Development of a TaqMan Array Card for Pneumococcal Serotyping on Isolates and Nasopharyngeal Samples

Streptococcus pneumoniae is both a commensal and a major pathogen that causes invasive disease in people of all ages. The introduction of serotype-specific pneumococcal vaccines has reduced the burden of disease but has also led to replacement with new strains; thus, serotyping remains important for vaccine-related disease surveillance. Conventional serotyping methods are laborious and expensive. We developed an easy-to-perform genotypic TaqMan array card (TAC) to identify S. pneumoniae strains, including lytA-based sequences, and 53 sequence-specific PCRs to identify 74 serotypes/serogroups covering all current vaccine types as well as prevalent nonvaccine types. The TAC method was evaluated on 146 clinical S. pneumoniae isolates and 13 nonpneumococcal species that naturally inhabit the upper respiratory tract and yielded 97% (142/146) sensitivity and 100% (13/13) specificity versus results of standard Quellung serotyping. The calculated limit of detection was 20 to 200 fg (∼8 to 84 genome equivalents) per reaction. On 23 blinded nasopharyngeal specimens that were pneumococcus culture positive, the TAC pan-pneumococcus lytA assay was positive in 21 (91% sensitivity versus culture). On TAC lytA-positive specimens, a serotype result was obtained on 86%, and the result was 95% accurate versus the subsequent culture's Quellung result. TAC also detected mixed serotypes in two specimens where Quellung detected only the predominant serotype. This TAC method yields fast and comprehensive serotyping compared to the standard method and may be useful on direct specimens.

S treptococcus pneumoniae (the pneumococcus) is a leading invasive pathogen of children and older adults, principally causing pneumonia, otitis media, and meningitis. The precursor to invasive disease is upper airway colonization (1). Existing vaccines are based upon capsular polysaccharide and are highly effective only against vaccine types (2)(3)(4)(5). Serotype replacement occurs with increased colonization and disease caused by nonvaccine strains (6,7). With over 90 different capsular serotypes, there is a constant race to add more capsular types to further expand coverage to reduce disease burden amid a headwind of changing strain replacement.
In this context it is important to epidemiologically follow pneumococcal serotypes, both in invasive strains to detect emergence of virulent serotypes and also in the upper airway to monitor strain replacement (1,8). However, serotyping of pneumococci with the Quellung method is technically difficult, requires expensive panels of polyclonal antisera and precise inocula (9), and may yield visually ambiguous reactions (10). Furthermore, a limited number of subcultured colonies are typed, limiting the ability to detect mixed infections, particularly from nasopharyngeal specimens (11,12).
Molecular serotyping methods are therefore emerging. After elucidation of the capsular biosynthetic locus (13), PCR assays for the capsular polysaccharide synthesis gene clusters have been devised. Sequencing-based assays of the cps and wzh genes (14,15) have been published, as have real-time PCR assays to detect 21 serotypes/serogroups (16,17). Nanofluidic, microarray, and Luminex-based systems have also been developed (18)(19)(20)(21). Recently, we optimized 53 singleplex reactions to discern most serotypes/ serogroups, including all vaccine types (22). However, performing that many reactions per specimen is onerous and difficult to implement in field settings; therefore, in this work we further opti-mized and configured the reactions to a single TaqMan array card (TAC).
Quellung standard serotyping. Quellung results were determined as described previously (22). Briefly, a fresh overnight bacterial culture in a blood agar plate was suspended in 1ϫ phosphate-buffered saline (PBS) and then mixed with antiserum on a glass slide and read microscopically at a magnification of ϫ100. Pneumococcus Neufeld antiserum was obtained from the Statens Serum Institute (Copenhagen, Denmark).
DNA extraction from bacterial cultures and nasopharyngeal specimens. A bacterial colony was suspended in 200 l of lysis buffer (Tris-EDTA [TE] buffer containing 0.04g/ml lysozyme and 75 U/ml mutanolysin), or 200 l of nasopharyngeal specimens (in STGG medium) was mixed with 100 l of lysis buffer. Samples were incubated for 1 h at 37°C. DNA was then purified using a QIAamp DNA minikit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions and eluted in 100 l. The quality and quantification of DNA preps obtained from bacterial cultures were further evaluated using a NanoDrop system (NanoDrop Technologies, Wilmington, DE).
Assay development on 384-well plates. We adopted 53 serotype/serogroup-specific primers and probes from published sources (16,22,(24)(25)(26) (Table 1) and, if needed, made modifications to accommodate the common cycling condition of the TaqMan array card (TAC) using Primer Express, version 3 (Applied Biosystems, Life Technologies Corp., Carlsbad, CA, USA). We also included one pan-pneumococcus assay (lytA) (26) and an assay for an internal control (27). Optimization of conditions and probe specificity testing were performed using the 384-well format of the ViiA7 platform (Applied Biosystems, Life Technologies Corp., Carlsbad, CA, USA). Each primer/probe set (0.09 l of each forward and reverse primer, 0.025 l of probe of a 50 M stock) was amplified in singleplex in a total of 5 l of PCR mixture containing 2.5 l of 2ϫ TaqMan universal master mix II with uracil-N glycosylase (UNG) (Applied Biosystems, Life Technologies Corp., Carlsbad, CA, USA), 1.295 l of nucleasefree water, and 100 pg of genomic DNA. Cycling conditions included UNG activation at 50°C for 2 min and initial denaturation at 95°C for 10 min, followed by 40 cycles of denaturation at 95°C for 10 s and annealing/ extension at 60°C for 1 min. We included 54 previously characterized serotypes in each run for specificity testing, and nuclease-free water was used for a nontemplate control.
Evaluation of the TaqMan array card. Primer and probe oligonucleotides were synthesized and spotted onto the TaqMan array card by Applied Biosystems (Life Technologies Corp., Carlsbad, CA, USA) as laid out in Fig. 1. Twenty microliters of input DNA (1 ng/l for isolates) was mixed with 50 l of 2ϫ TaqMan universal master mix II with UNG (Applied Biosystems, Life Technologies Corp., Carlsbad, CA, USA) and 30 l of nuclease-free water to a 100-l final volume. This was loaded into each port of the card, whereby each card included seven clinical samples and one synthetic positive-control plasmid (Genewiz, Inc., South Plainfield, NJ, USA) that we designed to contain the primer and probe region of all 55 assays (53 serotype-specific assays plus lytA and phocine herpesvirus [PhHV]). The loaded card was centrifuged twice at 1,200 rpm for 1 min and then sealed; the loading ports were excised, and the card was inserted into a ViiA7 instrument (Life Technologies Corp., Carlsbad, CA, USA) and run under the same cycling conditions as described above for 40 cycles.
Statistical analysis. Means or medians were compared using Student's t test or a Mann-Whitney test. Data are shown as means Ϯ standard deviations unless otherwise stated. A standard curve of lytA was generated with known DNA concentrations and plotted against the threshold cycle (C T ) to yield the copy number, calculated as 10 (CT Ϫ 33.701)/Ϫ3.4262 .

RESULTS
Analytical assay performance. The 53 serotype/serogroup-specific singleplex PCR assays as well as the S. pneumoniae lytA assay were tested against 54 serotyped pneumococcal isolates as well as several nonpneumococcal species, and 100% specificity with no cross-reactivity was observed (see Fig. 1 in the supplemental material). The PCR performance of each primer/probe assay was determined on both 384-well plates and the TaqMan array card formats. DNA from both individual and pooled serotypes was tested in serial dilution. The overall linearity of the 53 serotype assay targets including lytA was 0.997 Ϯ 0.01 and 0.986 Ϯ 0.02, and the PCR efficiencies were 93% Ϯ 4.9% and 97% Ϯ 9.7% using the 384-well plates and TaqMan array card formats, respectively. The limit of detection on 384-well plates was 10 to 100 fg/ reaction while that of TAC was 20 to 200 fg/reaction ( Table 2).
Evaluation of TaqMan array card serotyping on clinical isolates. The performance of the TaqMan array card was evaluated on 54 S. pneumoniae serotypes (previously serotyped by Quellung reactions) and then on 92 blinded isolates, which included sero-  (Table 3). There was 100% specificity, and all four discordant samples were indicated as serotype 22F by Quellung (performed twice) but negative with the TAC 22AF assay. This was not due to 22AF assay failure since it was positive for three serotype 22A strains.
TaqMan array card serotyping on direct specimens. Twentyeight nasopharyngeal specimens underwent DNA extraction and TAC testing, and the results were compared with the culture and Quellung results (22). In 23 of these nasopharyngeal specimens an S. pneumoniae strain had been isolated whereas five were culture negative. The TAC yielded a lytA-positive result in 21/23 culturepositive specimens and was positive in 1/5 culture-negative specimens (C T of 34), corresponding to a TAC lytA sensitivity of 91% and specificity of 80% versus culture ( Table 4). The serotype assays were exclusively positive in lytA-positive specimens (TAC serotype assays, 100% specificity versus lytA). In the 22 lytA-positive specimens, a serotype result was obtained for 19 specimens (TAC serotype assays, 86% sensitivity versus lytA). The TAC serotype result matched the Quellung result in 18 specimens (95% accuracy), the exception being one pneumococcus specimen nontypeable by Quellung that was serotype 4 by TAC (with a somewhat late C T of 30 compared with the other serotype 4 strains). There were two culture-positive specimens undetectable by lytA on the TAC, one of which was lytA-positive when larger volumes of DNA were used (1 l; C T of 37) in a plate format and one of which remained negative. Likewise, we retested the three lytApositive/serotype-negative samples with the serotype assays with larger volumes of DNA, and one became positive (1 l; C T of 36). We used the TAC lytA C T to extrapolate the pneumococcal concentration in the nasopharyngeal specimens, and the two discrepant results (e.g., lytA positive/serotype negative) were both at the lowest concentration of 3 ϫ 10 3 to 4 ϫ 10 3 CFU/ml.

DISCUSSION
In this work we describe the development of a TaqMan array card that compartmentalizes 53 reactions to detect 74 pneumococcal serotypes and that can be used on isolates or nasopharyngeal spec-   imens. Once developed, the TAC assay is as simple to perform as a single PCR. The TAC assays exhibited excellent linearity and limits of detection, albeit they were slightly less sensitive than the assays in a plate format, where more DNA template can be added. This slight sensitivity loss may not be clinically deleterious, and certainly the procedural advantage of the TAC versus setting up 54 singleplex PCRs is enormous.
For isolates where abundant DNA is available, performance remained excellent, with 97% accuracy versus the Quellung result. Indeed, the card had 100% accuracy on blinded isolates from a wide variety of 24 serotypes, including all of the PCV13 strains.
Discrepancies were observed only with 22F strains. Curiously, wcwA sequences, according to GenBank accession numbers CR931681.1 and CR931682.1, are identical between 22A and 22F in the primer region, yet the 22F strains did not amplify, while 22A did, suggesting that there is a disconnect between the available 22F GenBank sequence and these four strains (see Fig. 2 in the supplemental material). Future iterations of the assays can attempt to understand and improve this. This caveat aside, the assay is clearly robust for use on cultured isolates from invasive sites and should provide an important tool to document whether serotype replacement is occurring in invasive strains (28). F, forward primer; R, reverse primer; P, probe labeled with FAM (6-carboxyfluorescein) except for the probes for the serotypes marked with asterisks, which are labeled with VIC at the 5= end. All are 3= minor groove binder probes. b Atypical 19F.
The assay also worked well on direct nasopharyngeal specimens, with a 91% sensitivity versus culture and an 86% accuracy of the serotype result versus the Quellung reaction on lytA-positive specimens. A few samples had low levels of DNA at the lytA or serotype level that could be rescued with larger amounts of DNA. Thus, the serotype result was 100% accurate for any lytA result of a C T of 34 or below (corresponding to a nasopharyngeal density of 8 ϫ 10 3 CFU/ml), which is how we would propose using the assay. This assay is suitable for monitoring pneumococcus density and mixed infections in nasopharyngeal specimens, which is of great interest in the effort to better document the phenomenon of serotype replacement in the nasopharynx after vaccination (6). Regarding mixed infections, there was one discrepant nasopharyngeal specimen which was nontypeable by Quellung but serotype 4 by TAC, which we hypothesize was mixed. It is also plausible that this TaqMan array card includes eight sample ports, whereby each sample is aliquoted into 48 PCRs. Serotypes in the form AB or A-B indicate a common assay that detects multiple serotypes/serogroups. Serotypes in the form A/B* indicate a duplex assay. specimen represents a weakly expressing strain. We think that applying TAC to nasopharyngeal colonization will be particularly useful to monitor vaccine effectiveness in communities over time, ensuring that vaccine types are being eliminated as expected. Nasopharyngeal materials are much easier to obtain than invasive isolates, particularly in children and in resource-limited settings where the burden is highest. It has been proposed that the absence of a vaccine type in nasopharyngeal specimens in children with pneumonia could be used as a surrogate for vaccine effectiveness (11).
Limitations of our study are that the number of direct specimens with culture-and serotype-confirmed results was small; thus, the sensitivity and specificity estimates of the TAC assay are approximate, and additional evaluation will be beneficial. Other investigators have found that nonpneumococcal streptococcal species can interfere with serotyping assays (29), so more direct specimen testing is needed. Although the limit of detection by TAC was within the range of other reported assays (26), it was 2-fold higher than that of the regular real-time PCR format (22). Some serotypes could not be identified individually by single assay sets. For example, to infer serotype 6AB we must detect serotypes 6ABCD in the absence of serotypes 6CD (Fig. 1). We certainly suspect that the serotype reactions may need to be modified over time to include alternate types.
We embarked on this project because we have demonstrated excellent performance and reproducibility of the TAC platform in multisite field studies in Africa and Asia (30), areas of high pneumococcal carriage, coinfection, and variable serotype distributions. While the real-time PCR instrument is costly (ϳ$75,000), it also performs routine real-time PCR. To our knowledge, the TAC platform exists in at least 13 countries across sub-Saharan Africa and South Asia. The TAC cards are stable at 4°C for at least 2 years and cost about $50 per specimen, or approximately $1 per reaction, which compares favorably with conventional Quellung testing, which can cost up to $100 per colony (12). In conclusion, the