Background/Purpose: Pro-inflammatory fibroblasts have been independently implicated in the pathogenesis of rheumatoid arthritis (RA), inflammatory bowel disease (IBD), interstitial lung disease (ILD), and Sjogren’s syndrome (pSS). While these fibroblasts are likely affected by disease modifying antirheumatic drugs that target broad cytokine pathways such as TNF, IL6, and GM-CFS, recent work suggests the possibility of fibroblast-specific therapeutics, such as Notch3 inhibition to disrupt FAP⁺THY1⁺ inflammatory fibroblasts in arthritis. The presence of common fibroblast phenotypes expanded in multiple inflamed tissues would suggest that one fibroblast therapeutic could be effective for multiple indications. Clinically, these common states would provide a molecular basis for rheumatology “basket” trials to select patients based on common disease etiology across diverse diagnoses. We therefore constructed a cross-tissue atlas to (1) identify shared inflammatory fibroblast states and (2) map new tissue samples to define the composition of their fibroblast phenotypes.

Methods: Inflamed and non-inflamed tissue samples were collected from 71 donors in the (1) synovium: RA and osteoarthritis; (2) intestine: IBD (inflamed and adjacent non-inflamed) and healthy; (3) salivary gland: pSS and Sicca syndrome; (4) lung: ILD and non-chronic inflammatory conditions. We constructed and analyzed an integrated reference of 73,260 single cell RNAseq (10X Genomics) fibroblast profiles using novel methods based on our published Harmony and MASC algorithms. We mapped external scRNAseq data from mouse disease models (serum transfer arthritis, DSS-colitis, and bleomyocin lung fibrosis) onto our atlas using our novel Symphony algorithm and analyzed the distribution of inflammatory fibroblasts in each model.

Results:

1. Atlas of fibroblast phenotypes. We identified 9 conserved fibroblast states, including THY1⁺CD34⁺ interstitial, NOTCH3⁺ perivascular, and CCL19⁺ cytokine producing, and 3 tissue-specific states, including CD55⁺ synovial lining and MYH11⁺ myofibroblasts.
2. Phenotypes expanded in inflammation. Two fibroblast states were significantly expanded with inflammation in all four tissues. CCL19⁺ fibroblasts, enriched in interferon-gamma signaling, may be involved in T cell recruitment. NOTCH3⁺ fibroblasts, enriched in collagen production and proliferation, may orchestrate vascular remodeling.
3. Evaluation of mouse models. We mapped scRNAseq profiles from pre-clinical disease models to our fibroblast atlas. CCL19⁺ cells were expanded with inflammation in all tissues, while NOTCH3⁺ cells were expanded in lung and synovium but not in the intestine.

Conclusion: Our identification of two distinct inflammatory fibroblast states synthesize findings from multiple tissue-specific studies. Our comparison to mouse models predict that Notch3 based therapeutic approaches that are successful pre-clinical models of arthritis may also work on bleomyocin-induced pulmonary fibrosis but not in DSS-induced colitis. This study serves as a template for larger drug response studies to inform the feasibility of multi-tissue “basket” trials for Immune Mediated Inflammatory Diseases.

ACR Meeting Abstracts - https://acrabstracts.org/abstract/a-single-cell-stromal-atlas-identifies-conserved-fibroblast-phenotypes-expanded-in-the-inflamed-synovium-lung-intestine-and-salivary-gland/