Release of DNA into the medium by competent Streptococcus pneumoniae: kinetics, mechanism and stability of the liberated DNA

Authors

  • Miriam Moscoso,

    1. Laboratoire de Microbiologie et Génétique Moléculaires, UMR 5100 CNRS-Université Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex, France.
    Search for more papers by this author
    • Present address: Centro de Investigaciones Biológicas (CIB-CSIC), Campus Universidad Complutense, Ramiro de Maeztu, 9, 28040 Madrid, Spain.

  • Jean-Pierre Claverys

    Corresponding author
    1. Laboratoire de Microbiologie et Génétique Moléculaires, UMR 5100 CNRS-Université Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex, France.
    Search for more papers by this author

Summary

The release of chromosomal DNA into culture media has been reported for several naturally transformable bacterial species, but a direct link between competence development and the liberation of DNA is generally lacking. Based on the analysis of strains with mutations in competence-regulatory genes and the use of conditions favouring or preventing competence, we provide evidence that DNA release is triggered by the induction of competence in Streptococcus pneumoniae. Kinetic analyses revealed that whereas competence was maximal 20 min after addition of competence-stimulating peptide, and then decreased, the amount of liberated DNA continued to increase and reached a maximum in stationary phase, when cells are no longer competent for DNA uptake. These data are not consistent with the proposal that release of DNA by a fraction of the population is coordinated with uptake by the remainder. Moreover, we observed that an unidentified DNase was specifically induced or released in competent cultures, and that together with the major pneumococcal endonuclease, EndA, it could degrade released DNA. Nearby complete abolition of release in a mutant lacking both the major autolysin, LytA, and the autolytic lysozyme, LytC, indicated that DNA liberation occurs by LytA-LytC-dependent cell lysis. These observations suggest that competence-dependent DNA release is one facet of a more general phenomenon of sensitization to autolysis that reaches its maximum in stationary phase.

Introduction

Natural genetic transformation results in the stable acquisition of exogenous DNA by an organism. Since its discovery with Streptococcus pneumoniae (Griffith, 1928), this potential mechanism for intra- and interspecies gene transfer was reported 10 years ago in approximately 40 different species distributed among trophic and taxonomic bacterial groups, including Archaea (Lorenz and Wackernagel, 1994). To undergo transformation, most transformable species develop transiently a capacity for binding and taking up naked DNA present in their environment. This capacity, termed competence, requires the synthesis of proteins encoded by competence-specific (com) genes. Extensive studies of model transformable bacteria, including S. pneumoniae and Bacillus subtilis, identified and characterized numerous key com genes coding for competence-regulatory and DNA-uptake proteins. The detection in completely sequenced microbial genomes of intact homologues of key com genes may thus enable the prediction of new transformable species (Claverys and Martin, 2003).

In contrast to this solid body of information, one central question (with respect to transformation) has received much less attention, the source of transforming DNA. The presence of DNA in the environment as well as in the culture fluids of several bacteria has been described (Ottolenghi and Hotchkiss, 1960; 1962; Lorenz et al., 1991; Lorenz and Wackernagel, 1994). Interestingly, a coincidence of peaks of chromosomal DNA release and competence was often observed in experiments on S. pneumoniae and B. subtilis (Ottolenghi and Hotchkiss, 1960; 1962; Lorenz et al., 1991). Genetic exchange was even shown to occur in mixed cultures without experimental intervention (Ottolenghi and Hotchkiss, 1960; 1962). However, no evidence that DNA release and competence were causally related could be obtained as both transformable and untransformable strains of S. pneumoniae were observed to release active genetic material. Moreover, a transformable culture grown under conditions in which it could not develop competence still released DNA, leading to the conclusion that competence and DNA release were independent processes (Ottolenghi and Hotchkiss, 1962). The mechanism of release also remained completely unknown as no evidence of cell disintegration or death concomitant with DNA release was obtained (Ottolenghi and Hotchkiss, 1962).

In the years since these early studies, a battery of well-characterized com mutants has accumulated, and we have taken advantage of them to re-open the question of a causal relationship between competence development and DNA release in cultures of S. pneumoniae. The regulation of competence in S. pneumoniae is known to rely on a cell–cell signalling system involving five genes, comABCDE (for a review, see Claverys and Håvarstein, 2002). comC encodes the pre-CSP (competence-stimulating peptide), which is matured and exported by ComAB (Hui et al.. 1995) as an unmodified 17-residue-long peptide pheromone (Håvarstein et al., 1995). ComDE is a two-component regulatory system that responds to external CSP (Pestova et al., 1996). ComD, a membrane-bound histidine kinase, is the CSP receptor, which upon signal recognition activates its cognate response regulator, ComE. The latter controls the expression of ComX, an alternative sigma factor required for expression of the late com genes (Lee and Morrison, 1999; Luo and Morrison, 2003; Peterson et al., 2004). The entire com regulon consists of more than 100 genes (Dagkessamanskaia et al., 2004; Peterson et al., 2004).

Using comABCDE mutants, together with externally added CSP and culture conditions permissive or non-permissive for competence, we demonstrate here that DNA release is triggered by the induction of competence in S. pneumoniae. We compare our data with those of a parallel study by Havarstein and coworkers (Steinmoen et al., 2002; 2003), and assess the validity of their ‘coordinated release-and-uptake’ model. In contrast to this model, we show that CSP-induced DNA release reaches its maximum in stationary phase when cells are no longer competent. We also show that released DNA is subject to attack by the major endonuclease, EndA, and by at least one other DNase specifically induced or released in competent cultures. Finally, we provide evidence for the involvement of cell lysis, by showing that DNA release depends on the major autolytic amidase, LytA, and an autolytic lysozyme, LytC.

An initial oral communication of our data was presented at the 6th European Meeting on the Molecular Biology of the Pneumococcus (Moscoso and Claverys, 2002).

Results

Conditions permissive for competence favour DNA release

Preliminary experiments indicated that a significant level of extracellular DNA is generally present in pneumococcal cultures and that this could hinder accurate estimation of specific DNA release. We sought conditions that minimized this level. Age of the preculture, freeze-thawing of the preculture, and medium shift upon inoculation of the culture in which release was monitored all affected the amount of DNA in culture supernatants early after inoculation (data not shown). We therefore adopted a standard procedure, in which precultures were grown in conditions non-permissive for competence to OD550 of 0.2, frozen at −70°C in the presence of glycerol, and used within not more than 2 days by 1/50 dilution into fresh medium to start the experimental cultures. Furthermore, to minimize effects of variations in medium composition, all experiments were carried out using pairwise comparisons of conditions and/or strains.

The strains used harboured a com::luc transcriptional fusion to allow monitoring of competence, and a streptomycin resistance (SmR) chromosomal marker to assay DNA release by transformation of a recipient to SmR (see Experimental procedures). This assay is a measure not only of the quantity of DNA but also of its biological integrity and is thus appropriate to our aim of documenting the possible source of transforming DNA in nature.

We first investigated whether spontaneous competence development favoured chromosomal DNA release. S. pneumoniae com+ cells were grown in parallel under conditions permissive (C medium) and non-permissive (acid C medium; Experimental procedures) for competence. The results (Fig. 1) show that extracellular DNA began to increase 40–60 min after the onset of competence when cells were grown in permissive conditions but did not increase in the non-permissive culture. After 160 min following inoculation the amount of DNA liberated was approximately 220-fold lower in the non-permissive culture (Fig. 1). DNA release was also significantly lower in permissive cultures of comA, comD or comE mutant strains (data not shown).

Figure 1.

Conditions permissive for competence favour DNA release. Strain R1183 (comC+) was grown in C medium under permissive (initial pH 7.9; open symbols) or non-permissive (initial pH 6.9; filled symbols) conditions. To measure chromosomal DNA release, samples taken at intervals were filtered and SmR transforming activity was assayed (see Experimental procedures). Vertical bars: DNA release. Solid lines: competence expressed in relative luminescence units (RLU) divided by OD492. Broken lines: OD492. Values for DNA release and competence represent units of 103 SmR ml−1 and 1840 RLU/OD respectively.

Although suggestive, these data did not prove that competence development and DNA release are causally related. We therefore tested whether CSP could trigger DNA release in mutants that cannot make CSP (ΔcomC) or transport it (comA) and so depend on externally added CSP for competence development.

CSP induces DNA release in com mutants

A ΔcomC strain, R1343 (ΔcomC, comC::luc, str1), was inoculated in C medium, and CSP was added after 120 min of incubation (OD492 = 0.036). At all times after 20 min following CSP addition, the concentration of DNA in culture supernatants was much higher in the CSP-treated culture than in the control (Fig. 2A). CSP also induced DNA release in comA mutants, whereas it did not do so in mutants lacking either the CSP sensor, ComD, or its cognate response regulator, ComE (data not shown).

Figure 2.

CSP induces DNA release in a ΔcomC strain. Strain R1343 (ΔcomC) was grown in C medium and CSP1 was added at 100 ng ml−1 (arrow) after 120 min incubation to half of the culture. Open symbols: CSP. Filled symbols: no CSP. Vertical bars: DNA release. Solid lines: competence expressed in RLU/OD. Broken lines: OD492. Values for DNA release and competence represent units of 103 SmR ml−1 and 1267 RLU/OD respectively.
A. DNA release from 0 to 140 min after CSP addition.
B. DNA release throughout the experiment.

These observations provide evidence that DNA release and competence are regulated by the same cell–cell signalling regulatory circuit. In agreement with this conclusion, trt1 as well as ciaR mutant cells, which display a competence-upregulated phenotype (Martin et al., 2000), also released significantly higher amounts of DNA than wild-type cells (data not shown). However, the data in Figs 1 and 2A suggest that the kinetics of DNA release and of competence differ. In the experiment shown in Fig. 1, spontaneous competence started to develop at 80 min (OD492 = 0.015) and peaked 20 min later, whereas extracellular DNA was at a maximum at 160 min. At this time, the liberated DNA was 5.3-fold higher than it had been at the time competence peaked, while competence had decreased 2.8-fold. We next investigated this behaviour.

Kinetics of CSP-induced DNA release

Examination of release during prolonged incubation revealed that the liberation of DNA continued while competence disappeared (Fig. 2B). The amount of liberated DNA was thus approximately 210-fold higher after 4 h of incubation with CSP than after 20 min This maximum occurred in stationary phase, where cells are no longer competent for DNA uptake (Fig. 2B). We could calculate, assuming that released DNA and purified chromosomal DNA have similar transforming efficiencies, that it corresponded to the release of about 10 ng ml−1 DNA, which in turn corresponds to approximately 0.5% of total chromosomal DNA of the culture.

To investigate whether cell density could affect the kinetics of CSP-induced DNA release, CSP was added to aliquots of the same ΔcomC mutant culture at different ODs, chosen so as to mimic cell densities at which competence normally develops. Taking into account a previous observation that in cultures in which the initial pH was varied from 8.1 to 7.2, spontaneous competence first developed at OD550 ranging between 0.02 and 0.2 (Chen and Morrison, 1987), findings fully confirmed by our own upublished observations, CSP was added at three different OD492, 0.03, 0.09 and 0.23. In each case, competence peaked 20 min after CSP addition, then decreased while the amount of liberated DNA reached a maximum only in stationary phase (Fig. 3). Essentially identical results were obtained in a similar experiments using comA mutant cells (data not shown). These data demonstrate that CSP-induced DNA release and competence do not reach their maxima at the same time, contrary to the conclusion of Steinmoen et al. (2002).

Figure 3.

Kinetics of CSP-induced DNA release. Strain R1070 (ΔcomC) was inoculated in C medium and CSP1 was added at 100 ng ml−1 (arrows) to aliquots of the culture after 100 min (filled symbols), 140 min (open symbols) or 180 min (grey symbols) of incubation. Grey-filled squares: no CSP. Vertical bars: DNA release. Solid lines: competence expressed in RLU/OD. Broken lines: OD492. Values for DNA release and competence represent units of 103 SmR ml−1 and 1799 RLU/OD, respectively.

The major endonuclease EndA affects the stability of released DNA

A fall in transforming activity of culture filtrates with continued growth was previously reported and was thought to be caused by nucleolytic inactivation of previously released DNA (Ottolenghi and Hotchkiss, 1960; 1962). Our data also suggested a decrease in transforming activity of culture filtrates (Figs 1 and 2A). This observation prompted us to investigate the stability of DNA in cultures and culture supernatants. Naked chromosomal DNA harbouring a rifampicin resistance (RifR) genetic marker was added to cultures (10 ng ml−1 final concentration), samples taken at intervals were filtered, and the transforming activity of the cell-free filtrates was assayed using a RifS recipient strain. The amount of externally added DNA was chosen to approximate the maximum concentration attained through spontaneous release (see above).

The results are shown in Fig. 4A. DNA added to a ΔcomC culture remained stable for about 80 min (until the culture reached OD492 of ≈ 0.02), then decreased progressively to become essentially undetectable at t = 180 min (OD492 of 0.22; Fig. 4A). Because a non-competent strain (ΔcomC) was used in this experiment, DNA uptake could not account for the observed decrease in transforming activity. This experiment suggested that a DNase activity, which was not competence-induced, was responsible for the decrease in transforming activity of externally added DNA.

Figure 4.

The major endonuclease EndA affects the stability of externally added DNA. Strains were incubated in C medium at 37°C (A) or 30°C (B) and R304 DNA (RifR, SmR) was added (at 10 ng ml−1, A; or 20 ng ml−1, B) at time 0 (A and B) and again after 165 min of further incubation (B). The stability of added DNA was measured at intervals by filtering samples and assaying the filtrates for ability to transform strain R827 (A) or R394 (B) to RifR or SmR.
A. strain R1343 (ΔcomC; filled symbols) and R1360 (ΔcomC, endA; open symbols); broken lines: OD492.
B. strain R243 (comA). Black and grey bars: stability of DNA added at time 0 and 165 min respectively. White bars: stability of DNA added to and incubated with the supernatant taken at time 165 min. Solid line: OD492.

In another experiment, in which a comA (SmS) mutant was used instead of a ΔcomC mutant, we added SmR DNA either at t = 0 min (OD492 = 0.026) and at t = 165 min (OD492 = 0.22) (Fig. 4B). A faster rate of decrease in SmR transforming activity was observed in the latter case (<30 min for a 65% reduction, versus ≈120 min), probably reflecting a higher DNase concentration attributed to higher cell density. DNA was also added to the supernatant of a sample taken at 165 min, and incubated. Transforming activity decreased in parallel to that of the whole culture (Fig. 4B), demonstrating that at t = 165 min a nuclease had been released into the medium. Residual transforming activities 40 and 60 min after DNA addition were higher for DNA incubated with the supernatant (Fig. 4B) suggesting some instability of the nuclease in the supernatant or, more likely, continuous production of the nuclease in the culture.

In an attempt to identify this nuclease, a mutant lacking EndA, the major endonuclease of S. pneumoniae (Lacks et al., 1975), was investigated in parallel with its wild-type parent (Fig. 4A). Transforming activity of externally added DNA was stable within experimental error, in the endA mutant culture, at least until the culture reached the stationary phase (t = 260 min). This result strongly suggested that the decrease in transforming activity observed in non-competent endA+ exponential phase cultures was due to degradation by EndA. From its effect on externally added DNA, we deduce that EndA can affect the stability of released DNA, possibly leading to underestimation of the amount of chromosomal DNA liberated into the medium in our transformation assay. While transforming activity was stable in the endA log phase culture, it was about 20% lower in stationary phase (Fig. 4A). It is unclear whether this limited decrease was due to a new nuclease appearing in stationary phase or whether some DNA was bound to stationary phase cells and lost during filtration.

Appearance of a new DNase activity following CSP addition

Because our results indicated that the competence-dependent release of DNA occurred by autolysis (see next section), we reasoned that the addition of CSP might also trigger the appearance of one or more nucleases. We therefore investigated the effect of CSP on the stability of DNA added to endA mutant cultures. Addition of CSP at t = 100 min (OD492 of 0.02), but not at t = 280 min (i.e. when cells are in stationary phase and therefore unable to respond to CSP), resulted in a net decrease in transforming activity (Fig. 5). A comparison with data in Fig. 4 revealed that the rate of decrease was lower than that attributed to EndA in non-competent cultures. The persistence of the decrease in transforming activity in stationary phase, i.e. when competence has disappeared (Fig. 5), rules out the possibility that DNA was lost during filtration because of transient binding to the surface of competent cells. The CSP-induced reduction in transforming activity of externally added DNA is therefore attributed to the appearance of a new nucleolytic activity.

Figure 5.

Appearance of a new DNase activity following CSP addition. Strain R1360 (ΔcomC, endA) was incubated in C medium at 37°C. R304 DNA (RifR) was added at time 0 at 10 ng ml−1. CSP1 was added at 100 ng ml−1 after 100 min (black symbols) or 280 min (grey symbols) of incubation. A non-induced culture was run in parallel (open symbols; note that these data are the same as those shown in Fig. 4A). The stability of added DNA expressed as per cent initial activity (vertical bars) was measured in aliquots taken at intervals, as described in Fig. 4. Solid lines: competence expressed in RLU/OD. Broken lines: OD492. Values for DNA release and competence represent units of 103 RifR ml−1 and 1969 RLU/OD respectively.

Dependence of DNA release on two cell-wall hydrolases, LytA and LytC

A previous report indicated that no cell disintegration concomitant with DNA release was detected (Ottolenghi and Hotchkiss, 1962). Nevertheless, the observation that competence induced the release of pneumolysin, a cytoplasmic protein (S. Guiral, T. Mitchell, B. Martin, and J.-P. Claverys, in preparation), suggested to us that DNA release could occur through cell lysis, rather than through a specific mechanism of secretion. LytA (N-acetylmuramyl- l-alanine amidase), the major autolysin of S. pneumoniae (García et al., 1986), was therefore an obvious candidate for an essential role in the process. We first checked the effect of the presence of 2% choline in the medium, a condition known to inhibit LytA and thus prevent cell lysis (Briese and Hakenbeck, 1983; Giudicelli and Tomasz, 1984). While competence induction was unaffected by the presence of choline, DNA release was almost completely abolished (Fig. 6A). This encouraged us to investigate CSP-induced DNA release in a lytA mutant background. DNA release was significantly reduced in the absence of LytA, but by no mean abolished (Fig. 6B).

Figure 6.

Involvement of cell-wall hydrolases in CSP-induced DNA release. Each strain was inoculated in C medium and CSP1 was added at 100 ng ml−1 after 120 min of incubation to an aliquot of the culture. Vertical bars: DNA release. Solid lines: competence expressed in RLU/OD. Broken lines: OD492. Data for control cultures without CSP incubated and analysed in parallel have been omitted for clarity, except in D.
A. Strain R1343 (ΔcomC) grown in the absence (open symbols) or presence of 2% choline (filled symbols).
B. Strain R1367 (ΔcomC), open symbols. Strain R1380 (ΔcomC, lytA), filled symbols.
C. Strain R1369 (ΔcomC, lytC), open symbols. Strain R1372 (ΔcomC, lytA, lytC), filled symbols.
D. Strain R1372 (ΔcomC, lytA, lytC) with CSP, open symbols; no CSP, filled symbols.
Values represent units of 103 SmR ml−1 for DNA release and of 1267 (A), 1972 (B) and 962 (C and D) RLU/OD for competence. Note that the scales for DNA release differ in A–B, C and D.

Free choline does not inhibit only LytA. It also inhibits LytC, an autolytic lysozyme (García et al., 1999). We therefore checked a lytC mutant strain and observed that the absence of LytC reduced DNA release to an extent similar to that seen in the absence of LytA (Fig. 6C). Most interesting was the observation that DNA release was almost completely abolished in the absence of both LytA and LytC (Fig. 6C). This result strongly suggests that the combined action of the autolytic amidase and the autolytic lysozyme is required to promote DNA release in competent cultures of S. pneumoniae. It is worth pointing out that significant although very limited residual DNA release is still detected in the double mutant (Fig. 6D). This observation suggests that a third hydrolytic activity could contribute to the process.

Discussion

Our results in conjunction with those of Steinmoen et al., (2002; 2003) have established a causal relationship between competence development and the liberation of DNA in S. pneumoniae, and provide evidence that DNA release and competence are controlled by the same cell–cell signalling regulatory circuit.

Kinetics of CSP-induced DNA release

Although it is clear that competence and DNA release are causally related, their maxima never coincided. Thus, whereas competence routinely reached a maximum 20 min after CSP addition, the amount of DNA liberated was consistently approximately 200-fold higher in stationary phase than at the peak of competence. These results do not agree with the conclusions reached by Steinmoen et al., (2002). Based on a single kinetic experiment in which samples were assayed every 5 min for a period of 60 min after CSP addition, they concluded that the release of cytoplasmic β-galactosidase, used as a surrogate for DNA, ceased after approximately 25–30 min. By comparing their results with the kinetics of competence induction reported by another group (Peterson et al., 2000), they inferred that DNA uptake and DNA release reached their maxima at about the same time (Steinmoen et al., 2002). They further concluded that the release of DNA from the donor cells was coordinated in time and space with uptake by the recipients (Steinmoen et al., 2003).

The observation that CSP-induced release did not stop in any of our experiments when competence disappeared but rather continued for several hours does not fit this appealing model. The discrepancy could result from the use of different media (CAT versus C medium) and/or strains. Steinmoen et al., (2002) used strain CP, an Rx derivative, whereas our strains were derived from R6. These two sets of strains have different histories (Tiraby et al., 1975) and are clearly not strictly related (Dagkessamanskaia et al., 2004). A further explanation for the discrepancy may lie in the design of Steinmoen and coworkers’ experiments, which involved measurement of release at a single point, 30 min after CSP addition, thus preventing detection of the delayed release we have demonstrated here.

Cell-density and CSP-induced DNA release

As we detected DNA release following addition of CSP at widely differing cell-densities, from OD492 of 0.03–0.23, we conclude that the triggering of DNA release by CSP is not cell-density-dependent. This is in contrast to the report of Steinmoen et al., (2002) that release was readily detected when CSP was added to cultures at OD550 of 0.4 but, in most experiments, not after CSP addition at the beginning of logarithmic phase (OD550 of 0.1–0.2). We suggest that this reflected the failure to detect release with a single early measurement rather than a lack of induction at low cell-density. In contrast, addition of CSP close to stationary phase allowed detection of release 30 min later, because competence-triggered cell lysis is most efficient in stationary phase, as we have shown here.

Our observation that DNA release was highest in stationary phase could indicate that high cell-density increases the efficiency of release. This could be because of an increased probability of cell–cell contacts, which have been suggested to be required for release (Steinmoen et al., 2003). Alternatively, changes in cell physiology in stationary phase (e.g. modification of peptidoglycan) could favour cell lysis and therefore release. Indeed, there is evidence that two cell-wall hydrolases are under some sort of physiological control. Addition of purified LytA and LytC, respectively, to lytA and lytC mutant cultures in the exponential phase of growth resulted in autolysis only when cultures reached stationary phase (Tomasz and Waks, 1975; García et al., 1999).

DNases and the stability of released DNA

We identified EndA, the major endonuclease of S. pneumoniae, as one of the DNases potentially responsible for the degradation of released DNA. EndA was reported to be located in the membrane (Lacks and Neuberger, 1975). It does not normally have access to external DNA in non-competent cultures, and at competence requires the presence of other proteins of the DNA transport apparatus (e.g. ComGA and ComEA) to generate the DNA single strands that are taken up (Bergéet al., 2002). The observation that cell-free filtrates of non-competent endA+ cultures were able to degrade DNA strongly suggests that during normal growth some EndA molecules are liberated into the medium by autolysis or by an unknown mechanism.

While DNA added to non-competent endA mutant cultures remained essentially stable throughout the growth cycle, its transforming activity decreased following addition of CSP. This result strongly suggests that competence triggers appearance of an as yet unidentified DNase. Degradation of DNA was previously reported to occur at the surface of competent cells, resulting in the accumulation of low-molecular-weight (cold trichloroacetic acid-soluble) products and of DNA molecules heterogeneous in size, having undergone endonucleolytic incisions (Seto et al., 1975). Only the former disappeared almost completely in a strain harbouring a mutation (end1) in the endA gene, suggesting the presence of an additional endonucleolytic activity at the surface of competent cells. However, it was later shown that the end1 mutation reduced but by no mean abolished EndA activity (Lacks et al., 1975). It is therefore unclear whether this previous report identified the same nucleolytic activity as the one we have detected in the absence of EndA.

The unknown nuclease could be induced at competence (i.e. the corresponding gene belongs to the competence regulon) and surface-exposed (or released into the medium). It may correspond to the S. pneumoniae counterpart of NucA, the nuclease needed for double-strand cleavage of transforming DNA in B. subtilis (Provvedi et al., 2001). Alternatively, it could be constitutively expressed and membrane-bound, its specific detection following CSP addition implying some surface rearrangements in competent cells; or cytoplasmic, its liberation from competent cells resulting from competence-triggered autolysis (see below).

According to our calculation, the amount of DNA released would correspond to ≈ 0.5% of the total chromosomal DNA of the culture. Our estimate is much lower than the ∼1/5 of the total β-galactosidase activity found in the supernatant reported by Steinmoen et al. (2002). This strongly suggests that DNases affect the stability of released DNA, leading to underestimation of the amount of chromosomal DNA liberated into the medium in our transformation assay. Nevertheless, with respect to the question of the source of transforming DNA in nature, the direct measurement of the biologically active fraction of DNA is more relevant than an estimate based on the release of a cytoplasmic protein used as a surrogate for DNA.

Mechanism of competence-induced DNA release: cell lysis

Previous observations suggested that competent cells were prone to autolysis. Competence was shown to activate protoplast formation (Lacks and Neuberger, 1975; Seto and Tomasz, 1975) and a tendency of cells collected soon after competence to lyse was noticed (Morrison and Baker, 1979; our unpubl. obs.). In addition, recent data suggest the existence of a connection between competence induction, stress responses and autolysis (Dagkessamanskaia et al., 2004). In agreement with Steinmoen et al. (2002), we found that addition to the growth medium of 2% choline, which is known to inhibit both the autolytic amidase, LytA, and the autolytic lysozyme, LytC, almost completely blocked CSP-induced DNA release without affecting competence induction. Consistent with this observation, inactivation of lytA significantly reduced but did not abolish DNA release. A similar effect of lytA inactivation on release was reported previously (Ramirez, 1998; Steinmoen et al., 2002; 2003). However, the former author concluded that DNA release occurred independently of competence.

We also found that inactivation of lytC affected DNA release to the same extent as inactivation of lytA. This observation suggests that although LytC was reported to function optimally at 30°C (García et al., 1999), it is sufficiently active at 37°C to play a significant part in DNA release. A recent report concluded that LytC was not required for CSP-induced release of a cytoplasmic β-galactosidase (Steinmoen et al., 2003). At the moment, there is no explanation for the discrepancy. However, the same report concluded that release of β-galactosidase was two- to threefold higher at 30°C than at 37°C, which fits well with our conclusion that LytC contributes to DNA release. The almost complete abolition of DNA release in the absence of LytA and LytC provide evidence that the combined action of the two cell-wall hydrolases is required for DNA release.

It is unlikely that increased transcription of lytA or lytC in competent cells is responsible for autolysis. First, the previously reported 3.5-fold induction of lytC at competence (under the name cbp6; Rimini et al., 2000) was not confirmed in three independent reports (Steinmoen et al., 2003; Dagkessamanskaia et al., 2004; Peterson et al., 2004). Second, although lytA has been demonstrated to belong to the cinA-recA late com operon (Mortier-Barrière et al., 1998), a strain engineered so as to prevent CSP-induced expression of lytA showed no reduction in CSP-triggered DNA release (our unpubl. obs.). An alternative possibility is that competence-dependent sensitization to autolysis is due to the previously reported alteration(s) of the cell wall and/or the cytoplasmic membrane accompanying the differentiation to competence (Tomasz and Zanati, 1971). We reported recently that CSP induced complete autolysis of ciaR mutant cultures in stationary phase (Dagkessamanskaia et al., 2004). CiaR is the response regulator of a two-component regulatory system, CiaRH, which we proposed to be required for normal exit from competence. It is possible that despite the presence of a functional CiaRH regulatory system a fraction of a wild-type culture fails to exit normally from competence and undergoes autolysis.

Evolutionary significance of competence-triggered lysis

Because both DNA release and competence are controlled by the same regulatory circuit (ComABCDE) and because their observations suggested that DNA release and competence reached their maxima at about the same time, Steinmoen and coworkers proposed that competence-triggered lysis has been evolved to ensure coordination in time and space between release and uptake by competent cells, thus favouring genetic exchange (Steinmoen et al., 2002; 2003). Our kinetic data are not consistent with such a coordinated release-and-uptake model. In addition, the concomitant appearance of a new DNase activity in competent cultures that could degrade released DNA is difficult to reconcile with a model in which competence-induced release would have been tuned up during evolution to maximize gene exchange. We are thus led to consider other roles for competence-triggered lysis in the biology of S. pneumoniae. One interesting possibility would be that released DNA is important for colonization. DNA can contribute to biofilm structure (Sutherland, 2001) and was even reported to be required at early stages in the process of biofilm formation in the case of Pseudomonas aeruginosa (Whitchurch et al., 2002). Competence development is accompanied by the appearance of agglutination properties (Tomasz and Zanati, 1971). The new agglutinin could combine with liberated DNA to promote the formation of a biofilm, thus favouring pneumococcal colonization.

Alternatively, the release of large amounts of DNA could participate in the virulence of S. pneumoniae because bacterial DNA can cause septic shock (Sparwasser et al., 1997), elicit inflammatory cytokine production and stimulate lymphocyte proliferation (Krieg et al., 1995; Klinman et al., 1996; Chatellier and Kotb, 2000). Several other compounds potentially released by autolysis could also contribute to virulence. Competence-triggered release of pneumolysin, a major virulence factor of S. pneumoniae (Jedrzejas, 2001), has been established in our laboratory (S. Guiral, T. Mitchell, B. Martin, and J.-P. Claverys, in preparation). In addition, cell lysis produces teichoic and lipoteichoic acid, which are well-known inflammatory compounds (Nau and Eiffert, 2002). Competence could thus not only allow genetic plasticity via transformation (Claverys et al., 2000), but also participate in the virulence of S. pneumoniae by triggering autolysis, possibly when cells encounter environmental conditions in the host that do not permit normal exit from competence (Dagkessamanskaia et al., 2004). An additional, non-mutually exclusive possibility is that competent cells trigger lysis to feed on the nutrients thereby released, as recently described in the case of sporulating B. subtilis (González-Pastor et al., 2003).

Experimental procedures

Bacterial strains and growth conditions

S. pneumoniae strains, plasmids and oligonucleotides used in this work are listed in Table 1. All S. pneumoniae strains (except strain R6BC) were constructed by transformation of the R6 derivative strain R800. Pneumococcal strains were grown in C medium (Alloing et al., 1998) or in acid C medium (pH value of 6.7–6.9). A low initial pH value is known to prevent spontaneous competence development in wild-type cultures (Chen and Morrison, 1987). Cells were incubated at 37°C and growth was monitored by measuring OD at 492 nm or 550 nm.

Table 1. S. pneumoniae strains, plasmids and primers used in this study.
StrainGenotype and constructionaSource or reference
  • a

    . Only if not previously described.

  • A and C refer, respectively, to the anti- and cotranscribed orientation of the minitransposon antibiotic resistance cassette with respect to the target gene.

  • Cm, chloramphenicol; Ery, erythromycin; Kan, kanamycin; R, resistant; S, sensitive; Spc, spectinomycin; Tc, tetracycline.

S. pneumoniae
 R6Non-capsulated D39 derivativeLaboratory stock
 R6BC lytB::ermC, lytC::tet; EryR; TcR R6 derivativeP. García
 R800R6 derivative Martin et al. (1985)
 R243 comA::ermAM A; EryR Alloing et al. (1996)
 R304 nov1, rif23, str41; NovR; RifR; SmR Mortier-Barrière et al. (1998)
 R322 recA::pR322, recA+ Ind, hexAΔ3::ermAM; CmR; EmR Mortier-Barrière et al. (1998)
 R394 trt1 (this mutation corresponds to the ComDD299N change which upregulates competence; Martin et al., 2000)
R800 derivative
This study
 R825 comC::luc (pR414), comC +; EryR Bergéet al. (2002)
 R827 trt1, hexAΔ3::ermAM; EryR R394, but hexAΔ3::ermAM (R322 donor DNA)This study
 R895 ssbB::luc (pR424), ssbB +; CmR Chastanet et al. (2001)
 R924 lytA::kan4C; KanR R800, but lytA::kan by mariner mutagenesis (Experimental procedures)This study
 R938 endA::kan6C; KanR Bergéet al. (2002)
 R1036 str1 (rpsL1), ΔcomC::kan-rpsL+; KanR; SmS Sung et al. (2001)
 R1068ΔcomC, comDI176T, str1; SmR R1036, but ΔcomC (Experimental procedures)This study
 R1069ΔcomC, str1; SmR R1036, but ΔcomC (Experimental procedures)This study
 R1070 str1, ΔcomC, comC:: luc (pR414); SmR; EryR R1068, but comC::luc (pR414 donor DNA)This study
 R1183 str41, comC::luc (pR414), comC+ SmR; EryR R825, but str41 (R304 donor DNA)This study
 R1343 str1, ΔcomC, comC::luc (pR414); SmR; EryR R1069, but comC::luc (pR414 donor DNA)This study
 R1359 str1, ΔcomC, comC::luc (pR414), lytA::kan4C; SmR; EryR; KanR R1343, but lytA::kan (R924 donor DNA); SmR; EryR; KanRThis study
 R1360 str1, ΔcomC, comC::luc (pR414), endA::kan6C; SmR; EryR; KanR R1343, but endA::kan (R938 donor DNA)This study
 R1367 str1, ΔcomC, ssbB::luc (pR424), ssbB +; SmR; CmR R1069, but ssbB::luc (R895 donor DNA)This study
 R1369 str1, ΔcomC, comC::luc (pR414), lytC::tet; SmR; EryR; TcR R1343, but lytC::tet (R6BC donor DNA)This study
 R1372 str1, ΔcomC, comC::luc (pR414), lytA::kan4C, lytC::tet; SmR; EryR; KanR; TcR R1359, but lytC::tet (R6BC donor DNA)This study
 R1380 str1, ΔcomC, ssbB::luc, ssbB+, lytA::kan4C; SmR; CmR; KanR R1367, but lytA::kan (R924 donor DNA)This study
Plasmid
 pR410pEMcat derivative, ApR, KanR; carries a 1337-bp-long mini-transposon containing the inverted repeats (IRs) of the Himar1 transposon and ≈ 100 bp of
Himar1 transposon sequences flanking the kan gene
Sung et al. (2001)
 pR414p5.00 derivative carrying a 389 bp comC targeting fragment; insertion-duplication in S. pneumoniae generates a comC::luc (comC +) fusion; EryR Bergéet al. (2002)
PrimersSequence; gene; position within the deposited sequence indicated; GenBank/EMBL accession number 
 MMN9AGGGCTTGGGGCAAAAGATCCTT; upstream of lytA; Z34303 
 MMN10CCATCCGTGATATTCCTTTCACCT; downstream of lytA; Z34303 
 MP127CCGGGGACTTATCAGCCAACC; mariner cassette universal primer; internal to terminal IRs; outward orientation 
 MP128TACTAGCGACGCCATCTATGTG; mariner cassette universal primer; adjacent to IRL; outward orientation 

Transformation

Transformation of S. pneumoniae with chromosomal or plasmid DNA was performed as described previously (Martin et al., 2000), using precompetent cells treated at 37°C for 10 min with synthetic CSP1 (100 ng ml−1) to induce competence. In some cases, spontaneous competence was obtained in cultures grown in C medium as described previously (Alloing et al., 1996). Cells were incubated at 30°C during DNA uptake. Methods for preparing plasmid and S. pneumoniae chromosomal DNA were described previously (Martin et al. 1985). Transformants were selected by plating in CAT agar (Bergéet al., 2002) supplemented with 3% (v/v) horse blood, followed by challenge with a 10 ml overlay containing the appropriate antibiotic after phenotypic expression for 120 min at 37°C. Antibiotic concentrations used for the selection of transformants were: chloramphenicol, 2.5 µg ml−1; erythromycin, 0.05–0.2 µg ml−1; kanamycin, 250 µg ml−1; rifampicin, 4 µg ml−1; spectinomycin, 100 µg ml−1; streptomycin, 200 µg ml−1; tetracycline, 0.5 µg ml−1.

Monitoring of competence

Measurement of competence was achieved using a transcriptional fusion between a gene specifically induced at competence (comC or ssbB; Table 1) and the Photinus pyralis luc gene that encodes the firefly luciferase, as previously described (Bergéet al., 2002). A direct correlation between luciferase activity and competence level was observed under widely varying levels of competence (Bergéet al., 2002). For the detection of luciferase activity, 280 µl aliquots of cultures in C medium were distributed into a 96-well plate and incubated in an Anthos LucyI luminometer. After injection of 20 µl of firefly d-luciferin (10 mM solution in C medium), luminescence (expressed in RLU) and OD492 were measured at timed intervals for each aliquot.

Construction of mutant strains

A complete deletion of comCcomC) was generated by fusing the ATG of comC with the second triplet of comD. This was achieved by taking advantage of a previously constructed strain, R1036 (Table 1), in which the comC gene was substituted with Janus (Sung et al., 2001). Janus is a kan-rpsL+ cassette, which confers dominant SmS in a SmR background. Replacement of Janus by an arbitrary segment of DNA during a second transformation restores SmR (and KanS) (Sung et al., 2001). Strain R1036 was therefore transformed with PCR-ΔcomC, a PCR fragment overlapping the comC::Janus segment in which the ATG of comC was fused with the second triplet of comD, generated as previously described (Dagkessamanskaia et al., 2004). SmR transformants were selected and their comCD region was sequenced to retain strains R1068 and R1069. Both contained silent (third position in triplets) changes in comD; R1068 harboured in addition an I→T change in ComD (residue 176), which had no detectable effect on its response to CSP.

Insertions of the kan (KanR) gene cassette in lytA were generated by in vitro mariner mutagenesis as previously described (Martin et al., 2000). Briefly, the plasmid used as a source for the 1337 bp kan mariner mini-transposon was pR410 (Table 1). Plasmid DNA (approximately 1 µg) was incubated with a 2078 bp lytA fragment (≈ 1 µg) generated with primers MMN9 and MMN10 (Table 1), in the presence of purified Himar1 transposase (Lampe et al., 1996), in a total volume of 40 µl leading to random insertion of the mini-transposon within the fragment. Gaps in transposition products were repaired as described (Akerley et al., 1998) and the resulting in vitro-generated transposon insertion library was used to transform S. pneumoniae. Location and orientation of the mariner cassette was determined through PCR reactions using primers MP127 or MP128 (Table 1) in combination with either one of the two primers used to generate the lytA PCR fragment. Cassette-chromosome junctions were sequenced using the Thermo Sequenase cycle sequencing kit (USB) and primer MP128. Strain R924 which carries the lytA::kan4C mariner insertion (position 60 with respect to the ATG of lytA) was retained for this study.

Transformation assay for DNA release

The release of DNA into the culture medium by SmR strains was assayed by transformation of competent cells of an SmS recipient strain, R827 (Table 1), using cell-free filtrates from SmR cultures. For all experiments described in this article, stocks of SmR strains were grown in low-pH C medium (pH 6.7–6.9) to OD550 of 0.2. Glycerol (15% final concentration) was added and aliquots were frozen at −70°C. Frozen stocks were gently thawed, washed once or twice with C medium (to remove any extracellular DNA), diluted 50-fold in C medium and incubated at 37°C. Samples (0.2 ml) were withdrawn at timed intervals and put on ice for a very short time. Cells were then removed by filtration and centrifugation for 60 s, using Micropure-EZ enzyme removers (Millipore). To maximize DNA recovery, 50 µl of cold C medium was added to Micropure-EZ reservoir and spun 30 s. Cell-free filtrates were kept at 4°C until the transformation assay.

Competent cells of strain R827 (or R394) were prepared from a freshly subcloned stock as follows. Stocks in CAT medium at OD550 of 0.4 were diluted 20-fold in C medium and incubated for 150 min at 37°C. The culture was then frozen after addition of glycerol (15% final concentration) and aliquots were stored at −70°C. For the transformation assay of DNA release (or DNA stability), competent cells were thawed on ice and diluted 100-fold in C medium. An equal volume of competent cells and cell-free filtrates were mixed (total volume 0.2–0.4 ml). The mixture was incubated at 30°C for 20 min. After plating in CAT agar, plates were incubated for 120 min at 37°C followed by challenge with an antibiotic-containing overlay. Plates were then incubated overnight before counting SmR (or RifR) transformants.

Acknowledgements

The authors wish to thank Dave Lane for critical reading of the manuscript, Pedro García for kindly providing us with the lytB-lytC double mutant strain R6BC and Chantal Granadel for expert technical assistance. M.M. was the recipient of a Marie-Curie Individual Fellowship granted to a proposal entitled ‘Genetic control and mechanism of DNA release in cultures of the transformable human pathogen Streptococcus pneumoniae’ (October 2000–2002; QLK2-CT-1999-51509). This research was financed in part by the European Union (Grant QLK2-CT-2000-00543).

Ancillary