![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Journal of Clinical Microbiology, April 2005, p. 1706-1715, Vol. 43, No. 4
0095-1137/05/$08.00+0 doi:10.1128/JCM.43.4.1706-1715.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.
Institute of Molecular Medicine, Lisbon Faculty of Medicine, Lisbon,1 Biomathematics Group, Instituto de Tecnologia Química e Biológica, Oeiras, Portugal2
Received 17 August 2004/ Returned for modification 24 September 2004/ Accepted 6 December 2004
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Pulsed-field gel electrophoresis (PFGE) of chromosomal macrorestriction digests and multilocus sequence typing (MLST) are two complementary techniques widely used to characterize the relationships among pneumococcal isolates. PFGE allows rapid and cost-effective clustering of genetically related isolates, and MLST then can be used to assign easily comparable and portable clone identifiers to clusters of similar isolates. These sequence types (STs) can be compared to those in a public database to identify relationships to other isolates recovered in various geographic areas.
Although there is abundant information on the genetic structure of the population of pneumococci asymptomatically carried by children attending day care centers in Portugal (4), there is no current information on the molecular epidemiology of invasive disease. We have used PFGE and MLST to characterize a recently described collection of isolates recovered in Portugal during the years 1999 to 2002 (13) in order to identify the clones responsible for invasive disease.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
PFGE and MLST.
PFGE and MLST were performed as described previously (6, 14). Bionumerics software (Applied-Maths, Sint-Martens-Latem, Belgium) was used to make UPGMA (unweighted-pair group method using average linkages) dendrograms of fragment patterns. The Dice similarity coefficient was used with optimization and position tolerance settings of 1.0 and 1.5%, respectively, as described previously (8). PFGE-based clusters were defined as isolates with 
80% relatedness on the dendrogram. DNA fragments smaller than 19 kb were not considered for dendrogram construction.
PFGE profiling was done for 457 isolates; for the remaining 8 isolates, no PFGE pattern could be obtained in spite of multiple attempts. A major lineage was defined as 4 isolates whose PFGE patterns clustered with a UPGMA Dice coefficient of >80%. MLST analysis was performed for at least one isolate in each major lineage. For large PFGE clusters, we chose a variable number of isolates to represent the diversity observed. All non-penicillin-susceptible isolates and isolates recovered from the cerebrospinal fluid (CSF) with PFGE profiles not identical to those of other isolates for which MLST profiles were determined were also subjected to an independent MLST analysis. A subset of 104 isolates were characterized by MLST.
MLST alleles and STs were identified by searching the pneumococcal MLST database (spneumoniae.mlst.net). Whenever new alleles were identified, sequence traces of both strands were submitted electronically to the pneumococcal database curator for approval. Lineage assignment was done by use of e-BURST analysis (7) with default parameters and the complete S. pneumoniae database (spneumoniae.mlst.net).
Statistical tests were performed by use of the SPSS 10 software package, and P values of <0.05 were considered significant.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Lineage distribution within individual serotypes. Below we describe in detail the clones found among the 10 most frequent serotypes in our collection as well as the 10 most frequently found among isolates recovered from children <6 years old (Table 1). The importance of this age group relates to current recommendations suggesting vaccination of children who are <6 years old and who attend day care centers (3), since most children in Portugal are in this category. The data are summarized in Table 1, which also shows clones also found in a survey of U.S. invasive isolates carried out by Gertz et al. (8).
(i) Serotype 1. Two main lineages could be seen in serotype 1, each represented by two STs—a cluster including the majority of isolates representing ST306 and ST228 and another including ST304 and ST350, both double-locus variants of each other (Fig. 1). These four STs were previously shown to be members of different branches of a lineage found associated with invasive disease exclusively in Europe and North America—lineage A (2).
|
(iii) Serotype 4. Equal numbers of isolates (n = 11) were distributed in two different lineages, ST247 and ST1221 (Fig. 1). ST247 was found to be associated with invasive disease in a number of European countries (6) and belongs to the lineage with the largest number of related STs expressing serotype 4 presently in the database. ST1221 is a novel ST that belongs to the same lineage as ST993, which was previously found in an isolate from invasive disease in Scotland. Three isolates belong to a third lineage, ST1222, a novel ST belonging to the same lineage as ST800 and ST801, which were previously found in invasive isolates in the Czech Republic.
(iv) Serotype 6A. Only three small clusters of related PFGE profiles were observed in serotype 6A (Fig. 1). A cluster of three isolates included an isolate recovered from the CSF and representing ST395, which was previously found to be associated only with carriage in England (1). Two other clusters contained a non-penicillin-susceptible isolate with the novel ST1369 and an isolate recovered from the CSF and representing ST460, which was previously found to be associated with a meningitis case in England (spneumoniae.mlst.net). The two isolates from children were found in the ST460 cluster and had a PFGE profile unrelated to any other in this serotype.
(v) Serotype 6B. Two PFGE clusters contained the majority of the isolates expressing serotype 6B. The larger PFGE cluster (n = 8) included isolates related (ST1224 and ST273) to the internationally disseminated clone Greece6B-22 (www.sph.emory.edu/PMEN/pmen_clone_collection.html), with the exception of a non-penicillin-susceptible isolate belonging to the Spain6B-2 lineage (10) (Fig. 2). The smaller PFGE cluster (n = 3) included only non-penicillin-susceptible isolates related (ST887) to clone Poland6B-20 (www.sph.emory.edu/PMEN/pmen_clone_collection.html). Three isolates were not grouped in these clusters.
|
|
(viii) Serotype 9V. Serotype 9V included a large number of non-penicillin-susceptible isolates. All isolates were grouped in a single PFGE cluster related to the internationally disseminated clone Spain9V-3 (10), in spite of having STs that differed in up to three loci—ST156, ST557, ST644, ST838, and ST1225 (Fig. 4). e-BURST analysis of the MLST profiles confirmed that these isolates belong to a single lineage, with the ST of clone Spain9V-3 being the predicted founder.
|
(x) Serotype 12B. Although serotype 12B isolates were grouped in two PFGE clusters (Fig. 5), both included representatives of ST218, which was previously found to be associated exclusively with isolates expressing serotype 12F and recovered from patients with invasive disease in Europe, North America, and South America (6, 8). Both clusters contained eight isolates; one contained a representative of ST989 (which was previously found to be associated with isolates expressing serotype 6B and recovered from the nasopharynx), and the other contained the novel ST1365, a singleton (does not group with any other ST), as determined by e-BURST analysis. The MLST data suggested that these lineages constitute separate branches and that the observed PFGE clustering does not reflect true genetic relationships. Also of note is the fact that all isolates in this collection with ST218 express serotype 12B and not serotype 12F, as was found previously (8). Although this finding could be attributable to capsular switching (11), it is also possible that interconversion between these serotypes follows a route similar to the one recently described for serotypes 15B and 15C (16).
|
|
(xiii) Serotype 19A. The largest cluster (n = 8) contained two distinct branches—a set of closely associated isolates of ST81 (n = 5), belonging to the same lineage as clone Spain23F-1 (10), and another group (n = 3) with representatives of two unrelated lineages, ST876 and ST1201 (Fig. 5). Both of the latter STs had been reported only once previously (spneumoniae.mlst.net) and were found by e-BURST analysis to belong to groups with a large number of STs and different serotypes. This cluster contains 57% (n = 4) of the non-penicillin-susceptible isolates expressing serotype 19A, three related to Spain23F-1 and the other belonging to ST876; both of these lineages had already been associated with penicillin nonsusceptibility (spneumoniae.mlst.net). The remaining three non-penicillin-susceptible isolates were found in clusters of two isolates of ST276 and one isolate of ST199. Non-penicillin-susceptible serotype 19 isolates had been identified previously in both STs (6). The remaining five isolates included a cluster of two isolates; a cluster of three isolates representing lineage ST416, which was previously found in serotype 19A isolates recovered from blood and associated with carriage (1, 9); and an isolate with a PFGE profile unrelated to any other in this serotype.
(xiv) Serotype 23F. The majority of the isolates were grouped in a single cluster (n = 17) containing mostly non-penicillin-susceptible isolates of ST338, the same lineage as the widely distributed clone Colombia23F-26 (Fig. 7). We also observed a smaller cluster (n = 3) that also included non-penicillin-susceptible isolates related to clone Spain23F-1 (ST81).
|
Other serotypes. We analyzed by MLST representatives of a number of other serotypes expressed by significant numbers of isolates in our collection; the data are summarized in Table 1. Among serotype 5 isolates, the majority (11 of 12) belong to the novel ST1223, are included in a single cluster, and belong to the same lineage as the internationally disseminated clone Colombia5-19.
A single cluster contained all but one of the isolates expressing serotype 9N (n = 10). Isolates in this cluster belong to ST66, similar to what was recently found for isolates recovered from the nasopharynges of healthy children in the United Kingdom (1) but in contrast to the results of a study of invasive isolates in the United States, where ST632 dominated (8).
The majority (six of seven) of the isolates expressing serotype 11A were contained in a single PFGE cluster including members of ST62 and ST408, both of which were found by e-BURST analysis to belong to the same lineage. This lineage was also the most prevalent in both carriage isolates from the United Kingdom and invasive isolates from the United States (2, 8), suggesting that it is widely disseminated.
Although serotype 19F was only the 18th overall in our collection and a single isolate was recovered from children <6 years old, 44% (four of nine) of the isolates expressing this serotype were non-penicillin-susceptible. The largest cluster (n = 7) was found to be related to clone Portugal19F-21 both by PFGE and by MLST (ST177 and ST391); this clone was found frequently in isolates resistant to at least one antimicrobial agent and recovered from the nasopharynges of healthy children in Portugal (12).
Among the five isolates expressing serotype 18A, three were recovered in 2002 and were found to be identical by PFGE; they included a member of the novel ST1232. The other two isolates were grouped in another cluster. A similar distribution in two clusters was observed for the five isolates expressing serotype 20; the newly found ST in the largest cluster (ST1026) was predicted by e-BURST analysis to be the founder of a lineage of serotype 20 isolates including two other previously found STs (ST215 and ST824). The four isolates expressing serotype 16F contained a cluster of three isolates including a representative of ST414, which was previously associated with a carriage isolate (1).
For serotype 22F, the majority (9 of 11) of the isolates clustered in a single group representing the ST1012 lineage. This lineage was recovered in all 4 years of the study period exclusively from adults, and it is interesting that ST1012 had already been found in a serotype 11A invasive isolate recovered in 1998 in Poland (spneumoniae.mlst.net). Moreover, ST1012 is the predicted founder of an e-BURST group that contains isolates recovered from invasive disease in Europe but expressing a number of different serogroups (11A, 18, 33, 33A, and 33F). However, this is the first description of isolates in this lineage expressing serotype 22F. Another ST (ST1372) was found in the other PFGE cluster containing the remaining two isolates. Table 1 also shows the STs of isolates expressing other serotypes not discussed in the text.
Distribution within the study period and associations of serotypes with PFGE and MLST profiles.
Most (17 of 25) of the PFGE clusters with 4 isolates were represented in all years of the study period (1999 to 2002). Exceptions occurred in major clusters of the following serotypes or serogroups in which isolates were not recovered in the year(s) indicated in parentheses: 5 (2000), 8 (2002), 9N (2000), 10A (1999), 11A (2001), 18C (1999), 19F (2002), and 33F (2001 and 2002) (Table 1). A chi-square test of the association of serotypes with more than 10 isolates with the 4 study years failed to show any association (P = 0.09). Moreover, Kendall's coefficient of concordance (0.78) indicated a high degree of similarity in the rank orders of all serotypes in the four years studied (a coefficient of 1 indicates complete agreement, and a coefficient of 0 indicates no correlation). These results indicate that there are no substantial differences in serotype prevalence in the different years. Together with the identification of most of the major clones in every study year, this finding is consistent with a stable population of invasive pneumococci in Portugal during the years 1999 to 2002 with variations only in the prevalence of the major lineages observed each year.
As in previous studies, the associations between serotype and ST found in our collection were consistent with the extensive database found at spneumoniae.mlst.net. However, a few instances of isolates expressing serotypes different from the ones reported in the database were noted, and in these instances, the serotypes of the isolates were confirmed by the Quellung reaction. Notably, two serotype 22F isolates belonged to ST1012, which was previously found to be associated exclusively with a serotype 11A isolate. The combined analysis of the PFGE profiles of these two serotypes showed that most 22F and 11A isolates shared 80% similarity (data not shown). However, the STs that were found to be associated with isolates expressing serotype 11A in our collection shared only three or four alleles with ST1012, indicating that these isolates were not related. We also found two isolates expressing serotype 12B and sharing ST218 (Fig. 5), which was previously found to be associated exclusively with isolates expressing serotype 12F. Isolates belonging to ST557 and ST644 were also found to be associated with novel serotypes, 14 and 9V (Fig. 4 and 6). However, both of these STs belong to the same lineage, as defined by e-BURST analysis (predicted founder, ST156), and both of these serotypes had already been found in other STs of this lineage.
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The majority of the lineages identified were found in all years of the study period (1999 to 2002), suggesting that there were no major changes in overall diversity during this time. Widespread vaccine usage has the potential to change the serotype prevalence in the pneumococcal population causing invasive disease, as was recently demonstrated in the United States (17). This change in serotype composition will probably be paralleled by an alteration in the clonal composition of the population, but it is also possible that transformation to nonvaccine serotypes of the presently circulating clones will play an important role (11).
It is interesting that although large clusters of related isolates were frequently observed, the overall diversity within serotypes varied greatly (Fig. 1 to 6). Serotypes including almost exclusively a single lineage (7F) (Fig. 3) contrasted with others in which small unrelated clusters were found (6A) (Fig. 1). Even though it was previously suggested that isolates susceptible to antimicrobial agents are more diverse than penicillin-resistant ones (15), the intrinsic properties of each serotype clearly play an important role, since both serotype 6A and serotype 7F included mostly isolates susceptible to all antimicrobial agents tested. Moreover, the diversity found in serotype 1 isolates was less than that found in serotype 14 isolates in spite of the fact that the latter serotype included mostly isolates resistant to at least one of the antimicrobial agents tested and serotype 1 included overwhelmingly fully susceptible isolates. These observations could be a direct consequence of the diversity of the global pneumococcal population expressing each serotype, perhaps due to recent evolutionary history, or could be related to the selection of successful invasive clones in Portugal. Knowledge of the population of asymptomatically carried pneumococci is necessary to begin to address these questions. Although detailed studies of the pneumococci carried by children attending day care centers in Portugal have been published, they have so far been restricted to resistant isolates (4, 12), preventing a direct comparison of the two populations. In spite of these limitations, it is already clear that a lineage found frequently in our collection, Spain9V-3 (see below), was also frequently found colonizing the nasopharynges of children attending day care centers in the Lisbon area (4).
A number of internationally disseminated clones (www.sph.emory.edu/PMEN/pmen_clone_collection.html) were found in our collection. Isolates related to clone Spain9V-3 (expressing either serotype 9V or 14) and clone Colombia23F-26 accounted for 72% of non-penicillin-susceptible isolates. Most of the remaining non-penicillin-susceptible isolates were also related to other clones recognized by the PMEN (10). The dominance of internationally disseminated clones, including the two penicillin-susceptible clones Greece6B-22 and England14-9, was also found in erythromycin-resistant isolates; isolates belonging to clusters including clones recognized by the PMEN accounted for 59% of resistant isolates (27 of 46) (Fig. 2 and 6). Notwithstanding these data, other large clusters of related isolates that accounted for large proportions of the isolates expressing each serotype and that were not related to clones already accepted by the PMEN were also identified (e.g., the ST191 lineage in serotype 7F or the ST180 lineage in serotype 3). However, most of these isolates were fully susceptible to all antimicrobial agents tested and would therefore fall outside the stated aim of PMEN of identifying antibiotic-resistant clones (10). It will be interesting to monitor the expected changes in serotype prevalence due to vaccine use and to determine whether these will lead to an overall reduction in resistance rates due to negative selection for these clones expressing vaccine serotypes or whether capsular transformation events will allow the persistence of these genotypes. Also noteworthy is the absence of PMEN clones that were found in the United States by Gertz et al. (8), namely, those that comprised significant portions of the total population of their respective serotypes, specifically clones Taiwan19F-14 and Taiwan23F-16.
Gertz et al. (8) showed, using the same analysis parameters and methods that we applied, that most of the isolates belonging to clusters differing in individual PFGE patterns by a UPGMA Dice coefficient of 
20% had 5 shared alleles in their MLST profiles (8). However, these authors also described a number of exceptions; e.g., isolates differing at more than two loci had PFGE profiles with >80% relatedness. An analysis of our own collection largely confirms the correlation observed by Gertz et al. (8), but exceptions were also noted. In serotype 9V, ST838 and ST1225 are triple-locus variants of each other in spite of the fact that their PFGE profiles differ by 20% (Fig. 4). Similar situations were noted with ST790 and ST156 in serotype 14 (Fig. 6) and ST717 and ST100 in serotype 33F (Fig. 7). In serotype 19A and 12B isolates, two PFGE clusters with >80% relatedness contained STs sharing only a single allele—ST876 and ST81 and ST989 and ST218 (Fig. 5), respectively. In serotype 3, ST260 and ST180 did not share any of the alleles at the seven MLST loci, but their PFGE profiles differed by a UPGMA Dice coefficient of 20%. Notwithstanding these differences in MLST profiles, the STs differing at three loci in serotypes 9V and 14 belonged to the same lineage as that defined by e-BURST analysis, suggesting that the PFGE data reflect true relatedness. By combining PFGE and MLST data, we were able to infer the STs of the majority of isolates analyzed (Table 1).
In conclusion, this study provides a characterization of the pneumococcal population causing invasive disease in Portugal before widespread vaccine use. This information will allow investigators to address important issues, such as the impact of vaccination on the genetic structure of the population causing invasive disease and, together with studies of the carriage population, the relative weights of serotypes and other clonal properties in determining invasiveness.
|
ACKNOWLEDGMENTS |
|---|
We acknowledge the use of the pneumococcal MLST database, which is located at Imperial College London and is funded by the Wellcome Trust
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Antimicrob. Agents Chemother. | Clin. Microbiol. Rev. |
|---|---|
| Clin. Vaccine Immunol. | ALL ASM JOURNALS |
|---|
