Mechanotransduction unifies healthy nondiabetic wound healing over time by promoting a Cd14+/C1qa+ fibroblast subpopulation Authors: Katharina S Berryman, Abdelrahman M Alsharif, Mansi Singh, Filiberto Quintero, Amelia B Knochel, Maria Gracia Mora Pinos, Dharshan Sivaraj, Lulejeta Latifi, Fidel Saenz, Pedro Mora Pinos, Andrew C Hostler, Javier Gonzalez, Ben Litmanovich, Sultana M Mojadidi, Jonathan P Yasmeh, Hudson Kussie, Maisam S Jafri, Autumn Lester, William W Hahn, Maia B Granoski, Geoffrey C Gurtner, Kellen Chen Published: 2026 ABSTRACT AbstractDiabetes is a major growing public health concern, but the impact on cellular impairments in chronic wounds remains incompletely understood. Fibroblasts, key players in all phases of wound healing, are particularly responsive to mechanotransduction. In this study, we characterize fibroblast heterogeneity in genetically induced and pathophysiologic diabetic murine wounding models across all wound healing phases. Full-thickness excisional wounds were created on the dorsum of C57BL/6 nondiabetic wild-type mice (denoted as N-DB); wildtype mice fed a high-fat diet to create pathophysiologic diabetes (denoted as P-DB); and leptin-receptor—deficient mice, a genetically induced diabetic model (denoted as G-DB). Tissue was submitted for single-cell RNA sequencing at postoperative days 0, 2, 7, and 30. Both G-DB and P-DB significantly impaired wound healing compared with N-DB. N-DB expressed distinct myeloid-like Cd14+/C1qa+ angiogenic fibroblasts at post-operative days 2 and 7. Mechanotransduction pathways, focal adhesion kinase (FAK) and mitogen-activated protein kinase (MAPK), were consistently upregulated in N-DB fibroblasts and depleted in both diabetic models. G-DB upregulated WNT-activated Dpp4+ fibrotic fibroblasts at postoperative day 7 compared with N-DB. Thus, mechanotransduction pathways are present in physiologically healing fibroblasts and are selectively depleted in both pathophysiologic and genetically induced diabetes. In genetically induced diabetes, overactivation of canonical and noncanonical WNT pathways overwhelms cells and impairs healing processes. Transcriptomic insights into these mechanotransducive perturbations may lead to new therapies for diabetic wound repair.Keywords: Diabetes, Fibroblasts, Mechanotransduction, Single-cell transcriptomics, Wound healing INTRODUCTION Diabetes is a significant cause of morbidity and mortality, affecting approximately 38 million individuals in the United States and projected to double by 2030 (Lin et al, 2018). Up to 25% of all patients with diabetes develop microvascular complications that impair wound healing, leading to non-healing chronic ulcers and pressure ulcers that cost $40—60 billion annually (Berryman et al, 2026; Narayan et al, 2006). Beyond the healthcare costs, patients with impaired wound healing often face reduced QOL, limited mobility, and an increased need for long-term care. Products, such as silvercoated wound dressings, attempt to improve healing by restricting inflammatory cell expansion, but their effectiveness is limited by cytotoxic effects on fibroblasts and keratinocytes (Nešporová et al, 2020). Consequently, there remains a significant need to develop more effective treatments for chronic diabetic wounds. Our laboratory has previously investigated the importance of mechanical signaling in fibroblasts and myeloid cells during acute wound healing and fibrosis (Li et al, 2024; Ma et al, 2018; Wong et al, 2011). For example, we have found that inhibiting the focal adhesion kinase (FAK) mechanotransduction pathway reduces scar formation and accelerates wound healing in healthy, nondiabetic rodent and large animal models by targeting both mechanoresponsive myeloid cells and fibroblasts (Chen et al, 2025, 2022b, 2021). In addition, we have also previously observed that high-glucose conditions in vitro and diabetic wounds in vivo downregulate FAK expression and that inhibiting Calpain 1, a FAK-degrading enzyme, improves diabetic wound healing (Liu et al, 2017). Several previous studies have also shown the importance of fibroblast heterogeneity and plasticity in diabetic and nondiabetic wound healing using single-cell RNA sequencing (scRNA-seq) (Guerrero-Juarez et al, 2019), identifying processes such as myofibroblast regeneration into adipocytes or distinct prohealing fibroblast subsets (Ito et al, 2007; Theocharidis et al, 2022). However, these studies, including our own, did not specifically investigate mechanical signaling pathways or various diabetic states throughout the entire time course of wound healing. A better understanding of the cellular mechanisms underlying altered wound healing over time would enable the development of more efficacious treatments. In this study, we used our splinted excisional wound model (Fischer et al, 2023), which mitigates wound contraction to promote healing through re-epithelialization that is more similar to human healing, on nondiabetic mice, mice with pathophysiologic diabetes mellitus type 2, and mice with genetically induced diabetes mellitus type 2 to comprehensively characterize the transcriptomic changes of fibroblast ecology at multiple time points throughout all phases of wound healing. We identified that mechanotransduction signaling unifies healthy wound healing by promoting a prohealing, myeloid-like, angiogenic fibroblast (AF) subpopulation across time. We also show that impaired wound healing in mice with diabetes activates multiple WNT pathways within another distinct fibrotic subpopulation. These findings highlight the significance of mechanotransduction and fibroblast heterogeneity in diabetic wound healing. RESULTS Mice with pathophysiologic and genetically induced diabetes show significantly delayed wound healing compared with nondiabetic mice To analyze diabetic wound healing, we created excisional wounds on the dorsa of nondiabetic wild-type C57BL/6 mice (denoted as N-DB), C57BL/6 mice fed with a high fat diet for 16 weeks to generate pathophysiologic diabetes (denoted as P-DB), and mice with genetically induced diabetes (denoted as G-DB) (B6. BKS(D)-Lepr db /J db/db) (Figure 1a and b). We performed standardized gross photography every 2 days after wounding and found that both pathophysiologic and genetically induced diabetes significantly impaired wound healing compared with the nondiabetic state (Figure 2a and b). Specifically, N-DB mice demonstrated significantly smaller wound sizes than P-DB and G-DB mice at postoperative days (PODs) 8, 10, 12, 14, and 16 (POD 8: N-DB vs P-DB, P > .05; N-DB vs G-DB, P < .05; P-DB vs G-DB, P > .05; POD 10: N-DB vs P-DB, P < .001; N-DB vs G-DB, P < .05; P-DB vs G-DB, P > .05; POD 12: N-DB vs P-DB, P < .0001; N-DB vs G-DB, P < .001; P-DB vs G-DB, P > .05; POD 14: N-DB vs P-DB, P < .0001; N-DB vs G-DB, P < .001; P-DB vs G-DB, P > .05; POD 16: N-DB vs P-DB, P < .001; N-DB vs G-DB, P < .05; P-DB vs G-DB, P > .05) (Figure 2b). All N-DB mice healed by POD 14, whereas the wounds from both diabetic models remained open significantly longer, until POD 22 in P-DB mice and POD 24 in G-DB mice (N-DB vs P-DB, P < .0001; N-DB vs G-DB, P < .0001; P-DB vs G-DB, P < .05) (Figure 2c). Innovation schematic. (a) Experimental outline of the animal experiments and processing for single-cell analysis. (b) Analysis methods for single-cell RNA-sequencing data. D, day. Excisional wounding shows delay in wound closure in diabetic models. (a) Excisional wound model gross photography showing impaired diabetic wound healing. Genetically induced (G-DB) and pathophysiologic (P-DB) diabetic mouse models showed significantly impaired wound healing compared with healthy nondiabetic (N-DB) mice. (b) Percentage of wound closure over time. Genetically induced and pathophysiologic diabetic mouse models significantly increased wound areas compared with healthy nondiabetic mice at PODs 8, 10, 12, 14, and 16. (c) G-DB significantly increased days until wound closure compared with P-DB (P < .05) and N-DB (P < .001). P-DB significantly increased days until wound closure compared with N-DB (P < .001). (d) UMAP embedding of fibroblasts colored by experimental groups N-DB (gray), P-DB (orange), and G-DB (red). (e) Feature plots of fibroblast defining genes Col1a1, Vim, Col1a2, and Pdgfra. POD, postoperative day; UMAP, uniform manifold approximation and projection. Wound healing promotes heterogeneous fibroblast subpopulations over time in nondiabetic mice and mice with diabetes To fully identify the changes in cellular ecology caused by diabetes, we analyzed excisional wounds from PODs 2, 7, and 30 as well as unwounded skin (POD 0) using scRNA-seq. We identified a subset of 41,645 wound fibroblasts (Figure 2d), which were confirmed using the canonical cell markers Col1a1, Vim, Col1a2, and Pdgfra (Figure 2e). N-DB, P-DB, and G-DB fibroblasts consisted of 9 subpopulations demonstrating distinct transcriptomic signatures (Figure 3a and b and Supplementary Figure S1). Specifically, N-DB fibroblasts exhibited a strong enrichment in Cd14+/C1qa+ AFs at POD 2 (11% of all fibroblasts) and POD 7 (15%) compared with only 2% and 7% of P-DB and G-DB fibroblasts, respectively (Figure 3c). AFs specifically upregulated the angiogenic and myeloid-like genes Cd14, a major marker of myeloid lineage cells and monocyte activation that upregulates basic fibroblast GF (Song and Cho, 2007; Zamani et al, 2013), and C1qa, which increases vessel density in wounds (Bossi et al, 2014), within their top 10 differentiated genes (Figure 3b). AFs also upregulated additional genes commonly expressed in myeloid-derived cells, such as C1qb, Itgam, Pf4, and Lyz2, as identified by Guerrero-Juarez et al (2019) (Figure 3b). scRNA-seq identifies heterogenous fibroblast subpopulations, 2 of which are confirmed by immunofluorescence. (a) Top row: UMAP embedding of fibroblasts split by experimental group with N-DB (gray UMAP), P-DB (orange UMAP), and G-DB (red UMAP). Bottom row: UMAP embedding of fibroblasts colored by subclusters. (b) Dot plot of subpopulation defining genes. X-axis: genes, y-axis: clusters. (c) Temporal distribution of each group within subclusters. X-axes: time points, y-axes: percentages; colors: N-DB (gray), P-DB (orange), and G-DB (red). (d) Representative immunofluorescence images showing C1QA (green), PDGFRA (red), and DAPI (blue) expression at POD 2, 7, and 14 in G-DB and N-DB. (e) Line plot quantification of the immunofluorescence signal in d across POD 2, 7, and 14 for G-DB and N-DB. (f) Representative immunofluorescence images showing DPP4 (green), PDGFRA (red), and DAPI (blue) expression at POD 2, 7, and 14 in G-DB and N-DB. (g) Line plot quantification of the immunofluorescence signal in f across POD 2, 7, and 14 for G-DB and NDB. *P < .05, **P < .01, and ***P < .001. Bar = 50 μm. D, day; POD, postoperative day; scRNA-seq, single-cell RNA sequencing; UMAP, uniform manifold approximation and projection. In contrast, G-DB mice demonstrated a strong (27%) enrichment of Dpp4+/C3+ fibrotic fibroblasts (FFs) at POD 7 compared with only 12% in N-DB and 0% in P-DB mice (Figure 3c). This fibrotic cluster was defined by the WNT-responsive gene Dpp4, which is responsible for the remodeling of the dermis (Jussila et al, 2022), and C3, which drives an IL-17—mediated inflammatory response leading to fibrosis (Liu et al, 2018) (Figure 3b). At POD 7, G-DB mice showed a 35% upregulation of a Dkk3+/Lrrc15+ cluster potentially related to keloid formation and chronic inflammation compared with 20% in N-DB and 11% in P-DB mice (cluster 3) (Supplementary Figure S1). Cluster-defining genes Dkk3 and Lrcc15 are involved in the TGFβ pathway to regulate keloid proliferation and fibroblast activation, respectively (Li et al, 2017; Zeltz et al, 2022) (Figure 3b and Supplementary Figure S1). P-DB mice demonstrated a 41% enrichment in Krt17+/Fabp4+ fibroblasts (cluster 4), which were not present in NDB or G-DB mice (Figure 3b and Supplementary Figure S1). These fibroblasts upregulated both Krt17, which has been shown to promote endothelial-to-mesenchymal transition (EMT) through TGFβ−1 signaling in hepatic stellate cells (Chen et al, 2022a), and Fapb4, which is a marker for differentiated adipocytes and enhances cardiac fibrosis (Shan et al, 2013; Zhu et al, 2022). N-DB, P-DB, and G-DB all demonstrated low (<15%) expression of a Lamc3+/Sobp+ subpopulation (cluster 5) (Figure 3b and Supplementary Figure S1), defined by Lamc3 and Sobp, as found in Thompson et al (2022), and a Cyp26b1+/Cyp2f2+ subpopulation (cluster 6), defined by Cyp26b1 and Cyp2f2, which are expressed in epithelial progenitors after injury (Chen and Fine, 2016). N-DB, P-DB, and G-DB also demonstrated low (<3%) expression of Thbs4+/Nr2f2+ fibroblasts (cluster 7) (Figure 3b and Supplementary Figure S1), identified by the migrationpromoting gene Thbs4 through the β-catenin pathway (Klaas et al, 2021), and Nr2f2, which plays a cell-protective role against ROS (Hiebert and Werner, 2019). All 3 groups expressed a subpopulation of Rpl3+/Rps3+ fibroblasts (cluster 8) with abundant expression of cell growth-inhibiting ribosome-family genes (Rpl3, Rps3) (Kardos et al, 2014) (Figure 3b and Supplementary Figure S1) at 3% or lower. N-DB mice demonstrated a 6% expression of Krt6a+/Krt16+ fibroblasts (cluster 9) compared with <6% expression levels in P-DB and G-DB (Figure 3b and Supplementary Figure S1). This fibroblast subpopulation upregulated canonical markers of epithelial cells, such as Krt6a and Krt16 (He et al, 2020). Early AF activation is preserved in nondiabetic healing, whereas diabetic wounds show premature FFs To confirm our scRNA-seq data on the protein level, we first performed immunofluoresce (IF) staining of AF-defining marker C1QA and canonical fibroblast marker PDGFRA (Figure 3d and e and Supplementary Figure S2). At POD 2, there was no quantitative difference between N-DB and G-DB in the colocalized C1QA/PDGFRA area to total PDGFRA area. However, at POD 7, N-DBs showed significantly more C1QA+ fibroblasts than G-DB (Figure 3d and e), confirming our scRNA-seq results (Figure 3c). At POD 14, G-DB fibroblasts had upregulated C1QA+ fibroblasts compared with N-DB. We also performed protein staining to confirm our identified FF subpopulation using DPP4 colocalized with PDGFRA (Figure 3f and g and Supplementary Figure S3). We found that G-DB mouse wounds significantly upregulated DPP4+ fibroblasts compared with N-DB at PODs 2 and 7 (Figure 3f and g), confirming our scRNA-seq results at POD 7 (Figure 3c). By POD 14, there was no difference in the FFs between N-DB and G-DB cells (Figure 3g). Taken together, these findings confirm on the protein level that unimpaired, nondiabetic wound healing requires precise timing of fibroblast subpopulation differentiation to promote the proper wound healing process. C1QA+ AFs are required during the early stages of healing, potentially to secrete factors to endothelial cells and promote neovascularization. By later time points (POD 14), these cells subside during the proliferation and remodeling phases of wound healing, in which fibroblasts deposit collagen and fibrotic tissue to further close the wound with scar tissue. In the context of dysregulated diabetes, FFs are promoted too soon, during the still early stages of the inflammatory phase at POD 7. Then, AFs spike at later time points, well past when angiogenesis is required. Nondiabetic wound healing upregulates a prohealing, angiogenic, and myeloid-like fibroblast subpopulation that expresses mechanotransduction pathways We specifically analyzed the 8950 AFs, with 3722 at POD 2 and 5222 at POD 7 (Figure 4a), by performing gene set enrichment analysis (GSEA) on significantly upregulated genes in each condition (N-DB, P-DB, and G-DB). All AFs strongly expressed myeloid markers Cd14, C1qa, C1qb, Itgam, Pf4, and Lyz2 (Figure 4b); Cd14 and C1qa were upregulated as the top differentiating genes of this subpopulation. N-DB AFs demonstrated strong upregulation of the FAK pathway at PODs 2 and 7, whereas both P-DB and G-DB AFs demonstrated lower levels of FAK activation that moderately increased at POD 7 (Figure 4c). We have previously investigated the FAK pathway in fibroblasts during acute wound healing and observed that FAK initiates a strong fibrotic response through inflammatory signaling (Chen et al, 2025, 2022b, 2021; Wong et al, 2011) and is inhibited in diabetic wound healing (Liu et al, 2017). Specifically, AFs upregulated the FAK pathway through the genes Kitl, a mediator of keratinocyte-to-melanocyte signaling, cell migration, and proliferation; EMT-promoting Col4a1; and the extracellular matrix gene Tnc, which is essential for angiogenesis and reducing inflammation during wound healing (Wang et al, 2022) (Figure 4d). Angiogenic, myeloid-like fibroblast subpopulation shows upregulation of mechanotransduction pathways. (a) UMAP embedding of angiogenic fibroblast subcluster colored by (left) experimental groups and (right) time points. (b) Feature plots of myeloid-like genes. (c) Violin plot of FAK pathway. (d) Violin plots of FAK pathway—involved genes. (e) Violin plot of MAPK pathway. (f) Violin plots of MAPK pathway—involved genes. (g) Violin plot of IL-17 pathway. (h) Violin plot of IL-17 pathway—involved genes. D, day; FAK, focal adhesion kinase; UMAP, uniform manifold approximation and projection. N-DB AFs also showed a strong upregulation of the MAPK pathway at POD 2, whereas both P-DB and G-DB AFs reduced the expression of this pathway over time (Figure 4e). The MAPK pathway directs cell migration and proliferation (Lee et al, 2018) and consists of genes such as Mapk9 (also known as Jnk2), which is involved in cell migration, maintains cell polarity, and upregulates epidermal GFs (Mitra et al, 2011), as well as Il1r1, which activates fibroblasts through neutrophil infiltration after injury (Cordier-Dirikoc et al, 2022). Both Mapk9 and Il1r1 were highly upregulated in N-DB AFs at both time points, in P-DB AFs solely at POD 2, and in G-DB AFs at both time points but to lower levels than in N-DB AFs (Figure 4f). N-DB AFs also upregulated the MAPK-involved gene Fgf7, a fibroblast and keratinocyte GF (Kelley et al, 1992; Mattei et al, 1995), exclusively at POD 2, whereas both diabetic groups showed minimal expression (Figure 4f). In contrast, G-DB AFs specifically upregulated the IL-17 pathway at POD 2 compared with those of the other groups (Figure 4g). The IL-17 pathway is activated upon wounding and remains active specifically in chronic wounds (Hadian et al, 2019). G-DB AFs expressed IL-17 pathway—defining genes, including Mmp9, which promotes inflammation by maintaining continuous leukocyte infiltration (Ayuk et al, 2016), and Cxcl3, which aids in the differentiation of a prohealing macrophage phenotype (Lin et al, 2022), at POD 2 (Figure 4h). Both P-DB and G-DB AFs also upregulated Tnf, which causes decreased collagen production and impairs wound healing (Liu et al, 2006), at POD 2 (Figure 4h). Nondiabetic wound healing upregulates MAPK FFs We next analyzed the 10,973 FFs, which included 4415 FFs at POD 2 and 5777 at POD 7 (Figure 5a and b). All FFs strongly expressed fibrosis-associated markers Sox5, Lcn2, Serpinb2, Tnc, Mfap2, Mfap5, and H19 (Figure 5c). G-DB specifically upregulated the FFs compared with N-DB mice (29% of all G-DB fibroblasts vs 13% N-DB fibroblasts) at POD 7 (Figure 3c). Given that only 123 P-DB fibroblasts differentiated into FFs, we solely analyzed subsets of N-DB and G-DB FFs, containing 6383 and 3175 cells, respectively (Figure 5a and b). Gene set enrichment analysis of the enriched functional pathways showed a distinct upregulation of genes involved in the MAPK pathway in N-DB FFs during early wound healing (POD 2) (Figure 5d and e), such as Cxcl13, which is involved in mesenchymal differentiation through the mitogen-activated protein kinase—extracellular signal—regulated kinase (MAPK—ERK) pathway (Tian et al, 2015), and Sox5, which promotes cell migration in cancer (Sun et al, 2019) (Figure 5c). N-DB FFs upregulated the MAPK pathway at POD 2 and decreased its expression at POD 7 (Figure 5e), similar to AF MAPK expression shown in Figure 4d. G-DB FFs demonstrated lower levels of MAPK activation that moderately decreased at POD 7 (Figure 5e). We also observed that N-DB FFs promoted Malat1 expression, which increases the migration and proliferation of dermal fibroblasts (Huang et al, 2021), at both PODs compared with G-DB FFs (Figure 5f). Similarly, Cxcl13, which—as mentioned earlier—is linked to fibrotic cell formation and mesenchymal cell differentiation through MAPK—ERK signaling (Dubey et al, 2019; Tian et al, 2015), showed a high expression in N-DB FFs at PODs 2 and 7, with a noticeable downregulation in G-DB FFs at POD 7 (Figure 5f). In addition, Svep1, a profibrotic gene involved in epidermal differentiation (Samuelov et al, 2017), was upregulated in N-DB FFs at PODs 2 and 7 but showed only minimal expression in G-DB FFs at both time points (Figure 5f). Fibrotic fibroblast subpopulation shows upregulation of canonical and noncanonical WNT pathways in diabetic group. (a) UMAP embedding of fibrotic fibroblast subcluster colored by experimental groups. (b) UMAP embedding of fibrotic fibroblast subcluster colored by time points. (c) Differential gene heatmap of genes grouped by experimental group. (d) Bubble plot of select pathways identified by GSEA using EnrichR shows enrichment of MAPK pathway in N-DB and enrichment of WNT pathways in G-DB. (e) Violin plot of MAPK pathway. (f) Violin plot of WNT ß-catenin pathway. (g) Violin plot of WNT EMT pathway. (h) Violin plots of MAPK pathway—involved genes. (i) Violin plots of WNT ß-catenin pathway—involved genes. (j) Violin plot of WNT EMT pathway—involved genes. EMT, epithelial-to-mesenchymal transition; GSEA, gene set enrichment analysis; UMAP, uniform manifold approximation and projection. Diabetes upregulates WNT β-catenin and WNT EMT pathways in FFs G-DB FFs upregulated the canonical WNT β-catenin pathway compared with N-DB FFs at all time points, with both groups demonstrating higher expression at POD 7 than at POD 2 (Figure 5g). The WNT β-catenin pathway regulates cellular differentiation and proliferation and, specifically owing to the β-catenin activation, can also promote microvascular complications of diabetes mellitus type 2 (Sanabria-de la Torre et al, 2022). WNT β-catenin FFs upregulated Mfap2, which promotes invasion in cancer by enhancing EMT (Chen et al, 2020); C1qtnf6, which promotes proliferation, invasion, migration, and angiogenesis through the protein kinase B (AKT) pathway (Huang and Qi, 2023); and Wnt2, which activates the WNT pathway by binding to Frizzled receptors and can promote tumor growth (Bravo et al, 2013) (Figure 5h). G-DB FFs also upregulated the noncanonical WNT EMT pathway (Figure 5i), which is involved in all phases of physiologic wound healing. Interference with this pathway has been associated with the development of chronic, nonhealing wounds (Gumede et al, 2024). The WNT EMT pathway in FFs demonstrated upregulation of genes such as Sox4, which promotes cell migration in cancer (Ruan et al, 2017); Nurp1, which promotes fibrosis (Zhou et al, 2021); and Dmkn, which delays wound healing by inhibiting the expression of neutrophil-recruiting cytokines and thus impairing the acute inflammatory response in the early stages of wound healing (Hasegawa et al, 2013; Zhang et al, 2022) (Figure 5j). Together, these data indicate that diabetes overexpresses both canonical and noncanonical WNT pathways in this distinct FF subpopulation to potentially promote prolonged proliferation and tissue remodeling but ultimately delays wound closure. AFs demonstrate precisely timed FAK-mediated mechanotransduction signaling We performed IF staining for the FAK pathway activator KITL with C1QA and PDGFRA to further confirm our scRNA-seq subpopulation results (Figure 4d). Specifically, we quantified the ratio of KITL+ C1QA+ PDGFRA+ AFs to total C1QA+ PDGFRA+ AFs. At POD 2, when our scRNA-seq data revealed the most pronounced transcriptional changes in Kitl expression (Figure 4d), we observed that N-DB wounds significantly increased KITL expression in C1QA+ PDGFRA+ AFs compared with G-DB wounds (Figure 6a and b). These findings confirm at the protein level that N-DB AFs upregulated FAK mechanotransduction pathways through markers such as KITL (Figures 4d and 6a and b). This FAK pathway expression is especially important at the early POD 2 time point, and it seems likely that expression of the KITL-mediated FAK pathway may be necessary to promote the subsequent upregulation of overall AFs at POD 2 (transcriptionally) (Figure 3c) and POD 7 (both transcriptionally and at the protein level) (Figure 3c and e). Immunofluorescence shows increased KITL+C1QA+PDGFRA+ and decreased WNT2+DPP4+PDGFRA+ fibroblasts in N-DB versus G-DB. (a) Representative immunofluorescence images showing KITL (purple), C1QA (green), PDGFRA (red), and DAPI (blue). (b) Bar plot quantification of the immunofluorescence signal in a across POD 2 for G-DB and N-DB. (c) Representative immunofluorescence images showing WNT2 (purple), DPP4 (green), PDGFRA (red), and DAPI (blue) at POD 7 in G-DB and N-DB. (d) Bar plot quantification of the immunofluorescence signal in c across POD 7 for G-DB and N-DB. *P < .05. Bar = 100 μm. POD, postoperative day. Next, we performed IF staining for the canonical WNT pathway activation marker WNT2, together with FF subpopulation markers DPP4 and PDGFRA. At POD 7, we found that G-DB FFs significantly upregulated WNT2 expression compared with N-DB wounds (Figure 6c and d), confirming our transcriptional findings (Figure 5h). These findings confirmed at the protein level that WNT β-catenin FFs upregulate WNT2 expression at POD 7, potentially delaying wound healing through an increase in total FFs (Figures 3c and g and 6d). DISCUSSION Fibroblasts are key contributors to the cellular ecology of wound healing, and scRNA-seq can be used to investigate transcriptional perturbations within heterogeneous cell populations. Previous characterizations of dermal fibroblasts using scRNA-seq in various disease settings in humans and mice have identified heterogeneous, distinct subpopulations involved in beneficial healing outcomes (Deng et al, 2021; Ganier et al, 2022; Theocharidis et al, 2022). For example, Guerrero-Juarez et al (2019) identified heterogeneous Pdgfra+ wound fibroblasts, also observed in 7 of our 9 subpopulations, and myeloid-derived Cd14+ myofibroblasts persisting through wound regeneration. Similarly, we found Cd14+/C1qa+ AFs with myeloid-like features that support healthy wound healing and are depleted in diabetic models. Sinha et al (2018) corroborated these findings, demonstrating that the conversion of myeloid to fibroblast-like cells is impaired in diabetic wounds. We also observed several other subpopulations within our dataset that agree with the previous literature, such as the Lamc3+/Sobp+ cluster described by Thompson et al (2022). Most of our other identified fibroblast subpopulations did not demonstrate any significant differences in expression levels between our experimental groups. In addition, to our knowledge, these other studies did not investigate diabetic wound-healing fibroblasts throughout the entire wound-healing timeline at the single-cell level. Our laboratory has previously investigated how FAK-mediated mechanical signaling in fibroblasts drives profibrotic transcriptional pathways and how targeting FAK with an inhibitor can promote proper wound healing and mitigate fibrosis (Chen et al, 2022b, 2021; Ma et al, 2018). Furthermore, the MAPK pathway induces collagen production and alpha-smooth muscle actin transcription in fibroblasts (Davis et al, 2012; Stambe et al, 2004), and these processes seem to be important at a baseline level to drive proper fibroblast proliferation and collagen deposition in the dermis during wound healing. Previous studies have provided conflicting findings about whether MAPK provides beneficial or detrimental effects during diabetic wound healing (Kasowanjete et al, 2023; Medicherla et al, 2009). In our study, the myeloid-like, Cd14+/C1qa+ AF subpopulation seems to promote KITL-mediated FAK pathway activation specifically during nondiabetic, early wound healing. Consistent with the current literature, we observed delayed activation of FAK-mediated AFs in diabetic wounds. We also found that MAPK mechanical signaling is consistently upregulated in various healthy fibroblast subpopulations, including both nondiabetic AFs and FFs. Taken together, these findings seem to indicate that FAK- and MAPK-mediated AF activation may be important for unimpaired, physiologic wound healing and is downregulated during diabetes. Interestingly, our data show that AFs activate MAPK through a Mapk9+/Fgf7+/Il1r1+ gene expression profile to potentially promote vascularization and tissue repair, whereas FFs upregulate MAPK through Malat1+/Cxcl13+/Svep1+ genes to potentially induce inflammation, extracellular matrix remodeling, and mesenchymal differentiation (Sivaraj, 2022). Both angiogenesis and extracellular matrix remodeling have previously been found to be downstream of FAK signaling (Schmidt et al, 2021; Wong et al, 2011; Zhao and Guan, 2011), and these findings emphasize the multifaceted importance of FAK—MAPK signaling across different fibroblast phenotypes. In contrast, diabetic FFs upregulated canonical WNT β-catenin and noncanonical WNT EMT pathways. Overexpression of WNT signaling is well-studied and leads to the formation of multiple cancers as well as tumor growth and metastasis (Zhan et al., 2017), yet its effect on wound healing remains poorly understood. Current literature demonstrates conflicting reports about whether the WNT pathways promote diabetes (Lee et al, 2008) or might instead maintain insulin homeostasis (Xu et al, 2019; Yadav and Patel, 2024). The (non)canonical WNT classification differentiates the pathways into ß-catenin dependent (canonical, classic) and independent (noncanonical). The canonical, β-catenin—dependent WNT pathway was discovered first and has been studied in much more detail than the noncanonical pathway. Although there is debate about this binary classification owing to recent discoveries about crosstalk and interconnections between both WNT pathways, we chose to maintain the nomenclature in this study to better categorize our findings within the context of past and current literature. The canonical WNT β-catenin pathway mainly induces cell proliferation, differentiation, and angiogenesis through crosstalk with the protein kinase B (AKT) pathway, whereas the noncanonical WNT EMT pathway regulates cell migration and polarization and inhibits inflammation. Our data seem to indicate that early activation of the WNT pathways is necessary to initiate and coordinate physiologic wound healing. Consistent with these findings, previous studies have implicated that ulcer-derived fibroblasts in human diabetic wounds display aberrant extracellular matrix deposition and sustain chronic periwound remodeling, with concomitant β-catenin accumulation at the nonhealing wound edge (Maione et al, 2016; Stojadinovic et al, 2005). Interestingly, G-DB FFs overactivate both pathways, showing that cells could be in a constant state of proliferation, migration, differentiation, and inflammation that could ultimately overwhelm the cells, leading to the inhibition of physiologic wound healing functions, such as the formation of granulation tissue or angiogenesis. Overall, our study represents a single-cell characterization of dermal wound fibroblasts with confirmation at the protein level of 2 frequently used murine diabetic wound healing models. Our data show that nondiabetic healing upregulates FAK-mediated myeloid-like Cd14+/C1qa+ AFs, whereas both diabetic models downregulate this subpopulation and upregulate transcriptionally different fibroblast subpopulations. Thus, although the diabetic model type may be less important when identifying downregulated beneficial cells than nondiabetic models, careful consideration of the model should be taken in the future when identifying subpopulations that are upregulated during diabetes. Future work should be performed to interrogate other fibroblast subpopulations more closely, such as the altered-adipocyte and keloid-like clusters, as well as crosstalk with other cell types. Developing therapies to augment the FAK-responsive Cd14+/C1qa+ AF subpopulation could represent a promising avenue for improving diabetic wound care. Owing to our observed subpopulation heterogeneity across diabetic models, our data also suggest that a more personalized or model-specific approach could bring improvements to diabetic wound care. MATERIALS AND METHODS Induction of diabetes using genetically induced leptin receptor—knockout mouse model (G-DB) Female B6. BKS(D)-Lepr db /J db/db mice (strain 000697) were purchased from Jackson Laboratories at the age of 6—8 weeks and housed in a temperature-controlled room (71 ° F) on a 12-hour light—dark cycle with water and food ad libitum. Using a high fat diet for a pathologically induced diabetic mouse model (P-DB) Female C57BL/6 mice were purchased from Jackson Laboratories at ages 6—8 weeks and housed in a temperature-controlled room (71 ° F) on a 12-hour light—ark cycle with water and food ad libitum. The mice were started on a high fat feed consisting of 60 kcal% fat (D12492, Research Diets). Fasting oral glucose levels were measured once weekly after a 4-hour fasting period. After 16 weeks of high fat feeding, mice with glucose levels >200 mg/dl were considered diabetic and used for further analysis (Fajardo et al, 2014). Nondiabetic control mouse model (N-DB) Female C57BL/6 mice, purchased from Jackson Laboratories at ages 6—8 weeks, were housed in a temperature-controlled room (71 ° F) on a 12-hour light—dark cycle with water and fed a diet with 13.7 kcal% fat (NIH-31M Diet, Taconic Biosciences) ad libitum. Excisional wounding All mice were age matched for this study. Splinted full-thickness excisional wounds were created on age-matched mice from the above-mentioned mouse models, as previously described by Galiano et al (2004) and Fischer et al (2023). All mice received extended-release buprenorphine (Ethiqa XR, 3.25mg/kg, s.c.) once before surgery for perioperative analgesia. Anesthesia was induced with 5% isoflurane in 100% oxygen at 1L/min and maintained at 1–3% isoflurane throughout the procedure. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) under the protocol IACUC 2021–0828. In this model, a silicone ring is sutured around the wound margin to stent the skin, mimicking human physiologic wound healing by limiting the contraction of the murine panniculus carnosus muscle and allowing the wounds to heal by granulation tissue formation and re-epithelialization. Two full-thickness dermal wounds, 6 mm in diameter, were created on the dorsum of each mouse (N-DB, P-DB, and G-DB) using biopsy punches (Figure 1a). A silicone ring was fixed to the dorsal skin using adhesive glue (Vetbond, 3M, Saint Paul, MN), and 8 interrupted 6—0 nylon sutures were placed around the outer edge of the ring to prevent wound contraction. The wounds were covered with sterile dressings. Digital photographs were taken at the time of surgery and during every dressing change until wound closure. The wounds were harvested on PODs 2, 7, 14, and 30 (n = 5 per group). scRNA-seq The wounds were explanted, microdissected, and enzymatically digested in 20 ml Collagenase Type 1 (Thermo Fisher Scientific) in PBS at a concentration of 5 mg/ml (Januszyk et al, 2020). Subsequently, we minced and rotated the cell digests for 1 hour at 37 ° C and then used a vortex mixer to disrupt the remaining collagenous tissue (Figure 1b). Using a 100 and 70-μm nylon cell filter (Thermo Fisher Scientific), we filtered the cells and quenched the digests in 10% feal bovine serum DMEM. The solutions were centrifuged at 300g for 8 minutes at 4 ° C to pellet the cells, which were then resuspended to a final concentration of 1200 cells/μl in 0.04% BSA (Sigma-Aldrich) in PBS. To avoid overloading, this was done after the maximum capture concentration, per specifications from 10X Genomics (Pleasanton, CA). The cellular suspensions were submitted for droplet-based microfluidic scRNA-seq using the 10X Chromium Single-Cell platform (Single Cell 3′ v3, 10X Genomics) at Stanford Functional Genomics Facility or Arizona Genetics Core Facility at the University of Arizona. To process on the Chromium Controller, we combined a droplet of each suspension, reverse transcription master mix, and partitioning oil. Reverse transcription was performed at 53 ° C for 45 minutes. The cDNA was amplified for 12 cycles total (Bio-Rad C1000 Touch thermocycler), with the cDNA size selected using SpriSelect beads (Beckman Coulter) and a ratio of 0.6 of SpriSelect reagent volume to sample volume. We used the Agilent Bioanalyzer High Sensitivity DNA chip to analyze cDNA for quality control. The cDNA was then fragmented in the proprietary fragmentation enzyme blend for 5 minutes at 32 ° C, followed by end repair and A-tailing at 65 ° C for 30 minutes. cDNA was double-sided size selected using SpriSelect beads. Ligation of sequencing adapters to the cDNA occurred at 20 ° C for 15 minutes. The cDNA was then amplified using a sample-specific index oligo as a primer, followed by a second cycle of double-sided size selection using SpriSelect beads. An Agilent Bioanalyzer High Sensitivity DNA chip was used to analyze the final libraries for quality-control purposes. cDNA libraries were sequenced on a HiSeq 4000 Illumina platform aiming for 50,000 reads per cell. Data processing, FASTQ generation, and read mapping The Cell Ranger (10X Genomics; version 3.1) was used to convert the base calls into reads implementing mkfastq and then align them against the mm10 (mouse genome) using Cell Ranger’s count function with SC3Pv3 chemistry and 5000 expected cells per sample (Januszyk et al, 2020). Cell barcodes representative of quality cells were delineated from barcodes of apoptotic cells or background RNA on the basis of the following thresholds: <10,000 total transcripts, at least 200 unique transcripts, and <10% of their transcriptome of mitochondrial origin. Data normalization and cell population identification Unique molecular identifiers (UMIs) from each cell barcode were retained for all downstream analyses. The raw UMI counts were normalized with a scale factor of 10,000 UMI per cell and subsequently natural log transformed with a pseudo count of 1 using the R package Seurat (version 5.0.) (Hao et al, 2024). Highly variable genes were identified, and cells were scaled by regression to the fraction of mitochondrial transcripts. Aggregated data were evaluated using uniform manifold approximation (UMAP) and projection analysis over the first 15 principal components (Becht et al, 2019). Generation of characteristic subpopulation markers and enrichment analysis Using Seurat’s native FindMarkers function, cell-type marker lists were generated with a log fold change threshold of 0.25 using the receiver operating characteristic (ROC) test to assign predictive power to each gene (Hao et al, 2024). Gene set enrichment analysis (GSEA) was performed on the 100 most highly ranked genes for each cluster of this analysis against pathway databases in a programmatic manner using EnrichR (version 3.4.) (Kuleshov et al, 2016). IF staining and confocal microscopy Tissue was explanted and fixed in 4% paraformaldehyde for 24 hours. The tissue was then dehydrated in ethanol and embedded in paraffin. Sections of 4-μm thickness were generated using a microtome. Deparaffinization with acetone and rehydration with ethanol solutions of decreasing concentrations was performed prior to antigen retrieval with 0.01 M sodium citrate buffer in PBS (Abcam). This was followed by blocking for 2 hours in 5% goat serum (Invitrogen) in PBS. Sections were then incubated with the primary antibodies PDGFRA (1:50; Thermo Fisher Scientific; AB467491) in combination with either C1QA (1:50; AB2852305) or DPP4 (1:50; AB2902128) for double staining and PDGFRA in combination with C1QA and KITL (15 μg/ml; AB2608095) or with DPP4 and WNT2 (15 μg/ml; Novus Biologicals; AF3464) for the triple stain. All primary antibodies were diluted in 5% goat serum and incubated overnight at 4 ° C. Secondary antibodies were applied for 2 hours at room temperature (1:200) (Alexa Fluor 647—conjugated anti-goat [1:200; A-21447], Alexa Fluor 488—conjugated anti-rat [1:200; A-21208], and Alexa Fluor 594—conjugated anti-rabbit [1:200; A-21207]). A Zeiss LSM 880 confocal laser scanning microscope at the imaging facility at the University of Arizona was used for imaging. High-resolution images of the tissue cross-sections were obtained with ×25 magnification, and automatic tile scanning was used to generate a scan of the entire tissue. The ZEN Black software (Carl Zeiss) was used to stitch the individual images together. Quantitative immunofluorescence analysis was performed using a hierarchical normalization approach. For double-stained sections, the marker colocalization area was normalized to the total fibroblast-positive area. Similarly, triple-stained colocalization areas were normalized to the corresponding double-marker colocalization areas. All fluorescent signal quantification was performed within a fixed 2.25-mm 2 region of the dermis to ensure consistent and accurate comparisons across samples. Quantification of the IF staining was determined using MATLAB code developed in previous image analysis studies by the authors (Chen et al, 2018). Supplementary Material Supplementary material is linked to the online version of the paper at www.jidonline.org, and at 10.1016/j.jid.2026.02.027. ACKNOWLEDGMENTS This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (5R01DK074095-17 and U01DK119094) and the National Institute on Aging (R01AG073815). Single-cell RNA sequencing was supported by the Stanford Functional Genomics Facility with funds from the National Institutes of Health (S10OD018220 and 1S10OD021763). We thank the technical staff of the Office of Research, Innovation and Impact’s Arizona Genetics Core Facility at the University of Arizona for providing single-cell RNA sequencing and next generation sequencing services (Facility RRID: SCR_012429). Microscopy was performed at the University of Arizona Office of Research and Partnership’s Imaging Cores - Optical, RRID:SCR_023355. Figure schematics were created with www.BioRender.com. We gratefully acknowledge the invaluable support of our laboratory managers, Nicholas Dean Price and Theresa Carlomagno, whose expertise and dedication were essential to the successful completion of this work. KC is designated as the guarantor for this work and assumes responsibility for the integrity of the study and for addressing any concerns that may arise after publication. Abbreviations: angiogenic fibroblast epithelial-to-mesenchymal transition focal adhesion kinase fibrotic fibroblast mitogenactivated protein kinase immunofluorescence postoperative day single-cell RNA sequencing Footnotes ETHICS STATEMENT We have complied with all relevant ethical regulations for animal testing and research, and all animal work received ethical approval in accordance with the Institutional Animal Care and Use Committee (IACUC 2021–0828) approved by The University of Arizona. DECLARATION OF GENERATIVE ARTIFICIAL INTELLIGENCE (AI) OR LARGE LANGUAGE MODELS (LLMs) The author(s) did not use AI/LLM in any part of the research process and/or manuscript preparation. CONFLICT OF INTEREST The authors state no conflict of interest. DATA AVAILABILITY STATEMENT The single-cell RNA-sequencing data discussed in this publication are deposited in National Center for Biotechnology Information’s Gene Expression Omnibus (Edgar et al, 2002) and are accessible through the Gene Expression Omnibus Series accession number GSE326622. REFERENCES Associated Data Supplementary Materials Data Availability Statement The single-cell RNA-sequencing data discussed in this publication are deposited in National Center for Biotechnology Information’s Gene Expression Omnibus (Edgar et al, 2002) and are accessible through the Gene Expression Omnibus Series accession number GSE326622.