Emerging synovial cell states in rheumatoid arthritis as potential therapeutic targets

Abstract

Purpose of Review

To summarize recently discovered novel cell states in rheumatoid arthritis (RA) synovium that could have important implications for disease treatment.

Recent findings

The use of multiomic technologies including single cell and spatial transcriptomics and mass cytometry has led to the discovery of several novel cell states, which could have important implications for the treatment of RA. These cells can be found in patient blood, synovial fluid, or synovial tissue, and span several immune cell subsets as well as stromal cell types. These diverse cell states may represent the targets of current or future therapeutics, while their fluctuations may inform the ideal timing for therapy. Future efforts are needed to implicate how each cell state functions in the pathophysiologic network within affected joints and how medications perturb each cell state and ultimately the tissue.

Summary

Multiomic molecular technologies have afforded the discovery of numerous novel cellular states in RA synovium; the next challenge will be to link these states to pathophysiology and treatment response.

Introduction

Throughout the disease process, the synovial tissue of patients with rheumatoid arthritis (RA) undergoes extensive change, including increased vascularity, hyperplasia of the intimal lining, and infiltration of a diverse, variable population of immune cells ⁽¹⁾. While a standardized set of clinical parameters are used for diagnosis ⁽²⁾, underlying heterogeneity in the histologic features of synovial tissue suggest the cells comprising swollen, inflamed joints vary considerably between individuals diagnosed with RA. At one end of the spectrum, patient synovium can contain lymphocytes organized into ectopic lymphoid structures (ELSs) that resemble germinal centers, while others are relatively devoid of immune cells and instead largely comprised of fibroblasts and remodeled extracellular matrix (ECM) ^{(3, 4)}. This pathophysiologic heterogeneity may underlie variable responses to medications between patients and pose challenges in clinical trials that enroll patients solely based on a diagnosis of RA.

Recent advances in single cell transcriptomics, mass cytometry, and spatial multiomics have allowed for the identification of novel cellular states in the immune and stromal compartments. Their clinical significance, however, is not necessarily obvious, and further investigation is required to determine if they impart a protective or deleterious role in RA pathogenesis. Indeed, these findings bring us to what can be considered analogous to the sequencing of the human genome—a point that required a subsequent minimum of a decade to inform disease treatments ^{(5, 6, 7, 8, 9)}. The identification of the compendium of cell types in RA synovium thus represents a crucial early step in the discovery of novel therapeutics and treatment decision strategies. In this review, we will outline some of the most recently identified cells, and describe what is currently known about their role in the pathogenesis of RA. Importantly, where possible, we will focus on how these cells could potentially act as therapeutic targets with existing RA drugs, and how their presence in the synovium or blood of individuals may predict how they will respond to a therapy, or provide a guide for the timing of treatment.

STROMAL CELLS

Currently approved medications for RA can largely be categorized as “anti-inflammatory”; intended to target bone marrow-derived immune cells or their secreted products. The synovial stroma, however, has acquired attention as an untapped source of drug targets; encompassing lining and sublining fibroblasts, nerves, adipocytes, and vascular and lymphatic endothelial cells with their associated mesenchymal cells. Reluctance to target the stroma in the past centered on the assumption that these cell types have a similar role and phenotype regardless of the tissue in which they reside. Mounting evidence of tissue-specific features for stromal cells and their distinct influences on disease pathogenesis may provide the needed insight for disease-specific targeting ⁽¹⁰⁾. Further, for a tissue significantly affected by infiltrating immune cells such as the RA synovium, the production of inciting factors has a profound impact on the phenotype of the stroma, therein generating disease-specific resident cell states ripe for drug targeting. While the current anti-inflammatory medications likely intercept their activation to some extent, direct targeting of synovial fibroblast activity in the RA synovium could, for example, derail their own release of pro-inflammatory and angiogenic factors and their invasion into proximal tissue ^{(11, 12, 13, 14)}.

Using the contemporaneous technologies of single-cell (sc) RNA sequencing and mass cytometry, a sublining fibroblast activation state marked by high HLA-DR expression has been shown to be >15 times more abundant in RA compared to osteoarthritis (OA) synovium ⁽⁴⁾. The interferon-ɣ (IFN-ɣ) gene expression signature in these fibroblasts, including high levels of IL6, CXCL12, and MHC class II HLA-DR, suggests robust interactions with T and NK cells ⁽⁴⁾. Fibroblast subsets shared across autoimmune tissues, such as the RA synovium, include CXCL10⁺CCL19⁺ and SPARC⁺COL3A1⁺ subsets ⁽¹⁵⁾ (Table 1). Accordingly, in high-dimensional imaging experiments, CXCL10⁺CCL19⁺ fibroblasts colocalized with T cells, while SPARC⁺COL3A1⁺ fibroblasts were enriched in pathways suggestive of vasculature association and colocalized with perivascular and mural-cell enriched regions. In vitro culture with activated T cell supernatant or endothelial cells caused fibroblasts from distinct tissue origins to converge transcriptionally onto the CXCL10⁺CCL19⁺ or SPARC⁺COL3A1⁺ fibroblast phenotypes, respectively ⁽¹⁵⁾. Additional fibroblast heterogeneity has been previously described, with two distinct subsets of FAPα⁺ populations: FAPα⁺THY1⁺ found in the synovial sub-lining, known as immune effector fibroblasts which produce inflammatory cytokines and chemokines, and FAPα⁺THY1⁻ fibroblasts in the synovial lining layer, which mediate tissue damage with a less pronounced inflammatory signature ⁽¹⁶⁾. These studies demonstrate that when infiltrating immune cells interface with the stromal cells present in the synovial tissue, they mediate phenotypic changes in the stroma. Indeed, while Janus kinase (JAK)/ signal transducer and activator of transcription (STAT) targeting medications were likely intended to directly suppress immune cells, they also block IFN-ɣ and other JAK/STAT responses in fibroblast and endothelial cell ^{(17, 18)}. Further, current medications like tocilizumab, which targets the IL-6R, are very likely interfering with fibroblast-mediated pathways, considering their prolific production of IL-6 and autocrine and paracrine impacts of that cytokine. Lastly, as dominant resident sources of non-proteinaceous factors such as prostaglandins, hyaluronidase and much more, these stromal cells are a rich untapped resource for new insights and therapeutic targets.

Table 1:

Recently identified synovial cell types in rheumatoid arthritis

Cell Type	Markers	Proposed role or interactions in the synovium	Therapies that may target the cell state or interactions	Reference
Fibroblast	CXCL10 CCL19	Interacts with T cells	JAKi anti-IL-6R	Korsunsky et al. Med. 2022
Fibroblast	SPARC COL3A1	Interacts with vasculature	Anti-angiogenics	Korsunsky et al. Med. 2022
B cell	NR4A	Formation of ectopic lymphoid structures	JAKi anti-IL-6R anti-CD20	Meednu et al. Cell Rep. 2022
CD4+ TPH cell	GPR56, LAG-3 CXCR6, CD69	B cell help	anti-CD20	Argyriou et al. Nat Commun. 2022
CD8+ T cell	CD69 CD103	Recurrence of synovitis after remission	anti-TNF JAKi	Chang et al. Cell Rep. 2021
CD8+ T cell	GZMK	Production of TNFɑ and IFNɣ	anti-TNF JAKi	Jonsson et al. Sci Transl Med. 2022
CD8+ T cell	GZMB	Contribution to synovitis and joint tissue destruction	anti-TNF JAKi	Moon et al. Nat Commun. 2023
Macrophage	CXCL10 CCL2	Inflammatory cytokine production	anti-TNF JAKi	Zhang et al. Genome Med. 2021
Macrophage	SLAMF7	Inflammatory cytokine production	anti-TNF JAKi	Simmons et al. Sci Immunol. 2022
Dendritic cell	CD209 CD14	Inflammatory cytokine production, antigen presentation	JAKi	Marzaioli, et al. Front Immunol. 2021

^*Tph: T peripheral helper, IL-6R: IL-6 receptor, JAKi: JAK inhibitors.

Beyond the potential utility of stromal cells in the RA synovium as a therapeutic target, recent work by Orange et al. identified a cell population of stromal origin in the blood that can predict the onset of a period of symptom exacerbation, or “flare” ⁽⁴⁾. These circulating CD45⁻CD31⁻PDPN⁺ pre-inflammatory mesenchymal (PRIME) cells were identified through RNAseq from the blood of collected over a year along with carefully tracked clinical symptoms. Notably, this transcriptional signature was enriched 1 to 2 weeks prior to a flare onset, and was followed by the upregulation of a naïve B cell population. A model was proposed wherein prior to a flare, PRIME cells enter the blood before trafficking to the synovial sublining ⁽¹⁹⁾. If this model proves to be correct, PRIME cells in the blood could conceivably be used to predict the timing and therefore therapeutic prevention or amelioration of an impending flare.

B CELLS

B cells are the source of rheumatoid factor (RF) and anti-cyclic-citrullinated peptide (anti-CCP) autoantibodies (currently known as anti-citrullinated peptide antibodies, ACPA), which are well-established as RA diagnostic and disease activity indicators and contribute to immune complex formation and complement activation in joints. It is estimated that 50–80% of RA patients have elevated blood levels of RF, anti-CCP, or both ⁽²⁰⁾. Additionally, B cells have autoantibody-independent roles, such as secretion of cytokines that induce leukocyte infiltration, as well as activation of autoreactive T cells through antigen presentation ^{(21, 22)}.

Several subtypes of B cells that are expanded in RA synovium have been identified by scRNAseq, including IGHD⁺CD27⁻ naïve cells, IGHG3⁺CD27⁺ memory cells, and XBP1⁺ plasmablasts, which have high expression of immunoglobulin genes and XBP1, a transcription factor for plasma cell differentiation. ITGAX⁺ TBX21⁺ autoimmune-associated B cells, which express markers of interferon-stimulated genes in addition to markers for recently activated B cells, were found to be upregulated almost exclusively in leukocyte-rich RA. These autoimmune-associated B cells, or “ABCs”, are likely very heterogenous, with markers of both germinal centers and plasma cell differentiation ⁽⁴⁾.

Ectopic lymphoid structures (ELSs) that architecturally resemble secondary lymphoid organs can form in the synovial tissue of RA patients, and have been shown to support local production of antigen specific, somatically hypermutated autoantibodies ^{(23, 24, 25)}. Recent work identified a B cell population marked by expression of NR4A, a family of orphan nuclear receptors (Table 1). NR4A B cells express lymphotoxin ⍺ and β as well as IL-6, suggesting a role in the formation of ectopic lymphoid structures, and display somatic hypermutation, class switching, and clonal expansion, indicating in situ differentiation in the joint. A signature associated with these cells was also found to increase in the blood at the same time as flare occurrence, suggesting that naïve B cells are recruited from the blood into the synovial tissue and becoming activated to express NR4A. These cells then emigrate back into the blood where they become detectable by RNAseq. NR4A could represent a novel therapeutic target to ablate autoantibody producing B cells, but at the very least, NR4A B cells can serve as a biomarker for chronic autoantigen stimulation in the RA synovial tissue ⁽²⁶⁾. The IL-6 generated by NR4A B cells could also represent a promising target, as IL-6 is also known to be required for the spontaneous generation of germinal centers in other disease contexts ⁽²⁷⁾.

A recent clinical trial using joint biopsy features to stratify patients for randomized treatment examined the mechanisms of response and non-response to two targeted biologics: rituximab (an anti-CD20 monoclonal antibody) and tocilizumab (an anti-IL6R monoclonal antibody) ⁽²⁸⁾. Patients with little to no synovial B cell molecular signature had a lower response to rituximab compared to tocilizumab. These results represent a powerful clinical proof-of-concept for the molecular characterization of target tissue and subsequent alignment with the clinical response of patients who have tissues with an abundance of the cell types targeted by a specific drug. Studies such as this demonstrate the potential for rational, individualized medicine approaches for the treatment of RA.

T CELLS

CD4⁺ T cells impart a myriad of functions that impact RA pathogenesis, including helping the generation of local antibodies by promoting B cell activation and plasma cell differentiation. PD-1^hiCXCR5⁻ T peripheral helper (T_PH) cells were identified in RA first using a mass cytometry approach in RA synovial tissue and fluid ⁽²⁹⁾, then again using scRNAseq in synovial tissue ⁽⁴⁾. Like PD-1^hiCXCR5⁺ T follicular helper (T_FH) cells, T_PH cells have a B-cell-help program as evidenced by expression of CXCL13, IL-21, and other factors, and can induce plasma cell differentiation in vitro. Unlike T_FH cells, however, they also have a migratory transcriptional program, as evidenced by their expression of chemokine receptors that likely allow their entry into inflamed tissue to initiate the formation of ELSs. Imaging experiments on leukocyte-rich synovial tissue samples found that T_PH and T_FH cells could be found both within and outside lymphoid aggregates, but T_PH cells greatly outnumbered them ⁽²⁹⁾. A more recent study where scRNAseq was performed on RA patient synovial fluid (SF) found that T_PH cells in the SF express GPR56, a G-protein couple receptor, LAG-3, an inhibitory receptor, as well as CXCR6 and CD69, both markers for tissue-resident memory T cells (T_RM) (Table 1). Relative to ACPA- patients, T_PH cells were enriched in the SF of ACPA⁺ RA patients, with an upregulation in the expression of GPR56 and LAG-3. Further studies are needed, but therapies that modulate targets like GPR56 hold considerable potential, considering the distinct and robust expansion of T_PH cells in the RA synovium ⁽³⁰⁾.

The recent discovery of two CD8⁺ T cell subsets have particularly improved our understanding of the role of these cells in the pathogenesis of RA. Using an antigen-induced murine model of RA, Chang et al. showed that CD8⁺ T cells persisted in the joints, even during remission, and that those same CD8⁺ T cells mediated the recurrence of synovitis following secondary antigenic challenge (Table 1). These CD8⁺ T cells expressed T_RM markers on their surface, like CD69 and CD103, and did not egress from the joint following CCL21 injection. They were also found to still be capable of mediating flares 8 months after the initial injection of antigen, and flares could not be initiated if T cells were ablated from the affected joint during the remission period. Importantly, oligoclonal CD8⁺ T_RM cells were identified in RA patient synovial tissue, particularly in leukocyte-poor synovial tissue, suggesting that CD8⁺ T_RM cells also persist in human synovial tissue following the clearance of most immune cells that could conceivably mediate future flares ⁽³¹⁾.

Among the CD8⁺ T cell populations located in the synovium of RA patients, a novel population was recently identified with low cytotoxic potential, but the capacity to produce large quantities of inflammatory cytokines, namely IFN-ɣ and tumor necrosis factor (TNF). Such patterns immediately pinpoint CD8 T cells as critical areas to better understand, particularly as they produce factors targeted by efficacious RA medications like anti-TNFs and JAK inhibitors. The recently described CD8 subset is marked by high expression of Granzyme K (GzmK) and represented the majority of CD8⁺ T cells in RA synovial tissue (Table 1). Unlike GzmB, GzmK is not considered to be cytotoxic, and the authors found both GzmB⁺GzmK⁺ and GzmB⁻GzmK⁺ CD8 T cells had reduced cytotoxic potential, compared to GzmB⁺GzmK⁻ counterparts. Functionally, GzmK was shown to induce expression of cytokines like IL-6 and CCL2 in synovial fibroblasts in culture, alone or in combination with IFN-ɣ. Like innate lymphocytes, GzmK expressing CD8⁺ T cells had the capacity to respond rapidly in a TCR-independent manner to cytokine activation by IL-12 + IL-15 to produce IFN-ɣ. Given the abundance of GzmK expressing CD8⁺ T cells and their capacity to produce large amounts of inflammatory cytokines in response to both antigen dependent and antigen independent signaling, these cells could represent a promising therapeutic target for the treatment of RA, by either blocking GzmK directly to prevent fibroblast activation, or by blocking cytokines which activate this unique T cell population ⁽³²⁾. Additionally, a cytotoxic, clonally expanded GZMB⁺CD8⁺ subpopulation was demonstrated to be activated by citrullinated peptide antigens, in CCP+ RA synovium (Table 1). These cells may represent a new autoreactive T cell subset that directly contribute to joint destruction, and further provide strong rationale for additional development of therapeutics targeting cytotoxic CD8+ T cells ⁽³³⁾.

MYELOID CELLS

Macrophages are highly abundant in inflamed synovial tissue, and play a pivotal role in the pathogenesis of RA, as activated macrophages constitute a source of TNF-α and numerous other inflammatory factors ⁽³⁴⁾. The synovial lining contains tissue-resident macrophages responsible for the maintenance of tissue homeostasis during steady state and establishment and maintenance of remission in RA synovium ⁽¹¹⁾. During an inflammatory response, blood-derived monocytes can robustly invade the tissue and constitute a highly plastic cell type. Monocytes respond to cues in the environment and differentiate into a myriad of states with diverse phenotypes in the RA synovium. These include IL1β-expressing macrophages with a phenotype similar to LPS-activated pro-inflammatory M1 monocytes and macrophages ⁽⁴⁾, and HBEGF⁺ IL-1β⁺ macrophages that can be derived by TNF and prostaglandin E from TNF-exposed fibroblasts. This latter state can promote invasiveness of synovial fibroblasts in an epidermal growth factor (EGFR)-dependent manner ⁽¹⁴⁾. Notably, these two substates are functionally, epigenetically distinct and respond quite differently to medications used for RA ⁽¹⁴⁾. Recently, a CXCL10⁺CCL2⁺ inflammatory macrophage state was identified in a meta-analysis of inflammatory tissue from five different disease states, one of which was RA synovium, and from COVID-19 lung (Table 1). These macrophages are transcriptionally similar to human blood-derived macrophages that have been stimulated by IFN-γ and TNF-⍺ ⁽³⁵⁾, and could thus represent a cell type targeted by both anti-TNF and JAK inhibitor therapies.

An additional macrophage state which has recently been identified in several inflammatory tissues, including RA synovium and COVID-19 lung, and which requires IFN-ɣ for its polarization, are marked by the expression of signaling lymphocytic activation molecule family member 7 (SLAMF7) (Table 1). Engagement of SLAMF7 drives the expression of a highly inflammatory, so-called “superactivated” macrophage phenotype, through activating nuclear factor κB (NF-κB) and mitogen activated protein kinase (MAPK) signaling, and later, autocrine TNF⍺ signaling. These cells could play a role in the response to current anti-TNF therapies, while SLAMF7 itself could represent a novel therapeutic target for the treatment of RA and other inflammatory conditions ⁽³⁶⁾.

Dendritic cells (DCs) are a myeloid cell subset that mediate a crucial link between the innate and adaptive arms of the immune system, and are thus crucial to consider in the pathogenesis of RA. A recent study identified a CD209⁺CD14⁺ subset (CD209 is also known as DC-SIGN) of conventional DCs, which are derived from monocytes and are increased in the blood of RA and psoriatic arthritis (PsA) patients compared to healthy controls (Table 1). This subset also had increased expression of inflammatory cytokines and, when exposed to synovial fluid from RA or PsA patients, increased expression of specific chemokines and chemokine receptors. The activation state of these DCs was driven by JAK/STAT signaling, and while their exact function within the inflamed joint has not yet been elucidated, the inhibition of JAK signaling may act in part by reducing the inflammatory signals perpetuated by these DCs ⁽³⁷⁾.

CONCLUSION

The recent era of high dimensional molecular assays, particularly single-cell transcriptomics, has afforded the identification of a myriad of cell types and activation states in the RA synovium. Each will have with their own network of cell-cell interactions and potential roles in disease pathogenesis. The challenge moving forward will be to define how these cells impact disease. To determine whether they promote, suppress, or play a minor role in pathogenesis will take substantial well-considered effort. Simply finding a cell state in a patient tissue is not enough to justify targeting that cell, as their presence could indicate a role in clearing pathogenic material, for example. Powerful tools towards these efforts will include implementation of carefully designed experiments using ex vivo primary target tissue cultures and humanized animal models. In clinical studies, synovial biopsies can now be used to connect drug responses to molecular patterns at the level of an individual or a subgroup of individuals that may represent a putative disease endotype. Thus, the application of contemporaneous molecular technologies have been a critical first step to understanding the cellular makeup of the heterogenous RA synovium and will also likely serve as invaluable guides in individualized medicine approaches in RA and other rheumatic diseases.

Key Points.

• Multiomic molecular analyses of synovium from individuals with rheumatoid arthritis (RA) has allowed for the identification of novel cell states spanning immune and stromal identities

• Two fibroblast states found across autoimmune tissues, including RA synovium, can now be distinguished by expression signatures suggesting interactions with vasculature versus T cells

• A new subset of B cells expressing the orphan nuclear receptor NR4A has now been implicated in the establishment and/or maintenance of ectopic lymphoid structures in RA synovium

• Newly defined T cell states include a GPR56⁺LAG-3⁺CD4⁺ T_PH cell uniquely enriched in the synovial fluid of ACPA⁺ patients, and a Granzyme K expressing CD8⁺ T cell with low cytotoxic potential yet high production inflammatory factors such as IFN-ɣ and TNF⍺

• Defining which cellular states promote or restrict pathology will require considerable work, including precision clinical trials involving synovial cellular state analyses