The currently used empirical criteria for the diagnosis of BD in this study were the criteria proposed by the International Study Group for BD (abbreviated as 'International Criteria') 2. The study subjects of 39 Chinese BD patients comprised 17 males and 22 females, mean age 37 ± 11.3 years old, who were divided into two subgroups: 25 typical BD patients (Group I, satisfying the International Criteria) and 14 clinically diagnosed BD patients who had recurrent oral ulcers and one of the symptoms of genital ulcers, eye symptoms or skin lesions as defined by the International Criteria, as well as additional symptom(s) closely related to BD as listed in the International Criteria, that is, gastrointestinal ulcerations, deep vein thrombosis or arthralgia/arthritis without evidence that the latter symptoms might be related to any other disease (Group II, defined as 'probable BD' in this study). Disease controls included 10 patients with SLE and 10 with SjS, all satisfying corresponding international classification criteria. All BD patients and disease controls involved in the study were patients treated at the Rheumatology Department of Ren Ji Hospital, Shanghai, China, where their clinical data and serum samples were collected. Twenty normal control sera were randomly selected from healthy blood donors working in the same hospital. This study was approved by the institution review board of Ren Ji Hospital which is affiliated with Shanghai Second Medical University, and each patient involved gave informed consent. All serum samples were preserved at -20°C or -70°C until use.

HeLa (ATCC CCL 2.2) and T24 (human transitional cell bladder carcinoma) were obtained from the American Type Culture Collection (Rockville, MD, USA). A bovine aortic endothelial cell line was kindly provided by Dr Eugene G Levin from the Scripps Research Institute (La Jolla, CA, USA). Cells were cultured in DMEM containing 10% calf serum, harvested and extracted in Buffer A (150 mM NaCl, 10 mM Tris-HCl, pH7.2, 0.5% Nonidet P-40) with protease inhibitor (Complete™; Boehringer Mannheim, Indianapolis, IN, USA). For the preparation of whole cell extract, 10 volumes of Laemmli gel sample buffer 18 were added to the cell pellet, boiled for 3 min and stored at -20°C until use.

Whole cell lysates from bovine aortic endothelial cell, HeLa and T24 cells were resolved individually by discontinuous 7.5% gel SDS-PAGE according to Laemmli's method 18. Immunoblotting was performed as described by Towbin et al. 19 with modifications. Nitrocellulose strips were blocked with 3% nonfat milk in PBS containing 0.05% Tween-20 (PBS-T) and then incubated with BD patient sera and normal control sera (1:100 dilution) at room temperature for 1 h. Filters were washed extensively with PBS-T to remove any unbound antibodies. Bound antibodies were detected with polyvalent, peroxidase-conjugated goat anti-human Ig and visualized by incubating the nitrocellulose strips in chemiluminescent reagents (NEN Life Science Products Inc., Boston, MA, USA) and exposing to Kodak XAR-5 films.

Serum from a BD patient showing the highest antibody titer in immunoblotting was selected as a probe and used at a dilution of 1:300 for initial immunoscreening of approximately 10⁶ recombinants from a T24 cDNA expression library. The latter was constructed in λZAPExpress vector (Stratagene, La Jolla, CA, USA) and screened as previously described 202122. All screenings were performed on duplicate isopropyl β-D-thiogalactoside (IPTG) pre-impregnated nitrocellulose filters, and immunoreactive clones were detected by chemiluminescence. Positive phages were subsequently plaque purified to 100% by two repeated rounds of screening at low plaque densities. Before screening the cDNA library, the BD serum was extensively adsorbed against bacteria and wild-type λZAP phage mixture to reduce background binding.

Purified candidate plaques were subcloned in vivo into pBK-CMV plasmids using ExAssist™ helper phage as recommended in the manufacturer's instructions (Stratagene). The recombinant pBK-CMV plasmids were then purified using QIAprep Spin Minprep Kit (Qiagen, Valencia, CA, USA). Restriction enzyme digestion of plasmids with EcoRI and XhoI and electrophoresis in a standard 1.0% agarose gel was used to analyze the length of cDNA insert of each candidate plasmid. The complete nucleotide sequence was determined using Bigdye terminator sequencing and a semi-automated sequencer model 377 (ABI, Foster City, CA, USA). Both nucleotide and deduced amino acid sequences were analyzed for similarity with known sequences using BLAST search 23 and ExPASy Proteomics tools http://www.expasy.ch/www/tools.html. Secondary structure analysis for coiled-coil motifs was conducted with the software program COILS 24.

Candidate cDNA clones were used as templates for in vitro transcription and translation and the products were used as substrates for immunoprecipitation to confirm the specificity of reaction with BD sera. In brief, 1 μg of the pBK-CMV plasmid identified in the screening outlined above was added as template in a 50-μl reaction for the coupled in vitro transcription and translation reaction with a rabbit reticulocyte lysate system (Promega, Madison, WI, USA) in the presence of ³⁵S-methionine (Trans-³⁵S label; ICN Biochemicals, Costa Mesa, CA, USA) and RNasin^® Ribonuclease Inhibitor (Stratagene) as recommended by the manufacturer (Promega). Translation was carried out at 30°C for 1.5 h. Products were analyzed in a 12.5% gel SDS-PAGE and stored at -80°C for further immunoprecipitation analysis. The in vitro translation proteins were examined for reactivity by sera using immunoprecipitation described 825.

Initial examination of a group of 39 BD patients using indirect immunofluorescence (IIF) on a HEp-2 cell substrate did not yield any characteristic nuclear or cytoplasmic staining patterns. BD is thought by some to be a vasculitic disease involving pathophysiology of endothelial cells, and antibody to endothelial cell antigen (AECA) has been reported. Reports on the prevalence of AECA have varied largely and alpha-enolase was reported as one of the putative target antigens 26. In this study, the use of bovine aortic endothelial cells as substrate for IIF did not provide any additional data. However, Western blot analysis of the BD sera began to show some interesting autoreactivity using cell lysates from both HeLa and bovine aortic endothelial cells. HeLa cells were initially used for this analysis because they are commonly used in the laboratory as Western blot substrate. Fig. 1 illustrates the common reactivity to 49 kDa and 120 kDa proteins in the endothelial cell lysates. These antigens were also detected in HeLa and T24 cells; the latter cell line was analyzed because our laboratory at The Scripps Research Institute has produced an excellent expression cDNA library from the T24 line and the positive result with the T24 cell extracts allowed us to screen the T24 library. Ig isotype analysis showed that all reactivity was largely IgG antibodies. Since the 49 kDa and 120 kDa bands were observed in cell extracts from bovine as well as human cell lines, these autoantigens might be evolutionarily conserved.

In total, nine out of 39 BD sera (23%) had autoantibody to the 49 kDa antigen and eight (20%) to the 120 kDa antigen. Four BD sera (10%) reacted with both proteins. Additionally, sera that showed common reactivity to the 120 kDa protein also demonstrated a common band that migrated at ~150 kDa, although it appeared weaker than the 120 kDa band. These antigens appeared to have different molecular weights than those of the known autoantigens in systemic rheumatic diseases. In addition, other reactive bands were detected but they were not as commonly shared as the 49 kDa and 120 kDa bands. The 49 kDa protein was shown to be distinct from 48 kDa SS-B/La or 50 kDa Jo-1 proteins (Fig. 1). The 120 kDa antigen was also shown to migrate differently from alanyl tRNA synthetase in another Western blot analysis (data not shown) and did not share any apparent crossreactive epitopes with the 49 kDa antigen. Western blot analyses of 20 normal control sera did not show the reactivities observed with BD sera. In order to further characterize these autoreactivities, a serum sample from the Group I definitive BD patients with the strongest reactivity to 49 kDa and 120 kDa antigens (Fig. 1, lane 2) was selected as antibody probe for expression library screening.

After screening 500,000 clones from the T24 cell λZAPExpress expression library, seven immunoreactive clones were isolated and plaque purified in two to three rounds to achieve 100% homogeneity. The cDNA inserts were subcloned in vivo into pBK-CMV plasmids, analyzed by restriction digestion using EcoRI and XhoI enzymes, and submitted to direct nucleotide sequencing across the polylinker arms. The cDNA inserts represented six independent clones designated BD41 (identical to BD44), BD481, BD42, BD47, BD482 and BD49. Their identities were established as overlapping partial cDNAs of human kinectin, ranging from 1.9 kb to 3 kb (Fig. 2a). The full-length human kinectin (GenBank accession number NM_18292627) has 4,816 bases containing an open reading frame coding 1,357 amino acid residues with molecular mass 156 kDa. All six cDNAs lacked the 5' portion of the kinectin sequence to different degrees but spanned a sequence of kinectin that extended to the 3'-untranslated region. Secondary structure analysis of kinectin protein using the program COILS identified a long region of α-helical coiled-coil domain that extended from amino acid residue 327 to the C-terminus (Fig. 2a, hatched boxes). In vitro coupled transcription and translation of BD44 and BD42 clones directed the synthesis of [³⁵S]-methionine-labeled polypeptides that migrated at 95 and 60 kDa, respectively, in addition to smaller polypeptides (Fig. 2b). These products had predicted molecular weights of 103 kDa and 75 kDa.

Kinectin was initially identified in chick embryo brain microsome as an integral membrane protein anchored in endoplasmic reticulum and involved in kinesin-driven vesicle motility along microtubules 2829. Kinectin consists of a 120-kDa and a 160-kDa polypeptide interacting through the α-helical coiled-coil domain to form a heterodimer 30. The full-length kinectin is the 160 kDa polypeptide containing an N-terminal transmembrane helix followed by a bipartite nuclear localization sequence and two C-terminal leucine zipper motifs. We presume that the 120 kDa polypeptide detected in Western blot (Fig. 1) is the truncated version of the 160-kDa polypeptide, lacking the N-terminal first 232 amino acids 30. The N-terminus of the 160-kDa polypeptide consists of a transmembrane domain that anchors kinectin to endoplasmic reticulum 3031. This 120 kDa polypeptide is probably the predominant form detected in the Western blot analysis (Fig. 1) because of its preferential solubility due to the omission of the N-terminal transmembrane domain.

Other functions for kinectin have been reported. Yeast two-hybrid screen studies from several laboratories have revealed the interaction of the Rho family of GTPase with kinectin, and have shown the functional links among RhoG, kinectin and kinesin, with kinectin as a key effector of RhoG microtubule-dependent cellular activity 32. Kinectin was also identified as an important constituent of integrin-based adhesion complexes, which link integrins to the cytoskeleton and recruit signaling molecules 33. A new study reported that a kinectin isoform lacking a major portion of the kinesin-binding domain is very probably the most conservative form of kinectin; it does not bind kinesin but act as a membrane anchor for the translation elongation factor-1 delta in the endoplasmic reticulum 34.

The in vitro [³⁵S]-methionine-labeled translation product of BD44, representing the largest recombinant kinectin fragment available, was used as the antigen substrate in an immunoprecipitation assay. Out of 39 BD patient sera, nine (23%) recognized the BD44 translation product (Fig. 3), whereas sera from 20 normal controls, 10 SLE and 10 SjS patients did not show reactivity. Among the nine anti-kinectin positive patients, six (6/25, 24%) were from Group I (definitive BD) including the BD patient whose serum was used in the immunoscreening of expression cDNA library, and three (3/14, 21.4%) patients were from the Group II (probable BD) in this study. According to the Fisher Exact Probability calculation (P = 1.00), there is no statistically significant difference for antibody to kinectin between the two groups. The combined data substantiated the finding that kinectin is an autoantigen that can be recognized by sera from 23% of Chinese BD patients in this study with at least one immunoreactive region or autoepitope residing within the BD44 encoded polypeptide.

Currently, there are more than six diagnostic/classification criteria for BD, among which the International Criteria have been applied most extensively due to its relatively high sensitivity (91%) and specificity (96%) 2. As discussed above, differential diagnosis of BD might be confusing in clinical practice since no specific laboratory test is available, and some patients may have symptoms and signs strongly suggestive of BD but do not fully satisfy the International Criteria, as in the Group II (probable BD) patients in our study group. A number of investigators have pointed out that a comprehensive analysis of the clinical data for a given patient is very important for correct clinical diagnosis of BD, and that classification/diagnosis criteria, including the International Criteria, should be followed but should not be exclusive. The observation that three out of 14 patients in the probable BD group also had antibody to kinectin and the similar percentage of positive reactors between this group and Group I (21.4% versus 24%) supports this notion. The further use of non-clinical parameters such as immunological biomarkers as adjuncts to identify BD patients could be of help in the classification of this disease entity

While our work was ongoing, anti-kinectin antibodies were reported in sera from patients with hepatocellular carcinoma (HCC) 3536 and aplastic anemia 3738. The first HCC report 35 identified kinectin as a tumor-associated antigen from the screening of an autologous cDNA library constructed from the cancer of a 30-year-old patient from Guangxi, China. This report stated that four out of five HCC patients tested were positive for anti-kinectin antibody 35. In 2004, another laboratory also reported the cloning of kinectin as a tumor-associated antigen from a (presumably) different 30-year-old Chinese HCC patient 36. In contrast, anti-kinectin antibodies were not detected in other studies of HCC patients associated with our laboratory 3940. The reports of anti-kinectin antibodies in aplastic anemia are also very interesting 3738. The initial report by Hirano et al. identified kinectin by screening an aplastic anemia patient for candidate antigens using a Clontech human fetal liver cDNA expression library and it was concluded that seven out of 18 aplastic anemia patients were positive for anti-kinectin while none of the normal or disease controls had this antibody 37. In their recent report, Hirano et al. reported that anti-kinectin antibodies were found in 39% of aplastic anemia patients from the United States but only in three out of 30 (10%) cases in Japan 38. In our study reported here, kinectin antibodies were only detected in BD patients and not in normal controls and SLE and SjS disease controls. None of the BD patients with anti-kinectin had signs of HCC or aplastic anemia at the time of diagnosis and at up to 4 years of follow-up. Mapping of epitope(s) recognized by anti-kinectin antibodies may shed light on the question of whether different autoepitopes reside within the kinectin molecule recognized by sera from different diseases.

We have recently reviewed the literature on the growing number of cytoplasmic autoantigens rich in α-helical coiled-coil domains as typified from our study of Golgi autoantigens 41. Golgi autoantigens are generally high molecular weight proteins between 100 and 350 kDa and rich in coiled-coil domains in the central region with non-coiled-coil or globular domains at both N and C termini. Golgi autoantigens are displayed on the cytoplasmic face of the Golgi complex and are not localized to apoptotic blebs during apoptosis 42. Giantin, the highest molecular weight Golgi autoantigen reported, is the predominant target of human anti-Golgi complex antibodies and multiple non-cross-reactive epitopes have been mapped spanning the 350 kDa protein 43. Other high molecular weight autoantigens with similar features have been reported in cytoplasmic and mitotic organelles suggesting that these selected proteins become autoimmunogenic based on their subcellular association and molecular features 41. For example, in the endosomal compartment, the two known autoantigens are early endosomal protein EEA1 (180 kDa) 44 and CLIP-170 (170 kDa) 45. There is also a series of centrosomal autoantigens identified as coiled-coil-rich proteins including pericentrin, a 220 kDa protein 46, ninein, a protein with alternatively spliced products of 245 and 249 kDa 47, Cep250 (250 kDa) and Cep110 (110 kDa) 48. Centromere autoantigens have been described but the two interesting ones related to this discussion are CENP-E 49 and CENP-F 50; both are high molecular weight proteins (312 to 400 kDa) and have the same type of overall structure as discussed above. NuMA is another large coiled-coil protein located at the mitotic spindle pole and is the most common target autoantigen in sera with mitotic spindle apparatus staining 51. Non-muscle myosin (~200 kDa) is a cytoskeletal autoantigen 52 that falls in the same group of high molecular weight and coiled-coil-rich autoantigens. These endosomal, centrosomal, mitotic apparatus and intracellular autoantigens are, like the golgins, proteins with high molecular weights and an overall high content of coiled-coil domains. The combination of these two physical features in autoantigens may contribute to the induction and production of autoimmune antibodies in certain disease states. Kinectin is an integral membrane protein largely confined to the endoplasmic reticulum 2831 and it fits into this new category of autoantigens that are large coiled-coil rich proteins (≥100 kDa) in the cytoplasm.