Identification of a pro-angiogenic functional role for FSP1-positive fibroblast subtype in wound healing

Abstract

Fibrosis accompanying wound healing can drive the failure of many different organs. Activated fibroblasts are the principal determinants of post-injury pathological fibrosis along with physiological repair, making them a difficult therapeutic target. Although activated fibroblasts are phenotypically heterogeneous, they are not recognized as distinct functional entities. Using mice that express GFP under the FSP1 or αSMA promoter, we characterized two non-overlapping fibroblast subtypes from mouse hearts after myocardial infarction. Here, we report the identification of FSP1-GFP⁺ cells as a non-pericyte, non-hematopoietic fibroblast subpopulation with a predominant pro-angiogenic role, characterized by in vitro phenotypic/cellular/ultrastructural studies and in vivo granulation tissue formation assays combined with transcriptomics and proteomics. This work identifies a fibroblast subtype that is functionally distinct from the pro-fibrotic αSMA-expressing myofibroblast subtype. Our study has the potential to shift our focus towards viewing fibroblasts as molecularly and functionally heterogeneous and provides a paradigm to approach treatment for organ fibrosis.

Fig. 1

FSP1 and αSMA mark distinct populations after multiple types of tissue injury. a Confocal analysis of FSP1⁺ and αSMA⁺ cell populations from heart 2, 8, 12, 20, and 30 days post myocardial infarction (MI; ×40). Uninjured heart was used as control (n = 3). b Confocal analysis of representative 1 µM histologic sections from murine skin stained for FSP1⁺ and αSMA⁺ fibroblast populations 4 and 7 days after excisional full-thickness cutaneous wound. Uninjured skin was used as control (×40) (n = 3). c Confocal analysis of representative histologic sections of murine kidney stained for FSP1⁺ and αSMA⁺ fibroblast populations 7 and 14 days after unilateral ureteral ligation. Contralateral uninjured kidney was used as control (×40) (n = 3). d FSP1 and αSMA staining on histological sections of uninjured (n = 1) and post myocardial infarction human hearts (n = 3)

Fig. 2

FSP1⁺ and αSMA⁺ cells exhibit the molecular phenotype and functionality of fibroblasts. a Schematic of activated fibroblast isolation 10 days post myocardial infarction (MI) from murine hearts. b, c Representative dot plot for fluorescence assisted cell sorting (FACS) of FSP1-expressing (GFP⁺/CD31⁻/CD45⁻) and αSMA-expressing (GFP+) fibroblasts from FSP1-GFP and αSMA-GFP mice left ventricle 10 days after MI. Uninjured fibroblasts were used as unstained negative control. CD45 and CD31 antibodies were used to gate out hematopoietic and endothelial cells, respectively. d Immunofluorescence staining of GFP⁺ populations (P0) from left ventricle of uninjured, FSP1-GFP, or αSMA-GFP sorted by FACS showed FSP1⁺ and αSMA⁺ as unique fibroblast populations, which express fibroblasts markers, such as COL1α1, periostin, and vimentin, but not hematopoietic marker CD45 or the endothelial marker CD31. Nuclei were stained with DAPI (n = 3). Scale bar = 100 µM; ×10 magnification. e Representative FACS overlay histogram of uninjured, FSP1⁺, and αSMA⁺ fibroblasts (P3–P5) to evaluate expression of MEF-SK4. Rat IgG was used as negative control (n = 2). f Representative picture of collagen gel contraction in the presence of uninjured, FSP1⁺, and αSMA⁺ fibroblasts (P3–P5) (left panel, top). Percentage change in the initial gel area following 24 and 48 h of contraction in the presence of uninjured, FSP1⁺, and αSMA⁺ fibroblasts (left panel, bottom) (n = 2). Fold change in proliferative ability of FSP1⁺ and αSMA⁺ fibroblasts (P3–P5) compared with uninjured fibroblasts measured by Brdu proliferation assay (right panel). ^nsp > 0.05 was calculated by one-way ANOVA, n = 3 experiments were performed; ns = not significant. Cell isolation and FACS were performed at least three independent times for staining and proliferation assay and two independent times for contraction assay and MEF-SK4 flow analysis. For each sorting, cells were isolated from pooled homogenates from three to four injured murine hearts

Fig. 3

FSP1⁺ fibroblasts do not express pericyte marker, AN2/NG2. a Staggered FACS histogram representing AN2/NG2 expression in FSP1⁺ and uninjured fibroblasts (P3–P5). Human retinal pericytes were used as positive control and secondary only antibody was used as a negative control. b Relative expression of An2/Ng2 in cultured uninjured, FSP1⁺, and αSMA⁺ fibroblasts (P3–P5). p < 0.05 was calculated by one-way ANOVA, n = 3 experiments were performed independently; bar represents mean ± SD. Mesenchymal stem cells (MSCs) were used as a positive control. c Confocal analysis of FSP1 (green), AN2/NG2 (red), and DAPI (blue) staining on post myocardial infarcted human heart; n = 2

Fig. 4

FSP1⁺ fibroblasts do not differentiate into αSMA-expressing fibroblasts. a. Immunofluorescence staining for αSMA protein in fibroblasts (P0) isolated from uninjured and FSP1-GFP mice hearts 10 days post injury and cultured for 72 h in the presence or absence of TGFβ. Nuclei were stained with DAPI (n = 3). αSMA⁺ fibroblasts isolated from αSMA-GFP mice post injury were used a positive control (P0). Scale bar = 100 µM; ×10 magnification b Relative fold change of αSMA expression in both uninjured fibroblast and FSP1⁺ fibroblasts (P3–P5) cultured with or without TGFβ, *p < 0.05; **p < 0.005; ^nsp > 0.05 was calculated by two-way ANOVA, n = 2 experiments for uninjured fibroblasts and n = 3 experiments with FSP1⁺ and αSMA⁺ fibroblasts were performed; bar represents mean ± SD; ns = not significant

Fig. 5

Distinct molecular signatures of uninjured, FSP1⁺, and αSMA⁺ fibroblasts identified by fibrosis array and RNA sequencing. a Heat map from fibrosis array representing magnitude of gene expression of genes expressed in freshly isolated uninjured, FSP1⁺, and αSMA⁺ fibroblasts (n = 3 for uninjured and αSMA⁺ fibroblasts, n = 4 for FSP1⁺ fibroblasts; cell isolation and FACS were performed at least three independent times. For each sorting, cells were isolated from pooled homogenates from three to four injured murine hearts 10 days post MI). FB = fibroblasts b Relative fold change of selected highest expressing genes identified by fibrosis array expressed by FSP1⁺ and αSMA⁺ fibroblasts compared with uninjured fibroblasts. c Relative fold change of upregulated genes expressed in FSP1⁺ fibroblasts compared with αSMA fibroblasts by fibrosis array. d Relative fold change of downregulated genes expressed by FSP1⁺ fibroblasts compared to αSMA⁺ fibroblasts by fibrosis array. e Hierarchical clustering (partial heat map representing log₁₀ values) of RNA transcript reads per kilobase per million mapped reads for genes with significant differential expression (fold difference > 2, p value < 0.05) between uninjured, FSP1⁺, and αSMA⁺ fibroblasts using CLC Genomics Workbench 8.0. f Relative fold change of genes representing key pathways identified by RNA sequencing in αSMA⁺ vs. FSP1⁺ fibroblasts. Genes downregulated in αSMA⁺ fibroblasts are mentioned as genes upregulated in FSP1⁺ fibroblasts. KEGG pathway analysis was performed on the RNA sequencing data on DAVID bioinformatics Resource 6.7 platform

Fig. 6

FSP1⁺ fibroblasts exhibit pro-angiogenic protein signature in vitro. a Relative pixel density of selected angiogenic proteins significantly high in FSP1⁺ vs. αSMA⁺ fibroblast lysates (P3–P5) identified using mouse angiogenesis proteome profiler array. *p < 0.01 and ***p < 0.0001 were calculated by two-way ANOVA using Bonferroni post-test, n = 2 technical replicates were performed; bar represents mean ± SD. b Relative pixel density of selected angiogenic proteins significantly high in αSMA⁺ vs. FSP1⁺ fibroblast lysates (P3–P5) identified using mouse angiogenesis proteome profiler array. **p < 0.001 and ***p < 0.0001 were calculated by two-way ANOVA using Bonferroni post-test, n = 2 technical replicates were performed; bar represents mean ± SD. c (left panel) Quantification of secreted vascular endothelial growth factor (VEGF) by Quantikine VEGF ELISA in the conditioned media (CM) obtained from FSP1⁺ (n = 5) and αSMA⁺ (n = 4) fibroblasts cultured for 72 h at 37 °C. **p < 0.001 was calculated by unpaired t test; bar represents mean ± SD. (Right panel) Quantification of Gremlin 1 by ELISA in cell lysates (n = 2) and CM (n = 4) obtained from FSP1⁺ and αSMA⁺ fibroblasts cultured (P3–P5) for 72 h at 37 °C. ***p < 0.0001 was calculated by one-way ANOVA; bar represents mean ± SD. d Relative fold change of Vegfa (n = 3 for uninjured, n = 4 for FSP1⁺, and n = 2 for αSMA⁺ fibroblasts), Vegfb (n = 2 for uninjured, n = 4 for FSP1⁺, and n = 2 for αSMA⁺ fibroblasts), Grem1 (n = 3 for uninjured, n = 2 for FSP1⁺, and n = 3 for αSMA⁺ fibroblasts), Angpt1 ((n = 3 for uninjured, n = 2 for FSP1⁺, and n = 3 for αSMA⁺ fibroblasts), and Fgf1 (n = 4 for uninjured, n = 5 for FSP1⁺, and n = 4 for αSMA⁺ fibroblasts) transcripts measured by real-time RT-PCR in uninjured, FSP1⁺, and αSMA⁺ fibroblasts (P0–P5) . *p < 0.01, **p < 0.001, and ***p < 0.0001 was calculated by one-way ANOVA; bar represents mean ± SD

Fig. 7

FSP1⁺ fibroblasts are pro-angiogenic in vitro and in vivo. a Representative images from an endothelial cell assembly assay using HUVECs treated with conditioned media (CM) collected from fibroblasts (P3–P5) or FSP1 recombinant protein. In the top row, the endothelial cells were stained with cell tracker orange and the second row contains corresponding bright field images (×20). Scale bar = 100 nM. b Quantification from the endothelial cell assembly assay based on the number of intersecting branch points per field of view (×10). Data represent an average of HUVEC cell assembly from n = 16 for NT, n = 10 for positive control, n = 15 for FSP1⁺ FB CM, n = 7 for FSP1 protein, and n = 10 for αSMA⁺ FB CM images from three independent biological replicates; bar represents mean ± SD. c BrdU proliferation assay performed on HUVECs (P5) in response to either FSP1⁺ fibroblasts conditioned media (with 0.2% serum; n = 3) or FSP1 recombinant protein (10 nM; n = 4). HUVECs cultured in full serum (2%; n = 4) were used as positive control. d Representative images of FSP1⁺ and αSMA⁺ fibroblast soak loaded sponges stained with CD31 to analyze vascular density. SP = sponge matrix, arrows point at positive stain. Scale bar = 500 µM. e Vascular density graphed as percentage of immunopositive CD31 area/total tissue area in histologic sections from granulation tissue. Data represent averages of multiple 40× fields from unpaired samples (n = 6 for αSMA⁺ fibroblast loaded sponges and n = 5 for FSP1⁺ fibroblast loaded sponges). f BrdU proliferation assay performed on HUVECs (P5) in response to NT (n = 4), FSP1⁺ (n = 4), αSMA⁺ (n = 4), and uninjured (n = 3) fibroblast (P3–P5) conditioned media. g Immunofluorescence staining of GFP⁺ FACS sorted fibroblasts (P3–P5) isolated from left ventricle of FSP1-GFP or αSMA-GFP mice indicate that FSP1⁺ and αSMA⁺ cell populations do not express endothelial markers such as vWF or Flk-1 (VEGFR2). HUVECs were used as positive control (n = 3). ^nsp > 0.05, *p < 0.05, **p < 0.001, and ***p < 0.0001 was calculated by one-way ANOVA, n = 3 experiments were performed; bar represents mean ± SD