To our knowledge, our collection represents the largest historical pneumococcal collection ever compiled and includes the greatest number of early (1960s to 1980s) PNSP in a single study. This has provided us with a unique opportunity to study pneumococcal evolution. Within our diverse isolate collection, whole or part pbp
alleles that were identical or highly similar to those of the PMEN1 reference strain were identified from pneumococci representing a total of 24 distinct CCs. Two isolates, CGSP14 and PMEN3, shared pbp2x
sequences identically with PMEN1, and shared highly similar sequences in the pbp
flanking regions. The full extent of sharing all three of these PMEN1 pbp
sequences has never before been uncovered (nor have the large regions around all three pbps
), although previous authors have independently noted pbp
similarities between CGSP14, PMEN3, PMEN2 and PMEN1 [16
]. Highly similar pbp2x
nucleotide sequences have also been independently identified among numerous pneumococcal and viridans streptococcal clones [37
]. Shared resistance-conferring pbp
sequences other than those associated with PMEN1 have also been described previously [5
] and were identified among our isolate collection, though at much lower frequency (data not shown). Additionally, data from other authors have suggested that fluoroquinolone resistance determinants have been donated from PMEN1 to numerous unrelated pneumococcal clones [43
It is very unlikely that identical or highly similar sequences were assembled independently in different CCs, but rather that they have been transferred, as whole or part alleles, between unrelated pneumococcal clones at least once for each independent CC. Note that the establishment of defined CCs must have occurred prior to acquisition of these pbp alleles, since we also identified penicillin-susceptible members of these clones (that is, CC1514, CC156/1629A/V).
The predominance of the PMEN1-like pbp
sequences within our isolate collection suggests that the PMEN1 reference pbp2x
, pbp1a and pbp2b
sequence combination may provide an evolutionary selective advantage over other resistance-conferring pbp
alleles, which is consistent with recent experimental data suggesting certain combinations of pbp2x
alleles are more advantageous than others [44
]. A study of genomic evolution within the PMEN1 lineage also provided evidence consistent with the hypothesis that the pbp2x
alleles were positively selected [17
is upstream and pbp1a
is downstream of the capsular (cps
) locus in the pneumococcal genome. The authors noted that the boundaries of recombination events affecting the cps
locus appeared to be restricted by the nearby pbp2x
loci. It is known that both of these loci and the cps
locus can be transferred simultaneously between pneumococci [45
], and thus the restriction within the PMEN1 lineage is likely not entirely mechanistic.
In addition to the pbp
loci and flanking regions, CGSP14 and PMEN3 acquired multiple genomic regions, totaling approximately 9.5% and approximately 5.3% of the genome, respectively, from PMEN1 representative(s). This is consistent with two recent studies that demonstrated that multiple fragment recombination had taken place in vivo
, between pneumococci isolated from a single patient [46
] and among vaccine escape strains isolated in the USA [47
]. The suggestion that a significant proportion of the genome may be acquired from a single pneumococcal donor was recently supported mechanistically by Attaeich and colleagues [48
]. These authors showed that cellular levels of pneumococcal SsbB protein are sufficient to maintain an intracellular pool of internalized DNA of up to 1.15 Mb of length (approximately 50% of the genome). Such a pool of genetic material, potentially originating from a single donor cell, could then be used for successive rounds of recombination.
We cannot know for certain whether the CGSP14 and PMEN3 imported regions were acquired simultaneously in a single event or via successive recombination events. Either way, the data strongly suggest that they were acquired from a PMEN1 representative. It should also be noted that recombination events additional to those described here, and involving acquisition of DNA from other pneumococcal clones or other species, may also have altered the CGSP14 and PMEN3 genomes.
The genomic regions acquired from the PMEN1 donor may also have contributed to the virulence and host colonization ability of the recipients. CGSP14 was multidrug-resistant and isolated from a child suffering from necrotizing pneumonia complicated with hemolytic uremic syndrome [35
]. PMEN3 represents another globally distributed multidrug-resistant clone [15
]. The putative recombinogenic regions acquired by both of these pneumococci included several genes associated with virulence or host colonization potential. Interestingly, large sections of the CGSP14 transposon, which was shown previously to contain regions of similar sequence to those of the PMEN1 ICESp
23FST81 element [35
], were in fact identical to those of PMEN1. Two interpretations of the regions of similarity within the ICESp
23FST81 element are possible. One is convergent evolution between PMEN1 and CGSP14, with both strains independently acquiring similar, recently diverged elements. Alternatively, a more divergent element may have been acquired by CGSP14 that was subsequently modified through recombination with DNA from PMEN1. The mosaic nature of these variable conjugative elements [17
] makes distinguishing these two hypotheses difficult.
Finally, we showed that one of the first reported PNSP (23F/4) represented the ancestor of PMEN1. The year of isolation of 23F/4 (1967) correlated with the predicted date of emergence of the PMEN1 lineage (approximately 1970), although the country of isolation (Australia) did not match the predicted region of PMEN1 origin (Europe) [17
]. Despite its high overall genomic similarity to PMEN1, the 23F/4 ST, which differed from that of PMEN1, has not been identified amongst modern pneumococci. Perhaps the differences between these two genomes could explain the differences in disseminative success of these two STs. The acquisition of more favorable pbp2x
alleles in the PMEN1 genome provides a partial explanation; however, other changes and/or acquisition of other loci have also likely played an important role. The PMEN1 ICESp
23FST81 element, Na+
-dependent ATPase island and
MMI phage were absent from the 23F/4 genome. The ICESp
23FST81 element carries both tetracycline and chloramphenicol resistance determinants [30
] plus a umuDC
-like gene shown to provide protection against UV damage [32
]. The Na+
-dependent ATPase island and a
MMI phage closely related to that of PMEN1 have been associated with an increased incidence of invasive pneumococcal disease [33
] and increased cell adherence capabilities [31
Any or all of these regions may have provided PMEN1 with a selective advantage, allowing it to persist among the greater pneumococcal population for the past three decades. Alternatively, the combination of these regions with the PMEN1 reference pbp genes and the 23F/4 genetic background may have been the most important factor. In any case, the subsequent global dissemination of PMEN1 likely aided the spread of its pbp alleles and other regions of its genome to numerous unrelated pneumococci. Some of the recipient pneumococci might subsequently have transferred these PMEN1-like sequences to additional pneumococcal clones.
Recently, through comparison of multiple representatives within the PMEN1 CC, Croucher et al.
] showed that the PMEN1 lineage contains a vast amount of genetic diversity, largely resulting from horizontal acquisition of DNA via recombination. Through comparison of genomic sequences across a genetically diverse collection of pneumococci, our study has greatly increased our understanding of the emergence of the PMEN1 clone and its subsequent contribution to the genomic evolution and spread of penicillin resistance determinants among other pneumococci. It is well understood that many bacterial species, particularly the pneumococcus, frequently undergo horizontal gene transfer; however, the directional genetic promiscuity described here, from a single epidemiologically successful clone to numerous unrelated clones, has not been demonstrated before in pneumococci, or other bacterial species as far as we are aware.
Antibiotics are among the most influential global public health successes, but impose selective pressures that drive bacterial genomic evolution. PMEN1 is an excellent example of a bacterium that has become resistant to multiple antibiotics and that has evolved to become very successful in colonization, transmission, and causing disease. Moreover, PMEN1 has subsequently shared its advantageous DNA with other unrelated pneumococci. Studies such as this one may help to identify ways to counteract these biological changes and preserve the value of antibiotics for the future.