The ontogeny of myeloid-stromal synovial tissue niches in rheumatoid arthritis

Abstract

Recent single-cell multi-omic and spatial analyses of synovial biopsies have transformed our understanding of myeloid cell-driven mechanisms underlying human joint pathology and tissue homeostasis in Rheumatoid arthritis (RA). However, the developmental trajectories of synovial tissue macrophage (STM) subsets in humans remain poorly understood, due in part to the lack of models that faithfully replicate synovial tissue niches. This hinders the exploration of the therapeutic potential of targeting specific synovial myeloid cell clusters. Using multi-omics analyses of synovial tissue from an allogeneic bone marrow transplant recipient, we show that joint-specific tissue-resident STM subsets, including both health- and disease-associated clusters, can derive from peripheral blood monocytes. Analysis of embryonic synovial joints revealed that macrophage localization and maturation in the joints are preceded by local stromal niche specialisation, indicating that synovial fibroblasts (FLS) provide tissue-specific instructive cues to STM precursors. To elucidate human STM developmental trajectories, we established a SNP-based fate-tracking human synovial organoid system by embedding distinct blood-derived myeloid precursors, together with FLS clusters from RA synovial biopsies and endothelial cells, into 3D structures. These organoids reproduced key synovial tissue features, including lining and sublining architecture and stromal-myeloid cell cluster composition. Importantly, they supported differentiation of all resident STM subsets: homeostatic lining TREM2^pos macrophages, their pathogenic TREM2^lowSPP1^pos counterparts that characterize the RA hyperplastic lining, and both homeostatic and RA-associated perivascular LYVE1^pos STM clusters, all traced to monocytic precursors. In summary, we show that development of STM subsets is driven by fibroblast-conditioned spatial niches. We have established a novel, tractable ex vivo platform to dissect the niche-specific cues driving homeostatic versus pathogenic phenotypic clusters.

Figure 1.. PB Monocytes can give rise to all synovial tissue–resident macrophage clusters, including lining TREM2^pos and its pathogenic SPP1^pos phenotype.

(A) UMAP of integrated single-cell transcriptomic data from synovial tissue (ST) and matched peripheral blood (PB) of a patient with heterologous bone marrow (BM) transplant (BMT), and a reference dataset of synovial tissue from healthy controls (n=11), treatment-naïve RA (n=12), c/bDMARD-resistant RA (n=11), and RA in sustained remission (n=11). (B) Same as in (A) but split into PB (n= 8, 20,073 cells, and ST compartments n=45, 60,507 cells). (C) UMAP visualization of integrated single-cell transcriptomic data of ST myeloid cells and matched PB myeloid cells of a patient with a heterologous bone marrow transplant, and a reference ST/PB dataset from RA patients and healthy controls as in A. (D) Differential abundance showing changes in cell population frequencies between conditions. Data are visualized as MiloR neighbourhood graphs, where nodes represent neighbourhoods, coloured by their log fold change across conditions. Neighbourhoods with non-differential abundance (FDR > 10%) are coloured grey, and node size reflects the number of cells in each neighbourhood. (E) Density plots showing VSIG4 and SPP1 expression in myeloid cell atlas as in C. (F) Immunofluorescence images showing expression of CD68, VSIG4 and SPP1 in synovial tissue biopsies of healthy (HC) and of naïve-to-treatment RA patient, representative of at least n=3 healthy and n=3 RA patients. CD68 red; SPP1, and VSIG4 green (RNAscope); and Nuclei, blue. Scale bars, 50μm. (G) Histology of the synovial biopsy and UMAP visualization of synovial tissue cells from a heterologous BM transplant patient, showing donor- or recipient-derived origin as determined by SNP deconvolution (see Methods). (H) Visualization of synovial tissue myeloid cells from a patient with a heterologous BM transplant on the UMAP of the integrated synovial tissue myeloid cell dataset shown in (C), indicating whether the synovial myeloid cells are derived from the bone marrow donor or recipient.

Figure 2.. Establishment of synovial tissue niches in human embryonic joint and organoids.

(A-B) Immunofluorescence staining of PRG4 (yellow), CD68 (red), CD31 (cyan), TREM2 (green) and nuclei (DAPI, white) in human embryonic joints at 9–14 weeks post-conception (pcw), representing data from two joints at 9–10 pcw (A) and four joints at 13–14 pcw (B). (C) Schematic illustrating the workflow of synovial organoid (SO) generation using synovial fibroblasts, PB monocytes, and endothelial cells in a single droplet of Matrigel. All monocyte subsets were sorted from the PB of healthy donors by excluding dead cells and lineage-positive cells, followed by gating based on HLADR and CD88/CD89 expression. Synovial fibroblasts were isolated from synovial biopsies of patients with active RA, while endothelial cells were human umbilical vein endothelial cells (HUVECs). The macroscopic appearance of the SO structure, with clearly defined edges, was observed under bright-field microscopy using a Nikon inverted microscope. (D-E) Representative images of IHC staining of the SO for Collagen I (D) and Collagen III (E) from organoids with FLS derived from biopsies of n=6 patients with active RA. The inserts show isotype controls. The entire SO is shown at 10x magnification, while the enlarged regions of the LL of the SO are shown at 40x magnification. Collagens are brown, and nuclei are blue. (F) Representative confocal microscopy images (40x) showing IF staining for CD31 (red), macrophages (CD45, green) and nuclei stained with DAPI (blue) from organoids with FLS derived from biopsies of n=6 patients with active RA at different time points. The insert shows enlarged region of SO. Scale bars, 100 μm. (G) Representative images of myeloid cell-FLS interactions in the synovial organoids (SO), generated using CellTracker Red-labelled PB monocytes and CellTracker Green-labelled synovial fibroblasts. These images are derived from continuous live recordings conducted using the Omni imaging system at different time points. Data representative of one experiment with three technical replicates.

Figure 3.. Myeloid-synovial stroma organoids recapitulate synovial tissue topography.

(A) Representative confocal microscopy images (40×) showing immunofluorescence staining of the macrophage marker CD68 (green) and lining layer (LL) fibroblast markers PDPN, MMP3 (red), and PRG4 (red), with nuclei stained with DAPI (blue) in synovial tissue biopsy (ST). The inserts show enlarged cells in the selected regions of the tissue. Images are representative of ST biopsies from patients with active RA (n = 3–5). (B–E) Immunofluorescence staining of synovial organoids (SO) showing FLS with PRG4 (red) in (B), MMP3 (red) in (C), CD90 (red) in (D), and PDPN (red) in (E) as well as macrophages with CD45 staining (green) in (E) at different time points of SO development. Nuclei are stained with DAPI (blue) in all images. Data are representative of confocal microscopy images (40×). The inserts show enlarged cells in the selected regions of the SO. Data is representative of three independent experiments. Scale bars 50 μm for 40× magnification images. LL=lining layer, SL=sublining layer.

Figure 4.. Human synovial organoids support development of synovial tissue macrophage clusters.

(A) A representative gating strategy for the SO stromal and macrophage populations after digestion. (B) Flow cytometry validation of the frequency of lining layer (LL) and sublining layer (SL) FLS populations in synovial organoids (FLS from biopsies of n=3 RA patients with active disease, with 3–5 technical replicates). (C) UMAP visualization of integrated (scANVI) single-cell transcriptomics data of synovial organoid FLS and reference synovial tissue FLS. The dataset includes synovial fibroblasts from 15-day-old synovial organoids (31,101 cells) derived from synovial tissue biopsies of n=7 RA patients with active RA, and FLS directly sorted from synovial tissue (7,395 cells, n=7 Active RA tissues, reference). Data were collected across 14 independent experiments. Each dot represents an individual cell (42,179 cells in total). (D) UMAP visualization of integrated myeloid cell clusters from day 15 synovial organoids (11,349 cells) generated using FLS from biopsies of n = 7 patients with active RA, monocyte/DC populations from baseline peripheral blood samples (n = 7 healthy donors, including n=6 matching the 15-day organoids; 8,421 cells), and reference myeloid cells directly sorted from synovial tissue of 34 different RA patients with active disease (22,655 cells), across 14 independent experiments. (E) Proportional distribution of TREM2^pos, TREM2ˡᵒʷSPP1^pos, TREM2^negSPP1^pos, FOLR2^highLYVE1^pos, FOLR2^highEGR1^pos, and FOLR2^highCLEC10A^pos macrophage clusters at baseline (Day 0) and in synovial organoids (SO) on Day 15, based on n = 7 experiments as shown in (D). Data are presented as the median with the interquartile range. Each dot represents a patient derived SO (n=7). A paired T-test between matched baseline monocytes (Day 0) and matched myeloid cells at Day 15 organoids, with exact p-values displayed on the graph. SO=synovial organoids. (F) Schematic of SNP-based human synovial organoid system to deconvolute ST myeloid cell cluster differentiation trajectories. PB DC2, DC3, iDC3, and monocytes, each sorted from the blood of different healthy donors (with different SNP pattern), were embedded together in Matrigel with synovial biopsy-derived fibroblasts from active RA and endothelial cells (HUVEC). (G) UMAP visualizing scRNAseq data of CD45-positive cells (n=2273 cells) from SO as in (F) that was integrated with baseline PB DC2, DC3, iDC3 and monocytes (n=1646 cells) and with down-sampled reference scRNAseq data of RA synovial tissue myeloid cells (n=1915 cells). (H) UMAP as in (G) coloured by the cells from different donors based on a donor unique SNP pattern (Souporcell_v 2.4, as described in the methods). (I) Bar plots with SEM showing the frequency of different ST-DCs and macrophage clusters that developed from different blood precursors. One-way ANOVA with Dunnett’s corrections for multiple comparisons was used. The exact p-values are provided on the graphs. (J) UMAP visualization of integrated myeloid cells from organoid at day 15 and day 21 with matching baseline PB monocytes/DCs (n=2 of baseline and n=4 SO technical replicates). (K) Proportional distribution of TREM2^pos, TREM2ˡᵒʷSPP1^pos, TREM2^negSPP1^pos, FOLR2^highLYVE1^pos, FOLR2^highEGR1^pos, and FOLR2^highCLEC10A^pos macrophage clusters from (J) over time, comparing dynamic of their development from baseline monocytes to day 15 and day 21 synovial organoid macrophages. Data are presented as a connected Mean ± SEM of 4 technical replicates.

Figure 5.. Human synovial organoids support development of LL TREM2^pos, its pathogenic SPP1^pos phenotype and SL LYVE1^pos STMs.

(A) Representative FACS plots showing synovial organoid (SO) macrophage (SOM) gated based on CD45 and HLADR expression at day 15 SO. (B) The frequency of TREM2^posSPP1^neg and TREM2^pos/lowSPP1^pos macrophages within CD45^posHLADR^pos myeloid cells at the baseline (day 0) and day 15 of SO of active RA. Data are represented as Mean ± SEM from 3 biological replicates (FLS from n=3 patients with active RA). (C-D) Immunofluorescence images showing localisation of TREM2^pos (C) and SPP1^pos macrophages (D), CD68, green; TREM2 and SPP1, red; and Nuclei, blue. (E-F) Immunofluorescence images showing presence of LYVE1^pos (green) macrophages near CD31-positive (red) endothelial cell (EC) tube-like structures. (C-F) Data shows representative confocal microscopy images (40x) of SO generated from biopsies of n=3–5 RA patients with active disease in three independent experiments. Scale bars, 50μm. SOM=synovial organoid macrophages. Lining layer, LL; Sublining layer, SL.

Figure 6.. Synovial stroma drive development of lining layer TREM2^pos STMs and its pathogenic TREM2^lowSPP1^pos phenotype that is augmented by inflamed synovial fluid.

(A) SO were generated as in Figure 2C. To mimic acute inflammatory stimulation, at day 7, 10% of cell-free synovial fluids (SF) isolated from active RA joints were added to SO for subsequent 7 days. (B) UMAP visualization of single cell transcriptomics of myeloid cells from synovial organoids cultured with (985 cells) or without 10% synovial fluid (1,160 cells). (C) Proportional distribution of myeloid cell clusters identified by scRNAseq, comparing organoids cultured with and without SF. (D) UMAP feature plot showing the expression levels of SPP1 in myeloid cells at baseline and from synovial organoids (SO) cultured with or without 10% SF. (E) Flow cytometry data showing the percentage of FOLR2 and TREM2 positive macrophages within CD45^posHLADR^pos myeloid cells in synovial organoids cultured with 0%, 10%, 15%, or 20% SF. Expression of FOLR2 in baseline monocytes is also shown while TREM2 on monocytes is shown in Figure 5B. Data is presented as a bar plot showing Mean ± SEM of 2 biological replicates (FLS derived from biopsies of 2 RA patients with active disease) with two or three technical replicates each. Each FLS donor represented by different symbol. Kruskal-Wallis test with Dunn’s corrections for multiple comparison with exact p values on the graphs. (F) Representative dot plot showing surface and intracellular expression of TREM2 and SPP1 in macrophages derived from SO cultured in the absence or presence of different concentration of SF. Macrophages were gated as HLADR and FOLR2 positive. (G) Flow cytometry evaluation of frequency of TREM2^posSPP1^neg and TREM2^low/negSPP1^pos SOM in SO at day 15. Data is presented as a bar plot with Mean ± SEM of three technical replicates. (H-I) Representative immunofluorescence staining of TREM2 (H) and SPP1 (I) positive macrophages (CD68) in SO in the presence of 20% SF. CD68 (green) and Nuclei DAPI (blue) in H-I, TREM2 (red) in H and SPP1 (red) in I. Representative confocal microscopy images (40× magnification) of SO generated with FLS from biopsies of three active RA patients, obtained from three independent experiments. Scale bars, 50μm. Synovial fluid, SF pooled from n=5 RA patients with active diseases. SOM=synovial organoid macrophages.