Exploratory proteomics of human synovial fluid in knee osteoarthritis: towards biomarker identification

Abstract

Purpose: There is a lack of valid and robust biomarkers in the field of OA diagnosis, prognosis, and treatment evaluation. Thus, our aim was to perform mass spectrometry (MS) of human synovial fluid using a global discovery approach, to identify biomarker candidates associated with meniscus degradation and/or knee OA. Methods: Synovial fluid was sampled from 3 different subject groups: i) end-stage medial compartment knee OA patients undergoing arthroplasty (n=11, 3 men, 8 women; age range 55-80 years), ii) knee arthroscopy patients who had knee symptoms but no or only minor knee trauma, typically having a degenerative meniscal tear (n=7, 3 men, 5 women; age range 50-64 years), and iii) deceased human donors (within 48 hours post mortem) without known chronic knee disease (n=13, 8 men, 5 women; age range 19-79 years). These 3 groups were roughly assumed to represent i) end-stage knee OA, ii) early knee OA, and iii) healthy controls, respectively. For all 3 groups, we required no visual signs of blood contamination to be present in the synovial fluid. For the cadaver knee tissues (controls), we further required the meniscus and the weight-bearing femoral surface of cartilage/bone to be macroscopically intact. All synovial fluids were centrifuged and freshly frozen and stored at -80°C. For the analysis, 50 μL of synovial fluid from each sample was mixed with MS-safe proteinase inhibitor cocktail and hyaluronidase, and further depleted of the 7 most abundant proteins with the multiple affinity removal system (MARS Hu7 spin cartridge). After reduction and alkylation, the samples were precipitated and the pellets were further digested with sequencing grade trypsin (Promega. Following digestion, samples were cleaned with a 30kDa filter and desalted. The samples were further spiked with iRT peptides before analysis by MS. The samples were analyzed with an EASY-nLC 1000 coupled to an Orbitrap Fusion mass spectrometer. The samples were analyzed with data-independent acquisition settings. The raw MS data were further analyzed with Spectronaut™ software for protein identification and quant data extraction. The quant data extracted using Spectronaut™ were further imported into the R environment (http://www.R-project.org/) to facilitate data analysis and visualization. Differences in protein levels between the 3 different groups were analyzed using the limma package and multiple logistic regression analysis. Differentially expressed proteins between the groups were clustered by hierarchical clustering using Pearson correlation coefficient as the distance metric. For each cluster, the enriched Gene Ontology terms and Reactome pathway names were identified and the most representative ones were annotated. Results: In total, 529 proteins were identified in the 31 different synovial fluid samples. Principal component analysis suggested a profound difference between the protein profiles of synovial fluid from controls vs arthroplasty patients, while the arthroscopy group had a protein profile that was in-between controls and the arthroplasty group (Figure 1). Statistical differential analysis yielded significant differences in the levels of 45 proteins comparing controls vs arthroplasty patients, 43 proteins for controls vs arthroscopy patients, and 58 proteins for arthroplasty vs arthroscopy patients. Extracellular matrix proteins like collagens and aggrecan showed higher expression in controls than in arthroplasty patients. Collagen on the other hand was higher in arthroscopy patients compared to controls. Decorin, a previously suggested biomarker candidate for OA, was significantly higher in controls compared to both arthroscopy and arthroplasty patients. The differentially expressed proteins associated with the acute inflammatory response (Table 1) were higher in the arthroscopy group compared to both controls and arthroplasty patients. Generally, extracellular matrix proteins were negatively correlated with immune system proteins and positively correlated with proteins involved in wound healing (Figure 2). Conclusions: There is a profound difference in the protein profile of human synovial fluid in knee-healthy controls vs knee OA patients. The inflammatory response seems to be higher in the early stages of OA than in later stages of the disease. Our findings emphasize the importance of OA staging in the development and use of biomarkers.View Large Image Figure ViewerDownload Hi-res image Download (PPT)View Large Image Figure ViewerDownload Hi-res image Download (PPT)