Synovial tissues were obtained from patients with RA (56 ± 14 years old (mean ± SD); n = 16) or OA (73 ± 6 years old; n = 20) at total knee arthroplasty. Diagnosis of the patients with RA or OA was based on the 1987 revised American Rheumatism Association Criteria for RA 20 and the American Rheumatism Association Criteria for OA 21. Synovial specimens were fixed with periodate-lysine-paraformaldehyde, and paraffin sections stained with hematoxylin and eosin were analyzed by light microscopy according to our grading system of synovial lining cell hyperplasia, cellular infiltration and fibrosis 22. For the experimental use of the surgical specimens, written informed consent was obtained from the patients according to the hospital ethical guidelines.

Total RNA was extracted directly from RA (n = 16) and OA (n = 20) synovial tissues and evaluated by using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) as descried previously 9. By using a random hexamer of oligonucleotides (Takara Bio Inc., Otsu, Japan), cDNAs were prepared from total RNA with SuperScript II reverse transcriptase (Life Technologies Inc., Rockville, MD, USA). The reaction product was subjected to reverse transcription (RT)-PCR analysis on the expression of ADAMs 8, 9, 10, 12, 15, 17, 20, 21, 28 and 30, VEGFR-1, VEGFR-2, neuropilin-1 and β-actin for 25–30 cycles. PCR was carried out in 50 μl reaction volume containing 800 nM of each primer, 220 μM of dNTPs and 1 unit of ExTaq DNA polymerase (Takara Bio Inc.). The thermal cycle was 1 minute at 94°C, 1 minute at 62°C for ADAMs 8, 9, 10, 12, 17, 20, 21, 28 and 30, 67°C for ADAM15, 64°C for VEGFR-1, 63°C for VEGFR-2 and neuropilin-1 and 65°C for β-actin, and 1 minute at 72°C, followed by 3 minutes at 72°C for the final extension. The nucleotide sequences of the PCR primers and the expected sizes of the amplified cDNA fragments are shown in Table 1. Aliquots of the PCR products were electrophoresed in 2% agarose gels, and stained with ethidium bromide. For positive controls, total RNA was extracted from cancer cell lines as described previously 9. The specific amplification of these ADAMs, VEGFRs, neuropilin-1 and β-actin was confirmed by direct sequencing of the PCR products.

The mRNA expression levels of ADAM15 were evaluated in a TaqMan real-time PCR assay using the ABI Prism 7000 sequence detection system (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's protocols. Cycling conditions were 50°C for 2 minutes, 95°C for 10 minutes, and then 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. Primers were designed and selected using Primer Express software (Applied Biosystems). Sequences of the primers and TaqMan probe for ADAM15 were as follows: forward primer, 5'-GGCAATCGAGGCAGCAAAT-3'; reverse primer, 5'-TGGTGGAGATCAGCCCAAAC-3'; and TaqMan probe, 5'-FAM-CAGCTGTCACCCTCGAAAACTTCCTCC-TAMRA-3'. The relative quantification value of ADAM15 was normalized to an endogenous control, 18S ribosomal RNA, after confirming that ADAM15 and ribosomal 18S cDNAs were amplified with the same efficiency according to the manufacturer's protocol. The total gene specificity of the nucleotide sequences chosen for the primers and probe and the absence of DNA polymorphisms were ascertained using BLASTN and Entrez from the National Center for Biotechnology Information web site 23.

Paraffin sections of the RA synovial tissues (n = 5) were used for in situ hybridization of ADAM15 according to a modification of our methods used previously 19. Briefly, the sections were treated with proteinase K (5 μg/ml; Sigma-Aldrich Inc., St Louis, MO, USA) in 10 mM Tris-HCl buffer, pH 8.0, and 1 mM ethylenediaminetetraacetic acid, and endogenous peroxidase was blocked with 0.3% hydrogen peroxide in methanol. Single-stranded sense and anti-sense digoxigenin-labeled RNA probes were generated by in vitro transcription from the cDNA encoding ADAM15, nucleotides 1091 to 1331 (241 bp), using the DIG RNA labeling kit (Roche Diagnostics GmbH, Mannheim, Germany). BLASTN searches were performed to confirm the specificity of the probes. Hybridization with the probes was performed at 42°C for 16 h, and the sections were washed in 2× standard saline citrate/50% formamide, followed by digestion with 10 μg/ml ribonuclease A (Wako Pure Chemical Industries, Osaka, Japan). After washing once in 2× standard saline citrate and twice in 0.2× standard saline citrate and blocking nonspecific binding with blocking solution (DakoCytomation Norden A/S, Glostrup, Denmark), they were incubated with mouse anti-digoxigenin antibody (1/750 dilution; Roche Diagnostics GmbH), and subjected to the following steps using the Catalyzed Signal Amplification System (DakoCytomation Norden A/S) according to the manufacturer's protocol. Counterstaining was performed with hematoxylin.

A monoclonal antibody against human ADAM15 was developed using the synthetic peptide corresponding to the amino acid sequence of the cysteine-rich domain (residues 596 to 612, RDLLWETIDVNGTELNC) of human ADAM15 as an antigen according to our methods 24. Clones were screened by enzyme-linked immunosorbent assay using the peptide, and a clone 240-2C7 was selected as a candidate. The specific reactivity of the antibody was further evaluated by immunoblotting. The cysteine-rich domain of ADAM15 with FLAG-tag was expressed in Escherichia coli DH5α (Takara Bio Inc.) by transfecting with the expression vector pFLAG-ADAM15, which was prepared by inserting a cDNA fragment encoding the human ADAM15 cysteine-rich domain (nucleotides 1887 to 1937) 25 into pFLAG-CTC vector (Sigma-Aldrich Inc.). As a negative control, the vector pFLAG-CTC alone was transfected to DH5α (mock transfectants). Cell lysates were subjected to SDS-PAGE (15% total acrylamide) under reducing conditions. The resolved proteins were then transferred to polyvinylidene difluoride membranes (ATTO, Tokyo, Japan), which were reacted with anti-FLAG antibody (1 μg/ml; Sigma-Aldrich, Inc.), anti-ADAM15 antibody (1 μg/ml; 240-2C7) or non-immune mouse immunoglobulin G (IgG) (1 μg/ml; DakoCytomation Norden A/S) after blocking with 5% skim milk. They were then incubated with horseradish peroxidase-labeled anti-mouse IgG (1/5000 dilution; Amersham Biosciences Corp., Piscataway, NJ, USA). Immunoreactive bands were detected with ECL Western blotting reagents (Amersham Biosciences Corp.). As shown in Fig. 1, two protein bands of 15 kDa and 10 kDa were detected with anti-FLAG antibody in the cell lysates of ADAM15 transfectants (lane 1) but not mock transfectants (lane 2). On the other hand, anti-ADAM15 antibody (240-2C7) selectively reacted with the band of 15 kDa in the ADAM15 transfectants (Fig. 1, lane 3) but not mock transfectants (lane 4). The molecular weight of the immunoreactive 15-kDa band corresponds to that of the cysteine-rich domain of ADAM15 25. Importantly, the immunoreactivity of the 15-kDa band was blocked after absorption of the antibody with the antigen peptide (Fig. 1, lanes 5 and 6). Blotting with non-immune mouse IgG showed no reactive bands (data not shown). This indicates that the monoclonal antibody (240-2C7) is monospecific to the cysteine-rich domain of ADAM15.

RA synovial tissues (n = 5) were homogenized on ice in a lysis buffer (50 mM Tris-HCl buffer, pH 7.5, 150 mM NaCl, 10 mM CaCl₂ and 0.05% Brij35) containing a cocktail of proteinase inhibitors (Roche Diagnostics, GmbH). Supernatants of the homogenates were subjected to SDS-PAGE (10% total acrylamide) under reducing conditions, transferred onto polyvinylidene difluoride membranes and reacted with anti-ADAM15 antibody (240-2C7; 1 μg/ml) or non-immune mouse IgG (1 μg/ml) after blocking with 5% skim milk. They were then incubated with horseradish peroxidase-labeled anti-mouse IgG and immunoreactive bands were detected with ECL Western blotting reagents as described above.

Paraffin sections of the RA synovial samples were treated with 0.3% H₂O₂ and 10% normal goat serum to block endogenous peroxidase and non-specific binding, respectively. As antigen retrieval, the sections were subjected to microwave treatment at 500 W for 10 minutes in 10 mM citrate buffer, pH 6.0. They were then treated with mouse anti-ADAM15 antibody (240-2C7; 20 μg/ml), rabbit anti-VEGFR-2 antibody (Flk-1; 5 μg/ml; Santa Cruz Biotechnology, Santa Cruz, CA, USA), rabbit anti-von Willebrand factor (vWF; 15 μg/ml; DakoCytomation Norden A/S) or mouse anti-CD31 antibody (8 mg/ml; DakoCytomation Norden A/S). After the reaction with goat anti-mouse IgG or goat anti-rabbit IgG conjugated to peroxidase-labeled dextran polymer (no dilution; En Vison+ Mouse or En Vison+ Rabbit; DakoCytomation Norden A/S), the color was developed with 3,3'-diaminobenzidine tetrahydrochloride in 50 mM Tris-HCl buffer, pH 7.6, containing 0.006% H₂O₂. Counterstaining was performed with hematoxylin. As for a control, sections were reacted by replacing the first antibodies with non-immune mouse IgG or rabbit IgG.

Vascular density in RA and OA synovial tissues was evaluated by the morphometric analysis of RA and OA tissue sections immunostained with anti-CD31 antibody without any clinical information on each sample. Four fields were selected at random and vessels with a distinct lumen were counted to calculate the number of vessels per square millimeter as we described previously 14. The average vascular density (vessels/mm²) from the fields for each patient was processed for further statistical analysis.

RA synovial fibroblasts (SFs) were prepared from RA synovial tissues obtained at total knee arthroplasty. The tissues were minced and incubated with 0.04% bacterial collagenase type I (Worthington Biochemical Corp., Freehold, NJ, USA). Isolated cells were seeded in culture dishes and maintained in DMEM (Life Technologies) supplemented with 10% fetal bovine serum (Life Technologies) at 37°C in humidified 5% CO₂ in air. After the cells were cultured in confluence, they were trypsinized and reseeded in culture dishes. RA SFs at 5–9 passages were used for experiments. Human umbilical vein endothelial cells (HUVECs 7943; Cambrex Co., East Rutherford, NJ, USA) were grown in medium EBM-2 supplemented with EGM-2 (Cambrex Co.).

In order to examine the expression of VEGFR-2 and ADAM15 and exclude the possibility of contamination of cultured RA SF by endothelial cells, RA SFs at 5–9 passages were seeded on Lab-Tek II chamber slides (Nalge-Nunc International, Naperville, IL, USA) and subjected to immunohistochemistry for VEGFR-2, ADAM15, CD31 and vWF as described above. For a positive control, HUVECs were immunostained with these antibodies. In addition, mRNA expression of CD31 and vWF in cultured RA SFs was examined by RT-PCR using the PCR primers (Table 1).

RA SFs were plated on a 60 mm dish at a density of 3 × 10⁵ cells/dish in DMEM supplemented with 10% fetal bovine serum. The culture media were replaced with serum-free DMEM containing 0.2% lactalbumin hydrolysate and starved for 24 h before they were used in experiments. The cells were treated with tumor necrosis factor-α (TNF-α; Dainippon Pharmatheutical, Osaka, Japan), IL-1α (Dainippon Pharmatheutical), transforming growth factor-β (TGF-β; R&D Systems, Minneapolis, MN, USA; 0, 0.1, 1 or 10 ng/ml) or recombinant VEGF₁₆₅ (R&D Systems; 0, 1, 10 or 50 ng/ml) for 24 h. For co-stimulation of RA SFs with TNF-α and VEGF₁₆₅, the cells were first starved with serum-free DMEM containing 0.2% lactalbumin hydrolysate for 24 h, treated with TNF-α (1 or 10 ng/ml) for 24 h, and then stimulated with VEGF₁₆₅ (40 ng/ml) for 24 h. HUVECs were also stimulated with these cytokines or growth factors for 24 h after being starved with serum-free medium EBM-2 containing 1% bovine serum albumin for 24 h.

To block the signaling of VEGF₁₆₅, RA SFs that had been stimulated with TNF-α (10 ng/ml) for 24 h were incubated with SU1498 (1 or 10 μM; Calbiochem, San Diego, CA, USA), a selective VEGFR-2 tyrosine kinase inhibitor 2627, for 30 minutes and then treated with VEGF₁₆₅ (40 ng/ml) for 24 h. HUVECs were also treated with SU1498 in a similar way except for no treatment with TNF-α. To exclude the possible involvement of VEGFR-1 in ADAM15 expression, RA SFs and HUVECs were stimulated with recombinant human placenta growth factor (PlGF; 1, 10 or 50 ng/ml; R&D Systems), which selectively binds to VEGFR-1 28.

Comparisons between two independent groups were determined by Mann-Whitney U test. Spearman's rank correlation was used for analysis of the relationship between relative ADAM15 mRNA expression and vascular density. P-values less than 0.05 were considered significant.

The mRNA expression of 10 different ADAMs (8, 9, 10, 12, 15, 17, 20, 21, 28 and 30) with a putative metalloproteinase motif was screened by RT-PCR analysis in eight RA and eight OA synovial samples. ADAM9, ADAM10 and ADAM17 were expressed in more than 88% of RA samples, but their expression was also observed in more than 75% of OA samples (Fig. 2). ADAM12 was expressed in 38% and 13% of RA and OA samples, respectively. ADAMs 8, 20, 21, 28 and 30 were expressed in less than 13% of both RA and OA samples. On the other hand, ADAM15 was intensely expressed in all the RA synovial samples, whereas it was detected in 63% of OA samples (Fig. 2). Because of the more selective expression of ADAM15 in RA than in OA, we focused on ADAM15 for further studies. When the expression was examined in a larger number of RA and OA samples, ADAM15 was detected in 100% of the RA samples (16 of 16 cases) and in 60% of the OA samples (12 of 20 cases) (data not shown). By real-time quantitative PCR analysis, the expression levels (ratio of ADAM15 to ribosomal 18S RNA) were significantly higher in RA samples (0.344 ± 0.276; n = 10) than in OA samples (0.091 ± 0.030; n = 10) (p < 0.01; Fig. 3).

Cells expressing ADAM15 mRNA were examined by in situ hybridization. Synovial lining cells, endothelial cells of blood vessels and macrophage-like cells in the sublining layer were labeled with the anti-sense RNA probe (Fig. 4a), whereas the sense probe gave only a background signal in these cells (Fig. 4b). Immunohistochemical studies showed that ADAM15 was localized to synovial lining cells, endothelial cells of blood vessels and macrophage-like cells in the sublining layer of the RA synovium (Fig. 5a,b), confirming the findings from in situ hybridization. No staining was obtained with non-immune IgG (Fig. 5f). VEGFR-2 was immunolocalized to some synovial lining cells and endothelial cells of blood vessels in RA samples (Fig. 5c), but vWF and CD31 were almost exclusively localized to endothelial cells (Fig. 5d,e).

When homogenates of RA synovial tissues were subjected to immunoblotting analysis, eight major immunoreactive bands of 100, 76, 66, 47, 41, 38, 34 and 29 kDa were observed (Fig. 6, lane 2). Because the molecular weight of the 100-kDa band is similar to that of the precursor form of ADAM15 1225, this band appears to correspond to proADAM15. On the other hand, at least some of the other bands are considered to be active ADAM15 forms containing the metalloproteinase domain because of their positive immunoreactivity to the antibody specific to the cysteine-rich domain and their molecular weights. An immunoreactive band of 58 kDa was a non-specific reaction, because it was also detected with non-immune mouse IgG (Fig. 6, lane 1).

No definite correlation between relative mRNA expression levels of ADAM15 and the separate or total histological scores of RA and OA synovial tissues was observed (data not shown). Thus, we further evaluated vascular density in the RA and OA synovial tissues, by counting CD31-positive vessels in synovial tissue sections, and compared it with the mRNA expression levels of ADAM15. A linear correlation was found between expression levels and vascular density in RA and OA synovial tissues (r = 0.907, p < 0.001; n = 20; Fig. 7).

When the expression of ADAM15 in RA SFs was examined by RT-PCR, it was constitutively expressed regardless of the passage numbers (5 to 9) of the cells (Fig. 8a). Expression was decreased to low levels in a time-dependent manner, however, after starvation with serum-free medium for up to 72 h (Fig. 8b). To test the effect of cytokines and growth factors on ADAM15 expression, RA SFs were stimulated with TNF-α, IL-1α, TGF-β or VEGF₁₆₅; however, no changes in mRNA expression were found with these factors (Fig. 8c,d). In contrast, VEGF₁₆₅ appeared to selectively enhance ADAM15 expression in HUVECs (Fig. 8d), whereas TNF-α, IL-1α or TGF-β did not alter the expression (data not shown). Using real-time quantitative PCR, we found that the relative expression levels of ADAM15 mRNA (ratio of ADAM15 to ribosomal 18S) in HUVECs are significantly 2.2-fold higher after treatment with VEGF₁₆₅ (p < 0.05).

As previously reported 1529, HUVECs expressed the three major VEGF receptors (VEGFR-1, VEGFR-2 and neuropilin-1), but RA SFs expressed only neuropilin-1 under unstimulated conditions (Fig. 9a). Because the data that VEGF₁₆₅ stimulated ADAM15 expression only in HUVECs suggested that the effect is dependent on the expression of VEGF receptors, we tried to induce VEGF receptors by treating RA SFs with cytokines and growth factor and found that TNF-α, but not IL-1α or TGF-β, can induce VEGFR-2 expression without affecting the expression of VEGFR-1 or neuropilin-1 (Fig. 9b).

As TNF-α induced VEGFR-2 expression in RA SFs, we further examined whether VEGF₁₆₅ enhances ADAM15 expression in VEGFR-2-expressing RA SFs. After stimulation of RA SFs with TNF-α or VEGF₁₆₅ alone, ADAM15 mRNA expression, which was only weak after starvation of the cells, did not change (Fig. 10a). When the cells were sequentially treated with TNF-α and VEGF₁₆₅, however, the level of ADAM15 expression appeared to be increased (Fig. 10a). Real-time quantitative PCR analysis demonstrated that the expression levels are significantly 2.2-fold higher in RA SFs treated with TNF-α and VEGF₁₆₅ compared with the control without treatment (p < 0.05).

To examine the involvement of VEGFR-2 signaling in the stimulation of ADAM15 expression with TNF-α and VEGF₁₆₅, RA SFs were incubated with SU1498, a selective VEGFR-2 tyrosine kinase inhibitor, prior to the stimulation with VEGF₁₆₅ and ADAM15 mRNA expression was examined. ADAM15 expression decreased with 1 μM SU1498 and was completely suppressed with 10 μM SU1498, while VEGFR-2 expression was not affected by the treatment with such concentrations of the inhibitor (Fig. 10b). The enhanced expression of ADAM15 in HUVECs was also inhibited by the treatment with SU1498 (data not shown). PlGF, which selectively binds to VEGFR-1, did not affect the mRNA expression of ADAM15 in either RA SFs or HUVECs (data not shown).

Protein expression of ADAM15 and endothelial cell markers in cultured RA SFs was examined by immunohistochemistry. ADAM15 was immunolocalized to RA SFs and HUVECs (Fig. 11a,e), whereas no staining was observed with non-immune IgG (Fig. 11d,h). On the other hand, although VEGFR-2, vWF and CD31 were all immunostained in HUVECs (VEGFR-2 and vWF, Fig. 11f,g; CD31, data not shown), RA SFs were negative for these endothelial cell markers (vWF, Fig. 11c; VEGFR-2 and CD31, data not shown). When RA SFs were treated with TNF-α for 24 h, however, they were positively immunostained with anti-VEGFR-2 antibody (Fig. 11b). In accordance with the immunohistochemical data, the mRNA expression of CD31 and vWF in untreated RA SFs was not detected by RT-PCR (data not shown).