excepted)
Recently, Echenique et al. (2000) reported that microaerobic conditions inhibit competence development as a result of transcriptional control of comCDE by O2. In agreement with this report, we observed that, for wt cells, microaerobiosis abolished comCDE expression in T+ medium (Experimental procedures;Table 1). Interestingly, all our cup mutations allowed the expression of comCDE in T+ medium under microaerobiosis (Table 1). This observation suggested that the cup mutations could have been selected using microaerobic conditions instead of plating in T− medium and then regarded as affecting regulation by oxygen of comCDE expression. However, there is no indication that any of these mutations, including those isolated by Echenique et al. (2000) on the basis of O2-independent transformability, affects O2 control of comCDE. We propose that these mutations more generally affect the control of the basal level expression of comCDE and that the resulting overexpression of comCDE can be detected under a wide variety of conditions non-optimal for competence development, including microaerobiosis.
Interestingly, the cup10 mutation affected viability in C+Y medium. This mutant stopped growing at OD550 = 0.25, whereas it grew almost normally in C-maleate (Fig. 2, bottom). In addition, it exhibited an abnormal c.f.c./OD550 ratio, indicating that cell viability was affected. We suggest that the ComER120S mutant protein is so efficient at inducing competence that this interferes with normal growth. It is believed that the putative competence-specific transcription factor ComX displaces σ70, which accounts for the synthesis of a small set of proteins at competence (Morrison and Baker, 1979). This probably cannot be tolerated for long and, indeed, competence is rapidly shut off. The poor viability conferred by the cup10 mutation in transformation medium could therefore be a direct consequence of the alteration in the kinetics of shut-off later found to be associated with the mutation (see below). According to this interpretation, viability would be less affected in C-maleate because this medium reduces competence induction. The prediction that a comA− mutation would restore normal viability and growth in C+Y medium to a strain harbouring the cup10 mutation was also verified (data not shown).
Transposon insertions leading to overexpression of the comCDE operon
In vitro mariner mutagenesis of S. pneumoniae chromosomal DNA (see Experimental procedures) was used to generate a library of ≈ 120 000 mutants in strain R354, which carries a comC–lacZ transcriptional fusion (Table 2). Candidate ComCDEupmariner suppressors were isolated as blue colonies in Xgal-T−. From ≈ 33 000 colonies plated, 200 clones representing most of the truly blue colonies were selected in a first screen, of which 49 were retained after a second round of plating. Co-transfer (co-transformation) of the ComCDEup phenotype with the spc cassette was checked first (data not shown). Then, 42 insertions were characterized further by sequencing cassette–S. pneumoniae DNA junctions (see Experimental procedures). The 42 insertions were found to be distributed into nine different location groups. Two of these groups were analysed in more detail and are described below.
The first group comprised four independent insertions in the ciaR gene (Fig. 3A). Investigation of the transformation phenotype of the mutants revealed that they behaved similarly to strains harbouring point mutations in comD or comE described above. They displayed spontaneous transformation under acid growth conditions or in C-maleate (Table 1). The latter observation was fully consistent with a previous report that inactivation of ciaR by a frameshift mutation or of ciaRH by integration of a non-replicating plasmid (insertion-duplication mutagenesis) conferred the ability to develop competence in C-maleate (Giammarinaro et al., 1999). As examination of competence profiles in Fig. 2 suggested that strains harbouring cup mutations could also differ from the wt in displaying transformation at lower cell densities, we checked this for a ciaR− strain as well as for a trt1 strain. Both developed competence at a more than 20-fold reduced cell density compared with wt (data not shown; Table 1). A similar phenotype was reported for an Obl mutant and for a strain harbouring two copies of comC+ and was directly correlated to overexpression of comCDE (Alloing et al., 1998; Table 1). Altogether, these observations suggested that CiaR is directly or indirectly involved in repression of comCDE.
The second group consisted of two insertions of the mariner spc minitransposon immediately upstream of the comAB coding region (Fig. 3B). Insertions spc37C and spc190C (C denotes a co-transcribed orientation of the cassette with the operon) targeted adjacent TA pairs (nucleotides 1659–1600 and 1661–1662 in GenBank/EMBL accession number L15190) in the A-subunit of BOX, a highly conserved repeated element identified in the chromosome of S. pneumoniae (Martin et al., 1992). Interestingly, cassettes were located downstream of the ComE-binding direct repeats (Ween et al., 1999) and were in the co-transcribed orientation with respect to comAB (Fig. 3B). This raised the possibility that the synthetic promoter driving expression of the spc cassette (Claverys et al., 1995; Dintilhac et al., 1997) increased expression of the comAB operon (see below).
The CiaRH TCS and the control of comCDE expression
The observation that four independent comCDEup transposon insertions occurred in ciaR and none in ciaH, together with the fact that the four insertions shared the same cassette orientation (antitranscribed, denoted A; Fig. 3A), prompted us to mariner mutagenize a 5051-bp-long ciaRH PCR fragment generated using the MP144–MP145 primer pair (Table 2). Twenty-four additional insertions were isolated, and their effect on comCDE expression was investigated (Fig. 3A). To summarize, based on observations discussed below, CiaR inactivation seems to affect comCDE expression more drastically than CiaH inactivation. We suggest that either the non-phosphorylated form of CiaR is active in repressing, directly or indirectly, comCDE or that, in the absence of CiaH, enough phosphorylation of CiaR occurs to lead to repression of comCDE.
Fourteen insertions were located in ciaH. Six of these affected comCDE expression only partially (2C, 11C, 12C and 23C) or not at all (17C and 5A) (Fig. 3A). This observation indicates that inactivation of ciaH does not necessarily lead to strong comCDE overexpression. Eight ciaH insertions conferred a comCDEup phenotype similar to that of ciaR insertions. Interestingly, they all corresponded to the antitranscribed orientation of the spc cassette (Fig. 3A). In this orientation, ciaR antitranscripts are generated from the spc cassette promoter. In a similar situation at the comA locus, extinction of a gene as a result of the antitranscripts was observed (see below). It is therefore possible that the eight ciaH::spcA insertions confer a comCDEUP phenotype, not because of the inactivation of ciaH, but rather because of the extinction of ciaR. It would be interesting to carry out Western blot analysis of CiaR in strains harbouring ciaH::spcA insertions to clarify this point. The peculiar behaviour of insertion 5A, which, applying the same reasoning, should also lead to silencing of ciaR, is not understood.
Five out of six new insertions in ciaR, including one in which the spc cassette is co-transcribed with the operon, conferred a comCDEup phenotype (Fig. 3A). The only insertion that did not lead to activation of comCDE, 21A, is located close to the 5′ extremity of ciaR. Analysis of DNA sequence at the junction suggested that insertion 21A could lead to the production of a CiaR protein, the first 10 residues of which are substituted by 10 cassette-encoded residues. The ‘modified’ CiaR protein would lack Asp-9, which corresponds to Asp-13 of CheY, a residue that helps to co-ordinate a divalent metal ion required for phosphorylation (Stock et al., 1995; Lange et al., 1999). According to our interpretation, reduction of ciaR expression would be expected from the orientation of the cassette in insertion 21A. It is possible that, despite low-level expression, the ‘modified’ CiaR can still repress comCDE because it behaves as a super-repressor.
Finally, five cassette insertions were located outside ciaRH (Fig. 3A). Four of them did not affect comCDE expression. The fifth, 20A, which had inserted between the putative promoter of ciaRH and the start of ciaR, was in the antitranscribed orientation. As discussed above, extinction of ciaRH was expected in this situation. In complete agreement with our prediction, this insertion conferred a comCDEup phenotype (Fig. 3A).
CSP export capacity affects comCDE expression and vice versa
To account for the ComCDEUP phenotype resulting from insertion of the spc cassette (spc37C and spc190C) in front of comAB (Fig. 3B), we hypothesized that the promoter driving expression of the cassette increased the expression of the comAB operon. This hypothesis was checked by taking advantage of the existence of comA mutations constructed by insertion of ermAM, an erythromycin resistance gene (Fig. 3B). This gene has been shown to depend on local transcription for its expression and has been used as an indicator of local transcription on several occasions (Claverys et al., 1995).
In order to check the effect of the spc190C–comA+ cassette on comAB expression, the cassette was first transformed into strain R243, which carries the comA::ermAMA (Ω202) construct (Fig. 2 and Table 2). No colony displaying both SpcR and EryR could be obtained, whereas SpcR transformants were readily selected. This suggested that antitranscription initiated at the spc cassette promoter and extending into the ermAM region could be responsible for the silencing of the gene. To check this hypothesis, PCR reactions were carried out on SpcR transformants using the ERY1–comA4 primer pair (Table 2) to identify clones that had retained the ermAM cassette. One SpcR transformant out of 10 displayed an ermAM-specific PCR product, a proportion that would be expected from the distance between the two cassettes (≈ 950 bp). Chromosomal DNA extracted from a representative clone, R630, was then used to transform a wt strain. Transformants resistant to Ery or Spc were obtained, but no double transformant, demonstrating the silencing of the ermAM cassette when placed downstream of the spc cassette in the antitranscribed orientation. An additional independent proof that the spc190C–comA+ cassette enhanced comAB expression was obtained by monitoring its effect on the expression of an ermAM cassette placed downstream of it in the co-transcribed orientation. With this aim, the comA::ermAMC (Ω210) construct (Fig. 2 and Table 2) was transformed into strain R467. EryR transformants were readily obtained and appeared faster than in the wt control (data not shown, but see Fig. 3 for a similar experiment with cup3). Collectively, these data demonstrated that the spc190C–comA+mariner insertion permanently enhanced comAB expression. The resulting increase in CSP export capacity would account for the ComCDEup phenotype conferred by spc37C and spc190C. These observations also strongly suggested the idea that CSP export is rate limiting.
To investigate whether other comCDEUP mutations affected expression of the comAB operon, strains harbouring cup mutations were transformed with comA::ermAMC and comA::ermAMA donor DNA. Only the cup3 mutation (ComDT233I) was observed to increase comAB expression significantly, as judged from the reduction (ermAMC) and the increase (ermAMA) in the time required for the appearance of EryR transformants compared with wt recipient cells (Fig. 4). This observation demonstrated that overexpression of comCDE as a result of the cup3 mutation could in turn affect expression of the comAB operon. cup3 was unique in this respect, most probably because the ComDT233I mutant is fully CSP independent, whereas other mutants require circulating CSP for full expression of the Cup phenotype (see below). It is likely that other cup mutations also affect comAB expression, but this could only be detected by monitoring the expression of comAB in a comA+ background.
Requirement for circulating CSP for comCDE overexpression
As cup3 affected comAB expression in a comA− context, we concluded that the ComDT233I mutant protein could promote comCDE overexpression in the absence of circulating CSP. To investigate whether the other suppressors required circulating CSP, we constructed comA− derivatives. Expression of the comCDE operon was monitored in Xgal-T− (Table 1). The ComER120S RR mutant showed no expression of comCDE when combined with a comA knock-out, indicating that interaction with activated ComD was absolutely required for induction. The ciaR knock-out also did not express comCDE in this context. Only cup mutations in comD resulted in comCDE overexpression in the absence of circulating CSP. This observation suggested that all ComD mutant kinases were able to self-activate. However, when the ability of the various suppressors to develop spontaneous competence in a comA− background was examined, only cup3 turned out to be fully competent (Table 1). A 1000- to 10 000-fold reduction in competence was observed for other comD mutations. The various ComD mutants, except ComDT233I, were concluded to be only partially self-activated in the absence of circulating CSP, including ComDD299N (trt1) that was previously reported to restore transformability in a comA− context (D. Morrison, personal communication; cited in Claverys et al., 1997). These results also indicated that CSP-dependent autoinduction of comCDE contributed to the ComCDEUP phenotype of all suppressors, including ciaR knock-outs and the ComER120S mutant, except for the ComDT233I mutant.
The latter protein is likely to mimic a phosphorylated HK in the absence of CSP, a conclusion totally consistent with the effect of the cup3 mutation on comAB expression in a comA− context. It behaves like the ComEK38E mutant (Echenique et al., 2000) in being fully CSP independent.
Modification of the blind-to-CSP period in the ComER120S mutant
The kinetics of competence shut-off of strains harbouring the various cup mutations, except cup3, were investigated using comA− derivatives to avoid spontaneous induction of competence. The comA− parent displayed a typical decrease in the number of transformants after 20 min contact with CSP (Fig. 5). This decrease occurs in the presence of CSP and has been shown to parallel a transcriptional shut-off of late com genes (Alloing et al., 1998).
Only the cup10 strain behaved differently from the comA− parent, exhibiting a significant alteration in the kinetics of shut-off. Competence decreased only after 50–60 min exposure to CSP and at a slower rate than for the comA− parent (Fig. 5). The alteration of the blind-to-CSP period in the ComER120S mutant could result from a direct or indirect effect of the mutation on the mechanism of competence shut-off. It is possible that shut-off is controlled, at least in part, at the level of the RR ComE. As it was observed that the ComE protein is stable for about 80 min after induction with CSP (Ween et al., 1999), competence shut-off could not involve proteolysis of ComE, but rather dephosphorylation. According to this hypothesis, the ComER120S protein would be more resistant to dephosphorylation than the wt. Alternatively, the mutant protein could be more easily activated upon contact with phosphorylated ComD. The apparent change in the kinetics of shut-off would then reflect the capacity of ComER120S to be turned on by residual levels of phosphorylated ComD normally inactive in the wt.
It is worth pointing out that the kinetics of shut-off of the comA− parent were very similar to those of a wt strain (data not shown). This suggests that no peptide depending on ComAB for export, including the processed GG-containing moiety of ComC, plays an important role in the mechanism of competence shut-off.
Discussion of elements involved in the control of basal level expression of comCDE
A putative ComE binding site was identified upstream of comAB, and a specific shift was obtained upon incubation of ComE with a DNA probe containing this site (Ween et al., 1999). However, the shift was considerably weaker than that obtained with the comCDE site, suggesting that ComE is a weaker transcriptional activator of comAB than of comCDE. Nevertheless, two independent reports of comAB induction at competence (Lee and Morrison, 1999; Rimini et al., 2000) have suggested that activation of ComE at competence resulted in ComE-dependent induction of comAB.
Our finding that the ComDT233I mutant protein leads to overexpression of comAB strongly suggests that the ComE-mediated regulatory link between comAB and comCDE is not only important at the onset of competence, but also plays a role in the control of basal level expression of comAB and comCDE. The observation that mariner cassette-driven overexpression of comAB results in comCDE overexpression indicates that CSP export is rate limiting. It suggests that CSP export capacity directly affects the timing of comCDE autoinduction. Therefore, the control of comAB expression could constitute a key point in the early control of transformation (Fig. 6).
Competence shut-off could also contribute to the early control of competence development. This is suggested by the finding that the ComER120S mutant displayed an altered kinetics of competence shut-off (Fig. 5). This alteration could be directly connected to the ComCDEUP phenotype that led to its isolation. Because mutant cells could produce CSP for a longer time than wt cells, it is expected that a mutation affecting the kinetics of competence shut-off results in earlier accumulation of CSP. In a wt context, the phenomenon of competence shut-off could also play a regulatory role, at the level of the primary CSP producers in a culture, before the peak of competence.
The isolation of four independent mariner insertions in ciaR in a screen for comCDEup mutations suggested that CiaR affects the regulation of comCDE expression. A recent report similarly concluded that the CiaRH TCS negatively regulates comCDE transcription (Echenique et al., 2000). However, it remains to be established whether CiaR is indirectly or directly involved in repression. As the CSP, in conjunction with ComAB and ComDE, exerts a positive feedback on comCDE expression, CiaRH could affect the stability of one of these elements or of some other unidentified component of the feedback loop. If CiaR is directly involved in transcriptional regulation of elements of the competence regulatory cascade, its target could be comCDE or, as discussed above, comAB (Fig. 6). With respect to this question, we noticed that comCDE and ciaRH exhibit 83% identity over a 24-bp-long segment (ATTTt/gGTATAATg/aGTTTTt/ag/aTAAG) including the −10 promoter region of each operon. This segment could correspond to a CiaR binding site, in which case the target for regulation would be comCDE. If CiaR is an effector of competence repression, the question of its active form remains open. Single-residue substitutions in ciaH that led to inhibition of competence (Guenzi et al., 1994) were interpreted, in view of the effect of analogous mutations in the Escherichia coli EnvZ HK, as affecting the phosphatase activity of CiaH; it was suggested that competence is repressed when the CiaRH system is activated (Hakenbeck et al., 1999). One of the mutations in ciaH resulted in the T230P change. We noticed that an identical substitution (T253P) in the very same sequence context (SHELRTPL) in the E. coli CpxA101 protein has been demonstrated in vitro to alter phosphatase activity of the HK towards its cognate RR, CpxR-P (Raivio and Silhavy, 1997). This substantiates the hypothesis that CiaR could remain phosphorylated in the corresponding CiaH mutant and be implicated in competence inhibition. However, it is possible that the CiaHT230P context creates a special situation, as it has been reported that competence of the mutant could not (Guenzi et al., 1994) or could only partially (Hakenbeck et al., 1999) be restored upon addition of CSP.
On the other hand, the distribution and phenotypes of mariner insertions in ciaH (Fig. 3) raise some questions. In particular, the observation that the inactivation of CiaR seems to affect comCDE expression more drastically than CiaH inactivation (see above) could indicate that the non-phosphorylated form of CiaR is active in repressing competence. Another puzzling observation is the finding that transcription of ciaRH increased 10-fold when the K2HPO4 concentration in C+Y medium was raised from 1 to 50 mM, the transcription level at the latter concentration being similar to that in the CiaHT230P mutant (Giammarinaro et al., 1999). If this is taken as an indication that the CiaRH system was activated, most probably through phosphorylation of CiaR, how could this be reconciled with the fact that 50 mM K2HPO4 is precisely the concentration required for development of competence and, therefore, the concentration at which comCDE induction occurs? This might indicate that CiaR-P is not the active form of CiaR for the repression of competence. Finally, as competence is dramatically reduced when cells enter stationary phase, it was suggested that the CiaRH TCS is activated during this period in wt cells (Hakenbeck et al., 1999). Our observation that the competence of CiaRH knock-outs is still repressed when cells enter stationary phase (data not shown) suggests that other factor(s) are involved in competence repression during stationary phase.
Whatever the nature of the other factor(s), the mechanism of action of CiaR (direct or indirect) and the form of CiaR that is active in repression, these observations provide additional support for the hypothesis that basal level expression of the comCDE operon is under metabolic control (see also Claverys et al., 2000a). If it is confirmed that comCDE expression is adjusted by a complex network in response to metabolism, to growth conditions or, more generally, to environmental conditions (Claverys et al., 2000a), this will raise the question of whether the CSP, which has so far been regarded as a pheromone involved in quorum sensing, should not rather be considered as an alarmone, triggering transformation as an adaptive response of S. pneumoniae population via genetic plasticity (Claverys et al., 2000b).
Bacterial strains, growth conditions and competence
The bacterial strains and plasmids used are listed in Table 2. Microaerobiosis was obtained as described previously (Echenique et al., 2000). Competence profiles, which assess spontaneous transformability, were generated from cultures grown in C+Y medium as described previously (Alloing et al., 1996)). For inoculation, stocks of bacteria grown in CAT medium to an OD550 of 0.4 were diluted 20-fold in C+Y medium (except for the monitoring of cell density effects). For all experiments described here, stocks in CAT medium were prepared from freshly subcloned strains. Where indicated, C-maleate (the same as C+Y medium except that 0.05 M K2HPO4 was replaced by 0.05 M sodium maleate buffer, pH 7.6; Tomasz and Hotchkiss, 1964) was used instead of C+Y medium, or trypsin (2 µg ml−1) was added to C+Y medium, or the pH of the C+Y medium was adjusted to 7 (referred to as acid growth conditions). For CSP-induced competence, precompetent cells, prepared as described previously (Martin et al., 1995) but in THY (30 g l−1 Todd–Hewitt, 5 g l−1 yeast extract; Difco) instead of CTM 1, were incubated with synthetic CSP (25 ng ml−1) in CTM 2 (Martin et al., 1995) at 37°C for 10–15 min before the addition of DNA.
Cells were incubated at 30°C during DNA uptake. Transformants were determined by plating in 10 ml of D medium (Alloing et al., 1996), followed by challenge with a 10 ml overlay containing the appropriate antibiotic after phenotypic expression for 120 min at 37°C. E. coli strains were grown in LB medium. Antibiotic concentrations used for the selection of transformants were: chloramphenicol (Cm), 4.5 µg ml−1; erythromycin (Ery), 0.05–0.2 µg ml−1; kanamycin (Kan), 250 µg ml−1; spectinomycin (Spc), 100 µg ml−1; and streptomycin (Sm), 200 µg ml−1 for S. pneumoniae; ampicillin (Ap), 50 µg ml−1 and Spc, 100 µg ml−1 for E. coli.
For the monitoring of lacZ expression, cells were plated in a medium containing 50 µg ml−1 Xgal and described previously for in situ colony competence testing (Morrison et al., 1983). The complete medium (normally used for the middle layer but, in our case, without maltose and with 1 g of bacto agar 100 ml−1) is referred to here as T+ medium, whereas T− medium corresponds to the same medium from which CaCl2, NaOH and BSA have been omitted.
In vitro mariner mutagenesis
A minitransposon containing a gene conferring SpcR was constructed by substituting the cat gene (magellan2 minitransposon) of plasmid pEMcat for the spc gene from plasmid pR350 (Table 2). The gene was PCR amplified using the spcUP–spcDO primer pair (Table 2). The PCR fragment was mixed with MluI-digested pEMcat DNA, blunted using T4 DNA polymerase and ligated. The ligation mixture was transformed into E. coli strain DH5α. Plasmid pR412 was isolated from a SpcR transformant and used as a source of the 1146-bp-long spc mariner minitransposon. Transposition reactions were performed using purified Himar1 transposase as described previously (Lampe et al., 1996). Targets for transposition were S. pneumoniae strain R800 chromosomal DNA or PCR products. Gaps in transposition products were repaired as described previously (Akerley et al., 1998), and repaired transposition products were transformed into S. pneumoniae.
PCR reactions and characterization of mariner cassette–S. pneumoniae DNA junctions
PCR was performed using hot Tub DNA polymerase (Amersham), primers (1 µM) and 30 cycles of amplification (30 s denaturation at 94°C, 30 s annealing at 55°C and 5 min extension at 72°C). Ligation-mediated (LM)-PCR was used to characterize mariner cassette–S. pneumoniae DNA junctions (Palittapongarnpim et al., 1993). In short, the MP135–MP137 (Table 2) preannealed Sau3A linker was ligated to Sau3A-digested chromosomal DNA extracted from each strain harbouring a mariner insertion. PCR reactions were performed on ligated templates with the MP128–MP135 primer pair (Table 2). PCR products were purified with the QIAquick DNA Cleanup system (Qiagen) and sequenced using the ThermoSequenase cycle sequencing kit (USB) and primer MP128. Analysis of sequences was carried out using blast (Altschul et al., 1997) at The Institute for Genomic Research (http://www.tigr.org).
We thank A. Camilli for introducing one of us (M.P.) to mariner mutagenesis and for sharing with us plasmid pEMcat (magellan2) and the E. coli strain allowing mariner transposase production, S. Lacks for the gift of the trt1hex4noz1 strain, and D. Morrison for providing us with strains CP1200-Ω202 and CP1200-Ω210. We also thank The Institute for Genomic Research for granting access to the partially completed genome sequence of S. pneumoniae type 4 isolate. This research was financed by the Programme de Recherche Fondamentale en Microbiologie, Maladies Infectieuses et Parasitaires and by the European Union (grant BIO4-CT98-0424).
†The first two authors contributed equally to this study.