Streptococcus mitis Expressing Pneumococcal Serotype 1 Capsule

Streptococcus pneumoniae’s polysaccharide capsule is an important virulence factor; vaccine-induced immunity to specific capsular polysaccharide effectively prevents disease. Serotype 1 S. pneumoniae is rarely found in healthy persons, but is highly invasive and a common cause of meningitis outbreaks and invasive disease outside of the United States. Here we show that genes for polysaccharide capsule similar to those expressed by pneumococci were commonly detected by polymerase chain reaction among upper respiratory tract samples from older US adults not carrying pneumococci. Serotype 1-specific genes were predominantly detected. In five oropharyngeal samples tested, serotype 1 gene belonging to S. mitis expressed capsules immunologically indistinct from pneumococcal capsules. Whole genome sequencing revealed three distinct S. mitis clones, each representing a cps1 operon highly similar to the pneumococcal cps1 reference operon. These findings raise important questions about the contribution of commensal streptococci to natural immunity against pneumococci, a leading cause of mortality worldwide.

aged 65 years and older. Our results suggest that serotype 1-containing S. mitis may be common among US adults. In addition, we describe results of tests of genetic similarity and serologic cross-reactivity with serotype 1 pneumococci.

Methods
Study Design and Population. We conducted a cross-sectional survey from July 13, 2015 through March 31, 2016 among healthy adults 65 years of age and older enrolled from outpatient clinics, senior residential communities, senior day care centers and health fairs in four US states participating in the Centers for Disease Control and Prevention's (CDC) Active Bacterial Core surveillance: Georgia, Maryland, New York and Tennessee. We excluded adults if they reported severe immunocompromising conditions (e.g., hematologic malignancies, transplant recipient, end-stage renal disease) 13 or if they were residents in nursing homes or skilled nursing facilities. For each eligible adult who consented to participate, we collected one nasopharyngeal (NP) and one oropharyngeal (OP) swab and conducted an interview to obtain demographic, clinical and vaccination data. Participants without paired NP and OP specimens were excluded from the study. Trained staff followed up with healthcare providers to confirm PCV13 receipt and dates of immunization.
The study was approved by the institutional review boards of the Centers for Disease Control and Prevention (CDC -Protocol #6725), Emory University, Georgia Department of Public Health, Maryland Department of Health, Johns Hopkins School of Public Health, University of Rochester, Vanderbilt University, and Tennessee Department of Health. Written informed consent was obtained from all study participants, and the research was performed according to federal regulations for protection of human subjects.
Specimen collection and processing. Trained staff collected NP and OP specimens using sterile nylon tipped flocked swabs (Copan Diagnostics INC, Murrieta, CA). NP and OP specimens were placed separately into cryovials containing 1.0 ml skim milk-tryptone-glucose-glycerol (STGG) medium, vortexed for 10 to 20 seconds to disperse the organisms from the swab, and frozen at −70 °C within 4 hours of collection.
Each swab was processed separately. For pneumococcal isolation and identification, 200 µl of either STGG-NP or STGG-OP inoculated medium was transferred into 5.0 ml Todd Hewitt broth containing 0.5% yeast extract (THY) and 1 ml of rabbit serum, which was then incubated at 37 °C in a CO 2 incubator for 5-6 hours. After incubation, 10 µl of cultured broth was streaked on tryptone soy agar plates with 5% sheep blood (BAP) and incubated at 37 °C in CO 2 for 18-24 hours 14 . Alpha-hemolytic colonies resembling pneumococci were tested for susceptibility to optochin and bile solubility, and serotyped by Quellung using CDC pneumococcal typing antisera 15 . Pneumococcal culture-negative NP and OP specimens underwent real-time PCR targeting lytA for detection of pneumococcal DNA 16 . DNA extraction was performed using a MagNA Pure Compact with an external lysis protocol using buffer #4 (Roche isolation kit III) or NucliSENS ® EasyMag ® automated system for total nucleic acid extraction, according to manufacture instructions. Specimens were considered to be lytA-positive if the cycle threshold (Ct) value was ≤35. Real-time multiplex PCR for detection of 37 serotypes (including PCV13 serotypes) was performed on culture-negative specimens from 53 and 395 adults with lytA-positive and lytA-negative specimens, respectively. For lytA-negative specimens, we assumed that 30% (+/−5%) would have a serotype detected by real-time PCR leading to a target sample size of 395; while for lytA-positive specimens almost all of them (71%) were tested. PCR for serotyping was performed using DNA extracted from 200 µl of each NP and OP swab-inoculated STGG media 17 . Because real-time PCR for serotyping is a multiplex reaction, Ct ≤35 for a specific serotype was considered positive, while Ct values between 36 and 40 were considered equivocal and repeated three times for confirmation.
Non-pneumococcal species isolation and characterization. The five specimens with the lowest PCR cycle threshold for serotype 1-specific wzy1 gene, which lies within the pneumococcal cps1 operon, were cultured using broth enrichment followed by plating on blood agar for isolation of alpha-hemolytic streptococcal species; these five were all OP specimens, two lytA positive and three lytA negative. Non-pneumococcal alpha-hemolytic streptococci colonies grown on the plates that were optochin resistant and bile insoluble were screened for wzy1 by real-time and conventional multiplex PCR. wzy1-positive non-pneumococcal isolates (N = 5, 1 from each specimen) were subjected to lytA screening by PCR, immunochromatographic testing (BinaxNow), and whole genome sequencing on Illumina MiSeq 18 . We compared the sequences of the 16,000 bp non-pneumococcal streptococci cps1 operons and flanking regions with the corresponding sequence of the 22,000 bp pneumococcal serotype 1 reference strain (GenBank accession CR931632) using the SSEARCH program 19 .
Species assignment was done by phylogenetic analysis of seven previously described concatenated housekeeping gene sequences (total length of 3063 bases) 20 . A selection of 3063 bp sequences 20 that most closely matched the non-pneumococcal streptococcal strains were used to construct the phylogenetic tree. In addition, reference strain sequences were obtained from NCBI accessions for S. pseudopneumoniae (ATCCBAA960), S. mitis (B6 and NCTC 12261), S. infantis (ATCC 700779), and S. oralis (Uo5). Evolutionary distances were computed using the Maximum Composite Likelihood method 21  Capsule expression of non-pneumococcal streptococci isolates. Latex agglutination followed by the Quellung reaction using polyclonal serotype 1 S. pneumoniae typing antiserum prepared in rabbits was performed on the five isolates to assess capsule expression and compare capsule expression with that of S. pneumoniae serotype 1 (positive control). In addition, double immunodiffusion assays using the same antiserum assessed cross-reactivity of non-pneumococcal streptococci with serotype 1 pneumococcal antisera using the methods described by Sorensen et al. 23  Opsonophagocytosis of pneumococcal and non-pneumococcal streptococci isolates. An opsonophagocytic killing assay 26 was used to assess functional antibody activity against both the non-pneumococcal streptococci strains expressing serotype 1 capsular polysaccharide and the serotype 1 pneumococcus. S. mitis ATCC (NCTC 12261) strain (PCR negative for wzy1 gene) was used as a control. In a 96-well micro titer plate, 20 μl of S. mitis serotype 1 or serotype 1 pneumococcus was pre-opsonized with 10 μl anti-pneumococcal serotype 1 specific human polyclonal serum (8-dilutions, diluted 2-fold starting neat) or anti-S. mitis serotype 1 rabbit serum for 15 minutes at 37 °C with 5% CO 2 . After pre-opsonization, 10 μl baby rabbit complement was added followed by 40 μl of human promyelocyte (HL60) derived neutrophils (effector cells). Complement, neutrophil, and bacterial controls were maintained. After incubation for 45 minutes at 37 °C, 5 μl from each well was transferred onto air-dried Todd-Hewitt yeast extract agar. Opsonophagocytic titers were the reciprocal of the serum dilution with >50% killing compared with the growth in the complement control wells.
Data Analysis. Analyses were performed using SAS software, version 9.3. We calculated the prevalence of pneumococcal carriage as determined by culture. Among culture-negative patients, we calculated the prevalence of serotype genes detected by PCR among patients with lytA-positive or lytA-negative specimens. Comparison of the prevalence of pneumococcal vaccine-serotype homolog genes (serotypes 1, 3, 4, 5, 6A, 6B, 7F, 9V, 14, 18C, 19A, 19F, 23F) by PCV13 status was done using the chi-square test. Patients were considered vaccinated if PCV13 was received at least 14 days before swab collection. Serotypes not distinguished by PCR (e.g. 9A/9V, 7A/7F, The data that support the findings of this study are available from the corresponding authors upon reasonable request. The strains are stored in the CDC's Streptococcus laboratory and can be available to investigators upon discussion and approval by the authors and site investigators. The method (composition of the strain with the purpose to induce immune response) is the subject of a U.S. Provisional Patent Application, No. 62/627,602 filed on February 7, 2018; non-exclusive royalty-free (NERF) academic research use licenses are available upon request.
Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Results
We enrolled 1,299 adults who had both NP and OP specimens successfully collected; of these, 67.9% were female, median age was 74 years (range: 65-102 years), and 73.9% were either recruited in Georgia or New York. Pneumococcus was isolated from specimens from 16 (1.2%) participants; 12 from NP, 2 from OP and 2 from both NP and OP. Of the 1,283 participants with culture-negative samples, 75 (5.8%) had a specimen that was lytA-positive: 1 from NP only, 71 from OP only, and 3 from both NP and OP (Supplementary Material, Figure).
PCR-based serotype deduction was performed on both NP and OP specimens from 448 culture-negative adults. Among the 53 adults with lytA-positive specimens from either NP or OP, 8 (15.1%) had wzy1 detected in one or both specimens; among the 395 adults with lytA-negative specimens from both NP and OP, 41 (10.4%) had wzy1 detected. Of the 49 adults with wzy1-positive, culture-negative specimens, 42 (85.6%) had wzy1 detected only in the OP specimen, 4 (8.2%) in both NP and OP specimens, and 3 (6.1%) in the NP specimen only; 23 (46.9%) of these 49 were enrolled in Georgia, 11 (22.5%) in Tennessee, 10 (20.4%) in New York and five (10.2%) in Maryland.
Of the 53 and 395 adults with lytA-positive and lytA-negative specimens, 27(50.9%) and 131(33.2%) were positive by PCR for genes for other pneumococcal serotypes; although overall wzy1 was most commonly detected (Table 1). Compared to NP specimens, OP specimens were more likely to be PCR-positive for serotype-specific pneumococcal genes in adults with lytA-positive (50.9% [27/53]

Isolation of non-pneumococcal strains and characterization. Streptococcus mitis was identified from
culture from all 5 of the OP specimens, three lytA-negative (ID# L006, L115, L116) and two lytA-positive (ID# L121, L164), with the lowest cycle threshold values for wzy1 (between 22.0 and 27.5). The S. mitis isolates were positive for the serotype 1-specific wzy1 gene by both real-time and conventional PCR, and by latex agglutination for serotyping (Table 2). Two S. mitis isolates were Binax positive.
Whole genome sequence analysis of these five S. mitis isolates revealed three very similar, although distinct, capsule gene operons and adjacent genes. The lengths of the three loci, corresponding to the region between the flanking dexB and aliA segments of the 22,182 bp pneumococcal reference (GenBank accession CR931632), ranged between 16,613-16,626 bp and shared 98.0-98.2% sequence identity with one another. Each of the 11 genes within the polysaccharide synthetic cluster (wzg -ugd) were highly homologous to the S. pneumoniae sero-type1 counterparts and shared the same order (Fig. 1).
Three distinct S. mitis clones were found (Fig. 2). While 16 different pneumococcal strains representing 6 different serotypes formed a tight cluster based upon phylogenetic analysis of 3,063 bp concatenated housekeeping gene segments, the S. mitis strains formed a much more divergent cluster. At the level of direct DNA sequence comparison, the 3 serotype 1 S. mitis clones shared 97. pneumococcal strains shared 99.5-99.9% sequence identity. These S. mitis strains were all macrolide-resistant (all contained erm-methylase or mef genes) and cotrimoxazole-resistant (alterations within folA and/or folP genes). As in pneumococci, the cps loci in the S. mitis strains were situated between the penicillin binding protein genes pbp2x and pbp1a. All five S. mitis strains lacked the pneumococcal-specific target piaA but contained pneumolysin genes with 61-97% sequence identity to the pneumococcal ply gene; suggesting that piaA is pneumococcal-specific, whereas ply may be found within S. mitis strains with varying degrees of similarity with pneumococci.
Cross-reactivity between S. mitis and serotype 1 S. pneumoniae anti-serum and between S. pneumoniae serotype 1 and anti-S. mitis serotype 1 anti-serum. Although subtle, pneumococcal serotype 1 capsule was visualized by Quellung reaction in three of the five S. mitis strains (L006, L121, L164; Table 2, Fig. 3a). Strains L006 and L121 representing two of the three S. mitis clones were tested using double immunodiffusion assays and showed strong cross-reactivity with S. pneumoniae serotype 1 typing anti-serum. Figure 3b shows side-by-side S. mitis serotype 1 antigen and S. pneumoniae serotype 1 antigen that generated symmetrical and fused precipitates in reaction with serotype 1 typing S. pneumoniae antisera. Based on an opsonophagocytosis killing (OPK) assay, capsule expressed by serotype 1 S. mitis and S. pneumoniae strains both triggered anti-serotype 1 antibody mediated opsonophagocytosis, while the S. mitis control strain (wzy1-negative) did not. In addition, opsonophagocytosis of serotype 1 pneumococci was observed with anti-serotype 1 S. mitis serum obtained from rabbits using L006 strain (Fig. 4).

Discussion
In our large, 4-site study we found that few older US adults carried pneumococci, as detected either by culture or PCR for lytA on NP and OP specimens. We observed that, in spite of this lack of pneumococci, about 1 in 3 older adults had genes for known pneumococcal polysaccharide serotypes detected by PCR, and the gene for serotype 1 (wzy1) was the type most commonly detected. The finding that genes for PCV13 serotypes were significantly less common among adults who had received PCV13 suggested that the genes detected by PCR were likely linked to organisms that expressed capsules that cross reacted with antibodies generated by PCV13. Indeed, we found that S. mitis identified from 5 wzy1-positive specimens expressed capsules that were nearly identical to pneumococcal serotype 1 capsules by a range of common detection assays, genetic sequencing of the cps locus, and serologic assays. Over 10% of the adults without pneumococcus detected within their upper respiratory tracts screened positive for the wzy1. This observation, combined with the characterization of the three distinct serotype 1 clones found in our study, suggests that a wide diversity of serotype 1 S. mitis strains may circulate within the adult US population. In our study population of adults aged over 64 years, carriage of wzy1-positive non-pneumococcal streptococci appeared to be more common than overall pneumococcal carriage. S. mitis colonization of the upper respiratory tracts persists from early infancy through adulthood, while S. pneumoniae colonization declines significantly with age 27 . In contrast to S. pneumoniae, S. mitis is only occasionally associated with disease 28 . Because the polysaccharide capsule is thought to be an important virulence factor for pneumococci, whether S. mitis that cause disease are typically those expressing a capsule should be explored. Other recent studies have shown that many S. mitis strains express a polysaccharide capsule, and animal models demonstrated that the capsule has a protective function in the clearance of S. mitis 23,29,30 . In addition, Engen et al. 31   carriage and disease 31 . The commensal S. mitis strains expressing a serotype 1 capsule that we found in healthy older US adults may induce a T cell response against serotype 1 S. pneumoniae, which in turn could theoretically provide protection against serotype 1 pneumococcal disease among colonized individuals. Our OPK assay results indicate that anti-serotype 1 pneumococcal human antibodies can clear S. mitis expressing serotype 1 capsule and that anti-serotype 1 S. mitis rabbit antibodies can clear serotype 1 pneumococci. If colonization with S. mitis serotype 1 strains induces an antibody response that can clear serotype 1 pneumococci, the notion that immunity induced by these relatively abundant and naturally occurring strains may explain the limited incidence of serotype 1 pneumococcal disease in the United States. We found three distinct clones among the five S. mitis isolates. All three S. mitis clones shared wzy1 operons that were highly similar to wzy1 of the serotype 1 S. pneumoniae reference strain. We observed more genetic distance between the three S. mitis serotype 1 clones than among the 16 pneumococcal strains of different capsular serotypes, suggesting that S. mitis serotype 1 strain diversity is much higher than that observed between pneumococcal serotype 1 strains. Although some studies have suggested that capsular polysaccharides in S. pneumoniae evolved by importing relevant genes from a range of commensal Streptococcus species 23,32 , the cps1 locus may have been transferred from pneumococcal strains to S. mitis strains, given the high level of background diversity found among S. mitis with serotype 1 capsule genes, or have been derived from a species that is no longer circulating. Therefore, predicting the directionality of serotype capsule genes transfers between S. pneumoniae and S. mitis is difficult.
Differences in the distribution of pneumococcal serotypes causing disease have been observed among populations 4 . However, more work is needed to understand if the distribution of capsular polysaccharide serotypes in   commensal streptococci affect the pneumococcal serotypes causing disease. Cross-reactivity between encapsulated S. mitis and S. pneumoniae has recently been reported by the Danish group using immunodiffusion assays 23 . However, serotype 1 S. mitis were not observed among the 66 S. mitis isolates examined, which were mostly from Denmark (N = 42) and non-vaccine serotypes. In a Kenya carriage study, we were also unable to detect wzy1positive non-pneumococcal Mitis group strains from NP or OP specimens from 158 adults 33 . In both Denmark and Kenya, serotype 1 is a common serotype causing IPD, while in the United States this serotype has been an uncommon cause of disease in recent decades 2,34,35 .   Among our study participants, pneumococci were rarely isolated from OP specimens, yet these specimens were much more likely than NP specimens to be positive by PCR for vaccine-serotype polysaccharide capsule genes; often such specimens were lytA-negative, providing evidence that the source of the serotype genes were not pneumococci. The oral cavity contains a great diversity of streptococcal species, and pneumococcal PCR serotyping assays can often identify homologs among mitis group non-pneumococcal strains 33 . Pneumococcal carriage studies using PCR-based detection of serotype-specific genes directly on OP specimens can overestimate vaccine type pneumococcal carriage rates given the high prevalence of confounding non-pneumococcal species carrying homologs of pneumococcal-serotype specific genes. In addition, the lytA gene for pneumococcal detection is not entirely specific. A recent study from Portugal identified a lytA-positive S. pseudopneumoniae strain in sputum 36 . In our study, two of 224 non-pneumococcal streptococci colonies screened for lytA were positive (data not shown). The use of PCR-based detection directly on upper respiratory specimens is of questionable value for pneumococcal carriage studies designed to measure vaccine impact based on the high prevalence of serotype 1 and other serotype genes on specimens that did not contain pneumococci. Although OP specimens had higher positivity for wzy1, 7 NP specimens that were lytA-negative were wzy1-positive by real-time PCR. Other studies have also shown problems with real-time PCR for serotyping of NP specimens 37 .
In conclusion, we found that genes for polysaccharide capsules similar to those expressed by pneumococci, in particular genes for serotype 1, were common among our sample of older US adults. The serotype 1 genes belonged to S. mitis strains that expressed capsules immunologically indistinguishable from pneumococcal capsules. Our findings raise many questions, in particular about the role of commensal streptococci in the development and maintenance of natural immunity to pneumococci. Further work should focus on these questions and whether the presence of capsule-containing non-pneumococcal streptococci affects the serotype distribution and amount of disease caused by pneumococci in different population. These strains could provide a novel approach for enhancing the human microbiome's ability to protect against serotype 1 pneumococcal disease, which is responsible for 11.7% of all IPD cases in Africa 4 and continue to be the main serotype causing pneumococcal meningitis outbreaks in the African meningitis belt countries that have already introduced pneumococcal conjugate vaccine 8 . Given the genetic similarities between capsular loci, those working with molecular assays to evaluate pneumococcal serotypes in the upper respiratory tract should consider whether non-pneumococcal streptococci strains with capsular genes may affect their results. . Opsonophagocytosis killing (OPK) testing of S. mitis and serotype 1 S. pneumoniae isolates using anti-serotype 1 pneumococcal and non-pneumococcal serum. Serotype 1 S. pneumoniae = positive control; S. mitis ATCC (NCTC 12261) strain confirmed to be wzy1-negative = negative control. Antibody source for OPK assay: 007 S. pneumoniae human reference serum. Complement: baby rabbit complement; OPK titer: reciprocal of serum dilution with >50% growth compared to serum free complement control. (S. mitis strain L006 and L121 representing two distinct S. mitis clones tested). Error bars represent the lower and upper limit of OPK titer based on the test runs.