Detection of Large Numbers of Pneumococcal Virulence Genes in Streptococci of the Mitis Group

Abstract

Seven streptococcal isolates from the mitis group were analyzed for the presence of pneumococcal gene homologues by comparative genomic hybridization studies with microarrays based on open reading frames from the genomes of Streptococcus pneumoniae TIGR4 and R6. The diversity of pneumolysin (ply) and neuraminidase A (nanA) gene sequences was explored in more detail in a collection of 14 S. pseudopneumoniae and 29 mitis group isolates, respectively. The mitis group isolates used in the microarray experiments included a type strain (NCTC 12261), two S. mitis isolates from the nasopharynxes of children, one S. mitis isolate from a case of infective endocarditis, one S. mitis isolate from a dental abscess, and one S. oralis isolate and one S. pseudopneumoniae isolate from the nasopharynxes of children. The results of the microarray study showed that the 5 S. mitis isolates had homologues to between 67 and 82% of pneumococcal virulence genes, S. oralis hybridized to 83% of pneumococcal virulence genes, and S. pseudopneumoniae hybridized to 92% of identified pneumococcal virulence genes. Comparison of the pneumolysin, mitilysin (mly), and newly identified pseudopneumolysin (pply) gene sequences revealed that mly and pply genes are more closely related to each other than either is to ply. In contrast, the nanA gene sequences in the pneumococcus and streptococci from the mitis group are closely clustered together, sharing 99.4 to 99.7% sequence identity with pneumococcal nanA alleles.

Streptococcus pneumoniae, also known as the pneumococcus, is an important human pathogen causing high levels of mortality worldwide. Diseases caused include meningitis, pneumonia, and otitis media. S. pneumoniae is a member of the viridans group streptococci (VGS) and is closely related to Streptococcus mitis and Streptococcus oralis, sharing >99% homology by 16S rRNA gene analysis (13, 23). Recently, a subset of “atypical” pneumococci were identified that were distinct from both S. mitis and S. pneumoniae by housekeeping gene analysis (48). It is likely that these isolates belonged to the species Streptococcus pseudopneumoniae, which has since been identified, with isolates appearing phenotypically similar to S. mitis but undistinguishable from S. pneumoniae in genotypic tests (2). As a result of the close relationships between these species, clear identification can be problematic. A recent study showed that definitive species identification can be achieved by phylogenetic analysis of 4 housekeeping genes. Furthermore, although it has been reported that comparison of superoxide dismutase genes (sodA) was sufficient to identify most species, this did not distinguish between S. pneumoniae and S. pseudopneumoniae isolates (19).

Despite the close genetic relationships between these species, while S. pneumoniae can be a severe pathogen, S. oralis and S. mitis are usually commensal organisms of mucous membranes. However, on rare occasions, they are able to cause opportunistic disease in immunocompromised patients, particularly those with replacement heart valves, organ transplants, or cancer (4, 11, 28).

The reasons for the dramatic difference in pathogenicity observed between these species are not fully known, although the pneumococcus has a large number of well-characterized virulence factors with known roles in pathogenesis, and these may account for the differences in virulence between these species. However, VGS are naturally transformable, which allows exchange of genetic information between species by homologous recombination (16, 30). Indeed, an increase in penicillin resistance in the pneumococcal population has been attributed to acquisition of resistant penicillin-binding protein variants from S. mitis and S. oralis, resulting in a “pool” of penicillin resistance genes being transferred between these species (9, 12, 38, 41). Furthermore, homologous recombination with related VGS species has been observed in the pneumococcal virulence genes encoding IgA protease and neuraminidase A (nanA) (25, 37), and homologues of pneumococcal virulence factors such as pneumolysin (Ply) and autolysin A (LytA) have been identified in related VGS species (20, 32, 48).

Many pneumococcal virulence factors are currently being considered vaccine candidates for a protein-based pneumococcal vaccine affording cross-serotype protection. However, the presence of homologues of pneumococcal virulence genes in related species may allow evasion of a vaccine targeting a particular virulence factor by acquisition of a gene variant by homologous recombination (6, 36). As a result, it is important not only to increase our understanding of the biological differences between closely related commensals and invasive isolates but also to provide insight into selection of appropriate targets for future vaccine design.

The aim of this study was to assess the presence of pneumococcal virulence factors in a cohort of streptococci from the mitis group at a genome-wide level, using comparative genomic hybridizations. Furthermore, the diversity and phylogenetic relatedness present in homologues of the pneumococcal virulence genes and vaccine candidates Ply and NanA were explored.

MATERIALS AND METHODS

Bacterial strains and preparation.

All strains used in the study are listed in Table 1. Strains were grown in brain heart infusion (BHI) broth (Oxoid) and stored as frozen stocks on Protect beads (Technical Service Consultants). Cells were grown in 20 ml of BHI broth until mid-logarithmic phase (optical density at 600 nm [OD₆₀₀], 0.55 to 0.65). Cell lysates for hemolytic and neuraminidase assays were prepared from 20 ml of culture, as described elsewhere (26).

TABLE 1.

Strains used in this study

Strain	Species	Isolation site	Disease	Use	Location of isolation	% Virulence genes presenta
NCTC 12261	S. mitis	Reference isolate		Microarray study	Reference isolate	82.3
R75 A1	S. mitis	Nasopharynx	Commensal	Microarray study	Glasgow, UK	72.7
R77 C4	S. mitis	Nasopharynx	Commensal	Microarray study	Glasgow, UK	71.4
Col	S. mitis	Blood	Endocarditis	Microarray study	Glasgow, UK	77.9
990123	S. mitis	Pus	Dental abscess	Microarray study	Glasgow, UK	67.5
Sv29	S. oralis	Nasopharynx	Commensal	Microarray study	Belgium	83.1
R98 A5	S. pseudopneumoniae	Nasopharynx	Commensal	Microarray study	Glasgow, UK	92.2
RRR468	S. pseudopneumoniae	Unknown	Unknown	Ply study	Germany	NA
RRR471	S. pseudopneumoniae	Unknown	Unknown	Ply study	Germany	NA
RRR472	S. pseudopneumoniae	Unknown	Unknown	Ply study	Germany	NA
R82 E6	S. pseudopneumoniae	Nasopharynx	Commensal	NanA study	Glasgow, UK	NA
R83 E2	S. mitis	Nasopharynx	Commensal	NanA study	Glasgow, UK	NA
R89 G3	S. pseudopneumoniae	Nasopharynx	Commensal	NanA study	Glasgow, UK	NA
R89 G5	S. pseudopneumoniae	Nasopharynx	Commensal	NanA study	Glasgow, UK	NA
R98 A5	S. pseudopneumoniae	Nasopharynx	Commensal	NanA study	Glasgow, UK	NA

^aDetermined by microarray study. NA, not applicable.

Gene sequencing, phylogenetic trees, and species identification.

PCR amplification of ply, sodA, and nanA genes was carried out using flanking primer pairs 27R and S, 39I and J, and 31R and S, respectively (Table 2). Primers used to fully sequence these genes are also found in Table 2. Cleaning and sequencing of nanA genes were carried out at the Molecular Biology Sequencing Unit at the University of Glasgow, as described previously (21). Full gene sequences were aligned and translated into proteins by use of VectorNTI software (Invitrogen). Identification of species was done by sodA alignment as described previously (19). Where necessary, S. pseudopneumoniae isolates were further identified by optochin resistance and bile solubility (2). Phylogenetic trees were created using MEGA4 (molecular evolutionary genetics analysis) software (www.megasoftware.net).

TABLE 2.

Primers used for PCR amplification and sequencing of pneumococcal genes sodA, ply, and nanA

Primer	Name	DNA sequence (5′-3′)	Use
Primers for sodA
39I	sodA forward	TRCAYCATGAYAARCACCAT	PCR amplification and sequencing of sodA gene fragment
39J	sodA reverse	ARRTARTAMGCRTGYTCCCARACRTC	PCR amplification and sequencing of sodA gene fragment
Primers for ply
4T	ply internal forward	GTTGATCGTGCTCCGATGAC	Sequencing of ply gene
4V	ply internal forward	CAATACAGAAGTGAAGGCGG	Sequencing of ply gene
4W	ply internal forward	GATCATCAAGGTAAGGAAGTC	Sequencing of ply gene
9Y	ply internal forward	CGGGATCCGGCAAATAAAGCAGTAAATGACTTT	Sequencing of ply gene
27R	ply external forward	CTTGGCTACGATATTGGC	PCR amplification and sequencing of ply gene
27S	ply external reverse	TACTTAGTCCAACCACGG	PCR amplification and sequencing of ply gene
27T	ply internal reverse	ATAAGTCATCGGAGCACG	Sequencing of ply gene
Primers for nanA
22X	nanA internal forward	GAGGAACAGTATGAATCGGAGTG	Sequencing of nanA gene
31R	nanA external forward	TAACAGTACACCTTGACTGC	PCR amplification of nanA gene
31S	nanA external reverse	GTGTTCGATAAGGATTGAGC	PCR amplification of nanA gene
35E	nanA internal forward	TGACCGCTTCTCCAGCATGC	Sequencing of nanA gene
35F	nanA internal reverse	GAAACCAATGCTTCAAATGG	Sequencing of nanA gene
38P	nanA internal forward	TCCTCAGTAATGCAGGTGGAC	Sequencing of nanA gene
38Q	nanA internal reverse	TGGAGCCTTGTTGACCAATAC	Sequencing of nanA gene
62F	nanA internal forward	CCAGAAGAGGTACAAAAACGTAGTCAAC	Sequencing of nanA gene
62G	nanA internal forward	AGGGTGATCTATACAAGGGTGACCA	Sequencing of nanA gene
62H	nanA external reverse	TGCCTTTGGTTTTCGGAACTTT	Sequencing of nanA gene
62I	nanA internal reverse	CCCCTTCTTGAGCTAAAACAGGAGA	Sequencing of nanA gene
62J	nanA internal forward	AATCTTTACAAAGGTCAGGAATTGATT	Sequencing of nanA gene
62K	nanA internal forward	GCCGTCTTCATTCTAGTGACTGGGG	Sequencing of nanA gene
62L	nanA internal forward	CAACAAACCAAATAAAGATGGAAT	Sequencing of nanA gene
64I	nanA internal forward	GAATAAATGTCTTATTTCAGAAATC	Sequencing of nanA gene
64J	nanA internal reverse	TTATTGTTCTCTTTTTTCCCTAGC	Sequencing of nanA gene

Hemolytic and neuraminidase assays.

Hemolytic assays were carried out as previously described (21). Neuraminidase assays were carried out by a previously described method (29), with modifications. Isolates were grown to mid-log phase (OD₆₀₀ of 0.6) in BHI medium (Oxoid). Induction was achieved by growth in BHI medium plus 50 μg/ml N-acetylneuraminic acid (Neu5Ac). Pellets from 20-ml mid-log-phase cultures were tested for neuraminidase activity, and culture supernatants were also tested, after being concentrated ×20 by use of an Amicon centrifugal filter (Millipore). Sample plates were incubated for 30 min at 37°C. For lysates and supernatants, activity was measured in nanomoles of p-nitrophenol (pNP) released/mg total cell protein/hour. Isolates tested can be found in Table 1.

DNA-DNA microarray studies and validation.

In order to assess the presence of pneumococcal gene homologues in related VGS species, comparative genomic hybridization studies using S. pneumoniae DNA microarrays were performed on 1 S. oralis, 1 S. pseudopneumoniae, and 5 S. mitis isolates (Table 1). These isolates were chosen to represent a range of mitis group isolates from cases of colonization and infection (Table 1). Microarray studies were carried out using S. pneumoniae SPv1.1.0 microarrays, designed and produced by the Bacterial Microarray Group, St. George's University of London (BμG@S). These represent all open reading frames (ORFs) from the genomes of S. pneumoniae isolates TIGR4 and R6. The array design is available in BμG@Sbase (http://bugs.sgul.ac.uk/A-BUGS-14), under accession number A-BUGS-14, and also in ArrayExpress (accession number A-BUGS-14). DNA hybridizations were carried out per standard BμG@S protocols. Test isolate genomic DNA was labeled with Cy5 dye (GE Healthcare), and control TIGR4 genomic DNA was labeled with Cy3 dye, with reciprocal dye swaps performed to eliminate artifacts. In all cases, bacteria were incubated at 37°C in BHI medium to mid-log phase (OD₆₀₀, 0.6) before harvest of DNA for microarrays. Slides were scanned using ScanArray Express (Packard Biosciences), feature extraction and normalization were carried out using BlueFuse for Microarray 3.5 software (Bluegnome Ltd.), and gene lists were created using GeneSpring GX 7.3.1 (Agilent Technologies) software. Under the stringent microarray conditions used, sequence identities above 80% will result in hybridization, whereas identities below this level will likely result in nonhybridization. Validation of comparative genomic hybridizations was done by PCR, using the primers used to design the microarray probes.

Nucleotide sequence accession numbers.

The complete sequences of Ply and NanA homologues were deposited in GenBank under accession numbers FJ440109 to FJ440116.

Microarray data accession number.

Fully annotated microarray data have been deposited in BμG@Sbase under accession number E-BUGS-82 (http://bugs.sgul.ac.uk/E-BUGS-82) and may also be found in ArrayExpress (accession number E-BUGS-82).

RESULTS

Presence of pneumococcal gene homologues in streptococci of the mitis group.

The results of the mitis group microarray study showed that a large number of pneumococcal genes had homologues in the 7 isolates tested. The five S. mitis isolates tested had homologues to between 72 and 83% of pneumococcal genes, as determined by hybridization to the pneumococcal gene probe. Furthermore, the S. oralis isolate hybridized to 84% of the genes, and the S. pseudopneumoniae isolate hybridized to 91.4% of them. Comparison of these hybridization percentages to those for pneumococcal isolates hybridized in a previous study (unpublished data) showed differences in genetic content between the species, implying that S. pseudopneumoniae is more closely related to S. pneumoniae than the other species are (Fig. 1).

FIG. 1.

Comparison of percentages of genes hybridized to pneumococcal microarrays for different VGS species (reference genome, S. pneumoniae TIGR4). S. pneumoniae hybridization results are representative of hybridization values from previous studies of S. pneumoniae strains, using the same arrays, in our laboratory.

Pneumococcal virulence gene homologues present in streptococci of the mitis group.

For comparison, pneumococcal virulence genes were considered to be the 231 genes identified in a signature-tagged mutagenesis screen as essential for pneumococcal pneumonia in a murine model (15). Our results showed that homologues to 218/231 virulence genes were present in at least one mitis group isolate, many of which are also being considered as pneumococcal vaccine candidates (42). The presence of homologues to the most significant pneumococcal virulence genes is shown in Fig. 2, while a full spreadsheet, with hybridization results from all 7 isolates aligned to the pneumococcal gene list, can be found in Fig. S1 in the supplemental material. A number of capsular genes were present in the S. oralis (Sv29) and S. mitis (R77 C4) isolates. The capsular genes were not those of any known pneumococcal serotype, and the strains did not react with pneumococcal typing serum (data not shown). Conversely, 13/231 essential virulence genes did not hybridize with any of the mitis group isolates tested (Table 3). It is possible that these genes are unique to the pneumococcus but also that high levels of variation prevented the hybridization of these genes to the corresponding microarray probes. Many of the genes were hypothetical in function, but included were the transcriptional regulator gene rlrA, present in the RlrA islet encoding the pneumococcal pilus (3), and genes in the PsrP islet (33). Both of these islets were fully absent from all VGS isolates tested.

FIG. 2.

Presence of pneumococcal virulence gene homologues in related VGS species by comparative genomic hybridization. The presence of a homologue by hybridization is represented by a black box, and its absence is shown by a white box. The final five genes shown are essential virulence genes of unknown function present on identified regions of pneumococcal diversity (33).

TABLE 3.

List of pneumococcal virulence genes not identified in any of the VGS isolates tested, with names and functions where known

Gene no.	Name	Role/function
SP0461	rlrA	Positive regulator of pneumococcal pilus expression (3)
SP1143		Hypothetical protein
SP1321	ntpK	V-type sodium ATP synthase, subunit K
SP1328		Sodium:solute symporter family protein
SP1343		Prolyl oligopeptidase family protein
SP1344		Conserved hypothetical protein
SP1433		AraC family transcriptional regulator
SP1434		Permease/ATP-binding protein of ABC transporter
SP1760		Hypothetical protein
SP1770		Glycosyl transferase, family 8
SP1771		Glycosyl transferase, family 2/glycosyl transferase family 8
SP1772	psrP	Cell wall surface anchor family protein
SP1952		Hypothetical protein

Pneumolysin homologues in streptococci of the mitis group.

Previous work characterized a Ply homologue termed mitilysin (Mly) in a subset of S. mitis isolates, with the protein sequences of 7 alleles reported (20). We report the characterization of a further related toxin in S. pseudopneumoniae isolates, referred to as pseudopneumolysin (pPly). A collection of 14 S. pseudopneumoniae isolates were screened for the presence of pPly by PCR, with 13 positive isolates (data not shown). Of these, three were chosen for further study (Table 1). Full pply gene sequencing of 3 isolates showed that compared to wild-type ply allele 1 from isolate D39 (21, 45), mutations were present at 74 positions. Translation of these sequences resulted in only 7 amino acid changes, with many synonymous mutations present, and 3 unique pPly alleles. Of the 7 amino acid changes, 2 were shared by both Ply and Mly (N14D and K224R), while another further 4 were shared only by Mly (A2T, H386Y, D398N, and V423I). Culture lysates of these 3 strains were shown to have hemolytic activity, confirming the expression of the pPly toxin (data not shown). The genetic relatedness of the three toxins, i.e., Ply (21), Mly (20), and pPly, was explored using MEGA4 software to construct a minimum evolution tree to show the relationships between the toxin genes at the DNA level (Fig. 3). The results showed that while the ply alleles clustered together, the mly and pply alleles were part of the same branch of the tree. Therefore, in terms of genetic relationships, the mly and pply genes are more closely related to each other than either is to ply.

FIG. 3.

Minimum evolution tree of ply DNA alleles and homologues in related streptococcal species. ply and mly alleles were previously described (20, 21). The tree was drawn using MEGA4 software and demonstrates the relatedness of these toxins at the DNA level.

Neuraminidase A homologues in streptococci of the mitis group.

After PCR screening of 29 mitis group isolates of various species, nanA gene homologues were discovered in all five S. pseudopneumoniae isolates tested and in a subset of five S. mitis isolates. The nanA genes from four S. pseudopneumoniae isolates and one S. mitis isolate were fully sequenced, and the translated protein sequences were compared to NanA alleles discovered in S. pneumoniae. The results showed that while high levels of variation were present within these alleles, they displayed striking similarities with the NanA alleles of S. pneumoniae, sharing 99.4 to 99.7% amino acid sequence identity with known pneumococcal alleles (Fig. 4). Furthermore, all 5 of these isolates displayed neuraminidase activity, which was likely due to the expression of the NanA homologue. This activity was comparable to the neuraminidase activity detected in D39 and other pneumococcal isolates tested, while the S. mitis strains that were negative for nanA showed only residual activity in the assay (data not shown). Therefore, in contrast to the case for Ply, there was very little difference between the NanA proteins found in the pneumococcus and those found in the related species S. mitis and S. pseudopneumoniae.

FIG. 4.

Four VGS NanA alleles identified in isolates of S. mitis and S. pseudopneumoniae. Alleles 1 to 3 were aligned with a pneumococcal NanA sequence from a clinical isolate of serotype 19A, ST1201 (GenBank accession no. FJ440122.1), and allele 4 was aligned with the pneumococcal NanA allele of sequenced strain G54. Row 2 represents the amino acid position in the pneumococcal NanA protein, and row 3 represents the amino acid present in the pneumococcal NanA allele. Next to each strain is the mutant amino acid present at each position along the protein. The amino acid sequence identity between the NanA alleles is noted as a percentage.

The genetic relatedness of the NanA alleles from the pneumococcus and the mitis group isolates was explored using MEGA4 software to construct a minimum evolution tree to show the relationships between the nanA genes at the DNA level (Fig. 5). Analysis showed that the pneumococcal isolates clustered by allele in the tree. However, the nanA alleles from S. mitis and S. pseudopneumoniae are not separated from the pneumococcal alleles and are closely linked to their nearest pneumococcal relatives. Therefore, unlike the case for ply, no evolutionary distinction between these species was observed when we examined the NanA enzyme.

FIG. 5.

Minimum evolution tree of nanA DNA alleles and homologues in related streptococcal species. Pneumococcal nanA alleles include those from sequenced pneumococcal strains. The tree was drawn using MEGA4 software and demonstrates the relatedness of these enzymes at the DNA level.

DISCUSSION

Homologues of a large proportion of pneumococcal virulence genes are present in isolates of S. mitis, S. oralis, and S. pseudopneumoniae. Although some degree of similarity was to be expected, due to the close relationship to S. pneumoniae (2, 13, 14, 24), the large number of homologous virulence genes raises interesting biological questions in the context of the differing behaviors and pathogenic potentials of these microorganisms in the same niche. S. mitis and S. oralis are commensal organisms of mucous membranes, rarely causing human disease, yet the pneumococcus is a major human pathogen. The finding that a number of virulence gene islets, including the RlrA and PsrP islets, appear absent from related commensal species is interesting, although further study across a broader selection of related VGS isolates would be required to confirm the virulence gene islets as unique to pneumococci. It is also possible that rather than being absent, these regions are very diverse in the mitis group population and did not hybridize to the pneumococcal microarray probes. The RlrA islet and the PsrP islet have been shown to be involved in pneumococcal virulence (3, 33), although they are not universally present in the pneumococcal population. Indeed, the RlrA islet was shown to be present in only 27 to 31% of pneumococci (1, 31), while the PsrP islet was also shown to be variably present in S. pneumoniae isolates (Donald Inverarity, personal communication). Therefore, it is perhaps unsurprising that these islets are absent from related VGS species.

A number of other regions were identified as being fully absent from mitis group isolates, and although less is known about the roles of these regions, many essential virulence genes were located within them (15). These include the genes encoding a transcriptional regulator of the AraC family, a sodium symporter, and an ABC transporter, as well as a number of essential genes of unknown function. However, many of the proteins in these regions remain hypothetical, and further study may uncover genes of interest as pneumococcal vaccine candidates if they are found to be fully absent from mitis group isolates.

There was increased hybridization between the S. oralis isolate and the S. pneumoniae array compared to that for the S. mitis strains. This was unexpected but may have been due to the fact that the S. oralis isolate was chosen for study as an isolate of interest because it was seen to possess at least some pneumococcal capsular genes (data not shown). Therefore, it is possible that the S. oralis isolate used in this study is not representative of the S. oralis population, but the result shows that this level of similarity is possible between strains distinguished as S. pneumoniae and S. oralis. Hybridization of further S. oralis strains would shed light on this subject.

There are a number of drawbacks to the use of microarray technology for comparative genomic hybridization, and these must be appreciated in this study. First, only information on the presence or absence of genes present in the reference isolate (in this case, S. pneumoniae TIGR4) is available. Furthermore, nonhybridization of a particular gene does not confirm its absence from the test genome, as it may be a result of diversity in the target region of the probe. This is particularly important for comparative genomic hybridization across species, where levels of diversity are likely to be higher. As a result, further study is required to confirm whether nonhybridized genes in this study are indeed absent or simply display increased diversity in the tested isolates. For example, this is likely to be the case for the IgA proteases, which have been shown to display high levels of variation between mitis group isolates (37) and did not hybridize to the pneumococcal probes for any isolates. However, this does allow a certain degree of certitude for positive results, which must share at least 80% sequence identity with the array probes.

Two recent publications have also begun to address the genetic relationships between S. pneumoniae and related VGS species. One used microarrays generated from a sequenced S. mitis isolate to hybridize DNAs from a number of related species (10). The second study was a study of population genetics further exploring the evolutionary relationships between these species (24). Our results are in agreement with both studies, showing that there is far greater diversity in the S. mitis species than in S. pneumoniae and that many pneumococcal virulence factors are present in this population. However, there are a number of contrasting results presented. Denapaite and colleagues were unable to identify virulence factors such as pspA and pspC in the S. mitis population (10). This may be due to the diverse nature of these genes (5, 18) or simply to the variation in virulence gene content in the S. mitis population. Furthermore, Kilian and colleagues were able to identify capsular genes in the sequence of S. mitis strain NCTC 12261, while these were not identified in our microarray study (24). This may result from increased diversity in this region compared to that in S. pneumoniae, as a sequence identity of <80% would result in nonhybridization to the probe under our conditions.

The presence of at least a partial capsule locus in two isolates tested here, an S. mitis and an S. oralis isolate, demonstrates that remnants of capsule loci may be present in other VGS species. This is supported by the findings of Kilian and colleagues, who were able to identify a capsule locus in an S. mitis strain, as discussed previously (24). It would therefore be very relevant to carry out a similar study of capsule locus presence/diversity and expression in related VGS species compared to the pneumococcus, whose capsule locus is very well characterized. This study could provide further important information in relation to the use of the current pneumococcal vaccine, which targets the polysaccharide capsule, as the presence and expression of capsular genes in related species could allow vaccine escape to occur.

The results of the comparative genomic hybridization analysis allowed selection of homologues of pneumococcal vaccine candidates for further study in order to assess the diversity present in and genetic relationships between these genes. Recently, we identified and characterized a Ply homologue, termed mitilysin (20), in a subset of S. mitis isolates. Our results identify and further characterize a Ply homologue in S. pseudopneumoniae which is more closely related to mly than to ply. This segregation is in agreement with phenotypic traits such as bile solubility, lack of capsule, and optochin resistance, which all link S. pseudopneumoniae to S. mitis (2). ply is one of the genes used routinely to identify S. pneumoniae by PCR-based detection in clinical samples and, along with lytA, is considered one of the best targets to distinguish the pneumococcus (8). These genes have been shown to be well conserved in the pneumococcus (21, 47), but the identification of relatively similar homologues in related VGS species in this and other studies shows that these genes may not be ideal candidates for identifying the pneumococcus by PCR (10, 24, 39).

In comparing the sequences of nanA gene homologues identified in S. mitis and S. pseudopneumoniae isolates to pneumococcal nanA alleles, there was no species segregation, in contrast to the case for the ply alleles. These results further the argument that a smooth continuum may exist between these species, with pneumococcal virulence genes and homologues dispersed in isolates identified as S. mitis or S. pseudopneumoniae (14). Furthermore, the fact that nanA alleles are distributed in a non-species-specific manner may indicate that there has been genetic exchange between the species at this locus, which would have occurred after the species diverged from a common ancestor. Indeed, genetic exchange at the nanA locus in the pneumococcus has been documented between S. pneumoniae and S. oralis, in agreement with this hypothesis (25).

Our results demonstrate the presence of a NanA homologue in a subset of S. mitis and S. pseudopneumoniae isolates. However, sialidase genes have previously been found in S. oralis, a close relation of the pneumococcus (7, 25). The sequences of the protein alleles reported in our work were very similar to that of pneumococcal NanA, and this may indicate that tight control of the NanA structure is more important for biological activity than that of Ply and the Mly homologue. NanA has a highly defined role in pneumococcal colonization (29, 43, 44), and the presence of NanA in commensal relatives probably highlights its role in adherence and colonization of mucosal surfaces as well. In contrast, Ply has a much less clear role in pneumococcal colonization but has well-defined roles in invasive diseases such as pneumonia and meningitis (17, 22, 40, 46). Therefore, it may be expected that NanA would be important for the colonization of commensal organisms, whereas the role of a cytolytic toxin remains unclear. However, the presence of these homologues in the mitis group and their possible transfer into the pneumococcal population under the selective pressure of a vaccine demonstrate that the study of virulence factor homologues in related species is important in the context of selecting protein vaccine candidates. Overall, the diversity present in homologues of Ply and NanA remains relatively low, implying that the presence of these homologues in related species may not pose a problem if Ply or NanA is targeted with a pneumococcal vaccine. Further study, including study of further mitis group streptococcal isolates for diversity analysis and antigenic studies, would help to confirm this conclusion.

In conclusion, genomic microarray analysis has demonstrated the presence of a large number of important pneumococcal virulence genes in the commensal mitis group streptococcal population. Specific virulence factors such as RlrA and PsrP appeared absent from related species and may provide a useful starting point from which to explore mechanisms explaining the different biological behaviors between these groups of organisms. This approach may also inform future vaccine strategies for selection of targets less likely to be compromised by horizontal recombination between species.