Subjects
Ten patients with AS were recruited from an outpatient rheumatology clinic in Hanyang University Hospital for Rheumatic Diseases, Seoul, South Korea. An additional 30 patients treated at Keimyung University Dongsan Hospital, Daegu, South Korea were recruited as disease control groups (10 OA, 10 gout, and 10 RA). These patient SF samples (first cohort) were used for the LC-MS/MS experiments and the associated western blot verification. We further recruited five more patient samples per each patient group (second cohort) for the independent western blot verification experiments (5 AS, 5 OA, 5 gout, and 5 RA). The patients met the 1984 New York criteria for ankylosing spondylitis [17] and the American College of Rheumatology criteria for RA [16], knee OA [18], and acute gout [19]. We collected demographic and clinical data from the subjects, including age, gender, disease duration, blood chemistry, and concomitant treatment. We provide the summary statistics (Table 1) and the original data including the diagnosis, clinical data, and current treatments of each patient enrolled in this study (Supplementary Table S1). The samples were obtained after getting informed consent from the patients. The study was approved by the ethical committee of the Keimyung University Dongsan Hospital (IRB 2015-12-022).
Synovial fluid sample collection and processing
Synovial fluid was collected by arthrocentesis in patients with knee joint pain and swelling. Contaminated bloods during arthrocentesis and samples where synovial fluids were generated for reasons other than the respective diseases were excluded. After centrifugation at 500 x g for 10 min, five 1 mL vials of supernatant and 1 vial of sediment from each sample were stored at -80 °C through the Keimyung University Dongsan Hospital Human Resource Bank.
Immunodepletion of abundant proteins with MARS cartridge
We used a Multiple Affinity Removal System (MARS) Hu-14 spin cartridge (Agilent, 5188-6560), which contains a bulk of immobilized antibodies against the 14 most abundant proteins (albumin, IgG, antitrypsin, IgA, transferrin, haptoglobin, fibrinogen, alpha-2-macroglobulin, alpha-1-acid glycoprotein, IgM, apolipoprotein A1, apolipoprotein A2, complement C3 and transthyretin), which are known to occupy about 95% of plasma proteome [20], enabling immunodepletion of such abundant proteins in SF. The 0.22 μm membrane filter (Agilent) was used to remove particulates from the fluid samples by centrifugation at 100 x g for 1.5 min. The flow-through was mixed with Buffer A LOAD/WASH (Agilent) and depleted according to the manufacturer’s instructions. During the depletion, flow-through was collected, and protein concentration was determined using a BCA assay. The column was routinely regenerated by eluting bound high-abundance proteins with buffer B and neutralizing with buffer A before further use. The acquired proteins were directly digested for total proteomic analysis.
Peptide sample preparation
In-solution tryptic digestion and peptide cleanup were simultaneously performed in a 96-well plate for high reproducibility. Each depleted sample was supplemented with 8 M urea in 100 mM ammonium bicarbonate (ABC) (Sigma) and incubated at room temperature for 20 min. The samples were homogenized by vortexing and sonication twice. To each sample, dithiothreitol (Sigma) was added to be 10 mM for protein reduction at RT for 1 hr. Then, iodoacetamide (Sigma) was added to be 30 mM for the cysteine alkylation at RT for 30 min in the dark. Samples were then diluted with 100 mM ABC prior to the addition of trypsin (MS grade, Pierce) at 1:50 of trypsin:sample ratio (w/w), and incubated at 37℃ for overnight. The trypsin was inactivated by acidification with 0.4% trifluoroacetic acid (Sigma). The acidified digests were immediately processed using a Sep-Pak C18 96-well plate (100 mg C18 sorbent per well, Waters). The peptides were eluted with 80% acetonitrile and then dried in a vacuum centrifuge.
LC-MS/MS experiments
We performed a label-free quantitative proteomics using LC−MS/MS experiments in synovial fluid samples. Forty LC−MS/MS experiments were carried out on an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) coupled to a nanoACQUITY UPLC (Waters) with an in-house-packed trap (150 μm i.d. × 3 cm) and analytical column (75 μm i.d. × 100 cm) using 3 μm Jupiter C18 particles (Phenomenex). The LC gradient was as follows: from 5% to 40% solvent B (acetonitrile with 0.1% formic acid) for 130 min, then 40% to 80% solvent B for 5 min, holding at 80% solvent B for 10 min, and then equilibrating at 95% solvent A (water with 0.1% formic acid) for 30 min. Full MS data were acquired in a scan range of 375–1,575 Th at a resolution of 60,000 at m/z 200, with an automated gain control (AGC) target value of 4.0×105 and a maximum ion injection time of 50 ms. The maximal ion injection time for MS/MS was 50 ms at a resolution of 15,000 and an AGC target value of 5 x 104. The dynamic exclusion time was set to 30 s. The resulting forty MS raw files were analyzed using MaxLFQ in Maxquant software (Figure 1).
Data analysis
MaxQuant (v.1.5.1.2) [21] was used to compare the acquired spectra to the UniProt human database (obtained in June 2018). Carbamidomethylation of cysteine was selected as a fixed modification, while N-terminal acetylation and methionine oxidation were set as variable modifications. 1% of false discovery rate (FDR) cutoff was applied at the levels of peptide-spectrum match and protein. An initial precursor mass deviation of up to 4.5 ppm and a fragment mass deviation of up to 20 ppm were allowed. To maximize the protein identification in initial discovery stage, the criterion with minimum stringency, at least one unique peptide, was applied for the protein identification. Proteins were quantified using the label-free quantification (LFQ) algorithm in MaxQuant [22]. The option of ‘match between runs’ was used for nonlinear retention time alignment. The match time window was 0.7 min, and the alignment time window was 20 min. Further statistical and bioinformatics analysis was performed using Perseus software (v.1.5.3.2) [23]. Proteins hit to the decoy database and contaminants, or proteins only identified by the peptide with modification sites such as methionine oxidation were filtered out prior to further analysis. A minimum of three valid protein quantification values across each clinical group was required for protein quantification. The LFQ intensity of individual proteins in the AS group was compared with that of such proteins in the other clinical groups following log2 transformation, i.e., AS vs OA, AS vs gout, and AS vs RA. Statistical analysis of the log2-transformed data was performed using two samples t-test (p value < 0.05). Proteins with fold change ≥ 1.5 or ≤ 0.67 between AS and other patient groups were considered to be differentially expressed proteins (DEPs) for AS group. The resulting DEPs were submitted to gene ontology (GO) annotation enrichment analysis.
Enrichment and network analysis
To explore functional enrichment in the identified proteins in each of the clinical groups, GO analysis was performed using DAVID (version 6.8) [24]. We identified biological processes and KEGG pathways [25] that were enriched in our DEP list with p values of less than 0.05. To construct a network representing the enriched GO terms, we selected DEPs that are involved in the enriched cellular processes. We then built a protein network model using protein interaction information obtained from STRING version 11 database [26]. The interaction network models were visualized using Cytoscape [27].
Western blot verification
We carried out western blot based verification experiments for the candidates as AS-specific proteins from LC-MS/MS proteomic data. To reduce the viscosity of SF, samples were treated with hyaluronidase (Sigma, H3884) at room temperature for 5 or 10 min, and then diluted 1:10 in RIPA buffer (Thermo Fisher) containing both protease and phosphatase inhibitor (Roche Diagnostics GmbH). The diluted sample was incubated on ice for five min, then transferred into new tube and centrifuged at 10,000 x g at 4°C for 15 min. Equal amounts of protein (30 μg) in each sample were aliquoted following protein quantitation by BCA assay (Thermo Fisher). Samples were mixed with sodium dodecyl sulfate (SDS) loading buffer and separated using SDS polyacrylamide gel electrophoresis (Western Blotting Kit, Hoefer Inc.). Proteins were then transferred onto nitrocellulose membranes (Amersham), blocked with 5% non-fat dried milk in tris-buffered saline containing 0.05% Tween-20 and immunoblotted with the appropriate primary and secondary antibodies. The antibodies used were as follows: rabbit polyclonal antibodies for C9 (1:5000, PA5-29093, Thermo Fisher), CFHR5 (1:500, ab175254, Abcam), MMP3 (1:5000, ab52915, Abcam), mannose-binding protein C (MBL2) (1:5000, ab189856, Abcam), complement C4-A (C4A) (1:5000, ab66790, Abcam), serum amyloid P-component (APCS) (1:5000, ab45151, Abcam), matrix metalloproteinase-1 (MMP1) (1:3000, ab38929, Abcam), anti-transferrin (1:5000, ab109503, Abcam), and peroxidase-conjugated AffiniPure Donkey anti-rabbit IgG (H+L) (1:10,000, Jackson Immunoresearch). Anti-transferrin (77 kDa) was used as a loading control [28].