The data presented in this study supports previous research that has identified S. pneumoniae
as a genetically highly heterogeneous species, even within closely related strains 
. The ability to colonize and cause disease did not differ depending on the isolate's serotype when tested in the chinchilla model. However, isolates in the disease clade were associated with a decreased time to cause bacteremia compared to isolates in the carriage clade. One hypervirulent serotype 19A isolate was associated with a short time to bacteremia and high mortality. The genome of this strain contains a 20 kb genetic region that is absent in the other CC199 isolates tested in the chinchilla. This hypervirulent strain may have acquired a bacteriophage with novel virulence factors, although further analysis is required to identify the precise role of this sequence, if any, in the observed hypervirulence of isolate 06-01-003MEF. Taken together, these data indicate that genetic factors, other than capsule polysaccharide, modulate virulence potential in CC199. Serotype 15B/C is not included in PCV13 
. Given the similar genetic background and similar pathogenicity in the chinchilla, CC199 strains of serotype 15B/C may increasingly cause acute otitis media in children.
We hypothesize that CC199 strains are successful at the population level because they are a generalist lineage adept at carriage, otitis media, and invasive disease. Within CC199, isolates of both serotype 15B/C and 19A were equally able to colonize and cause otitis media in the chinchilla model. Furthermore, there was little difference in the time taken to cause invasive disease by serotype. These data expand upon a previous study that indicated similar fitness for otitis media when serotype 19A and 15B/C isolates were inoculated together in competition 
. While isolates belonging to the carriage clade have the potential to cause disease, isolates from the disease clade were associated with faster development of invasive disease. It remains to be seen if the identified pattern of separate carriage and disease clades in CC199 is repeated for other established clones, such as CC156 or CC176.
This study has identified genes of potential importance in tissue-specific virulence, which will require additional study to assess their potential and function as virulence factors. A larger collection more representative of the pneumococcal population was screened to determine whether those associations identified in CC199 isolates would be maintained in a more genetically diverse collection. Given that S. pneumoniae is a highly recombinogenic species, tissue-specific associations that are preserved throughout the population as a whole would be expected to be of biological importance to the species. However, it is worth noting that for two of the genetic loci, spr0282 and RD8a, the tissue-specific associations differed in the overall pneumococcal population when compared to the CC199 data, although no statistical significance was given to the CC199 data due to small sample size. This would seem to indicate that the effect of a gene may be dependent on the combination of genes found in the remainder of the pneumococcal genome.
Previous studies have identified individual genes associated with otitis media, pneumonia, and meningitis 
. We identified RD8a as more common among our middle ear and blood isolates in comparison to carriage. Obert et al
previously described the correlation between the region of diversity RD8a and invasive isolates 
. SP0463 (rrgB
) is on the pilus locus 
and has previously been identified in a minority of invasive pneumococcal isolates 
. Our finding that SP0463 is found significantly more often in middle ear isolates than carriage isolates seems at variance with other data that pneumococcal isolates do not contain genes encoding pili at a greater frequency in otitis media isolates than in invasive disease isolates 
. It is likely that the disparity arises through the specific isolate collections examined or the difference in target genes used by each study. Moschioni et al 
targeted entire pilus encoding islets, this study focussed on a single gene within the islet.
The specific combination of putative tissue-specific genes differed for each of the isolates tested in the chinchilla model. Furthermore, we did not identify a single genetic factor from the accessory genome, present in all strains from a specific tissue source. A supragenome pool exists for S. pneumoniae
, which consists of the total number of genes available to the species 
. A given gene may associate with S. pneumoniae
isolates from a specific tissue-source. Furthermore, a given gene may enhance fitness for tissue-specific disease, yet an isolate lacking this gene can still cause disease. These data may be explained by high levels of recombination and functional redundancy amongst the accessory regions within the pneumococcal genome. Blomberg et al
did not identify a unique pattern of accessory regions among invasive disease isolates and concluded that redundancy existed amongst the accessory regions required to cause invasive pneumococcal disease 
. Our data indicate that this redundancy holds for otitis media as well as invasive disease.
One hypervirulent CC199 19A (06-01-003 MEF) strain caused rapid mortality. The bacteriophage sequence within this region may encode virulence determinants. A majority of S. pneumoniae
strains contain lysogenic phage 
. Researchers have speculated that phage-encoded proteins modulate S. pneumoniae
. Further analysis of this sequence is required to identify if this is indeed the basis for the differences in virulence observed between the isolates.
This study had limitations. Genomic diversity of CC199 was evaluated using CGH. As mentioned previously, microarrays are limited in that genetic regions that are not included on the array will be missed. Genes that are highly variable in sequence may also fail to hybridize and will be counted as absent. Additional unmeasured factors, such as level of capsule expression, may also contribute to differences in virulence. As with any animal model, our data may not be fully reflective of S. pneumoniae carriage and disease in humans. A small number of chinchillas were used in each experiment and the chinchilla model does not use a genetically pure line. However, the advantage to the chinchilla model is it closely mirrors pneumococcal disease aetiology where colonization is established before disease.
In summary, the approval of PCV13 by the FDA in 2010 
means that isolates belonging to serotype 19A should be protected against in the future. However, this expanded vaccine does not include protection against serotype 15B/C. Recent studies have indicated an increase in the prevalence of serogroup 15 isolates amongst carriage and otitis media 
. Small but significant increases in the proportion of serotype 15B/C isolates have been observed among invasive disease cases 
. While immunization with PCV13 is likely to result in disease reduction, the similarities in terms of virulence between 19A and 15B/C isolates suggest that a vaccine based on genetic factors other than serotype is necessary, especially for otitis media and nonbacteremic pneumonia. Such genetic factors could belong to the core genome or comprise several from the accessory genome, which together cover the entire pneumococcal population. Alternatively, future vaccines could be targeted towards tissue-specific genes, allowing physicians to protect patients from invasive pneumococcal disease, while not affecting those pneumococcal isolates that exist among carriage isolates in the nasopharynx