ABSTRACT
Pneumococcus (Streptococcus pneumoniae) remains a significant cause of morbidity and mortality, especially among those at the extremes of age. Its capsular polysaccharide is essential for systemic virulence. Over 90 serologically distinct pneumococcal capsular polysaccharides (serotypes) are recognized, but they are unequal in prevalence. Because antibodies against the capsule are protective, polysaccharide conjugate vaccines, which are constructed against the most prevalent serotypes, have caused great reductions in pneumococcal disease caused by these serotypes. In response, however, the relative prevalences of serotypes have shifted. Certain previously rare serotypes, such as serotype 35B, are increasing in prevalence. Serotype 35B is thus a likely future vaccine candidate, but due to their previous rarity, serotype 35B strains have not been scrutinized for underlying heterogeneity. We studied putative serotype 35B clinical isolates to assess the uniformity of their serological reactions. While most isolates exhibited the accepted serology of serotype 35B, one isolate failed to bind to critical serotyping reagents. We determined that the genetic basis for this aberrant serology was the presence of inactivating mutations in the O-acetyltransferase gene wciG. Complementation studies in a wciG deletion strain verified that the mutant WciG was nonfunctional, and the serology of the mutant could be restored through complementation with a construct encoding a functional WciG. Nuclear magnetic resonance studies confirmed that the capsule of the WciG-deficient isolate lacked O-acetylation but was otherwise identical to serotype 35B. As this isolate expresses a unique serology with unique biochemistry and a stable genetic basis, we named its novel capsule serotype 35D.
INTRODUCTION
Streptococcus pneumoniae (pneumococcus) is an important human pathogen that can express many different polysaccharide capsules that protect the organism from opsonophagocytic killing (1). As antibodies to the capsule are protective, currently available vaccines against pneumococcus are designed to elicit antibodies to the capsule and have been highly successful in reducing the incidence of invasive pneumococcal disease (IPD). A pneumococcal polysaccharide-protein conjugate vaccine (PCV) with polysaccharide from the seven most commonly observed serotypes (PCV7) was introduced in the United States in 2000 and reduced the incidence of IPD in children younger than 5 by over 75% (2); the updated 13-valent PCV (PCV13) has reduced the incidence of IPD by an additional 64% in this age group in the United States (3). However, the protection by these vaccines is serotype specific, and their widespread implementation has brought shifts in the serotypes found in both nasopharyngeal carriage and IPD (see, e.g., reference 4).
Because pneumococcal vaccines provide serotype-specific protection, the efficacy of a given vaccine is inherently dependent upon the serotype distribution in the target population. Thus, it is essential to monitor the epidemiology of pneumococcal serotypes to formulate pneumococcal vaccines and to assess their efficacy. The classical serotyping method, the Quellung reaction, is based on the reaction of serotype-specific antisera with the corresponding capsule (5); this method is time-consuming, requires many reagents, and requires a high degree of operator expertise. Since the capsular genetic loci (cps loci) for all known serotypes have been described (6–12), many serotyping methods are now based on detecting the genetic hallmarks of each serotype; such methods are less expensive and require less operator expertise. Yet, current genetic assays rely on the association of certain sequences with certain serotypes (reviewed in reference 13) without evidence that those sequences are responsible for those serotypes. Small genetic differences differentiate serotypes that have only subtle differences in structure yet profoundly different interactions with immunity, both vaccine-elicited and innate. Thus, indirect genetic “serotyping” methods may fail to identify novel serotypes.
Invasive disease by serotype 35B has been rare historically, accounting for 0.5% of IPD in the 2 years prior to PCV7 introduction in the United States (2); consequently, 35B is not included in current pneumococcal vaccines. However, the prevalence of serotype 35B in both IPD and nasopharyngeal carriage has been increasing with the use of PCVs in the United States (see, e.g., references 3 and 4) and other countries (14, 15). Because the genetic approach is an indirect method, we studied the serologic properties of a collection of isolates putatively assigned as serotype 35B genetically. We found that most of these putatively serotype 35B isolates exhibit serology typical of serotype 35B. However, one isolate reproducibly failed to bind to factor serum 35a. We investigated the genetic basis and biochemical structure of this isolate's capsule and discovered a novel serotype, which we have named 35D.
RESULTS
Heterogeneity exists among isolates assigned to serotype 35B by genetic means.To investigate serologic properties of putative serotype 35B isolates, we identified 30 clinical isolates determined to be serotype 35B by a genetic method (16) (Table 1) and examined them for their abilities to bind commercial serotyping antisera. By flow cytometry, all isolates but one bound to pool G, group 35 antiserum, and factor sera 35a, 35c, and 29b, consistent with the accepted serotype 35B serology (see Table 2 for the accepted serological profiles [17]; representative flow cytometric data are shown in Fig. 1). The remaining isolate, 3431-06, reproducibly failed to bind to group 35 antiserum and factor serum 35a while retaining serotype 35B-like properties with respect to the other factor sera (Fig. 1 and Table 2). (It is worthwhile to note that serotype 29 reacted with factor serum 35c [Table 2]; while 35c is not officially part of the serotype 29 antigenic formula, this binding was reproducible in our hands.) Notably, all putative serotype 35B isolates bound serotype 29 antiserum; thus, as a consequence of lost factor serum 35a binding, strain 3431-06 serologically resembled serotype 29 but lacked the factor serum 35b reactivity observed with serotype 29 (Fig. 1, “29/2,” and Table 2). Therefore, we concluded that strain 3431-06 represented a novel serotype that is similar to, but distinct from, serotypes 29 and 35B. We refer to this novel serotype as serotype 35D.
Strains used in this study
Serological profiles of serotypes in this studya
Isolate 3431-06 fails to bind to group 35 antisera and factor serum 35a by flow cytometric serotyping assay. Histograms are presented for the staining of indicated strains (serotype in parentheses) by the indicated serological reagent. Gray shading indicates staining of an irrelevant serotype control (serotype 4 strain TIGR4 [34]) to control for noncapsular antibodies in the typing antisera. Staining of strain 35B/2 (a serotype 35B reference strain) is representative of results obtained with strains listed in Table 1. fs, factor serum.
Isolate 3431-06 is a wciG-deficient variant of serotype 35B.To determine the genetic basis for the novel serologic properties of isolate 3431-06, we sequenced its cps locus (submitted to GenBank accession no. KY084476 ) and compared it to a serotype 35B reference cps locus (GenBank accession no. CR931705 ). The comparison showed that the 3431-06 cps locus has two missense mutations in wciG, which encodes a putative O-acetyltransferase. The mutations result in amino acid substitutions R178K and A227T in WciG. The cps locus showed no further deviations from the reference sequence.
To verify that wciG functionality was responsible for the novel capsule of 3431-06, we deleted wciG in a serotype 35B clinical isolate, 3009-06. The resulting strain, KAG1014, was serologically identical to 3431-06 (Fig. 2). Complementation of KAG1014 with wciG of 3431-06 failed to restore factor serum 35a binding (Fig. 2, KAG1019), while complementation with wciG from serotype 35C (used for technical reasons, see Materials and Methods) was able to restore binding in both KAG1014 and 3431-06 (Fig. 3, KAG1023 and KAG1022). This strongly suggested that the mutant WciG of 3431-06 is nonfunctional and that WciG functionality is the determinant of both group 35 antiserum and factor serum 35a recognition.
The unique serology of serotype 35D is due to loss of WciG functionality in a serotype 35B background. The flow cytometric staining of bacteria (strain description in parentheses) by the indicated serological reagents is presented as histograms. The serotype 4 isolate TIGR4 (34), with gray shading, was used to control for binding by noncapsular antibodies. A strain staining at or below the irrelevant serotype control was considered negative for binding. fs, factor serum.
(A) Overlay of 1H NMR spectra of serotypes 35B (top) and 35D (bottom) polysaccharides at 25°C. The serotype 35D spectrum has lost the O-acetyl peak at 2.12 ppm (arrowhead), and shifts in the anomeric region (especially ≥5 ppm [1H]) are observed in consequence (see Table 3). (B) The 1H-13C HMQC spectra of 35B (left) and 35D (right) show the loss of the O-Ac peak at 2.12 ppm (1H).
Isolate 3431-06 capsular polysaccharide lacks O-acetylation.Genetic and serological evidence suggested that the 35D capsular polysaccharide is a de-O-acetylated serotype 35B capsule. To confirm this biochemically, we purified the capsule from isolate 3431-06 and analyzed it via nuclear magnetic resonance (NMR) spectroscopy. For a direct comparison with the serotype 35D isolate, identical NMR experiments were also collected for serotype 35B polysaccharide. Complete assignments of 1H and 13C signals (Table 3) for serotype 35B and 35D polysaccharides were achieved by two-dimensional (2D) NMR experiments, as described in Materials and Methods. The chemical shifts are very similar to those described for serotype 35B polysaccharide and its O-deacetylated form (18). Although the spectra of serotypes 35B and 35D are largely similar, some significant differences are observed in the anomeric and acetyl regions (Fig. 3A). As shown in Fig. 3B, the 1H-13C signal corresponding to the O-acetyl group at 2.12/20.24 ppm (1H/13C) is observed for serotype 35B but not for serotype 35D, consistent with the loss of capsular O-acetylation for serotype 35D. A second signal corresponding to the N-acetyl group is observed at 2.04/22.37 ppm (1H/13C). Because the chemical shifts of the polysaccharide are very similar except for groups directly affected by acetylation, our data demonstrate that loss of the O-acetyl group does not alter the structure of the polysaccharide, a finding that is consistent with the previously reported structure of the serotype 35B repeat unit.
1H and 13C chemical shifts (ppm) of 35B and 35D obtained at 25°C
DISCUSSION
Herein, we describe a new serotype, 35D, which is genetically very similar to serotype 35B but differs in having inactivating mutations in wciG. wciG encodes a putative O-acetyltransferase, and two minor mutations apparently render WciG inactive. As a result, serotype 35D produces a capsule that differs from 35B polysaccharide in structure and immunoreactivity. The serotype 35D capsule is identical to that of serotype 35B except for the absence of an O-acetyl group at one galactofuranose (Galf) (Fig. 3). Immunologically, serotype 35D is identical to serotype 35B except for its nonreactivity to factor serum 35a and the group 35 antiserum (Fig. 1 and Table 2). Because serotype 35D has a heritable genetic alteration, a unique polysaccharide structure, and a distinct immunologic profile, it represents a novel serotype.
Selecting a name for this serotype was problematic. The new serotype does not react with group 35 antiserum or factor serum 35a, both of which are accepted to react with all members of serogroup 35 (17). In addition, serotype 35D is serologically and chemically similar to serotype 29. Serologically, they differ only by their reactions with factor serum 35b (Table 2), and their polysaccharide structures differ by only one residue (glucose [Glc] versus galactose [Gal]). Indeed, classical serotyping approaches may have mistyped serotype 35D as serotype 29 because of their serologic similarity and the failure of 35D to react with factor serum 35a, a requisite for assignment to serogroup 35 using Quellung serotyping (19). However, the cps locus of 35D is almost identical to the serotype 35B cps locus and quite different from the serotype 29 cps locus. Consequently, many popular genotyping methods, which are increasingly favored over Quellung serotyping, would have assigned a serotype 35D isolate as serotype 35B. Thus, we have chosen to include this new and 98th (by our count) serotype in serogroup 35 as serotype 35D. Serotyping schemes of either type should be revised to take this closely related serotype into account.
Using the serologic properties of serotype 35D, we have attempted to deduce the minimum chemical structures targeted by several factor sera (presented as a schematic in Fig. 4) while excluding currently known structures that are not known to cross-react (13, 17). Since WciG-mediated O-acetylation correlates with factor serum 35a, factor serum 35a must target the O-acetyl group conserved in serogroup 35 members other than 35D. Similarly, factor serum 20b has been shown to recognize WcjE-mediated O-acetylation (20), which is present in both the serotype 35A and 35C repeat units (13; K. A. Geno, C. A. Bush, M. H. Nahm, and J. Yang, unpublished data). Binding of factor serum 35b distinguishes serotype 29, whose first repeat unit sugar is Gal, from serotype 35D, whose first repeat unit sugar is Glc. Therefore, factor serum 35b must target the Galf-Gal(β1→6) motif present in serotypes 29 and 35F. This is consistent with the observation that serotype 47F, which has the same motif (13), contains 35b in its antigenic formula (17). All serotypes presented but 35F contain Galf in (1→1) linkage with a sugar alcohol, a motif that correlates with factor serum 35c reactivity. Factor serum 29b reacts with only serotypes 29, 35B, and 35D, whose capsules contain a ribitol-(5→P→4)-N-acetyl galactosamine motif that seems a likely target for factor serum 29b. Finally, factor serum 42a differentiates serotype 35C from serotype 35A; these serotypes are differentiated only by the presence of a side chain also present in serotype 42 (13); thus, this side chain (and its associated attachment point) is the likely target of factor serum 42a.
Predicted epitope correlations among serogroup 35 capsules. Capsular structures of serotypes 35F, 35A, 35B, and 35C were previously described (13). Epitopes were deduced by comparisons within serogroup 35 and with the other known structures of pneumococcal capsules (13). *, we have determined that the serotype 35C capsule is acetylated at the indicated sites (K. A. Geno, C. A. Bush, M. H. Nahm, and J. Yang, unpublished data). Gal, galactose; Ac, acetyl; Rib-ol, ribitol; P, phosphate; f, furanose; Man-ol, mannitol; Glc, glucose; GalNAc, N-acetyl galactosamine.
Our studies identified two mutant residues of WciG in strain 3431-06 that appear to render the enzyme inactive, as wciG of 3431-06 was not able to complement a WciG deletion. Because all 17 predicted amino acid sequences for WciG among pneumococci are highly conserved, it is difficult to assess the two residues' relative importance to WciG function. WciG is a predicted integral membrane O-acetyltransferase with 10 transmembrane helices that may form a pore. Thus, we examined the predicted topology of WciG by submitting the WciG protein sequence derived from GenBank accession no. CR931705 to two different prediction programs (TMPRED [http://embnet.vital-it.ch/software/TMPRED_form.html ] and Protter [http://wlab.ethz.ch/protter/start/ ]). Both models predict a protein with 10 transmembrane helices with residue 178 as the first or second residue of outside-in transmembrane helix six and residue 227 as the last residue of inside-out transmembrane helix seven. We thus speculate that the two residues, which are located at the exterior surface of the cell on adjacent helices, disrupt pore function in the membrane.
Membrane-bound O-acetyltransferases modify a wide array of bacterial polysaccharides in both Gram-positive and Gram-negative organisms, including capsular polysaccharides (see, e.g., reference 11), O antigens (21), and peptidoglycan (22). Our discovery of serotype 35D further illustrates that genetic inactivation of membrane-bound O-acetyltransferases is a commonly used mechanism for generating variation in capsular polysaccharides. For example, wcjE-positive and -negative pairs are known to exist in many serotypes (6, 11, 23), and serotype pairs 15B/15C and 18C/18B are differentiated by O-acetyltransferase functionality (6). Here, we have presented the first described serotype arising from wciG inactivation. As 14 other serotypes encode wciG (6), the possibility of such variation within these types should be explored.
As serotype 35B is increasing in prevalence in the wake of conjugate vaccine usage in some parts of the world (3, 4, 14, 15), it is likely to be included in a future generation of PCVs. WciG-mediated O-acetylation seems to be an antigenically dominant epitope for this serogroup, as evidenced by loss of group 35 antiserum recognition in wciG-deficient isolates. Therefore, the serotype 35B polysaccharide is likely to elicit antibodies to the acetyl group. Furthermore, if vaccination elicits antibodies primarily targeting the O-acetyl group of serotype 35B, it may selectively remove serotype 35B while permitting serotype 35D to become prevalent. Thus, serotypes 35B and 35D should be distinguished during vaccine design and manufacturing, as well as for epidemiologic studies performed before and after the introduction of future vaccines.
After the submission of this paper, we became aware of a report from Australia describing four clinical isolates whose genetic and serological profiles match those of serotype 35D (24). In addition, the CDC Active Bacterial Core surveillance (ABCs) program has uncovered two serotype 35B wciG variants exhibiting the serology we describe for serotype 35D (Bernard Beall, CDC ABCs, personal communication). Thus, it appears that at least seven isolates of the novel serotype herein described have been recognized globally, and retrospective studies of serotype 35B or serotype 29 isolates may uncover additional isolates. Notably, six of the seven isolates have unique inactivations of wciG. Thus, as the authors of the Australian report noted, these serotype 35D isolates may reflect microevolution during serotype 35B invasive disease, similar to that proposed for serotype 11A microevolution to serotype 11E (25).
MATERIALS AND METHODS
Bacterial strains and cultivation.The strains used are listed in Table 1. Unless otherwise specified, strains were streaked to blood agar plates with 5% sheep's blood (Remel) from frozen glycerol stocks and incubated overnight at 37°C with 5% CO2. Fresh colonies were inoculated into Todd-Hewitt broth with 5% yeast extract (THY) and grown to mid-log density before a 1:1 mixture with fresh THY and supplementation to 16% glycerol; 500-μl aliquots were frozen at −80°C for use in the assays described below. When indicated, kanamycin was used at 100 μg/ml, and erythromycin was used at 0.3 μg/ml.
Flow cytometric serotyping assays.Flow cytometric serotyping assays were performed as previously described (10), with the modification that HBC buffer (Hanks' balanced salt solution [HBSS]-bovine serum albumin-calcium buffer; 1× HBSS without magnesium or calcium, 0.5% bovine serum albumin, 2.2 mM CaCl2 [26]) was used instead of FACS buffer. Briefly, bacterial stocks were washed in HBC, incubated with 1:10,000 dilutions of antisera, washed by centrifugation, stained with a phycoerythrin-conjugated anti-rabbit immunoglobulin secondary antibody (catalog no. 4010-09; Southern Biotech) at a 1:1,000 dilution, washed by centrifugation, and read on a BD FACSCalibur, BD FACSArray, or BD Accuri flow cytometer. Data were analyzed in FCS Express versions 3.0 and 6.0. The indicated serological reagents were obtained from Statens Serum Institut.
Genetic manipulations.Primer sequences are presented in Table 4. wciG was deleted in serotype 35B isolate 3009-06 through allelic exchange of wciG with Janus cassette (27) with selection for kanamycin resistance. Upstream and downstream flanking regions were amplified from 3009-06 chromosomal DNA using primer sets 5300/3151 and 5093/3150. Janus cassette was amplified from TIGR-JS (8) using primers 5092/3152; fragments were successively joined by overlap extension PCR. The agarose gel-purified product was mixed with an equal volume of 3009-06 lysate (prepared as in reference 28) and transformed into 3009-06, as previously described (29), with the exception that competence-stimulating peptide 2 was used; transformants were plated on THY agar (THY plus 1.5% agar) and overlaid with THY plus 0.8% agar supplemented with kanamycin to 100 μg/ml and 25 μg/ml 2-,3-,5-triphenyltetrazolium chloride to aid visualization. Plates were incubated at 37°C with 5% CO2 overnight. Agar plugs containing colonies were inoculated into THY, grown to mid-log phase, and stocked as described above. Uptake of the construct was verified by DNA sequencing from purified chromosomal DNA.
Primers used in this study
For complementation of wciG, we utilized a previously described strategy that places the gene to be complemented immediately downstream of aliA (30), which is present downstream of the cps locus. wciG from isolate 3431-06 was amplified using primers 5224/3182. For complementation of wciG in 3431-06, homology to the existing wciG gene prohibited use of the serotype 35B wciG, as it is larger than the aliA flanking region of pAG269, preferentially targeting the plasmid to wciG rather than aliA; thus, wciG from serotype 35C, whose product is predicted to perform the same acetylation but whose gene has lower nucleotide homology, was used. Primer set 5240/3196 was used to amplify wciG from the serotype 35C clinical isolate 5705-06 (GenBank accession no. KX470740 ), which was obtained from the CDC ABCs program. PCR products were cloned into pCR2.1 using the TOPO TA cloning kit (Invitrogen) and maintained in Escherichia coli TOP10 cells (Invitrogen). Sequences of cloned PCR products were verified by DNA sequencing. Correct plasmids were digested with SpeI/SphI and ligated into pAG269 prior to transformation into E. coli DB11, as previously described (30). Prepared plasmids were transformed into KAG1014 or 3009-06 as described above, substituting erythromycin (0.3 μg/ml) for kanamycin.
DNA sequencing.DNA sequencing was performed by the Heflin Center Genomics Core Lab at the University of Alabama at Birmingham.
Polysaccharide purification.Serotype 35B and 35D polysaccharides were purified from strains 3009-06 and 3431-06 by anion-exchange chromatography, as previously described (31). Polysaccharide-containing fractions were determined through an inhibition enzyme-linked immunosorbent assay (ELISA) competing soluble serotype 35B/35D polysaccharide fractions against serotype 35B polysaccharide (ATCC)-coated wells for binding of factor serum 29b. Polysaccharide-containing fractions were pooled, dialyzed against water, and lyophilized.
NMR spectroscopy.NMR data were collected at 25°C on a Bruker Avance II (700 MHz 1H) spectrometer equipped with a cryogenic triple-resonance probe. NMR samples were prepared by dissolving 2 to 5 mg of lyophilized polysaccharide in 0.5 ml of 99.99% D2O. Complete assignments of 1H and 13C signals for 35B and 35D polysaccharides were achieved by two-dimensional nuclear Overhauser spectroscopy (1H-1H NOESY), correlation spectroscopy (1H-1H COSY), total correlation spectroscopy (1H-1H TOCSY), and heteronuclear multiple quantum coherence (1H-13C HMQC). Data were processed with NMRPIPE (32) and analyzed with NMRView (33). Trimethylsilylpropanoic acid was used as an external reference.
Accession number(s).Sequence data for relevant isolates were deposited to GenBank under the accession numbers found in Table 1.
ACKNOWLEDGMENTS
This work was funded by NIH grants HL105346 (to K.A.G.) and AG050607 (to M.H.N.) and NIH contract HHSN272201200005C (to M.H.N.). The NIH did not participate in experimental design, data collection and interpretation, or the publication of these findings.
The University of Alabama at Birmingham (UAB) has intellectually property rights to some reagents developed in the laboratory of M.H.N., and all study authors are UAB employees. We declare no additional conflicts of interest.
FOOTNOTES
- Received 10 January 2017.
- Returned for modification 7 February 2017.
- Accepted 9 February 2017.
- Accepted manuscript posted online 15 February 2017.
- Copyright © 2017 American Society for Microbiology.
