INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Human inflammatory bowel disease (IBD) represents a set of a chronic, relapsing, and remitting intestinal inflammatory disorders involving T-cell-mediated mucosal and mural destruction and polygenic familial susceptibility (34, 35). Several spontaneous and transgenic mouse strains have been established with susceptibility for chronic colitis similar to human IBD (4, 5, 18, 20, 26, 32). In all evaluated models, normal resident enteric bacteria were found to be required for disease pathogenesis (13, 27, 40). Similarly, in human IBD several lines of evidence implicate enteric bacteria as a pathogenic factor in clinical disease, particularly in Crohn's disease (CD) (3, 33, 48).

Immunologic studies have demonstrated that antibody and T-cell reactivity to commensal enteric bacteria is a distinguishing feature of colitic mouse strains (6, 10). However, the bacterial species and antigens recognized by colitigenic T cells have not yet been defined. Moreover, monoassociation studies have not yet revealed pathogenic bacterial species for colitis-prone mouse strains (27). A systematic approach to this issue is hampered by the limited understanding of gastrointestinal microflora ecosystem. In addition, immune recognition of this commensal microflora in normal individuals is attenuated or undetectable. Accordingly, it has been difficult to highlight bacterial species or antigens for evaluation in IBD pathogenesis.

Marker antibodies have been used successfully to identify disease-relevant antigenic targets in several immune-mediated diseases (38, 45). In IBD, approximately 60 to 70% of ulcerative colitis (UC) patients and 25% of CD patients have elevated levels of pANCA, an antineutrophil cytoplasmic antibody with distinctive morphological and antigenic fine specificity (15, 30, 36, 41, 51). An immunochemical study has associated pANCA with enteric bacterial antigens by serum cross-reactivity in human and mouse (39). This observation provides a precedent for the hypothesis that pANCA antibody would identify the antigenic proteins responsible for the pathogenic mucosal inflammation.

We have addressed this hypothesis using two pANCA monoclonal antibodies (Fab 5-3 and 5-2) to search for bacterial cross-reactive proteins associated with the UC-specific immune response. These antibodies were isolated by phage display technology from lamina propria lymphocytes of UC patient. Their concordance with serum pANCAs was validated by the criteria of immunofluorescence, confocal microscopy, and DNase I sensitivity (17). To our knowledge, these are the only reported pANCA monoclonal antibodies. Experience has shown that antigen discovery with individual monoclonal antibodies can be misleading and is best pursued with a diverse monoclonal antibody panel. However, because of the unique disease association of pANCA, we proceeded with the available monoclonal antibodies in a bacterial antigen search.

Using Fab 5-3, we identified two candidate bacterial pANCA antigens in a search of colonic bacterial clinical isolates from an IBD patient: OmpC of Escherichia coli, and a novel 100-kDa protein of Bacteroides caccae (8). In the present study, the gene encoding the 100-kDa protein is cloned and characterized. We show that this protein is a new TonB-linked outer membrane protein (termed OmpW) closely related to the RagA virulence factor of Porphyromonas gingivalis. Immunologic assessment demonstrated that immunoglobulin A (IgA) anti-OmpW is elevated in a subset of CD patients. These finding suggest that B. caccae and OmpW may be bacterial targets of the disease-related immune response in IBD.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. A panel of Bacteroides clinical isolates (3955-3, 4536, 4552, 4556, 4562, 4570, 4578, and 4579) were stored and cultured in the clinical laboratories at University of California, Los Angeles (UCLA). B. vulgatus strains LG-1, LG1-33, and CPT-6 were kind gifts from R. B. Sartor, University of North Carolina at Chapel Hill. B. caccae strains 43185 and p2Lc3 were from the American Type Culture Collection and a colonic isolate of a Crohn's disease patient (8), respectively. Clinical isolates of E. coli, Salmonella enterica serovar Typhi, and Shigella flexneri were provided by UCLA Clinical Laboratories. All Bacteroides strains were grown on brucella blood agar in an anaerobic chamber with 10% CO₂-90% N₂ atmosphere at 37°C.

Recombinant cloning reagents. E. coli XL-1 Blue strain (Stratagene, La Jolla, Calif.) was used for all cloning and recombinant expression experiments. The pBluescript vector (Stratagene) and pCR 2.1 plasmid (Invitrogen, Carlsbad, Calif.) vector were used for cloning in E. coli XL-1 Blue and selected on Luria-Bertani (LB) medium agar plate (1% tryptone, 0.5% yeast extract, 0.5% NaCl, 1.5% agar) supplemented with ampicillin (100 µg/ml). Blue or white colony color selection was used to distinguish between nonrecombinant and recombinant E. coli clones by spreading X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) and IPTG (isopropyl--D-thiogalactopyranoside) on LB plates in the cloning procedures. All restriction enzymes used in this study were purchased from New England Biolabs (Beverly, Mass.). Human monoclonal anti-pANCA antibody (Fab 5-3) was used to analyze recombinant OmpW protein (17). Alkaline phosphatase-conjugated goat anti-human F(ab)₂ (Pierce, Rockford, Ill.) was used as the secondary antibody in Western blot analysis.

Construction and screening of genomic library. Genomic library of a B. caccae clinical isolate (p2Lc3) was constructed using lambda ZAPII  phage vector (Stratagene, La Jolla, Calif.). Briefly, chromosomal DNA was purified from p2Lc3 using the ASAP genomic DNA isolation kit (Boehringer Mannheim, Indianapolis, Ind.), partially digested with EcoRI, and fragments ranging from 1.5 to 9 kb were ligated into lambda ZAPII EcoRI-digested, calf intestinal alkaline phosphatase (CIAP)-treated vector. Recombinant lambda phages were packaged, and the PFU of packaging reaction (primary library) were titrated. The primary library was plated at appropriate dilution on large 150-mm agar plates to 5 × 10⁴ PFU/plate and incubated at 37°C for 8 h. The phage plaques from each plate were transferred onto duplicated Hybond-N membranes (Amersham Pharmacia Biotech, Piscataway, N.J.), denatured in 0.5 N NaOH-1.5 M NaCl, and neutralized in 0.5 M Tris-Cl-1.5 M NaCl (pH 8.0), rinsed in 0.2 M Tris-Cl-2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) buffer solution, and fixed using the Stratalinker UV cross-linker (Stratagene).

Cloning and sequencing the full-length 100-kDa protein gene (OmpW). Based on primary sequence information obtained from the original genomic library clones, genome-walking strategy (rapid amplification of cDNA ends [RACE]) was employed to clone full-length gene encoding 100-kDa protein using Universal GenomeWalker kit (Clontech Laboratories, Palo Alto, Calif.). Reverse GSP and Forward GSP oligonucleotide pairs were designed to obtain upstream and downstream flanking sequences of the primary gene fragment, respectively (Table 1).

OmpW-specific PCR. To test ompW distribution in enteric bacterial strains, PCR was performed on bacterial chromosomal DNA samples to amplify an ompW-specific amplicon of 468 bp. Chromosomal DNA was extracted from bacterial strains and used as the templates in the PCR analysis with ompW-specific primers (OmpW-f and OmpW-r) spanning nucleotides 893 to 1361 of the open reading frame (Table 1). A phylogenetically conserved segment of the bacterial 16S RNA was used to quantitate bacteria DNA sample in PCR (16S primers, Table 1) (31). The PCRs were performed in a 50-µl volume consisting of 1 µg of genomic DNA, 0.5 U of Taq polymerase, 2 mM deoxynucleotide triphosphate mixture, and 1 µM each of primers in 1× PCR buffer using GeneAmp PCR System 9700 (PE Applied Biosystems): 5 min at 95°C followed by 30 cycles of 95°C for 60 s, 65°C for 60 s, and 72°C for 60 s. After the final cycle, the reaction was extended at 72°C for 5 min and cooled at 4°C.

Southern blot analysis. Southern blot analysis was employed to identify positive phages screened from genomic library and to analyze ompW gene distribution in bacterial clinical isolates. To identify insert-containing phagemids, digested plasmid DNAs were electrophoresed on 0.8% agarose gel and depurinated in 0.5 M HCl, denatured in 0.5 M Tris-Cl-1.5 M NaOH, and neutralized in 0.5 M Tris-Cl-1.5 M NaCl buffer, and then transferred onto Hybond N+ membrane (Amersham) by capillary blotting. After blotting, the membranes were prehybridized for at least 1 h at 60°C in Rapid-Hyb buffer (Amersham Life Sciences) and hybridized at 60°C for 6 h in the same buffer with a mixture of ³²P-labeled Pep1 and Pep2 oligonucleotide probes. Membranes were washed for 15 min three times with 2× SSC containing 0.1% SDS at room temperature and then washed with 0.1× SSC containing 0.1% sodium dodecyl sulfate (SDS) at 60°C for 15 min three times. Autoradiography was carried out at 70°C with intensifying screens for an optimized exposure time on Hyper Film (Amersham Life Sciences).

Expression of recombinant OmpW protein. To express the OmpW proteins, pairs of primers were designed to clone the full-length and truncated OmpW open reading frame segments directly from p2Lc3 genome by PCR. Forward primers used to amplify the full-length ORF (OmpW1) and the OmpW2 and OmpW3 truncated proteins were designated OmpW 1f, 2f, and 3f, respectively. OmpW-R was used as the reverse primer for cloning three segments (Table 1). In order to facilitate the construction of expression vectors, a BamHI site was designed in the forward primers and a SacI site in reverse primer for each fragment. The PCR products were inserted in frame into His-taggged expression vector pQE-30 (Qiagen). The full-length and truncated OmpW proteins were expressed in E. coli XL-1 Blue strain as 6× His-tagged proteins and purified by HisTrap column (AmershamPharmacia) under denatured conditions according to manufacturer's instructions.

Western blot analysis. The full-length and truncated recombinant OmpW proteins were quantified using the Bradford assay (Bio-Rad Laboratories, Hercules, Calif.), and equivalent protein amounts (2 µg/well and 0.5 µg/well) were separated on 12% polyacrylamide gels under reducing conditions. Electrophoresed proteins were transferred overnight onto nitrocellulose membranes (Amersham Life Sciences) in Tris-glycine buffer (National Diagnostics, Atlanta, Ga.) and verified by Ponceau S red staining (Sigma Chemicals, St. Louis, Mo.) or Coomassie blue staining. Membranes were blocked in 5% nonfat milk (Carnation, Glendale, Calif.) in PBS with 0.1% Tween-20 (PBS-Tween) for 1 h. Fab 5-3 antibody diluted in 1% milk-PBS-Tween was incubated with membranes for 1 h. Immunoblots were detected by alkaline phosphatase-conjugated goat anti-human F(ab)₂ and developed with 5-bromo-4-chloro-3-indolylphosphate-nitro blue tetrazolium (BCIP/NBT) liquid substrate system (Sigma Chemicals).

OmpW serum ELISA. Human antibodies that bind OmpW were detected by enzyme-linked immunosorbent assay (ELISA). Plates (USA Scientific, Ocala, Fla.) were coated overnight at 4°C with 100 µl/well of OmpW recombinant protein at 5 µg/ml in borate-buffered saline, pH 8.5. After three washes in 0.05% Tween 20 in PBS, the plates were blocked with 150 µl/well of 0.5% bovine serum albumin in PBS, pH 7.4 (BSA-PBS), for 30 min at room temperature (RT) and washed again prior to incubation with sera. Then 100 µl/well of serum from CD patients, UC patients and normal controls at various dilutions were added in duplicate and incubated for 2 h at RT. The plates were washed and incubated with alkaline phosphatase-conjugated goat anti-human IgA or anti-human IgG (Jackson ImmunoResearch, West Grove, Pa.) at a dilution of 1:1,000 in BSA-PBS for 2 h at RT. The plates were washed three times with 0.05% Tween 20 in PBS followed by another three washes with Tris-buffered normal saline, pH 7.5. Substrate solution (1.5 mg/ml of disodium P-nitrophenol phosphate; Amresco, Solon, Ohio) in 2.5 mM MgCl₂-0.01 M Tris, pH 8.6, was added at 100 µl/well to allow color development. The absorbances were measured at 405 nm. Values for pANCA activity was determined by neutrophil ELISA and categorized by neutrophil immunofluorescence, as previously described (37).

Human subjects. Serum samples from 69 subjects (23 each of UC patients, CD patients, and healthy controls) were obtained from the serum archive of the Cedars-Sinai IBD Research Center. Sera were produced from standard phlebotomy blood specimens, anonymously number coded, aliquoted, and stored at 80°C until use. The UC and CD patient specimens were drawn from an ongoing genetic case-control study; the demography, disease ascertainment methods, disease activity, and treatment profile of this population have been described previously (47, 51). At the time of our analysis, this study contained 240 UC and 213 CD probands, respectively. Each patient was diagnostically validated by clinical history, endoscopic and radiologic examination, and histopathology findings.

Nucleotide sequence accession number. The nucleotide sequence data for the ompW gene of B. caccae reported in this study have been assigned GenBank accession number AF305878.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning of full-length gene encoding 100-kDa protein of B. caccae. The first goal of this study was to clone the gene encoding 100-kDa protein of B. caccae. A genomic library of p2Lc3 was constructed by ligating EcoRI-restricted chromosomal DNA into ZAPII phage vector. Two oligonucleotide probes were designed using Bacteroides preferred codons from two N-terminal peptide sequences (Pep1 and Pep2) of the tryptic 100-kDa protein (8). To ensure a positive result in library screening, a mixture of ³²P-labeled Pep1 and Pep2 probes was used to hybridize with phage library under low-stringency conditions (40°C). Approximately 10⁶ phage plaques of the library were screened, and eight positive clones were identified.

View larger version (39K): [in this window] [in a new window]   FIG. 1.   Pep1 hybridization of a 2-kb AccI genomic segment of B. caccae. Three clones (5-2, 3-2, and 1-1) were isolated from a primary screen of a genomic B. caccae  phage library. The clones were digested with AccI, electrophoresed, transferred to nitrocellulose membranes, and hybridized with the Pep1 or Pep2 oligonucleotide probe. (Left panel) Ethidium bromide-stained gel; (middle panel) membrane hybridization with Pep1 oligonucleotide; (right panel) membrane hybridization with Pep2 oligonucleotide.

Sequence analysis and structural features of deduced OmpW protein. The major cloning steps and the structural features of the OmpW genomic segment are summarized in Fig. 2. The nucleotide sequence analysis of 4,034-bp p2Lc3 genomic fragment revealed a 2,844-bp-long open reading frame (ORF) encoding a putative outer membrane protein with a predicted molecular mass of 105 kDa and isoelectric point of 5.96. The Pep1 sequence was identified within this ORF. A 189-bp AT-rich region was identified upstream of the ORF (G+C content of 34%, compared to 48% for the ompW ORF), perhaps representing a promoter region. No homologies to known prokaryotic promoter sequences were found in this putative ompW promoter region. However, this might reflect the structural divergence of regulatory elements in Bacteroides compared to the more commonly characterized E. coli elements (43).

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Cloning of the ompW gene. A genomic  phage library of p2Lc3 was constructed and screened using Pep1 and Pep2 probes. Southern analysis identified an authentic Pep1-positive genomic clone, designated 5-2 (Fig. 1), which contained an artifactual insert Tn10 sequence. The native ompW gene was isolated by genome walking using p2Lc3 genomic DNA as substrate. Sequence analysis of a 4,043-bp fragment revealed an open reading frame encoding 947 amino acids and an upstream AT-rich putative promoter locus.

View larger version (80K): [in this window] [in a new window]   FIG. 3.   Sequence homology of OmpW with SusC and Rag A outer membrane proteins. The amino acid sequence of OmpW long open reading frame is aligned with its two closest database homologues, SusC of B. thetaiotaomicron and RagA of P. gingivalis. The peptide 1 sequence, used as a cloning probe, is also shown. Alignments were performed using the ClustalW multiple-sequence alignment program and are displayed using the GenDoc program. Residues with similarity (identical or conservative amino acid changes) among all sequences are denoted by an uppercase letter, and with single discordances by a lowercase letter or number. Gradations of similarity frequency are denoted by dark to pale shading.

View larger version (72K): [in this window] [in a new window]   FIG. 4.   OmpW contains a bacterial TonB box. Homologous sequences of TonB-linked proteins from different bacterial species were identified by BlastX and aligned and displayed as described in Fig. 3. E.coli, E. coli; Porphyromo, P. gingivalis; Bacteroide, Bacteroides fragilis; Yersinia, Yersinia enterocolitica; Shigella, Shigella dysenteriae; Haemophilu, Haemophilus influenzae; Salmonella: Salmonella enterica serovar Typhimurium; Helicobac, Helicobacter pylori; Citrobacte, Citrobacter freundii; Pseudomona, Pseudomonas sp; Synorhizob, Sinorhizobium meliloti; Synechocys, Synechocystis sp; Campylobac, Campylobacter jejuni. The TonB box is labeled.

OmpW is specific to B. caccae. To determine the species selectivity of the OmpW, the distribution of the ompW gene in different Bacteroides strains and other gram-negative coliforms (E. coli, Salmonella enterica serovar Typhi, and Shigella flexneri) was analyzed by PCR and Southern blot. PCR was performed on chromosome DNAs of the enteric bacterial strains (Fig. 5). The 468-bp ompW-specific sequence was detectable in both available B. caccae isolates, but was undetectable among four Bacteroides species and eight isolates, and the E. coli, S. enterica serovar Typhi, and S. flexneri strains. In accord with PCR results, only B. caccae strains showed a strong hybridizing band by Southern blot analysis, demonstrating that ompW is a B. caccae-specific gene (Fig. 6). The result also indicated that the gene encoding 100-kDa protein existed partially in a 10-kb HindIII restriction fragment of the B. caccae genome as a single copy.

View larger version (32K): [in this window] [in a new window]   FIG. 5.   PCR detection of ompW distribution in enteric bacteria. Chromosomal DNA samples from a panel of Bacteroides strains and gram-negative coliforms (E. coli, Salmonella enterica serovar Typhi, Shigella flexneri) were detected by PCR for the prevalence of ompW gene. The DNA samples were quantitated by the detection of conserved bacterial 16S rRNA with 16S-f and 16S-r primers (Table 1) in the same PCR. Negative control, no template DNA; positive control, ompW plasmid clone.

View larger version (52K): [in this window] [in a new window]   FIG. 6.   Southern blot analysis of ompW gene distribution in bacterial genomic DNA. A total of 5 µg of the HindIII-restricted bacterial genomic DNAs were separated on 0.8% agarose gel and transferred onto nitrocellulose membrane. The membranes were hybridized with a mixture of 32P-labeled ompW-specific fragment (0.7 kb) and DNA markers under low-stringency conditions (40°C). The membranes were exposed for an optimized time on Hyper Film. OmpW control, 10 pg of ompW restriction fragment.

The Fab 5-3 pANCA monoclonal antibody recognizes the recombinant OmpW protein. To characterize the immunological properties of the deduced OmpW protein, three gene fragments encoding full-length and truncated OmpW proteins (as shown in Fig. 7A) were cloned directly from p2Lc3 genome by PCR and inserted in frame into pQE30, a His-tagged protein expression vector. The OmpW1, OmpW2, and OmpW3 proteins (947, 805, and 648 amino acids, respectively) were purified by nickel chromatography, separated on 12% acrylamide gel (Fig. 7B), and evaluated for reactivity to the original Fab 5-3 pANCA monoclonal antibody by Western blot analysis (Fig. 7C). It was indicated that Fab 5-3 recognized OmpW1 and OmpW2, but not OmpW3. These findings confirm that we had cloned the intended Fab5-3 monoclonal antibody-pANCA-reactive Bacteroides protein. They also suggest that the Fab5-3 monoclonal antibody-pANCA epitope is located towards the N terminus of the protein, between amino acid positions 143 and 300. 

View larger version (26K): [in this window] [in a new window]   FIG. 7.   Expression and analysis of full-length and truncated OmpW proteins. (A) Strategy for expression of OmpW proteins. The gene fragments encoding the full-length and truncated OmpW proteins were cloned directly from p2Lc3. The OmpW-1 ORF starts from the first putative start codon, located at site 1097 in the 4.34-kb cloned fragment. OmpW-2 and OmpW-3 start from positions 1523 and 1994, respectively. The cloned OmpW segments were inserted into vector pQE30, expressed as His-tagged recombinant proteins, and purified by nickel chromatography. (B) SDS-PAGE analysis of purified recombinant OmpW proteins. The purified recombinant OmpW proteins were separated at 2 µg/well and 0.5 µg/well on a 12% acrylamide gel and probed with human pANCA antibody (Fab 5-3) in Western blot analysis. (C) Western blot analysis of expressed OmpW proteins with Fab5-3 pANCA antibody. The recombinant OmpW proteins were probed with Fab5-3 human monoclonal antibody and then detected by alkaline phosphatase-conjugated anti-human F(ab')2 and developed by NCIP/NBT substrate.

OmpW is an antigenic target in CD patients. Purified recombinant OmpW1 and OmpW2 proteins were used for ELISA evaluation of human anti-OmpW antibody levels in human subjects (Fig. 8). The mean OD units of IgA anti-OmpW were numerically elevated in the CD group (median, 0.168) compared to the UC (0.097) and normal (0.082) groups. Compared to normals, this difference was significant for the CD group (P < 0.002) but not the UC group. Similarly, the frequency of positive seroreactivity was elevated in the CD group (35%, 8 of 23), compared to the UC group (13%, 3 of 23) and normal subjects (4%, 1 of 23). The frequency was significantly greater in the CD versus normal group (P < 0.009); no significant difference was observed between UC and normals (Fig. 8A). These findings reflect an antibody response which is CD associated but of moderate scale and expressed in a minority of CD patients.

View larger version (27K): [in this window] [in a new window]   FIG. 8.   Serum anti-OmpW antibody levels in normal and IBD patients. ELISA wells coated with OmpW2 (A, B, D, E, and F) or OmpW1 (B and C) were reacted with patient sera, and binding was detected by IgA- (A, B, D, E, and F) or IgG- (C) specific secondary reagents and expressed as absorbance units (OD405). Levels of pANCA activity were expressed as enzyme units (EU) per milliliter. (A) Levels of anti-OmpW2 IgA in normal, UC, and CD patients. Quantitative values for each IBD group compared to normals were compared by a Mann-Whitney (unpaired, two-tailed). Positive sera were defined as those exceeding the mean + 2 standard deviations for the normal group (dashed line). The frequencies of positive individuals in each IBD group compared to normals were compared by a Fisher's exact test. (B) Levels of IgA antibodies to OmpW1 and OmpW2. Correlation of the values was assessed by linear regression. (C) Levels of anti-OmpW1 IgG in normal, UC, and CD patients. Dashed line is the positive cutoff value, and solid lines are the arithmetic means for each group. The standard deviations for normal, UC, and CD values are: (A) 0.044, 0.026, and 0.87; (C) 0.45, 0.40, and 0.31. (D) Levels of anti-OmpW2 IgA in patients stratified for UC-pANCA positive (+) and negative () immunofluorescence. The standard deviations are normal (0.044), UC+ (0.35), UC (0.089), CD+ (0.11), and CD (0.080). (E and F) Levels of IgA anti-OmpW2 and pANCA in UC (E) and CD (F) patients. Patients were stratified for UC-pANCA positive (solid symbols) and negative (open symbols) immunofluorescence.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, the B. caccae protein identified by an IBD-associated pANCA monoclonal antibody was characterized by molecular cloning and immunologic evaluation of the recombinant gene product. The protein, termed OmpW, was found to be a new member of the TonB-linked outer membrane protein family. OmpW was also closely related to RagA, a virulence factor of the periodontal disease pathogen P. gingivalis. Evaluation of patient sera demonstrated increased anti-OmpW IgA levels in CD patients. The issues raised by this study are the relationship of OmpW to the outer membrane protein family, its relationship to candidate pANCA antigens, and the role of OmpW and B. caccae in IBD pathogenesis.

Identification of the 100-kDa antigen. OmpW was cloned from a B. caccae genomic library probed with oligonucleotide derived from 100-kDa tryptic peptide sequences. OmpW corresponds to the 100-kDa antigen by several criteria. First, the predicted molecular mass was 105 kDa, and this size was confirmed by SDS-PAGE analysis of recombinantly expressed OmpW. Second, the recombinant protein was immunoreactive with the 5-3 pANCA monoclonal antibody. Third, OmpW was encoded by a single-copy gene and was detected exclusively in B. caccae (versus other species of Bacteroides and various gram-negative coliforms). These findings agree with the size and species distribution of the 100-kDa protein, as originally defined using Western analysis of colonic bacterial species with the 5-3 pANCA monoclonal antibody (8).

Relationship to TonB-linked outer membrane protein family. The OmpW protein sequence was notable for the presence of a TonB box and other clusters of homology with TonB-linked receptors. The TonB complex is an energy transduction system which powers high-affinity active transport of certain membrane receptors across the gram-negative outer membrane (7, 25). The TonB box is a conserved domain of TonB-linked outer membrane proteins, mediating their association with the TonB complex. In Enterobacteriaceae, this system plays an important role in iron acquisition (siderophore transport), necessary for growth in iron-limited environments, including host cell tissues (23, 25).

Relationship of OmpW to other pANCA antigens. A number of bacterial and mammalian proteins have been advanced as candidate pANCA antigens. The OmpW epitope recognized by the 5-3 pANCA monoclonal antibody was localized by truncation mutants. Thus, antigenicity was lost by the truncation from OmpW2 to OmpW3 (residues 143 to 300). This is the unique segment of OmpW bearing KKAK motifs, which were previously identified as core epitopes in three other candidate pANCA antigens (mammalian histone H1, HMG1/2, and mycobacterial HupB) (9, 16). In contrast, other candidate pANCA antigens lack the KKAK motif (E. coli OmpC).

Bacteroides, OmpW, and IBD pathogenesis. The immunologic finding in this paper is that there was an elevation of anti-OmpW IgA levels in a subset of patients with CD (compared to healthy controls and UC patients). The number of subjects in the present study was insufficient to address the potential correlation of anti-OmpW IgA with clinically or genetically defined CD patient subpopulations. A larger population study and alternate randomization strategies may also resolve potential type 1 statistical error in our conclusions regarding disease association. The immunoreactivity might be secondary to mucosal damage and increased bacterial antigen delivery across the disrupted epithelial barrier to inductive immunologic sites. In support of this idea, Bacteroides spp. (including B. caccae) are a major component of the colonic microflora and typically display a commensal, nonvirulent phenotype (19). Monoassociation studies have implicated Bacteroides vulgatus as a pathogenic factor in a rat transgenic model of colitis (27). CD is distinguished not only by antibody immunoreactivity to several colonic bacterial taxa (1, 3, 50), but also by a striking expansion of T cells reactive to autologous colonic bacteria, including B. thetaiotaomicron (14). Gut-associated T lymphocytes are remarkably anergic to colonic bacterial antigens. Similar observations have been made in mouse models of IBD, notably the capacity of such T cells to establish disease by cell transfer (10). These observations strongly implicate immune responses to one or more enteric bacterial species as an important element of mucosal damage in IBD. The present work provides specific bacterial species and protein antigens for experimental evaluation.