Transcriptional profiling of PVNS using cDNA microarray and RNA-seq
Transcriptional profiling of PVNS was analyzed using RNA microarray and RNA-seq, each analysis included three PVNS and three OA patients. In RNA-seq analysis, 28,753 human cDNA clones were detected (Fig. 1A-D), whereas in RNA microarray analysis, sample RNA and a common reference RNA were labeled with fluorescent dye and hybridized on a microarray containing 12,762 human cDNA clones (Fig. 1E-H)
Principle component analysis (PCA) was conducted to visualize the sample relationships, which revealed that the samples of the PVNS or OA group can be separated into two clusters according to their gene expression profile, no matter in microarray or RNA-seq analysis. To further obtain an overview of the similarity in all samples, hierarchical cluster analysis was performed by determining the sample-to-sample distances. The samples of PVNS and OA patients were built one cluster at a time in the two analysis. Based on these results, the differentially expressed genes (DEG) in PVNS and OA groups were identified (Fig. 1A, B, E, F).
After the weakly expressed genes were filtered out and correction for multiple testing was performed, genes with q-value < 0.01, fold change (FC) > 2, and fragments per kilobase of transcript per million mapped reads (FPKM) > 1 were identified as DEGs. RNA-seq analysis identified 1,220 DEGs in the PVNS group compared to the OA group, whereas RNA microarray analysis reported 1,282 DEGs (Fig. 1C, D, G, H).
Venn diagram presentation of the DEGs in RNA-microarray and RNA-seq analyses revealed 195 common DEGs (Fig. 1I) with consistent trend in expression changes (Fig. 1J, K), confirming that these are important DEGs. Of the 195 common DEGs, 103 were upregulated and 92 were downregulated in PVNS patients compared to OA patients. The following analyses were based on these common DEGs.
Differential expression and pathway enrichment analyses
The DEGs in PVNS were grouped in Gene Ontology (GO) categories using DAVID software. Functional enrichment analysis included immune response, cytokine production, osteoclast development, and cell migration (Fig. 2A). Analysis using cytoscape reflected the relationship between the GO terms and summarized the main biological processes in which the DEGs expressed in PVNS were involved. As shown in Fig. 2C, the primary GO term focused on “immune effector process”, “cytokine production”, “lipopolysaccharide response”, “leukocyte migration”, “osteoclastogenesis”, and “nervous system development” (Fig. 2C).
Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis showed that the DEGs were mainly enriched in the immune cell interactions, hematopoietic cell lineage, peroxisome proliferator-activated receptor signaling pathway, neutrophil degranulation and several others (Fig. 2B).
Results of the global transcriptional profile analysis of the DEGs in PVNS compared to OA observations were consistent with previous reports that revealed PVNS as an invasive arthritis characterized by both inflammation and tumor phenotype (3). Although inflammation was associated with high immune response and cytokine production, cell proliferation and migration resulted in the manifestation of tumor characteristics. Furthermore, activated osteoclast and macrophage were observed in the joint. These three main aspects of the transcriptional profile in PVNS are presented and demonstrated below.
High immune cell infiltration and increased cytokine secretion
Proliferative lesion and synovitis are considered as the critical pathogenic characteristics of PVNS. The protein-protein interaction of immune response-related genes were analyzed using string and shown in cytoscape (Fig. 3A). Heatmaps of these genes were also generated to compare the difference in the synovium of PVNS and OA patient samples (Fig. 3B). It is worthy of mentioning that majority of the upregulated membrane proteins were markers of immune cells, including CD3, CD6, colony-stimulating factor (CSF)2R, CSF3R, and receptor activator of nuclear factor kappa B (RANK), which is also known as TNFRSF11A, indicating the increasing immune cell infiltration in PVNS synovium.
Immunohistochemical experiments confirmed the presence of CD45+ inflammatory cells in the PVNS synovium, suggesting severity of the local inflammation (Fig. 3C). Analysis of the composition of local cells using flow cytometry revealed that in addition to the CD45− local fibroblasts in joints, there was a high proportion of CD45+ inflammatory cells, such as T cells (CD3+), natural killer (NK) cells (CD3−CD56+), and NKT cells (CD3+CD56+) (Fig. 3D). Checkpoint molecules, including programmed cell death protein 1(PD-1), T cell immunoglobulin and mucin domain 3 (TIM-3), lymphocyte-activation gene 3 (LAG-3), and cytotoxic T lymphocyte-associated protein 4 (CTLA-4), were detected on the surface of CD4+ and CD8+ T cells (Fig. 3E).
We analyzed the related pathways or cell types of highly expressed DEGs using gene set enrichment analysis (GSEA) and found that the ones in PVNS had a relation to activated neutrophil, B cells, T cells, and monocytes. This finding suggests that the inflammatory cells activated in the joints primarily comprises immune cells from myeloid and lymphoid lineages and that the local inflammatory responses include both innate and adaptive immunity (Fig. 3F).
Increased cell proliferation and cell migration in PVNS
H&E staining of PVNS synovium showed obvious synovial hyperplasia and local hemosiderin pigmentation in tissues. Higher expression of Ki-67 confirmed cell proliferation in local joint (Fig. 4A). GSEA revealed that the DEGs in PVNS were mostly associated with the cell cycle-related pathways, including G2M checkpoint and transcript factor E2F (Fig. 4B).
Protein-protein interaction analyzed using cytoscape and heatmap of gene expression revealed that the large number of DEGs were related to cell migration, including cell adhesion, extracellular matrix, actin assembly, and cell chemotaxis (Fig. 4C, D). RT-PCR validated and verified the relative expression levels of the fifteen cell migration-related genes that were identified as DEGs (Fig. 4E). Corresponding GSEA pathway analysis indicated that the epithelial-mesenchymal transition and integrin beta-2 were correlated with PVNS characteristics (Fig. 4F).
Increasing osteoclastogenesis-related genes in PVNS
Among the DEGs, highly upregulated genes in PVNS, such as matrix metallopeptidase 9/11 (MMP9/11), TNFRSF11A, and osteoclast stimulatory transmembrane protein (OCSTAMP), were closely associated with osteoclastogenesis and bone resorption (Fig. 5A). The high expression of bone resorption-related genes was consistent with the high frequency of bone erosion in PVNS. RT-PCR demonstrated the increased expression of OCSTAMP, sialic-acid-binding immunoglobulin-like lectin 15 (SIGLEC15), and TNFRSF11A. in PVNS compared to OA samples (Fig. 5B).
The correlation between the highly expressed DEGs in PVNS and osteoclast were confirmed using GO term functional enrichment (Fig. 2A) and GSEA (Fig. 5C). H&E staining of the PVNS synovial tissues revealed the typical multinucleated giant cell morphology of the osteoclast (Fig. 5D).
CD14+ monocytes purified from synovial tissue of PVNS and OA samples were made to differentiate into osteoclasts with M-CSF and RANK ligand (RANKL) treatment, respectively, for 2 weeks. Tartrate-resistant acid phosphatase-positive multinucleated cells with > 3 nuclei were considered as osteoclasts. As shown in Fig. 5E, the number of mature osteoclasts that differentiated from PVNS were significantly higher than from OA synovial tissue.
Since macrophages are precursors of osteoclasts, the functions of macrophage were also analyzed. GSEA pathway analysis revealed activation of several macrophage functions in PVNS, such as interferon (IFN) response, Fc receptor (FcR)-mediated phagocytosis, antigen presentation, and toll-like receptor (TLR)4 signaling pathway (Fig. 5F).