Adaptation to the environment: Streptococcus pneumoniae, a paradigm for recombination-mediated genetic plasticity?

Authors

  • Jean-Pierre Claverys,

    1. Laboratoire de Microbiologie et Génétique Moléculaire CNRS-UPR 9007, Université Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex, France.
    Search for more papers by this author
  • Marc Prudhomme,

    1. Laboratoire de Microbiologie et Génétique Moléculaire CNRS-UPR 9007, Université Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex, France.
    Search for more papers by this author
  • Isabelle Mortier-Barrière,

    1. Laboratoire de Microbiologie et Génétique Moléculaire CNRS-UPR 9007, Université Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex, France.
    Search for more papers by this author
  • Bernard Martin

    1. Laboratoire de Microbiologie et Génétique Moléculaire CNRS-UPR 9007, Université Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex, France.
    Search for more papers by this author

Jean-Pierre Claverys. E-mail claverys@ibcg.biotoul.fr; Tel. (+33) 561 33 59 11; Fax (+33) 561 33 58 86.

Abstract

Genetic plasticity plays a central role in the biology of the human pathogen Streptococcus pneumoniae. This is illustrated by the existence of at least 90 different capsular types (the polysaccharide capsule has an essential antiphagocytic function) as well as by the rapid emergence of penicillin-resistant (PenR) pneumococcal isolates. Natural genetic transformation is believed to be essential for this genetic plasticity; capsular types can be switched by intraspecies transformation, whereas interspecies transformation is responsible for the appearance, in the PenR isolates, of mosaic pbp genes, which encode proteins with reduced affinity for penicillin. Data on the regulation of competence for transformation in S. pneumoniae, on the control of intra- and interspecies genetic exchange and on the shuffling and capture of exogenous sequences during transformation are reviewed. Possible links between transformation and changes in environmental conditions are discussed, and the adaptive ‘strategy’ deduced for S. pneumoniae is compared with that of Escherichia coli.

Introduction

Microorganisms require a certain degree of flexibility to adapt to changes in their environment (e.g. exhaustion of nutriments or, for pathogens, triggering of host cell defences). Cellular adaptation can be achieved through the activation of regulatory networks as well as through changes in the genetic material. Genetic variability can be provided by both endogenous mechanisms and horizontal transfer. Only the latter will be considered here. The control of gene transfer between bacterial species has been dissected in some detail, but most of the information comes from the study of enterobacteria (Matic et al., 1996). Naturally transformable bacteria have received little attention from this point of view, despite the fact that transformation with naked DNA is probably the most frequent mode of horizontal transfer among bacteria. The number of species identified as naturally transformable appears to be much larger than initially thought (for a review, see Lorenz and Wackernagel, 1994), and recent evidence suggests that this number will soon increase significantly (Håvarstein and Morrison, 1999).

In this review, we summarize data available for Streptococcus pneumoniae on the regulatory aspects of competence for genetic transformation as well as on recombination, including its regulation, the possible role of mismatch repair (MMR) in avoiding interspecies genetic transfer and the type of recombination events identified. Although some aspects of transformation and/or recombination are better documented in other transformable species (e.g. Bacillus subtilis or Neisseria gonorrhoeae), S. pneumoniae offers a unique combination of data. Collectively, these data are consistent with the view that natural transformation has evolved in S. pneumoniae to maximize genetic flexibility, thus favouring adaptation of this human pathogen to environmental changes. Finally, this adaptive strategy is compared with that deduced for Escherichia coli (Taddei et al., 1997).

Genetic variability and adaptation to changes in environmental conditions of S. pneumoniae

Two examples below illustrate the flexibility potential of S. pneumoniae under selective pressure from host defences or as a result of antibiotic treatment.

S. pneumoniae colonizes man and, occasionally, other mammals. It is a commensal whose ecological niche is the nasopharynx, and it becomes a pathogen only after moving to other areas (Austrian, 1986). The polysaccharide capsule, which protects pneumococcal cells from engulfment and digestion by polymorphonuclear leucocytes, is the main virulence factor of this microorganism (for a review, see García and López, 1997). As many as 90 different capsular serotypes have been described up to now. Early observations indicated that the genes responsible for capsule synthesis were linked in the pneumococcal chromosome. Consistent with this observation, studies of several serotypes at the molecular level revealed a cassette-type organization. The cap (or cps) cassette is inserted at the same chromosomal location in all serotypes investigated so far (with the remarkable exception of the type 37 capsule, which is directed by a single gene, tts, distant from the cap locus by at least 820 kb; Llull et al., 1999) and contains up to 19 genes (≈ 20 kb of DNA for the longest), several of them being type-specific (García and López, 1997; Coffey et al., 1998).

Antibiotic treatment can be considered as a major change in environment for both commensal microorganisms and pathogens. Resistance to β-lactam antibiotics has emerged in S. pneumoniae through the development of altered penicillin-binding proteins (PBPs) with decreased affinity for β-lactam antibiotics (Dowson et al., 1994; Hakenbeck et al., 1999). Mosaic pbp genes containing regions from the corresponding genes from other streptococcal species have been found in penicillin-resistant (PenR) clinical isolates of S. pneumoniae. Mosaics that differ by up to 25% in DNA sequence have been detected so far in three out of the five pbp genes encoding high-Mr PBPs. In addition, a huge diversity was reported for each pbp gene; for example, in a collection of 45 PenR, 18 different mosaic pbp2x genes and 16 different mosaic pbp2b genes were detected (Dowson et al., 1994).

Genetic transformation as a major source of genetic variability for S. pneumoniae

Among the three selected natural strategies for the generation of genetic diversity, nucleotide substitution, DNA rearrangements and gene acquisition, only the first is not promoted by genetic transformation in S. pneumoniae. Unlike Haemophilus influenzae or N. gonorrhoeae, which take up only DNA carrying a species identifier sequence, S. pneumoniae takes up any native DNA, in the form of single strands (ss) (Lacks, 1962). The probability of homology-dependent integration of these ss varies from close to 1 for homologous DNA to 0.1–0.5 for 2–10% divergent DNA (Humbert et al., 1995). In contrast to mutations that are frequently neutral or deleterious, transformation is expected to introduce (by homology-dependent recombination) sequences that were functional in the donor. Several independent fragments (up to 1–5% of the genome) can be taken up per cell, which allows each cell to experiment with new combinations of genes. Simultaneous substitution of components of protein complexes (even when encoded by genes unlinked on the chromosome) could also occur in one transformation event. Uptake of pneumococcal DNA as well as DNA from related species present in the same ecological niche (e.g. oral streptococci) is therefore an extremely powerful mechanism for rapid evolution of the genome, and can contribute significantly to genome fluidity in S. pneumoniae.

Capsular serotype transformation in vivo (i.e. in man) could account for the identification in epidemiological studies of clinical isolates differing only in the capsular polysaccharide (Coffey et al., 1998). There is also little doubt that interspecies transformation is responsible for the evolution of virulence factors (Poulsen et al., 1998) and for the rapid emergence of antibiotic resistance. Streptococcus oralis and Streptococcus mitis are likely donors of some of the sequences that replace parts of ‘sensitive’ genes in the mosaic pbp genes of PenR pneumococcal isolates (Dowson et al., 1994; Hakenbeck et al., 1999). Laboratory transformation of the pbp1b and pbp2a genes from S. mitis into S. pneumoniae has been described recently (Hakenbeck et al., 1999). Transformation of S. pneumoniae from optochin sensitivity (a characteristic trait of the species) to optochin resistance has also been observed under laboratory conditions using S. oralis chromosomal DNA as donor (Fenoll et al., 1994).

The greater evolutionary potential of transformation over mutation is well illustrated by mosaic PBPs. Although various combinations of the five high-Mr PBPs of S. pneumoniae can be changed into low-affinity variants in one transformation event with chromosomal DNA from a β-lactam-resistant S. mitis, evolution of a single PBP towards low affinity requires, in the laboratory, the accumulation of as many as five independent mutations (Hakenbeck et al., 1999).

Interestingly, transformation provides each pneumococcal cell with access to a ‘global genome’, the size of which greatly exceeds that of an individual genome. A simple calculation on polysaccharide capsule DNA, based on the reasonable assumption that two types differ by ≈ 5 kb DNA segments, indicates that capsular DNA could amount to at least 450 kb for the species, i.e. 25% of an individual genome size. It is noteworthy that a recent survey of 214 isolates revealed the ubiquitous distribution of the competence control genes (see below), comA and comC (for a review, see Håvarstein and Morrison, 1999), strongly suggesting that transformation is conserved within the entire species.

Shuffling and capture of exogenous sequences during transformation

Although substitutive recombination is the most frequent event, a wide variety of recombination events, such as cassette insertion (e.g. capsular type transformation), insertion–duplication (e.g. integration of non-replicative recombinant plasmids) and insertion–deletions (see below), can also take place in transformation (for a review, see Mortier-Barrière et al., 1997). Shuffling of sequences can occur in the course of substitutive recombination, as deduced from a molecular analysis of recombinants formed between partially divergent sequences. Chromosomal DNA carrying a unique divergent region (3.2 kb long, 4.6% divergent nucleotides) including the pbp2b gene was used as donor in transformation. Some 14% of the 81 PenR recombinants analysed exhibited alternating blocks of donor and recipient sequences (Mortier-Barrière et al., 1997; O. Humbert, G. C. Dowson and J.-P. Claverys, in preparation). These structures were reminiscent of the mosaic pbp genes found in S. pneumoniae PenR clinical isolates. It was concluded that, at least under laboratory conditions, a single-step recombination event can give rise to mosaic recombinants during transformation with partially divergent DNA. In addition, these data indicated that shuffling of pre-existing mosaics can occur in subsequent transformation(s), leading to sequence diversification and therefore to the production of new proteins. Interestingly, mosaic recombinants were obtained in both MMR-proficient and -deficient strains (O. Humbert, G. C. Dowson and J.-P. Claverys, in preparation).

Insertion–deletion, another event potentially able to promote the capture of and to shuffle exogenous DNA, was observed in transformation of S. pneumoniae under laboratory conditions. Using as donor hybrid DNA molecules consisting of recombinant E. coli bacteriophage λ carrying a pneumococcal insert, illegitimate recombinants were detected at a frequency of about 0.5% of that of homologous recombinants (Claverys et al., 1980). The illegitimate recombination events corresponded to the simultaneous insertion of heterologous (i.e. λ) DNA and deletion of chromosomal sequences. The homologous region in the hybrid molecule (i.e. the pneumococcal insert) was absolutely required for the production of insertion–deletions. The underlying recombination mechanism was therefore named homology-directed illegitimate (HDI) recombination (Mortier-Barrière et al., 1997). A model for HDI recombination postulated that pairing between pneumococcal donor and recipient sequences could favour transient pairing between heterologous (i.e. λ or any DNA) sequences (adjacent to S. pneumoniae DNA in the donor) and resident sequences. Resolution of the heteroduplex would result in integration of heterologous DNA and concomitant loss or resident sequences (Fig. 1). Recent analysis of several novel joints confirmed the presence of small (4–10 bp) regions of sequence identity at the point of cross-over (M. Prudhomme and J.-P. Claverys, in preparation).

Figure 1.

. Creation of novel joints during transformation of S. pneumoniae: proposed mechanism for the formation of insertion–deletions. RecA catalyses invasion of the homologous donor strand (D) into the recipient duplex (RC). The resulting donor–recipient heteroduplex structure is proposed to favour transient pairing between heterologous (λ; thick line) donor DNA and the complementary strand (C) of the recipient chromosome, within a short stretch of sequence identity (vertical bars). Resolution of the heteroduplex is postulated to require concomitant incision of the heterologous (i.e. λ or any DNA) strand and of the displaced chromosomal recipient strand (R). A hypothetical resolvase (open triangle), specific for three-stranded structures, could carry out incision, followed by ligation to create a novel joint (NJ). DNA replication then generates a wild-type and a mutant chromosome. *Deleted in the R strand.

It should be emphasized that any piece of homology, including a mobile genetic element (e.g. an insertion sequence, IS), present in both the incoming DNA and the recipient chromosome can provide the homology required for the integration of completely heterologous sequences. Therefore, such a mechanism is of great potential in terms of genome evolution as, depending on the location of the novel joint, it can lead to the production of chimeras or to the acquisition of completely heterologous genes.

Repeated sequences, transformation and genome rearrangements

Transformation can also constitute a means for entry of IS. ISs belonging to the IS605, ISL3, ISNCY and IS5 families have been identified in S. pneumoniae (for a review, see Mahillon and Chandler, 1998). A non-exhaustive analysis of the partially completed genome sequence of S. pneumoniae type 4 revealed that a large number of ISs (and fragments thereof) are scattered over the genome and identified additional ISs belonging to the IS3, IS66 and IS630 families (Oggioni and Claverys, 1999; our unpublished observations). ISs can contribute to genome plasticity by promoting illegitimate recombination events (Berg and Howe, 1989). They can also participate in homology-dependent reactions by providing homology for insertion–deletions (see above) or for additive recombination events (Mortier-Barrière et al., 1997).

In addition to ISs, the S. pneumoniae genome contains two families of highly repeated small extragenic elements, BOX and RUP (Oggioni and Claverys, 1999). Several features of RUPs led to the proposal that this 105-bp-long element is an IS derivative that could be trans-mobilized by the transposase of IS630-Spn1. Examination of sequences flanking RUP revealed that homologous recombination between directly repeated RUPs could promote sequence rearrangements (Oggioni and Claverys, 1999). Interestingly, the recently identified tts gene, which directs synthesis of the type 37 capsule, is flanked by RUP elements in opposite orientation (Fig. 2; Llull et al., 1999). In type 37 strains, the tts gene is located between the gpmA and the pyrDA genes, which are separated for at least 380 kb in the physical map of the S. pneumoniae R6 chromosome. It is not known whether RUPs played any role in this genome rearrangement. However, chromosomal rearrangements were detected upon transformation of a non-capsulated laboratory strain with type 37 chromosomal DNA (Llull et al., 1999). At least two categories of strains were obtained. In one of them, integration of the tts gene immediately downstream of gpmA resulted in an insertion–deletion event removing 2400 out of 2412 bp of orf3, the gene next to gpmA in the recipient chromosome (Fig. 2). In the other, the gpmA gene moved close to pyrDA as in the donor DNA (Fig. 2). Confirmation that a large genomic rearrangement had occurred was obtained by pulsed-field gel electrophoresis (Llull et al., 1999).

Figure 2.

. Diagrammatic representation of the genomic rearrangement caused by type 37 capsule transformation. In the donor strand (shown in black), the tts gene, which directs synthesis of type 37 capsule, is located between the gpmA and pyrDA genes, while the latter genes are at least 380 kb apart in the recipient chromosome. Synapsis between gpmA in the donor and in the recipient (red–pink chromosome moiety) on one side and between pyrDA in the donor and in the recipient (light–dark blue chromosome moiety) on the other side, followed by cut-and-seal reactions (arrows), leads to type 37 transformation and concomitant chromosome rearrangement. It has not been investigated whether a reciprocal exchange that would link metE to orf3 occurred simultaneously in the rearranged type 37 transformant. The RUPC element (white triangle) in type 37 DNA is separated from the stop codon of gpmA by 269 nucleotides (black rectangle). Two blocks of sequence identity between donor and recipient DNA are indicated. The first one (96 nucleotides; blue rectangle) is adjacent to RUPA (blue triangle). The other (128 nucleotides; red rectangle) overlaps the 5′ extremity of orf3 in the recipient and is located immediately downstream of tts in the donor. The presence of a copy of IS1167 between the two blocks in type 37 DNA suggests that this IS could have played a role in the initial type 37 chromosomal rearrangement.

Genetic control of competence for transformation in S. pneumoniae

In S. pneumoniae, competence, the ability of cells to take up DNA, is a transitory property that develops suddenly during the exponential phase of growth. The development of competence is cell density dependent. The quorum-sensing signal is an unmodified heptadecapeptide named CSP (for competence stimulating peptide; for a review, see Håvarstein and Morrison, 1999) that accumulates extracellularly. It is encoded by the 3′ moiety of comC and exported, most probably by the dedicated ABC transporter ComA/B, which recognizes and processes the amino-terminus of ComC. The CSP pheromone acts through a two-component regulatory system composed of ComD, the CSP receptor and ComE, a response regulator (Håvarstein and Morrison, 1999). The addition of CSP resulted in transcriptional induction of the comCDE operon, as demonstrated in two genetic backgrounds (Alloing et al., 1998; Håvarstein and Morrison, 1999), thus explaining the autocatalytic accumulation of CSP and the synchronous development of competence in pneumococcal cultures. Most recently, a probable alternative sigma factor required for expression of the late com genes, which encode the transformation machinery (i.e. genes for DNA uptake and recombination such as recA), was characterized as the product of comX, a ComE-dependent gene (Lee and Morrison, 1999). Transcriptional activation of the late com genes most probably requires recognition by the ComX–RNA polymerase complex of the cin boxes previously identified in front of the late com genes (Håvarstein and Morrison, 1999).

A survey of 256 encapsulated S. pneumoniae strains revealed the existence of three CSP pherotypes (Håvarstein and Morrison, 1999). Nucleotide changes occurred mainly in the CSP-encoding 3′ region of comC (suggesting that each ComC allele could be processed/exported by the same ComA/B transporter) and were correlated with changes in the 5′ region of comD that is postulated to encode the CSP-binding domain of ComD (Håvarstein and Morrison, 1999; Whatmore et al., 1999). Interestingly, competence autoinduction was shown to be delayed upon mixing a wild-type strain with a CSP non-producing mutant (Alloing et al., 1998). Together with the existence of different CSP pherotypes and the documented simultaneous colonization by more than one pneumococcal capsular type (Austrian, 1986), this observation raised some intriguing questions regarding the development of competence in mixed natural populations. Would competence of the most abundant subpopulation be delayed? Would the minor fraction remain non-transformable? It has been argued that pherotypes divide competent pneumococci (as well as streptococci) into bacterial populations that cannot communicate (Håvarstein et al., 1997). However, it should be pointed out that communication (i.e. exchange of DNA) does not require simultaneous competence, but only DNA availability.

Little is known regarding the presence and persistence of DNA in nature. Free DNA exposed to human saliva has been shown to be progressively degraded (Mercer et al., 1999). However, bacteria in the oral cavity are organized in biofilms; such matrix-enclosed complex bacterial populations may offer some protection to DNA. Interestingly, type 9 pneumococci in a healthy human carrier were transformed to streptomycin resistance by DNA released from living unencapsulated pneumococci sprayed onto the pharynx (Ottolenghi-Nightingale, 1972). It has long been recognized that pneumococcal cells release DNA spontaneously relatively early in the exponential phase of growth (Ottolenghi and Hotchkiss, 1960; 1962). Cell lysis or death could not be detected at the time of maximal DNA release, which coincided with the development of maximal competence. A decline in the amount of DNA in ageing cultures was attributed to the presence of DNases. Altogether, these observations suggested that the release of DNA could be a programmed event, coupled to competence development. However, a recent reinvestigation of this phenomenon revealed that maximal release could occur somewhat earlier than (sometimes independently of) competence development (our unpublished observations). Taking into account the existence of various CSP pherotypes and the behaviour of mixed cultures with respect to competence development (see above), the latter observation suggested that, depending on the rate of degradation, released DNA could be available for interpherotype or interspecies exchanges rather than for intrapherotype transfer.

Control of intra- and interspecies genetic exchange

Neither DNA uptake nor restriction of incoming DNA (see below) are barriers to gene exchange by transformation in S. pneumoniae. The major limitation would then be on the recombination process itself, which depends on the recombination machinery and on sequence divergence. In E. coli, a minimal length of sequence identity is required by the recombination enzymes at the initial stage of the strand exchange process (Matic et al., 1996). Once initiated, strand exchange can accommodate a large number of mismatches. A key protein in the initiation of strand exchange is RecA. Interestingly, transcriptional induction of recA occurs in competent cells of S. pneumoniae (Martin et al., 1995). Competence-induced expression of recA is essential for the achievement of maximal recombination; its abolition results in a 20-fold decrease in homologous recombination (Mortier-Barrière et al., 1998). The presence of a cin box in front of the recA operon (Håvarstein and Morrison, 1999) accounts for induction and suggests that recA expression has been tuned to optimize genetic exchanges during transformation.

MMR systems can be potent inhibitors of recombination between related species (Matic et al., 1996). In S. pneumoniae, the generalized MMR system Hex can reduce transformation frequencies for point mutations in homologous DNA up to 20-fold by repairing mismatches at the donor–recipient heteroduplex stage (Claverys and Lacks, 1986). However, the Hex system is unable to prevent interspecies transformation (Humbert et al., 1995). In a range from 1.7% to 10.3% divergence, the Hex system became saturated (inhibited) as a result of an excess of mismatches. This phenomenon is illustrated in Fig. 3, which shows that, in a Hex+ recipient, transformation efficiency is proportional to divergence, whereas the opposite would have been predicted if mismatches prevented recombination. The increase in efficiency reflects the progressive inability of Hex to correct mismatches as their number increases (from about 40 to 149 over a 3205-bp-long region). HexA (the mismatch recognition protein) and HexB, which are both essential for MMR, appeared to be limiting, as an increase in either hexA or hexB gene copy number improved MMR (Humbert et al., 1995; and Fig. 3). It is noteworthy that Northern as well as Western blotting experiments failed to detect induction of hexA or hexB expression during competence (Humbert et al., 1995). This is consistent with the lack of cin box upstream of the two genes. In addition, a recent measurement of the cellular concentrations of RecA and HexB revealed a twofold reduction for the latter in competent cells (I. Mortier-Barrière, B. Martin and J.-P. Claverys, unpublished observations), consistent with the use of the alternative sigma factor ComX (see above). These data indicated that the Hex system has not been tuned to cope with excess mismatches and therefore to abort interspecies recombination during transformation, whereas at the same time, recA expression has adjusted to favour recombination.

Figure 3.

. Saturation of the Hex MMR system of S. pneumoniae by excess mismatches during transformation: both HexA and HexB are limiting. A series of chromosomal DNAs carrying the str41 reference marker (resistance to streptomycin, Str) and a unique divergent region was used as donor in transformation of an MMR-proficient strain (wt). The divergent region contained from 42 to 149 nucleotide changes (scattered over a 3.2 kb segment) and included the pbp2b gene, which confers resistance to penicillin. Transforming efficiencies are given as the ratio between PenR and StrR transformants (filled triangles). Chromosomal DNA containing the most divergent PenR marker was used to transform either wt (open triangle) or strains with multiple copies of hexA+ (phexA; open circle) or hexB+ (phexB; open square).

Further evidence that optimization of genetic exchanges by transformation has been selected during evolution is provided by a very recent observation on the DpnII restriction–modification system (Lacks, 1999). Two complementary restriction systems, DpnI and DpnII, are found in different strains of S. pneumoniae. Interestingly, they are encoded by cassettes located at the same position in the chromosome (Lacks et al., 1986). The DpnI and DpnII endonucleases recognize and cleave the methylated and unmethylated sequence 5′-GATC respectively. It has long been thought that these restriction systems do not affect substitutive recombination simply because, first, incoming donor DNA, being ss, is protected from restriction endonucleases and, secondly, upon synapsis (pairing) with resident sequences, donor DNA is embedded in hemimethylated, double-stranded regions, which protects it from restriction. Although this holds true for the DpnI system, the recent finding that dpnA is induced at competence (Lacks, 1999) suggests an additional explanation. The dpnA gene encodes a single-stranded DNA methylase acting at the N6 position of adenine in 5′-GATC sequences. Lacks (1999) proposed that the DpnA methylase is induced to protect incoming plasmid strands from the DpnII endonuclease to allow plasmid establishment. We suggest that methylation of single-stranded DNA is primarily important for chromosomal exchange of genetic cassettes in the chromosome (e.g. capsular type transformation). As heterologous DNA in a cassette has no counterpart in the recipient genome, it is expected to remain single stranded even after the formation of donor–recipient heteroduplexes on both sides of the cassette. Its conversion to unmethylated double-stranded DNA by chromosomal replication would render it sensitive to the DpnII endonuclease, unless methylation by DpnA occurs before replication. The competence-specific induction of dpnA would therefore be essential for efficient cassette substitution in DpnII strains.

Quorum sensing and genetic variability

Several hypotheses have been advanced to explain the evolutionary origin of transformation, including a role in nutrient acquisition, in repair of chromosome damage and in genome evolution (Dubnau, 1991). The induction of recA at competence is consistent only with the latter two hypotheses. On the other hand, the observation of a co-ordination of competence development is difficult to fit in with the repair hypothesis. Co-ordination would be most consistent with the hypothesis that transformation has evolved as a means of increasing fitness of a bacterial population by facilitating the acquisition of new genetic traits. As mentioned above, each cell in a culture can take up several independent fragments randomly and can therefore experiment with new combinations of genes. As only a few new combinations are expected to be beneficial, high cell density is important to increase the probability of the appearance of rare combinations. From this point of view, the rationale for quorum sensing in competence regulation would be to ensure that a sufficient number of pneumococcal cells experiences the process.

On the other hand, the advantage of a tight regulation of competence development, as observed in S. pneumoniae, would be twofold. First, interspecies gene exchanges can be detrimental, either because of modifications of regulatory regions or because the new activity interferes with cell metabolism (e.g. in the case of PBPs, modification of a single pbp gene can lead to dramatic changes in peptidoglycan structure). This represents a biological cost, particularly for a species in which transformation is highly efficient. A tight regulation of transformation would lower this cost. Second, modulating the evolutionary rate in response to change in environmental conditions would provide a potential for the fine tuning of the rate of sequence diversification.

Transformation in response to environmental changes?

What evidence connects competence induction to changes in environmental conditions? It has long been recognized that competence development responds to aspects of metabolism and environment. Under laboratory conditions, competence is influenced by temperature, the initial growth medium pH or the concentrations of magnesium and calcium. In addition, substitution of maleate for phosphate buffer and zinc or manganese starvation prevent competence induction (Håvarstein and Morrison, 1999). These observations hint at a regulatory link between metabolism and competence development. Another potentially interesting feature is the location of comCDE (and comX) very close to ori, the putative origin of chromosome replication (Gasc et al., 1998). It is known that, in E. coli, the frequency of initiation at ori is dependent on the richness of the growth medium. In S. pneumoniae, a change in initiation frequency would directly affect the comCDEcomX gene copy number per cell. It is tempting to speculate that the comori co-location has evolved as a means of sensing the rate of replication and adjusting competence to this parameter.

Cell genotype also affects endogenous competence induction (Hakenbeck et al., 1999; Håvarstein and Morrison, 1999). One of the most intriguing observations is the effect of oligopeptide permease mutations. Several lines of evidence have indicated that sensing of oligopeptides could be important for triggering competence (Claverys et al., 2000). Particularly striking was the observation that quorum sensing was severely affected in a mutant lacking all three oligopeptide-binding lipoproteins, as competence developed in this mutant at a > 50-fold reduced cell density (Alloing et al., 1998). It was suggested that uptake of oligopeptides acts as a central element of metabolic regulation in S. pneumoniae by providing information on nutrient availability. Exhaustion of nutrients (or the presence of inhibitory compounds) could be detected through the peptide permease. A central transcriptional regulator sensing amino acid pools could serve to integrate the signal(s) (Claverys et al., 2000).

In connection with these observations, what could be the significance of the cell density-dependent control of competence? Although competence develops abruptly during the exponential phase of growth, a change in growth rate is frequently detected at about the same time (our unpublished observations), suggesting some modification in cell metabolism. Taking into account the multiple auxotrophy of S. pneumoniae, the cell density signal may thus constitute a ‘near-starvation’ signal. An alternative hypothesis relies on the biological meaning of the critical cell density at which competence develops under laboratory conditions. How does it compare with the cell density in the nasopharynx, under normal carrier state conditions? If the cell density critical for competence is close to that which triggers host defences, competence could then serve for adaptation of S. pneumoniae to host defence-generated stress. Future progress in our understanding of the pneumococcal carrier state and the unravelling of the regulation of competence and of virulence will undoubtedly answer these questions.

Comparison with the adaptive ‘strategy’ of E. coli

It is interesting to compare the situation in S. pneumoniae (scenario A, Table 1) with that in E. coli (scenario B, Table 1). Specific induction of recA in connection with recombination has been reported in both species, and they are both equipped with evolutionarily related MMR systems (for a review, see Claverys and Lacks, 1986). However, despite this seemingly identical situation, the two species differ in a number of points. First, the induction of recA occurs independently of and precedes DNA uptake in S. pneumoniae, which, as pointed out above, suggests a fine tuning of the transformation machinery to favour recombination. In E. coli, recA induction is subsequent to conjugation and most probably results from a slowing down of recombination between diverged DNA sequences, which in turn induces the SOS system (Matic et al., 1996; Taddei et al., 1997). This is more consistent with repair of potential ‘lesions’ than with optimization of recombination. Second, although MMR does not (and has not been tuned to) prevent interspecies recombination in S. pneumoniae, the frequency of recombination between E. coli and Salmonella typhimurium is reduced as much as 1000-fold in MMR-proficient cells. The latter observation led to the conclusion that the major limitation to gene exchange in bacteria is genomic divergence (Matic et al., 1996). Third, there is no SOS-induced mutagenic response in S. pneumoniae, probably because of the lack of error-prone repair function (Martin et al., 1995), which suggests that mutation has a minor contribution to genetic plasticity in S. pneumoniae.

Table 1. . Genetic variability of S. pneumoniae and E. coli: similar casting but different scenarios. a. No species identifier sequence is required.b. The SOS response.c. MMR deficient mutants (mut) are frequent (over 1%) in natural populations of E. coli (Matic et al., 1997; and references therein). It is not known whether natural populations of S. pneumoniae contain a high proportion of hex mutants.Thumbnail image of

The respective contributions of mutation and interspecies recombination to genetic variability is difficult to assess in the case of E. coli. MMR and SOS systems have antagonistic functions with respect to both types of events in this species. Under laboratory conditions, inhibition of interspecies recombination indicates that MMR is dominant over SOS. A fine tuning of both systems would therefore be required to achieve maximal genetic flexibility in response to stress. Although regulation of SOS is well documented, the evidence for regulation of MMR (Taddei et al., 1997) is really quite weak. Rather, the observation that populations of commensal and pathogenic E. coli contain a high proportion of MMR-deficient strains (Matic et al., 1997; and references therein) suggests ‘random’ inactivation of MMR as an alternative to regulation.

The comparison between S. pneumoniae and E. coli indicates that, although both species might respond to stress by increasing their genetic variability, they achieve this goal through completely different strategies with respect to the role of recombination, SOS and MMR systems. The overall picture emerging from this comparison is that of a programmed variability for S. pneumoniae, possibly involving the entire pneumococcal population in a given host, as opposed to a ‘random’ variability, efficient in only a fraction of the population, for E. coli (Table 1).

Transformable species and genetic plasticity

Does transformation play a similar role with respect to genetic variability in other transformable species? The answer is probably yes for many oral streptococcal species, including close relatives of S. pneumoniae. On the other hand, in H. influenzae or N. gonorrhoeae, the genetic flexibility conferred by transformation is clearly restricted by the existence of species identifier sequences for DNA uptake. B. subtilis represents an interesting intermediate situation. Although only a small fraction of B. subtilis cultures becomes competent (Dubnau, 1991), SOS-induced mutagenesis (error-prone repair) is operative in this species (Yasbin et al., 1993). This suggests that mutation and DNA exchange could both contribute to genetic variability in the case of B. subtilis. Thus, even among transformable species, the relative contribution of mutation and of horizontal transfer to genetic flexibility appears to be organism dependent. This precludes generalizations based on the study of a unique ‘model’ organism.

Acknowledgements

We thank past members of the team whose work contributed to this review. This work was supported in part by a Contrat de Recherche Externe (no. 920102) from the Institut National de la Santé et de la Recherche Médicale, by a Subvention (no. RECH/9407531) from the Région Midi Pyrénées and by fellowships from the Ministère de l'Enseignement Supérieur et de la Recherche and from the Association pour la Recherche contre le Cancer.

Ancillary