This work was almost entirely carried out during a sabbatical stay (August 2004–June 2005) of L.S.H. in Toulouse.
New insights into the pneumococcal fratricide: relationship to clumping and identification of a novel immunity factor
Article first published online: 20 DEC 2005
Volume 59, Issue 4, pages 1297–1037, February 2006
How to Cite
Håvarstein, L. S., Martin, B., Johnsborg, O., Granadel, C. and Claverys, J.-P. (2006), New insights into the pneumococcal fratricide: relationship to clumping and identification of a novel immunity factor. Molecular Microbiology, 59: 1297–1037. doi: 10.1111/j.1365-2958.2005.05021.x
- Issue published online: 20 DEC 2005
- Article first published online: 20 DEC 2005
- Accepted 28 November, 2005.
In 1971, Tomasz and Zanati discovered that competent pneumococci have a tendency to form aggregates when pelleted by centrifugation and resuspended in 0.01 N HCl by brief vortexing. Interestingly, no clumping was observed with parallel cultures of non-competent cells treated in the same way. We set out to elucidate the mechanism behind this striking phenomenon, and were able to show that it depends on extracellular DNA that is presumably released by so-called competence-induced cell lysis. Competence-induced cell lysis, which was first described a few years ago, seems to rely on the concerted action of several murein hydrolases. Our results confirmed and extended previous findings by showing that competence-induced aggregation is abolished in a lytA–lytC double mutant, and absolutely requires CbpD and its N-terminal CHAP amidase domain. Furthermore, we discovered a novel competence stimulating peptide (CSP)-induced immunity protein, encoded by the early competence gene comM (spr1762), which protects competent pneumococci against their own lysins. Together, the murein hydrolases and the immunity protein constitutes a CSP-controlled mechanism that allows competent pneumococci to commit fratricide by killing non-competent pneumococci sharing the same ecological niche. Through such predatory behaviour, pneumococci can get access to transforming DNA and nutrients, promote the release of virulence factors, and at the same time get rid of competitors.
Induction of natural genetic transformation in Streptococcus pneumoniae is controlled by a secreted 17 aa peptide pheromone called the competence stimulating peptide (CSP) (Håvarstein et al., 1995). Extracellular CSP is sensed by the polytopic membrane-embedded histidine kinase ComD (Håvarstein et al., 1996; Pestova et al., 1996). By analogy with other two-component regulatory systems, it is assumed that CSP binding stimulates ComD to autophosphorylate and transfer a phosphoryl group to its cognate response regulator ComE. Use of DNA microarrays has revealed that ComE, presumably in its phosphorylated state, activates transcription of about 20 early com genes. Among these are the two identical comX genes encoding an alternative σ factor, σX (Luo and Morrison, 2003), that directs transcription of a large number of so-called late com genes. So far, close to 105–124 CSP-responsive genes have been identified in S. pneumoniae (Dagkessamanskaia et al., 2004; Peterson et al., 2004). Curiously, only 22 of them have been shown to be required for natural transformation (Guiral et al., 2006). They encode proteins involved in the regulation of competence, and in the uptake and processing of homologous transforming DNA into recombinants.
Recently, the products of the late com genes cbpD (Guiral et al., 2005; Kausmally et al., 2005), lytA (Moscoso and Claverys, 2002; Steinmoen et al., 2002), cibA and cibB (Guiral et al., 2005), all of which are dispensable for natural transformation, have been implicated in competence-induced cell lysis. This CSP-controlled mechanism causes release of DNA from a subfraction of the cells in a culture of competent pneumococci (Moscoso and Claverys, 2002; 2004; Steinmoen et al., 2002; 2003). In liquid culture, disruption of the putative murein hydrolase cbpD has been shown to essentially abolish competence-induced cell lysis (Kausmally et al., 2005). Disruption of the gene encoding the autolysin LytA (García et al., 1986) was reported to affect lysis to various extent (Moscoso and Claverys, 2002; 2004; Steinmoen et al., 2002; 2003; Guiral et al., 2005), whereas the simultaneous inactivation of lytA and lytC (which encodes an autolytic lysozyme; García et al., 1999) abolished both DNA release in liquid culture (Moscoso and Claverys, 2004) and pneumolysin release within agar plates (Guiral et al., 2005). Efficient lysis within blood agar plates also depends on the bacteriocin-like peptides CibA and CibB, whereas the contribution of CbpD seems to be less important (Guiral et al., 2005). In the latter case, the competent subpopulation which triggers the lysis of non-competent cells was shown to protect itself from the lytic process via CibC (Guiral et al., 2005). Several biological roles have been proposed for the selective lysis of siblings, also termed microbial fratricide (Gilmore and Haas, 2005). Pneumococcal fratricide could provide access to nutrients, increase the efficiency of gene exchange between different pneumococcal strains through the release of chromosomal DNA, exacerbate an infection by releasing virulence traits and inflammatory mediators, or stabilize the relationship with its host by activating host innate defences (Gilmore and Haas, 2005; Guiral et al., 2005).
Tomasz and Zanati (1971) demonstrated more than three decades ago that competent pneumococcal cells pelleted by centrifugation and resuspended in 0.01 N HCl display strong and immediate clumping. Interestingly, this remarkable effect was shown to be competence-dependent, and did not take place with non-competent cells (Tomasz and Zanati, 1971). The authors proposed that the observed aggregation resulted from the appearance of a trypsin-sensitive agglutinin on the surface of competent cells. However, they did not pursue this idea, and the mechanism behind the clumping reaction has therefore remained unknown. In this paper, we present evidence that aggregation requires a mixture of competent and non-competent cells and relies on the release of DNA into the medium. We used the clumping assay to demonstrate the requirement for LytA or LytC and CbpD, and to establish that this competence-induced phenomenon requires the CHAP (cysteine, histidine-dependent amidohydrolase/peptidase) domain of CbpD. Finally, we show that protection of competent cells against their own lytic weaponry relies on a competence-regulated immunity function, encoded by the early gene comM (spr1762).
Clumping requires a mixture of competent and non-competent pneumococci
The clumping assay was carried out essentially as previously described (Tomasz and Zanati, 1971), except that the cells were resuspended in citrate/phosphate buffer pH 3.5 (referred to as ‘acid’, hereafter) instead of 0.01 N HCl. The initial experiments were carried out with the R704 (comA) strain, which lacks the CSP-specific ABC-transporter ComA. Consequently, a culture of this strain will not develop the competent state spontaneously, but relies on exogenously added synthetic CSP for competence induction. When vigorously growing bacteria of this strain were subjected to synthetic CSP (150 ng ml−1) for about 12 min and then pelleted by centrifugation, they aggregated upon resuspension in acid (data not shown). Surprisingly, this clumping effect was essentially lost when one of the central components of the C + Y medium, the casein hydrolysate (from Qbiogen, CQ hereafter), was exchanged with the corresponding product from another company (Merck, CM hereafter). We suspected that this change had affected competence development in the culture, and therefore decided to compare the level of competence reached in CQ- and in CM-based medium by exploiting a strain (R1521) containing the luciferase reporter gene under control of an early com gene promoter (comC::luc). Upon CSP treatment, the pneumococci grown in CM-based medium emitted twice as much light as those grown in CQ-based medium (data not shown). As a direct relationship between luciferase activity and transformation frequency was observed under widely varying levels of competence (Bergéet al., 2002), this result suggested that the number of competent cells in the new medium had increased by a factor of two. We therefore hypothesized that not all pneumococci grown in the CQ-based medium became competent even when subjected to 150 ng ml−1 CSP and, consequently, that aggregation might require a mixture of competent and non-competent cells in the culture. To test this hypothesis, we grew the CSP-inducible R391 (comA) and the competence-deficient R631 (comE) strains in CM-based medium at 37°C to an OD550∼0.15, and mixed them in equal amounts just before addition of CSP. Separate cultures of R391 and R631 were run in parallel as controls. The latter strain lacks the key transcriptional activator ComE, and is therefore unable to develop competence even in the presence of CSP. No clumping whatsoever was obtained with pure cultures of R391 or R631, whereas strong clumping was observed with the mixed culture (Fig. 1, top). The observed clumping depended on addition of CSP, as a parallel control culture without CSP did not display any clumping (Fig. 1, top). These results, which were highly reproducible, show that in a culture containing both competent and non-competent pneumococci, the competent cell fraction interacts with the non-competent population in a way that gives rise to aggregates when the pelleted cells are resuspended in acid.
Clumping requires the presence of DNA in the medium
The kinetics of the clumping reaction was investigated by exposing mixed cultures, consisting of approximately equal numbers of R704 (comA) and R484 (comE) cells, to CSP for periods ranging from 3 to 24 min before pelleting the cells and resuspending them in acid. The results showed that clumping is transient, reaching a maximum after about 10 min under the conditions used (Fig. 1, bottom). We also noticed that upon resuspension in acid, the cells occasionally stuck together in a large slimy clump. This observation suggested that polysaccharides or long DNA strands could participate in clumping. We favoured the latter hypothesis because DNA release is known to occur in competent cultures of pneumococci (Steinmoen et al., 2002; Moscoso and Claverys, 2004). To test this idea, DNase I was added to cultures immediately after CSP induction to see if this would influence clumping. Clumping was reduced by 80% in the presence of 20 Kunitz units per ml of DNase I, and completely abolished when DNase I concentration was raised to 50 Kunitz units per ml (Fig. 2), strongly indicating that clumping requires the presence of DNA in the medium.
In view of the involvement of DNA in the observed aggregation, we hypothesized that the transient nature of the phenomenon could be due to nuclease activity. We therefore ran a clumping reaction in the presence of 5 mM EDTA to inhibit nucleases. Clumping was no longer transient but appeared to be permanent (Fig. 1, bottom). Obviously, this effect of EDTA could be interpreted in several different ways. However, in the context of the DNase I experiment, the stabilizing effect of EDTA suggests that the transient nature of the observed aggregation could result from a nuclease activity degrading (or more likely reducing the length of; see Discussion) extracellular DNA.
An attractive hypothesis would be that aggregation results from attachment of competent pneumococci to released DNA through their DNA-uptake apparatus. However, this does not appear to be the case, as normal clumping was observed with a strain lacking ComGA (strain R1741), a protein previously shown to be essential for DNA binding and uptake (Bergéet al., 2002), or ComGB (strain R1918). The endA strain R1049, a mutant known to accumulate DNA at the surface when it is competent, because it fails to process and import bound DNA (Bergéet al., 2002), displayed normal clumping as well (data not shown). The observation of a normal clumping kinetics in the absence of the major pneumococcal endonuclease EndA ruled out its possible contribution to the transient nature of the clumping phenomenon.
An alternative to the active binding of DNA by the DNA-uptake machinery is that genomic DNA (presumably long stranded; see Discussion) somehow glue the cells together under the acidic conditions used.
Clumping requires LytA, LytC and CbpD
Recent studies have established that the release of DNA in competent cultures of S. pneumoniae requires cell lysis. Competence-induced cell lysis in liquid culture is virtually abolished in a lytA–lytC double mutant (Moscoso and Claverys, 2004) as well as in a cbpD mutant (Kausmally et al., 2005), whereas inactivation of lytA was reported to reduce release of DNA and of a cytoplasmic protein (β-galactosidase) about 10-fold (Steinmoen et al., 2002) and two to fourfold (Steinmoen et al., 2003) respectively. Within blood agar plates, CbpD, LytA, LytC and the two-peptide bacteriocin, CibAB, were all found to be involved in competence-induced lytic events (Guiral et al., 2005). Lysis (of non-competent cells) was virtually abolished in a lytA–lytC double mutant as well as in a cibA (or cibB) mutant (Guiral et al., 2005).
We therefore tested the effect of cbpD, cibA, lytA and lytC mutations to investigate their possible involvement in clumping. Disruption of the cbpD gene (strain R1924) abolished clumping in a standard mixed assay with the R484 strain. In contrast, a positive control, consisting of equal amounts of the parental R704 strain and strain R484, displayed strong clumping. This result was highly reproducible, demonstrating that the observed clumping reaction depends on CbpD (data not shown). Similarly, while disruption of only lytA did not prevent aggregation, no clumping was observed when LytA and LytC were both missing, which indicated that clumping requires the presence of either LytA or LytC in addition to CbpD. Interestingly, abolition of clumping in the mixed assay required the simultaneous inactivation of lytA and lytC in both the CSP-inducible and the competence-deficient strains (mixed assay with strains R1721 and R1355; data not shown), a situation reminiscent of the pneumococcal fratricide which readily occurred even when LytA and LytC were provided only by the competent cells or the targeted cells (Guiral et al., 2005). Disruption of cibB had no detectable effect on clumping (mixed assay with strains R1914 and R1839; data not shown), suggesting that the two-peptide bacteriocin is not required.
The involvement of the major autolysin LytA or the lysozyme LytC, and of the putative murein hydrolase CbpD strongly suggests that clumping depends on cell lysis and DNA release. The observation of residual clumping when only the competence-deficient cells harboured intact lytA and lytC genes is most consistent with CbpD being a key element to initiate the cascade of events required for clumping.
Clumping requires a functional CHAP domain in CbpD
CbpD consists of three types of domains; an N-terminal CHAP domain, two SH3-like domains, and a C-terminal choline binding domain with four choline binding repeats (Anantharaman and Aravind, 2003; Bateman and Rawlings, 2003; Rigden et al., 2003). Several members of the CHAP family of enzymes are known to act as amidases attacking the bacterial cell wall (Bateman and Rawlings, 2003; Rigden et al., 2003). We therefore hypothesized that the amidase function of CbpD was required for clumping and wished to determine if the active site cysteine in its CHAP domain was essential. The cysteine residue was changed into alanine by transforming with a recombinant plasmid carrying a cbpD fragment containing the desired mutation (see Experimental procedures). The resulting CbpDC75A mutant strain (R1902) was tested in the clumping assay together with the competence-deficient strain R484. No clumping was observed, demonstrating that the enzymatic activity of the CHAP domain is required for clumping and suggesting that the lytic properties of CbpD is essential for competence-induced cell lysis.
Competence-induced immunity against clumping
As a pneumococcal population consisting only of competence-proficient cells does not display any detectable clumping, competent cells must be less susceptible to lysis than non-competent cells. We therefore speculated that competent cells express immunity against their own lysins, and that this immunity function depends on a CSP-responsive gene. To check whether the hypothetical immunity protein is encoded by an early or a late com gene, we took advantage of a comX mutant strain, R1128 (comA, comX1, comX2; Table 1). As this strain can only express the early com genes in response to CSP, it could become immune to clumping only if the immunity gene belongs to the early class.
|EK4253||comM::pEVP3 insertion-duplication mutant (229 bp internal comM fragment cloned into pEVP3)||Knutsen et al. (2004)|
|R6||Non-capsulated D39 derivative||Laboratory stock|
|R391||R800 but comA::kanC; KanR||Guiral et al. (2005b)|
|R484||R800 but trt1, comE::kan102A; KanR||Laboratory stock|
|R631||R800 but malM::gfp, comE::kan102A; CmR, KanR||Laboratory stock|
|R704||R800 but comA::ermAMC; EryR||Laboratory stock|
|R800||R6 derivative||Martin et al. (1985)|
|R895||R800 but ssbB::luc (pR424), ssbB+ CmR||Chastanet et al. (2001)|
|R1049||R895 but comA::ermAMA, endA::kan6C; CmR, EryR, KanR||Bergéet al. (2002)|
|R1128||R391 but comX1::ermAM, comX2′::(pEVP3)::′comX2 (using as donor chromosomal DNA from strain CPM4; Lee and Morrison, 1999); KanR, EryR, CmR||Laboratory stock|
|R1278||R800 but ssbB::luc (pR424), ssbB+, hexA-Δ3::ermAM, comA::kanC; CmR, EryR, KanR||Bergéet al. (2003)|
|R1355||R800 but lytC::tet, lytA::cat4C, comE::kan102A; CmR, KanR, TcR||Guiral et al. (2005b)|
|R1502||R800 but ΔcomC, ssbB::luc (pR424), ssbB+ CmR||Dagkessamanskaia et al. (2004)|
|R1521||R800 but ΔcomC, comC::luc (pR414); EryR||Laboratory stock|
|R1721||R1502 but lytA::kan1A, lytC::tet; CmR, KanR, TcR||Laboratory stock|
|R1741||R1502 but comGA::spc3C (from strain R1062; Bergéet al., 2002); CmR, SpcR||This study|
|R1839||R1502 but comE::spc2C; CmR, SpcR||This study|
|R1865||R704 but comM::spc28A; EryR, SpcR||This study|
|R1866||R704 but comM::spc1C; EryR, SpcR||This study|
|R1868||R704 but spr1760::spc19C; EryR, SpcR||This study|
|R1879||R1502 but comM::spc1C; CmR, SpcR||This study|
|R1902||R1278 but cpbDC75A; CmR, EryR, KanR||This study|
|R1914||R1502 but cibB::kan9C; CmR, KanR||This study|
|R1915||R704 but comM::pEVP3 (by transformation with EK4253 chromosomal DNA); CmR, EryR||This study|
|R1918||R1502 but comGB::kan4C; CmR, KanR||This study|
|R1924||R704 but ΔcbpD::kan (by transformation with a PCR fragment amplified from the L3 strain using cbpD1 and cbpD4 as primers; see Kausmally et al., 2005); EryR, KanR||This study|
|LE392||F– e14– (McrA–) hsdR514 (rK–mK+) glnV44 supF58 lacY1 galK2 galT22 metB1 trpR55||Sambrook et al. (1989)|
|pGBDU-C1||ColE1 derivative; ApR||U70021 (NCBI nucleotide)|
|pR412||ColE1 derivative carrying a SpcRmariner minitransposon; ApR, SpcR||Prudhomme et al. (2005)|
|pR424||ColE1 derivative carrying an S. pneumoniae 5′-ssbB targeting fragment adjacent to luc; CmR||Chastanet et al. (2001)|
|pR451||pGBDU-C1 carrying the cbpDC75A mutagenic fragment (see Experimentalprocedures); ApR||This study|
We designed an experiment in which R1128 cells had been treated with CSP for various periods of time (2–21 min) and looked for induction of immunity against clumping by mixing them with R704 cells that had been subjected to CSP for about 8 min, i.e. R704 cells at about their maximum clumping capacity. The results clearly showed that R1128 cells that have been subjected to CSP for ∼9 min or more are fully immune (Fig. 3). Strong clumping, however, was observed in the mixed samples where the R1128 cells had been subjected to CSP for less than 3–4 min. This experiment proves that comX competent cells develop immunity against clumping, and therefore that the immunity function resides among the early com genes.
Fortunately, microarray analyses have shown that there are only about 20 early genes in the S. pneumoniae genome. Around half of these have unknown functions, and were therefore considered to be possible immunity proteins. The largest early com operon, consisting of four open reading frames (spr1762–spr1759) with no assigned function (Dagkessamanskaia et al., 2004; Peterson et al., 2004), was selected as our initial target for gene disruption mutagenesis. The first gene of this operon (spr1762) potentially encodes an integral membrane protein previously termed ComM (Knutsen et al., 2004). The comM gene of strain R704 was disrupted by integration of the pEVP3 plasmid (Table 1), an insertion that presumably exerts a polar effect on the three downstream genes in the operon (Fig. 4). The immunity of the resulting strain (R1915) towards competence-induced cell lysis was then tested in a standard clumping assay with strain R704. The results showed that transient clumping took place, and that the kinetics followed a pattern almost identical to the one previously obtained with mixed cultures of the R704 and R484 strains (Fig. 5). It is clear from these data that R1915 competent cells have become susceptible to clumping, and that one or more of the proteins encoded by the comM operon transcript are responsible for this loss of immunity.
As described above, a pure culture of strain R704 will not display any clumping when induced to competence. In contrast, a culture of strain R1915, which has lost immunity, should form aggregates on its own. This is indeed what happens when a clumping assay is run with the R1915 strain. Very strong clumping was observed, demonstrating that competent cells of this mutant were no longer protected from clumping (Fig. 5).
ComM is the immunity protein
Due to the probable polar effect of the pEVP3 insertion in comM on downstream gene expression, it was not possible to pinpoint the gene(s) responsible for immunity. In an attempt to disrupt each gene separately, in vitro mariner mutagenesis was carried out on the comM chromosomal region (Fig. 4 and Experimental procedures). Unexpectedly, no insertions were obtained in spr1761 and spr1759, strongly suggesting that disruption of these genes is lethal. The presence of an extended −10 (i.e. σA-dependent) promoter (Pc in Fig. 4) between comM and spr1761 suggests that spr1761–spr1759 constitute an operon constitutively expressed separately of comM. This would readily explain why the insertion of pEVP3 in comM is tolerated despite its probable polar effect on expression of spr1761 and spr1759. Insertions, however, were obtained in the spr1760 gene (Fig. 4), which is consistent with its previous classification as nonessential (Thanassi et al., 2002). The spr1760::spc19C insertion mutant (R1868) was still immune when tested in a pure culture clumping assay, demonstrating that the spr1760 gene is not involved in immunity. Several mariner insertions were obtained in the comM gene (Fig. 4). Two of them, one in each orientation (spc28A, strain R1866; spc1C, strain R1865), were characterized in pure culture clumping assays, and both displayed loss of immunity (data not shown). Together, these data show that ComM constitutes the immunity protein that protects pneumococci against their own lysins during competence development. The alternative explanation that CSP-treated cultures of the comM mutant naturally consist of a mixture of competent and non-competent cells because the comM mutation reduces competence (a formal possibility not previously excluded) was ruled out by demonstrating similar expression of the ssbB::luc reporter in both the comM mutant R1879 and its parent R1502 (data not shown).
The effect of disrupting the comM gene can be detected just by comparing the growth curves of the wild type and the comM mutant before and after induction of competence (Fig. 6). In the wild-type strain, addition of CSP results in a small but distinct reduction in growth rate compared with the non-competent culture (Fig. 6, top). The reason for this is unknown. We tentatively attribute it to the fact that expression of ComX in competent cells causes a global shift in the gene expression pattern, with a temporally reduced expression of housekeeping genes. In contrast, a drop in optical density is observed between 9 and 10 min after CSP addition in the comM mutant (Fig. 6, bottom). This drop can only be explained by cell lysis resulting from lack of immunity towards components of the lytic machinery, CbpD and/or LytA–LytC.
The strong and immediate clumping of competent pneumococcal cells resuspended in 0.01 N HCl was tentatively attributed to the appearance of an agglutinin on the surface of competent cells (Tomasz and Zanati, 1971). Our reinvestigation of this phenomenon allowed us to establish that the clumping reaction relies in fact on the presence of DNA in the medium. We also established that clumping requires the presence of a mixture of competent and non-competent cells. The simplest interpretation of these data is that the competent fraction interacts with the non-competent population so as to trigger the release of chromosomal DNA from the latter. The finding that clumping requires the presence of the major autolysin LytA or the autolytic lysozyme LytC, and a functional amidase (CHAP) domain in CbpD demonstrates the involvement of the pneumococcal lytic machinery in the process. The clumping reaction therefore most likely results from the lysis of non-competent cells triggered by competent cells, a phenomenon reminiscent of pneumococcal fratricide. Although it has not been formally demonstrated here that the extracellular DNA required for clumping comes from the non-competent cells alone, the finding that the competent cells develop immunity towards clumping is consistent with this interpretation. The additional observation that inactivation of the early gene comM not only abolishes immunity but also allows strong clumping of pure competent cultures of the mutant is fully consistent with this view. Altogether, these data suggest that the presence of ComM somehow renders competent cells resistant to the combined lytic action of CbpD and LytA–LytC. This interpretation is reinforced by the observation of a drop in optical density following competence induction in the comM mutant strain, a finding that is most consistent with lysis of comM competent cells.
The clumping phenomenon is related to pneumococcal fratricide through several aspects including the requirement for a mixture of competent and non-competent cells, the involvement of CbpD, LytA and LytC, and the release of DNA. However, it also differs from that regarding the relative importance of CibAB and CbpD. Allolysis, the lysis of non-competent pneumococci triggered by competent pneumococci, depends on the two-peptide bacteriocin CibAB, whereas the contribution of CbpD seems to be less important (Guiral et al., 2005). On the other hand, LytA and LytC from the targeted cells have been implicated in both clumping and allolysis. We suggest that CbpD triggers lytic events in liquid culture whereas cell immobilization and cell density within plates favour the transfer of CibAB through cell–cell contacts. CibAB could thus substitute for CbpD to activate LytA and LytC in the latter case.
Another apparent difference between clumping and pneumococcal fratricide resides in the kinetics. Clumping is quite fast compared with the previously reported kinetics of lysis (Steinmoen et al., 2002; Moscoso and Claverys, 2004). It is somewhat puzzling that maximum clumping occurs about 10 min after CSP addition, when only 15–20% of total β-galactosidase release (used as a surrogate for DNA) has occurred (Steinmoen et al., 2002). A possible explanation to the paradox that clumping disappears while efficient DNA release continues could be that clumping requires really long DNA strands. As the assays for monitoring DNA release through transformation (Steinmoen et al., 2002; Moscoso and Claverys, 2004) as well as by quantitative polymerase chain reaction (PCR) (Kausmally et al., 2005) involved a sterile filtration step, it is possible that long DNA strands remained attached to the lysing cells and were therefore not detected, leading to a general underestimate of release. Few cuts introduced by low amounts of active nuclease(s) could then destroy long DNA strands and abolish the clumping reaction, whereas shorter pieces of DNA (but with a length still compatible with detection by transformation or PCR) could persist and/or accumulate through continuing lysis. This interpretation would be consistent with the effect of EDTA, which could stabilize clumping by inhibiting a nuclease. The appearance of a nuclease activity, which differs from EndA, at the surface of competent cells or in the supernatant of competent cultures (Moscoso and Claverys, 2004) would also fit with this interpretation.
Concerning the mechanism of clumping, it is unclear why extracellular DNA pelleted with living bacteria and bacterial debris gives rise to clumps when redissolved in acid. The observation that inactivation of comGA or comGB, which abolish DNA binding and uptake, did not affect clumping led us to reject the hypothesis that clumping results from the active binding of DNA by the DNA-uptake machinery. An alternative hypothesis, which does not involve a specific DNA receptor, would be that long DNA strands create a sort of fishing net or glue the cells together under the acidic conditions used. In any case, a brief cell–cell contact between the competent killer and the non-competent target cells would seem necessary to trigger the events leading to clumping. However, it remains unknown whether this contact occurs by mere chance or whether it involves a specific receptor. Our finding that clumping relies on the presence of extracellular DNA does not rule out the hypothesis that competent pneumococci synthesize a new protein agglutinin (a receptor?) as previously proposed (Tomasz and Zanati, 1971) but it merely provides no support for it. The existence of an agglutinin was postulated because agglutination was found to be trypsin sensitive. Our observation that CbpD and LytA or LytC are required for clumping could readily account for the sensitivity to trypsin.
It has been pointed out that it is puzzling that competence for genetic transformation is regulated by a cell–cell communication mechanism that resembles quorum-sensing (Håvarstein and Morrison, 1999). Most of the quorum-sensing systems co-ordinate the release of products whose effectiveness for the producer would be compromised by dilution, i.e. products that require mass action. Considering that the proposed major roles for transformation (nutrition, repair or gene transfer) were more difficult to reconcile with regulation of competence through a quorum-sensing mechanism, it was suggested that this mechanism could provide a means for sensing potential gene donors of the same species (Håvarstein and Morrison, 1999). However, CSP could function as a sensor of donors only if CSP producers (i.e. competent cells) release DNA. Recent evidence that a CSP-controlled lysis mechanism operating during competence is directed against non-competent cells and our identification of a novel immunity factor protecting competent cells from lytic events provide no support to this theory. These observations rather suggest that CSP-controlled fratricide of non-competent cells is primarily a predatory mechanism that has evolved as a weapon to kill off close relatives competing for the same resources. In this process, cell–cell communication could be important to favour protection of competence-proficient cells. In addition, synchronization of competence could ensure a maximum efficacy of lytic mechanisms.
A previous suggestion that DNA released through CSP-induced cell lysis constitutes a source of transforming DNA in nature (Moscoso and Claverys, 2002; 2004; Steinmoen et al., 2002) is nevertheless not ruled out by the recent findings. By promoting the lysis of non-competent cells, the predators not only get rid of potential competitors but also can benefit from various compounds released by the lysed cells (see Introduction). Considering only DNA release, the co-regulation of fratricide and DNA uptake might indicate that a coupling of these processes is beneficial for the bacterium. The benefit could lie in the use of incoming DNA as a source of nucleotides (although in limited amounts; see Guiral et al., 2006), for recombinational repair of the recipient's genome and/or chromosomal transformation, the latter contributing to increased genetic diversity and the acquisition of novel traits.
Bacterial strains, growth and transformation conditions
Streptococcus pneumoniae strains and plasmids used in this study are described in Table 1, whereas the primers are listed in Table S1. Precultures were grown at 37°C in CAT medium (Martin et al., 2000) to OD550 = 0.3; after centrifugation, cells were concentrated to OD550 = 0.4 in fresh CAT medium containing 15% glycerol and kept frozen at −70°C. Transformation was performed as described previously (Martin et al., 2000), using pre-competent cells treated at 37°C for 10 min with synthetic CSP-1 to induce competence. Unless otherwise indicated, synthetic CSP-1 was used at 150 ng ml−1. After addition of transforming DNA, cells were incubated for 20 min at 30°C. Transformants were selected by plating in 10 ml of CAT agar supplemented with 4% horse blood, followed by challenge with a 10 ml of overlay containing chloramphenicol (9 µg ml−1), erythromycin (0.1 µg ml−1), kanamycin (500 µg ml−1) or streptomycin (400 µg ml−1), after phenotypic expression for 120 min at 37°C.
The clumping assay was performed essentially as described by Tomasz and Zanati (1971). Pneumococcal strains were grown in C + Y medium (Martin et al., 2000) at 37°C. When vigorously growing cultures (generation time approximately 35 min) reached OD550 = 0.15–0.2, samples of 10 ml (5 + 5 ml in the case of mixed cultures) were withdrawn and added to test tubes containing 150 ng ml−1 CSP-1. Samples without CSP-1 were run in parallel as a negative control if necessary. After being subjected to CSP-1 for 3–24 min (kinetics assay) at 37°C, the tubes were immediately centrifuged for 2 min at 5000 rpm in a pre-cooled Beckman rotor JA-25.50. After centrifugation, the tubes were promptly put on ice. Then, working as quickly as possible, the supernatants were discarded and the emptied tubes were put back on ice. To detect possible aggregation, 2 ml of 0.1 M citric acid/phosphate (Na2HPO4) buffer pH 3.5 was added, followed by immediate vortexing for about 3 s. After vortexing, the tubes were placed in a rack for 20–30 min to let possible aggregates precipitate to the bottom of the tubes. Then, the top layer of the cell suspension (1 ml; Fig. 1, top) was carefully removed without disturbing the precipitate. Finally, OD550 of the collected samples were determined. Unless otherwise indicated, throughout the manuscript ‘clumping assay’ refers to a mixed culture experiment (e.g. equal amounts of strain R704 and strain R484; Fig. 1).
Construction of CbpDC75A, a CHAP domain mutant of CbpD
To determine if the active site cysteine (position 75 in the sequence RQCTS) in the CHAP domain of CbpD was essential for clumping, this cysteine was changed into alanine as follows. Two PCR fragments overlapping the CHAP domain were first generated with primer pairs cpbD1-cbpD4 and cbpD3-cbpD2 (Table S1) using R800 chromosomal DNA as template and the Phusion™ High-Fidelity DNA Polymerase (Finnzymes). Primers cbpD3 and cbpD4 are complementary and incorporate nucleotide substitutions (TGT to GCG; Table S1) corresponding to codon 75 in the cbpD gene potentially changing the cysteine into an alanine to generate CbpDC75A. After purification with a QIAquick nucleotide removal kit (Qiagen), both fragments (respectively 883 bp and 846 bp in length) were mixed and used as template to amplify a unique cbpD1-cbpD2 fragment. This fragment was digested with EcoRI and BamHI, and cloned into EcoRI–BamHI digested plasmid pGBDU-C1 (Table 1). As the nucleotide changes introduced in the cbpD segment were expected to eliminate an RsaI site (cagTGTact changed to cagGCGact), RsaI digestion of PCR fragments amplified from Escherichia coli ApR transformants with the cbpD1-cbpD2 primer pair was used as a diagnostic. The cloned fragment of an RsaI-resistant transformant was entirely sequenced to retain plasmid pR451. Then, pR451 plasmid DNA was used to transform the S. pneumoniae strain R1278. After allowing a few hours for phenotypic expression and cell segregation in liquid culture without selection, aliquots of the transformed culture were plated and 10 independent colonies were isolated. RsaI digestion of PCR fragments amplified from individual colonies was similarly used to identify candidate CbpDC75A mutants. Strain R1902 was retained after sequencing the entire cbpD segment.
A 3.8 kb PCR fragment encompassing the comM and spr1761–spr1759 genes was amplified from R800 chromosomal DNA with primers plcR1 and cinA1 (Table S1), and mutagenized in vitro with the mariner transposase as described previously (Prudhomme et al., 2006), using plasmid pR412 as donor of the SpcRmariner minitransposon. The resulting transposition products were used to transform S. pneumoniae strain R800. Each SpcR transformant was subcloned, and the exact site of insertion of the minitransposon in the recipient's genome was determined as described (Prudhomme et al., 2006). Selected insertions were transformed into strain R704 using as donor PCR fragments amplified with the appropriate primer pair, 1762F-1762R (internal to comM) or 1760F-1760R (internal to spr1760), to avoid the presence of point mutation(s) in the flanking regions.
L.S.H. benefited from a position de Professeur Invité from the Université Paul Sabatier and received support from the French Ministère de la Jeunesse, de l’Education Nationale et de la Recherche. This work was financed in part by European Union Grant QLK2-CT-2000-00543.
- 2003) Evolutionary history, structural features and biochemical diversity of the NlpC/P60 superfamily of enzymes. Genome Biol 4: R11. , (
- 2003) The CHAP domain: a large family of amidases including GSP amidase and peptidoglycan hydrolases. Trends Biochem Sci 28: 234–237. , (
- 2002) Uptake of transforming DNA in Gram-positive bacteria: a view from Streptococcus pneumoniae. Mol Microbiol 45: 411–421. , , , , (
- 2003) Transformation of Streptococcus pneumoniae relies on DprA- and RecA-dependent protection of incoming single strands. Mol Microbiol 50: 527–536. , , , (
- 2001) Regulation of Streptococcus pneumoniae clp genes and their role in competence development and stress survival. J Bacteriol 183: 7295–7307. , , , (
- 2004) Interconnection of competence, stress and CiaR regulons in Streptococcus pneumoniae: competence triggers stationary phase autolysis of ciaR mutant cells. Mol Microbiol 51: 1071–1086. , , , , , , et al. (
- 1986) Nucleotide sequence and expression of the pneumococcal autolysin gene from its own promoter in Escherichia coli. Gene 43: 265–272. , , , (
- 1999) The molecular characterization of the first autolytic lysozyme of Streptococcus pneumoniae reveals evolutionary mobile domains. Mol Microbiol 33: 128–138. , , , , (
- 2005) The selective advantage of microbial fratricide. Proc Natl Acad Sci USA 102: 8401–8402. , (
- 2005) Competence-programmed predation of noncompetent cells in the human pathogen Streptococcus pneumoniae: genetic requirements. Proc Natl Acad Sci USA 102: 8710–8715. , , , (
- 2006) Competence in Streptococcus pneumoniae: what is it for? In Molecular Biology of Streptococci. Hakenbeck, R. and Chhatwal, G.S. (eds). Norwich: Horizon Scientific Press, in press. , , , (
- 1999) Quorum sensing and peptide pheromones in streptococcal competence for genetic transformation. In Cell–cell Signaling in Bacteria. Dunny, G.M., Winans, S.C. (eds). Washington, DC: American Society for Microbiology Press, pp. 9–26. , (
- 1995) An unmodified heptadecapeptide pheromone induces competence for genetic transformation in Streptococcus pneumoniae. Proc Natl Acad Sci USA 92: 11140–11144. , , (
- 1996) Identification of the streptococcal competence-pheromone receptor. Mol Microbiol 21: 865–871. , , , (
- 2005) Choline-binding protein D (CbpD) in Streptococcus pneumoniae is essential for competence-induced cell lysis. J Bacteriol 187: 4338–4345. , , , , (
- 2004) Two separate quorum-sensing systems upregulate transcription of the same ABC transporter in Streptococcus pneumoniae. J Bacteriol 186: 3078–3085. , , (
- 1999) Identification of a new regulator in Streptococcus pneumoniae linking quorum sensing to competence for genetic transformation. J Bacteriol 181: 5004–5016. , (
- 2003) Transient association of an alternative sigma factor, ComX, with RNA polymerase during the period of competence for genetic transformation in Streptococcus pneumoniae. J Bacteriol 185: 349–358. , (
- 1985) Cloning of the hexA mismatch repair of Streptococcus pneumoniae and identification of the product. Gene 34: 293–303. , , (
- 2000) Cross-regulation of competence pheromone production and export in the early control of transformation in Streptococcus pneumoniae. Mol Microbiol 38: 867–878. , , , , (
- 2002) Spontaneous release of DNA and induction of competence in cultures of Streptococcus pneumoniae. 6th European Meeting on the Molecular Biology of the Pneumococcus. Siena, Italy. URL http://www.unisi.it/eventi/europneumo/abstract.doc. , (
- 2004) Release of DNA into the medium by competent Streptococcus pneumoniae: kinetics, mechanism and stabiliy of the liberated DNA. Mol Microbiol 54: 783–794. , (
- 1996) Regulation of competence for genetic transformation in Streptococcus pneumoniae by an auto-induced peptide pheromone and a two-component regulatory system. Mol Microbiol 21: 853–864. , , (
- 2004) Identification of competence pheromone responsive genes in Streptococcus pneumoniae. Mol Microbiol 51: 1051–1070. , , , , , , et al. (
- 2006) In vitro mariner mutagenesis of Streptococcus pneumoniae: tools and traps. In The Molecular Biology of Streptococci. Hakenbeck, R. and Chhatwal, G.S. (eds). Norwich: Horizon Scientific Press, in press. , , (
- 2003) Amidase domains from bacterial and phage autolysins define a family of γ-D,1-glutamate-specific amidohyrolases. Trends Biochem Sci 28: 230–234. , , (
- 1989) Molecular Cloning: A Laboratory Manual, 2nd edn. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press. , , (
- 2002) Induction of natural competence in Streptococcus pneumoniae triggers lysis and DNA release from a subfraction of the cell population. Proc Natl Acad Sci USA 99: 7681–7686. , , (
- 2003) Competence-induced cells of Streptococcus pneumoniae lyse competence-deficient cells of the same strain during cocultivation. J Bacteriol 185: 7176–7183. , , (
- 2002) Identification of 113 conserved essential genes using a high-throughput gene disruption system in Streptococcus pneumoniae. Nucleic Acids Res 30: 3152–3162. , , , , (
- 1971) Appearance of a protein ‘agglutinin’ on the spheroplast membrane of pneumococci during induction of competence. J Bacteriol 105: 1213–1215. , (