3.1 ScRNA-seq analysis of synoviocytes samples from patients with osteoarthritis and rheumatoid arthritis identified different cell types.
In this study, we obtained synovial tissue single-cell transcriptome data from the published dataset GSE200815 for RA patients (4 cases) and selected synovial tissue single-cell transcriptome data from GSE152805 for OA patients (3 cases) as a control for integration analysis. We then utilized RBP for single-cell subclustering to analyze the mechanisms of RBP regulation at the single-cell level that differed between rheumatoid arthritis and OA. Moreover, we collected bulk RNA-seq data from the GSE89408 dataset, and analyzed variable splicing events that were significantly regulated in RA compared to OA controls using the newly published SUVA software. We constructed a co-variant regulatory network of RBP and variable splicing events, as illustrated in Fig. 1A. The analyses offered significant insights into the unique mechanisms of RBP regulation and variable splicing that differentiate RA and OA. These insights contribute to a deeper understanding of the underlying factors of RA.
Following rigorous data quality control, a total of 33,226 single cells were included in the transcriptome profile analysis. Following normalization of the transcriptomic expression of these cells, principal component downsizing analysis was applied, and the top 50 principal components were selected for UMAP (Uniform Manifold Approximation and Projection) downsizing and visualization. Through unbiased cluster analysis, we identified 17 cell subgroups (Fig. 1B, Figure S1A-1B). Based on the characteristic expressed genes of the clusters, combined with previously reported synovial cell marker genes, we identified 9 different cell types (Fig. 1C-D).
We proceeded to compare the relative abundance of cell subtypes across all sample groups, and found that the RA and OA groups exhibited notably more pronounced alterations. In the RA group, we observed marked increases in proportions of cells including C2, C12, C7: Endothelial, C6: T, and C14: B, comparing to the OA group. Conversely, proportions of cells such as C15: Mastcells, C8, C4, and C0: Fibroblasts were found to decrease (Fig. 1E-F). Gene ontology enrichment analysis showed that up-regulated and down-regulated genes of each cell type in two comparison groups were enrich in several biological processes (Figure S1C-1D). We conducted a differential expression analysis between RA and OA groups for each cell type separately, and Fig. 1G shows the number of genes and differentially expressed RBPs in each cell type. Our analysis revealed that in each cell type, the count of down-regulated RBPs exceeded that of up-regulated ones. Nonetheless, when it came to fibroblasts, endothelial cells, and macrophages, the overall number of up-regulated genes vastly outnumbered that of down-regulated genes. Furthermore, In Fig. 1H, it can be observed that the overall AUC levels of RBPs were comparatively lower in the RA group as opposed to the OA group, suggesting reduced RBP activity in the RA group, which may affect transcriptional and post-transcriptional gene regulation. Collectively, these results provide a comprehensive understanding of the changes in cell composition and changes in rheumatoid arthritis and OA control synovial cells.
3.2 ScRNA-seq analysis identified heterogeneity and regulatory module of cellular specific RBP expression module.
Our aim was to investigate changes in RBP expression patterns in both the OA control group and the RA group. We conducted unsupervised clustering using Seurat based on the expression of 2,141 reported RBP genes. We observed that the resulting clusters of RBP expression (referred to as RBP-expressing cell clusters) were highly cell-type specific and correlated with disease status (Fig. 2A, Figure S2A-B). By analyzing the composition of RBP-expressing cellular taxa in different cell types, we found that for most cell types, the pattern of RBP expression was highly specific to that particular cell type, with an overwhelmingly dominant RBP-expressing cellular taxon (Fig. 2B). These findings provide evidence for cell type-specific RBP expression patterns in both disease states and control groups. Gene ontology enrichment analysis showed that these RBPs were enrich in several biological processes (Figure S2C). By shedding light on the function of RBPs, our findings help advance the comprehension of their role in synovial tissue during rheumatoid arthritis and osteoarthritis.
In addition, we found that the composition of RBP expression taxa changed significantly between the RA and OA groups for almost all cell types. For example, the main RBP expression taxon in the RA group was R1 for endothelial cells, whereas it was R13 for the OA group (Fig. 2B). This result suggests that different cell types have specific RBP expression patterns that are highly correlated with disease status. Specifically, different RBP expression taxa have specific marker RBP genes, and we demonstrated the expression of these RBPs in different sample subgroups and cell types (Fig. 2C). For instance, osteoglycine (OGN) and Filamin B (FLNB) is one of the markers of RBPs of R2, and we observed that these genes are mainly highly expressed in fibroblasts (Fig. 2D, Figure S2D). On the other hand, Muscleblind-like splicing regulator 2 (MBNL2) is one of the markers of RBPs of R1 and is predominantly expressed in endothelial cells of the RA group (Fig. 2E).
3.3 Functional RBPs are largely regulated in Fibroblasts cells between RA and OA samples
In particular, fibroblast-like synoviocyte (FLS) in RA have been identified as key actors in both the activation and maintenance of the RA-induced NF-κB pathway, which in turn promotes local proliferation, production of pro-inflammatory cytokines, and cartilage invasion. To better understand the potential functions of RBPs, which are key regulators of transcriptional and post-transcriptional processes, we isolated fibroblast cell classes and conducted secondary clustering and analysis using RBP-Genes. The resulting RBP-expressing cell classes differed significantly between the OA and RA groups. Specifically, the RA group primarily expressed FR1, FR3, FR4, FR5, FR5, FR6, and FR7, while the OA group mainly expressed FR0, FR2, FR8, and FR9. These observations indicate that the expression pattern of RBPs differs considerably between RA and OA and may have significant implications for the transcriptional and post-transcriptional regulation of fibroblasts (Fig. 3A-B, Figure S3A-B).
In our analysis comparing fibroblasts from RA and OA patients, we identified 105 upregulated RBPs and 133 downregulated RBPs. Figure 3C displays the top 20 upregulated and downregulated RBPs. These differentially expressed RBPs were co-expressed with genes enriched in various functional pathways, such as extracellular matrix organization, cell adhesion, collagen fibril organization, and cytokine signaling. These pathways suggest potential roles for RBPs in regulating fibroblast proliferation, antigen presentation, and pro-inflammatory responses (Fig. 3D-E, Figure S3C). Our study also revealed specific RBPs associated with important cellular pathways in RA, such as YBX3 and EIF4A1 (Fig. 3F-G). Additionally, we identified significant differences in the expression of important splicing factors, including U2AF1, SF3B6, and SF3B14, between RA and OA groups (Figure S3D, Figure S2E). These observations suggest that variations in splicing regulation may contribute to the disease pathology of RA.
3.4 RBP‑mediated fibroblasts subpopulations contributed to the aberrant activation of signaling pathways with immune cells during rheumatoid arthritis.
scRNA-seq has been applied successfully to predict potential LR interactions, revealing the importance of crosstalk between different cell types in various disease mechanisms. To investigate further the potential role of RBPs in the interactions between fibroblasts and immune cells, we performed cellular communication analysis of different RBP-expressing cell populations of fibroblasts and immune cells. Based on our research, we discovered that the level of interconnection and strength between fibroblasts and immune cells was noticeably higher in the RA group when compared to the OA group (Fig. 4A). In addition, fibroblasts and macrophages displayed the highest communication strength among the different immune cells (Fig. 4B). We then identified several signaling pathways, such as CXCL12-CXCR4, that were enhanced in RA and known to be associated with the disease (Fig. 4C-D). Lastly, changes in ligand expression levels of these key signaling pathways could be regulated by RBPs (Fig. 4E).
3.5 Identification of highly conserved RA-associated AS events co-disturbed with differentially expressed RBPs in RA and OA patients.
Disruption of normal RBP function can lead to cellular dysfunction by affecting post-transcriptional processes, such as variable splicing of RNA. To explore further the role of RBPs associated with RA in variable splicing regulation, we downloaded bulk RNA-seq data from the GSE89408 dataset, which included five RA and five OA samples as controls. Given the complex nature of human variable splicing, we utilized the SUVA software tool, which is a recently developed tool, to identify significant differences in variable splicing events between OA controls and RA. Our analysis using SUVA identified 715 differential variable splicing events, mainly alternative 5' and 3' splicing (Fig. 5A). Among the SUVA-identified splice events, alternative 5' splice site, cassette exon, exon skipping, and alternative 3' splice site were the most frequently detected differentially variable splicing events (Fig. 5B).
A single splicing event usually encompasses two transcripts, which make up only a small portion of the overall gene expression. To identify the leading transcripts where splicing occurs, we screened out 556 splicing events (pSAR > = 50%) that accounted for the dominant transcripts for further analysis (Fig. 5D). PCA analysis based on the splicing ratio of these dominant transcripts clearly separated the two sample sets, suggesting that the RNA splicing landscape is closely associated with the development of rheumatoid arthritis (Fig. 5D). These 556 transcripts hosting the identified splicing events are enriched in multiple functional pathways, including cytoskeleton organization, GTPase activation (which is associated with cellular processes such as programmed cell death regulation), and damage repair (Fig. 5E).
RBPs have been identified as crucial regulators of variable splicing. To predict the potential regulatory relationship between RBPs and variable splicing during rheumatoid arthritis development, Initially, we identified genes that exhibited a significant difference in expression levels between RA and OA controls. Our findings indicated that a large number of genes were activated during disease progression, with 2183 genes showing up-regulation and 823 genes showing down-regulation, as illustrated in Fig. 5F. Out of these genes, 209 RBPs showed up-regulation, while 232 were down-regulated, as shown in Fig. 5G-H. We then overlapped the differential RBPs obtained from scRNA-seq and bulk RNA-seq to identify conserved regulated RBP genes, finding three co-up-regulated (SYNE2, S100A9, IFIT3) and four co-down-regulated genes (RNASE1, GRN, FN1, SORBS2) (Fig. 5G, H, I). To predict the potential regulatory role of RBPs on variable splicing, we performed a co-variation analysis using these co-regulated RBPs and RAS. After screening (|correlation| >= 0.6, p-value < = 0.01), we identified RASs that were significantly associated with these RBPs (Fig. 5J).