Metabolic reprogramming and macrophage expansion define ACPA-negative rheumatoid arthritis: insights from single-cell RNA sequencing

Abstract

Background: Anti-citrullinated peptide antibodies (ACPA)-negative (ACPA-) rheumatoid arthritis (RA) presents significant diagnostic and therapeutic challenges due to the absence of specific biomarkers, underscoring the need to elucidate its distinctive cellular and metabolic profiles for more targeted interventions.

Methods: Single-cell RNA sequencing data from peripheral blood mononuclear cells (PBMCs) and synovial tissues of patients with ACPA- and ACPA+ RA, as well as healthy controls, were analyzed. Immune cell populations were classified based on clustering and marker gene expression, with pseudotime trajectory analysis, weighted gene co-expression network analysis (WGCNA), and transcription factor network inference providing further insights. Cell-cell communication was explored using CellChat and MEBOCOST, while scFEA enabled metabolic flux estimation. A neural network model incorporating key genes was constructed to differentiate patients with ACPA- RA from healthy controls.

Results: Patients with ACPA- RA demonstrated a pronounced increase in classical monocytes in PBMCs and C1QChigh macrophages (p < 0.001 and p < 0.05). Synovial macrophages exhibited increased heterogeneity and were enriched in distinct metabolic pathways, including complement cascades and glutathione metabolism. The neural network model achieved reliable differentiation between patients with ACPA- RA and healthy controls (AUC = 0.81). CellChat analysis identified CD45 and CCL5 as key pathways facilitating macrophage-monocyte interactions in ACPA- RA, prominently involving iron-mediated metabolite communication. Metabolic flux analysis indicated elevated beta-alanine and glutathione metabolism in ACPA- RA macrophages.

Conclusion: These findings underscore that ACPA-negative rheumatoid arthritis is marked by elevated classical monocytes in circulation and metabolic reprogramming of synovial macrophages, particularly in complement cascade and glutathione metabolism pathways. By integrating single-cell RNA sequencing with machine learning, this study established a neural network model that robustly differentiates patients with ACPA- RA from healthy controls, highlighting promising diagnostic biomarkers and therapeutic targets centered on immune cell metabolism.

Keywords: ACPA; beta-alanine and glutathione metabolism; rheumatoid arthritis; single-cell RNA sequencing; synovial macrophage.

Figure 1

(A) Sample size distribution of scRNA-seq data from HC, ACPA+ RA, and ACPA− RA individuals. (B) UMAP clustering of immune cells from PBMCs. (C) Wilcoxon test comparing immune cell proportions between ACPA+ and ACPA− groups. (D) UMAP clustering of monocyte subpopulations. (E) Wilcoxon test comparing monocyte subpopulations between ACPA+ and ACPA− groups. (P-values are expressed as follows: * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001.).

Figure 2

(A) UMAP plot of scRNA-seq data from synovial cells. (B) UMAP plot of macrophage subtypes from synovial cells. (C) Markers for hematogenous macrophages and tissue-resident macrophages. (D) AUCell activity scoring of the 11 macrophage subtypes in the KEGG Rheumatoid Arthritis pathway. (E) Wilcoxon test comparing synovial cell populations between ACPA+ and ACPA− groups. (F) Wilcoxon test comparing macrophage subpopulations from synovial cells between ACPA+ and ACPA− groups. (P-values are expressed as follows: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and NS indicates no significance.).

Figure 3

(A) Pseudotime trajectory analysis of macrophage subtypes in patients with ACPA− RA. (B) Pseudotime progression of macrophage subtypes in patients with ACPA− RA. (C) Pseudotime trajectory analysis of macrophage subtypes in patients with ACPA+ RA. (D) Pseudotime progression of macrophage subtypes in patients with ACPA+ RA. (E) Heatmap of transcription factor activity in ACPA− (left panel) and ACPA+ (right panel) RA macrophages. (F) GSVA analysis of macrophage subpopulations.

Figure 4

(A) Correlation heatmap of seven gene co-expression modules identified by WGCNA in macrophage subtypes. (B) Dot plot showing gene expression within the brown module across macrophage subtypes. (C) Enrichment analysis of genes in the brown module. (D) Venn diagram showing the overlap of 426 genes from the brown module with differentially expressed genes in PBMCs and macrophage subtypes. (E) Training loss curve for the deep neural network model distinguishing patients with ACPA− RA from healthy controls. (F) ROC curve displaying the neural network model’s performance on the test set.

Figure 5

(A) Bar chart depicting the number of inferred interactions and interaction strengths. (B) Relative information flow of key signaling pathways mediating macrophage-monocyte communication. (C) CD45 signaling pathway network for ACPA− (left) and ACPA+ (right) RA. (D) Box plot of PTPRC expression, a critical component of the CD45 signaling pathway. (E) Box plot of MRC1 expression, another essential component of the CD45 signaling pathway. (P-values are expressed as follows: * p ≤ 0.05, *** p ≤ 0.001, and NS indicates no significance.).

Figure 6

(A) Number of metabolite-sensor communication events in ACPA− RA. (B) Number of metabolite-sensor communication events in ACPA+ RA (C) Metabolite-mediated communication pathways in ACPA− RA. (D) Metabolite-mediated communication pathways in ACPA+ RA. (E) Violin plots showing the mean abundance of metabolites in ACPA− RA. (F) Violin plots showing the mean abundance of metabolites in ACPA+ RA.

Figure 7

(A) Convergence of the loss function during scFEA analysis. (B) T-values of metabolite modules across macrophage subtypes, visualized with color codes: red for t-values > 0 (indicating upregulation in the ACPA− group) and blue for t-values < 0 (indicating downregulation in the ACPA− group or upregulation in the ACPA+ group). (C) Summary table of the top in-and-out metabolites for significant metabolic modules in ACPA− RA. (D) KEGG enrichment analysis of output metabolites from macrophage subtypes. (E) KEGG enrichment analysis of input metabolites from macrophage subtypes.