Sequence Diversity and Antigenic Variation at the rag Locus of Porphyromonas gingivalis
http://www.100md.com
感染与免疫杂志 2005年第7期
Centre for Infectious Disease, Barts and The London School of Medicine and Dentistry, Queen Mary, University of London, Turner Street, London E1 2AD, United Kingdom
Clinical and Diagnostic Oral Sciences, Barts and The London School of Medicine and Dentistry, Queen Mary, University of London, Turner Street, London E1 2AD, United Kingdom
ABSTRACT
The rag locus of Porphyromonas gingivalis W50 encodes RagA, a predicted tonB-dependent receptor protein, and RagB, a lipoprotein that constitutes an immunodominant outer membrane antigen. The low G+C content of the locus, an association with mobility elements, and an apparent restricted distribution in the species suggested that the locus had arisen by horizontal gene transfer. In the present study, we have demonstrated that there are four divergent alleles of the rag locus. The original rag allele found in W50 was renamed rag-1, while three novel alleles, rag-2 to rag-4, were found in isolates lacking rag-1. The three novel alleles encoded variants of RagA with 63 to 71% amino acid identity to RagA1 and each other and variants of RagB with 43 to 56% amino acid identity. The RagA/B proteins have homology to numerous Bacteroides proteins, including SusC/D, implicated in polysaccharide uptake. Monoclonal and polyclonal antibodies raised against RagB1 of P. gingivalis W50 did not cross-react with proteins from isolates carrying different alleles. In a laboratory collection of 168 isolates, 26% carried rag-1, 36% carried rag-2, 25% carried rag-3, and 14% carried rag-4 (including the type strain, ATCC 33277). Restriction profiles of the locus in different isolates demonstrated polymorphism within each allele, some of which is accounted for by the presence or absence of insertion sequence elements. By reference to a previously published study on virulence in a mouse model (M. L. Laine and A. J. van Winkelhoff, Oral Microbiol. Immunol. 13:322-325, 1998), isolates that caused serious disease in mice were significantly more likely to carry rag-1 than other rag alleles.
INTRODUCTION
Porphyromonas gingivalis is a gram-negative, anaerobic coccobacillus that is strongly associated with destructive periodontal disease. P. gingivalis is frequently isolated in high numbers from severely inflamed periodontal pockets in patients with periodontitis; it is isolated less frequently and in lower numbers from periodontally healthy individuals. Molecular typing studies have suggested that strains of some genotypes may be more commonly associated with disease than others (2, 11, 21, 24). Experiments using animal models and in vitro systems have also indicated that certain strains are more pathogenic than others; the significance of fimbrial genotypes and capsular serotypes has been subject to particular attention (9, 17, 25, 31), but neither accounts fully for the differences in pathogenicity between strains. A greater understanding of the factors governing strain variation in pathogenic potential has implications for the development of improved diagnostic tools and therapeutic strategies.
Analysis of the immunoglobulin G serum antibody response of periodontitis patients has led to the identification of three immunodominant surface antigens (115, 55, and 47 kDa) expressed by P. gingivalis W50 (8). We have previously cloned the gene ragB (receptor antigen gene B), encoding the 55-kDa antigen, and found it to be located immediately downstream of a cotranscribed gene, ragA, coding for a putative tonB-dependent receptor (12). Several factors led to the conclusion that the rag locus has arisen by horizontal gene transfer and may represent a pathogenicity island in P. gingivalis W50: the locus has a G+C content of 42%, compared to a mean value of 48% for the complete genome; it is flanked by an insertion sequence (IS1126, now known as ISPg1) and a sequence with similarity to mobile elements; PCR and Southern blots indicated that the locus has a restricted distribution within the species; and PCR of subgingival samples suggested an association of the locus with deep periodontal pockets (12). The aim of the present study was to investigate the chromosomal location and extent of polymorphism of the rag locus in a diverse collection of isolates of P. gingivalis.
MATERIALS AND METHODS
Bacterial strains and culture conditions. Isolates were grown on fastidious anaerobic agar plates with 5% defibrinated horse blood or brain heart infusion (BHI) broth with 5 μg ml–1 hemin at 37°C in an anaerobic cabinet (Don Whitley Scientific) with an atmosphere of 80% N2, 10% H2, and 10% CO2.
A total of 168 isolates of P. gingivalis from our laboratory collection were investigated, including isolates generously supplied by numerous colleagues. All isolates had been recovered from human sources, mostly from patients with periodontal disease, and were from 15 countries. Isolates were confirmed to be P. gingivalis by PCR with species-specific primers to 16S rRNA genes as described previously (3, 12).
DNA manipulations. Total genomic DNA was isolated from stationary-phase bacteria using a Puregene DNA isolation kit reagent (Flowgen) according to the manufacturer's instructions. Briefly, 1 ml of brain heart infusion cultures grown overnight were lysed with 600 μl cell lysis solution at 80°C for 5 min. Cells were treated with 3 μl RNase A solution at 37°C for 1 h. Protein was removed by precipitation with guanidine thiocyanate, and the DNA was precipitated and cleaned with successive washes of 100% isopropanol and 70% ethanol. Purified DNA was resuspended in 50 μl TE (10 mM Tris, 1 mM EDTA, pH 8.0).
DNA purification to remove primers, enzymes, and other reagents was undertaken using a QIAgen gel extraction kit. Briefly, DNA was bound to a silicon-based ion-exchange matrix under high-salt conditions in Qiagen buffer PB. The DNA was cleaned with two washes of Qiagen buffer PE and eluted in 50 μl TE.
Restriction digestion of genomic DNA and PCR products was performed using Amersham Pharmacia or New England Biolabs enzymes and buffers. Reaction mixtures were incubated at 37°C for at least 2 h. DNA electrophoresis was performed in 0.8% agarose with Tris-borate-EDTA (0.09 M Tris-borate, 0.002 M EDTA). Ethidium bromide-stained gels were viewed under UV light, and the image was captured with a Syngene Imager. Southern hybridizations were performed by standard methods, with DNA immobilized on HyBond N+ (Amersham Pharmacia). PCR products used as probes were labeled with digoxigenin and detected by color precipitate using a DIG labeling and hybridization kit (Roche). Hybridization was at 65°C, and final washes were in 0.2x SSC (1x SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% sodium dodecyl sulfate for 15 min at 65°C.
PCR. All amplification reactions were performed by using an Omnigene Life Sciences International thermal cycler. Standard PCRs used 50-μl volumes with 0.5 μg each primer and 0.5 μl chromosomal DNA template in ABgene PCR Reddy Load Master Mix (ABgene) with 1.5 mM MgCl2. Amplification of longer products was performed with ABgene Extensor Long Master Mix 2 (ABgene).
The following primers (with positions of primers and products indicated in Fig. 1) were used: PG0183F (forward) (CAA CTA CCT TCA CTT CCC), PG0183R (reverse) (AAC CTT CAA TCT CCT CCA C), ISPg1F (GGC TCC CCC TTG ATT TTT CTC), ISPg1R (AAG GAG GTG GAG GGA AGA GGA) W83 ragAF (TGT CTG GTG CGA TAG CGA ATA G), W83 ragAR (TAC ATA GGT GAG TTT GAG ATT C), W83 ragBF (AAT ACT GAA AAT CCA CGA), W83 ragBR (TAG GGG CTG CGA CAA AAA), ELPCRF (TAT CGT TTC GCC CAG CCA AGA), ELPCRR (TCC TTC GCA TCG CTC TTT CCC), LPCRF (CAA AGT CCT GCC ACG AGT AGC), and LPCRR (CGT TTT CTC GCC ACT TTC GTC). Primers designed to be specific to ragB from each of the four rag alleles were rag1F (CGC GAC CCC GAA GGA AAA GAT T) and rag1R (CAC GGC TCA CAT AAA GAA CGC T) (positions 70 to 697 of ragB1), rag2F (GCT TTG CCG CTT GTG ACT TGG) and rag2R (CCA CCG TCA CCG TTC ACC TTG) (positions 56 to 1034 of ragB2), rag3F (CCG GAA GAT AAG GCC AAG AAA GA) and rag3R (ACG CCA ATT CGC CAA AGC T) (positions 76 to 498 of ragB3), and rag4F (CCG GAT GGA AGT GAT GAA CAG A) and rag4R (CGC GGT AAA CCT CAG CAA ATT) (positions 82 to 820 of ragB4).
Standard programs for PCR comprised 30 cycles of denaturation for 1 min at 95°C, annealing for 1 min at 50°C, and extension for 1 to 3 min at 72°C. The annealing temperature was changed to 55°C for ragA primers and 60°C for ISPg1 primers. For extra-long PCR (ELPCR), the annealing temperature was 60°C, the extension time was 20 min, and 25 cycles were performed. For long PCR (LPCR), 10 cycles with annealing at 52°C and an extension time of 6 min 40 s at 68°C were followed by 15 cycles with annealing at 60°C and a 7-min extension step at 68°C.
Pulsed-field gel electrophoresis (PFGE). Bacterial growth from 4-day-old cultures grown on fastidious anaerobic agar was washed and resuspended in 0.5 ml of PIV buffer (10 mM Tris [pH 7.6], 1 M NaCl). The cell suspension was added to 0.5 ml molten 2% CHEF agarose (Bio-Rad) at 50°C, mixed, and gently pipetted into a CHEF plug mold (Bio-Rad). Plugs were incubated at 37°C in 2 ml lysis buffer (6 mM Tris [pH 6.7], 1 M NaCl, 100 mM EDTA, 0.5% Brij 58 solution, 0.2% deoxycholate, 0.5% N-lauroylsarcocine, 20 μg/ml RNase A, 100 μg/ml lysozyme) for 2 h. Plugs were washed in 2 ml TE (CHEF) (10 mM Tris [pH 7.5], 0.1 mM EDTA) at 50°C for 5 min and then incubated in 2 ml ESP buffer (0.5 M EDTA [pH 9 to 9.5], 1% N-lauroylsarcosine, 50 μg/ml proteinase K) at 50°C for 2 h and then washed twice in 2 ml TE (CHEF) at 37°C for 30 min. Plugs were stored in TE at 4°C in the dark. For restriction, plug slices approximately 1 mm thick were taken from three separate blocks and placed into 1.5-ml microcentrifuge tubes and equilibrated in restriction buffer, and then DNA was digested overnight at 37°C in 200 μl buffer plus XbaI. Electrophoresis was performed with a CHEF DRII system (Bio-Rad) for 19 h at 6 V cm–1, with a switch time of 1 to 30 s and with a 1% agarose gel in 0.5x Tris-borate-EDTA buffer at 15°C.
Antiserum production and Western blotting. Preparation of extracts, electrophoresis on 10% polyacrylamide gels, and detection of the 55-kDa outer membrane antigen RagB with monoclonal antibodies B15 and E38 were performed as previously described (20). Polyclonal antiserum to recombinant W50 RagB, synthesized as previously described (29), was produced in rabbits. Pooled dialyzed fractions of recombinant protein in phosphate-buffered saline (0.7 mg/ml) were incubated with a fresh precipitate of alum. The precipitate was washed and resuspended in saline for injection. Primary immunization was with 175 μg protein subcutaneously over four sites. Two more immunizations of 100 μg were given 3 weeks apart. Blood was withdrawn and separated for antiserum after an additional 3 weeks.
DNA sequence determination. Long PCR products from A011/9, QM220, and ATCC 33277T (strain obtained as NCTC11834T) were purified and sent to MWG-Biotech for sequence determination using their publication-grade sequencing service. Open reading frames were identified with FramePlot 2.3.2 online software (13).
Nucleotide sequence accession numbers. The nucleotide sequence data reported in this study have been assigned GenBank accession numbers AY842852 (NCTC11834T), AY842853 (QM220), and AY842854 (A011/9).
RESULTS AND DISCUSSION
Detection of rag genes and genomic location in diverse isolates. We previously proposed that P. gingivalis W50 had acquired the rag locus by horizontal gene transfer (12), raising the possibility that the site of insertion could differ between isolates. We had also demonstrated that the rag locus of W50 was not present in all isolates. In order to identify isolates that were genetically diverse for further investigation of the locus, we first typed 42 isolates originating from several countries. PFGE identified 13 different types (A to M) among 37 isolates (Table 1). Four types were shared by two or more isolates, while nine were unique. Five isolates consistently failed to yield readable band patterns; 23A4 was retained for further investigation, as this is a well-characterized isolate, but the other four untyped isolates were not studied further.
The presence of genes corresponding to ragA and ragB of W83/W50 was investigated by PCR using primers located as illustrated in Fig. 1. In 20 isolates of nine PFGE types, both genes were detected, while in 18 isolates (including the type strain) of four PFGE types or that were untyped, ragA and ragB PCR failed to yield a product (Table 1). Southern hybridization of chromosomal DNA with probes comprising PCR products from W50 ragA and ragB (Fig. 1) confirmed the PCR results in all cases. PCR with primers in the gene upstream of the rag locus in W83, PG0183 (separated from ragA by ISPg1 [Fig. 1]), was also negative in 5 of the 18 rag-negative isolates. In all cases, identical PCR results were obtained for isolates having the same PFGE type (Table 1).
Western blots were performed on representative isolates of rag-positive PFGE types A, B, C, D, and I and rag-negative PFGE types F, G, and J plus untyped isolate 23A4 by using monoclonal antibodies (MAbs) specific to RagB (MAb B15 and MAb E38). The 55-kDa RagB was detected in all rag-positive isolates and was absent in all rag-negative isolates.
Publication of the complete genome sequence of P. gingivalis W83 allowed examination of the region surrounding the locus (26). PCR was performed on representatives of all PFGE types with primers from sequences in PG0183, upstream of the rag locus in the sequenced isolate W83, and PG0190 (uppS), downstream of the locus (ELPCR in Fig. 1). A PCR product of similar length (13.5 to 15 kb, established by adding the sizes of fragments after digestion with BglII and NspI) was obtained with isolates of the nine rag-positive PFGE types and a ragB probe hybridized to a 2.5-kb BclI fragment from the PCR product in all cases. This indicated that the rag locus was in the same genomic location (i.e., flanked by the same genes) in diverse isolates. Variations in the length of the long PCR product and in the sizes of restriction fragments suggested that there was nevertheless some polymorphism between isolates in this region.
Polymorphism among rag-positive isolates. Polymorphism in the rag locus of isolates representing the nine rag-positive PFGE types was investigated further. Long PCR products spanning the rag locus (ELPCR, PG0183 to PG0190), as described above, were digested with NspI and BglII. Fragments were identified by size comparison to those of W83 and by hybridization with probes from ragA, ragB, and a fragment overlapping ISPg1. All isolates with PFGE types other than type A had a deletion of 1.3 kb between PG0183 and ragA, relative to W50 and W83. The restriction fragments obtained were compatible with the absence of ISPg1 in these isolates. Isolate LB13D3 had an insertion of approximately 1 kb in the intergenic region between ragB and PG0188, and a number of restriction sites were absent in individual isolates, suggesting a degree of sequence polymorphism in the locus.
The locus in rag-negative isolates. The same long PCR primers (ELPCR) were used to amplify DNA from isolates in which ragA and ragB gave negative results but PG0183 PCR was positive. In the absence of ragA and ragB, a PCR product of no more than 10 kb would be predicted; unexpectedly, products of 12.6 to 14 kb were obtained. A simple insertion/deletion event was therefore unlikely to account for the difference between these rag-negative isolates and rag-positive isolates. Southern hybridization of digested ELPCR products with W50 ragA and ragB probes again confirmed the absence of sequences with homology to these genes, except for isolate YH522, in which a weak but distinct hybridization signal to a 6.6-kb BglII fragment was detected with the W50 ragA probe.
An alternative primer pair was designed to amplify the rag locus with less additional flanking sequence by using primers closer to the 3' end of PG0183 and, in PG0188, the gene immediately downstream of ragB (LPCR in Fig. 1). These should result in an 8.35-kb product in W83. Amplification with the alternative primer pair yielded fragments of 6 to 8.5 kb in isolates representative of all PFGE types, including those that had been negative for PG0183 by PCR. Again, this suggested that the locus had been replaced rather than deleted in isolates that both PCR and Southern blotting had indicated did not contain W50 ragA and ragB. It also suggested that the failure to detect PG0183 by PCR in some isolates might be due to local sequence polymorphism rather than absence of the gene.
Novel variants of the rag locus. Digestion of LPCR products spanning the region between PG0183 and PG0188 in rag-negative isolates yielded restriction fragments that were of very different sizes from rag-positive isolates. To determine the extent of polymorphism in this region, a much larger collection of isolates was screened by digestion of the LPCR product with NcoI, SacI, and SphI. From a total of 168 isolates, 15 different restriction profiles were obtained; for two isolates, no specific amplification product could be obtained. The profiles fell into four main groups on the basis of common restriction fragments, designated 1 to 4; within each group, profiles differed by up to three bands, consistent with a single polymorphism, and were designated with a lowercase letter (1a, 1b, etc.) (Table 2). The profiles most frequently observed were 2b (43 isolates), 3a (40 isolates), 1a (25 isolates), and 1b, 2a, and 4a (14 isolates each) (Fig. 2). W83 and W50 both had profile 1a; profile 1b was highly related and was observed in isolates for which the rag locus had already been shown to differ from W50 only in the absence of ISPg1.
The LPCR product extending from the 3' end of PG0183 to the 5' end of PG0188 was sequenced in isolates A011/9 (profile 2a), QM220 (profile 3a), and ATCC 33277T (profile 4a) in order to find out what genes had replaced ragA and ragB of W50/W83. Analysis of open reading frames revealed that each of the three isolates carried different novel genes with homology to ragA and ragB of W50/W83. Sites of PCR primers used to detect ragA and ragB were absent or divergent in each isolate, and the regions corresponding to probes were less than 60% identical to the corresponding genes in W50/W83, accounting for the failure to detect rag genes in these isolates.
Between PG0183 and ragA in A011/9 was a gene with 98% nucleotide homology to the transposase of ISPg3 (formerly IS195), an insertion sequence found previously in P. gingivalis (18). Examination of the flanking sequences identified a 10-bp direct repeat element (TTACATCACA), possibly a target for insertion. On the basis of restriction analysis described above, it could be deduced that isolates with LPCR restriction profile 2b carried the same rag locus as A011/9 but lacked ISPg3.
In ATCC 33277T, an additional open reading frame was found downstream of ragB. The protein encoded had no significant homology to known proteins, except for a C-terminal 80-amino-acid sequence with 50% similarity to the C-terminal domain shared by gingipain precursor proteins and hemagglutinins of P. gingivalis (7).
To distinguish the different rag loci, the following designations were assigned: rag-1, representing the locus found in W50 and W83; rag-2, representing the locus of A011/9; rag-3, representing the locus of QM220; and rag-4, representing the locus of ATCC 33277T. PCR primers were designed to specifically amplify ragB of each of the four rag loci and distinguish between them (further details of this PCR assay and its use in clinical material will be described in a separate publication). When applied to the full collection of isolates, PCR confirmed the groups assigned by LPCR restriction profile (Table 2); rag-3 was detected in the two isolates for which LPCR was unsuccessful.
Antigenic variation and expression. Western blots of isolates with different PFGE types had already demonstrated that two monoclonal antibodies recognizing RagB1 from W50 did not cross-react with proteins from isolates lacking the rag-1 locus. To further investigate antigenic cross-reactivity between the different variants of RagB, polyclonal antiserum was raised against recombinant RagB1 from W50. Western blots with representative isolates carrying the four alleles of the rag locus demonstrated that neither the monoclonal antibodies nor the polyclonal antiserum to RagB1 were able to recognize RagB2, RagB3, or RagB4 (Fig. 3).
In a previous study, we had investigated outer membrane proteins in strains W50, W83 (both rag-1), and NCTC11834T (rag-4) (8). Retrospective examination of the published gel images demonstrates that while W50 and W83 both produced a prominent band at 55 kDa, which later proved to represent RagB, at least one protein of similar molecular weight but which was apparently less abundant was also expressed by NCTC11834T. In the absence of a specific antibody to Rag4, it was not possible to confirm the identity of this protein. However, Murakami et al. recently determined by N-terminal sequencing that both RagA and a variant of RagB that can now be confirmed to be RagB4 are expressed in ATCC 33277T (rag-4), where they represent two of the seven major outer membrane proteins (22, 23).
Sequence comparisons of the four rag alleles. Multiple alignment of the four RagA proteins using ClustalW (32) demonstrated a high level of sequence homology at the N terminus of the proteins, with only one amino acid polymorphism in the first 150 residues (Fig. 4). Through the rest of the protein, there were local regions of high homology interspersed with regions with little sequence conservation. The RagB proteins were more diverse, with only short regions of sequence conservation (Fig. 5). Alignment of the DNA sequences revealed that the region between PG0183 and ragA was highly conserved between isolates, apart from the presence of insertion sequences in W50/W83 and A011/9. Only 29 (3.1%) of 939 nucleotides in this intergenic region were found to be polymorphic. Conversely, there was no significant sequence homology detected between any of the isolates in the intergenic region between ragB and PG0188, apart from the last 46 nucleotides upstream of PG0188 (Fig. 1).
All four variants of RagA and RagB were submitted to the NCBI Conserved Domain Search and BLAST analysis (1, 19). Homology to the N-terminal region of conserved domains "ligand_gated_channel, TonB dependent (cd01347)," "TonB_dep_rec, TonB dependent receptor (pfam00593)," "FepA, outer membrane receptor for ferrienterochelin and colicins (COG4771)," "CirA, outer membrane receptor proteins, mostly Fe transport (COG1629)," and "BtuB, outer membrane cobalamin receptor protein" was detected from approximately amino acids 100 to 250 of all RagA variants. Further homology to "ligand_gated_channel, TonB_dep_rec" and CirA was also detected over the less conserved part of the protein, especially amino acids 500 to 800 of all variants. No conserved domains were detected in the RagB proteins.
All RagA variants demonstrated extensive homology to members of the large families of putative outer membrane proteins (over 50 in each) from the complete genome sequences of Bacteroides thetaiotaomicron (35) and Bacteroides fragilis (15). Many of these outer membrane protein genes have a putative lipoprotein gene immediately downstream, in an arrangement analogous to the rag locus, and RagB in turn had significant homology to some of these lipoproteins. For all Rag variants, the homology to other P. gingivalis Rag protein variants was always much greater than the homology to the most closely related Bacteroides protein (Tables 3 and 4); the same B. thetaiotaomicron locus ranked as the highest score for each of the RagA and RagB variants.
The most similar protein to RagA with an experimentally proven function was SusC (27), involved in starch uptake in B. thetaiotaomicron, as noted previously (12) (Tables 3 and 4). SusD, a lipoprotein functionally associated with SusC and encoded immediately downstream (28) had significant similarity to RagB2 over a short region, but similarity to the other RagB variants was scored as nonsignificant by standard BLASTP analysis. By analogy to SusC/D, the families of related outer membrane proteins and lipoproteins in both Bacteroides species have been interpreted as systems for the binding and uptake of a wide range of polysaccharides from their environment in the large intestine (15, 35). Circumstantial support for this role comes from the finding of susCD/ragAB-related genes on a plasmid required for agar degradation by a marine bacterium identified as a Microscilla species (36). The occurrence of such a function in P. gingivalis is surprising given that this species is asaccharolytic, requiring protein as a carbon source.
The RagA sequences were submitted to the web-based program for protein fold recognition, 3D-PSSM (14). In accordance with the results of the Conserved Domain Search, all four RagA variants had the greatest structural homology (>95% confidence level) to the TonB-dependent outer membrane transporter proteins FepA, FhuA, and BtuB from Escherichia coli. FepA, FhuA, and BtuB all form 22-stranded antiparallel -barrels, with an N-terminal domain that forms a plug within the barrel and interacts with TonB (5, 6, 10). Alignment of the sequences of the four RagA variants, the B. thetaiotaomicron homologue BT0206, and SusC demonstrated strong conservation in the position of predicted barrel region -strands despite extensive sequence divergence (Fig. 4). (The structural similarity of RagA1 to FepA had previously been noted by Wexler et al. [34], who have characterized the locus encoding the omp200 porin complex of Bacteroides fragilis and demonstrated a gene organization similar to that of the rag locus.)
Analysis of the RagB sequences by 3D-PSSM revealed a predicted similarity to the alpha/alpha-6 barrel structure of 1AYX, a glucoamylase of yeast belonging to a sequence family 15 of glycosyl hydrolases (30), although the significance score was over 95% only for RagB4. This prediction was intriguing given the role of the Sus operon in polysaccharide degradation and uptake (27, 28). Sequence alignment of the RagBs, BT0207 (the putative lipoprotein encoded downstream of BT0206), and SusD demonstrated close correspondence in the positions of 11 predicted alpha helices (Fig. 5), but the location and spacing of alpha helices did not match well to those of 1AYX (not shown).
Distribution of the rag alleles. The collection of P. gingivalis investigated here comprises isolates from many sources, including well-characterized laboratory strains as well as more recent clinical isolates. A majority of isolates are from western European countries, with few from North America. The collection may not be fully representative of the P. gingivalis population currently circulating, nor is it possible to rule out that some isolates may be duplicates either from the same patient or from laboratory strains that have been renamed by different laboratories. Nevertheless, some general comments may be made about the prevalence and geographic distribution of the four rag alleles.
All four of the rag alleles were found among isolates from several countries; none was restricted to isolates from a single continent. Their relative prevalence in this collection was comparable, except that rag-4 was found less frequently than the other alleles. rag-1 was detected in isolates obtained from Germany, The Netherlands, Romania, Sweden, the United Kingdom, and Kenya. However, LPCR restriction profile 1a was frequently found in established laboratory strains such as W50 and W83, some of which may have a common origin, while profile 1b (lacking ISPg1) was more frequent among isolates thought to have a recent clinical source. Profile 2a (rag-2) was found exclusively among a set of isolates from Thailand, but profile 2b, representing the rag-2 locus without ISPg3, was found in isolates from across Europe (Finland, The Netherlands, Sweden, and the United Kingdom). rag-3 was found in numerous isolates from Europe and one isolate each from Japan and Thailand. Profile 4a (rag-4) was represented by strains ATCC 33277T and 381, both of which are long-established laboratory strains, and in isolates from European countries.
Relationship of rag alleles to other markers of disease and virulence. The distribution of rag alleles was investigated relative to other polymorphic markers that have been used to type P. gingivalis and determine associations with disease (Table 5). A set of isolates kindly provided by M. L. Laine representing two to three isolates each of the six K-capsular serotypes (K1 to K6) plus unencapsulated types (K0) of P. gingivalis (16, 33) was included in the analysis. There was a limited degree of correlation between the rag allele and K serotype: K3 and K5 were found only in rag-3 isolates, and K4 was found only in rag-1 isolates, but K1, K2, and K0 were each associated with two different rag alleles; each rag allele was found in isolates of at least two different K serotypes (Table 5). The fimA genotype was known for eight isolates (25) and showed no clear association with the rag allele. Genotypes fimA II and Ib have been reported to be associated with periodontitis and recovery from deep pockets in epidemiological studies in Asia (2, 24), although the same association was not found in a study in Europe (4); these genotypes were found among isolates with three different rag alleles. The ribosomal intergenic spacer heteroduplex types were too diverse among the few isolates for which they were known to draw any conclusions, but again, types with an epidemiological association with periodontitis (11) were found in isolates of both rag-1 and rag-2. We had previously demonstrated the association of the rag locus, corresponding to rag-1 in this study, with deep pockets in patients with periodontal disease (12). An epidemiological study comparing these various markers in isolates from patients with detailed clinical assessment is required to evaluate their relative strength of association with disease. Such a study would have the potential to indicate the significance of the three polymorphic surface antigens, RagAB, capsular polysaccharide, and fimbriae, to the outcome of P. gingivalis infection.
An alternative strategy to assess the significance of putative virulence factors is to use an animal model of infection, and Laine and van Winkelhoff (17) have reported the virulence of isolates with different serotypes in a mouse model. Encapsulated strains were significantly more virulent than K0 strains, but variability in virulence between strains with the same serotype suggested the importance of additional virulence factors that must differ between strains. A measure of virulence from the Laine and van Winkelhoff study has been included in Table 5, and it was apparent that most of the rag-1 isolates had high scores for virulence. Chi-square analysis demonstrated that there was a significant correlation (P < 0.05) between carriage of rag-1 (versus carriage of any other rag allele) and high virulence (2/3 or 3/3 mice seriously ill within 72 h of inoculation).
Further discussion. In our initial investigation of the rag locus of P. gingivalis, it was concluded that the genes encoding this immunodominant surface antigen had a restricted distribution in the species (12). The present study has demonstrated that there are actually four distinct variants or alleles of the rag locus, each of which is common among the P. gingivalis population and none of which appear to be confined to specific geographic regions.
Although the function of RagA/B has not yet been experimentally determined, the similarity to SusC/D family proteins in Bacteroides species suggests a possible role in polysaccharide uptake. Expression of some of the SusC/D loci in B. fragilis has been found to be controlled by DNA inversion systems, including three clusters of loci where inversion of a promoter mediates alternative expression of different SusC/D cassettes (15). Other DNA inversion systems in B. fragilis affect the expression of other surface-expressed molecules such as capsular polysaccharides. These systems are likely to represent a mechanism for the high-frequency generation of antigenic variation and are thus implicated in evasion of the host response (15). While there is no evidence for high-frequency variation of RagA/B, the rag locus bears many of the hallmarks characteristic of genes involved in the production of variable surface antigens. These features include an association of insertion elements with the locus and a G+C content lower than what is typical of the species and are generally interpreted to suggest that the locus and its variant alleles have been acquired by horizontal gene transfer. The large number of genes in Bacteroides spp. with sequence homology to the rag locus would suggest a ready reservoir of genes available for recombination, although none of the genes available to date in sequence databases are sufficiently similar to suggest a close ancestry in common with the P. gingivalis locus.
The presence of variant alleles of the rag locus leads directly to the question of whether specific alleles are associated with different clinical conditions. We had previously demonstrated that rag-1 was associated with deep periodontal pockets (12). The present study has revealed that strains shown by others to be more virulent in a mouse model (17) are significantly more likely to carry rag-1 than other rag alleles. Further work is in progress to investigate the links between the rag locus and pathogenesis in more detail.
ACKNOWLEDGMENTS
This work was funded by development grant G9900398 and program grant PG9318173 from the Medical Research Council (United Kingdom) and a Ph.D. studentship (for S.C.F.) from the Special Trustees of the Royal London Hospital.
We thank Jenny Slaney for her assistance with immunological techniques.
REFERENCES
1. Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang, W. Miller, and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25:3389-3402.
2. Amano, A., I. Nakagawa, K. Kataoka, I. Morisaki, and S. Hamada. 1999. Distribution of Porphyromonas gingivalis strains with fimA genotypes in periodontitis patients. J. Clin. Microbiol. 37:1426-1430.
3. Ashimoto, A., C. Chen, I. Bakker, and J. Slots. 1996. Polymerase chain reaction detection of 8 putative periodontal pathogens in subgingival plaque of gingivitis and advanced periodontitis lesions. Oral Microbiol. Immunol. 11:266-273.
4. Beikler, T., U. Peters, S. Prajaneh, K. Prior, B. Ehmke, and T. F. Flemmig. 2003. Prevalence of Porphyromonas gingivalis fimA genotypes in Caucasians. Eur. J. Oral Sci. 111:390-394.
5. Buchanan, S. K., B. S. Smith, L. Venkatramani, D. Xia, L. Esser, M. Palnitkar, R. Chakraborty, D. van der Helm, and J. Deisenhofer. 1999. Crystal structure of the outer membrane active transporter FepA from Escherichia coli. Nat. Struct. Biol. 6:56-63.
6. Chimento, D. P., A. K. Mohanty, R. J. Kadner, and M. C. Wiener. 2003. Substrate-induced transmembrane signaling in the cobalamin transporter BtuB. Nat. Struct. Biol. 10:394-401.
7. Curtis, M. A., H. K. Kuramitsu, M. Lantz, F. L. Macrina, K. Nakayama, J. Potempa, E. C. Reynolds, and J. Aduse-Opoku. 1999. Molecular genetics and nomenclature of proteases of Porphyromonas gingivalis. J. Periodont. Res. 34:464-472.
8. Curtis, M. A., J. M. Slaney, R. J. Carman, and N. W. Johnson. 1991. Identification of the major surface protein antigens of Porphyromonas gingivalis using IgG antibody reactivity of periodontal case-control serum. Oral Microbiol. Immunol. 6:321-326.
9. Dierickx, K., M. Pauwels, M. L. Laine, J. Van Eldere, J. J. Cassiman, A. J. van Winkelhoff, D. van Steenberghe, and M. Quirynen. 2003. Adhesion of Porphyromonas gingivalis serotypes to pocket epithelium. J. Periodontol. 74:844-848.
10. Ferguson, A. D., E. Hofmann, J. W. Coulton, K. Diederichs, and W. Welte. 1998. Siderophore-mediated iron transport: crystal structure of FhuA with bound lipopolysaccharide. Science 282:2215-2220.
11. Griffen, A. L., S. R. Lyons, M. R. Becker, M. L. Moeschberger, and E. J. Leys. 1999. Porphyromonas gingivalis strain variability and periodontitis. J. Clin. Microbiol. 37:4028-4033.
12. Hanley, S. A., J. Aduse-Opoku, and M. A. Curtis. 1999. A 55-kilodalton immunodominant antigen of Porphyromonas gingivalis W50 has arisen via horizontal gene transfer. Infect. Immun. 67:1157-1171.
13. Ishikawa, J., and K. Hotta. 1999. FramePlot: a new implementation of the frame analysis for predicting protein-coding regions in bacterial DNA with a high G+C content. FEMS Microbiol. Lett. 174:251-253.
14. Kelley, L. A., R. M. MacCallum, and M. J. Sternberg. 2000. Enhanced genome annotation using structural profiles in the program 3D-PSSM. J. Mol. Biol. 299:499-520.
15. Kuwahara, T., A. Yamashita, H. Hirakawa, H. Nakayama, H. Toh, N. Okada, S. Kuhara, M. Hattori, T. Hayashi, and Y. Ohnishi. 2004. Genomic analysis of Bacteroides fragilis reveals extensive DNA inversions regulating cell surface adaptation. Proc. Natl. Acad. Sci. USA 101:14919-14924.
16. Laine, M. L., B. J. Appelmelk, and A. J. van Winkelhoff. 1996. Novel polysaccharide capsular serotypes in Porphyromonas gingivalis. J. Periodont. Res. 31:278-284.
17. Laine, M. L., and A. J. van Winkelhoff. 1998. Virulence of six capsular serotypes of Porphyromonas gingivalis in a mouse model. Oral Microbiol. Immunol. 13:322-325.
18. Lewis, J. P., and F. L. Macrina. 1998. IS195, an insertion sequence-like element associated with protease genes in Porphyromonas gingivalis. Infect. Immun. 66:3035-3042.
19. Marchler-Bauer, A., J. B. Anderson, C. DeWeese-Scott, N. D. Fedorova, L. Y. Geer, S. He, D. I. Hurwitz, J. D. Jackson, A. R. Jacobs, C. J. Lanczycki, C. A. Liebert, C. Liu, T. Madej, G. H. Marchler, R. Mazumder, A. N. Nikolskaya, A. R. Panchenko, B. S. Rao, B. A. Shoemaker, V. Simonyan, J. S. Song, P. A. Thiessen, S. Vasudevan, Y. Wang, R. A. Yamashita, J. J. Yin, and S. H. Bryant. 2003. CDD: a curated Entrez database of conserved domain alignments. Nucleic Acids Res. 31:383-387.
20. Millar, D. J., E. E. Scott, J. M. Slaney, S. U, P. Benjamin, and M. A. Curtis. 1993. Production and characterisation of monoclonal antibodies to the principle sonicate antigens of Porphyromonas gingivalis W50. FEMS Immunol. Med. Microbiol. 7:211-222.
21. Missailidis, C. G., J. E. Umeda, C. Ota-Tsuzuki, D. Anzai, and M. P. Mayer. 2004. Distribution of fimA genotypes of Porphyromonas gingivalis in subjects with various periodontal conditions. Oral Microbiol. Immunol. 19:224-229.
22. Murakami, Y., M. Imai, Y. Mukai, S. Ichihara, H. Nakamura, and F. Yoshimura. 2004. Effects of various culture environments on expression of major outer membrane proteins from Porphyromonas gingivalis. FEMS Microbiol. Lett. 230:159-165.
23. Murakami, Y., M. Imai, H. Nakamura, and F. Yoshimura. 2002. Separation of the outer membrane and identification of major outer membrane proteins from Porphyromonas gingivalis. Eur. J. Oral Sci. 110:157-162.
24. Nakagawa, I., A. Amano, Y. Ohara-Nemoto, N. Endoh, I. Morisaki, S. Kimura, S. Kawabata, and S. Hamada. 2002. Identification of a new variant of fimA gene of Porphyromonas gingivalis and its distribution in adults and disabled populations with periodontitis. J. Periodont. Res. 37:425-432.
25. Nakano, K., M. Kuboniwa, I. Nakagawa, T. Yamamura, R. Nomura, N. Okahashi, T. Ooshima, and A. Amano. 2004. Comparison of inflammatory changes caused by Porphyromonas gingivalis with distinct fimA genotypes in a mouse abscess model. Oral Microbiol. Immunol. 19:205-209.
26. Nelson, K. E., R. D. Fleischmann, R. T. DeBoy, I. T. Paulsen, D. E. Fouts, J. A. Eisen, S. C. Daugherty, R. J. Dodson, A. S. Durkin, M. Gwinn, D. H. Haft, J. F. Kolonay, W. C. Nelson, T. Mason, L. Tallon, J. Gray, D. Granger, H. Tettelin, H. Dong, J. L. Galvin, M. J. Duncan, F. E. Dewhirst, and C. M. Fraser. 2003. Complete genome sequence of the oral pathogenic bacterium Porphyromonas gingivalis strain W83. J. Bacteriol. 185:5591-5601.
27. Reeves, A. R., J. N. D'Elia, J. Frias, and A. A. Salyers. 1996. A Bacteroides thetaiotaomicron outer membrane protein that is essential for utilization of maltooligosaccharides and starch. J. Bacteriol. 178:823-830.
28. Reeves, A. R., G. R. Wang, and A. A. Salyers. 1997. Characterization of four outer membrane proteins that play a role in utilization of starch by Bacteroides thetaiotaomicron. J. Bacteriol. 179:643-649.
29. Scragg, M. A., A. Alsam, M. Rangarajan, J. M. Slaney, P. Shepherd, D. M. Williams, and M. A. Curtis. 2002. Nuclear targeting of Porphyromonas gingivalis W50 protease in epithelial cells. Infect. Immun. 70:5740-5750.
30. Sevcik, J., A. Solovicova, E. Hostinova, J. Gasperik, K. S. Wilson, and Z. Dauter. 1998. Structure of glucoamylase from Saccharomycopsis fibuligera at 1.7 A resolution. Acta Crystallogr. D Biol. Crystallogr. 54:854-866.
31. Sugano, N., K. Ikeda, M. Oshikawa, Y. Sawamoto, H. Tanaka, and K. Ito. 2004. Differential cytokine induction by two types of Porphyromonas gingivalis. Oral Microbiol. Immunol. 19:121-123.
32. Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680.
33. van Winkelhoff, A. J., B. J. Appelmelk, N. Kippuw, and J. de Graaff. 1993. K-antigens in Porphyromonas gingivalis are associated with virulence. Oral Microbiol. Immunol. 8:259-265.
34. Wexler, H. M., E. K. Read, and T. J. Tomzynski. 2002. Characterization of omp200, a porin gene complex from Bacteroides fragilis: omp121 and omp71, gene sequence, deduced amino acid sequences and predictions of porin structure. Gene 283:95-105.
35. Xu, J., M. K. Bjursell, J. Himrod, S. Deng, L. K. Carmichael, H. C. Chiang, L. V. Hooper, and J. I. Gordon. 2003. A genomic view of the human-Bacteroides thetaiotaomicron symbiosis. Science 299:2074-2076.
36. Zhong, Z., A. Toukdarian, D. Helinski, V. Knauf, S. Sykes, J. E. Wilkinson, C. O'Bryne, T. Shea, C. DeLoughery, and R. Caspi. 2001. Sequence analysis of a 101-kilobase plasmid required for agar degradation by a Microscilla isolate. Appl. Environ. Microbiol. 67:5771-5779.(Lucinda M. C. Hall, Stuar)
Clinical and Diagnostic Oral Sciences, Barts and The London School of Medicine and Dentistry, Queen Mary, University of London, Turner Street, London E1 2AD, United Kingdom
ABSTRACT
The rag locus of Porphyromonas gingivalis W50 encodes RagA, a predicted tonB-dependent receptor protein, and RagB, a lipoprotein that constitutes an immunodominant outer membrane antigen. The low G+C content of the locus, an association with mobility elements, and an apparent restricted distribution in the species suggested that the locus had arisen by horizontal gene transfer. In the present study, we have demonstrated that there are four divergent alleles of the rag locus. The original rag allele found in W50 was renamed rag-1, while three novel alleles, rag-2 to rag-4, were found in isolates lacking rag-1. The three novel alleles encoded variants of RagA with 63 to 71% amino acid identity to RagA1 and each other and variants of RagB with 43 to 56% amino acid identity. The RagA/B proteins have homology to numerous Bacteroides proteins, including SusC/D, implicated in polysaccharide uptake. Monoclonal and polyclonal antibodies raised against RagB1 of P. gingivalis W50 did not cross-react with proteins from isolates carrying different alleles. In a laboratory collection of 168 isolates, 26% carried rag-1, 36% carried rag-2, 25% carried rag-3, and 14% carried rag-4 (including the type strain, ATCC 33277). Restriction profiles of the locus in different isolates demonstrated polymorphism within each allele, some of which is accounted for by the presence or absence of insertion sequence elements. By reference to a previously published study on virulence in a mouse model (M. L. Laine and A. J. van Winkelhoff, Oral Microbiol. Immunol. 13:322-325, 1998), isolates that caused serious disease in mice were significantly more likely to carry rag-1 than other rag alleles.
INTRODUCTION
Porphyromonas gingivalis is a gram-negative, anaerobic coccobacillus that is strongly associated with destructive periodontal disease. P. gingivalis is frequently isolated in high numbers from severely inflamed periodontal pockets in patients with periodontitis; it is isolated less frequently and in lower numbers from periodontally healthy individuals. Molecular typing studies have suggested that strains of some genotypes may be more commonly associated with disease than others (2, 11, 21, 24). Experiments using animal models and in vitro systems have also indicated that certain strains are more pathogenic than others; the significance of fimbrial genotypes and capsular serotypes has been subject to particular attention (9, 17, 25, 31), but neither accounts fully for the differences in pathogenicity between strains. A greater understanding of the factors governing strain variation in pathogenic potential has implications for the development of improved diagnostic tools and therapeutic strategies.
Analysis of the immunoglobulin G serum antibody response of periodontitis patients has led to the identification of three immunodominant surface antigens (115, 55, and 47 kDa) expressed by P. gingivalis W50 (8). We have previously cloned the gene ragB (receptor antigen gene B), encoding the 55-kDa antigen, and found it to be located immediately downstream of a cotranscribed gene, ragA, coding for a putative tonB-dependent receptor (12). Several factors led to the conclusion that the rag locus has arisen by horizontal gene transfer and may represent a pathogenicity island in P. gingivalis W50: the locus has a G+C content of 42%, compared to a mean value of 48% for the complete genome; it is flanked by an insertion sequence (IS1126, now known as ISPg1) and a sequence with similarity to mobile elements; PCR and Southern blots indicated that the locus has a restricted distribution within the species; and PCR of subgingival samples suggested an association of the locus with deep periodontal pockets (12). The aim of the present study was to investigate the chromosomal location and extent of polymorphism of the rag locus in a diverse collection of isolates of P. gingivalis.
MATERIALS AND METHODS
Bacterial strains and culture conditions. Isolates were grown on fastidious anaerobic agar plates with 5% defibrinated horse blood or brain heart infusion (BHI) broth with 5 μg ml–1 hemin at 37°C in an anaerobic cabinet (Don Whitley Scientific) with an atmosphere of 80% N2, 10% H2, and 10% CO2.
A total of 168 isolates of P. gingivalis from our laboratory collection were investigated, including isolates generously supplied by numerous colleagues. All isolates had been recovered from human sources, mostly from patients with periodontal disease, and were from 15 countries. Isolates were confirmed to be P. gingivalis by PCR with species-specific primers to 16S rRNA genes as described previously (3, 12).
DNA manipulations. Total genomic DNA was isolated from stationary-phase bacteria using a Puregene DNA isolation kit reagent (Flowgen) according to the manufacturer's instructions. Briefly, 1 ml of brain heart infusion cultures grown overnight were lysed with 600 μl cell lysis solution at 80°C for 5 min. Cells were treated with 3 μl RNase A solution at 37°C for 1 h. Protein was removed by precipitation with guanidine thiocyanate, and the DNA was precipitated and cleaned with successive washes of 100% isopropanol and 70% ethanol. Purified DNA was resuspended in 50 μl TE (10 mM Tris, 1 mM EDTA, pH 8.0).
DNA purification to remove primers, enzymes, and other reagents was undertaken using a QIAgen gel extraction kit. Briefly, DNA was bound to a silicon-based ion-exchange matrix under high-salt conditions in Qiagen buffer PB. The DNA was cleaned with two washes of Qiagen buffer PE and eluted in 50 μl TE.
Restriction digestion of genomic DNA and PCR products was performed using Amersham Pharmacia or New England Biolabs enzymes and buffers. Reaction mixtures were incubated at 37°C for at least 2 h. DNA electrophoresis was performed in 0.8% agarose with Tris-borate-EDTA (0.09 M Tris-borate, 0.002 M EDTA). Ethidium bromide-stained gels were viewed under UV light, and the image was captured with a Syngene Imager. Southern hybridizations were performed by standard methods, with DNA immobilized on HyBond N+ (Amersham Pharmacia). PCR products used as probes were labeled with digoxigenin and detected by color precipitate using a DIG labeling and hybridization kit (Roche). Hybridization was at 65°C, and final washes were in 0.2x SSC (1x SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% sodium dodecyl sulfate for 15 min at 65°C.
PCR. All amplification reactions were performed by using an Omnigene Life Sciences International thermal cycler. Standard PCRs used 50-μl volumes with 0.5 μg each primer and 0.5 μl chromosomal DNA template in ABgene PCR Reddy Load Master Mix (ABgene) with 1.5 mM MgCl2. Amplification of longer products was performed with ABgene Extensor Long Master Mix 2 (ABgene).
The following primers (with positions of primers and products indicated in Fig. 1) were used: PG0183F (forward) (CAA CTA CCT TCA CTT CCC), PG0183R (reverse) (AAC CTT CAA TCT CCT CCA C), ISPg1F (GGC TCC CCC TTG ATT TTT CTC), ISPg1R (AAG GAG GTG GAG GGA AGA GGA) W83 ragAF (TGT CTG GTG CGA TAG CGA ATA G), W83 ragAR (TAC ATA GGT GAG TTT GAG ATT C), W83 ragBF (AAT ACT GAA AAT CCA CGA), W83 ragBR (TAG GGG CTG CGA CAA AAA), ELPCRF (TAT CGT TTC GCC CAG CCA AGA), ELPCRR (TCC TTC GCA TCG CTC TTT CCC), LPCRF (CAA AGT CCT GCC ACG AGT AGC), and LPCRR (CGT TTT CTC GCC ACT TTC GTC). Primers designed to be specific to ragB from each of the four rag alleles were rag1F (CGC GAC CCC GAA GGA AAA GAT T) and rag1R (CAC GGC TCA CAT AAA GAA CGC T) (positions 70 to 697 of ragB1), rag2F (GCT TTG CCG CTT GTG ACT TGG) and rag2R (CCA CCG TCA CCG TTC ACC TTG) (positions 56 to 1034 of ragB2), rag3F (CCG GAA GAT AAG GCC AAG AAA GA) and rag3R (ACG CCA ATT CGC CAA AGC T) (positions 76 to 498 of ragB3), and rag4F (CCG GAT GGA AGT GAT GAA CAG A) and rag4R (CGC GGT AAA CCT CAG CAA ATT) (positions 82 to 820 of ragB4).
Standard programs for PCR comprised 30 cycles of denaturation for 1 min at 95°C, annealing for 1 min at 50°C, and extension for 1 to 3 min at 72°C. The annealing temperature was changed to 55°C for ragA primers and 60°C for ISPg1 primers. For extra-long PCR (ELPCR), the annealing temperature was 60°C, the extension time was 20 min, and 25 cycles were performed. For long PCR (LPCR), 10 cycles with annealing at 52°C and an extension time of 6 min 40 s at 68°C were followed by 15 cycles with annealing at 60°C and a 7-min extension step at 68°C.
Pulsed-field gel electrophoresis (PFGE). Bacterial growth from 4-day-old cultures grown on fastidious anaerobic agar was washed and resuspended in 0.5 ml of PIV buffer (10 mM Tris [pH 7.6], 1 M NaCl). The cell suspension was added to 0.5 ml molten 2% CHEF agarose (Bio-Rad) at 50°C, mixed, and gently pipetted into a CHEF plug mold (Bio-Rad). Plugs were incubated at 37°C in 2 ml lysis buffer (6 mM Tris [pH 6.7], 1 M NaCl, 100 mM EDTA, 0.5% Brij 58 solution, 0.2% deoxycholate, 0.5% N-lauroylsarcocine, 20 μg/ml RNase A, 100 μg/ml lysozyme) for 2 h. Plugs were washed in 2 ml TE (CHEF) (10 mM Tris [pH 7.5], 0.1 mM EDTA) at 50°C for 5 min and then incubated in 2 ml ESP buffer (0.5 M EDTA [pH 9 to 9.5], 1% N-lauroylsarcosine, 50 μg/ml proteinase K) at 50°C for 2 h and then washed twice in 2 ml TE (CHEF) at 37°C for 30 min. Plugs were stored in TE at 4°C in the dark. For restriction, plug slices approximately 1 mm thick were taken from three separate blocks and placed into 1.5-ml microcentrifuge tubes and equilibrated in restriction buffer, and then DNA was digested overnight at 37°C in 200 μl buffer plus XbaI. Electrophoresis was performed with a CHEF DRII system (Bio-Rad) for 19 h at 6 V cm–1, with a switch time of 1 to 30 s and with a 1% agarose gel in 0.5x Tris-borate-EDTA buffer at 15°C.
Antiserum production and Western blotting. Preparation of extracts, electrophoresis on 10% polyacrylamide gels, and detection of the 55-kDa outer membrane antigen RagB with monoclonal antibodies B15 and E38 were performed as previously described (20). Polyclonal antiserum to recombinant W50 RagB, synthesized as previously described (29), was produced in rabbits. Pooled dialyzed fractions of recombinant protein in phosphate-buffered saline (0.7 mg/ml) were incubated with a fresh precipitate of alum. The precipitate was washed and resuspended in saline for injection. Primary immunization was with 175 μg protein subcutaneously over four sites. Two more immunizations of 100 μg were given 3 weeks apart. Blood was withdrawn and separated for antiserum after an additional 3 weeks.
DNA sequence determination. Long PCR products from A011/9, QM220, and ATCC 33277T (strain obtained as NCTC11834T) were purified and sent to MWG-Biotech for sequence determination using their publication-grade sequencing service. Open reading frames were identified with FramePlot 2.3.2 online software (13).
Nucleotide sequence accession numbers. The nucleotide sequence data reported in this study have been assigned GenBank accession numbers AY842852 (NCTC11834T), AY842853 (QM220), and AY842854 (A011/9).
RESULTS AND DISCUSSION
Detection of rag genes and genomic location in diverse isolates. We previously proposed that P. gingivalis W50 had acquired the rag locus by horizontal gene transfer (12), raising the possibility that the site of insertion could differ between isolates. We had also demonstrated that the rag locus of W50 was not present in all isolates. In order to identify isolates that were genetically diverse for further investigation of the locus, we first typed 42 isolates originating from several countries. PFGE identified 13 different types (A to M) among 37 isolates (Table 1). Four types were shared by two or more isolates, while nine were unique. Five isolates consistently failed to yield readable band patterns; 23A4 was retained for further investigation, as this is a well-characterized isolate, but the other four untyped isolates were not studied further.
The presence of genes corresponding to ragA and ragB of W83/W50 was investigated by PCR using primers located as illustrated in Fig. 1. In 20 isolates of nine PFGE types, both genes were detected, while in 18 isolates (including the type strain) of four PFGE types or that were untyped, ragA and ragB PCR failed to yield a product (Table 1). Southern hybridization of chromosomal DNA with probes comprising PCR products from W50 ragA and ragB (Fig. 1) confirmed the PCR results in all cases. PCR with primers in the gene upstream of the rag locus in W83, PG0183 (separated from ragA by ISPg1 [Fig. 1]), was also negative in 5 of the 18 rag-negative isolates. In all cases, identical PCR results were obtained for isolates having the same PFGE type (Table 1).
Western blots were performed on representative isolates of rag-positive PFGE types A, B, C, D, and I and rag-negative PFGE types F, G, and J plus untyped isolate 23A4 by using monoclonal antibodies (MAbs) specific to RagB (MAb B15 and MAb E38). The 55-kDa RagB was detected in all rag-positive isolates and was absent in all rag-negative isolates.
Publication of the complete genome sequence of P. gingivalis W83 allowed examination of the region surrounding the locus (26). PCR was performed on representatives of all PFGE types with primers from sequences in PG0183, upstream of the rag locus in the sequenced isolate W83, and PG0190 (uppS), downstream of the locus (ELPCR in Fig. 1). A PCR product of similar length (13.5 to 15 kb, established by adding the sizes of fragments after digestion with BglII and NspI) was obtained with isolates of the nine rag-positive PFGE types and a ragB probe hybridized to a 2.5-kb BclI fragment from the PCR product in all cases. This indicated that the rag locus was in the same genomic location (i.e., flanked by the same genes) in diverse isolates. Variations in the length of the long PCR product and in the sizes of restriction fragments suggested that there was nevertheless some polymorphism between isolates in this region.
Polymorphism among rag-positive isolates. Polymorphism in the rag locus of isolates representing the nine rag-positive PFGE types was investigated further. Long PCR products spanning the rag locus (ELPCR, PG0183 to PG0190), as described above, were digested with NspI and BglII. Fragments were identified by size comparison to those of W83 and by hybridization with probes from ragA, ragB, and a fragment overlapping ISPg1. All isolates with PFGE types other than type A had a deletion of 1.3 kb between PG0183 and ragA, relative to W50 and W83. The restriction fragments obtained were compatible with the absence of ISPg1 in these isolates. Isolate LB13D3 had an insertion of approximately 1 kb in the intergenic region between ragB and PG0188, and a number of restriction sites were absent in individual isolates, suggesting a degree of sequence polymorphism in the locus.
The locus in rag-negative isolates. The same long PCR primers (ELPCR) were used to amplify DNA from isolates in which ragA and ragB gave negative results but PG0183 PCR was positive. In the absence of ragA and ragB, a PCR product of no more than 10 kb would be predicted; unexpectedly, products of 12.6 to 14 kb were obtained. A simple insertion/deletion event was therefore unlikely to account for the difference between these rag-negative isolates and rag-positive isolates. Southern hybridization of digested ELPCR products with W50 ragA and ragB probes again confirmed the absence of sequences with homology to these genes, except for isolate YH522, in which a weak but distinct hybridization signal to a 6.6-kb BglII fragment was detected with the W50 ragA probe.
An alternative primer pair was designed to amplify the rag locus with less additional flanking sequence by using primers closer to the 3' end of PG0183 and, in PG0188, the gene immediately downstream of ragB (LPCR in Fig. 1). These should result in an 8.35-kb product in W83. Amplification with the alternative primer pair yielded fragments of 6 to 8.5 kb in isolates representative of all PFGE types, including those that had been negative for PG0183 by PCR. Again, this suggested that the locus had been replaced rather than deleted in isolates that both PCR and Southern blotting had indicated did not contain W50 ragA and ragB. It also suggested that the failure to detect PG0183 by PCR in some isolates might be due to local sequence polymorphism rather than absence of the gene.
Novel variants of the rag locus. Digestion of LPCR products spanning the region between PG0183 and PG0188 in rag-negative isolates yielded restriction fragments that were of very different sizes from rag-positive isolates. To determine the extent of polymorphism in this region, a much larger collection of isolates was screened by digestion of the LPCR product with NcoI, SacI, and SphI. From a total of 168 isolates, 15 different restriction profiles were obtained; for two isolates, no specific amplification product could be obtained. The profiles fell into four main groups on the basis of common restriction fragments, designated 1 to 4; within each group, profiles differed by up to three bands, consistent with a single polymorphism, and were designated with a lowercase letter (1a, 1b, etc.) (Table 2). The profiles most frequently observed were 2b (43 isolates), 3a (40 isolates), 1a (25 isolates), and 1b, 2a, and 4a (14 isolates each) (Fig. 2). W83 and W50 both had profile 1a; profile 1b was highly related and was observed in isolates for which the rag locus had already been shown to differ from W50 only in the absence of ISPg1.
The LPCR product extending from the 3' end of PG0183 to the 5' end of PG0188 was sequenced in isolates A011/9 (profile 2a), QM220 (profile 3a), and ATCC 33277T (profile 4a) in order to find out what genes had replaced ragA and ragB of W50/W83. Analysis of open reading frames revealed that each of the three isolates carried different novel genes with homology to ragA and ragB of W50/W83. Sites of PCR primers used to detect ragA and ragB were absent or divergent in each isolate, and the regions corresponding to probes were less than 60% identical to the corresponding genes in W50/W83, accounting for the failure to detect rag genes in these isolates.
Between PG0183 and ragA in A011/9 was a gene with 98% nucleotide homology to the transposase of ISPg3 (formerly IS195), an insertion sequence found previously in P. gingivalis (18). Examination of the flanking sequences identified a 10-bp direct repeat element (TTACATCACA), possibly a target for insertion. On the basis of restriction analysis described above, it could be deduced that isolates with LPCR restriction profile 2b carried the same rag locus as A011/9 but lacked ISPg3.
In ATCC 33277T, an additional open reading frame was found downstream of ragB. The protein encoded had no significant homology to known proteins, except for a C-terminal 80-amino-acid sequence with 50% similarity to the C-terminal domain shared by gingipain precursor proteins and hemagglutinins of P. gingivalis (7).
To distinguish the different rag loci, the following designations were assigned: rag-1, representing the locus found in W50 and W83; rag-2, representing the locus of A011/9; rag-3, representing the locus of QM220; and rag-4, representing the locus of ATCC 33277T. PCR primers were designed to specifically amplify ragB of each of the four rag loci and distinguish between them (further details of this PCR assay and its use in clinical material will be described in a separate publication). When applied to the full collection of isolates, PCR confirmed the groups assigned by LPCR restriction profile (Table 2); rag-3 was detected in the two isolates for which LPCR was unsuccessful.
Antigenic variation and expression. Western blots of isolates with different PFGE types had already demonstrated that two monoclonal antibodies recognizing RagB1 from W50 did not cross-react with proteins from isolates lacking the rag-1 locus. To further investigate antigenic cross-reactivity between the different variants of RagB, polyclonal antiserum was raised against recombinant RagB1 from W50. Western blots with representative isolates carrying the four alleles of the rag locus demonstrated that neither the monoclonal antibodies nor the polyclonal antiserum to RagB1 were able to recognize RagB2, RagB3, or RagB4 (Fig. 3).
In a previous study, we had investigated outer membrane proteins in strains W50, W83 (both rag-1), and NCTC11834T (rag-4) (8). Retrospective examination of the published gel images demonstrates that while W50 and W83 both produced a prominent band at 55 kDa, which later proved to represent RagB, at least one protein of similar molecular weight but which was apparently less abundant was also expressed by NCTC11834T. In the absence of a specific antibody to Rag4, it was not possible to confirm the identity of this protein. However, Murakami et al. recently determined by N-terminal sequencing that both RagA and a variant of RagB that can now be confirmed to be RagB4 are expressed in ATCC 33277T (rag-4), where they represent two of the seven major outer membrane proteins (22, 23).
Sequence comparisons of the four rag alleles. Multiple alignment of the four RagA proteins using ClustalW (32) demonstrated a high level of sequence homology at the N terminus of the proteins, with only one amino acid polymorphism in the first 150 residues (Fig. 4). Through the rest of the protein, there were local regions of high homology interspersed with regions with little sequence conservation. The RagB proteins were more diverse, with only short regions of sequence conservation (Fig. 5). Alignment of the DNA sequences revealed that the region between PG0183 and ragA was highly conserved between isolates, apart from the presence of insertion sequences in W50/W83 and A011/9. Only 29 (3.1%) of 939 nucleotides in this intergenic region were found to be polymorphic. Conversely, there was no significant sequence homology detected between any of the isolates in the intergenic region between ragB and PG0188, apart from the last 46 nucleotides upstream of PG0188 (Fig. 1).
All four variants of RagA and RagB were submitted to the NCBI Conserved Domain Search and BLAST analysis (1, 19). Homology to the N-terminal region of conserved domains "ligand_gated_channel, TonB dependent (cd01347)," "TonB_dep_rec, TonB dependent receptor (pfam00593)," "FepA, outer membrane receptor for ferrienterochelin and colicins (COG4771)," "CirA, outer membrane receptor proteins, mostly Fe transport (COG1629)," and "BtuB, outer membrane cobalamin receptor protein" was detected from approximately amino acids 100 to 250 of all RagA variants. Further homology to "ligand_gated_channel, TonB_dep_rec" and CirA was also detected over the less conserved part of the protein, especially amino acids 500 to 800 of all variants. No conserved domains were detected in the RagB proteins.
All RagA variants demonstrated extensive homology to members of the large families of putative outer membrane proteins (over 50 in each) from the complete genome sequences of Bacteroides thetaiotaomicron (35) and Bacteroides fragilis (15). Many of these outer membrane protein genes have a putative lipoprotein gene immediately downstream, in an arrangement analogous to the rag locus, and RagB in turn had significant homology to some of these lipoproteins. For all Rag variants, the homology to other P. gingivalis Rag protein variants was always much greater than the homology to the most closely related Bacteroides protein (Tables 3 and 4); the same B. thetaiotaomicron locus ranked as the highest score for each of the RagA and RagB variants.
The most similar protein to RagA with an experimentally proven function was SusC (27), involved in starch uptake in B. thetaiotaomicron, as noted previously (12) (Tables 3 and 4). SusD, a lipoprotein functionally associated with SusC and encoded immediately downstream (28) had significant similarity to RagB2 over a short region, but similarity to the other RagB variants was scored as nonsignificant by standard BLASTP analysis. By analogy to SusC/D, the families of related outer membrane proteins and lipoproteins in both Bacteroides species have been interpreted as systems for the binding and uptake of a wide range of polysaccharides from their environment in the large intestine (15, 35). Circumstantial support for this role comes from the finding of susCD/ragAB-related genes on a plasmid required for agar degradation by a marine bacterium identified as a Microscilla species (36). The occurrence of such a function in P. gingivalis is surprising given that this species is asaccharolytic, requiring protein as a carbon source.
The RagA sequences were submitted to the web-based program for protein fold recognition, 3D-PSSM (14). In accordance with the results of the Conserved Domain Search, all four RagA variants had the greatest structural homology (>95% confidence level) to the TonB-dependent outer membrane transporter proteins FepA, FhuA, and BtuB from Escherichia coli. FepA, FhuA, and BtuB all form 22-stranded antiparallel -barrels, with an N-terminal domain that forms a plug within the barrel and interacts with TonB (5, 6, 10). Alignment of the sequences of the four RagA variants, the B. thetaiotaomicron homologue BT0206, and SusC demonstrated strong conservation in the position of predicted barrel region -strands despite extensive sequence divergence (Fig. 4). (The structural similarity of RagA1 to FepA had previously been noted by Wexler et al. [34], who have characterized the locus encoding the omp200 porin complex of Bacteroides fragilis and demonstrated a gene organization similar to that of the rag locus.)
Analysis of the RagB sequences by 3D-PSSM revealed a predicted similarity to the alpha/alpha-6 barrel structure of 1AYX, a glucoamylase of yeast belonging to a sequence family 15 of glycosyl hydrolases (30), although the significance score was over 95% only for RagB4. This prediction was intriguing given the role of the Sus operon in polysaccharide degradation and uptake (27, 28). Sequence alignment of the RagBs, BT0207 (the putative lipoprotein encoded downstream of BT0206), and SusD demonstrated close correspondence in the positions of 11 predicted alpha helices (Fig. 5), but the location and spacing of alpha helices did not match well to those of 1AYX (not shown).
Distribution of the rag alleles. The collection of P. gingivalis investigated here comprises isolates from many sources, including well-characterized laboratory strains as well as more recent clinical isolates. A majority of isolates are from western European countries, with few from North America. The collection may not be fully representative of the P. gingivalis population currently circulating, nor is it possible to rule out that some isolates may be duplicates either from the same patient or from laboratory strains that have been renamed by different laboratories. Nevertheless, some general comments may be made about the prevalence and geographic distribution of the four rag alleles.
All four of the rag alleles were found among isolates from several countries; none was restricted to isolates from a single continent. Their relative prevalence in this collection was comparable, except that rag-4 was found less frequently than the other alleles. rag-1 was detected in isolates obtained from Germany, The Netherlands, Romania, Sweden, the United Kingdom, and Kenya. However, LPCR restriction profile 1a was frequently found in established laboratory strains such as W50 and W83, some of which may have a common origin, while profile 1b (lacking ISPg1) was more frequent among isolates thought to have a recent clinical source. Profile 2a (rag-2) was found exclusively among a set of isolates from Thailand, but profile 2b, representing the rag-2 locus without ISPg3, was found in isolates from across Europe (Finland, The Netherlands, Sweden, and the United Kingdom). rag-3 was found in numerous isolates from Europe and one isolate each from Japan and Thailand. Profile 4a (rag-4) was represented by strains ATCC 33277T and 381, both of which are long-established laboratory strains, and in isolates from European countries.
Relationship of rag alleles to other markers of disease and virulence. The distribution of rag alleles was investigated relative to other polymorphic markers that have been used to type P. gingivalis and determine associations with disease (Table 5). A set of isolates kindly provided by M. L. Laine representing two to three isolates each of the six K-capsular serotypes (K1 to K6) plus unencapsulated types (K0) of P. gingivalis (16, 33) was included in the analysis. There was a limited degree of correlation between the rag allele and K serotype: K3 and K5 were found only in rag-3 isolates, and K4 was found only in rag-1 isolates, but K1, K2, and K0 were each associated with two different rag alleles; each rag allele was found in isolates of at least two different K serotypes (Table 5). The fimA genotype was known for eight isolates (25) and showed no clear association with the rag allele. Genotypes fimA II and Ib have been reported to be associated with periodontitis and recovery from deep pockets in epidemiological studies in Asia (2, 24), although the same association was not found in a study in Europe (4); these genotypes were found among isolates with three different rag alleles. The ribosomal intergenic spacer heteroduplex types were too diverse among the few isolates for which they were known to draw any conclusions, but again, types with an epidemiological association with periodontitis (11) were found in isolates of both rag-1 and rag-2. We had previously demonstrated the association of the rag locus, corresponding to rag-1 in this study, with deep pockets in patients with periodontal disease (12). An epidemiological study comparing these various markers in isolates from patients with detailed clinical assessment is required to evaluate their relative strength of association with disease. Such a study would have the potential to indicate the significance of the three polymorphic surface antigens, RagAB, capsular polysaccharide, and fimbriae, to the outcome of P. gingivalis infection.
An alternative strategy to assess the significance of putative virulence factors is to use an animal model of infection, and Laine and van Winkelhoff (17) have reported the virulence of isolates with different serotypes in a mouse model. Encapsulated strains were significantly more virulent than K0 strains, but variability in virulence between strains with the same serotype suggested the importance of additional virulence factors that must differ between strains. A measure of virulence from the Laine and van Winkelhoff study has been included in Table 5, and it was apparent that most of the rag-1 isolates had high scores for virulence. Chi-square analysis demonstrated that there was a significant correlation (P < 0.05) between carriage of rag-1 (versus carriage of any other rag allele) and high virulence (2/3 or 3/3 mice seriously ill within 72 h of inoculation).
Further discussion. In our initial investigation of the rag locus of P. gingivalis, it was concluded that the genes encoding this immunodominant surface antigen had a restricted distribution in the species (12). The present study has demonstrated that there are actually four distinct variants or alleles of the rag locus, each of which is common among the P. gingivalis population and none of which appear to be confined to specific geographic regions.
Although the function of RagA/B has not yet been experimentally determined, the similarity to SusC/D family proteins in Bacteroides species suggests a possible role in polysaccharide uptake. Expression of some of the SusC/D loci in B. fragilis has been found to be controlled by DNA inversion systems, including three clusters of loci where inversion of a promoter mediates alternative expression of different SusC/D cassettes (15). Other DNA inversion systems in B. fragilis affect the expression of other surface-expressed molecules such as capsular polysaccharides. These systems are likely to represent a mechanism for the high-frequency generation of antigenic variation and are thus implicated in evasion of the host response (15). While there is no evidence for high-frequency variation of RagA/B, the rag locus bears many of the hallmarks characteristic of genes involved in the production of variable surface antigens. These features include an association of insertion elements with the locus and a G+C content lower than what is typical of the species and are generally interpreted to suggest that the locus and its variant alleles have been acquired by horizontal gene transfer. The large number of genes in Bacteroides spp. with sequence homology to the rag locus would suggest a ready reservoir of genes available for recombination, although none of the genes available to date in sequence databases are sufficiently similar to suggest a close ancestry in common with the P. gingivalis locus.
The presence of variant alleles of the rag locus leads directly to the question of whether specific alleles are associated with different clinical conditions. We had previously demonstrated that rag-1 was associated with deep periodontal pockets (12). The present study has revealed that strains shown by others to be more virulent in a mouse model (17) are significantly more likely to carry rag-1 than other rag alleles. Further work is in progress to investigate the links between the rag locus and pathogenesis in more detail.
ACKNOWLEDGMENTS
This work was funded by development grant G9900398 and program grant PG9318173 from the Medical Research Council (United Kingdom) and a Ph.D. studentship (for S.C.F.) from the Special Trustees of the Royal London Hospital.
We thank Jenny Slaney for her assistance with immunological techniques.
REFERENCES
1. Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang, W. Miller, and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25:3389-3402.
2. Amano, A., I. Nakagawa, K. Kataoka, I. Morisaki, and S. Hamada. 1999. Distribution of Porphyromonas gingivalis strains with fimA genotypes in periodontitis patients. J. Clin. Microbiol. 37:1426-1430.
3. Ashimoto, A., C. Chen, I. Bakker, and J. Slots. 1996. Polymerase chain reaction detection of 8 putative periodontal pathogens in subgingival plaque of gingivitis and advanced periodontitis lesions. Oral Microbiol. Immunol. 11:266-273.
4. Beikler, T., U. Peters, S. Prajaneh, K. Prior, B. Ehmke, and T. F. Flemmig. 2003. Prevalence of Porphyromonas gingivalis fimA genotypes in Caucasians. Eur. J. Oral Sci. 111:390-394.
5. Buchanan, S. K., B. S. Smith, L. Venkatramani, D. Xia, L. Esser, M. Palnitkar, R. Chakraborty, D. van der Helm, and J. Deisenhofer. 1999. Crystal structure of the outer membrane active transporter FepA from Escherichia coli. Nat. Struct. Biol. 6:56-63.
6. Chimento, D. P., A. K. Mohanty, R. J. Kadner, and M. C. Wiener. 2003. Substrate-induced transmembrane signaling in the cobalamin transporter BtuB. Nat. Struct. Biol. 10:394-401.
7. Curtis, M. A., H. K. Kuramitsu, M. Lantz, F. L. Macrina, K. Nakayama, J. Potempa, E. C. Reynolds, and J. Aduse-Opoku. 1999. Molecular genetics and nomenclature of proteases of Porphyromonas gingivalis. J. Periodont. Res. 34:464-472.
8. Curtis, M. A., J. M. Slaney, R. J. Carman, and N. W. Johnson. 1991. Identification of the major surface protein antigens of Porphyromonas gingivalis using IgG antibody reactivity of periodontal case-control serum. Oral Microbiol. Immunol. 6:321-326.
9. Dierickx, K., M. Pauwels, M. L. Laine, J. Van Eldere, J. J. Cassiman, A. J. van Winkelhoff, D. van Steenberghe, and M. Quirynen. 2003. Adhesion of Porphyromonas gingivalis serotypes to pocket epithelium. J. Periodontol. 74:844-848.
10. Ferguson, A. D., E. Hofmann, J. W. Coulton, K. Diederichs, and W. Welte. 1998. Siderophore-mediated iron transport: crystal structure of FhuA with bound lipopolysaccharide. Science 282:2215-2220.
11. Griffen, A. L., S. R. Lyons, M. R. Becker, M. L. Moeschberger, and E. J. Leys. 1999. Porphyromonas gingivalis strain variability and periodontitis. J. Clin. Microbiol. 37:4028-4033.
12. Hanley, S. A., J. Aduse-Opoku, and M. A. Curtis. 1999. A 55-kilodalton immunodominant antigen of Porphyromonas gingivalis W50 has arisen via horizontal gene transfer. Infect. Immun. 67:1157-1171.
13. Ishikawa, J., and K. Hotta. 1999. FramePlot: a new implementation of the frame analysis for predicting protein-coding regions in bacterial DNA with a high G+C content. FEMS Microbiol. Lett. 174:251-253.
14. Kelley, L. A., R. M. MacCallum, and M. J. Sternberg. 2000. Enhanced genome annotation using structural profiles in the program 3D-PSSM. J. Mol. Biol. 299:499-520.
15. Kuwahara, T., A. Yamashita, H. Hirakawa, H. Nakayama, H. Toh, N. Okada, S. Kuhara, M. Hattori, T. Hayashi, and Y. Ohnishi. 2004. Genomic analysis of Bacteroides fragilis reveals extensive DNA inversions regulating cell surface adaptation. Proc. Natl. Acad. Sci. USA 101:14919-14924.
16. Laine, M. L., B. J. Appelmelk, and A. J. van Winkelhoff. 1996. Novel polysaccharide capsular serotypes in Porphyromonas gingivalis. J. Periodont. Res. 31:278-284.
17. Laine, M. L., and A. J. van Winkelhoff. 1998. Virulence of six capsular serotypes of Porphyromonas gingivalis in a mouse model. Oral Microbiol. Immunol. 13:322-325.
18. Lewis, J. P., and F. L. Macrina. 1998. IS195, an insertion sequence-like element associated with protease genes in Porphyromonas gingivalis. Infect. Immun. 66:3035-3042.
19. Marchler-Bauer, A., J. B. Anderson, C. DeWeese-Scott, N. D. Fedorova, L. Y. Geer, S. He, D. I. Hurwitz, J. D. Jackson, A. R. Jacobs, C. J. Lanczycki, C. A. Liebert, C. Liu, T. Madej, G. H. Marchler, R. Mazumder, A. N. Nikolskaya, A. R. Panchenko, B. S. Rao, B. A. Shoemaker, V. Simonyan, J. S. Song, P. A. Thiessen, S. Vasudevan, Y. Wang, R. A. Yamashita, J. J. Yin, and S. H. Bryant. 2003. CDD: a curated Entrez database of conserved domain alignments. Nucleic Acids Res. 31:383-387.
20. Millar, D. J., E. E. Scott, J. M. Slaney, S. U, P. Benjamin, and M. A. Curtis. 1993. Production and characterisation of monoclonal antibodies to the principle sonicate antigens of Porphyromonas gingivalis W50. FEMS Immunol. Med. Microbiol. 7:211-222.
21. Missailidis, C. G., J. E. Umeda, C. Ota-Tsuzuki, D. Anzai, and M. P. Mayer. 2004. Distribution of fimA genotypes of Porphyromonas gingivalis in subjects with various periodontal conditions. Oral Microbiol. Immunol. 19:224-229.
22. Murakami, Y., M. Imai, Y. Mukai, S. Ichihara, H. Nakamura, and F. Yoshimura. 2004. Effects of various culture environments on expression of major outer membrane proteins from Porphyromonas gingivalis. FEMS Microbiol. Lett. 230:159-165.
23. Murakami, Y., M. Imai, H. Nakamura, and F. Yoshimura. 2002. Separation of the outer membrane and identification of major outer membrane proteins from Porphyromonas gingivalis. Eur. J. Oral Sci. 110:157-162.
24. Nakagawa, I., A. Amano, Y. Ohara-Nemoto, N. Endoh, I. Morisaki, S. Kimura, S. Kawabata, and S. Hamada. 2002. Identification of a new variant of fimA gene of Porphyromonas gingivalis and its distribution in adults and disabled populations with periodontitis. J. Periodont. Res. 37:425-432.
25. Nakano, K., M. Kuboniwa, I. Nakagawa, T. Yamamura, R. Nomura, N. Okahashi, T. Ooshima, and A. Amano. 2004. Comparison of inflammatory changes caused by Porphyromonas gingivalis with distinct fimA genotypes in a mouse abscess model. Oral Microbiol. Immunol. 19:205-209.
26. Nelson, K. E., R. D. Fleischmann, R. T. DeBoy, I. T. Paulsen, D. E. Fouts, J. A. Eisen, S. C. Daugherty, R. J. Dodson, A. S. Durkin, M. Gwinn, D. H. Haft, J. F. Kolonay, W. C. Nelson, T. Mason, L. Tallon, J. Gray, D. Granger, H. Tettelin, H. Dong, J. L. Galvin, M. J. Duncan, F. E. Dewhirst, and C. M. Fraser. 2003. Complete genome sequence of the oral pathogenic bacterium Porphyromonas gingivalis strain W83. J. Bacteriol. 185:5591-5601.
27. Reeves, A. R., J. N. D'Elia, J. Frias, and A. A. Salyers. 1996. A Bacteroides thetaiotaomicron outer membrane protein that is essential for utilization of maltooligosaccharides and starch. J. Bacteriol. 178:823-830.
28. Reeves, A. R., G. R. Wang, and A. A. Salyers. 1997. Characterization of four outer membrane proteins that play a role in utilization of starch by Bacteroides thetaiotaomicron. J. Bacteriol. 179:643-649.
29. Scragg, M. A., A. Alsam, M. Rangarajan, J. M. Slaney, P. Shepherd, D. M. Williams, and M. A. Curtis. 2002. Nuclear targeting of Porphyromonas gingivalis W50 protease in epithelial cells. Infect. Immun. 70:5740-5750.
30. Sevcik, J., A. Solovicova, E. Hostinova, J. Gasperik, K. S. Wilson, and Z. Dauter. 1998. Structure of glucoamylase from Saccharomycopsis fibuligera at 1.7 A resolution. Acta Crystallogr. D Biol. Crystallogr. 54:854-866.
31. Sugano, N., K. Ikeda, M. Oshikawa, Y. Sawamoto, H. Tanaka, and K. Ito. 2004. Differential cytokine induction by two types of Porphyromonas gingivalis. Oral Microbiol. Immunol. 19:121-123.
32. Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680.
33. van Winkelhoff, A. J., B. J. Appelmelk, N. Kippuw, and J. de Graaff. 1993. K-antigens in Porphyromonas gingivalis are associated with virulence. Oral Microbiol. Immunol. 8:259-265.
34. Wexler, H. M., E. K. Read, and T. J. Tomzynski. 2002. Characterization of omp200, a porin gene complex from Bacteroides fragilis: omp121 and omp71, gene sequence, deduced amino acid sequences and predictions of porin structure. Gene 283:95-105.
35. Xu, J., M. K. Bjursell, J. Himrod, S. Deng, L. K. Carmichael, H. C. Chiang, L. V. Hooper, and J. I. Gordon. 2003. A genomic view of the human-Bacteroides thetaiotaomicron symbiosis. Science 299:2074-2076.
36. Zhong, Z., A. Toukdarian, D. Helinski, V. Knauf, S. Sykes, J. E. Wilkinson, C. O'Bryne, T. Shea, C. DeLoughery, and R. Caspi. 2001. Sequence analysis of a 101-kilobase plasmid required for agar degradation by a Microscilla isolate. Appl. Environ. Microbiol. 67:5771-5779.(Lucinda M. C. Hall, Stuar)