Skin-derived myeloid precursors and joint-resident fibroblasts spread psoriatic disease from skin to joints

Abstract

Psoriatic disease initially affects the skin and later extends to the joints. Here, we show a two-step process that orchestrates the spread of inflammation from the skin to the joints. Induction of psoriatic skin disease in photoconvertible mice, followed by sequencing and computational characterization of skin-derived cells in the joints, was used to identify a population of CD2⁺MHC-II⁺CCR2⁺ myeloid precursors that builds a skin-derived myeloid cell compartment in the joints. Single-cell cross-species reference mapping and mitochondrial variant tracing showed an orthologous human cell population. Interactome analysis of the joints showed that in a second step, resident regulatory CD200⁺ fibroblasts regulate the priming of CD2⁺MHC-II⁺CCR2⁺ myeloid precursors, which subsequently control IL-17 expression in T cells. Hence, the spread of inflammation requires a distinct migratory myeloid precursor population and a permissive local tissue environment, similar to tumor metastasis.

Fig. 1. CD11b⁺ myeloid cells migrate from the skin to joints in a model of psoriatic disease.

a, Left, representative micrographs of H&E-stained skin sections of the hind paws of BALB/c and C57BL/6 mice at day 21 with and without IL-23OE. Right, quantification of epidermal thickness at day 21. The graph shows the median, quartiles and minimum–maximum; N = 4 per condition. P values were calculated by one-way analysis of variance (ANOVA) with a Tukey’s post hoc test. b, Left, representative micrographs of H&E-stained ankle sections of BALB/c and C57BL/6 mice at day 21 with and without IL-23OE. Right, quantification of arthritis at day 21. The graph shows the median, quartiles and minimum–maximum; N = 8 per condition. P values were calculated by one-way ANOVA with a Tukey’s post hoc test. c, Representative micrographs of MRI-scanned ankles of BALB/c and C57BL/6 mice at day 21 with and without IL-23OE used for the quantification of arthritis at day 21. Arrowheads indicate inflammation, and stars indicate the talar bone. The graph shows the median, quartiles and minimum–maximum; N = 5 per condition. P values were calculated by one-way ANOVA with a Tukey’s post hoc test; T1w CE, T1-weighted contrast-enhanced. d, Representative micrographs of light sheet fluorescence microscopy of Kaede^tg ankles from BALB/c (C-Kaede^tg) and C57BL/6 (B6-Kaede^tg) background strains at day 7 with and without IL-23OE and after photoconversion of cells localized in the skin. Arrowheads indicate accumulations of photoconverted Kaede^RED cells. Graphical drawings of the tibia (Ti), calcaneus (Ca) and Achilles tendon (Te) are included. e, Representative flow cytometry plots for the quantification of Kaede^RED skin-derived cells in the joint. f, Quantification of Kaede^RED skin-derived cells in the joints. The graph shows the mean and standard error of the mean; N = 4 per time point and condition. g, Representative micrographs of imaging flow cytometry for the typing of Kaede^RED skin-derived cells in the joint at day 7; BF, brightfield. h, Quantification of CD45⁺ Kaede^RED skin-derived cell types in the joints at day 7. The graph shows medians, quartiles and minimum–maximum; N = 4 per condition. P values were calculated by one-way ANOVA with a Tukey’s post hoc test. Source data

Fig. 2. Skin-derived myeloid cells are precursors of mononuclear phagocytes in the joints.

a, UMAP of hashtag-identified ankle joint cells in the scRNA-seq dataset from Kaede^tg mice on BALB/c (4,371 cells, N = 1 library from six pooled animals) and C57BL/6 (4,867 cells, N = 1 library from nine pooled animals) backgrounds on day 7 of the IL-23OE psoriasis model. Left, Kaede^GREEN and Kaede^RED cells are highlighted. Right, subcluster visualization of myeloid phagocytes (BALB/c 1,615 cells; C57BL/6 1,238 cells) with apparent highest abundance of Kaede^RED cells; Mφ, macrophages. b, Scatter plot corresponding to a showing the log transferred cell abundance and the Kaede^RED percentage in each of the identified clusters versus the Benjamini and Hochberg (BH)-adjusted P values calculated using quasibinomial models. The dashed line indicates the significance threshold (BH-adjusted P value of <0.05); T_reg cells, regulatory T cells; NK, natural killer; Mo, monocytes; GMP, granulocyte–monocyte progenitors; pDCs, plasmacytoid DCs. c, Bar plot comparing the ratio of clusters of myeloid cells to the highest Kaede^RED percentage between strains (BALB/c, 1,023 cells; C57BL/6, 819 cells). d, Reference-based annotation of each single cell of the Kaede^RED myeloid cells (BALB/c, 498 cells; C57BL/6, 407 cells) based on the ImmGen mononuclear phagocyte dataset (GSE122108) identified by singleR. e, Velocity streams (left) and pseudotime (right) identified by CellRank of the Kaede^RED myeloid cells visualized over the UMAP. f, UMAP visualization of Kaede^RED myeloid cells with subclusters of H2⁺ phagocytes after manual annotation; PCs, precursors. g, Expression of the most relevant marker genes among each myeloid cluster from f. h, Significantly enriched GO-BP terms in the CD2⁺MHC-II⁺CCR2⁺ myeloid precursor cluster compared to all other cell types in the joint. The selection criteria included a BH-adjusted P value of <0.05; Pos., positive; reg., regulation; stim., stimulus; SF, superfamily. i, Comparison of significantly enriched GO-BP terms between Kaede^RED and Kaede^GREEN myeloid cells. The selection criteria included a BH-adjusted P value of <0.05; Resp. response; IFNγ, interferon-γ; Neg., negative. j, Identified CD2⁺MHC-II⁺CCR2⁺ myeloid precursors using UCell highlighted on the UMAP plot of the hashtag-identified skin leukocytes in the scRNA-seq dataset from Kaede^tg animals with BALB/c (2,494 cells, N = 1) and C57BL/6 (3,024 cells, N = 1) background strains on day 7 of IL-23OE. k, Representative flow cytometry plots for the quantification of CD2⁺MHC-II⁺CCR2⁺ skin-derived myeloid precursors in the joint (left) and their quantification in BALB/c mice over time (right). The graph shows the median, quartiles and minimum–maximum; N = 5, N = 4 and N = 5 per time point and condition. P values were calculated by one-way ANOVA with a Tukey’s post hoc test. l, Quantification of CD2⁺MHC-II⁺CCR2⁺ skin-derived myeloid precursors in the joints of BALB/c and C57BL/6 mice on day 7. The graph shows the median, quartiles and minimum–maximum; N = 5 and N = 6 per time point and condition. P values were calculated by two-sided Mann–Whitney U-test. Source data

Fig. 3. Evidence for a skin–joint axis in humans.

a, UMAP plot of the scVI-integrated scRNA-seq datasets of human PsA synovium (N = 5; E-MTAB-11791) and healthy synovial tissue (N = 3). Identified clusters of myeloid cells are highlighted. b, Heat map of co-occurrence of psoriatic mouse myeloid cell clusters with human PsA myeloid cell clusters when mouse joint annotations were mapped to human cells using scANVI. c, Heat map of co-occurrence of human synovial myeloid cell clusters (PsA N = 5, healthy N = 3) with skin myeloid cell clusters from a human skin scRNA-seq dataset (healthy participants N = 5, participants with psoriasis N = 3, participants with AD N = 4; E-MTAB-8142) when synovial annotations were mapped to skin annotations (retrieved from metadata) using scANVI; Mig., migrating; moDC, monocyte-derived DC 3; Inf. Mac, inflammatory macrophages; Macro, macrophages; Mono macro, monocyte-derived macrophages. d, Proportions of shared mitochondrial variants between synovial and skin clusters of the same lineage for the major clusters in the synovia. The graph shows the median, quartiles and minimum–maximum; N = 3. P values were calculated by one-versus-rest two-sided Wilcoxon rank-sum test with a BH correction. e, Box plot corresponding to d. The synovial myeloid compartment is shown in detail; Prol., proliferating. f, Proportion of CD2⁺MHC-II⁺CCR2⁺ myeloid precursors in psoriasis at risk and PsA compared to healthy in the scRNA-seq dataset of human synovial tissue. The graph shows the median, quartiles (for N > 3) and minimum–maximum; psoriasis at risk N = 2; PsA N = 5; healthy participants (HC) N = 3. Statistically credible (Cred.) and noncredible (N. cred.) changes in abundance were identified using scCODA. g, Proportion of CD2⁺MHC-II⁺CCR2⁺ myeloid precursors in AD and psoriasis compared to healthy in the scRNA-seq dataset of human skin. The graph shows the median, quartiles and minimum–maximum; AD N = 4; psoriasis N = 3; healthy N = 5. Statistically credible and noncredible changes in abundance were identified using scCODA. h, Heat map of co-occurrence of human myeloid cell clusters from the joint (PsA N = 5, healthy participants N = 3) with the PBMC myeloid cell cluster from the proteogenomic dataset (healthy participants N = 20, PsA N = 19, psoriasis N = 24; GSE194315) when joint annotations were mapped to preannotated PBMCs using scANVI. i, Proportion of CD2⁺MHC-II⁺CCR2⁺ myeloid precursors in psoriasis and PsA compared to healthy in the scRNA-seq dataset of human PBMCs. The graph shows the median, quartiles and minimum–maximum; psoriasis N = 24; PsA N = 19; healthy N = 20. Statistically credible and noncredible changes in abundance were identified using scCODA. Source data

Fig. 4. Joint fibroblasts prime migrating precursors to proinflammatory or anti-inflammatory phenotypes.

a, TradeSeq-fitted smooth expression of UCell scores of genes dynamically associated with either BALB/c (top) or C57BL/6 (bottom) mice over pseudotime from Fig. 2e in the scRNA-seq dataset of ankle joints from Kaede^tg mice (BALB/c and C57BL/6 strains) on day 7 of the IL-23OE model; P_adj, adjusted P value. b, UCell scores for GO-BP terms GO:0050728 (anti-inflammatory response) and GO:0050729 (proinflammatory response) on the phagocyte compartment from Fig. 2a stratified by stain and photoconversion status. P values were calculated with a two-sided Wilcoxon signed-rank test and adjusted for multiple comparisons using the BH method. c, Circle plot showing the communication probability between myeloid cells (Kaede^GREEN or Kaede^RED) and other cell types in the joints associated with either BALB/c (top) or C57BL/6 (bottom) mice. Line thickness corresponds to communication probability determined using CellChat; ECs, endothelial cells; Mφ, macrophage. d, UMAP plot of Seurat-identified subclusters among fibroblasts in the scRNA-seq dataset of ankle joints from Kaede^tg mice (BALB/c and C57BL/6 strains) on day 7 of the IL-23OE model. e, Expression of the most relevant marker genes among each fibroblast cluster identified in d. f, Comparison of UCell scores of the C57BL/6-associated gene signature (fibroblast-only differentially expressed genes in C57BL/6 mice compared to BALB/c mice) between different subclusters of joint fibroblasts; DEG, differentially expressed gene. g, C57BL/6-associated relative likelihood of different subclusters of fibroblasts determined by MELD. h, Quantification of CD200⁺ fibroblasts in the joints of BALB/c and C57BL/6 mice on day 7. The graph shows the median, quartiles and minimum–maximum; N = 4 per time point and condition. P values were calculated by one-way ANOVA with a Tukey’s post hoc test. i, Quantification of CD200⁺ fibroblasts in the joints of BALB/c mice on days 2, 4 and 7. The graph shows the median, quartiles and minimum–maximum; N = 4 per time point and condition. P values were calculated by one-way ANOVA with a Tukey’s post hoc test. j, Expression of Cd200r1 among the different cell clusters in the scRNA-seq dataset of ankle joints from Kaede^tg mice (BALB/c and C57BL/6 strains) on day 7 of the IL-23OE model. k, Heat map of median fold change gene expression in macrophage–fibroblast cocultures treated with anti-CD200 antagonist (ant.; OX-90) or isotype control; N = 7; FC, fold change. l, Representative images of MRI scans and micrographs of H&E-stained ankles of C57BL/6 animals at day 21 after IL-23OE with or without anti-CD200 antagonist (OX-90) treatment. MRI quantification of arthritis at day 21 is shown on the right. Arrowheads indicate inflammation, and stars indicate the talar bone. The graph shows the median, quartiles and minimum–maximum; N = 4 per condition. P values were calculated by two-sided Mann–Whitney U-test. m, Representative images of MRI scans and micrographs of H&E-stained ankles of BALB/c animals at day 21 after IL-23OE with or without anti-CD200R1 agonist (ago.; OX-110) treatment. MRI quantification of arthritis at day 21 is shown on the right. Arrowheads indicate inflammation, and stars indicate the talar bone. The graph shows the median, quartiles and minimum–maximum; N = 4 per condition. P values were calculated by two-sided Mann–Whitney U-test. Source data

Fig. 5. CD200–CD200R1 control the onset of joint inflammation.

a, Violin plots of ALRA-imputed gene expression among the identified clusters of human synovial myeloid cells from Fig. 3a. b, Representative histogram and quantification of the median fluorescence intensity (MFI) of marker expression on CD2⁺HLA-DR⁺CD14⁺ circulating monocytes in PBMCs of individuals with PsA; N = 10. P values were calculated by two-sided Mann–Whitney U-test between target markers and isotype controls. c, Representative micrographs of H&E-stained sections of human synovial tissue and selected IMC-stained markers from psoriasis (N = 6) and early PsA (N = 5). d, IMC-based quantification of CD200⁺ fibroblasts and CD2⁺MHC-II⁺CCR2⁺ myeloid precursors in synovial tissue from individuals with psoriasis or early PsA. The graphs show median, quartiles and minimum–maximum; N = 6 and N = 5, respectively. P values were calculated by two-sided Mann–Whitney U-test. e, CD200 expression correlated to NOTCH3 expression in synovial fibroblasts (gray dots) from Extended Data Fig. 4a. The blue line shows the linear fit, and the gray region shows the confidence interval. Two-sided Pearson correlation and BH-adjusted P value are shown. f, Pairwise interaction test between CD2⁺MHC-II⁺CCR2⁺ myeloid precursors, CD200⁺ fibroblasts and T cells with the surrounding immune and mesenchymal cells in the IMC dataset of synovial tissue from individuals with psoriasis (N = 6) or PsA (N = 5); cDCs, conventional DCs. g, Proportion of CD2⁺MHC-II⁺CCR2⁺ precursors interacting with CD200⁺ fibroblasts in the IMC dataset in c. The graphs show the median, quartiles and minimum–maximum; N = 6 and N = 5, respectively. P values were calculated by two-sided Mann–Whitney U-test. h, Distance between CD8a⁺ and CD4⁺ T cells and CD2⁺MHC-II⁺CCR2⁺ myeloid precursors in the IMC dataset from synovial tissue of individuals with psoriasis (N = 6) or PsA (N = 5). The graphs show the median, quartiles and minimum–maximum. P values were calculated by two-sided Mann–Whitney U-test. i, Hierarchically clustered heat map showing the CellChat-calculated interaction probability between subclusters in the scRNA-seq dataset of Kaede^tg animals. Gray tiles are excluded interactions. j, Representative flow cytometry plots showing IL-17A expression in CD8⁺ and CD4⁺ T cells after coculture with CD2⁺MHC-II⁺CCR2⁺ or CD2⁻MHC-II⁺CCR2⁺ myeloid cells. The graphs show quantification of IL-17A expression, correspondingly, as mean, quartiles and minimum–maximum; N = 19. P values were calculated by two-way ANOVA with a Tukey’s post hoc test. Source data

Extended Data Fig. 1. Assessment of psoriasis and arthritis over time.

(a) Macroscopic evaluation of skin-scaling in mice with/without IL-23OE for 21 days (left). Total scaling quantified as area under the curve (AUC, right). Graphs: mean ± SEM (left) and median, quartiles, min-max (right); N = 12 per condition. P values calculated by one-way ANOVA with Tukey´s post hoc test. (b) Histological inflammation scores for ankle regions based on H&E staining in mice with/without IL-23OE for 21 days (left). Quantification of arthritis as AUC (right). Graphs: mean ± SEM (left) and median, quartiles, min-max (right); N = 8 per time point and condition. P values calculated by one-way ANOVA with Tukey´s post hoc test. (c) MRI ankle measurements over time in mice with/without IL-23OE for 21 days (left). Quantification of arthritis as AUC (right). Graphs: mean ± SEM (left) and median, quartiles, min-max (right); N = 5 per condition. P values calculated by one-way ANOVA with Tukey´s test. (d-f) Representative micrographs of H&E-stained sections and representative photographs from skin of ear and tail (d), distal paws (e) and ankles (f) of mice at day 21 with/without IL-23OE. In (e): dashed lines highlight dactylitis; S = skin, B = bone. In (f): dashed squares in the overview identify the areas of corresponding magnifications; black arrows indicate sites of inflammation. Images representative for three independent experiments with similar results. (g) Histological inflammation score of synovia in ankle and tarsal joints based on H&E-stainings in mice with/without IL-23OE for 21 days. Graphs: median, quartiles, min-max.; N = 4 per condition; P values calculated by one-way ANOVA with Tukey´s post hoc test. (h) Micro-CT quantification of tibia from mice at day 21 with IL-23 OE. BV/TV = bone volume/total volume; Tb. N = trabecular number; Tb. Th = trabecular thickness. Graphs: median, quartiles, min-max; N = 10 per condition; P values calculated by one-way ANOVA with Tukey´s post hoc test. (i) Representative micro-CT images of the tibial trabecular network of mice at day 21 with/without IL-23 OE. (j) Representative micro-CT images from hind paws of mice at day 21 with/without IL-23OE. Red and green dashed squares indicating corresponding magnification of metatarsal bone and calcaneus. Source data

Extended Data Fig. 2. Validation of photoconversion.

(a) Graphical cartoon explaining skin-photoconversion of the Kaede^tg mice upon ultraviolet (UV) exposure. (b) Representative micrographs of immunofluorescence microscopy of synovial tissue from hind paw joints of photoconverted (exposed) or not photoconverted (unexposed) Kaede^tg at day 7 with IL-23OE. (c) Quantification of Kaede^RED skin-derived cells in the joints at day 7. Graph shows median, quartiles and min-max; N = 4 per condition; P values were calculated by two-sided Mann-Whitney U-test. (d) Quantification of CD45⁺ Kaede^RED skin-derived cell types in the joints at steady state (day 0 of the IL-23OE model). Graph shows median, quartiles and min-max; N = 4 per condition; P values were calculated by one-way ANOVA with Tukey´s post hoc test. Source data

Extended Data Fig. 3. Mouse single-cell landscape in joints and skin.

(a) Graphical cartoon showing the strategy to cell sort target populations used for scRNA-seq. Target populations were individually hash-tagged before generating the scRNA-seq library. (b) UMAP plot of the hash-tag identified ankle joint cells in the scRNA-seq dataset from Kaede^tg mice (BALB/c and C57BL/6) on day 7 of the IL-23OE model. Refers to Fig. 2a. Detailed annotation with Seurat-identified clusters shown. (c) CellRank identification of terminal (left) and initial states (right) in Kaede^RED myeloid cells. (d) Coarse-grained matrix of transition probabilities between states calculated by CellRank. (e) Significantly enriched GO-BP terms in Ly6c2⁺ monocyte cluster, when compared to all other cell types in the joint. Selection criteria: BH-adjusted P value < 0.05. (f) UMAP plot of the hash-tag identified skin leukocytes in the scRNA-seq dataset from Kaede^tg mice (BALB/c 2,494 cells, N = 1 and C57BL/6 3,024 cells, N = 1) on day 7 of the IL-23OE model. Annotated Seurat identified clusters shown (left). Kaede^GREEN and Kaede^RED cells are highlighted (right). (g) Significantly enriched GO-BP terms in skin CD2⁺MHC-II⁺CCR2⁺ myeloid precursors (PCs) cluster, when compared to other cell types. Selection criteria: BH-adjusted P value < 0.05. (h) Representative flow cytometry plots showing the gating strategy for the identification of CD2⁺MHC-II⁺CCR2⁺ skin-derived PCs in the joint. Red gate on T/B lymphocytes as internal control for CD2 and MHC-II. (i) Quantification of CD2⁺MHC-II⁺ myeloid cells in the skin of BALB/c animals after IL-23OE over time. Graph shows median, quartiles, min-max.; N = 5, per timepoint and condition; P-values calculated by one-way ANOVA with Tukey´s post hoc test. (j, k) Representative flow cytometry plots showing the gating of CD2⁺MHC-II⁺ skin-derived myeloid PCs in the bone marrow (j, left) or small intestine (k, left) and their respective quantification after IL-23OE on day 7 in BALB/c and C57BL/6 mice (j, k right). Graph (j) shows median, quartiles, min-max.; N = 5, per condition; P values calculated by one-way ANOVA with Tukey’s post hoc test. Graph (k) shows median, min-max.; N = 2, per condition; no P values calculated. Source data

Extended Data Fig. 4. Human myeloid compartment in synovium, skin and PBMC of psoriatic disease.

(a) UMAP plot of the identified clusters of major cell types in the scVI-integrated scRNA-seq datasets of human (PsA, N = 5, E-MTAB-11791) and healthy synovial tissue (N = 3). (b) Violin plots of relevant marker genes based on ALRA-imputed gene expression among the identified clusters of human synovial myeloid cells from Fig. 3a. Abbreviation: precursors (PCs). (c) Violin plots of ALRA-imputed gene expression of selectively CD2 and CCR2 among the identified clusters of human synovial myeloid cells from Fig. 3a. (d) Comparison of the UCell score of the gene signature for human joint CD2⁺MHC-II⁺CCR2⁺ myeloid precursors in the myeloid phagocytes of the scRNA-seq dataset of mouse ankle joint among dataset from Fig. 2a. (e) Proportions of shared mitochondrial variants between synovial and skin clusters of the same lineage for all myeloid, T and B cell, fibroblast and endothelial clusters in the synovia. Graph shows median, quartiles, min-max.; N = 3; P values calculated by one-vs-rest two-sided Wilcoxon rank-sum test with BH correction. (f) Normalized network strength for the shared mitochondrial variants among subclusters of myeloid cells in the synovia, when all shared variants are included (left) or shared variants with skin are excluded (right). Graph shows median, quartiles, min-max.; N = 3; P-values calculated by one-versus-rest two-sided Wilcoxon rank-sum test with BH correction. (g-j) Median rank (N = 3) in each iteration of synovia-skin shared variant identification among major cell clusters (g-h) and among myeloid cell clusters (i-j) shown as heatmap (g, i) or boxplot (h, j). Graphs show median, quartiles, min-max. (k) UMAP plot of a scRNA-seq dataset of synovial myeloid cells of treatment naive individuals with rheumatoid arthritis (RA, N = 5, E-MTAB-8322). (l) Violin plots of the ALRA-imputed signature genes for each cluster identified in the RA and PsA synovial myeloid cell scRNA-seq datasets. (m) UMAP plot of integrated CITEseq dataset of human PBMC myeloid cells. Publication annotations along with scANVI identified clusters of CD2⁺MHC-II⁺CCR2⁺ myeloid precursors shown. (n) Expression levels of most relevant CITEseq antibodies among publication clusters of human PBMC dataset along with scANVI identified clusters of CD2⁺MHC-II⁺CCR2⁺ myeloid precursors. Source data

Extended Data Fig. 5. Pro-inflammatory responses in BALB/c ankle joints.

(a) Comparison of the significantly enriched GO-BP terms in the genes dynamically associated with either BALB/c or C57BL/6 strain over pseudo-time as shown in Fig. 4a. Selection criteria: BH-adjusted P value < 0.05. (b) Mean fold change (FC) gene expression in macrophage-fibroblasts co-cultures treated with CD200 antagonist (OX-90) or isotype; N = 7 per condition; P values were calculated by two-sided Wilcoxon matched-pairs signed-rank test. (c) Representative images of MRI-scans and H&E-stained ankles of BALB/c mice at day 14 after IL-23OE with or without adoptive transfer of skin-derived myeloid cells at day 7 with IL-23OE. MRI-quantification of arthritis at day 21. Arrowheads indicate inflammation. Stars indicate the talar bone. Graph shows median, quartiles and min-max; N = 4 per condition; P values were calculated by Mann-Whitney U-test (d) Quantification of CD2⁺MHC-II⁺CCR2⁺ skin-derived myeloid precursors and of CD200⁺ fibroblasts in the joint of BALB/c on day 7 with or without imiquimod (IMQ). Graph shows median, quartiles and min-max.; N = 4 per condition; P values were calculated by Mann-Whitney U-test. (e) Representative micrographs of H&E-stained ankles of BALB/c mice on day 21 after IMQ with or without anti-CD200 antagonist (OX-90) treatment and corresponding quantification of arthritis score; N = 3 mice treated with IMQ and N = 6 mice with IMQ and CD200 antagonist; P values were calculated by Mann-Whitney U-test. Mann-Whitney U-tests in (c), (d) and (e) were all two-sided. Source data

Extended Data Fig. 6. CD2⁺MHC-II⁺CCR2⁺ myeloid precursors trigger IL-17 expression in T cells.

(a) Representative flow cytometry plots showing the gating strategy for the identification of CD2⁺HLA-DR⁺CD14⁺ circulating monocytes (pre-gated on FSC/SSC) in PBMCs of PsA patients. (b) Representative flow cytometry plot showing CD2⁺ HLA-DR⁺ gate on lymphocytes. (c) Scaled CD200 signal intensity correlated to scaled CD55 signal intensity in synovial of fibroblasts from psoriasis (yellow dots) or individuals with PsA (orange dots) from Fig. 4i. Blue line shows the linear fit. Two-sided Pearson correlation and P value are shown. (d) Comparison of the UCell score of the GO-BP “positive regulation of T cell chemotaxis; GO:0010820” among subclusters of myeloid cells in the joint of BALB/c mice in scRNA-seq dataset from Kaede^tg mice. (e) UMAP plot of the subsets of T cells and ILCs in the joint in the scRNA-seq dataset from Kaede^tg mice. (f) Expression of the most relevant marker genes among each subclusters identified in e.