henhao
Sign in
PathologyArthritis & Rheumatology
Peer-reviewed journal article

Integrative scRNA-seq and spatial transcriptomics uncovers distinct macrophage-fibroblast cross-talk in human hip synovium between patients with femoroacetabular impingement and osteoarthritis

Gulzada KulzhanovaiD, Alexis Klee, Mina BotrosiD, Victoria L. HanseniD, John E. ReuteriD, Eliya Tazreena TashbibiD, Eloise FadialiD, Benjamin F. RicciardiiD, Brian D. GiordanoiD, Chia‐Lung WuiD

Abstract

Femoroacetabular impingement (FAI) and synovitis have been recognized as essential factors for developing osteoarthritis (OA) in the hip joints. However, little is known about altered synovial cellular compositions, their associated transcriptomic profiles, and cell-cell interactions between patients with FAI and hip OA. In the current study, by using integrative single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (Spatial-seq), we identified the molecular mechanisms by which synovial cells promote hip OA pathogenesis from FAI. Compared to FAI synovium, epiregulin (EREG)-enriched lining fibroblast-like synoviocytes (FLS) were significantly increased in the hip OA synovium. These EREG+ FLS are pro-inflammatory due to their high expression of CXCL1, IL8 (CXCL8), and MMPs. Furthermore, pseudotime analysis predicts that EREG+ FLS are potentially derived from DPP4+PI16+ sublining FLS. Importantly, analysis of cell-cell interactions indicates that fibroblast growth factor 2 (FGF2) secreted from COL1A1+IGFBP5+ fibrotic macrophages may signal through syndecan 4 (SDC4) expressed by EREG+ lining FLS, inducing the expression of IL6, IL8, MMP1, and PTGS2. The GO term analysis of activated genes downstream of FGF2-SDC4 signaling revealed that biological processes associated with inflammation and angiogenesis were upregulated in hip OA, while mechanical stimulus and skeletal muscle differentiation were dominant in FAI. Moreover, we also found that EREG+CCL20+MMP3hi lining FLS as well as most M{Phi} and monocyte populations are unique to hip OA patients when compared to knee OA and RA patients. The findings of this study offer a groundwork in tailoring novel targets and therapies for FAI and hip OA patients.

PDFDOI
Full-text digest ↓

02 /Specification & Classification

Click any tag to browse matching papers
Citations
12
FWCI
43.94
Field percentile
top 1%
OpenAlex Topics

In-depth summary

✦ Multi-agent verified full-text synthesis

Claude Opus 5.5 · max effort · 8 Nov 2026 · from the full text of version 1 (11 Dec 2024)

In hip osteoarthritis (OA) versus femoroacetabular impingement (FAI) synovium, inflammatory EREG+ lining fibroblasts expand and co-localize more with COL1A1+ fibrotic macrophages; FGF2-SDC4 is their top predicted interaction.

FAI is regarded as an early-phase model of hip OA, but whereas knee OA and rheumatoid arthritis (RA) synovium have been profiled extensively by sequencing, hip synovial cell composition and signaling across FAI-to-OA progression are largely unknown, and synovial-fluid cytokines suggest hip OA differs immunologically from knee OA. The study pairs scRNA-seq with Visium spatial transcriptomics to compare fibroblast and macrophage states, and their ligand-receptor links, between FAI and hip OA synovium.

Key results

  1. ShownEREG+ lining fibroblast-like synoviocytes (FLS) expand in hip OA relative to FAI synovium and carry a pro-inflammatory, matrix-degrading profile; immunofluorescence places the increase in the lining, not the sublining.

    66,658 integrated non-hematopoietic cells (6 FAI, 6 OA patients): PRG4+EREG+ FLS were 24% in OA vs 0.78% in FAI (normalized per condition) and expressed more CXCL1, CXCL8, MMP1 and MMP3 in OA; clusters F-1 (EREG+CCL20+MMP3hi) and F-12 rose significantly in OA regardless of sex (two-way ANOVA, Fisher's LSD). Immunofluorescence: EREG+ FLS significantly increased in OA lining (unpaired t-test), not sublining. · Hip OA synovium exhibited increased inflammation…; The increased proximity of EREG+ FLS and…; Figure legends

  2. SuggestedInferred macrophage-to-lining-FLS interactions are more numerous in FAI, whereas in OA COL1A1+ fibrotic macrophages become the main senders; FGF2-SDC4 is the top-ranked link from COL1A1+IGFBP5+ macrophages to EREG+CCL20+MMP3hi FLS, is upregulated in OA, and the two populations co-localize more there.

    MultiNicheNet: main senders M-4 (MERTK+LYVE1+) and M-7 (LTB+CD69+) in FAI, M-8 and M-11 in OA; predicted FGF2-SDC4 target genes are enriched for inflammation and angiogenesis in OA, mechanical stimulus in FAI. Visium mapping shows more F-1/M-11 co-localization in OA; immunofluorescence (EREG+ FLS and CD68+COL1A1+ macrophages ≤ 5 μm apart) shows more proximity in the OA lining only, not the sublining. · Abstract; COL1A1+IGFBP5+ fibrotic MΦ may modulate…; The increased proximity of EREG+ FLS and…; Figure legends

  3. ShownMyeloid composition shifts with disease: alarmin-expressing S100A8+S100A9+ monocytes and LTB+CD69+ myeloid cells increase in hip OA, COL1A1+IGFBP5+ fibrotic macrophages trend up in OA, and MERTK+LYVE1+ macrophages trend up in FAI.

    9813 integrated CD45+CD14+ myeloid cells (6 FAI, 6 OA patients) fell into 7 conserved groups; clusters M-7 (LTB+CD69+) and M-10 (S100A8+S100A9+) were significantly higher in OA (two-way ANOVA, Fisher's LSD); M-11 (up in OA) and M-4 (up in FAI) were trends only. Female OA patients had significantly more LTB+CD69+ cells than male OA patients. · FAI and hip OA synovial synovium exhibited…; Figure legends

  4. SuggestedTrajectory analysis places EREG+CCL20+MMP3hi lining FLS downstream of DPP4+PI16+ sublining FLS, and motif analysis nominates PBX3, KDM5A and E2F4 as candidate regulators of that transition.

    Monocle 3 trajectory rooted (from prior studies) at DPP4+PI16+ cluster F-3: DPP4 and PI16 fall and PRG4 and EREG rise along pseudotime toward F-1. RcisTarget on F-1-enriched gene modules flagged PBX3, KDM5A, E2F4; PBX3 and KDM5A expression declined, while E2F4 levels stayed similar though the share of E2F4-expressing cells grew. · Differentiation trajectory and transcription…; EREG+CCL20+MMP3hi lining FLS may derive from…; Figure legends

  5. SuggestedHip OA synovium shares several sublining FLS, mural and dendritic-cell programs with knee OA or RA synovium, but its EREG+CCL20+MMP3hi lining FLS carry a distinct signature and most macrophage and monocyte populations diverge.

    Chou et al. knee OA scRNA-seq (GSE152805, n = 3) was re-processed; Zhang et al. heatmaps (n = 73 RA, 9 OA samples) were reused alongside hip heatmaps built from the same knee OA/RA marker genes; DEG analysis showed hip EREG+CCL20+MMP3hi FLS uniquely enriched for IL6, CHI3L2, CXCL9 and IGFBP6, though otherwise similar to knee PRG4+ lining and intimal FLS, and M-2 resembled RA MERTK+S100A8+ macrophages. · Identification of distinct and similar synovial…; Distinct and comparable synovial cell…

Methods, data and reproducibility

Hip synovium from 6 Cam-type FAI and 6 hip OA (secondary to FAI) patients, 3 per sex each, was profiled by 10x Chromium scRNA-seq, and 4 FAI and 6 OA samples by 10x Visium. Seurat integration and label transfer onto Visium spots, Monocle 3 pseudotime with RcisTarget motifs, and MultiNicheNet ligand-receptor inference followed, with immunofluorescence validation and comparison to published knee OA and RA data.

  • Hip synovium scRNA-seq (10x Chromium 3′ v3.1), Cam-type FAI (arthroscopy) vs advanced hip OA secondary to FAI (arthroplasty); 66,658 non-hematopoietic and 9813 myeloid cells after integration · human · n = 12 samples: 6 FAI and 6 hip OA patients, 3 per sex each
  • 10x Visium spatial transcriptomics on 8 μm formalin-fixed paraffin-embedded (FFPE) hip synovium sections, at least 50,000 reads per spot · human · n = 4 FAI and 6 hip OA samples
  • Immunofluorescence on FFPE FAI and hip OA synovium for EREG, COL1A1 and a macrophage marker (CD68 in Results, CD16 antibody listed in Methods); pairs ≤ 5 μm apart scored as proximal · human
  • Public knee OA synovium scRNA-seq (Chou et al.); non-hematopoietic and myeloid cells re-processed and integrated with the same pipeline · human (knee OA patients) · n = 3 knee OA samples · GSE152805
  • Published myeloid and stromal marker heatmaps from Zhang et al. (RA and OA synovium from different joints), reused with the journal's permission · human (RA and OA patients) · n = 73 RA and 9 OA samples
Methods & tools
Cell Ranger, Seurat v5.0.1 (SCTransform, anchor integration, label transfer), Monocle 3, RcisTarget, MultiNicheNet, edgeR, DAVID, GraphPad Prism 10, 10x Genomics Chromium Single Cell 3′ v3.1, 10x Genomics Visium, Illumina NovaSeq 6000
Reagents & environment
Seurat v5.0.1 run in RStudio; Monocle 3; GraphPad Prism 10. Versions of Cell Ranger, RcisTarget, MultiNicheNet, edgeR and R are not given.
Reproducibility gaps
No accession or URL for the new scRNA-seq and Visium data and no code link; feature-count QC cutoffs and number of principal components are not given numerically; Methods list a CD16 antibody while Results report CD68+COL1A1+ macrophages; immunofluorescence sample size not stated.
Limitations (6)
  • Sample size is small, as the authors acknowledge: scRNA-seq on 6 FAI and 6 hip OA patients (3 per sex), with 4 FAI and 6 OA Visium samples. · stated by the authors
  • No healthy hip synovium was profiled; the authors suggest adding healthy tissue to map the full course of hip OA development. · stated by the authors
  • Findings still need validation in primary FAI and hip OA synovial cells and/or transgenic mice; how E2F4, PBX3 and KDM5A steer FLS differentiation, and how EREG drives inflammation in hip OA, are left for future work. · stated by the authors
  • Whether hip subsets resembling knee OA or RA populations are functionally equivalent is unresolved, and the authors note that prior evidence on the role of FGF2 in OA is conflicting. · stated by the authors
  • Tissue digestion differed by condition (0.3% collagenase for three hours for FAI, 0.4% overnight for OA), which could itself change recovered cell-type proportions, the basis of most FAI-versus-OA comparisons here. · noted in this summary
  • FGF2-SDC4 signaling and its downstream genes (IL6, IL8, MMP1, PTGS2) are MultiNicheNet predictions; no perturbation of the ligand or receptor, and no protein-level readout of the pathway, is reported. · noted in this summary

Who should read:Read this if you study fibroblast-macrophage crosstalk in joint disease and want hip-specific, FAI-versus-OA candidates (EREG+ lining FLS, FGF2-SDC4 signaling) to test experimentally.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISKuhns '22Thavaratnam '25Kim '26Xu '26Kragstrup '18Knights '22
66% semantic matchBioinformatics
Benjamin D. Kuhns, John M. Reuter, Victoria L. Hansen+5 more·bioRxiv·May 23, 2022