INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteroides fragilis is an important cause of intra-abdominal abscesses arising from introduction of this organism into the peritoneal cavity via surgical, traumatic, or disease-induced bowel perforation. The capsular polysaccharide complex of the B. fragilis strain used for the study of abscess formation, NCTC 9343, is a potent inducer of abscesses. It is composed of at least three distinct capsular polysaccharides: polysaccharide A (PS A), PS B, and PS C. Purified PS A and PS B are each capable of inducing intra-abdominal abscesses in a rodent abscess model, although PS A is more potent: the dose of purified PS A causing abscesses in 50% of animals (AD₅₀) is 0.67 µg, whereas the AD₅₀ of PS B is 25 µg (30). PS C has not yet been purified to homogeneity.

The structures of the PS A and PS B capsular polysaccharides of strain 9343 have been determined; each of these polysaccharides contains both positively and negatively charged groups. The repeating unit of PS A has one free amino and one carboxyl group, whereas the repeating unit of PS B has one free amino and two negatively charged (carboxyl and phosphonate) groups (1). The genetic composition of the PS C biosynthesis locus suggests that the PS C repeating unit has one carboxyl group (5).

The presence of both a free amino and at least one negatively charged group is essential for these polysaccharides to induce abscesses (30). Chemical conversion of a free amino group to a positively charged tertiary amine or to a neutral N-acetyl group results in loss of abscessogenic activity (29). We have shown that polysaccharides with repeating unit structures that differ from those of PS A or PS B but that possess both free amino and negatively charged groups are also capable of inducing abscesses in animals. Two surface polysaccharides from Streptococcus pneumoniae, the type 1 capsular polysaccharide and teichoic acid (C substance), each contain the same free amino sugar component as PS A (10, 16) and have AD₅₀s of 31 and 5 µg, respectively (30). In addition, repeating units with one negatively charged group acquire the ability to induce abscesses in the animal model when chemically converted to possess free amino groups (30).

Although both PS A and PS B in purified form have been shown to induce abscesses in animal models, there has been no demonstration that these are the only, or most potent, abscess-inducing molecules produced by B. fragilis. The contribution of PS B to abscess formation was analyzed using a mutant in which 5.7 kb of the PS B biosynthesis region was deleted. This PS B-negative mutant was fully virulent in the rodent abscess model (9). The contribution of PS C to the abscess-inducing ability of 9343 was similarly analyzed using a mutant with a chromosomal deletion in the PS C biosynthesis locus. This mutant was also not attenuated in its ability to induce abscesses (6).

In this study, we identified and characterized the PS A biosynthesis locus of 9343. Through the creation of a PS A-negative mutant strain, the singular importance of this polysaccharide for abscess induction by B. fragilis was revealed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteria, plasmids, and media. Bacterial strains and plasmids are described in Table 1. Escherichia coli strains were grown in L broth or on L agar plates. B. fragilis strains were grown anaerobically in basal medium (23) or on brain heart infusion plates supplemented with hemin (50 µg/ml) and menadione (0.5 µg/ml) (BHIS). The following concentrations of antibiotics were added when appropriate: ampicillin, 100 µg/ml; kanamycin, 20 µg/ml; erythromycin, 5 µg/ml; trimethoprim, 100 µg/ml; gentamicin, 200 µg/ml.

Transposon mutagenesis to locate the PS A biosynthesis locus. As 9343 is somewhat resistant to the introduction of foreign DNA, B. fragilis strain B16 (also reactive with a monoclonal antibody [MAb] to the PS A of 9343, MAb CE3 [23]) was used for transposon mutagenesis. Transposon suicide vector pNJR64 was conjugally transferred, as previously described (6), from E. coli DH5 into B. fragilis B16 using helper plasmid R751 (27). B. fragilis transconjugants acquiring resistance to erythromycin encoded by pNJR64 were screened for loss of MAb CE3 reactivity. One such CE3-negative B16 clone, mutant 65, was further analyzed.

Isolation of 9343 cosmid clones containing the PS A biosynthesis locus and generation of subclones. The chromosome of B. fragilis B16 mutant 65 (MAb CE3 negative) yielded a HindIII fragment containing the origin of replication and Km^r gene of pNJR64, as well as several kilobases of B. fragilis B16 DNA. This fragment was cloned by ligation of HindIII-digested mutant 65 chromosomal DNA, transformation of DH5, and selection with kanamycin. The resulting plasmid, pLEC12, has a very low copy number; therefore, a PCR product designated P65 containing a portion of the B16 junctional DNA was synthesized. Using the P65 PCR product as a probe, two pHC79-based cosmid clones, pLEC15 and pLEC21 (Fig. 1A), were selected from a B. fragilis 9343 gene bank. Restriction fragments of pLEC15 and pLEC21 were subcloned into pBluescript II SK (Stratagene, La Jolla, Calif.) and used as sequencing templates (Fig. 1A).

View larger version (18K): [in this window] [in a new window]   FIG. 1.   (A) Cosmids and subclones used in sequencing the 9343 PS A locus. B, BamHI; E, EcoRI; P, PstI; B/S, compound site resulting from cloning Sau3AI-digested DNA into the vector BamHI site. (B) Diagrammatic representation of the PCRs showing the positions of the primers and the span of the product. PCRs marked with an asterisk were successful in more than 98% of the strains tested. (C) ORF map of the B. fragilis 9343 PS A locus. The region deleted in mutant strain 9343PSA and the positions of the primers used for EPCR are indicated above the ORFs. Flanking ORFs not contributing to the production of PS A are shaded. Arrows within the ORFs indicate direction of transcription. (D) G+C content of each of the 9343 PS A locus ORFs.

Creation of deletion mutant 9343PSA. To create mutant 9343PSA, DNA upstream and downstream of the region to be deleted was ligated into the Bacteroides conjugal suicide vector pNJR6 (28). The upstream flank consists of a 4.3-kb EcoRI fragment of pLEC21.2 (the upstream EcoRI site is vector-based), and the downstream flank consists of the 4.2-kb EcoRI insert of pLEC15.1 (Fig. 1A). These EcoRI fragments were cloned into the EcoRI site of pNJR6 in a three-way ligation, and E. coli DH5 was transformed with this ligation mixture. Kanamycin-resistant colonies were screened by PCR for proper orientation of the left and right flanking DNA. Plasmid pMJC15 thus contains both flanks of the 9343 PS A locus ligated in the correct orientation (representing bp 1 through 4350 and 12536 through 16792 of the sequence reported here, for the left and right flanks, respectively).

PCR. The locations of the PCR primers and the DNA amplified by each reaction in relation to the 9343 PS A locus are diagrammed in Fig. 1B. The sequences of the primers and the parameters used for each PCR are shown in Table 2. 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. Each PCR was performed at least twice. Extended PCRs (EPCRs) were carried out as previously described (7).

DNA sequencing and analysis. DNA sequencing and analysis of the B. fragilis PS A region were performed as previously described (6). Comparison of WcfR and WcfS with the S. pneumoniae unfinished microbial genome was performed using TBLASTN with sequence obtained from The Institute for Genomic Research through the website at http://www.tigr.org.

SDS-polyacrylamide gel electrophoresis and immunoblotting. Bacterial cell lysates were run on discontinuous sodium dodecyl sulfate (SDS)-polyacrylamide gels (4 to 20% gradient; ESA, Inc. Chelmsford, Mass.) and transferred to polyvinylidene difluoride Immobilon membranes (Millipore Corp., Bedford, Mass.) by standard techniques. Immunoblotting was performed as described previously (6). Culture supernatants of MAbs G9 (9343 PS B specific [22]) and QUBF7 (B. fragilis 9343 PS C specific [18]) were used at 1:50; ascites fluid containing the PSA-specific MAb CE3 was diluted 1:2,000 before use.

Mouse abscess experiments. A previously described animal model of intra-abdominal infection (26) was modified for these studies. Briefly, male C57BL/6 mice (4 to 6 weeks old; Charles River Laboratories, Wilmington, Mass.) were challenged via the intraperitoneal route with 0.1 ml of inoculum containing 10-fold serial dilutions of 9343 or 9343PSA mixed 1:1 (vol/vol) with sterilized rat fecal contents. Rat fecal contents are used as an adjuvant for abscess formation and do not induce abscesses when implanted alone into the peritonea of mice. Six days after challenge, animals were necropsied and examined for intra-abdominal abscesses. The presence of one or more abscesses in an animal was scored as a positive result. To compare the abscess-inducing potentials of these strains, the AD₅₀ was estimated for each strain using linear logistic regression. The likelihood ratio test was used to test for distinct AD₅₀s for the two strains (8).

Nucleotide sequence accession number. The nucleotide sequence described in this work has been submitted to GenBank and assigned accession number AF189282.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning, characterization, and mutation of the PS A locus. As a first step in determining the contribution of PS A to the abscess-inducing capabilities of B. fragilis 9343, the PS A locus was cloned, sequenced, and mutated. In all, a 16,792-bp region of the 9343 chromosome was sequenced in both directions, using as templates two cosmid clones isolated from the B. fragilis 9343 gene bank (pLEC15 and pLEC21) and subclones derived from them (Fig. 1A).

View larger version (38K): [in this window] [in a new window]   FIG. 2.   Western immunoblot of B. fragilis wild-type and mutant strains. Bacterial cultures were grown to an optical density at 600 nm of 0.8, pelleted by centrifugation, and lysed by boiling in 1× loading buffer. Bacteria equivalent to 40 µl of the original culture were added to each well of the 4-to-20% gradient SDS-polyacrylamide gel. After separation, the samples were transferred to polyvinylidene difluoride membranes and probed with PS A-specific MAb CE3 (A), PS B-specific MAb G9 (B), and PS C-specific MAb QUBF7 (C). Lanes (all panels): 1. B. fragilis 9343; 2. 9343PSA; 3. B. fragilis B16; 4. B. fragilis B16 mutant 65.

Virulence studies. The contribution of PS A to the ability of B. fragilis 9343 to cause abscesses was studied using a mouse abscess model. As outlined in Table 4, the PS A mutant was severely attenuated in its ability to induce abscesses. The wild-type strain consistently induces abscesses in at least 80% of animals at a dose of 10⁸ organisms, with a titrating effect as the dose is lowered. In contrast, the PS A mutant, even at the highest dose, was unable to induce the formation of abscesses in 50% of animals. The AD₅₀ of the wild-type was 10^5.2 organisms, whereas the AD₅₀ of 9343PSA was extrapolated to be 10^14.9 organisms (P < 0.0001).

View this table: [in this window] [in a new window]   TABLE 4.   Abscessogenic potential of B. fragilis 9343 and 9343PSA

Genetic complement of the PS A locus. PS A is a polymer of the repeating unit {3)-d-AATGalp(14)[-d-Galf(13)]-d-GalpNAc(13)-d-Galp(1}, where AATGal is acetamido-amino-2,4,6-trideoxygalactose, and the galactopyranosyl residue is modified by a pyruvate substituent spanning O-4 and O-6 (1). One gene of the PS A locus, wcfR, encodes a product that is highly similar to a series of proteins involved in amino sugar biosynthesis (Table 3). This gene product is likely responsible for transfer of the amino group on the AATGal residue of the PS A repeating unit. As this free amino group is essential for the abscess-inducing capabilities of PS A, wcfR is expected to be crucial for virulence.

Conservation of the PS A region among various B. fragilis strains. MAb analyses revealed that not only does B. fragilis produce multiple capsular polysaccharides, but there is interstrain heterogeneity in serotypes (22). Because of the importance of the PS A molecule for virulence of B. fragilis 9343, the level to which this genetic region is conserved among B. fragilis isolates was investigated. A diverse collection of 50 B. fragilis strains was used, which included strains isolated from the United States, England, and Italy and from a wide spectrum of clinical infections and fecal samples. PCR primer pairs were selected according to the sequence of the 9343 PS A locus such that each primer pair would amplify various portions of this locus or its flanking DNA (Fig. 1B).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The mechanism by which displacement of B. fragilis into the peritoneal cavity results in the formation of abscesses is a subject of intense scrutiny in our laboratory. Though much work has been directed at identifying the biochemical, structural, and immunogenic features of the capsular polysaccharides that account for the unusual propensity of B. fragilis infection to culminate in the formation of abscesses, comparatively few studies have addressed the genetic features underlying this tendency.

The PS A locus is the third capsular polysaccharide biosynthesis region sequenced from strain 9343. Genes involved in conferring charges to the capsules are present in all three of these loci (6, 9). Recent data demonstrated that homologs of genes whose products confer these charge groups are present in all B. fragilis strains tested and suggest that the synthesis of capsular polysaccharides with charged groups is a characteristic of the species B. fragilis (7).

The synthesis of at least three capsular polysaccharides by B. fragilis and the demonstration that purified PS A and PS B each induce abscesses in the animal model suggested that more than one of the capsular polysaccharides were contributing to the organism's ability to induce abscesses. If so, a mutant devoid of multiple capsular polysaccharides might be necessary to produce an attenuated strain. The first mutants created, 9343wcfF and 9343PSB, which are unable to produce PS C and PS B, respectively, supported this contention. These mutants were each found to be fully virulent, with AD₅₀s statistically indistinguishable from that of wild-type 9343 (6, 9). The data presented here demonstrate that the capsular polysaccharides of 9343 do not contribute equally to abscess induction; rather, the synthesis of PS A is essential for B. fragilis 9343 to facilitate abscess formation (Table 4). The 9343PSA mutant strain had an AD₅₀ that was extrapolated to be approximately 10 log units higher than the wild-type strain. These data definitively demonstrate that PS A is the major virulence determinant in the formation of abscesses by B. fragilis 9343.

PCR studies and hybridization analysis of the PS C locus of 50 B. fragilis strains demonstrated that no polysaccharide biosynthesis gene was conserved in the PS C locus of all strains, although the flanking DNA was conserved (7). Therefore, the presence of wcfR and wcfS in the PS A locus of all strains tested was an unexpected finding. Our conclusion that WcfR and WcfS are involved in the synthesis and transfer of the acetamido-amino-2,4,6-trideoxygalactose (AATGal) of PS A is strongly supported by homology data. Close homologs of both of these proteins are encoded by the region responsible for the synthesis of the O17 O antigen of Plesiomonas shigelloides and by the Shigella sonnei form I antigen locus (4) (Table 3). Both of these polysaccharides contain the rare monosaccharide AATGal (15). Homologs of WcfR and WcfS were also detected in several species for which the polysaccharide structures have not been determined. With few exceptions, the WcfR and WcfS homologs are encoded by adjacent genes, suggesting that they participate in a common pathway, as predicted for WcfR and WcfS. The teichoic acid polymer (C substance) of S. pneumoniae also contains AATGal (2, 10). Based on the putative functions of WcfR and WcfS in AATGal biosynthesis and transfer, homologs of these genes should be present in the genome of S. pneumoniae. To test our hypothesis, a search of the unfinished S. pneumoniae type 4 genome (available from www.tigr.org) using the TBLASTN program with WcfR and WcfS as query sequences was performed. This search revealed two adjacent ORFs encoding putative products with similarity to WcfR and WcfS, as predicted (Table 3). These genes are not flanked by other polysaccharide biosynthesis genes and therefore do not appear to be contained in a capsular polysaccharide biosynthesis locus. Since the S. pneumoniae type 4 capsule repeating unit does not contain AATGal (13), we predict that these genes are involved in the synthesis and transfer of the AATGal of the S. pneumoniae teichoic acid. As this polymer is common to this species, these genes are likely contained in the genome of all S. pneumoniae strains, regardless of capsular serotype.

The importance of the free amino group of AATGal to the ability of PS A to cause abscesses is well documented (29-31). The conservation of wcfR and wcfS in the PS A locus of B. fragilis indicates that this unusual monosaccharide is likely present in the PS A repeating unit of this species as a whole. The only other B. fragilis strain for which the structure of the PS A repeating unit has been determined is 638R. We have shown that the PS A locus of 638R is ~24 kb, compared with the ~10.7-kb PS A locus of 9343 (Table 5). As predicted by this large locus, the 638R PS A repeating unit (termed PS A2) is more complex than the PS A of 9343. PS A2 contains five monosaccharides (compared with the four residues in the repeating unit of 9343 PS A [1]), and several of these are deoxymonosaccharides (32). Despite the great structural differences between PS A and PS A2, PS A2 also contains AATGal, as the results presented here predict it should. The possibility that this sugar is conserved throughout the species, as the genetic analyses suggest, has profound biological and pathogenic implications.