Characterization of Recombinant Fluoroquinolone-Resistant Pneumococcus-Like Isolates

Abstract

Fourteen fluoroquinolone-resistant streptococcal isolates with recombinant DNA topoisomerase genes, preliminarily identified as pneumococci, were further characterized using phenotypic and genotypic approaches. Phenotypic tests classified them as atypical pneumococci. Phylogenetic relationships were analyzed by using the sequences of seven housekeeping alleles from these isolates and from isolates of Streptococcus pneumoniae, Streptococcus mitis, Streptococcus oralis, and Streptococcus pseudopneumoniae. Four isolates grouped with S. pneumoniae, seven grouped with S. pseudopneumoniae, and three grouped with S. mitis. These results generally agreed with those obtained with an optochin susceptibility test and with the organization of the atp operon chromosomal region, encoding the F_oF₁ H⁺-ATPase (the target of optochin). All seven isolates grouping with S. pseudopneumoniae share the same spr1368-atpC-atpA gene order; all four grouping with S. pneumoniae share the spr1368-IS1239-atpC-atpA order, and two out of the three grouping with S. mitis share the spr1284-atpC-atpA order. In addition, evidence for recombination within the seven housekeeping alleles of the S. pseudopneumoniae population was provided by several methods: the index of association (0.4598, P < 0.001), the pairwise homoplasy index, and the split-decomposition method. This study confirms the existence of pneumococci among the alpha-hemolytic streptococci with DNA topoisomerase genes showing a mosaic structure and reveals a close relationship between atypical pneumococci and S. pseudopneumoniae.

INTRODUCTION

Streptococcus pneumoniae (the pneumococcus), an important pathogen associated with various human diseases such as pneumonia, otitis media, septicemia, and meningitis, is a member of the mitis group of streptococci (MGS), which includes commensal microorganisms of the oral cavity. MGS species most closely related to S. pneumoniae are Streptococcus mitis, Streptococcus oralis, and the recently reported species Streptococcus pseudopneumoniae (1). These species are especially difficult to differentiate using the nucleotide sequences of their 16S rRNA genes since their identity is greater than 99% (1, 2). Given that S. pneumoniae is often isolated along with other upper respiratory tract streptococci that have different clinical significance, accurate discrimination among closely related species is essential for proper diagnosis and treatment of pneumococcal infections.

Four phenotypic characteristics are classically used for the presumptive identification of S. pneumoniae: colony morphology (colonies showing alpha-hemolysis on sheep blood agar), optochin (OPT) susceptibility, deoxycholate (DOC) solubility (commonly referred to as bile solubility), and a positive reaction with antipneumococcal polysaccharide capsule antibodies. Although their colony morphologies can be similar, nonpneumococcal streptococci are classically OPT resistant and bile insoluble and do not react with specific pneumococcal antipolysaccharide antibodies. Isolates that give inconsistent results in one or more of these tests are named atypical pneumococci (3–5). Moreover, S. pseudopneumoniae isolates were found to be resistant to OPT when incubated in an atmosphere enriched in CO₂ but were OPT susceptible when incubated in ambient atmosphere (1), making proper identification even more difficult.

In addition to standard phenotypic tests, various genotypic methods have been used to identify pneumococcal isolates, such as pulse-field gel electrophoresis (6), PCR tests for virulence genes (7), and multilocus sequence typing (MLST) (8). Additionally, MLST results can be used to perform phylogenetic analyses, referred to as multilocus sequence analysis (MLSA). With this approach, using phylogenetic trees inferred from concatenated sequences of MLST genes (9, 10) or alternative sets of genes (11, 12), it is possible to identify clusters among a large number of closely related strains.

Most members of the MGS are naturally transformable. Homologous recombination is believed to play a role in the evolution of these bacteria, which is reflected in mosaic structures in gene sequences. Recombination processes yielding antibiotic resistance have been identified in the genes of penicillin-binding proteins (13), rpoB genes of rifampin-resistant strains (14), and DNA topoisomerase genes of fluoroquinolone-resistant isolates (15, 16). Recombination, together with the spread of a few international clones, is a main factor involved in antibiotic resistance in S. pneumoniae. Fluoroquinolone (Fq) resistance occurs mainly by alteration of intracellular drug targets, i.e., DNA topoisomerase IV and DNA gyrase. Fq resistance is acquired by point mutation as well as by intraspecific (17) or interspecific recombination with MGS (15–17). We along with others have reported that specific Fq-resistant (Fq^r) mutations confer a fitness cost to S. pneumoniae (18, 19). However, compensation of this fitness cost in isolates carrying recombinant topoisomerase genes has been observed (20). In this scenario, the future spread of Fq^r recombinant clones is not unexpected. These recombinant isolates are scarce, given the frequency of Fq^r pneumococci (lower than 2.6%) and that of recombinants among Fq^r isolates (lower than 14%) (15, 21, 22).

Fq^r recombinant isolates, which were preliminarily identified as pneumococci, showed high levels of nucleotide variation (>4%) in some of their topoisomerase genes (parC, parE, and gyrA), and the genetic organization of the parE-parC chromosomal region was different from that of S. pneumoniae strains, indicative of the probable recombination origin of these Fq^r isolates (15). Since these isolates carry nonpneumococcal sequences, accurate identification is essential. The present study was carried out to accurately identify these isolates at the species level and to assess the usefulness of some currently used tests for the identification of pneumococcal isolates. Our collection of 14 isolates and several control strains were further characterized. We used a combination of phenotypic and genotypic approaches, which included routine tests for pneumococcal identification, MLSA, atpCA sequencing, detection by hybridization of the genes coding for pathogenic determinants (autolysin and pneumolysin), and detection of the ant gene, a gene specific to nonpneumococcal MGS (15). Additionally, we explored the genetic relationship of these isolates with the most closely related MGS species.

We report here that some of the isolates analyzed are pneumococci which have acquired portions of DNA topoisomerase genes from MGS. However, most of the isolates studied cannot be considered pneumococci although they have phenotypic and genetic characteristics that are frequently associated with S. pneumoniae. The results generated in the present study indicate that, due to the occurrence of interspecific recombination, no single methodology can be used to provide an accurate identification to the species level of MGS isolates.

MATERIALS AND METHODS

Identification tests.

Isolates were confirmed as S. pneumoniae bacteria by standard methods, and serotypes were determined by a Quellung reaction at the Spanish Reference Laboratory. OPT susceptibility was tested by disk diffusion, using commercially available OPT disks (5-μg, 6-mm diameter disk; Oxoid, Hampshire, England) applied to blood agar plates (Trypticase soy agar supplemented with 5% sheep blood). Plates were incubated at 37°C in either 5% CO₂ atmosphere or in ambient atmosphere as described previously (1). Isolates were considered OPT resistant if they displayed inhibition zones smaller than 14 mm. For detection of the rRNA genes, an AccuProbe S. pneumoniae culture identification test (Gen-Probe, San Diego, CA) was used according to the manufacturer's instructions. S. pneumoniae ATCC 6303, S. pneumoniae R6, S. mitis NCTC 12261^T, S. oralis NCTC 11427^T, and two S. pseudopneumoniae strains (CCUG49455^T and CCUG48465) were used as controls.

MLST, multilocus sequence analysis (MLSA), and phylogenetic analysis.

A multilocus sequence typing (MLST) scheme developed for S. pneumoniae was used (8). Tentative allele number assignment was performed by comparing sequences to those in the pneumococcal MLST database (http://spneumoniae.mlst.net/). The population structure was analyzed using eBURST (23). This program groups related sequence types (STs) into clonal complexes (CCs), identifies the probable ancestor of each CC, and outputs a graphical representation of these relationships. The most restrictive group definition was used, in which STs were included within the same group only if they shared a minimum of six of the seven MLST loci with at least one other ST in the group. Allelic combinations not described in the MLST database were assigned new ST numbers. Phylogenetic analysis was performed with the MEGA program (version 4.0.2), using the minimum-evolution algorithm (24).

Recombination analysis.

The index of association (I_A) was calculated using the LIAN software program, version 3.5 (www.pubmlst.org). Significant I_A values were determined using the Monte-Carlo method with 1,000 resamplings. The Neighbor-Net implemented in the software program SplitsTree, version 4.0 (25), with 1,000 bootstrap replicates was used to create the phylogenetic network for the individual loci and for concatenated sequences. Further, we used the pairwise homoplasy index (PHI) (26) implemented in SplitsTree, version 4.0, to test the role of past recombination in generating allelic variation.

Southern blot identification of strains.

Restriction fragments carrying lytA (autolysin) and ply (pneumolysin) DNA probes and PCR products carrying the ant probe (a homolog of genes coding for aminoglycoside adenylyltransferases) were obtained as described previously (4, 16). Probes were labeled with a Phototope-Star detection kit (New England BioLabs). Southern blotting and hybridization were carried out according to the manufacturer's instruction.

Organization and sequence of the atp chromosomal region.

PCR amplifications were performed using 0.5 to 1 U of Thermus thermophilus thermostable DNA polymerase (Biotools), 0.1 μg of chromosomal DNA, 1 μM each synthetic oligonucleotide primer, and 0.2 mM each deoxynucleoside triphosphate in a final volume of 50 μl in the buffer recommended by the manufacturer. Amplifications were achieved with an initial cycle of 5 min of denaturation at 94°C, followed by 25 to 30 cycles of 1 min of denaturation at 94°C, 1 min of annealing at 55°C, and 1 to 4 min of polymerase extension at 72°C, with a final 8-min extension at 72°C and slow cooling at 4°C. The atpC upstream regions from S. pneumoniae and S. pseudopneumoniae were amplified with oligonucleotides spr1368F (5′-CTGTAAGGTGAGTATGGAGG-3′) and atpC1 (5′-CTAAAATCCTCCCTTTTCTTCTCG-3′), located in spr1368 and atpC, respectively. PCR amplifications yielded fragments of about 3 kb for S. pneumoniae isolates and of about 1 kb for S. pseudopneumoniae isolates. Fragments of about 0.5 kb from S. mitis and S. oralis isolates were amplified with oligonucleotides pepti368 (3) and atpC1, located in spr1284 and in atpC, respectively. PCR amplifications of a 1,499-bp fragment containing atpC, atpA, part of atpB, and 114 bp located upstream of atpC were carried out using conditions and oligonucleotide primers described elsewhere (3). PCR products were purified using a geneJet PCR purification kit (Fermentas) and sequenced on both strands with an Applied Biosystems Prism 377 DNA sequencer.

RESULTS

Phenotypic characterization.

The 14 clinical isolates analyzed in this study were ciprofloxacin resistant (Cip^r; ciprofloxacin MICs of ≥4 μg/ml) and showed high levels (>4%) of nucleotide sequence variation in the quinolone resistance-determining regions (QRDRs) of their parC, parE, and gyrA genes. These 14 isolates represent most (14 out of 16) Fq^r recombinants selected in three epidemiological studies performed by us, one during 1991 to 2001 (six strains), one in 2002 (five strains), and one in 2006 (three strains). All of the isolates except isolate 5305 have been previously described (15, 16, 21, 22). Eleven isolates carried changes in the QRDR of ParC or ParE responsible for Fq resistance. High-level Cip^r isolates (MICs of >8 μg/ml) had additional changes in GyrA (Table 1). Isolates 5305, CipR-71, and CipR-6.74 had an efflux mechanism as the single cause of resistance (data not shown). Among these 14 isolates, 6 (42.9%) were nontypeable, with the remaining 8 belonging to six different serotypes (6A, 6B, 19A, 23F, 24, and 35). Serotype 23F was the most prevalent, accounting for 3 of the 14 isolates (21.4%). Isolates were sent to the Spanish Reference Laboratory, presumptively identified as S. pneumoniae. However, further analysis (data not shown) of their OPT susceptibility and DOC solubility revealed that all except isolate 3180 showed some discrepancies in these tests. Isolate 3180, although nontypeable, showed hybridization with the AccuProbe (Table 1). Moreover, most isolates showed hybridization with the AccuProbe.

Table 1.

Phenotypic and genotypic properties of the strains used in this study

Strain (yr of isolation)	Amino acid change at the QRDR ofa:			Typeb	OPT susceptibility (in CO2, in O2)c	APd	STe	CCf	MLSTh							Hybridization with:			atp region(s)g
	ParC	ParE	GyrA						aroE	gdh	gki	recP	spi	xpt	ddl	lytA	ply	ant
S. pneumoniae R6	None	None	None	NT	IR, S	+	128		7	5	1	5	10	7	15	+	+	−	SPN
S. pneumoniae 6303	None	None	None		S, S	+										+	+	−	SPN
3180 (1994)	S79F	None	S81Y	NT	S, S	+	595	342	7	5	1	1	10	7	15	+	+	+	SPN
CipR-72 (2002)	S79I	None	S81F	6B	IR, S	+	1624	1	5	6	1	2	6	3	1	+	+	−	SPN
CipR-73 (2002)	S79F	None	S81F	19A	IR, S	+	202	2	8	16	19	15	6	40	26	+	+	+	SPN
CipR-6.76 (2006)	S79Y	None	S81F	35	IR, S	ND	1884	81	45	13	6	12	9	14	14	+	+	+	SPN
CCUG49455T	None	None	None	NT	IR, S	+			139	F	E	E	G	F	E	+	+	−	SSP
CCUG48465	None	None	None	NT	IR, S	+			E	E	H	B	F	A	E	+	+	+	SSP
4391 (1997)	S79Y	None	S81F	23F	IR, S	+		S	103	166	B	B	6	A	B	+	+	+	SSP
5237 (1999)	S79N	None	S81Y	23F	IR, S	+		S	A	A	C	C	B	B	C	+	+	+	SSP
5305 (1999)	None	None	None	23F	IR, S	+		S	B	8	D	D	C	C	D	+	+	+	SSP
CipR-71 (2002)	None	None	None	NT	IR, S	+		S	C	B	E	E	D	D	E	+	+	−	SSP
CipR-75 (2002)	S79R	None	S81F	24	IR, S	+		S	139	D	G	G	B	E	6	+	+	+	SSP
CipR-6.39 (2006)	None	D435N	S81F	NT	IR, S	ND		S	F	G	I	H	H	G	101	+	+	+	ND
3870 (1996)	S79F	None	S81Y	NT	R, S	+		S	A	A	A	A	A	1	A	+	+	+	SSP
4589 (1998)	S79F	None	S81F	NT	R, S	+		S	A	A	A	A	A	1	A	+	+	+	SSP
S. mitis 12261T	None	None	None	NT	R, R	−			H	I	K	J	J	I	H	−	−	+	SMI
S. oralis 11427T	None	None	None	NT	R, R	−			I	J	L	K	K	J	I	−	−	+	SOR
CipR-74 (2002)	S79F	None	S81F	NT	S, S	−		S	D	C	F	F	E	46	F	+	+	+	SMI/SOR
CipR-6.74 (2006)	None	None	None	6A	R, R	ND		S	G	H	J	I	I	H	G	+	+	+	SMI/SOR

^aOnly residues involved in Fq resistance are indicated. Underlining indicates that the residue is located in a gene with a mosaic structure.

^bSerotype. NT, not typeable.

^cOPT susceptibility phenotypes were categorized as follows: resistant (R), MIC of ≥6 μg/ml or zone diameter of <10 mm; intermediate resistant (IR), MIC of 3 to 6 μg/ml or zone diameter of 10 to 14 mm; susceptible (S), MIC of ≤1.5 μg/ml or zone size diameter of >14 mm.

^dAP, AccuProbe; +, positive; −, negative; ND, not determined.

^eST, sequence type.

^fCC, clonal complex; S, singleton.

^gGenetic structure of the atp region. SPN, S. pneumoniae; SSP, S. pseudopneumoniae; SMI, S. mitis; SOR, S. oralis. ND, not determined.

^hAlleles assigned numbers have been previously deposited in the pneumococcal database. Alleles assigned letters diverged up to 5% from the closest available match.

To determine if the isolates showed the OPT phenotype described for S. pseudopneumoniae, we examined OPT susceptibility in both ambient air and in a CO₂-enriched atmosphere. While isolates 3180 and CipR-74 were susceptible under both conditions, only one isolate (CipR-6.74) was resistant to OPT in both environments. Unexpectedly, 11 isolates showed identical results to those described for S. pseudopneumoniae; they were OPT susceptible in ambient air and OPT resistant in the presence of 5% CO₂. A similar result was previously reported for the S. pneumoniae R6 strain (3). Therefore, more tests were needed before these 11 isolates were classified as S. pseudopneumoniae.

Genotyping by MLST, MLSA, and eBURST.

Sequence analysis of the seven housekeeping genes included in the S. pneumoniae MLST scheme showed that, among 92 alleles obtained, only 34 had been previously deposited in the pneumococcal database (Table 1, alleles assigned numbers). All other alleles diverged up to 5% from the closest available match (Table 1, alleles assigned letters). Each isolate is defined by seven alleles, the combination of which constitutes an allelic profile defining a specific ST. While STs were assigned to four isolates (3180, CipR-72, CipR-73, and CipR-6.76), no STs could be assigned to the remaining 10 isolates, given that their alleles were 1 to 5% divergent from those of pneumococci included in MLST database. The simultaneous occurrence of four to seven novel alleles in each strain suggested that they were not pneumococci (10). The presence of identical alleles in isolates 3870 and 4589 suggests a clonal origin.

To further examine the relationship between the 14 isolates and other closely related Streptococcus spp., we aligned concatenated 2,751-bp sequences of six of the seven MLST alleles (ddl was not included) with those of 91 S. pneumoniae, S. pseudopneumoniae, S. mitis, and S. oralis strains, including their species type strains. Phylogeny based on MLSA (Fig. 1A) clearly placed our isolates apart from S. pneumoniae, with the exception of four isolates (3180, CipR-72, CipR-73, and CipR-6.76). Three isolates (CipR-74, CipR-6.39, and CipR-6.74) clustered in a subgroup that included the S. mitis type strain, while the seven remaining isolates appeared to be most closely related to S. pseudopneumoniae. The increase in the number of genes included in the analysis, with 4,893-bp concatenated sequences of the MLSA genes plus the 16S rRNA-sodA-rpoB-hlpA genes, provided a result entirely consistent with that provided by MLSA but not significantly improved bootstrap values (Fig. 1B).

Fig 1.

Phylogenetic relationships of isolates. The minimum evolution algorithm in MEGA, version 4.0.2, was used. Only bootstrap values, based on 1,000 replications, exceeding 80% are shown. The scale bar refers to genetic divergence. (A) Tree made with 2,751-bp concatenated sequences of six genes included in the MLSA scheme. Isolates characterized in this work (in bold face) and type strains of individual species were compared with 87 sequences of pneumococcal (black circles) and nonpneumococcal (white squares) streptococci available in the MLST database. (B) Tree made with 4,893-bp concatenated sequences of the six genes of the MLSA scheme plus the 16S rRNA, sodA, rpoB, and hlpA genes. Isolates were compared with S. pneumoniae (ST 128, ST 236, ST 568, and ST 1982), S. pseudopneumoniae (IS7493 and CCUG49455^T), S. mitis (SMI) NCTC 12261^T, and S. oralis (SOR) NCTC 11427^T strains.

To display the relationships among related STs, the 14 isolates were examined by eBURST using the MLST database. Four isolates (CipR-72, CipR-73, CipR-6.76, and 3180) were assigned by eBURST into four clonal complexes, CC1, CC2, CC81, and CC342, respectively. The clonal complexes CC1 and CC2 comprised 1,049 and 257 different STs and included 1,050 and 258 S. pneumoniae isolates, respectively. The 10 isolates that could not be assigned to any clonal complex were defined as singletons (Table 1).

Evidence of recombination.

Recombination within the S. pseudopneumoniae population was investigated by several methods. The index of association (I_A) estimates the level of linkage disequilibrium between alleles at different loci. In a completely clonal population, alleles will be in linkage disequilibrium, and the I_A approaches 1. If the I_A approaches zero, it suggests that the organism is in linkage equilibrium and is therefore freely recombining. The I_A value across the analyzed population was 0.4598 (P < 0.001), which suggests a high rate of recombination. The PHI test, conducted separately for each locus and for the concatenated sequences, also produced statistically significant evidence of recombination (P < 0.01). The split-decomposition method provided additional support. Split graphs of each locus (data not shown) and the split graph of concatenated sequences showed a “rectangular” network shape (Fig. 2), indicating that the seven genes were significantly affected by intragenic recombination. Comparison of this phylogenetic network with the phylogenetic trees (compare Fig. 1 and 2) showed that the 14 isolates were found in the same clusters.

Fig 2.

Evidence of extensive recombination within the isolates and type strains. Split-decomposition analysis was conducted with SplitsTree, version 4.0, of MLSA concatenated sequences. SMI, S. mitis; SOR, S. oralis.

Genetic analysis of lytA, ply, and ant.

We have previously characterized strains 3180, 3870, 4391, 4589, and 5237 as S. pneumoniae isolates by hybridization of their DNA with pneumococcal-specific lytA and pnl probes (15). However, they also hybridized with the ant probe, a gene that is not normally present in S. pneumoniae but that is found in S. mitis and S. oralis (15, 27). Hybridization assays of the remaining nine isolates showed that all of them harbored lytA and ply genes (data not shown). The presence of ant was also analyzed, showing that all isolates, except CipR-71 and CipR-72, hybridized with the ant probe (Table 1). S. pseudopneumoniae strains CCUG49455^T and CCUG48465 also hybridized with the lytA and pnl probes but only CCUG48465 hybridized with the ant probe (3).

Organization and sequence of the atp chromosomal region.

In a previous study we reported that the gene organization upstream of atpC appears to be quite variable depending on the streptococcal species studied. In type strains of S. mitis and S. oralis, atpC is preceded by a gene highly similar to S. pneumoniae spr1284, putatively encoding a protease. In S. pneumoniae, however, atpC and spr1284 are located approximately 65 kb apart. In S. pseudopneumoniae strains, spr1368 is located immediately upstream of atpC, an organization equivalent to that found in S. pneumoniae, but lacking IS1239 (4). Among the 14 isolates studied here, 4 share the same spr1368-IS1239-atpC-atpA pneumococcal-specific organization, two had the spr1284-atpC-atpA organization of the S. mitis and S. oralis type strains, and the remaining 9 isolates would be classified as S. pseudopneumoniae on the basis of their spr1368-atpC-atpA gene organization (Fig. 3). Gene organization upstream of atpC in the CipR-6.39 isolate could not be determined since PCR amplifications failed.

Fig 3.

Genetic structure of the atp region. Big arrows indicate the genes, which are named according to their S. pneumoniae homologues. The oligonucleotides used to amplify the region upstream of atpC are indicated by small black arrows.

Resistance to OPT in pneumococci has been reported as a consequence of point mutations that change amino acid residues located in either one of the two transmembrane α-helices of the c subunit or one of the two last α-helices of the a subunit of the F₀F₁ H⁺-ATPase (5, 28–30). The sequences of the atpC and atpA genes from 11 of 14 isolates were determined. These sequences showed almost complete identity (>99% identical) with those of S. pneumoniae. All carried the amino acid change E104G in AtpA. In addition, isolates 3870, 4589, CipR-71, and CipR-75 showed the AtpC F5Y change. Both amino acid changes are located outside the regions involved in OPT resistance; their role in OPT resistance needs to be determined.

DISCUSSION

This study describes several phenotypic and genotypic properties of a collection of streptococcal isolates, initially classified as pneumococci, obtained in three different studies designed to describe the incidence of Fq resistance among S. pneumoniae strains. All isolates were tentatively classified as atypical pneumococci because they gave inconsistent results in one or more tests conventionally used to identify pneumococcus isolates. To establish the relationship between our strains and pneumococci, we characterized them by MLSA and compared results with those obtained for control strains of S. pneumoniae, S. pseudopneumoniae, S. mitis, and S. oralis. On the basis of MLSA, four isolates were identified as pneumococci, three clustered closely with S. mitis and S. oralis, and the rest clustered with S. pseudopneumoniae. These results confirm that some of the isolates with mosaic structures in their topoisomerase genes are pneumococci despite showing some atypical characteristics. Some of them (6 of 14, or 42.9%) were found to be nonencapsulated. Despite lacking capsule, a major virulence factor for S. pneumoniae, their putative ability to produce disease cannot be dismissed since some nonencapsulated pneumococci have been associated with conjunctivitis (9, 21, 27), otitis media (31), and, less frequently, with invasive disease (6, 27). Other studies have shown that colonization was also promoted by the absence of capsule, facilitating adherence to epithelial cells and biofilm formation (32). Despite the putative pathogenic potential of these isolates, they are frequently omitted in epidemiologic studies due to their nontypeability, and their prevalence is underestimated (33).

The results presented herein reflect the complex taxonomy of the S. pseudopneumoniae species, due in part to its phenotypic and genetic heterogeneity, which has made it difficult to define a gold standard technique that would provide accurate identification to the species level (34). Among the different approaches used, only the MLSA was able to discriminate S. pneumoniae from S. pseudopneumoniae isolates. In contrast to one-target identification tests (OPT susceptibility, DOC solubility, AccuProbe, or determination of the presence of ply, lytA, or ant), which are severely affected by horizontal gene transfer, the MLSA scheme appeared to be the most useful instrument in the discrimination of species with high transformation rates since the distortion due to an interspecies event of recombination at one locus could be buffered by the other loci. However, an increase in the number of genes analyzed is not always worthwhile. While the distance value between S. pneumoniae and S. pseudopneumoniae using the MLSA approach was 0.052 ± 0.003 (mean ± standard error of the mean), the analysis based on the concatenated sequences of MLSA genes plus the 16S rRNA, sodA, rpoB, and hlpA genes also resolves the species although the distance value between S. pneumoniae and S. pseudopneumoniae decreased (0.035 ± 0.002), probably as a result of the reduction in the number of strains analyzed. In conclusion, if the purpose of the study is to discriminate S. pseudopneumoniae, the MLSA technique is suitable because it uses a smaller number of genes and because it allows a comparison of the results with an extensive database, which provides robustness to the analysis. Because speciation is a gradual process, it is not always easy to discriminate between species. In fact, the boundaries among these two species are fuzzy due to the presence of hybrid isolates. That is the case of isolate 4391, which is placed in the branch between S. pneumoniae and S. pseudopneumoniae (Fig. 1), representing an intermediate genotype. MLSA results totally agreed with those obtained with the analysis of the upstream region of the atp operon, coding the F_oF₁ H⁺-ATPase, the target of OPT and other amino-alcohol antimalarial drugs. This technique can also discriminate between S. pneumoniae and S. pseudopneumoniae and even helps to explain some inconsistencies obtained in the phenotypic characterization of isolates. For instance, the chromosomal organization of the atp region indicated that CipR-74 is not a pneumococcus, and the determination of the nucleotide sequences of the atpC and atpA genes strongly suggests that the OPT susceptibility of this isolate was due to the acquisition of atpC and atpA genes from S. pneumoniae by horizontal gene transfer. However, this technique could not be used with CipR-6.39.

The molecular characterization of the S. pseudopneumoniae isolates by MLST identified 62 alleles, indicating high genotypic diversity. Moreover, most STs analyzed were singletons, which reflects a highly diverse population. The lack of CCs and the large number of single STs suggested that during the evolution of S. pseudopneumoniae the expansion of clonal groups by accumulation of point mutations was limited, while strains with remote genotypes appeared frequently, most likely as a result of genetic recombination. Split-decomposition analysis conducted with individual loci and concatenated sequences showed highly statistically significant evidence of recombination. Intragenic and intergenic recombination among housekeeping genes, responsible for the diversification of genotypes, was supported also by visual inspection of the complex parallelogram network structure formed. The role of recombination in the diversification of S. pseudopneumoniae has been suggested but not yet demonstrated using multilocus sequence typing. Our data confirmed the importance of recombination in S. pseudopneumoniae evolution, where divergence among genotypes appeared to be driven mainly by recombination. The high frequency of recombination exhibited by S. pseudopneumoniae could be due to the absence of capsule, a phenomenon already suggested for nontypeable pneumococci (35, 36). S. pseudopneumoniae could act as an intermediary for horizontal gene transfer between streptococcal species, being a hybrid species between S. pneumoniae and S. mitis (37). Therefore, S. pseudopneumoniae isolates must play a major role in the pneumococcal ecology.