Proteomic surveillance of retinal autoantigens in endogenous uveitis: implication of esterase D and brain-type creatine kinase as novel autoantigens

Patients

Serum samples were obtained from 36 BD patients with uveitis (31 men, 5 women; average age 36.9±10.5 years), 16 patients with VKH (12 men, 4 women; average age 39.2±12.6 years), and 17 patients with sarcoidosis (4 men, 13 women; average age 59.6±17.6 years). All patients were followed up by the outpatient clinic of the Department of Ophthalmology in Tokyo Medical University Hospital, Tokyo, Japan. Participants were considered to have BD if their symptoms fulfilled the diagnostic criteria of the Bechet’s Disease Research Committee of Japan [17]. At the time of sample collection, BD patients were determined to be in the active phase of the disease if they presented with one of the following indicators: iridocyclitis with hypopyon, dense vitreous opacity, obscure fundi observation, and retinal exudates with hemorrhages. All VKH patients showed typical exudative retinal detachment confirmed by fluorescein angiography. Sera were collected in these patients before aggressive systemic corticosteroid therapy was begun. Three of the VKH patients were examined when they were experiencing a recurrent phase of ocular inflammation. Ocular sarcoidosis was diagnosed based on the criteria established by the Japanese Committee for Diffuse Lung Disease for the systemic sarcoidosis [18,19]. In total, 64 healthy volunteers (35 men, 29 women), who were age- and sex-matched for each uveitis group, were recruited as healthy controls (HC). The investigation was conducted in accordance to the tenets of the Declaration of Helsinki.

Animals and reagents

C57BL/6 female mice, aged 6 to 9 weeks, were obtained from Sankyo Laboratory Service Corp (Tokyo, Japan). Animals were handled in accordance with the Association for the Vision and Ophthalmology statement for the use of Animals in Ophthalmic and Vision research. Horseradish peroxidase (HRP)-conjugated rabbit antimouse IgG and goat antihuman IgG were obtained from Zymed Laboratories (San Francisco, CA). Human IRBP peptide 1–20 (hIRBP-p) was synthesized by conventional solid-phase techniques on a peptide synthesizer (Takara Bio Inc., Shiga, Japan). Complete Freund’s adjuvant (CFA) and Mycobacterium tuberculosis H37Ra were purchased from Difco (Detroit, MI), Bordetella pertussis (PTX) was from Sigma-Aldrich Corp (St. Louis, MO), purified human β-actin (bAct) was from Cytoskeleton (Denver, CO), recombinant human esterase D (EsteD) was from GenWay Biotech (San Diego, CA), brain-type creatine phosphokinase (BB-CK) was from Biogenesis (Poole, England), and pETBlue-2 and Turner (DE3) pLacIcompetent cell were from Novagen (Darmstadt, Germany).

Induction and scoring of experimental autoimmune uveoretinitis

Induction of EAU in C57BL/6 mice and scoring of ocular inflammation were performed as previously described [19]. Briefly, mice were immunized subcutaneously in the occipital region with 200 μg hIRBP-p in 200 μl emulsion in CFA that had been supplemented with H37Ra to 5 mg/ml. Next, 1 μg PTX was given intraperitoneally as an additional adjuvant. On day 18 after immunization, ocular inflammation was clinically evaluated on a scale of 0 to 4 according to the previous report [20]. Briefly, minimal vasculitis with few very small lesions were scored as 0.5; mild vasculitis with 5 or less small focal chorioretinal lesions as 1; multiple (more than 5) chorioretinal lesions or severe vasculitis as 2; pattern of linear choreioretinal lesions, subretinal neovascularization, and haemorrhages as 3; and large retinal detachment and retinal atrophy as 4. The sera of the mice scoring over 1.5 were collected six weeks after immunization and used as the primary antibody in 2D-WB. Control mice were immunized with only CFA and PTX and then processed in the same way as the EAU mice.

Protein extract preparation

Mice were killed by cervical vertebra dislocation, and then retina from ten normal C57BL/6 female mice were separated immediately after enucleation and whole-tissue lysates were prepared by the freeze–thaw method using a lysis buffer that contained 7 M urea, 2 M thiourea, and 4% CHAPS [21]. The supernatant of each sample was collected, mixed together, and then stored at −80 °C until used.

Two-dimensional western blotting

Two-dimensional electrophoresis was performed as described previously [21,22]. Briefly, 100 μg extracted protein was applied on 11 cm linear Immobiline Drystrips, pH 3–10 (GE Healthcare Bio-Science Corp., Piscataway, NJ) and rehydrated at room temperature for 10 h. Isoelectric focusing (IEF) was performed using Ettan IPGphorII (GE Healthcare Bio-Science Corp.) for 12 h. After IEF, the strips were equilibrated for 30 min in a denaturation buffer that contained 6 M urea, 2% sodium dodecyl sulfate (SDS), 50 mM Tris-HCl, 30% glycerol, and 100 mg/10 ml dithiothreitol (DTT). Strips were then embedded onto 12.5% sodium dodecyl sulfate (SDS)-polyacrylamide gels, and gels were run for 2.5 h. After electrophoresis, the separated proteins on the gels were transferred onto nitrocellulose membranes. For 2D-WB, membranes were blocked in phosphate-buffered saline with 0.1% Tween20 (PBST) containing 0.5% bovine serum albumin (BSA) for 30 min and washed three times with PBST (10 min each time). Membranes were incubated for 1 h in 1:500 times diluted murine serum samples in 0.5% BSA-PBST. Next, membranes were washed with PBST before they were incubated for 1 h with 1:2000 diluted HRP rabbit antimouse IgG. Bound antibodies were visualized using diaminobenzidine (DAB; Dojindo Laboratories, Kumamoto, Japan).

Protein identification

Protein spots, which corresponded to the EAU-positive spots on the 2D-WB, were recovered from a 2D gel stained with SYPRO Ruby (Molecular Probes, Eugene, OR) and digested in gel with trypsin. The digested peptides were extracted from the gel pieces using trifluroacetic acid and acetonitrile. After centrifugation, the supernatant was recovered. After three more cycles of this extraction, the supernatant was filtered and concentrated down to 10 μl in an evaporator [23,24]. Masses of the digested peptides were determined using a matrix-assisted laser desorption/ionization-time of flight mass spectrometer (MALDI-TOF MS; Ultraflex, Bruker Daltonics, Germany). A list of the determined peptide masses were subjected to mass fingerprinting by using the MASCOT software program version 2.1.0 (Matrix, Science, London UK), in which the National Center for Biotechnology Information (Bethesda, MD) protein data base was searched [25]. Only those proteins with a Mowse score above statistical significant identity (p<0.05) were accepted as identified ones.

Recombinant protein

Based on the nucleotide sequence of the human tubulin-β (Tub-b) cDNA, the entire protein-coding region of the Tub-b cDNA from the murine retina was amplified using specific pairs of oligonucleotide primers. The amplified cDNA fragment was subcloned into the plasmid expression vector pETBlue-2, and then the protein was expressed in Escherichia coli Turner (DE3) pLacIcompetent cells. Thereby, recombinant Tub-b with a tag of six histidines in its COOH-terminal was produced. The purification of Tub-b was performed as follows by brief modification of the previous report [26]. The bacteria was harvested from a 500 ml culture by centrifugation, lysed in 10 ml of lysozyme solution (0.2 mg/ml lysozyme, 1% NP-40 in PBS), and then sonicated to fragment the bacterial DNA. 0.1 mg DNase and 10 mmol MgCl₂ were added to the sonicated sample. 25 min after, 20 ml of denaturation buffer was added to make final concentration of 8 M urea, 12 mM 2-mercaptoethanol, and 10 mM imidazol. After centrifugation, the sample was filtered, and then the protein was purified using a histidine-Ni+ affinity column (His Trap HP; GE Healthcare Bioscience Corp.) The sample was applied to the column and eluted with an elution buffer (10 mM-500 mM imidazol, 8 M Urea in PBS, pH 7.6).

One-dimensional western blotting

One-dimensional (1D)-WB was performed as follows. Each protein (4 μg) was separated by 12.5% SDS–PAGE in a buffer that contained DTT, and then transferred onto nitrocellulose membranes. The serum samples were diluted to 1:200 in 1% BSA-PBST and used for incubation with the membranes for 1 h. Incubation of the membranes with the secondary antibody and treatment with DAB were performed as described in the previous section for 2D-WB.

Enzyme-linked immunosorbent assay (ELISA)

Ninty-six-well microtiter plates (Thermo Labsystems, Franklin, MA) were coated with 5 ug/ml of EsteD, bAct, or BB-CK at 4 °C for 16h in a carbonate buffer which contained 50 mM sodium carbonate. The plates were washed three times with PBST then incubated in 1% BSA-PBST at room temperature for 1 h. The plates were again washed three times, and 50 μl of 1:200 diluted serum samples were added to each well. The plates were incubated for 2 h, and then washed five times with PBST. Following incubation with 1:2000 HRP-goat anti-human IgG for 1 h, the plates were washed eight times with PBST. Color development was achieved by adding 100 μl of 0.04% o-phenylenediamine as a substrate. The reactivity of the serum sample to each protein was expressed using the arbitrary binding units, calculated according to the following formula: sample (binding units)=(OD sample/[mean OD control sera + 2SD of control sera] × 100), where OD is the optical density. According to the formula, 100 binding units were used as the cutoff point.

IgG1 and IgG2a subclasses of anti-EsteD and anti-BB-CK antibodies were determined using the mouse IgG1 and IgG2a quantitation kits (Bethyl laboratories, Montgomery, TX). Briefly, EsteD or BB-CK was coated onto the microtiter plates as described in the previous section, and serum samples diluted to 1:100 were used for the detection of IgG1 and undiluted samples were used for the detection of IgG2a.

Statistical analysis

Fisher’s exact probability test was used to compare differences in the prevalence of autoantibodies between the HC and each disease group. Fisher’s exact probability test or Student’s t-test was used to compare differences in the clinical parameters or clinical symptoms between the antibody-positive and -negative patient groups.

Detection of retinal autoantigens involved in experimental autoimmune uveoretinitis

EAU was induced in C57BL/6 mice by immunization with hIRBP-p. Clinical severity was evaluated 18 days after immunization. As a result, approximately 70% of hIRBP-p immunized mice developed ocular inflammation, with the clinical score of more than 1.5. We used sera from ten EAU and ten control mice to comprehensively detect retinal autoantigens by 2D-WB, and we obtained a total of 151 protein spots that reacted positively with any of the tested sera. Though the control and EAU sera showed similar positive pattern on 2D-WB, we detected seven candidate autoantigens (Figure 1A and Table 1) by comparing the positive rate of each spot between the EAU and control by referring to the respective p value. For this purpose, the p value for a protein spot reacting positively to three EAU sera and no control sera was determined to be 0.105, and we chose all the spots as candidate autoantigens whose p values were 0.105 or less. Spot positions of five major candidate autoantigens are shown in Figure 1B.

The MALDI-TOF MS and MS/MS analysis were used to identify the seven candidate autoantigens and the results of MS/MS analysis are summarized in Table 2. One spot (number 103), which reacted with 100% of the EAU sera but none of the control sera, was identified as the commonly known retinal binding protein 3 (rbp3). Rbp3 contains the 20 amino acid sequence of the hIRBP-p used for the induction of EAU. Thus, the positive reaction to this protein spot in all the EAU sera was due to the initial immunoresponse against the hIRBP-p used for immunization. It is therefore possible that the remaining six proteins were candidate retinal autoantigens induced as a result of EAU-mediated retinal destruction.

Confirmation of immunoreactivity of candidate autoantigens in experimental autoimmune uveoretinitis

Among the candidate autoantigens, we chose the four antigens, bAct, EsteD, Tub-b, and BB-CK, for further analysis. We selected bAct, EsteD, and Tub-b because their p values were low (Table 1), and BB-CK because it was a relatively retina- and brain-specific protein. To confirm the aforedescribed results of the 2D-WB, we evaluated autoantibody titers against recombinant bAct, EsteD, Tub-b, and BB-CK in the EAU mice by using 1D-WB and ELISA. 1D-WB analysis showed that 8/10, 6/10, and 4/10 EAU mice were positive for autoantibodies against the antigens of bAct, EsteD, and BB-CK, respectively (Figure 2A,B,C); in contrast, none of the control mice reacted positively with any of these three antigens. Similarly, ELISA analysis showed that 4/10, 7/10, and 4/10 EAU mice were positive for autoantibodies against the antigens of bAct, EsteD, and BB-CK, respectively (Figure 2D,E,F). The results of the 2D-WB, 1D-WB, and ELISA corresponded closely to one another. Similar ELISA results were obtained using another set of control and EAU sera collected six weeks after immunization (data not shown). In addition, when we compared the antibody level against bAct, EsteD, and BB-CK using the murine sera collected three, six, and ten weeks after the immunization, we found that the sera collected six weeks after the immunization had the highest levels of antibodies against these three antigens (data not shown). As Tub-b did not react with the EAU sera in either 1D-WB or ELISA (data not shown), it was not used for any further experiments.

IgG subclass of anti-EsteD and anti-BB-CK antibodies

IgGs are subclassified depending on the types of Th immune responses. Th1-mediated immune response promotes IgG2a production, while Th2-mediated immune response promotes IgG1 production. Thus, we next determined the subclass of the anti-EsteD and the anti-BB-CK antibodies. Interestingly, most of the anti-EsteD and anti-BB-CK antibodies in all the tested EAU serum samples were IgG1, and the amount of IgG2a type of the anti-EsteD and anti-BB-CK antibodies were much lower than their respective IgG1 antibodies (Figure 3). All control serum samples collected from the mice immunized without hIRBP-p were negative for both IgG1 and IgG2a (data not shown).

Autoantigenicity in human uveitis

As shown in the previous section, bAct, EsteD, and BB-CK were confirmed as retinal autoantigens in the murine model of autoimmune uveitis. Because the amino acid sequence of these three autoantigens are highly conserved in human and mouse (approximately 91–100%), we next examined whether autoantibodies against these antigens could be detected in human endogenous uveitis.

We examined the immunoreactivity against bAct, EsteD, and BB-CK by ELISA in 36 BD, 16 VKH, and 17 sarcoidosis patients with uveitis and age- and sex-matched HC. A profile of these patients is given in Table 3. Patients with the medical history of all four symptoms associated with BD (oral aphthae, skin lesion, genital ulcer, and uveitis) were 36.1%. For the VKH group, patients with active ocular inflammation before the starting of aggressive corticosteroid therapy were used, because high dose corticosteroid therapy would dramatically suppress immune response and antibody titers would also be strongly suppressed. As shown in Figure 4, EsteD reacted with the sera from 25% of BD, 25% of VKH, and 17.6% of sarcoidosis patients, and the percentages were significantly higher in BD (p=0.016) and VKH (p=0.015) groups than in the control group. BB-CK reacted with sera from 5.6% of BD, 25% of VKH, and 38.4% of sarcoidosis patients. Percentages were significantly higher in VKH (p=0.015) and sarcoidosis (p=0.020) groups than in the control group. One sample each in BD and sarcoidosis and two samples in VKH were positive for both EsteD and BB-CK. However, bAct did not react positively with any of the uveitis sera with any statistical significance (data not shown).

Next, we compared the clinical parameters of patients between the anti-EsteD antibody and anti-BB-CK antibody positive- and negative-patient groups in each uveitis group. As shown in Table 4, every anti-EsteD antibody positive VKH patient in this group was female. Results from this group were statistically significant compared to the anti-EsteD antibody negative VKH group (p=0.034). In the case of the BD or sarcoidosis patients, there were no statistically relevant correlations between the examined parameters (age, sex, disease duration, visual acuity, activity of ocular inflammation, serum IgG levels, systemic medications at the time of sample collection, presence of HLA-B51, or involvement of systemic diseases) and the presence of either anti-EsteD or anti-BB-CK antibody (results not shown).