CXCL10 secreted by SPRY1-deficient epidermal keratinocytes fuels joint inflammation in psoriatic arthritis via CD14 signaling

Abstract

Psoriatic arthritis (PsA) is a multifaceted, chronic inflammatory disease affecting the skin, joints, and entheses, and it is a major comorbidity of psoriasis. Most patients with PsA present with psoriasis before articular involvement; however, the molecular and cellular mechanisms underlying the link between cutaneous psoriasis and PsA are poorly understood. Here, we found that epidermis-specific SPRY1-deficient mice spontaneously developed PsA-like inflammation involving both the skin and joints. Excessive CXCL10 was secreted by SPRY1-deficient epidermal keratinocytes through enhanced activation of JAK1/2/STAT1 signaling, and CXCL10 blockade attenuated PsA-like inflammation. Of note, CXCL10 was found to bind to CD14, but not CXCR3, to promote the TNF-α production of periarticular CD14hi macrophages via PI3K/AKT and NF-κB signaling pathways. Collectively, this study reveals that SPRY1 deficiency in the epidermis is sufficient to drive both skin and joint inflammation, and it identifies keratinocyte-derived CXCL10 and periarticular CD14hi macrophages as critical links in the skin-joint crosstalk leading to PsA. This keratinocyte SPRY1/CXCL10/periarticular CD14hi macrophage/TNF-α axis provides valuable insights into the mechanisms underlying the transition from psoriasis to PsA and suggests potential therapeutic targets for preventing this progression.

Keywords: Arthritis; Chemokines; Dermatology; Immunology; Macrophages.

Figure 1. Epidermis-specific SPRY1-deficient mice spontaneously develop psoriasis-like skin lesions and arthritis.

(A) Macroscopic views of the ears, shaved back skin, and paws from control and Spry1-cKO mice. (B) Ear and back skin thickness; PASI scores of back skin; digit, and paw thickness; and arthritis scores of control and Spry1-cKO mice (n = 6). (C) Incidence of inflammation in ears, back skin, and paws of control and Spry1-cKO mice after tamoxifen treatment (n = 20). (D) SPRY1 gene expression in lesional and nonlesional skin from patients with PsA from the NCBI’s GEO database (GSE205748). (E) SPRY1 gene expression in lesional and nonlesional skin from patients with psoriasis or PsA, and normal skin from patients with ankylosing spondylitis from the GEO database (GSE186063). (F and G) Representative H&E staining of the back skin and ears of control and Spry1-cKO mice. Lower panels show quantification of epidermis thickness respectively (n = 6). Scale bar: 100 μm. (H–K) Representative H&E staining of the digits from paws of control and Spry1-cKO mice (H), scale bar: 500 μm. Boxed areas magnified in (I–K) are the following: epidermal hyperplasia (I), fibroplasia (J), and inflammatory infiltrate (K), black arrows indicate enthesitis, scale bar: 250 μm. Lower panels show histological scores (n = 6). Data are shown as mean ± SEM. P values were determined using 2-tailed unpaired Student’s t test (B, the right panel of E, and F–K), 2-tailed paired Student’s t test (D and the middle panel of E), and 1-way ANOVA (the left panel of E). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 2. Skin and joint inflammation in Spry1-cKO mice exhibit hallmarks of psoriasis and psoriatic arthritis.

(A) Percentages of immune cell subsets in ear skin from control and Spry1-cKO mice determined by flow cytometry (n = 4). (B and C) Heatmap of selected genes based on RNA-Seq data from epidermis (B) and dermis (C) of control and Spry1-cKO mice (adjusted P < 0.05 and |log₂FC| > 0.5, n = 3). (D) GSEA of upregulated or downregulated DEGs in human psoriatic skin (GSE13355) on genes of mouse skin (epidermis plus dermis) ranked by log₂FC between Spry1-cKO and control mice. (E and F) GO pathway enrichment of upregulated DEGs (Spry1-cKO vs. control) in epidermis (E) and dermis (F) (adjusted P < 0.05). (G) Heatmap of selected cytokines in ear skin from control and Spry1-cKO mice detected by Luminex assays (n = 4). (H and I) ELISA quantification of CRP (H) and rheumatoid factor (RF) (I) in plasma from control and Spry1-cKO mice (n = 15). (J and K) T2-weighted MRI (J) and micro-CT (K) images of paws from control and Spry1-cKO mice. Red arrows indicate bone erosions; right, quantification of bone volume/total volume (BV/TV) in erosion areas (n = 3). (L) Safranin O-Fast green staining of articular cartilage in paws of control and Spry1-cKO mice. Scale bar: 100 μm. Right, quantification of Safranin-O intensity (n = 4). (M) TRAP staining of osteoclasts in interphalangeal joints of paws from control and Spry1-cKO mice (scale bar: 100 μm); boxed areas are magnified below (scale bar: 50 μm); red arrows indicate TRAP⁺ osteoclasts. Right, quantification of TRAP⁺ osteoclasts (n = 4). (N) Relative mRNA expression of genes associated with PsA in periarticular tissue from control and Spry1-cKO mice (n = 4). (O) Heatmap of selected cytokines in periarticular tissue from control and Spry1-cKO mice detected by Luminex assays (n = 4). Data are shown as mean ± SEM. P values determined using 2-tailed unpaired Student’s t test (A, H, I, and K–N). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 3. Excessive CXCL10 secreted by SPRY1-deficient keratinocytes promotes psoriatic arthritis–like inflammation.

(A) Network of representative terms of all epidermal DEGs and all dermal DEGs (Spry1-cKO vs. control) commonly enriched by Metascape (P < 0.01). (B) Heatmap of top 10 upregulated chemokine genes in epidermis and dermis of Spry1-cKO mice (adjusted P < 0.05 and |log₂FC| > 0.5, n = 3). (C–E) CXCL10 gene expression in human datasets: normal vs lesional epidermis (psoriasis, GSE166388); lesional vs nonlesional skin (PsA, GSE205748); and comparison across psoriasis, PsA, and ankylosing spondylitis (GSE186063). (F–H) ELISA quantification of CXCL10 in plasma, periarticular tissue, and epidermis of back skin from control and Spry1-cKO mice (n = 4). (I) Relative mRNA expression of Cxcl10 in mouse keratinocytes (n = 4). (J) Flow cytometric histograms and quantification of CXCL10 MFI of CD45^–K14⁺ mouse keratinocytes (n = 4). (K) ELISA quantification of CXCL10 in supernatant of mouse keratinocytes (n = 4). (L) Immunoblotting analysis of protein levels associated with JAK1/2/STAT1 pathway in keratinocytes (left) or NHEKs transfected with siSPRY1 (right), both treated with 50 ng/mL recombinant IFN-γ for 24 hours (n = 3). ELISA quantification of CXCL10 in the supernatant of keratinocytes (n = 3). (M–P) Representative macroscopic views (M), H&E staining (scale bar: left, 500 μm; right, 250 μm) (N), Safranin O-Fast green staining (scale bar: 100 μm) (O), and TRAP staining (scale bar: 100 μm) (P) of paws from age- and sex-matched Spry1-cKO mice treated with anti-CXCL10 antibody and isotype antibody (IgG). Quantification shown below (n = 4). Data are shown as mean ± SEM. P values determined using 2-tailed unpaired Student’s t test (C, right panel of E and F–K, lower left panel of L, and M–P), 2-tailed paired Student’s t test (D, middle panel of E, lower right panel of L), and 1-way ANOVA (left panel of E). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 4. The pathogenic role of keratinocyte-derived CXCL10 in psoriatic arthritis–like inflammation is independent of CXCR3 and TLR4.

(A) Flow cytometric plots (left) and percentages (right) of CD45⁺CXCR3⁺ cells in the periarticular tissue of control and Spry1-cKO mice (n = 4). (B) UMAP plots of immune cells from the periarticular tissue of control and Spry1-cKO mice by scRNA-Seq, showing 15 clusters (n = 3). (C) Dot plots showing the scaled expression of selected marker genes for all immune cell types defined in B. (D) Bar plots showing the distribution of all immune cell types. (E) UMAP plots (left) and violin plots (right) showing Cxcr3 expression in all immune cell types. (F) Representative macroscopic views (left) and arthritis scores (right) of the paws from age- and sex-matched Spry1-cKO mice treated with anti-CXCR3 antibody and isotype antibody (IgG) (n = 4). (G) UMAP plots (left) and violin plots (right) showing Tlr4 expression in all immune cell types. (H) Representative macroscopic views (left) and arthritis scores (right) of the paws from age- and sex-matched Spry1-cKO mice treated with TAK-242 (an inhibitor of TLR4 signaling) and vehicle (n = 4). Data are shown as mean ± SEM. P values were determined using 2-tailed unpaired Student’s t test (A, F, and H). ** P < 0.01.

Figure 5. Periarticular CD14^hi macrophages play a predominant proinflammatory role in the development of psoriatic arthritis–like inflammation.

(A) UMAP plots of macrophages from the periarticular tissue of control and Spry1-cKO mice by scRNA-Seq, showing 5 clusters (n = 3). (B) Dot plots showing scaled expression of selected marker genes for macrophage subsets defined in A. (C) Comparison of the proportions of periarticular macrophage subsets between control and Spry1-cKO mice. (D) Dot plots showing the scaled expression of genes linked to activation and polarization in periarticular CD14^hi macrophages from control and Spry1-cKO mice. (E) KEGG pathway enrichment of upregulated DEGs (Spry1-cKO vs. control) in periarticular CD14^hi macrophages (adjusted P < 0.01). P values were determined using 2-tailed unpaired Student’s t test (C). *P < 0.05, ***P < 0.001.

Figure 6. Periarticular CD14^hi macrophages aggravate psoriatic arthritis–like inflammation by producing TNF-α.

(A) Immunofluorescence staining of CD68 and CD14 in the digits from paws of control and Spry1-cKO mice. Boxed areas are magnified below. Scale bar: 50 μm. (B and C) Flow cytometric plots (B) and frequencies (C) of CD68⁺CD14^hi macrophages, TNF-α⁺CD68⁺CD14^hi macrophages, IL-1β⁺CD68⁺CD14^hi macrophages, and CD68⁺CD14^lo/– macrophages in the periarticular tissue from control and Spry1-cKO mice (n = 4). (D) Flow cytometric analysis of TNF-α, IL-1β, and CD86 expression in RAW264.7 cells treated with blank keratinocyte medium, control KC-CM, and Spry1-cKO KC-CM for 24 hours, followed by incubation with 50 ng/mL LPS and brefeldin A (a protein transport inhibitor) for 6 hours (n = 3). (E) Relative mRNA expression of genes associated with macrophage activation and polarization in RAW264.7 cells treated with blank KC medium, control KC-CM, and Spry1-cKO KC-CM for 24 hours (n = 3). (F–I) Representative macroscopic views (F), H&E staining (left scale bar: 500 μm; right scale bar: 250 μm) (G), Safranin O-Fast green staining (scale bar: 100 μm) (H), and TRAP staining (scale bar: 100 μm) (I) of the paws from age- and sex-matched Spry1-cKO mice treated with anti-CD14 antibody and isotype antibody (IgG). Lower panels show quantification of arthritis scores, total histological scores, Safranin-O intensity, and TRAP⁺ osteoclasts, respectively (n = 4). Data are shown as mean ± SEM. P values were determined using 2-tailed unpaired Student’s t test (C and F–I) and 1-way ANOVA (D and E). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 7. Keratinocyte-derived CXCL10 binds to CD14 and mediates the proinflammatory response in periarticular CD14^hi macrophages.

(A and B) Flow cytometric analysis of TNF-α, IL-1β, and CD86 expression in RAW264.7 cells treated with blank keratinocyte medium plus IgG, control KC-CM plus IgG, Spry1-cKO KC-CM plus IgG, and Spry1-cKO KC-CM plus 2 μg/mL anti-CXCL10 antibody for 24 hours (A), or treated with or without 100 ng/mL recombinant murine CXCL10 for 24 hours (B), followed by incubation with 50 ng/mL LPS and BFA for 6 hours (n = 3). (C) Dot plots showing scaled expression (scRNA-Seq) of potential CXCL10 receptor genes mouse periarticular CD14^hi macrophages. (D) Co-immunoprecipitation of TLR2, TLR4, CD14, and CXCL10 in RAW264.7 cells after recombinant murine CXCL10 treatment. (E) Immunoprecipitation of recombinant murine CD14 and CXCL10 proteins. (F) Confocal images of CD14 and CXCL10 colocalization in RAW264.7 cells after treatment with recombinant murine CXCL10. Boxed areas are magnified below. Scale bar: 100 μm. (G) Optimized binding mode with the lowest binding energy generated by HDOCK, and key residues for interaction between mouse CXCL10 and CD14. Boxed areas are magnified below. Hydrogen bonds are shown as yellow dashed lines and salt bridges as red dashed lines. (H) Immunoblotting analysis of protein levels associated with PI3K/AKT and NF-κB signaling pathways in RAW264.7 cells treated with NC or siCd14, followed by recombinant murine CXCL10 stimulation. (I) Flow cytometric analysis of CD14, CD86, TNF-α, and IL-1β expression in RAW264.7 cells treated with NC or siCd14, followed by recombinant murine CXCL10 stimulation and subsequent incubation with LPS and BFA (n = 3). (J) TSNE plots of 8 immune cell clusters in synovial fluid (SF) from PsA patients (GSE161500). (K) TSNE plots showing CD14 and CXCR3 expression across clusters defined in J. (L) TSNE plots showing single cell–level enrichment of inflammatory response and TNF-α signaling via NF-κB signatures in all immune clusters defined in J by GSVA. (M and N) GSVA for pathway enrichment of CD14⁺ versus CD14^– macrophages (M) and CD14⁺ macrophages versus CXCR3⁺ T cells (N) in PsA SF (GSE161500). Data are shown as mean ± SEM. P values determined using 1-way ANOVA (A) and 2-tailed unpaired Student’s t test (B and I). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.