The Serotype of Type Ia and III Group B Streptococci Is Determined by the Polymerase Gene within the Polycistronic Capsule Operon

Abstract

Streptococcus agalactiae is a primary cause of neonatal morbidity and mortality. Essential to the virulence of this pathogen is the production of a type-specific capsular polysaccharide (CPS) that enables the bacteria to evade host immune defenses. The identification, cloning, sequencing, and functional characterization of seven genes involved in type III capsule production have been previously reported. Here, we describe the cloning and sequencing of nine additional adjacent genes, cps_IIIFGHIJKL, neu_IIIB, and neu_IIIC. Sequence comparisons suggested that these genes are involved in sialic acid synthesis, pentasaccharide repeating unit formation, and oligosaccharide transport and polymerization. The type III CPS (cpsIII) locus was comprised of 16 genes within 15.5 kb of contiguous chromosomal DNA. Primer extension analysis and investigation of mRNA from mutants with polar insertions in their cpsIII loci supported the hypothesis that the operon is transcribed as a single polycistronic message. The translated cpsIII sequences were compared to those of the S. agalactiae cpsIa locus, and the primary difference between the operons was found to reside in cps_IIIH, the putative CPS polymerase gene. Expression of cps_IIIH in a type Ia strain resulted in suppression of CPS Ia synthesis and in production of a CPS which reacted with type III-specific polyclonal antibody. Likewise, expression of the putative type Ia polymerase gene in a type III strain reduced synthesis of type III CPS with production of a type Ia immunoreactive capsule. Based on the similar structures of the oligosaccharide repeating units of the type Ia and III capsules, our observations demonstrated that cps_IaH and cps_IIIH encoded the type Ia and III CPS polymerases, respectively. Additionally, these findings suggested that a single gene can confer serotype specificity in organisms that produce complex polysaccharides.

Group B streptococci (GBS) (Streptococcus agalactiae) are the leading cause of serious bacterial infections (bacteremia, pneumonia, and meningitis) in newborns, causing two to three cases per 1,000 live births (47). An indispensable GBS virulence determinant is the production of a type-specific capsular polysaccharide (CPS), which prevents the deposition of host complement factor C3b and inhibits opsonophagocytosis (45). Nine distinct capsular serotypes, Ia, Ib, and II to VIII, have been identified (54), and their chemical compositions and structures have been determined (16–19, 51, 55, 56, 58). Type Ia, Ib, II to V, and VII CPS consist of the monosaccharides glucose, galactose, N-acetylglucosamine, and N-acetylneuraminic acid. Serotypes VI and VIII lack N-acetylglucosamine, and type VIII contains rhamnose (19). Although serotypes Ia, III, and V are currently the most common isolates from the United States associated with early-onset disease (within 1 week of birth), comprising 82% of isolates (27), type III GBS are the most prevalent isolates associated with neonatal disease (5).

We previously identified a region of the GBS chromosome encoding genes involved in type III capsule production (the cpsIII locus) by screening genomic transposon libraries for CPS mutants (45, 59). DNA sequence analysis of the acapsular type III mutants led to the identification of four genes designated cpsA, cpsB, cpsC, and cpsD (43). To conform to the emerging consensus nomenclature in the CPS literature, these genes have been designated cps_IIIB, cps_IIIC, cps_IIID, and cps_IIIE, respectively. Cps_IIIE was assigned a function as a galactosyltransferase, but due to potential endogenous C-4 epimerase activity, the possibility that Cps_IIIE may act as a glucosyltransferase could not be ruled out (43). Mutants which produced a capsule lacking sialic acid were also identified (25, 57, 59). The asialo mutants had a common transposon insertion site in a gene approximately 9 kb downstream of cps_IIIE, which was designated cpsF (now referred to as neu_IIIA). The neu_IIIA gene was shown to encode a CMP-N-acetylneuraminic acid synthetase (15, 46) which complemented an Escherichia coli K1 neuA mutant, restoring synthesis of the polysialic acid capsule (15). A gene homologous to E. coli K1 neuD (formerly designated cpsE and now called neu_IIID) was found adjacent to the 5′ end of neu_IIIA (8). Additionally, upstream of cps_IIIB, two genes similar to regulatory proteins, designated cpsY and cpsX (referred to here as cps_IIIA to conform to the usage for the S. agalactiae type Ia homologue), have been identified, although their functions have not been confirmed (24). Thus, seven genes of the cpsIII locus have previously been described, along with the divergently transcribed cpsY, whose role in CPS synthesis is not known.

The type III CPS produced under the direction of the cpsIII operon possesses a backbone of repeating [→6)-β-d-N-acetylglucosamine-(1→3)-β-d-galactose-(1→4)-β-d-glucose-(1→]_n trisaccharide units. Each repeating unit carries a disaccharide side chain of α-d-N-acetylneuraminic acid-(2→3)-β-d-galactose, 1→4 linked via the galactose to the backbone N-acetylglucosamine. The type Ia CPS is composed of a polymer of β-1,4-linked lactose to which a trisaccharide side chain of α-d-N-acetylneuraminic acid-(2→3)-β-d-galactose-(1→4)-β-d-N-acetylglucosamine is attached to each repeating unit galactose by a β-d-N-acetylglucosaminyl-(1→3) linkage. Traditionally, the GBS CPS repeating unit structures have been depicted to emphasize these differences and to reflect their conformational structure in the native polysaccharide configuration (16, 17). However, if the repeating unit structures are redrawn as in Fig. 1, it can be seen that the type Ia and type III CPS oligosaccharide structures differ only by the linkage between repeating units. Therefore, the essential difference between the two capsules is due to the glycosidic bond formed during CPS polymerization. The structural similarity of these capsule oligosaccharides has led us to hypothesize that there is a high degree of genetic relatedness between their capsule synthesis loci. This hypothesis is supported by two further observations: first, the CPS structures of S. pneumoniae type 14 and type III GBS are identical except for the lack of the terminal sialic acid side chain residue of the former (Fig. 1); second, the cps loci of S. pneumoniae type 14 and GBS type Ia are highly homologous within their regulatory, repeat unit, and oligosaccharide transport genes (60).

FIG. 1.

GBS type Ia and III CPS repeating unit structures. Ia, S. agalactiae type Ia CPS subunit structure; III, S. agalactiae type III CPS subunit structure; Gal, galactose; Glc, glucose; GlcNAc, N-acetylglucosamine; NeuNAc, N-acetylneuraminic (sialic) acid. The critical linkages differentiating the type Ia and III CPS are shown in the shaded boxes.

In this study, we have completed the molecular characterization of nine additional genes that make up the remainder of the cpsIII operon. We present the organization of the 15.5 kb of DNA containing the 16 genes involved in CPS III synthesis, describe its similarity to other complex polysaccharide synthesis loci, and demonstrate its expression as a polycistronic operon. Last, we provide evidence that a single polymerase gene determines the unique bond formed during polymerization of the type Ia or type III repeating units and thus confers CPS serotype specificity.

MATERIALS AND METHODS

Bacterial strains and plasmids.

Bacterial strains and plasmids used for this study are listed in Table 1.

TABLE 1.

Bacterial strains and plasmids used

Strain or plasmid	Genotype/phenotypea	Reference or source
S. agalactiae
A909	Wild type/cpsIa	29
COH1	Wild type/cpsIII, Tetr	32
COH1-13	COH1, Tn916ΔE cpsE/CPS− Tetr Emr	43
COHY-104	COH1, cpsIIIB, ΩKm-2 via pHY108/CPS− Tetr Kanr Cms	This study
COHY-105	COH1, cpsIIIC, ΩKm-2 via pHY113/CPS− Tetr Kanr Cms	This study
COHY-106	COH1, cpsIIID, ΩKm-2 via pHY109/CPS− Tetr Kanr Cms	This study
COHY-107	COH1, cpsIIIA, ΩKm-2 via pHY120/CPS− Tetr Kanr Cms	This study
E. coli
DH5α	F−endA1 (rK− mK+) supE44 thi-1 hsdR17 gyrA (Nalr) recA1 Δ(lacZYA-argF)U169 deoR relA1 [φ80Δlac(lacZ)M15]	BRL, Inc.
MC1061	F−araD139 Δ(ara-leu)7696 Δ(lac)X74 galU galK hadR2 (rK− mK+) mcrB1 rpsL (Strr)	53
Plasmids
pBS	Cloning vector, Apr 2.8 kb	Stratagene
pBS9.0	pBS, cpsIIIIJKLneuBCDA/Apr 11.8 kb	This study
pCER107	pBS, cpsIIIABCDE/Apr 5.7 kb	25
pCER108	pBS, cpsIIIEFGHI/Apr 6.1 kb	25
pDC128	Cmr, pDC123 with a 7,122-bp XbaI fragment containing cpsIIIKLneuBCDA; 11.6 kb	This study
pCIV2	Kanr/ΩKm-2 5.9 kb	38
pHY101	Cmr, 597 bp ‘cpsIIIB’ fragment cloned from pCER107 via ApaI/HindIII ligation into pVE6007	This study
pHY104	Cmr, 412-bp ‘cpsIIID’ fragment cloned from pHY204 via KpnI/HindIII ligation into pVE6007	This study
pHY105	Cmr, 537-bp ‘cpsIIIC’ fragment cloned from pHY205 via KpnI/HindIII ligation into pVE6007	This study
pHY108	pHY101 with ΩKm-2 from pCIV2 cloned into FspI site of ‘cpsIIIB’/Kanr	This study
pHY109	pHY104 with ΩKm-2 from pCIV2 cloned into BglII site of ‘cpsIIID’/Kanr	This study
pHY113	pHY105 with ΩKm-2 from pCIV2 cloned into BglII site of ‘cpsIIIC’/Kanr	This study
pHY115	Cmr, 648-bp ‘cpsIIIA’ fragment cloned into pVE6007 from pHY208 via KpnI/HindIII ligation	This study
pHY120	pHY115 with ΩKm-2 from pCIV2 cloned into NsiI site of cpsIIIA using a BamHI-NsiI adapter/Kanr	This study
pHY204	Apr, 345-bp ‘cpsIIID’ amplicon cloned into pT7Blue vector	This study
pHY205	Apr, 470-bp ‘cpsIIIC’ amplicon cloned into pT7Blue vector	This study
pHY208	Apr, 583-bp ‘cpsIIIA’ amplicon cloned into pT7Blue vector	This study
pSH101	Apr, 541-bp ‘neuIIIA’ amplicon cloned into pT7Blue vector	This study
pSH102	Cmr, 525-bp ‘neuIIIA’ cloned from pSH101 via ApaI/EcoRV ligation into pVE6007	This study
pSH104	Cmr, plasmid rescue construct containing pSH102 and sequences downstream of neuIIIA, 7.3 kb	This study
pT7Blue	Apr 2.9-kb T-overhang PCR cloning vector	Novagen
pVE6007	Cmr temperature-sensitive shuttle vector, 3.4 kb, ori[Ts]	30
pDC123	Cmr, phoZ, gram-positive/negative blue/white screening expression vector	8
pDC123(GHIa)	Cmr, 2,350-bp PCR product from type Ia GBS strain A909 containing cpsIaGH cloned into AflII/ApaI-cut pDC123	This study
pDC123(HIa)	Cmr, 1,767-bp XbaI-cut PCR product containing cpsIaH from type Ia GBS strain A909 cloned into SmaI/XbaI-cut pDC123	This study
pDC123(GHIII)	Cmr, 2,244-bp PCR product containing cpsIIIGH from type III GBS strain COH1 cloned into AflII/EcoRV-cut pDC123	This study
pDC123(HIII)	Cmr, 209-bp BclI deletion of pDC123(GHIII)	This study

^aTet^r, tetracycline resistance; Em^r, erythromycin resistance; Ap^r, ampicillin resistance; Cm^r, chloramphenicol resistance; Cm^s, chloramphenicol susceptible; Kan^r, kanamycin resistance; Nal^r, nalidixic acid resistance; CPS⁻, capsule negative. 

Nucleotide sequencing and DNA analysis.

Sequencing was performed on plasmid subclones containing COH1 genomic DNA (25), purified PCR products, and a subclone generated by plasmid insertion/rescue (Fig. 2). Fluorescent dye terminator sequencing reactions were performed using a Ready Reaction sequencing kit (Perkin-Elmer Cetus, Foster City, Calif.) according to the manufacturer's specifications and analyzed on an Applied Biosystems 373A automated sequencer. Both strands of each DNA template were sequenced.

FIG. 2.

Organizational map of the type III GBS cps operon. ORFs within the operon, with the direction of transcription, are indicated by the open arrows. The corresponding gene designation is shown above each ORF. Restriction site designations: E, EcoRI; H, HindIII; S, SacI; X, XbaI. Plasmids used in sequencing the locus are depicted above the operon map. PCR products used to derive additional sequence are shown as solid bars above the operon map. The cpsIII transcriptional start site is indicated with an arrow in the cpsY-cpsA intergenic region. The ΩKm-2 insertion sites are indicated with triangles: filled triangle, cps_IIIA insertion; open triangle, cps_IIIB insertion; striped triangle, cps_IIIC insertion; dotted triangle, cps_IIID insertion. Sequences used in generating RNA probes are marked with open boxes. The location and orientation of the oligonucleotide used to initiate the primer extension reaction (primer 5a) is also shown.

cps_IIID-cps_IIIE and IS861-cps_IIIB intergenic regions were amplified from the COH1 chromosome by PCR. Primers used for amplifying the IS861-cps_IIIB intergenic region hybridized to a region downstream of IS861 (5′-GAAAGGTTTGCTTTGTCGTGTCGGATATAG-3′) (44) and to the 5′ end of cps_IIIB (5′-CTTCAACGCTTTTGGGCCCATCATCTACATC-3′) (43). The second PCR product was an amplified region of cps_IIID-cps_IIIE that was produced by using primers 5′-GAAGTAAGGGACTCTGGTATTGA-3′, which hybridizes to the 3′ end of cps_IIID, and 5′-GAGCAAACCTATAATAGCACCCGT-3′, which hybridizes to cps_IIIE. The PCR products were prepared for sequencing by purification using a QiaQuick PCR purification kit (Qiagen Inc., Valencia, Calif.).

DNA sequence analysis and FASTA homology searches were performed using the Genetics Computing Group (University of Wisconsin) software on a UNIX computer at the University of Washington. ENTREZ and BLAST homology searches were performed using the National Center for Biotechnology Information Internet server.

Plasmid insertion/rescue.

To clone and sequence chromosomal DNA 3′ of neu_IIIA, an intragenic ‘neu_IIIA’ fragment was amplified and cloned into the pT7Blue vector (Novagen Inc., Madison, Wis.) to create pSH101. An ApaI/EcoRV fragment containing the ‘neu_IIIA’ sequence was isolated from pSH101 and ligated into the temperature-sensitive vector pVE6007 (30) previously digested with the same restriction enzymes. This new plasmid was designated pSH102 and transformed into competent COH1 as described elsewhere (13). Transformants were screened for plasmid integration within the chromosomal wild-type (wt) neu_IIIA as described elsewhere (61). Subsequent digestion of integrant chromosomal DNA with SacI, followed by intramolecular ligation, rescued the plasmid and flanking chromosomal sequences. The ligated DNA was transformed into E. coli DH5α, and chloramphenicol-resistant clones were isolated by growth at 30°C on L agar containing 10 μg of chloramphenicol per ml.

Generation of cps gene allelic exchange plasmids.

Intragenic fragments of cps_IIIA, cps_IIIC, and cps_IIID were amplified from COH1 chromosomal DNA by PCR and directly ligated into the vector pT7Blue. The intragenic fragments were excised from pT7Blue with KpnI and HindIII and ligated into KpnI- and HindIII-digested pVE6007. To clone the intragenic fragment of cps_IIIB, pCER107 (43) was digested with ApaI and HindIII, removing 47 nucleotides (nt) from the 5′ end and 425 nt from the 3′ end of cps_IIIB. This 616-bp intragenic fragment was separated by agarose gel electrophoresis, extracted (QiaEx gel purification kit; Qiagen), and ligated into ApaI- and HindIII-digested pVE6007.

The ΩKm-2 fragment (38) from pCIV2 was subsequently cloned into sites within each of the intragenic cps gene fragments in the pVE6007-based vectors described above. For the cps_IIIC and cps_IIID constructs, the ΩKm-2 fragment was digested from pCIV2 with BamHI and cloned directly into the BglII site within each gene fragment. To insert ΩKm-2 within the cps_IIIB construct, an FspI site in cps_IIIB was ligated to the ΩKm-2 cassette previously prepared by BamHI digestion and end filled with Klenow fragment. The ΩKm-2 fragment was ligated into the NsiI site in the cps_IIIA construct via a BamHI-NsiI adapter, 5′-GATCAGCGGCCGCTTGCA-3′. Plasmid constructs containing ΩKm-2 cassettes in the same orientation relative to the chromosomal cps genes were chosen for deriving the allelic exchange mutants in COH1. After transformation of the allelic exchange vectors into GBS, the cells were grown under conditions promoting homologous recombination as described previously (61). Recombination occurring on both sides of the ΩKm-2 cassette resulted in a double-crossover recombination event and an allelic exchange mutation within the recipient gene.

PCR amplification.

PCRs were performed using standard conditions (4) and Taq polymerase (Promega Corp., Madison, Wis.). To amplify larger (>2.0-kb) PCR products, Taq extender (5 U; Stratagene Cloning Systems, La Jolla, Calif.) was added to a standard 100-μl reaction according to the manufacturer's instructions.

RNA analysis.

Total bacterial cell RNA was extracted from GBS using a rapid cell disruption method with a dental amalgamator and glass beads as described previously (62). Dot blots were generated as follows. A 2.5-μg aliquot of total cellular RNA was applied to Magnagraph nylon membrane (Micron Separations Inc., Westborough, Mass.) using a Minifold I dot blot vacuum apparatus (Schleicher & Schuell, Keene, N.H.) according to the manufacturer's recommendations and UV cross-linked. Blots were hybridized using standard 50% formamide buffers (31) and ³²P-labeled RNA gene probes. After hybridization, the blot was washed twice for 15 min with 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–0.1% sodium dodecyl sulfate (SDS) at room temperature and twice with 0.1× SSC–0.1% SDS at 65°C, excess buffer was removed, and the filter was subjected to autoradiography.

Primer extension reactions were used to identify the 5′ end of cps-specific mRNA. Primer 5a (5′-CACATACACCTCTACGG-3′) was end labeled using [γ-³²P]ATP and T4 polynucleotide kinase, hybridized to 50 μg of total cellular RNA, and extended, using displayTHERMO-RT RNA polymerase (Display Systems Biotech, Inc., Vista, Calif.). The sample was ethanol precipitated, resuspended in 8 μl of RNase A (200 μg/ml), and incubated for 20 min at 23°C. A reference standard sequencing ladder was generated using [γ-³²P]ATP-labeled primer 5a with a cycleSEQ manual cycle sequencing kit (Display Systems Biotech) and loaded next to the primer extension reaction on a 7% acrylamide–8 M urea–1× Tris-borate-EDTA gel. The samples were separated by electrophoresis at 45 W for 2.5 h; the gel was transferred to Whatman 3MM filter paper, dried, and exposed to X-ray film.

Generation of cps type Ia and III gene expression constructs.

Plasmid constructs were generated to test the effect on capsule production of heterologous cps type Ia and III gene expression. The cps_IaGH and cps_IIIGH genes were PCR amplified from strain A909 and COH1 chromosomal DNA, respectively, using primer Pol1f, which contained a 5′ ApaI adapter (5′-CCCGGGCCCAGATGTTATCATATCA-3′), and Pol1r, containing a 5′ AflII adapter (5′-CCAGATCTTAAGTTTCGTCTTTTCTTC-3′) (adapters are underlined). The 2.35-kb cps_IaGH amplicon was digested with ApaI and AflII and ligated to the ApaI/AflII-cut pDC123 expression vector, forming pDC123(GH_Ia). The 2.24-kb cps_IIIGH amplicon was digested with AflII and ligated with EcoRV/AflII-cut pDC123, forming pDC123(GH_III). The cps_IaH gene was amplified from the GBS A909 chromosome using primers Pol2f (5′-CTGAGATTGTTATCACAC-3′) and Pol1r. The cps_IaH amplicon (1.95 kb) was digested with XbaI and cloned into SmaI/XbaI-cut pDC123, forming pDC123(H_Ia). Constructs for the expression of cps_IIIH were derived from pDC123(HI_III) by excision of a 0.21-kb BclI fragment intragenic to cps_IIIG, forming pDC123(I_III). The constructs were transferred to the appropriate GBS hosts as described (13), and their identity was subsequently confirmed by PCR and restriction analyses.

Serotype determination of cell-associated CPS.

A modification of the colony immunoblot method described by Rubens et al. (45) was used to identify the type of CPS produced on recombinant strains compared to the wt strains. Aliquots (3 μl) of stationary-phase cultures were spotted onto a NitroPlus membrane (Micron Separations), fixed for 5 min with 70% ethanol, air dried, and blocked with BLOTTO (5% nonfat dry milk [NFDM] in phosphate-buffered saline [PBS]) for 30 min. The blots were washed three times for 5 min each with PBS and then incubated for 1 h with either type Ia or III rabbit anti-CPS antiserum (kindly provided by Michael Wessels, Channing Laboratory, Boston, Mass.) diluted 1:20,000 (type Ia) or 1:30,000 (type III) in BLOTTO. The blots were washed three times (5 min each in PBS) and then incubated for 1 h with horseradish peroxidase (HRP)-conjugated goat anti-rabbit antibody, diluted 1:20,000 in BLOTTO (Sigma, St. Louis, Mo.). After three 5-min PBS washes, 1 ml of SuperSignal chemiluminescent substrate for HRP (Pierce, Rockford, Ill.) was added to the surface of the filter according to the manufacturer's directions. Chemiluminescence was detected by exposing the blot to X-ray film (Kodak, Rochester, N.Y.).

Quantification of cell-associated CPS.

Mutanolysin digests of bacterial cells for extraction of CPS were carried out as described by Paoletti et al. (39), with minor modifications as follows. GBS were grown overnight in 10 ml of Todd-Hewitt broth, with antibiotic selection if necessary, harvested by centrifugation at 10,000 × g for 10 min at 4°C, and washed twice with 1.5 ml of 50 mM sodium phosphate buffer (pH 7.0). The cell pellets were resuspended in a final volume of 0.6 ml of 50 mM phosphate buffer (pH 7.0); 200 μl of the cell solution was transferred in duplicate to fresh microcentrifuge tubes; the cells were pelleted at 12,000 × g for 10 min at 4°C and then resuspended in 750 μl of 40% sucrose (wt/vol) in 50 mM sodium phosphate buffer (pH 7.0). To each tube, 250 μl of mutanolysin (1 mg/ml in 50 mM sodium phosphate buffer [pH 7.0]) was added. The samples were incubated at 37°C with end-over-end mixing for 1 h. Protoplasts and cell debris were removed by centrifugation at 13,000 × g for 4 min at 4°C, and the supernatants were transferred to fresh tubes and stored at −20°C.

Plates for quantitative enzyme-linked immunosorbent assay (ELISA) were prepared by coupling poly-l-lysine to purified type Ia or III CPS (kindly provided by Lawrence Paoletti, Channing Laboratory, Brigham and Women's Hospital, Harvard Medical School, Boston, Mass.), which was used to coat 96-well microtiter plates as described elsewhere (45). Coating solution (CPS [1.0 μg/ml] coupled to poly-l-lysine in 40 mM sodium phosphate buffer [pH 7.0]; 10 ml/plate) was prepared. To each well of the microtiter plate, 100 μl of the coating solution was dispensed, and the plates were incubated at 37°C for 1 h. The coating solution was discarded, and the plates were washed three times with 40 mM sodium phosphate (pH 7.0)–0.05% Tween 20 (PBT; 200 μl/well) and then blocked with 0.5% (wt/vol) NFDM in 40 mM sodium phosphate buffer (pH 7.0). The blocking solution was decanted, the wells were washed three times with 200 μl of PBT; and the plates were stored with 200 μl of PBT per well for up to 48 h before use at 4°C.

The amount of cell-associated, serotype-specific CPS produced by each strain was determined by competitive inhibition ELISA assay as described elsewhere (39). Assays were performed with duplicate plates in at least three independent experiments for each strain and serotype combination. CPS standards were prepared by dissolving purified type Ia or III CPS in H₂O to a final concentration of 10 ng/μl. Then 160 μl of CPS standard or CPS mutanolysin extract was transferred to the first column of the plate, and serial fourfold dilutions were performed across the plate from left to right into 40 mM sodium phosphate (pH 7.0). Rabbit anti-CPS antiserum (diluted 1:10,000 in 5% NFDM–50 mM sodium phosphate [pH 7.0]) was dispensed into the plate (120 μl/well), which was then allowed to stand at room temperature for 30 min, allowing the antibody to bind. From each well of the dilution series plate, 200 μl was transferred to the corresponding well (i.e., A1 to A1) of a CPS–poly-l-lysine-coated microtiter plate. One well of the plate, chosen as a blank, received 5% NFDM–50 mM Na phosphate (pH 7.0) instead of an antibody-CPS dilution. The ELISA was then developed as described elsewhere (39). The ELISA data were fitted to the four-parameter model of Rodbard (42), and the concentration of type-specific CPS in the undiluted extracts was calculated (50% inhibitory concentration of standard/50% inhibitory concentration of unknown) × concentration of undiluted standard).

Nucleotide sequence accession number.

The cpsIII DNA sequence has been deposited in GenBank with accession no. AF163833.

RESULTS

Organization of the type III GBS cps locus.

We have completed the sequencing of the entire GBS type III capsule synthesis locus by subcloning or amplifying the genomic DNA 5′ and 3′ of the cps genes identified previously (Fig. 2). Analysis of the compiled sequence revealed a total of 16 cps-related open reading frames (ORFs), of which 9 were previously undescribed. Additionally, a divergently transcribed monocistronic ORF 5′ of the type III locus containing cps_IIIY was observed. To facilitate comparison with the type Ia S. agalactiae and type 14 S. pneumoniae cps loci, we have changed the gene designations for cpsABCDEF, and thus the gene order for the CPS synthesis locus is cps_IIIABCDEFGHIJKL neu_IIIBCDA (Fig. 2). All of the ORFs except cps_IIIY are preceded by recognizable ribosome binding sites. Most of the ORFs are closely linked, with stop codons that overlap the translational initiation sites of the adjacent ORFs. There are significant gaps, however, between cps_IIID and cps_IIIE (51 bp), cps_IIII and cps_IIIJ (33 bp), cps_IIIJ and cps_IIIK (84 bp), and neu_IIIB and neu_IIIC (75 bp). We have examined these intergenic gaps and found that they do not contain potential regions of dyad symmetry, transcriptional terminators, or consensus promoter sequences. If we allow the ORFs to initiate with alternative start codons such as CUG or UUG, the intragenic gaps between cps_IIID and cps_IIIE, cps_IIII and cps_IIIJ, and neu_IIID and neu_IIIA decrease in size from 51 to 7, 33 to 0, and 69 to 21 bp, respectively. The alternative start codons for these putative ORFs also contain upstream ribosome binding sites.

The cpsIII locus is nearly identical to the cpsIa operon of S. agalactiae type Ia (Fig. 3 and Table 2). Identity between translated ORFs ranges between 94 and 100% with the exception of Cps_IIIG and Cps_IIIH, where it falls to 72 and 24%, respectively. A homologue for Cps_IIIY was not reported for type Ia S. agalactiae. The arrangement of genes from cps_IIIA to cps_IIIL of the GBS type III cps operon is also strikingly similar to that of the entire capsule locus from S. pneumoniae type 14 (Fig. 3). The nucleotide sequence similarity over the length of the common genes is 52% (7,784 identical nt out of 15,133 nt, with 215 gaps) when the type 14 capsule gene sequence is aligned with the GBS type III locus. The resemblance between the genes when translated is also high at approximately 50% identity and 60% similarity as anticipated, based on their identical capsular structures if the terminal sialic acid residue present on the type III CPS is excluded.

FIG. 3.

Schematic representation of the organization of the CPS synthesis loci of S. agalactiae type III (GBS type III), S. agalactiae type Ia (GBS type Ia) (60), and S. pneumoniae type 14 (Spn 14) (23). The unfilled arrows indicate ORFs within the operons and the direction of transcription. Filled arrows indicate ORFs not involved in CPS synthesis that flank the cps gene regions. Letters below the arrows indicate the gene designations. The organization of the cps loci into regions of associated gene function is indicated above the ORF diagrams.

TABLE 2.

Sequence comparisons to the NCBI database^a

Gene product	Proposed function	Similar gene products	% Identity	Range (amino acid)	Accession no.
CpsIIIY	Transcriptional regulatorb	Azorhizobium caulinodans, NACc	42	1–80	AJ006238
		Klebsiella pneumoniae, NACc	44	1–74	L01114
		Comamonas testosteroni, TsaRf	37	1–80	U32622
		H. influenzae Rd KW20, OxyRc	27	1–205	U49355
		E. coli, NACc	43	1–74	U56736
CpsIIIA	Transcriptional regulatorb	S. agalactiae type Ia, CpsIaAf	95	1–486	AB028896
		S. thermophilus Sfi6, EpsAf	47	1–486	U40830
		S. thermophilus, EpsAf	52	78–486	AF053346
		S. pneumoniae, Cps14fAf	42	5–485	X85787
		S. pneumoniae, Cps23fAf	42	5–485	AF030373
		S. pneumoniae, Cps33fAf	42	5–485	AJ006986
CpsIIIB	Tyrosine phosphataseb	S. agalactiae type Ia, CpsIaBg	99	1–243	AB028896
		S. thermophilus, EpsBg	65	1–243	AF053346
		S. thermophilus Sfi6, EpsBg	64	1–243	U40830
		S. salivarius, CpsBg	64	1–243	X94980
		S. pneumoniae, Cps19fBc	60	1–243	AF030367
CpsIIIC	Chain length regulatorb	S. agalactiae type Ia, CpsIaCf	99	1–230	AB028896
		S. thermophilus MR-1C, EpsCf	50	9–230	AF053348
		S. salivarius, CpsCf	50	1–230	X94980
		S. thermophilus Sfi6, EpsCf	49	1–230	U40830
		S. pneumoniae, Cap33fCf	45	1–230	AJ006986
		S. pneumoniae, Cps23fCf	44	9–230	AF030373
CpsIIID	Tyrosine kinase, CPS chain length regulator/exporterb	S. agalactiae type Ia, CpsIaCf	99	1–229	AB028896
		S. salivarius, CpsDf	57	1–222	X94980
		S. thermophilus Sfi6, EpsDf	56	1–222	U40830
		S. pneumoniae, Cps23fDf	51	1–229	AF030373
		S. pneumoniae, Cps19fDc	53	1–224	U09239
CpsIIIE	Glc-1-P transferasec	S. agalactiae Ia, CpsIaEc	99	1–449	AB028896
		S. salivarius, CpsEf	47	1–432	X94980
		S. pneumoniae, Cap33fEf	48	33–432	AJ006986
		S. pneumoniae, Cps19fEf	44	1–432	U09239
		S. pneumoniae, Cps14fEc	44	1–432	X85787
CpsIIIF	β-1,4-Gal transferase enhancerb	S. agalactiae Ia, CpsIaFg	96	1–149	AB028896
		S. pneumoniae, Cps14fFf	83	1–149	X85787
		Lactococcus lactis NIZO B40, EpsEc	38	2–149	U93364
		Sphingomonas strain s88, SpsKc	32	1–142	U51197
		E. coli, N-GlcAc	30	3–146	AF013583
CpsIIIG	β-1,4-Gal transferased	S. agalactiae Ia, CpsIaGc	72	1–147	AB028896
		S. pneumoniae, Cps14fGc	63	1–126	X85787
		L. lactis NIZO B40, EpsFc	32	1–157	U93364
		Rhizobium leguminosarum, PssEc	34	1–111	X99850
		R. leguminosarum, PssEc	34	1–111	AF014054
CpsIIIH	CPS polymerasee	S. pneumoniae, Cps14fHf	25	1–375	X85787
		S. agalactiae Ia, CpsIaHf	24	137–363	AB028896
		S. suis Cps1Hf	23	6–361	AF155804
CpsIIII	β-1,3-GlcNAc transferased	S. agalactiae Ia, CpsIaIc	98	5–320	AB028896
		S. pneumoniae, Cps23fEf	53	1–276	AF030373
		S. thermophilus Sfi6, EpsIf	35	1–255	U40830
		S. pneumoniae, Cap33ff	36	5–237	AJ006986
		S. pneumoniae, Cps14fEc	27	1–290	X85787
CpsIIIJ	β-1,4-Gal transferased	S. agalactiae Ia, CpsIaJc	99	1–315	AB028896
		S. pneumoniae, Cps14fJc	38	7–224	X85787
		S. thermophilus Sfi6, EpsIf	37	7–242	U40830
		S. pneumoniae, Cps23fEf	37	9–237	AF030373
		S. pneumoniae, Cap33ff	27	9–247	AJ006986
CpsIIIK	Sialyltransferaseb	S. agalactiae Ia, CpsIaKg	99	1–460	AB028896
		H. ducreyi Lstc	27	143–299	AF101047
		H. influenzae Rd KW20, OrfYg	26	102–298	HI0871
CpsIIIL	Repeating unit transporterb	S. agalactiae Ia, CpsIaL	100	1–460	AB028896
		H. influenzae Rd KW20g	26	102–298	M94855
		Staphylococcus aureus, Cap1Ff	24	183–400	U10927
		Streptococcus pneumoniae, Cps23ff	19	4–456	AF030373
NeuIIIB	NeuNAc synthetasec	S. agalactiae Ia, NeuBf	99	1–341	AB028896
		E. coli, NeuBc	56	2–337	U05248
		Methanococcus jannaschiif	47	3–329	H64432
		N. meningitidis, SiaCc	38	4–315	M95053
		Bacillus subtilis, SpsEf	34	2–334	X73124
NeuIIIC	UDP-GlcNAc 2-epimerasec	S. agalactiae Ia, NeuCf	94	1–320	AB028896
		E. coli, NeuCc	39	1–380	M84026
		N. meningitidis, SiaAf	38	4–315	M95093
		Rattus norvegicusc	29	1–318	Y07744
		Methanococcus thermoautotrophicumf	47	3–329	AE000860
NeuIIID	Sialic acid synthesisc	S. agalactiae type Ia, NeuDf	100	1–209	AB028896
		E. coli, NeuDc	30	1–209	U05248
		B. subtilis, YvfDg	25	3–208	Z71928
		Caulobacter crescentus, LpsBg	27	56–206	AF062345
		Campylobacter jejuni, WlaIg	27	65–207	Y11648
NeuIIIA	CMP-NeuNAc synthetasec	S. agalactiae type Ia, NeuAb	94	1–370	AB028896
		E. coli, NeuAc	37	2–395	J05023
		N. meningitidis, SiaBc	35	4–213	U04328
		H. influenzae, NeuAc	33	1–213	U32807
		H. ducreyi, NeuAc	29	1–209	U54496

^aTranslations of cpsIII ORFs were aligned using BLAST 2.0 with the BLASTP algorithm using the nonredundant database and the BLOSUM62 matrix at the National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, Md. The extent of the alignment returned by BLAST is given as the range. 

^bPredicted by sequence similarity. 

^cEnzymatic activity determined. 

^dPredicted based on similarity to S. agalactiae type Ia. 

^eDetermined in this study. 

^fProposed function. 

^gUnknown function. 

Upstream of the GBS CPS synthesis locus, but in the opposite orientation, is cps_IIIY. It is separated by an intergenic span of 189 nt from the cpsABCDEFGHIJKL neu_IIIBCDA (cpsIII) operon. At the 3′ end of the operon, downstream of neu_IIIA is a gap of 170 nt before another ORF is detected. This putative ORF has amino acid sequence homology with the central portion of the E. coli gene evgS, a putative sensor protein from a two-component regulatory system, and is therefore probably not a component of the capsule synthesis genes.

Sequence similarities of the CPS ORFs to other genes.

All of the cps gene sequences were translated, and the amino acid sequences were compared to protein sequence databases at the National Center for Biotechnology Information using the gapped BLAST version 2.0 program (1). Similarities to proteins in the sequence databases are presented in Table 2. The translated sequences of cps_IIIA through cps_IIIJ are similar to those of a number of genes from gram-positive organisms involved in CPS or exopolysaccharide synthesis.

The cps_IIIY and cps_IIIA (previously cpsX) genes coding for homologues of LysR and LytR, respectively, were previously described (24). The genes cps_IIIB, cps_IIIC, cps_IIID, and cps_IIIE have also been described (43). Cps_IIIB is homologus to S. pneumoniae CpsB, a tyrosine phosphatase involved in CPS regulation (36). Capsule expression is reduced when CpsB dephosphoryates CpsD, an autophosphorylating protein kinase involved in CPS chain length determination and transport. Cps_IIID is homologous to the S. pneumoniae CpsD and also shares amino acid similarity with the carboxy terminus of Wzc, an autophosphorylating tyrosine kinase for lipopolysaccharide synthesis in E. coli. Cps_IIIC, transcribed from the ORF immediately 5′ of cps_IIID, is homologous to the amino-terminal end of Wzc (40). Cps_IIIE is a glycosyl-1-phosphate transferase (60). The function of Cps_IIIF is unknown, but its S. pneumoniae type 14 homologue, Cps14F, is thought to enhance the activity of Cps14G. Cps_IIIG is homologous to Cps14G of S. pneumoniae and Cps_IaG from S. agalactiae type Ia, β-1,4-galactosyltransferases that catalyze the addition of the second monosaccharide in the formation of the CPS repeating unit.

The cps_IIIH gene is predicted to encode a protein with eight transmembrane domains which is homologous to the putative polysaccharide polymerase encoded by cps14H from S. pneumoniae type 14 (22). This homology led us to speculate that Cps_IIIH was responsible for polymerization of the oligosaccharide precursors in the type III CPS and thereby introduction of the structural determinant distinguishing type III from type Ia capsule. The aligned sequences of the cps_IIIH gene product, Cps14H from S. pneumoniae, and Cps_IaH, a GBS type Ia homologue (60), are depicted in Fig. 4. The translated type Ia and type III gene products are strikingly similar outside the Cps_IIIH region (99.3% similarity and 99.2% identity) but diverge markedly from the carboxy-terminal end of Cps_IIIG to the amino-terminal end of Cps_IIII (41% similarity and 28% identity). Further analysis of this region is discussed below. The Cps_IIII and Cps_IIIJ protein sequences are highly homologous to sequences of several glycosyltransferases, including Cps14I and Cps14J from S. pneumoniae type 14 (20, 22, 23).

FIG. 4.

Capsule subunit assembly locus. Regions of divergent sequence between type Ia and III strains are shown either striped (type III) or checked (type Ia). The degree of homology between the conserved and nonconserved regions is shown at the bottom. Gene designations are shown as single letters above each ORF map, and the CPS polymerase (H) is also indicated. The plasmids used in the polymerase expression constructs are shown beneath each map. Sequences deleted in the construction of pDC123(H_III) are shown with dashes.

The cps_IIIK gene shares homology with its type Ia GBS homologue, cps_IaK, with the lipooligosaccharide synthetase orfY (HI0871) from Haemophilus influenzae, and lst, which encodes a novel sialyltransferase in H. ducreyi (7). Lst, like Cps_IIIK, lacks homology to sialyltransferases in E. coli, H. influenzae, Neisseria gonorrhoeae, N. meningitidis, and mammals but shows significant homology to the H. influenzae OrfY. Interestingly, Lst catalyzes an α(2→3) linkage between the terminal sialic acid and a galactose residue in the lipooligosaccharide of H. ducreyi, the same specific glycosidic bond between these residues in the GBS capsular polysaccharides. These observations suggest that Cps_IIIK is the cpsIII sialyltransferase.

The cps_IIIL gene encodes a protein which is weakly similar to several known and presumed oligosaccharide transport proteins. The genes neu_IIIB, neu_IIIC, neu_IIID, and neu_IIIA encode the final four proteins in the locus. Neu_IIID (formerly called CpsE) and Neu_IIIA have been previously described (8, 15). These four genes are most similar to the neuABCD genes from the gram-negative organism E. coli K1 (6) and the siaCAB genes of N. meningitidis serogroup B (11), which are responsible for the synthesis and activation of sialic acid for capsule production in these species. We previously demonstrated that neu_IIIA encodes a CMP-NeuNAc synthetase, catalyzing the activation of sialic acid with CMP (15). Homologues of Neu_IIIB are NeuNAc synthetases (2). neuC of E. coli K1 has been characterized as a UDP-N-acetylglucosamine→ManNAc epimerase (R. P. Silver et al., submitted for publication). In this study, complementation of an E. coli K1 neuC mutant by neu_IIIC indicated that Neu_IIIC is functionally equivalent to the E. coli epimerase. Although the exact function of neu_IIID is unknown, a role in sialic acid synthesis has recently been established, since a nonpolar mutation in the E. coli K1 homologue neuD can be complemented by the addition of exogenous sialic acid or expression of neu_IIID in trans (9a).

Identification of the cpsIII transcriptional start site.

We hypothesized that the start of transcription occurred proximal to the cps_IIIA gene. To investigate this possibility, we used COH1 mRNA and a radiolabeled primer complementary to the 5′ end of cps_IIIA to identify the 5′ transcriptional start site of the cpsIII locus by primer extension analysis. The products of the primer extension reaction are shown in Fig. 5A. We mapped the 5′ end of the transcript (indicated by the arrow) to a G nucleotide on the DNA template 38 bp upstream of the cps_IIIA start codon. We examined the sequence directly upstream of the G nucleotide to identify canonical consensus promoter sequences. A sequence (TATAAT) identical to the −10 region of the consensus ς⁷⁰ E. coli promoter sequence was detected 7 bp above the 5′ end of the cpsIII transcript. A sequence 17 bp 5′ of the −10 region (TTGAAT) matched the canonical −35 region from E. coli at four of six positions. It also matched a consensus sequence for streptococcal promoters which is similar to the E. coli promoter, except that the −10 region is extended on the 5′ side to include the sequence TGN (with N being any nucleotide) (12, 52). We concluded this sequence is the cpsIII promoter.

FIG. 5.

(A) Primer extension reaction. Primer extension reactions were carried out on total cellular RNA derived from GBS strain COH1, using an antisense primer that hybridized to cps_IIIA mRNA. A sequencing ladder was generated from cps_IIIA double-stranded DNA using the same primer and was separated on a sequencing gel in lanes adjacent (lanes CTAG) to the primer extension products (p.ext.). The position of the apparent first ribonucleotide of the primer extension product (+1) is indicated to the right. The DNA sequence surrounding the start position of the mRNA is depicted vertically on the left, with the consensus −10 promoter sequence and the mRNA start site highlighted in bold. (B) RNA dot blots of total cellular RNA derived from the wt GBS strain COH1 and the ΩKm-2 cassette allelic exchange mutant strains COHY-107 (cps_IIIA), COHY-106 (cps_IIID), COHY-105 (cps_IIIC), and COHY-104 (cps_IIIB). The blots were hybridized to intragenic probes derived from the wt COH1 cps_IIIB, cps_IIIE, and neu_IIIA genes and exposed to X-ray film.

Polar effects of allelic exchange mutagenesis within the cpsIII locus.

To determine if there were additional transcriptional start sites within the cpsIII locus, we developed strains incorporating transcriptional terminators within their cps loci by allelic exchange mutagenesis (26, 61). To ascertain if these mutations were polar, we tested each clone for type III capsule production by immunoblot analysis as described. The type III antiserum recognizes both sialylated and asialo forms of the type III CPS. The wt strain, COH1, reacted with the antibody as expected; however, each of the CPS mutants failed to react, indicating they were acapsular (data not shown). These data demonstrated that mutations generated by homologous recombination in the 5′ region of the CPS synthesis locus prevented expression of capsule synthesis. These results also suggested that mutations in these upstream genes were polar and altered expression of downstream cps genes. This was expected, as a transposon mutant in cps_IIIB was previously shown to be acapsular (43).

Since each of the allelic exchange mutations above abolished capsule production, we sought to determine the effect of the mutations on the expression of downstream cps genes. RNA dot blot analysis was performed on total cellular RNA isolated from COH1 and each of the isogenic mutants. Equal amounts of RNA from each strain were hybridized to three different cps gene probes independently. The probes were specific for different regions of the CPS synthesis locus: the 5′ end (cps_IIIB), the central portion (cps_IIID), and the 3′ end (neu_IIIA). As shown in Fig. 5B, each of the probes hybridized to the RNA isolated from COH1 (wt) as expected and demonstrated comparable levels of mRNA transcribed from cps_IIIB, cps_IIIE, and neu_IIIA. In contrast, virtually no hybridization was observed between the three probes and the RNA from the cps_IIIA mutant. Since all of the probes are specific for genes 3′ of the ΩKm-2 insertion in cps_IIIA, these data suggest that this mutation was polar and blocked expression of all downstream genes.

RNA dot blot analysis using cps_IIIB-, cps_IIIC-, and cps_IIID-derived probes demonstrated hybridization proximal but not distal to each mutation as far downstream as neu_IIIA (data not shown). These results confirmed the observation that the cps genes are transcribed up to the site of the ΩKm-2 mutation but not distal to the site of the ΩKm-2 insertion for any of the mutants. Therefore, the ΩKm-2 cassette mutations interrupted transcription of the distal genes, providing further evidence that the CPS synthesis locus is transcribed as a large polycistronic message from cps_IIIA to neu_IIIA.

Development of strains expressing heterologous CPS.

Based on the observation that the GBS type Ia and type III CPS oligosaccharide repeating units are structurally identical in vivo, the serotype specificity of these two polysaccharides would be determined by the unique linkage introduced during polymerization of the repeating units into high-molecular-weight CPS (Fig. 1). Homology analysis between the CPS type Ia and III loci revealed a striking divergence between Cps_IaH and Cps_IIIH (Fig. 4), both of which share homology with repeating unit polymerase genes in other polysaccharide operons. In addition, we observed that the homology between the cpsIa and cpsIII loci diverged at amino acid 110 of the upstream glycosyltransferase gene cps_IaG (Fig. 4). We hypothesized that the putative polymerases are responsible for catalyzing the linkage between the type Ia or III subunits, which confer structural and therefore serotype specificity.

To address this hypothesis, we investigated how expression of the cps_IaH and cps_IIIH genes in the heterologous serotype strain would affect the type of CPS produced. cps_IaH and cps_IIIH were cloned independently into the gram-positive expression vector pDC123 (Fig. 4; Table 1). The cps_IaH construct, pDC123(H_Ia), was transformed into the type III capsule-producing strain, COH1. The type Ia-producing strain, A909, was transformed with pDC123(H_III) containing cps_IIIH. Due to the divergence in their 3′ termini, we also explored the contribution of the glycosyltransferase genes, cps_IaG and cps_IIIG, to the specific serotype of CPS synthesized. Therefore, additional constructs containing either cps_IaGH [pDC123(GH_Ia)] or cps_IIIGH [pDC123(GH_III)] were generated and transformed into the respective heterologous wt hosts. Immunoblot analysis was performed using rabbit type Ia or type III CPS polyclonal antisera to determine the nature of the CPS being expressed by the recombinant strains compared to the wt strains as controls. As seen in Fig. 6, cross-reactivity was not observed between A909 and the type III antisera or COH1 and antisera against type Ia CPS. When cps_IIIGH or cps_IIIH was expressed in A909, however, reactivity with the type III antiserum was observed, but reactivity with the type Ia antiserum was absent. Similarly, expression of cps_IaGH or cps_IaH alone in COH1 resulted in reactivity with the type Ia antisera (in addition to reactivity to the type III antisera for cps_IaGH). These data suggested that expression of cps_IIIH in A909 alone was capable of altering the type of CPS produced from Ia to III. Likewise, expression of cps_IaH in COH1 converted the CPS produced from type III to type Ia. Inclusion of the glycosyltransferase gene(s) (cps_IaG or cps_IIIG) just 5′ of the putative polymerase gene did not influence the nature of the CPS produced, as was anticipated from their roles as glycosyltransferases (60).

FIG. 6.

Immunoblots directed against immobilized whole GBS utilizing rabbit polyclonal anti-CPS antisera, detected with HRP-conjugated, goat anti-rabbit secondary antibody and chemiluminescent HRP substrate. Strains used: COH1, GBS type III strain; A909, GBS type Ia strain; COH1(GH_Ia), strain COH1 containing pDC123(GH_Ia); COH1(H_Ia), strain COH1 containing pDC123(H_Ia); A909(GH_III), strain A909 containing pDC123(GH_III); A909(H_III), strain A909 containing pDC123(H_III). (A) Blot exposed to type Ia antisera; (B) blot exposed to type III antisera.

The amount of CPS produced by the wt strain and the strains carrying the heterologous polymerase genes was quantified by competitive ELISA to determine if any native CPS in the latter strains was produced. Mutanolysin CPS extracts from the wt and recombinant strains were used to competitively inhibit the binding of serotype-specific polyclonal antisera to purified type Ia or type III CPS immobilized on microtiter plates. The degree of inhibition was compared to that of purified CPS similarly preincubated with the antibody, prior to transfer to the coated plates (Fig. 7). The CPS extracts from the wt strains A909 and COH1 inhibited type Ia and III antibody binding, respectively, as expected, and failed to inhibit reactivity of the heterologous CPS serotype antisera. For A909 containing pDC123(GH_III) or pDC123(H_III), production of type Ia CPS was reduced (1.7 and 1.2 ng/μl, respectively, versus 63.5 ng/μl for A909 alone). Both of these strains demonstrated a >37-fold decrease in type Ia-reactive CPS material compared to wt but produced a substantial amount of CPS which reacted with the type III antisera (20.8 and 26.1 ng/μl). COH1 containing pDC123(H_Ia), expressing the type Ia CPS polymerase, produced less type III CPS (10.5 ng/μl) and synthesized a quantity of anti-type Ia-reactive CPS equal to 72% of the wt A909 levels (45.4 ng/μl). A less dramatic effect was seen with COH1 containing pDC123(GH_Ia), expressing both the cps_IaG and cps_IaH alleles. The reason for this disparity is unclear but could be due to decreased expression of the CPS polymerase in this construct. Nevertheless, an anti-type Ia-reactive CPS was produced (3.2 ng/μl) with a concomitant decrease in type III CPS production (61.9 versus 91.0 ng/μl). These data demonstrated that expression of the heterologous CPS polymerase gene alone, in A909 or COH1, is sufficient to confer binary synthesis of CPS that are immunoreactive to both type Ia and type III antisera. Furthermore, inclusion of cps_IaG or its type III homologue cps_IIIG in the expression constructs was unnecessary for production of the altered CPS phenotype. These data also support the hypothesis that the putative polymerase gene is responsible for the glycosidic linkage between the repeating units of type Ia or III CPS and ultimately confers the serotype specificity of the CPS.

FIG. 7.

Competitive ELISA inhibition curves produced from mutanolysin extracts of wt and heterologous CPS polymerase-expressing strains. CPS extracts were used to compete with immobilized purified type Ia CPS for anti-type Ia polyclonal antibody (A) or with immobilized purified type III CPS for anti-type III polyclonal antibody (B). Strains A909 and COH1 produce native type Ia and type III CPS, respectively. Strain A909(H_III) expresses the cps_IIIH allele from COH1, and strain COH1(H_Ia) expresses the cps_IaH allele from strain A909.

DISCUSSION

The complete DNA sequences of several bacterial CPS and exopolysaccharide synthesis operons have been reported (10, 23, 28, 33, 35, 41, 48, 49). The organization of the regulatory and structural genes within these loci is remarkably well conserved across CPS and exopolysaccharide-producing species. The typical arrangement begins with regulatory determinants, then genes coding for control of polymer chain length and export, followed by structural genes for repeating unit assembly (glycosyltransferases) and polymerization (polymerases), and finally additional gene(s) involved in subunit transport and/or activated monosaccharide biosynthesis. The basic organization of the GBS type III operon reported here is similar to that of polysaccharide operons in other species and parallels that reported for the S. agalactiae type Ia operon (60).

Identification and characterization of the glycosyltransferases responsible for the structural organization of the oligosaccharide repeating units of S. pneumoniae types 3 (3, 10) and 14 (20, 22, 23) and the eps genes of Lactococcus lactis (49, 50) have recently been reported (Table 2). In addition, the S. agalactiae type Ia glycosyltransferases have recently been characterized and found to be functionally equivalent to their S. pneumoniae type 14 homologues Cps14E, Cps14G, Cps14I, and Cps14J (22, 60). In both species, these enzymes are responsible for the assembly of a tetrasaccharide that is chemically identical to the asialo type III GBS CPS repeating unit. We previously reported that cps_IIIE most likely encodes a galactosyltransferase. We could not, however, rule out the possibility that it may encode a glucosyltransferase (43). Given the homology of the type Ia glycosyltransferase genes to those in the cpsIII operon, and the identical structure of the oligosaccharide repeating unit, it is likely that these genes encode the same functions in both serotypes. Subsequent biochemical analysis of Cps_IaE and Cps14E in E. coli has shown they are glucosyl-1-phosphate transferases (22, 60), which catalyze the transfer of glucose to a receptor lipid intermediate to begin synthesis of the repeating unit. Based on these observations collectively, cps_IIIE should also code for a glucosyltransferase, and the assignment of cps_IIIE as a probable galactosyltransferase must be reevaluated.

Since the glycosidic linkage between the Ia and III pentasaccharide repeating units is the only apparent difference between the two CPS structures, and the type Ia and type III operons are virtually identical except for the region encoding the CPS polymerase, we hypothesized that the cpsH gene in both serotypes encoded the polymerase responsible for the repeating unit linkage. Although the carboxy-terminal domain of the upstream-encoded β-1,4-galactosyltransferases, Cps_IIIG and Cps_IaG diverged, our results confirm that these enzymes are not involved in linking the repeating units. Overexpressing the cps_IaH and cps_IIIH alleles in the heterologous serotype strains resulted in decreased expression of the native CPS and promoted synthesis of the alternative serotype CPS. We postulate that since plasmid pDC123 replicates at ∼90 copies/cell (8), the cps gene dosage of the recombinant plasmids was significantly higher than that of the chromosomal copy. Hence, depletion of the CPS precursor pool by the episomal polymerase gene product redirected polymerization of the repeating units into the alternative CPS structure. These results demonstrated that Cps_IaH and Cps_IIIH were responsible for the linkage between the repeating units of each CPS polymer.

Despite a number of recent publications characterizing the CPS glycosyltransferases in various gram-positive species, there is a lack of studies describing the CPS polymerases. The identification of genes encoding CPS polymerases within gram-positive cps loci has been based on the weak homology of the translated genes to the O-antigen polymerases of Shigella flexneri (37) and Salmonella enterica serovar Typhimurium (9). Direct evidence of polymerase function has been difficult to obtain presumably because biochemical approaches require significant amounts of the precursor oligosaccharides, which are difficult to isolate or synthesize. The need to reconstitute the polymerase with potential membrane-bound accessory proteins in a host suitable for carrying out the assays, such as E. coli, presents another difficulty. As a result, direct in vitro evidence of CPS polymerase activity has not been reported for gram-positive hosts producing complex, branched CPS. Only one other gene product from any gram-positive CPS- or exopolysaccharide-producing species has been unambiguously assigned CPS polymerase function, cap3B of type 3 S. pneumoniae (14). This enzyme does not share homology with Cps_IIIH and differs in activity, as it processively polymerizes an unbranched β(1→3)-linked glucuronic acid-(1→4)-β-d-glucose heteropolymer (7a). Examining CPS polymerase function by heterologous gene expression, as done here, provides the advantage of using the original host's native machinery in an intact cell for oligosaccharide production and provides a novel approach to assigning CPS gene function.

Divergent CPS polymerase genes have been detected in several pneumococcal serotypes and may be responsible for altered CPS structures in this species. In a study examining the genetic diversity of pneumococcal CPS synthesis loci (21), DNA from 26 different S. pneumoniae serotypes was probed with cps14H, the gene encoding the putative CPS polymerase. Only serotypes 15b and 15c, which have the same CPS core structure as type 14, hybridized to this probe. The structural differences between S. pneumoniae type 19A and type 19F CPS have been attributed to the potential activities of their respective CPS polymerases (34). Transformation of a type 19F strain with DNA containing the cps19AH and cps19AI genes resulted in a strain that produced a type 19A serotype. These data suggest that although CPS structural diversity within pneumococci is generally due to modification of glycosyltransferase and monosaccharide precursor usage, it can also be achieved through changes in the specificity of the polysaccharide polymerases. The S. agalactiae type Ia and type III data presented here provides compelling evidence that the polymerase gene is responsible for generating the linkage which confers the type-specific glycosidic bond between the oligosaccharide subunits and hence the serotype CPS produced.

Mutagenesis of the first four genes in the cpsIII operon and RNA dot blot analysis failed to show transcription starting distal to the promoter upstream of cps_IIIA, suggesting the absence of additional active promoters in the rest of the operon. These data indicated that the cpsIII operon is transcribed as a large polycistronic message. In contrast, Yamamoto et al. (60) identified an additional promoter in the type Ia locus between cps_IaD and cps_IaE (homologues of cps_IIID and cps_IIIE, respectively). This promoter may arise from differences between the type Ia and type III nucleotide sequences within the region where the type Ia secondary promoter was located. The most significant of these is the insertion of an additional A at base 3894 in the Ia nucleotide sequence that introduces a frameshift and subsequent stop codon. This results in the truncation of Cps_IaE compared to the type III homologue (the type III cps_IIIE ORF extends an additional 58 codons 5′ of the start site of the homologue type Ia cps_IaE gene). The additional adenine is located within the 17-bp spacer between the −35 and −10 regions of the type Ia promoter sequence. The shorter spacer in the type III locus may account for the lack of secondary promoter activity in our transcriptional analysis.

We are continuing to investigate the cps loci of the other GBS serotypes by analyzing the sequence differences of their operons. As might be expected from their CPS structure, we have obtained preliminary evidence, using the same genetic approach, that the CPS of serotype Ib S. agalactiae may be converted to type Ia by transferring and overexpressing a single type III galactosyltransferase gene. Similarly, an antigenically indistinguishable type III CPS can be produced from a type VI strain by transfer of a single type III glycosyltransferase gene. This approach promises to enable a rapid means of assigning CPS gene function and investigating polysaccharide biosynthesis in GBS and other gram-positive organisms.