Background/Purpose Osteoarthritis (OA) is the leading cause of chronic disability affecting 40% of individuals over the age of 70 and costing $128 billion annually in the US alone. Little is known regarding changes in gene expression that occur regionally within these affected joints. Herein, we perform RNA-seq analysis of eroded and intact cartilage from human OA, and correlate transcript levels with histopathologic disease severity.
Methods Six femoral heads were obtained at the time of hip arthroplasty for primary OA. Articular cartilage tissue was dissected from grossly affected and grossly normal areas from the same joints, flash frozen in liquid nitrogen, and RNA was extracted. Tissue samples from grossly affected and normal joint regions were histologically examined for OA severity using Mankin scoring. Following confirmation of RNA quality (RIN value ≥6), samples were sequenced with the Illumina TruSeq system on a MiSeq sequencer. Raw data were analysed and mapped using the GeneSifter software package. Genes with GeneSifter quality score <1.0 were excluded. For categorical analysis, EdgeR p-values <0.05 with Benjamini-Hochberg q ≤ 0.1 and expression ratios ≤0.83 or ≥ 1.2 between affected and normal tissues were considered significant. For correlations with histologic score, Pearson’s r > 0.75 or <-0.75 with p ≤ 0.05 were considered significant. Gene ontology and pathway analysis was performed using Ingenuity IPA and DAVID.
Results Categorical analysis identified 43 overexpressed and 313 underexpressed genes in eroded compared to intact cartilage. Both over- and underexpressed genes were overrepresented in the fibroblastic growth factor (FGF) signalling pathway (p = 0.004). FGFR2 demonstrated aneroded to intact cartilage expression ratio = 0.46, was highly inversely correlated with OA histologic score severity (r = 0.92), and is hypermethylated in eroded OA cartilage. The WNT pathway genes, WNT11 (ratio 0.27) and WNT9A (ratio 0.45), and STAT3 pathway was also overrepresented, including both over- and underexpressed genes (p = 0.001). A top predicted upstream regulator in differentially expressed genes was mir-9 (p = 0.005), known to be associated with metalloproteinase production. Further, we identified 1576 genes positively correlated with histopathologic score. Among these, the NFAT pathway was highly overrepresented including both positively and inversely correlated genes (24, p = 0.002). The RIG-I innate immunity pathway was also overrepresented among inversely correlated genes (p = 0.008), as were several genes related to chromatin remodelling (overall p = 0.009: HDAC1, r = -0.89, MECP2 r = -0.78, RBBP4 r = -0.82, SAP130 r = -0.81, SIN3A r = -0.80).
Conclusions Using RNA-seq we detected significant changes in gene expression in eroded compared to intact OA cartilage, as well as expression changes correlated with histologic disease progression in OA. Our data strongly suggest involvement of several signalling pathways, many of which are potential therapeutic targets for OA. This work reinforces the heterogeneity of the disease process and provides novel insights into OA pathogenesis.