Plasma proteins present in osteoarthritic synovial fluid can stimulate cytokine production via Toll-like receptor 4

Serum and synovial fluid samples were obtained from patients with knee OA, patients with rheumatoid arthritis (RA), or healthy individuals under protocols approved by the Stanford University Institutional Review Board and with the patients' informed consent. Synovial fluid aspiration was performed by a board-certified rheumatologist by fine-needle arthrotomy, and the synovial fluid samples obtained were free from obvious contamination with blood or debris. OA serum and synovial fluid samples were obtained from patients diagnosed with knee OA (of Kellgren-Lawrence score 2 to 4 [17]) according to the 1985 criteria of the American Rheumatism Association [18]. For mass spectrometric analysis, OA synovial fluid samples were from five Caucasian men aged 50 to 75 years who met the 1985 OA criteria [18]; exclusion criteria included radiographic evidence of chondrocalcinosis or evidence of crystals under polarizing microscopy. Demographics and clinical characteristics of these five individuals are shown in Table 1. Synovial fluids from the other OA patients and from the RA patients were provided as de-identified remnant clinical samples, and patient demographics were therefore unavailable for these samples. All RA patients met the 1987 American Rheumatism Association criteria for RA [19] and had RA of less than 6 months' duration; exclusion criteria included concurrent infectious or crystal arthritis. Samples of "normal" serum were obtained from healthy individuals who had no joint pain and no radiographic evidence of knee arthritis [20]. OA and normal sera were matched by age, sex, and BMI. Serum and synovial fluid samples were not matched but were derived from patients with the characteristics described earlier. All samples were aliquoted and stored at -80°C.

Synovial fluid proteins were separated by 1D or 2D polyacrylamide gel electrophoresis (PAGE), trypsinized, and identified by liquid chromatography tandem mass spectrometry (LCMS), as follows. Fifty microliters of frozen synovial fluid was diluted to a final volume of 1 ml in phosphate buffered saline (PBS) containing Halt protease and phosphatase inhibitor (Thermo Fisher Scientific), and then depleted of the highly abundant proteins albumin and immunoglobulin G (IgG) by using the ProteoPrep Immunoaffinity Albumin & IgG Depletion Kit (Sigma-Aldrich) according to the manufacturer's instructions. In brief, synovial fluids were twice passed over spin columns prepacked with a mixture of two beaded mediums containing recombinantly expressed, small, single-chain antibody ligands. The flow-through fractions containing synovial fluid depleted of albumin and IgG were diluted 1:1 with Laemmli Sample Buffer (BioRad) and then subjected to 1D-PAGE or 2D-PAGE analysis. Because a small number of proteins other than albumin and IgG may bind to the medium in the spin columns, the bound proteins were eluted with Laemmli sample buffer and also subjected to PAGE analysis. For 1D-PAGE analysis, proteins were boiled for 10 minutes and separated on Precast Criterion XCT gels (4% to 12% linear gradient, BioRad). After electrophoresis, the gels were stained for 1 hour with Gelcode blue (Pierce) and destained overnight. For 2D-PAGE analysis, methods were as previously described [21]. In brief, 100 μg of synovial fluid proteins was dissolved in 150 μl of isoelectric focusing (IEF) buffer (ReadyPrep Sequential Extraction Kit Reagent 3, BioRad). For the first-dimension electrophoresis, 150 μl (at a 1 μg/μl concentration) of sample solution was applied to an 11-cm Ready-Strip Immobilized pH Gradient (IPG) strip, pH 3 to 10 (BioRad). The IPG strips were soaked in the sample solution for 1 hour, to allow uptake of the proteins, and then actively rehydrated in the Protean IEF cell (BioRad) for 12 hours at 50 V. IEF was performed for 1 hour at each of 100, 200, 500, and 1,000 V, and then for 10 hours at 8,000 V. For second-dimension electrophoresis, IPG strips were equilibrated for 20 minutes in 50 mM Tris-HCl, pH 8.8, containing 6 M urea, 1%SDS, 30% glycerol, and 65 mM dithiothreitol (DTT), and then re-equilibrated for 20 minutes in the same buffer containing 260 mM iodacetamide in place of DTT. Precast Criterion XCT gels (4% to 12% linear gradient, BioRad) were used for the second-dimension electrophoresis, as was done for the 1D-PAGE. After electrophoresis, the gels were stained for 1 hour with Gelcode blue (Pierce) and destained overnight.

The stained protein bands and spots (from the 1D-PAGE and 2D-PAGE, respectively) were cut out of the gels, immersed in 10 mM ammonium bicarbonate containing 10 mM DTT and 100 mM iodoacetamide, treated with 100% acetonitrile, and then digested overnight at 37°C with 0.1 mg trypsin (Sigma-Aldrich) in 10 mM ammonium acetate containing 10% acetonitrile. The trypsinized proteins were identified with LCMS by using the Agilent 1100 LC system and the Agilent XCT Ultra Ion Trap (Agilent Technologies, Santa Clara, CA) as previously described [22]. We scanned the LCMS data against the SwissProt database by using the SpectrumMill software (Agilent). We required the detection of at least two peptides for identification of a protein, and a significance level of P ≤ 0.05 for identification of each peptide. The significance level of peptide identification takes into account the number of ionization forms of the fragmented peptide that match with a particular protein in the SwissProt database (with penalties for ionization forms not identified), as well as the total intensity of each ionization form [23].

Multiplex analysis of cytokines and chemokines in human serum and synovial fluid samples was performed by using both the 27-plex and the 21-plex Bio-Plex Pro Human Cytokine Assay (BioRad) run on the Luminex 200 platform, as recommended by the manufacturers. Performing the Bio-Plex assay with the kit reagents, we found that several commercial reagents designed to block the confounding effect of heterophilic antibodies, including ones we used previously with other cytokine assay kits [24], did not significantly affect the readout of the Bio-Plex assay; we therefore did not use such blocking reagents with the Bio-Plex assay. Data processing was performed by using Bio-Plex Manager 5.0, and analyte concentrations (in picograms per milliliter) were interpolated from standard curves. Statistical differences in cytokine levels were calculated with significance analysis of microarrays (SAM [25]), and the SAM-generated results with a false discovery rate (FDR) of less than 10% were selected. To identify relations and to display our results most effectively, we normalized the analyte concentrations as follows: all values less than 1 were designated as 1, and the mean concentration of each analyte in the "normal serum" samples was calculated; the analyte value in the sample was then divided by the mean analyte value in normal serum, and finally, a log-base-2 transformation was applied. Results were subjected to unsupervised hierarchic clustering by using Cluster 3.0, which arranges the SAM-generated results according to similarities in cytokine levels, and the clustering results were displayed by using Java Treeview (Version 1.1.3).

To generate mouse macrophages, we differentiated bone-marrow cells isolated from wild-type C57BL/6 mice and from B6.B10ScN-Tlr4 ^lps-del mice (Jackson Laboratory) according to standard procedures [26]. In brief, the femur and tibia were flushed with α-minimal essential medium (MEM; Invitrogen) by using a 1-ml syringe and a 25-gauge needle. The resulting cell suspension was lysed with ACK Lysing Buffer (Invitrogen) for removal of erythrocytes. Cell clumps were removed by filtering through a 70-μm cell strainer (BD). The remaining cells in the suspension were cultured on 100-mm culture dishes in α-MEM supplemented with 10% fetal bovine serum (FBS), 100 units/ml of penicillin, 100 μg/ml of streptomycin, and 2 mM glutamine (Invitrogen) for 16 to 24 hours in 5% CO₂ at 37°C. Nonadherent cells were collected, plated on 100-mm dishes, and differentiated into bone-marrow-derived macrophages (BMMs) for 6 days in the presence of 30 ng/ml of macrophage colony-stimulating factor (PeproTech). To generate human monocyte-derived macrophages (MDMs), we collected peripheral blood mononuclear cells (PBMCs) by performing density-gradient centrifugation of LRS chamber content (Stanford Blood Center) over Ficoll (Invitrogen), purified human monocytes by negative selection by using a monocyte isolation kit (Miltenyi Biotec), and differentiated the monocytes into macrophages by culturing them for 7 days in RPMI containing 10% FBS and 30 ng/ml of human M-CSF.

For stimulation assays, mouse BMMs were plated in 96-well plates at 1 × 10⁵ cells/well, and human macrophages at 7 × 10⁴ cells/well. Cells were incubated for 24 hours with lipopolysaccharide (LPS; Sigma-Aldrich), peptidoglycan (InvivoGen), α₁-microglobulin (Cell Sciences), α2-macroglobulin (EMD Chemicals), α1-acid glycoprotein (EMD Chemicals), Gc-globulin (also known as vitamin D-binding protein; Abcam), haptoglobin (Sigma-Aldrich), or human serum albumin (Sigma-Aldrich). We measured levels of interleukin-1β (IL-1β), interleukin-6 (IL-6), and vascular endothelial growth factor (VEGF) in the culture supernatants with Luminex analysis, by using a 27-plex Bio-Plex Pro Human Cytokine Assay kit (BioRad) according to the manufacturer's instructions. We measured TNF levels with enzyme-linked immunosorbent assay (ELISA; PeproTech). For the TNF ELISA, the limits of detection were 16 to 2,000 pg/ml for mouse TNF, and 23 to 1,500 pg/ml for human TNF. For the Luminex assay, the limits of detection were 3.2 to 3,261 pg/ml for IL-1β, 2.3 to 18,880 pg/ml for IL-6, and 5.5 to 56,237 pg/ml for VEGF. To exclude a contribution of endotoxin contamination, we included 10 μg/ml of polymyxin B (Sigma-Aldrich) in some of the stimulation assays. As an additional control for endotoxin contamination, we tested whether preincubating the plasma proteins with proteinase K and β-mercaptoethanol at 55°C for 4 hours (and then at 100°C for 10 minutes to inactivate the proteinase K) abrogated their ability to induce the production of cytokines (the plasma proteins, but not any contaminating endotoxin, would be denatured under these conditions).

One-way ANOVA and unpaired t test (Graph-Pad Software) were used to analyze differences in levels of cytokines. P values less than 0.05 were considered significant.