DNA Polymorphism and Molecular Subtyping of the Capsular Gene Cluster of Group B Streptococcus

Abstract

Serotyping and other phenotypic methods are often used to characterize the capsular polysaccharide of group B streptococci (GBS). We describe a capsular genotyping method that utilizes PCR of capsular polysaccharide synthesis genes (cps) and restriction enzyme digestion. This method facilitates the detection of DNA polymorphism in cps genes and correlates well with serotyping.

Group B streptococci (GBS; Streptococcus agalactiae), although generally carried asymptomatically, can cause invasive disease in newborns, pregnant women, and immunocompromised adults. A crucial factor in GBS virulence is the production of an antigenically variable polysaccharide capsule, also used for strain typing. Certain serotypes of the nine known types (Ia, Ib, and II to VIII) are more prevalent in invasive disease, e.g., serotypes Ia, II, and III, and, since the early 1990s, serotype V (2, 4, 13, 16, 29).

Because of the role of capsule in GBS virulence, several phenotypic methods have been devised for serotyping, including the Lancefield capillary precipitin method (19), the predominant serotyping scheme, latex agglutination (22, 25), coagglutination (15), double immunodiffusion (17), and enzyme immunoassays (1). These, however, generally have limited accuracy and applicability, are expensive, and result in numerous (∼2 to 18%) nontypeable (NT) isolates (3, 4, 12). Genotypic methods, including PCR and sequencing of serotype-specific gene fragments within the cps genes (8, 18), DNA dot blot hybridization (5), and PCR-based restriction fragment length polymorphism (RFLP) analyses (24), utilize genetic polymorphisms in the capsular polysaccharide synthesis (cps) gene cluster to classify GBS strains into the corresponding serotypes. Genotypic methods complement phenotypic approaches and avoid problems of unreliable capsule expression, NT phenotypes, and new antigenic variants. Here, we evaluate a new PCR-based method that utilizes RFLP of the cps cluster to detect DNA polymorphisms in the cps cluster.

The cps cluster comprises genes cpsA-O, cpsR, cpsS, and cpsY (6, 8, 28), most of which are conserved across serotypes (8). cpsG-K, however, are highly variable in serotypes Ia, Ib, and II to VII, whereas cpsE to -K are variable in serotype VIII (see Fig. 2a in Cieslewicz et al. [8]). In addition, not all serotypes contain all cps genes. We devised a set of PCR primers (cpsG-F97, 5′-GTGTTCATTCAAACGGGTTACTCA-3′; cpsL-R200, 5′-GAACTTAAAGAACCGCTCGTCTG-3′) in the conserved cpsG and cpsL flanking the variable region of the cps cluster that yield amplicons ranging in size from 4,577 bp (serotype VI cps cluster) to 6,248 bp (serotype V cluster) (Table 1). For serotype VIII, we developed another forward primer targeting cpsR (cpsR8-F40, 5′-CCGGCAAGATATGGTGGAT-3′), which when used with the cpsL-R200 primer, produces a 4,551-bp PCR product (Table 1).

FIG. 2.

PCR-based RFLP capsular typing method illustrating the observed banding patterns specific for each of the nine group B streptococcal serotypes. Lanes 2 to 6, serotypes Ia, Ib, and II to IV, respectively; lanes 8 to 11, serotypes V to VIII; lanes 1 and 12, 1-kb plus ladder; lane 7, 100-bp ladder.

TABLE 1.

Expected DNA fragment sizes of a 4,551- to 6,248-bp region of the group B streptococcal cps gene cluster after PCR and digestion with DdeI

Capsule serotype	GenBank accession no.	Reference(s)	Predicted cps amplicon length (bp)	Expected no. of cut sites	Expected restriction fragment lengths (bp)a
Ia	AB028896	28	4,814	7	14, 21, 63, 192, 287, 1,100, 1,566, 1,571
Ib	AB050723	27	4,717	12	14, 21, 63, 98, 100, 192, 288, 324, 390, 682, 720, 885, 940
II	AY375362	8	4,645	11	14, 21, 54, 63, 122, 165, 192, 380, 521, 609, 936, 1,568
III	AF163833, NC_004368	14,23	4,702	9	14, 21, 63, 122, 165, 192, 521, 936, 1,100, 1,568
IV	AF355776	7	5,304	14	11, 14, 21, 43, 63, 82, 241, 245, 337, 458, 460, 693, 806, 839, 1,027
V	AF349539, NC_004116	7,>26	6,248	14	11, 12, 14, 21, 63, 167, 192, 325, 368, 484, 511, 530, 883, 1,191, 1,476
VI	AF337958	7	4,577	11	14, 21, 51, 63, 192, 211, 372, 409, 505, 521, 764, 1,454
VII	AY376403	8	5,364	10	11, 14, 21, 114, 192, 49 8, 599, 642, 882, 1,011, 1,380
VIII	AY375363	8	4,551	9	41, 163, 170, 192, 214, 224, 366, 728, 823, 1,630

^aDdeI restriction fragments of cps PCR product. Underlining denotes fragments that can be resolved on a 1.5% agarose gel.

Because of the sequence variability across serotypes within the region amplified, in silico digestion of the cps amplicons yields extensive RFLP among serotypes (Fig. 1). Based on the published cps sequences of nine GBS strains, digestion with DdeI of the predicted cps amplicons should generate 8 to 15 restriction fragments (Table 1).

FIG. 1.

Diagram representing the serotype-specific variable region within the group B streptococcal capsular polysaccharide synthesis (cps) gene cluster that is amplified by PCR. Homologous cps genes are illustrated by colors. The vertical lines and numbers represent the DdeI restriction sites and the expected restriction fragment sizes, respectively. Solid circles mark fragments affected in allelic variants.

We examined 141 GBS strains representing nine serotypes and NT strains as determined previously by the Lancefield method (19) at the National Center for Streptococcus (Edmonton, Alberta, Canada). A total of 45 strains, collected in Alberta between 1998 and 2000, were isolated from patients with invasive GBS disease (10), and 36 strains were from colonized pregnant women (9). An additional 60 strains were obtained from the American Type Culture Collection (NEM316) (14), the University of Michigan (n = 26) (5, 20, 21), Czech National Collection of Type Cultures (n = 2), Norrlands Universitetssjuk-Ius (n = 20) (24), and Channing Laboratory (n = 11) (26). Approval to obtain isolates from some pregnant women and newborns in the present study was granted by the Conjoint Medical Ethics Board of the University of Calgary. Informed consent was obtained from all study participants.

DNA was isolated by growing strains overnight in Todd-Hewitt broth incubated at 37°C in CO₂ (Mo Bio Laboratories, Inc., Solana Beach, CA). The three primers (10 μM each) were pooled, and DNA fragments were amplified via PCR using 1.5 U of LA Taq (Takara Bio, Inc., Kyoto, Japan). The thermal cycle, initiated by a 1-min soak at 94°C, was run for 30 cycles (94°C for 30 s, 58°C for 30 s, and 68°C for 6 min), followed by a 10-min soak at 72°C. The amplicons were digested with DdeI (New England Biolabs, Inc., Beverly, MA) overnight at 37°C, followed by electrophoresis on a 1.5% agarose gel and ethidium bromide staining.

The cps region was amplified in 139 (98.5%) strains representing nine serotypes, and DdeI digestions revealed 12 distinct restriction fragment patterns (Fig. 2). With the exception of serotype II strains, the observed fragment sizes estimated using PRO-SCORE (11) matched the predicted sizes. RFLP of the cps region was concordant with the serotype in 138 (97.9%) of 141 strains (Table 2); therefore, we designated the distinct restriction patterns with numbers and letters corresponding to the serotype (e.g., cps1a for serotype Ia).

TABLE 2.

Capsule genotypes based on RFLP of the cps gene cluster

Capsule	No. of strains	Capsule genotype (no. of strainsa)
Ia	8	cps1a (8)
Ib	7	cps1b (7)
II	13	cps2 (13)
III	38	cps3 (36), cps1a (1), cps5.1 (1)
IV	16	cps4.1 (9), cps4.2 (6), cps6 (1)
V	11	cps5.1 (11)
VI	9	cps6 (9)
VII	8	cps7.1 (6), cps7.2 (2)
VIII	10	cps8 (10)
NT	21b	cps1a (2), cps1b (4), cps5.1 (9), cps5.2 (2), cps6 (1), cps8 (1)

^aA complete strain list with capsule genotypes is provided in the supplemental material.

^bThe capsular genotype could not be determined for two NT strains.

Two strains originally designated as serotype III matched the cps1a and cps5 genotypes. Blinded re-serotyping using hyperimmune rabbit antisera to the GBS polysaccharide types (21) was performed by Carol Baker. Re-serotyping confirmed the cps5 genotype in one strain, suggesting an error in the original serotype. The cps1a strain was re-serotyped as NT, a result that may be a consequence of overexpression of cpsH (6).

All 13 serotype II strains produced the identical cps2 pattern, which differed substantially from the predicted DdeI digestion of the serotype II cps sequence in GenBank (AY375362 [8]). The basis for this discrepancy is problematic; however, we suspect that the GenBank strain does not represent the primary serotype II GBS clone in circulation.

Among the 21 NT strains examined, 19 could be assigned a cps genotype after PCR amplification and DdeI digestion (Table 2). In two strains, however, no amplicons of the cps region were produced. Further evaluation of these strains with additional primer sets, including primers in cpsE and neuB paired with the reversed cpsG and cpsL primers and previously described (24) cps primers (lorfXfo and loEFrev), yielded no specific PCR products. These observations, along with reports of rare GBS strains that cannot be typed by other methods (5, 24), suggest that there are clones in the GBS population that have diverged in the conserved regions of cps locus.

One advantage of the RFLP-based method is its ability to identify cps genotypic variants: we recovered cps variants for serotypes IV, V, and VII but not for serotype III as described previously (24) (Fig. 3). Presumably, the polymorphism underlying the type III variant described by Sellin et al. (24) lies outside the region examined by our method. For serotypes IV and V, the observed restriction patterns were similar to the predicted patterns (Fig. 3). The cps4.2 (serotype IV, allele 2) variant can be explained by an additional restriction site in the 839-bp fragment of cpsK (Fig. 1), whereas cps5.2 (serotype V, allele 2) appears to have an ∼700-bp DNA insertion with additional restriction sites between cpsO and cpsJ (Fig. 1). The serotype VII variant pattern (cps7.2), originally identified by Sellin et al. (24), differs substantially from the predicted pattern (Fig. 3); the structure of this cps variant warrants further analysis.

FIG. 3.

Restriction fragment banding patterns of cps allelic variants. (A) Lane 2, cps4.1; lane 3, cps4.2; lane 4, cps5.1; lane 5, cps5.2. (B) Lane 2, cps6; lane 3, cps5.1; lane 4, cps7.1; lane 5, cps7.2. In both panels, lanes 1 and 6 show the 1-kb plus ladder and 100-bp ladder, respectively. Capsule serotypes are indicated above in Roman numerals.

The amplification and RFLP analysis of the cps gene cluster accurately reveals the genetic variation underlying the capsular variation of GBS strains. Genotyping the cps locus as described is inexpensive and easy to perform, and can be used to classify NT isolates and identify cps allelic variants. In addition, it can identify novel, emergent, capsular types based on restriction pattern differences.