Potential Nonpneumococcal Confounding of PCR-Based Determination of Serotype in Carriage

LETTER

Monitoring pneumococcal carriage serotype distributions is increasingly used to study pneumococcal biology, disease epidemiology, and vaccine impact. Potentially complicating DNA-based carriage assessment is the well-documented history of genetic recombination between pneumococci and related species that colonize the upper respiratory tract (4).

Recently, when assessing pneumococcal carriage in western Kenya (1) with a multiplex PCR (mPCR) assay for 40 capsular serotypes or serogroups (2), we noted a high frequency of mPCR positivity for cocarried serotypes. This was especially evident in adults who had naso- and oropharyngeal swab specimens combined; from these, 122 of 158 (77%) exhibited 4 or more mPCR serotypes (1). These included 24 of 27 specimens that were negative for the pneumococcus-specific lytA PCR assay (3; unpublished data). The mPCR amplicons sg10F/10C/33C and sg18A/18B/18C/18F occurred among 59 to 80% of adults (1). Sequencing of sg10F/10C/33C amplicons from 11 carriage specimens revealed five 192-bp sg10F/10C/33C sequence types, including three from lytA-negative specimens, which shared 90.1 to 93.8% identity to corresponding pneumococcal reference 10F, 33C, and 35B sequences (Table 1). Analysis of sg18A/18B/18C/18F amplicons from 17 specimens (including three lytA-negative specimens) revealed five sequence types with 89 to 93% identity to the 4 serogroup 18 reference sequences (Table 2). In contrast, sg10F/10C/33C and sg18A/18B/18C/18F mPCR amplicons from carriage specimens that yielded corresponding serotype 18A, 18C, or 10F isolates revealed sequence identity in each instance with the corresponding pneumococcal reference sequence (Tables 1 and 2).

Table 1.

Sequence identities (percent) for 192-bp wzx gene sequences from nonpneumococcal species and from carriage specimens that coamplify with targeted PCR assay for pneumococcal sg10F/10C^h

Sequence	% identity
	Spn 10Fa	Spn 33Cb	Spn 35Bc	S. infantis SS1641d	S. gordonii SS1245d	S. salivarius SS1061d	S. oralis SS911, SS1236, and 32e,f	Specimen 49f	Specimen 248f	Specimen 265f	Specimen 300f
Spn 10Fa		88.0	87.0	87.0	92.7	92.2	92.7	92.7	93.2	92.7	93.2
Spn 33Cb			90.1	90.6	94.3	93.2	95.3	93.2	93.8	93.2	93.2
Spn 35Bc				95.8	89.1	90.6	89.1	90.1	90.6	91.1	91.1
S. infantis SS1641d					90.6	89.6	89.6	89.6	90.1	90.6	90.6
S. gordonii SS1245d						97.4	99.0	97.9	98.4	97.9	97.9
S. salivarius SS1061d							97.4	97.4	97.9	97.4	96.4
32f								97.9	98.4	97.9	97.9
49f									99.5	99.0	97.9
248f										99.5	98.4
265f											97.9
83g	100

^aOnly available pneumococcal serotype 10F reference wzx sequence (GenBank accession no. CR931652), from which mPCR assay primers derived (192 bp after subtracting mPCR primer sequences) (2), which is also identical to the serotype 10C reference mPCR amplicon sequence (GenBank accession no. CR931651) .

^bThe corresponding pneumococcal serotype 33C wzx sequence (192 bp, from GenBank accession no. cr931700).

^cThe corresponding pneumococcal serotype 35B wzx sequence (192 bp, from GenBank accession no. cr931705).

^dThese sequences are available at the CDC Streptococcus Lab website at http://www.cdc.gov/ncidod/biotech/strep/pcr.htm (see PCR deduction protocols, point 3).

^eThese 3 amplicons shared sequence identity and were also identical within their overlap with GenBank submission AB289547 describing the S. oralis surface polysaccharide synthetic gene cluster. All included amplicons were of the exact same length of 248 bp.

^fSequences corresponding to amplicons representative of multiple carriage specimens PCR positive for sg10F/10C. Specimens either were lytA negative or were positive for confirmed pneumococcal isolates of nonrelated serotypes. Sequences corresponding to specimens 49, 248, and 300 are available at the CDC Streptococcus Lab website. Specimens 32 and 265 are identical within their overlaps with recent S. oralis GenBank submissions AB289547 and FR720602, respectively.

^gSequence corresponds to amplicon from the single carriage specimen that yielded a serotype 10F pneumococcal isolate. Specimen was mPCR positive for 10F/10C and was lytA positive.

^hShading indicates corresponding pneumococcal reference sequences for serotypes 10F, 33C, and 35B. Spn, S. pneumoniae.

Table 2.

Sequence identities (percent) for suspected nonpneumococcal 507-bp wzy gene sequences from carriage specimens that coamplify with targeted PCR assay for pneumococcal sg18A/18B/18C/18F with pneumococcal references 18A, 18C, and 18F^e

Sequence(s)	% identity
	18Aa	18Ca	18Fa	257b	262b	269b	368b
18A		97.6	97.8	89.5	90.5	92.1	90.7
18C			99.0	89.3	90.7	91.9	90.9
18F				90.3	90.9	92.5	91.1
257					92.1	92.7	91.9
262						96.3	99.4
269							96.4
8, 33, 103, 323c	100
117, 369, 395d		100

^aThe indicated 507-bp reference amplicon sequences (after subtracting primer sequences [2]) are within GenBank accessions CR931671 (18A), CR931673 (18C), and CR931674 (18F). The 18C and 18B reference amplicons share the identical 507-bp sequence. All amplicons in the table are of the same length (573 bp).

^bSequence corresponds to mPCR amplicon for sg18A/18B/18C/18F from indicated carriage specimen. Specimens either were lytA negative or were positive for confirmed pneumococcal isolates of non-sg18 serotypes. These sequences are available in GenBank as JX104734 (specimen 257), JX105735 (specimen 262), JX105736 (specimen 269), and JX105737 (specimen 368).

^cSequence corresponding to amplicons representative of carriage specimens PCR positive for sg18A/18B/18C/18F. Specimens were lytA positive and yielded serotype 18A pneumococcal isolates.

^dSequences corresponding to amplicons representative of carriage specimens PCR positive for sg18A/18B/18C/18F. Specimens were lytA positive and yielded serotype 18C pneumococcal isolates.

^eShading indicates pneumococcal references 18A, 18C, and 18F.

Within a reference collection of 54 strains comprised of 16 nonpneumococcal species, we found 5 mPCR-positive results for sg10F/10C/33C within strains of Streptococcus oralis (2 of 3), Streptococcus infantis (1 of 1), Streptococcus gordonii (1 of 5), and Streptococcus salivarius (1 of 3). Strains of Streptococcus mitis (4), Streptococcus pseudopneumoniae (21), Streptococcus parasanguinis (3), Streptococcus sanguinis (3), Streptococcus cristatus (3), Streptococcus vestibularis (2), Streptococcus peroris, Streptococcus australis, Streptococcus intestinalis, Streptococcus oligofermentans, Streptococcus sinensis, and Dolosigranulum pigrum tested negative. Sequence comparisons of the 5 amplicons revealed 87.0 to 95.8% identity to corresponding 10F, 33C, and 35B reference sequences (Table 1). We found that the amplicon from both S. oralis reference strains shared sequence identity with an S. oralis polysaccharide biosynthetic locus (5) and also with the amplicon obtained from lytA-negative specimen 32. We did not find positive mPCR results for sg18A/18B/18C/18F within these reference strains; however, it is unlikely that this collection provided adequate representation of commensal streptococcal species diversity.

In summary, mPCR amplicon sequences for sg10F/10C/33C and sg18A/18B/18C/18F within lytA-negative specimens were divergent compared to the reference pneumococcal serotype sequences (Tables 1 and 2). From within the same population, multiple specimens that yielded isolates that expressed the serotypes in question invariably revealed amplicons (from the specimens and isolates) with sequence identity to the corresponding 10F, 18A, and 18C references. We found divergent homologs of a pneumococcal capsular biosynthetic locus within a small nonpneumococcal species collection. We also encountered similar issues in this population with additional vaccine serotypes besides 18C (2, 5, 7F, and 33F) that are vaccine components (unpublished data). Until further clarification, PCR-based serotyping for carriage studies should be employed upon identified pneumococcal isolates only.

This preliminary letter describes our concern. There has been an unforeseen delay in the availability of these carriage specimens to us for isolation and characterization of potentially nonpneumococcal confounding species. While aware that this letter represents an incomplete study, we feel that further delay in conveying this information is unacceptable.