Human pericardial macrophages suppress cardiac fibrosis through cystatin C signaling after myocardial infarction

Figures

• 

  Figure 1. MI elicits a distinct inflammatory response in the pericardial space. (A and B) Schematic of in vitro coculture setup (A) and quantification of cardiac fibroblast–induced (CFb-induced) collagen gel contraction in response to TGF-β1 stimulation in the absence (CFb) or presence of human PF cells (PFc) (CFb:PFc) (B). n = 31 patient pericardial samples and 27 independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001, 1-way ANOVA with Tukey’s multiple-comparison test. (C) Quantification of CFb-induced collagen gel contraction in response to acellular PF without PFcs (PFsup) or addition of PFcs (PFsup + PFc) from the matching patient. Normalized to CFb contraction in culture medium. n = 23 patient pericardial samples and independent experiments. *P < 0.05, paired 2-tailed t test. (D) Quantification of CFb-induced collagen gel contraction in response to TGF-β1 stimulation in the presence of PFcs from NOCAD or MI patients. Normalized to CFb-alone (without PFc) contraction in response to TGF-β1 stimulation. n = 6 NOCAD patient pericardial samples; n = 8 MI patient pericardial samples, and 14 independent experiments. *P < 0.05, unpaired 2-tailed t test.

• 

  Figure 2. MI promotes phenotypic changes in specific subpopulations of pericardial macrophages. (A) Schematic of PF collection from the 2 patient groups, NOCAD (n = 4) and MI (n = 4), and subsequent analysis using scRNA-seq. (B) UMAP projection of 54,765 PFcs from the 8 patients, identifying 19 distinct states/populations of immune and non-immune cells. (C) Bubble plot with scRNA-seq expression of canonical lineage and subset markers. (D) UMAP projections of PFcs from NOCAD and MI patient samples and stacked plot quantification of cell clusters from NOCAD and MI patient samples. *P < 0.05, generalized linear mixed-effects model. (E) Kernel density estimates depicting magnitude of molecular response elicited by pericardial immune cell subsets in both patient groups (left) and those upregulated (right) compared with downregulated (middle) in the MI group relative to the NOCAD group, calculated by summing of differentially expressed gene fold changes for each cell cluster.

• 

  Figure 3. MI dictates maturation trajectories for pericardial macrophages. (A) Louvain clustering of pericardial macrophages represented by state, disease group, and individual patient and overlaid with velocity vector fields to generate a latent time plot. (B) UMAP plotting RNA velocity analysis of subclustered pericardial macrophages undergoing state transitions, colored by cluster ID. (C) Stacked bar plot depicting cluster composition of clinical cohorts examined. (D) UMAP colored by pericardial macrophage clusters and overlaid with summary path curves based on vector fields and pericardial macrophage state composition to determine macrophage states. (E and F) RNA gene expression of selected genes driving divergent maturation trajectories (E) and associated defining markers (F). (G) Flow cytometry analysis and quantification of pericardial immune populations in combined NOCAD and MI samples. (H and I) Flow cytometry–based representation of pericardial macrophage populations depicted by t-SNE plots (H) and expression of protein markers by histogram plots in FN1hi and FN1lo pericardial macrophage populations (I). (J) Quantification of FN1hi and FN1lo pericardial macrophage populations in…

• 

  Figure 4. Gene regulatory and signaling networks for pericardial macrophages. (A) Gene regulatory network (GRN) clustering and regulon density in cluster 0 (c0) and cluster 1 (c1) pericardial macrophages. (B and C) Gene regulatory networks preferentially driving cluster 0 (IRF2, HOXB6; B) and cluster 1 (PPARG, MXD1, XBP1; C) pericardial macrophage profiles. (D and E) UMAP plotting of signaling pathways associated with pericardial macrophages by NICHES analysis (D) and representation of cluster 0 and 1 pericardial macrophages within these signaling clusters (E). (F and G) Stacked bar plots depicting signaling cluster representation between disease states (F) or pericardial macrophage clusters (G). (H) UMAP plotting of expression distribution for PDGFB and FN1; CCL2, CCL13, CALM3, CALM1, and CALR; C3; WNT5A; VEGFA; and PLAU.

• 

  Figure 5. FN1+ pericardial macrophages inhibit cardiac fibroblast activity via cystatin C. (A and B) Schematic of in vitro coculture setup (A) and quantification of cardiac fibroblast–induced (CFb-induced) collagen gel contraction in response to TGF-β1 stimulation in the absence (CFb) or presence of enriched FN1+ pericardial macrophages or enriched pericardial T cells (B). n = 10 T cell–enriched patient pericardial samples; n = 13 FN1+ macrophage–enriched patient pericardial samples, and 13 independent experiments. **P < 0.01, 1-way ANOVA with Tukey’s multiple-comparison test. (C) Quantification of CFb-induced collagen gel contraction in response to TGF-β1 stimulation in the absence (CFb) or presence of direct or indirect (Transwell) pericardial macrophages (CD14+). n = 12 patient PF samples and 11 independent experiments. *P < 0.05; ****P < 0.0001, unpaired 2-tailed t test. (D) scRNA-seq quantification of top differentially expressed secreted factors in cluster 0 (FN1+) pericardial macrophages in either NOCAD or MI patients. (E) Schematic representation of IRF2 target sequence and RNA gene expression of its downstream activated gene, CST3, overlaid on the gene regulatory network…

• 

  Figure 6. Myeloid-derived Cst3 alters cardiac fibrosis. (A) Representative flow cytometry analysis of mouse GPCM expression of FN1, TIM4, and LYVE1. Representative of n = 4. (B) RNA-seq quantification of top-expressed genes by sorted mouse GPCMs. n = 4. (C) qPCR quantification of Cst3 mRNA expression in pericardial macrophages, cardiac macrophages, and Ly6Chi monocytes from WT (Cst3fl/fl) and Cst3ΔLyz2 mice (left), and ELISA quantification of serum (middle) and pericardial lavage (right) CST3 levels from WT and Cst3ΔLyz2 mice. n = 3 for qPCR; n = 8 for serum samples; n = 6 and 4 for pericardial lavage samples from WT and Cst3ΔLyz2 mice, respectively. *P < 0.05, **P < 0.01, unpaired 2-tailed t test for ELISA; ****P < 0.0001, 1-way ANOVA with Tukey’s multiple-comparison test for qPCR. (D) Immunohistochemistry staining and quantification for α-smooth muscle actin (α-SMA) and collagen I (Col I) in the infarct and border zones from WT (Cst3fl/fl) and Cst3ΔLyz2 mice at 7 days after MI. Scale bars: 1,000 μm. Data are represented as mean ± SEM; n = 6 and 8 for WT and Cst3ΔLyz2 mice, respectively. Unpaired 2-tailed t test, non-parametric (Mann-Whitney) for α-SMA and parametric (Welch’s correction)…