We have shown a small portion of what is an as yet unquantified upper respiratory reservoir of non-pneumococcal mitis group streptococcal strains that carry homologs of a large percentage of the known pneumococcal serotype or serogroup-specific genes that encode enzymes for specific polymerization and export functions (wzy
genes; Aanensen et al., 2007
). These genes serve as targets for the majority of the 40 individual primer sets that we employ. For example, we suspect that the majority of the cmPCR type 2-positive specimens reflects non-pneumococcal strains, given overall positivity in >30% of A-NP/OPs, while no serotype 2 pneumococcal strains were recovered. It is quite likely that many individual cmPCR sequence subtypes even within the same cmPCR type represent distinct non-pneumococcal strains of one or more species (as judged by MLSA ()). Among only 5 cmPCR type 2 amplicons, we found 4 distinct sequence subtypes (). The remaining 51 cmPCR-positive specimens are predicted to represent numerous additional type 2 subtypes. Strain and even species diversity within such cmPCR types remains to be investigated. For example, with relatively little sampling we have now found 4 different cmPCR-positive mitis group species for the cmPCR type 10F/10C/33C amplicon (S. oralis
, S. parasanguinis
, S. infantis
, and S. gordonii
) (this study and Carvalho et al., 2012
). In a previous study, we reported associating a Streptococcus salivarius
(salivarius group) reference strain with this cmPCR type (Carvalho et al., 2012
). Subsequently, we have found that our records were in error and that this strain (SS1061) is a strain of the mitis group species Streptococcus gordonii
While culture-independent cmPCR-serotyping of A-OP or combined A-OP/NP specimens added a very large number of false-positive results into our study, this technique also added valuable pneumococcal serotype detection data to the C-NP portion of this study. We believe that culture-independent cmPCR of these enriched specimens added valuable missed data for each of the vaccine-targeted serotypes 19F, 23F, 6A, 6B, 14, 1, 3, 4, and 7F (). Even though the evidence is not quantitatively supported by pneumococcal isolation data, it does suggest that few, if any, confounding non-pneumococcal amplicon results were obtained for these serotypes. Within C-NPs none of these “cmPCR serotypes” were observed that were not represented in the overall sampling by cultured pneumococci. For each of these targets, amplicon sequences were identical whether they were derived from specimens that were culture-positive for the corresponding pneumococcal serotype or not.
Unfortunately, the false-positive information that this culture-independent method introduced is difficult to quantitate. While real time PCR-serotyping is predicted to add somewhat more specificity for pneumococcal targets than cmPCR, real time PCR may similarly detect non-pneumooccal strains among carriage specimens. For example, we applied our recently developed triplexed real time assay (Pimenta et al., 2012
) on the non-pneumococcal strains depicted in that were cmPCR-positive. While we found that the S. oralis
strains with the cmPCR subtypes 376.12F and 291.33F were each strongly positive for the corresponding real-time PCR assay in triplex or monoplex format (for detecting 12F/12A/44/46 and 33F/33A/37 respectively), the other 7 strains were uniformly negative for their respective real time PCR assay. It is possible that any single PCR assay used for detection of pneumococci in the upper respiratory tract has a risk of cross-reaction with related mitis group streptococci. We have this concern for the CDC lytA
assay, however, we presently have no data suggesting that it cross-reacts with non-pneumococcal species. Currently we can only state from the data shown in that while we found lytA
-positivity for numerous specimens from which we did not recover pneumococci, especially from adult NP/OPS, we did not encounter any lytA
-negative specimens that yielded pneumococcal isolates. The nine cmPCR-positive non-pneumococcal strains described in this study were found to be lytA
-negative. Although this finding is not conclusive, it is consistent with observations that indicate the specificity of the CDC lytA
assay for pneumococcal identification (Carvalho et al., 2007
The majority of these non-pneumococcal homolog sequence subtypes have not been documented or characterized at this time. Indeed, except for the cmPCR type 10F/10C/33C subtypes which are highly homologous to known mitis group counterparts (Yoshida et al., 2008
; Yang et al., 2009
), all of the subtype sequences depicted in most closely matched their known pneumococcal counterparts. Similar known S. oralis
amplicon sequences lie within operons quite similar to their pneumococcal sg10 cps
operons and encode the apparatus responsible for synthesis of coaggregation receptor polysaccharides (Yoshida et al., 2008
; Yang et al., 2009
). The limited sequence-based associations made here in no way preclude identical amplicon subtypes from being shared between pneumococci and other related species. On the contrary, such findings are entirely expected, and a case in point is a suspected non-pneumococcal source for the single st39.39 subtype found within a lytA
-negative specimen (). While all colony types were screened for non-pneumococcal sources of cmPCR amplicons, only pneumococci and other alpha-hemolytic mitis group species were implicated as cmPCR-positive.
One issue that concerned us was the possibility that the real time lytA PCR assay might cross-react with non-pneumococcal species present in the upper respiratory tract. This was especially concerning in view of the relatively high frequency of lytA-positive A-NP/OPs relative to pneumococcal culture-positive A-NP/OPs (approximately 2-fold and 5-fold more lytA-positives relative to culture-positives in HIV-positive and HIV-negative, respectively, as shown in ). In part this discrepancy could be due to greater technical difficulty in isolating pneumococci from oropharyngeal flora relative to nasopharyngeal flora, as shown by our relatively poor isolation rates from retrospectively tested OP specimens (). From the limited re-testing results within our laboratories, it appears that we have under-estimated pneumococcal carriage within the A-NP/OPs described here ( and ). In the cross-comparison of adult NP and OP specimen testing results, we could project a total of 17 more culture-positives among the 70 HIV-positive culture-negative combined NP/OP specimens, which would have resulted in a 55.1% carriage frequency (rather than 40.7% as shown in ). Similarly, among the HIV-negative specimens we missed 5 positive results (4 NPs and 1 OP) corresponding to 28 A-NP/OP specimens that were originally found to be culture-negative. This would translate to 6 additional positives among the original 35 culture-negative NP/OPs (), more than doubling our original culture-based findings to 27.5%.
While the magnitude of putative non-pneumococcal cmPCR-positive results were evident within the A-NP/OPs ( and ), we demonstrated that the majority of this confounding data was conferred from A-OP specimens ( and ).While we believe that the majority of the cmPCR data shown in from C-NPs accurately predicts pneumococci of corresponding sequence types, at least a small percentage of non-pneumococcal cmPCR –positive results can be found within the pediatric nasopharyngeal reservoir, since we isolated a S. mitis
strain of cmPCR subtype 257.18 from one C-NP specimen.We quantitatively investigated the cmPCR-18C/A/B/F data from C-NPs and found that of the 4 amplicons not corresponding to st18C or st18A pneumococcal isolates, 2 amplicons shared an identical sequence with the published st18A reference (http://www.cdc.gov/ncidod/biotech/files/pcr-oligonucleotide-primers.pdf
), one shared sequence identity with the published st18C reference, and one was divergent (257.18) from lytA
-negative specimen from which the causal S. mitis
was recovered. Subtype 32.10 is also likely to be present within non-pneumococcal pediatric nasopharyngeal flora, since it was originally identified from S. oralis
reference strains (Carvalho et al., 2012
) and was also associated with a single lytA
-negative C-NP (). In contrast to cmPCR type 18C/A/B/F, several other cmPCR types corresponding to important serotypes included in conjugate vaccines (cmPCR types 1, 19A, 19F, and 23F) were reflected by single sequence subtypes among multiple sequenced amplicons, were associated only with pneumococcal isolates, and were not found among lytA
-negative specimens in this study (). These results indicate that these particular cmPCR reactions are potentially pneumococcal-specific.
Although the nasopharynx is believed to be the principal carriage reservoir of S. pneumoniae
in children, the organism also resides in the oropharynx. Although NP sampling is believed to be more representative overall of carriage strains than OP sampling, using both NP and OP sampling in adults modestly enhanced the detection of pneumococcal carriage (Watts et al., 2004
). In contrast, a large-scale study performed in Burkina Faso indicated that adding OP swab data to NP swab data increased culture-based carriage detection by 60% (Mueller et al., 2012
). While our small cross-comparison of 56 NP, OP, and combined NP/OP specimens indicate that NP specimens were the preferred specimen for pneumococcal isolation, it is important to note that our methods differ in that we employ broth-preculture before plating for isolation. Additionally we do not employ gentamycin selection in our isolation plates, however, we have found that using this selection does not improve our results in recovering pneumococci from NP or OP specimens (data not shown).
In conclusion, while usage of culture-independent cmPCR for more sensitive detection of pneumococcal serotypes in broth-enriched NP specimens appears promising, more analysis is necessary. Presently it is our opinion that pneumococcal strain isolation is a necessary component of carriage studies, and that serotyping by conventional or molecular methods should be done on colonies confirmed to be pneumococcal. When performing cmPCR to assess pneumococcal serotype distribution on specimens directly, it is necessary to correlate specific “cmPCR-serotype” amplicon sequence subtypes with their existence in pneumococcal strains of the concordant serotype. We admit that even this precaution is not completely satisfactory, due to the possible presence of specific amplicon sequence subtypes in both pneumococcal and non-pneumococcal strains. Enrichment culture to enhance pneumococcal recovery and detection that combines NP specimens together with OP specimens should be avoided, since OP specimens are apparently a much richer source of non-pneumococcal mitis group strains that confound cmPCR serotype assessment and potentially mask the presence of pneumococcal strains.