Genetic Diversity of the Capsular Polysaccharide C Biosynthesis Region of Bacteroides fragilis

Abstract

A genetic approach was used to assess the heterogeneity of the capsular polysaccharide C (PS C) biosynthesis locus of Bacteroides fragilis and to determine whether distinct loci contain genes whose products are likely to be involved in conferring charged groups that enable the B. fragilis capsular polysaccharides to induce abscesses. A collection of 50 B. fragilis strains was examined. PCR analysis demonstrated that the genes flanking the PS C biosynthesis region are conserved, whereas the genes within the loci are heterogeneous. Only cfiA⁺ B. fragilis strains, which represent 3% of the clinical isolates of B. fragilis, displayed heterogeneity in the regions flanking the polysaccharide biosynthesis genes. Primers were designed in the conserved regions upstream and downstream of the PS C locus and were used to amplify the region from 45 of the 50 B. fragilis strains studied. Fourteen PS C genetic loci could be differentiated by a combination of PCR and extended PCR. These loci ranged in size from 14 to 26 kb. Hybridization analysis with genes from the PS C loci of strains 9343 and 638R revealed that the majority of strains contain homologs of wcgC (N-acetylmannosamine dehydrogenase), wcfF (putative dehydrogenase), and wcgP (putative aminotransferase). The data suggest that the synthesis of polysaccharides that have zwitterionic characteristics rendering them able to induce abscesses is common in B. fragilis.

Bacteroides fragilis is one of the anaerobes most commonly isolated from clinical infections and is the predominant anaerobe in intra-abdominal abscesses. B. fragilis induces the formation of abscesses in animal models of intra-abdominal infection. The prototype strain used to study abscess formation, NCTC9343, produces a capsular polysaccharide complex that is composed of at least three distinct high-molecular-weight polysaccharides termed PS A, PS B, and PS C (4, 9). In animals, purified PS A and PS B induce intra-abdominal abscess formation in the absence of organisms (13). PS C has not yet been purified; therefore, the structure of the repeating unit has not yet been elucidated. The most striking feature of PS A and PS B is the presence of both positively and negatively charged groups on each repeating unit. Structure-function studies have demonstrated that this zwitterionic charge motif is required for these polysaccharides to induce abscesses (13). Other bacterial polysaccharides that contain both positive and negative charges are also able to induce abscess formation (13).

Studies using monoclonal antibodies (MAbs) have revealed interstrain heterogeneity among the capsular polysaccharides of B. fragilis (6, 7). We have sequenced the PS C biosynthesis loci of two strains, 9343 and 638R (2 [the 638R PS C locus is referred to as PS B2 locus in this reference], 3), which produce immunologically distinct PS C's. These loci are flanked by genes common to both strains; however, the genes involved in polysaccharide biosynthesis are distinct. The structure of the repeating unit of the PS C of strain 9343 (PS C1) has not been determined; however, the biosynthesis locus contains a gene, wcfF, which encodes a product similar to sugar dehyrogenases of other bacterial polysaccharide biosynthesis loci, which would confer a negative charge to one of the monosaccharides of the PS C1 repeating unit (3).

The structure of the repeating unit of the PS C of strain 638R (PS C2) is currently being determined. The genetic locus encoding PS C2 contains two genes (wcgH and wcgP) that encode products similar to aminotransferases. Each of these products likely transfers a positively charged free amino group to the repeating unit of PS C2. In addition, the locus contains a gene, wcgC, that has been shown by complementation to encode a UDP–N-acetylmannosamine dehydrogenase (2). WcgC converts UDP–N-acetylmannosamine to UDP–N-acetylmannosaminuronic acid, a negatively charged monosaccharide. On the basis of these findings, it is likely that PS C2 contains the necessary charge motif for the induction of abscess formation.

Although the loci encoding the PS Cs of strains 9343 and 638R have been shown to be distinct, the degree of genetic heterogeneity of the PS C biosynthesis loci of the species as a whole is unknown. In addition, it is not known whether all or most B. fragilis strains produce capsular polysaccharides that have both a positive and a negative charge and therefore can induce abscesses if released into the peritoneal cavity. Since both 9343 and 638R are clinical isolates, it is possible that only a subset of intestinal isolates contain this charge motif. In this study we examined the genetic heterogeneity of the PS C region from a variety of B. fragilis strains representing clinical and fecal isolates and from strains of various genetic backgrounds. In addition, we investigated the presence of genes whose products are involved in conferring charged groups to polysaccharides.

MATERIALS AND METHODS

Bacterial strains.

The B. fragilis strains used in this study are listed in Table 1. The other Bacteroides species used in the study are B. thetaiotaomicron 29741 (1), B. ovatus 8483 (5), B. vulgatus 8482 (5), B. uniformis 1001 (12), and B. distasonis 8503 (5). Bacteroides strains were grown anaerobically in supplemented brain heart infusion (BHI) broth (BHIS; Randolph Biomedical, West Warwick, R.I.) or on BHI agar plates supplemented with hemin (50 μg/ml) and menadione (0.5 μg/ml).

TABLE 1.

B. fragilis strains used in this study and PCR results for each strain

Strain	Sourceb	PCR resulta
		cepA	cfiA	bft	PCRs in capsule regions
					1	9343 specific				6	7	638R specific					PS C size (kb)	PS C type
						2	3	4	5			9	10	11	12	13
9343	Abscess, NCTC	*			*	*	*	*	*	*	*						16.9	1
US52540	Clinical, BWH	*		*	*	*	*	*	*	*	*						16.9	1
PA5	Fecal, ISS	*		*	*	*	*	*	*	*	*						16.9	1
US332	Abscess, BWH	*		*	*	*	*	*	*	*	*						16.9	1
B68	Clinical, ISS	*			*	*	*	*	*	*	*						16.9	1
1277810I	Kidney, BWH	*		*	*	*	*	*	*	*	*						16.9	1
38310	Middle ear, BCH	*			*	*	*	*	*	*	*				O		16.9	1
US388	Ischial wound, BWH	*			*	*		*	*	*	*						14.8	3
US398	Blood, BWH	*			*	*		*	*	*	*						14.8	3
1279155I	Palm, BWH	*			*	*		*	*	*	*						14.8	3
1287245I	Foot, BWH	*			*			*	*	*	*						14.8	4
1281262I	Peritoneal fluid, BWH	*			*				*	*	*						17	5
1285531I	Peritoneal fluid, BWH	*			*			*		*							22.5	6
US365	Peritoneal fluid, BWH	*			*			O	*	*							14	7
US357	Appendix, BWH	*			*			O	*	*							14	7
17905	Placenta, BCH	*			*			O	*	*							14	7
US390	Abscess, BWH	*			*				*	*							17	8
US2244	Clinical, BWH	*			*				*	*							17	8
US324	Peritoneal fluid, BWH	*			*				*	*							17	8
B272	Clinical, ISS	*			*				*	*							17	8
CM11	Fistula, RDR	*			*				*	*							17	8
CMR2896	Intestinal, RDR	*			*				*	*							17	8
379	Abscess, BCH	*			*				*	*							17	8
B110	Fecal, ISS	*			*				*	*							17	8
IL89375II	Peritoneal fluid, BWH	*			*				*	*							17	8
B35	Fecal, ISS	*			*				*	*							17	8
B124	Fecal, ISS	*			*				*	*					O		17	8
1291662III	Peritoneal fluid, BWH	*			*				*	*							18	9
B356772I	Blood, BWH	*			*			O	*	*							18	9
1284-2	Fecal, BWH	*			*				*	*							20	10
US1206	Clinical, BWH	*			*				*	*							20	10
1279-2	Fecal, BWH	*			*				*	*					*		18	11
2429	Clinical, WVAH	*			*			O		*	*	*	*	*	*	*	23.2	2
1283931I	Great toe, BWH	*			*			O		*	*	*	*	*	*	*	23.2	2
12905-23V	Abscess, BWH	*			*			O		*	*	*	*	*	*	*	23.2	2
12775L1II	Peritoneal fluid, BWH	*			*					*	*	*	*	*	*	*	23.2	2
638R	Clinical	*			*					*	*	*	*	*	*	*	23.2	2
B16	Clinical, ISS	*			*					*	*	*	*	*	*	*	23.2	2
13141	Placenta, BCH	*		*	*					*	*	*	*	*	*	*	23.2	2
CM3	Ascitic fluid, RDR	*		*	*					*	*	*	*	*	*	*	23.2	2
CM12	Abscess, RDR	*		*	*					*	*			*	*	*	23.2	12
45703	Subdural fluid, BCH	*		*	*					*							24.5	13
44355	Lochia, BCH	*		*	*					*							24.5	13
26877	Clinical, BCH	*			*					*							24.5	13
B117	Fecal, ISS	*			*					*							26	14
US326	Peritoneal fluid, BWH	*			*					*							NAc
1277476	Peritoneal fluid, BWH		*		*			O									NA
1281550II	Peritoneal fluid, BWH		*		*												NA
TAL2480	Clinical, TAL		*		*												NA
3636	Clinical, TAL		*		*												NA

^a∗, the PCR product was amplified for that strain; O, the PCR product was amplified for that strain, but from a region outside the PS C locus. 

^bISS, Laboratory of Bacteriology and Medical Mycology, Instituto Superiore di Sanita, 00161 Rome, Italy; BWH, Brigham and Women's Hospital, Clinical Microbiology Laboratory, Boston, Mass.; BCH, Boston City Hospital, Boston, Mass.; TAL, Tufts Anaerobe Laboratory, Boston, Mass.; RDR, R. Di Rosa, Policlinico Umberto I, Rome, Italy; and WVAH, Wadsworth VA Hospital, Los Angeles, Calif. 

^cNA, no extended PCR product was amplified for this strain. 

PCR.

The PCR mixtures contained 20 μl of PCR Supermix (Life Technologies, Gaithersburg, Md.), an additional 0.5 U of Taq polymerase (Life Technologies), 3.2 pmol of each primer, and 20 ng of chromosomal DNA. Primers to amplify wcfF were designed using accession no. AF048749 from the following 5′→3′ bp: 12161 to 12181 and 13060 to 13040. Primers to amplify areas of PS C2 (AF125164) were from the following 5′→3′ sequences: wcgC (bp 8805 to 8826 and 9424 to 9403), wcgH (bp 12965 to 12986 and 13727 to 13705), wcgP (bp 20701 to 20721 and 21372 to 21351). bft primers and program were as described previously (8). The internal portion of cepA was amplified with primers comprising bp 470 to 490 and bp 1188 to 1168 from GenBank entry L13472. The internal portion of cfiA was amplified with primers comprising bp 749 to 769 and bp 1256 to 1236 of GenBank entry M34831. The thermal cycler programs all have an initial segment at 94°C for 2 min and were run for 30 cycles. Each PCR was performed at least twice.

Extended PCR.

The primers used for extended PCR were EPCR-F2 (5′-TACGAACGTAGTCTTGGAGACAACAGATAG) and EPCR-R (5′-CTGACGATTTAGTTACCCGTGAAGTAATCT). The PCR mixtures contained 50 μl of PCR Supermix High Fidelity (Life Technologies), 2 U of Elongase enzyme mix (Life Technologies), 8.2 pmol of each primer, and 50 ng of chromosomal DNA, with 30 cycles of 94°C for 30 s, 61°C for 30 s, and 68°C for 20 min.

Southern blotting and hybridizations.

Agarose gels containing extended PCR products were treated with 0.1 N HCl for 10 min prior to denaturation and transfer to nylon. The DNAs used to probe these blots and EcoRI-digested chromosomal blots were PCR products representing internal portions of wcfF, wcgC, wcgH, and wcgP or PCR-4 and PCR-12 products. All probes were gel purified with the Qiaquick gel extraction kit (Qiagen, Valencia, Calif.). Probes were labeled with the ECL Direct Labeling Kit (Amersham, Piscataway, N.J.), and hybridizations were carried out according to the manufacturer's recommendations for stringent conditions with wash solution containing 0.1× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate–4% sodium dodecyl sulfate (SDS).

SDS-PAGE and Western blotting.

Bacterial cell lysates were subjected to discontinuous SDS-polyacrylamide gel electrophoresis (PAGE) on a 4 to 20% acrylamide gel gradient (ESA, Inc., Chelmsford, Mass.) and transferred to Immobilon (Millipore, Bedford, Mass.) by standard techniques. The blots were blocked in Tris-buffered saline (TBS) containing 5% nonfat dry milk (TBS-milk) and then incubated for 1 h in TBS-milk containing primary antibody. The blots were washed and incubated with alkaline phosphatase-conjugated goat anti-rabbit or anti-mouse immunoglobulin G (Sigma Chemical Co., St. Louis, Mo.) at 1:1,000 in TBS. The blots were washed and developed with a colorimetric solution. Culture supernatant containing MAb 4D5 was used at 1:10. Antiserum to PS C1, prepared as described elsewhere (3), was used at 1:40.

RESULTS

PCR analysis of the PS C biosynthesis region.

Comparison of the sequence of the PS C1 and PS C2 biosynthesis loci of strains 9343 and 638R has demonstrated that the genes flanking these loci are conserved, whereas the genes involved in polysaccharide biosynthesis are distinct (Fig. 1) (2). To study the genetic heterogeneity of this region of the chromosome within the species B. fragilis, PCR was initially used. Primers were chosen that would generate products falling into three categories: those that were entirely upstream or downstream of the PS C biosynthesis loci (PCR-1 and PCR-6), those that spanned the junctions between the genes involved in polysaccharide biosynthesis and the adjacent flanking genes (PCR-2, PCR-5, PCR-7, and PCR-13), and those that spanned several genes within the PS C1 or PS C2 biosynthesis loci (PCR-3, PCR-4, PCR-9, PCR-10, PCR-11, and PCR-12) (Fig. 1). Fifty strains of B. fragilis and one strain each of B. thetaiotaomicron, B. vulgatus, B. ovatus, B. uniformis, and B. distasonis were subjected to these PCRs. To ensure that strains representing the genetic heterogeneity of B. fragilis were included in this study, we chose isolates of both clinical and fecal origin, isolates containing the enterotoxin gene (bft) and nontoxigenic strains, and isolates containing the cephalosporinase gene cepA or cfiA.

FIG. 1.

Comparison of the PS C1 and PS C2 biosynthesis regions of B. fragilis strains 9343 and 638R demonstrating the locations of primers used for PCR. The numbers above the primers indicate the base-pair location of the primers 5′→3′ in regard to the submitted GenBank sequences AF048749 (9343 PS C1) or AF125164 (638R PS C2). The direction of transcription of each open reading frame is designated with an arrow. The open reading frames that do not exhibit homology with genes involved in polysaccharide biosynthesis are shaded. The area between the diagonal lines indicates where the two chromosomes diverge.

The results of these PCRs are presented in Table 1. A product was amplified from all B. fragilis strains in PCR-1 and from all strains except the four cfiA⁺ strains in PCR-6. The DNA homology upstream of the 9343 and 638R PS C loci has been shown to continue into the first gene of each region, rmlA, and the two chromosomes begin to diverge at bp 338 of this gene (2). If this conserved portion of rmlA was present in the PS C locus of all B. fragilis strains, a product would be amplified from all strains in PCR-7. In fact, a product was amplified from only 21 of the 50 B. fragilis strains. This result indicated that the homology does not continue into rmlA for all B. fragilis strains.

PCR products were amplified from some of the strains in the PCRs where the primers spanned the junctions and when the primers were completely within the polysaccharide biosynthesis loci. The results of these PCRs differentiated the strains into 10 PS C genetic types. A PCR-5 product was detected from 31 of the 50 strains examined. Since the primers for this PCR span the downstream junction and includes wcfL, these data indicate that more than half of B. fragilis strains contain wcfL as the last gene in the PS C region. The product encoded by wcfL is similar to undecaprenol-phosphate N-acetylglucosaminyltransferases; therefore, this finding suggests that the first sugar of the repeating unit of the PS C's of these strains is common.

PCR products were detected from six strains (other than 9343) in all of the 9343-specific PCRs (PCRs 2 to 5), a result suggesting that these strains may synthesize PS C1. Products were detected from only 12 strains in any of the 638R PS C2-specific PCRs. With only four exceptions, if a product was detected from a strain in any of the PS C2-specific PCRs, it was amplified in all of them (PCRs 9 to 14). In addition, amplification of a PCR product for PS C1 or PS C2 was generally exclusive. When a PCR product was detected from a strain in both a PS C1-specific and a PS C2-specific PCR, hybridization analysis demonstrated that one of the products was derived from sequences not contained in the PS C biosynthesis locus of that strain (with the exception of strain 1279-2).

No products were detected from the other Bacteroides species in any of the PCRs.

Amplification of the PS C loci of B. fragilis strains.

The results of PCR-1 and PCR-6 revealed that the upstream and downstream regions flanking the PS C biosynthesis loci are conserved in all cepA⁺ B. fragilis strains. Therefore, primers EPCR-F2 and EPCR-R were designed to PCR amplify the entire PS C biosynthesis locus from the collection of B. fragilis strains used in this study (Fig. 1). The results of these PCRs are shown in Fig. 2. A product was amplified for 45 of the 46 cepA⁺ strains. The products ranged in size from 14 to 26 kb. The expected 16.5-kb product amplified from strain 9343 was also amplified from the six strains that produced a PCR pattern identical to that produced by strain 9343. Likewise, an approximately 23-kb product was amplified from the seven strains with a PCR profile identical to that of strain 638R.

FIG. 2.

Ethidium bromide-stained 0.5% agarose gel of the PCR products generated from the indicated B. fragilis strains with primers EPCR-F2 and EPCR-R. MW indicates the lanes containing high-molecular-weight markers (Life Technologies). Kilobase sizes are indicated to the left. The PS C type of each strain is listed above.

In many instances, the product size was consistent for strains with the same PCR profile; in others, the product size demonstrated increased heterogeneity. The combined results of PCR and extended PCR divided the strains into 14 PS C types designated PS C1 to PS C14 (Table 1).

Correlation of PCR profile with antibody reactivity.

To determine whether strains that appear to have PS C regions identical to that of strain 9343 (PS C1) or strain 638R (PS C2) actually synthesize an immunologically similar polysaccharide, antibodies specific to PS C1 or PS C2 were used to probe all the B. fragilis strains. The only six strains of the collection that demonstrated reactivity with PS C1-specific antiserum were those that were grouped as PS C1, i.e., strains US52540, PA5, US332, B68, 1277810I, and 38310; however, the latter three strains reacted only weakly with this antiserum (Fig. 3A). None of the other 43 strains reacted at all with this antiserum. Similarly, the only strains that reacted with 4D5, the MAb specific to the PS C of 638R, were 2429, 1283931I, 12905-23V, 12775L1II, B16, 13141, and CM12. It was expected that CM3 would also react with this antiserum since its PCR profile was identical to that of the other PS C2 strains; however, this strain did not react with this MAb. Conversely, strain CM12, from which a product was detected in only four of the six 638R PS C-specific PCRs, reacted with this antiserum at a level comparable to the other positive strains. None of the other 42 strains reacted with this MAb. These data demonstrate a strong correlation between the PCR profile and the polysaccharide produced (PS C1 or PS C2) in these two cases.

FIG. 3.

Western immunoblot of cell lysates of B. fragilis strains. (A) Strains probed with antiserum specific to PS C1. (B) Strains probed with MAb 4D5 (PS C2 specific). The 220-kDa marker is indicated on the left.

Presence of genes conferring charge groups.

The PCR data have documented the heterogeneity of the PS C region of the B. fragilis chromosome but have not indicated whether different loci contain genes encoding products that may confer charge groups to the repeating units. To begin to address this question, genes from the (PS C1) an (PS C2) loci encoding products likely to confer charge groups were used as probes. Blots of the PCR products amplified with primers EPCR-F2 and EPCR-R, representing the PS C loci of the various strains, and chromosomal EcoRI digests were probed with internal portions of wcfF and wcgC (dehydrogenase homologs) and of wcgH and wcgP (aminotransferase homologs). The results of these hybridizations are shown in Table 2. Homologs of wcfF, wcgC, and wcgP were present in most or all of the B. fragilis strains examined, including the cfiA⁺ strains; however, these genes were rarely present in the PS C biosynthesis region of the various chromosomes. A wcgH homolog was present less frequently in the chromosome of B. fragilis, and in most cases this gene was present in the PS C locus. With few exceptions, only strains of the PS C1 type contained wcfF in the PS C locus, and only strains of the PS C2 type contained wcgC, wcgH, and wcgP in the PS C locus. In those strains for which the wcfF or wcgP probes hybridized to DNA not contained in the PS C locus, the hybridizing EcoRI fragments were of various sizes among the different B. fragilis strains, a result suggesting the presence of these genes in regions of the chromosome that are variable from strain to strain. In addition to strains of PS C2 and PS C12 types which contain wcgC in the PS C locus, 47 of the 50 B. fragilis strains analyzed contained an EcoRI fragment (outside of the PS C locus) of approximately 9 kb that hybridized with the wcgC probe. For many of these strains, an additional fragment that varied in size was also detected with the wcgC probe.

TABLE 2.

Hybridization of gene probes with the chromosome or PS C locus of B. fragilis strains

Strain (PS C type)	Hybridizationa of:
	9343 wcfF		638R wcgC		638R wcgH		638R wcgP
	Chrom	PS C	Chrom	PS C	Chrom	PS C	Chrom	PS C
9343 (1)	*	*	*				*
US52540 (1)	*	*	*				*
PA5 (1)	*	*	*				*
US332 (1)	*	*	*				*
B68 (1)	*	*					*
1277810I (1)	*	*	*				*
38310 (1)	*	*	*				*
US388 (3)	*		*				*
US398 (3)			*		±		*
1279155I (3)	*		*				*
1287245I (4)			*				*
1281262I (5)	*		*				*
1285531I (6)	*	*	*				*
US365 (7)	*						*
US357 (7)	*		*				*
17905 (7)	*		*				*
US390 (8)			*				*
US2244 (8)	*		*				*
US324 (8)	*		*				*
B272 (8)			*				*
CM11 (8)			*				*
CMR2896 (8)	*		*				*
379 (8)	*		*		±		*
B110 (8)			*				*
IL89375II (8)	*		*				*
B35 (8)	*		*				*
B124 (8)			*				*
1291662III (9)			*				*
B356772I (9)	*		*				*
1284-2 (10)			*		*		*
US1206 (10)							*
1279-2 (11)	*		*				*
2429 (2)	*		*	*	*	*	*	*
1283931I (2)	*		*	*	*	*	*	*
12905-23V (2)	*		*	*	*	*	*	*
12775L1II (2)	*		*	*	*	*	*	*
638R (2)			*	*	*	*	*	*
B16 (2)	*		*	*	*	*	*	*
13141 (2)	*		*	*	*	*	*	*
CM3 (2)	*		*	*	*	*	*	*
CM12 (12)	*		*	*	*	*	*	*
45703 (13)	*		*				*
44355 (13)	*		*				*
26877 (13)	*		*				*
B117 (14)	*		*		*		*
US326	*		*				*
1277476	*		*				*
1281550II	*		*				*
TAL2480	*		*				*
3636	*		*				*

^a∗, the designated probe hybridized under stringent conditions with the chromosome (Chrom) or PS C region of the listed strains; ±, hybridization occurred, albeit to a lesser extent than that demonstrated with the chromosome from which the probe was derived. 

DISCUSSION

The biosynthesis loci of the capsular polysaccharides of B. fragilis are beginning to be studied. We took advantage of the sequence data for two published polysaccharide biosynthesis regions of B. fragilis, PS C1 of 9343 and PS C2 of 638R, to study the heterogeneity of the PS C region of the chromosome among the species B. fragilis. This study focused particularly on the degree of heterogeneity and the question of whether genes whose products likely confer charged groups to PS C1 and PS C2 are present in strains that produce other PS C types. These data will help to determine whether other strains that produce distinct capsular polysaccharides may also have the ability to induce abscesses.

PCR-1 results indicate that the DNA upstream of the PS C biosynthesis region is conserved in all B. fragilis strains tested. Since no product was amplified for any of the other Bacteroides species, PCR-1 may be useful for species determination. In strains 9343 and 638R, the homology to the left of the PS C region continues into rmlA, where the first 338 bp of the genes are identical. PCR-7 results indicate that this homology does not exist in all strains; rather, the last homologous gene upstream of the PS C locus in all B. fragilis strains is upcZ.

Downstream of the PS C loci, interstrain homology begins with orf5 for all cepA⁺ strains tested. No product was amplified in PCR-6 in the four cfiA⁺ strains. cfiA⁺ strains represent only 3% of clinical isolates, and ribosomal DNA sequencing and other genetic analyses have shown that cepA⁺ and cfiA⁺ strains are distinct subspecies of B. fragilis (10, 11). Our studies extend the heterogeneity of the cfiA wild-type and mutant strains to include the PS C downstream flanking region.

The results of the PCR amplification of portions of the PS C1 and PS C2 biosynthesis regions differentiated the 50 strains into 10 PS C types. Extended PCR analysis further differentiated these strains into 14 PS C types. It is intriguing that 31 of the 46 cepA⁺ strains contain wcfL at the end of the PS C locus (PCR-5 positive). WcfL is similar to other products that transfer a sugar to undecaprenol-phosphate, the first step in assembly of the repeating unit. It is unknown which monosaccharide is transferred to undecaprenol-phosphate by WcfL in the synthesis of the repeating unit of PS C1; however, these data suggest that this sugar is present in PS C's of types 1 and 3 to 11.

According to these genetic analyses, 13 of the 50 strains fall into the same PS C type as the 9343 PS C1 or the 638R PS C2. With only a few exceptions, antibody typing showed that these strains produce immunologically similar PS C's. Strains B68, 1277810I, and 38310 all react only slightly with PS C1-specific antiserum, although they contain a PS C locus that appears to be genetically identical to that of PS C1. The presence or absence of a gene product encoded outside the biosynthesis region may account for polysaccharides that have immunological differences but contain the same genetic material in the PS C locus. Alternatively, the genetic analyses used may not be sufficiently sensitive to detect subtle differences in PS C loci that may account for immunologically altered polysaccharides.

The presence of both a positively and a negatively charged group on a bacterial polysaccharide is crucial to the polymer's ability to induce abscesses. We have previously identified four genes in the 9343 or 638R PS C loci encoding products that may confer charge groups to their respective polysaccharides. WcfF, a putative dehydrogenase, and WcgC, an N-acetylmannosamine dehydrogenase, would each confer a negative charge; WcgH and WcgP, both similar to aminotransferases, would confer positive charges. Because we were able to amplify the entire PS C locus from 45 of the 50 strains, we could determine not only whether a strain contained these particular genes but also whether the gene was contained in the PS C locus. The results demonstrate the conservation of these genes within the species. Most notable were the results of hybridization with wcfF, wcgC, and wcgP. These genes were present in all (wcgP) or most of the strains tested (wcgC, 94%; wcfF, 78%). However, for the majority of strains, these genes were not contained in the PS C region. The wcgP and wcfF probes hybridized with EcoRI fragment(s) that differed in size between strains. These results suggest that this DNA comprises a region of the chromosome that is variable between strains, as would be expected for a region encoding serologically distinct PS C types among strains. As the PS A and PS B loci of all strains except 9343 are unsequenced, it is possible that some of these genes are contained in these capsule loci. The probing results with wcgC were different from those with wcgP and wcfF. In addition to hybridizing with EcoRI fragments of differing size between strains, this probe also hybridized with an EcoRI fragment of approximately 9 kb in 42 of the 50 strains. Therefore, a conserved area of the B. fragilis chromosome contains a gene encoding an N-acetylmannosamine dehydrogenase.

The data indicate that the synthesis of polysaccharides with both positively and negatively charged groups is a characteristic of the species B. fragilis as a whole. The charged polysaccharides are crucial to abscess formation by B. fragilis and thus provide an enormous advantage to the organism for survival in extraintestinal sites. The presence of B. fragilis in extraintestinal sites, however, is accidental and does not contribute to the organism's propogation from person to person. Therefore, the presence of charged polysaccharides would not be selected for because of the advantage they provide for extraintestinal survival. It is most probable that the synthesis of charged polysaccharides by B. fragilis is conserved because these polymers confer a selective advantage in their natural environment of the colon.