Novel integrated multiomics analysis reveals a key role for integrin beta-like 1 in wound scarring

Abstract

Exacerbation of scarring can originate from a minority fibroblast population that has undergone inflammatory-mediated genetic changes within the wound microenvironment. The fundamental relationship between molecular and spatial organization of the repair process at the single-cell level remains unclear. We have developed a novel, high-resolution spatial multiomics method that integrates spatial transcriptomics with scRNA-Seq; we identified new characteristic features of cell-cell communication and signaling during the repair process. Data from PU.1^-/- mice, which lack an inflammatory response, combined with scRNA-Seq and Visium transcriptomics, led to the identification of nine genes potentially involved in inflammation-related scarring, including integrin beta-like 1 (Itgbl1). Transgenic mouse experiments confirmed that Itgbl1-expressing fibroblasts are required for granulation tissue formation and drive fibrogenesis during skin repair. Additionally, we detected a minority population of Acta2^high-expressing myofibroblasts with apparent involvement in scarring, in conjunction with Itgbl1 expression. IL1β signaling inhibited Itgbl1 expression in TGFβ1-treated primary fibroblasts from humans and mice. Our novel methodology reveal molecular mechanisms underlying fibroblast-inflammatory cell interactions that initiate wound scarring.

Keywords: Inflammation; Itgbl1; Multiomics; Scarring; Skin Wound Healing.

Figure 1. Dynamic temporal changes in single-cell-derived populations and gene expression at skin wound sites in mice.

(A) Overview of single-cell analysis. (B) UMAP plot of integrated data from 40,024 cells collected on Days 3 (3 d; 4117 cells), 7 (7 d; 21,476 cells), and 14 (14 d; 14,431 cells) post skin wounding, showing each cell type. N neutrophils, BC B cells, TC T cells, M macrophages, D dendritic cells, F fibroblasts, SC Schwann cells, EpC epithelial cells, MC mesenchymal cells, P pericytes, EC endothelial cells. (C) Bar plots of unsupervised hierarchical clustering showing relatedness of wound fibroblast and macrophage subclusters, and proportions of fibroblast and macrophage subclusters on Days 3 (blue), 7 (red), and 14 (orange). (D) UMAP of mRNA expression levels across 40,024 cells at Days 3, 7, and 14 (left panel), and a violin plot showing total mRNA expression levels in the various subclusters (right panel). N: n = 2567; M1: n = 4567; M2: n = 4520; M3: n = 3791; M4: n = 2126; M5: n = 282; M6: n = 270; D1: n = 1,359; D2: n = 734; D3: n = 467; TC: n = 744; BC: n = 122; F1: n = 4265; F2: n = 3849; F3: n = 2218; F4: n = 1421; F5: n = 1369; F6: n = 977; F7: n = 804; F8: n = 532; MC: n = 829; EpC: n = 854; EC: n = 507; P: n = 732; SC: n = 118. Source data are available online for this figure.

Figure 2. Expression of inflammation-related gene candidates in the fibroblast subclusters.

(A) UMAP plots (left) and violin plots (right) of normalized expression levels of the indicated genes from integrated data collected on Days 3 (n = 4117), 7 (n = 21,476), and 14 (n = 14,431) post skin wounding. (B) UMAP plot (left) and violin plot (right) of Col1a1 expression levels from integrated data collected on Days 3 (n = 4117), 7 (n = 21,476), and 14 (n = 14,431) post skin wounding. (C) Normalized Col1a1 expression level in each fibroblast subcluster (F1–F8). F1: n = 4265; F2: n = 3849; F3: n = 2218; F4: n = 1421; F5: n = 1369; F6: n = 977; F7: n = 804. (D) Normalized expression levels of Clec3b, Itgbl1, Ccl11, Lrrc17, and Pi16 in F1 to F8 from integrated data collected on Days 3, 7, and 14 post skin wounding. F1: n = 4265; F2: n = 3849; F3: n = 2218; F4: n = 1421; F5: n = 1369; F6: n = 977; F7: n = 804. Source data are available online for this figure.

Figure 3. Spatial transcriptome analysis of inflammation-related genes using the Visium platform.

(A) Flowchart of the Visium process. (B) Spatial expression of the indicated inflammation-related genes at skin wound sites on Days 3, 7, and 14 post wounding. W wound site, I intact skin, M muscle. H&E staining images H&E staining images in Figs. 3B and  4B–D were obtained using single sets of Visium data (Appendix Fig. S4B,G,L). Source data are available online for this figure.

Figure 4. Cell location, cell–cell communication, and signaling analysis of spatial transcriptomics data with spatialomics deconvolution.

(A) Overview of the spatialomics deconvolution process. (B–D) Identification of cell location in spatial transcriptomics data with spatialomics deconvolution at skin wound sites on Days 3 (B), 7 (C), and 14 (D) post wounding. Expression of S100a9 (neutrophil marker), Lyz2 (macrophage marker), Dcn (fibroblast marker), Krt10 (epithelial cell marker), and Pecam1 (endothelial cell marker) from Visium. The ratio of cell types is predicted by spatial deconvolution. The color bars in each panel represent relative values, with the range varying from panel to panel. S scab, W wound site, I intact skin, M muscle. H&E staining images in Figs. 3B and 4B–D were obtained using single sets of Visium data (Appendix Fig. S4B,G,L). (E, F) The signaling directions of five major signaling pathways using COMMOT on Days 7 (E) and 14 (F) post wounding. Source data are available online for this figure.

Figure 5. Delayed skin wound healing and altered collagen fibril formation at scar sites of the Itgbl1 defect in mice.

(A) qPCR measurement of the temporal expression of murine Itgbl1 relative to that of GAPDH during skin wound healing (n = 2–4 wounds). P values: Intact skin vs. Day 7 = 0.0043, Intact skin vs. Day 10 = 0.0068, Intact skin vs. Day 7 = 0.0005. (B–D). ISH of Itgbl1 at murine wound sites on Day 7 post injury. The area indicated with a rectangle in (B) is shown at higher magnifications in (C, D). Many Itgbl1-positive cells can be observed in granulation tissue, but not in the epidermis, which corresponded to the result from Visium analysis. Black arrowheads indicate the wound margin. (D) Expression of Itgbl1 predominantly observed in wound-infiltrating spindle-shaped cells presumed to be fibroblasts (black arrowheads), but not in endothelial cells (black arrows) or macrophages (white arrowheads). Scale bars; (B) 500 μm, (C, D) 50 μm. (E) Schematic diagram of the generation of Itgbl1^−/− mice illustrating the Itgbl1 genomic construct containing a deletion of exon (Ex) 2. (F) qPCR analysis of Itgbl1 in wound sites confirming that Itgbl1^−/− mice were completely deficient in expression of Itgbl1 (WT and Itgbl1^−/−, n = 10). ND none detected. (G) Representative photographic images of the gross appearance of excisional wounds in WT (left) and Itgbl1^−/− (right) mice. (H) Proportion of the wound area remaining open at each time point relative to the initial wound area (WT; n = 11 wounds, Itgbl1^−/−; n = 12 wounds). P values: WT Day 7 vs. Itgbl1^−/− Day 7 < 0.0001. (I) H&E staining of wound site on Days 7 and 14 post injury (wound margin [arrowheads]). Scale bars: 500 μm. (J) Measurement of granulation tissue area on Days 7 and 14 post injury in WT (Day 7; n = 11), Day 14; n = 7) and Itgbl1^−/− mice (Day 7; n = 12, Day 14; n = 7). P values: Day 7 = 0.0048, Day 14 = 0.0391. (K) Polarized light microscopy differential interference contrast images of picrosirius red-stained sections of excisional wound sites on Day 14 post injury for analysis of collagen fibers and alignment. Arrowheads indicate the wound edge. Images are representative of eight independent experiments. Low-magnification images (upper) and high magnification of boxed areas (lower) are shown. Scale bars; (top) 200 μm, (bottom) 100 μm. (L) TEM images of collagen fibrils in connective tissue from mid-wound sites on Day 14 post injury (left). Histogram of the total range of fibril diameters in the wound site on Day 14 post injury (right); n = 863 fibrils from four WT mice; n = 1092 fibrils from four Itgbl1^−/− mice. Scale bars: 100 nm. (M) Hydroxyproline content at skin wound sites on Day 14 post injury in WT and Itgbl1^−/− mice (n = 12 wounds). P value: WT vs. Itgbl1^−/− = 0.0069. Data information: All values represent the mean ± SD. One-way ANOVA followed by Dunnett’s multiple comparisons test (intact skin vs. wound samples) (A), two-way ANOVA followed by Sidak’s multiple comparisons test (H), and unpaired t-tests (J, M) were used to generate the indicated P values; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 6. Regulation of Itgbl1 expression, myofibroblast differentiation, and fibrogenesis through antagonism between TGFβ1 and IL1β signaling.

(A) qPCR measurement of the temporal expression of ITGBL1 in HDF stimulated with TGFβ1 (100 pg/mL) (n = 4 independent cell cultures). P values: control (C) vs. 1 day (d) = 0.0121, C vs. 2 d = 0.0015, C vs. 3 d < 0.0001. (B) qPCR measurement of expression of ITGBL1 in HDF normal and keloid-derived fibroblasts (n = 4 independent cell cultures). P value: Control vs. Keloid = 0.0011. (C) qPCR of Itgbl1, Col1a1, and Acta2 expression in MDF stimulated with IL1β for 1 day (n = 3 independent cell cultures). P value (Col1a1/GAPDH): C vs. 100 pg/mL = 0.0032. P values (Acta2/GAPDH): C vs. 1 pg/mL = 0.0270, C vs. 10 pg/mL = 0.0186, C vs. 100 pg/mL = 0.0082. (D) qPCR of Itgbl1, Acta2, Col1a1, and Col3a1 expression in MDF stimulated with TGFβ1 (100 pg/mL) and IL1β (100 pg/mL) for 1 day (n = 6 independent cell cultures). P values: Itgbl1/GAPDH <0.0001, Itgbl1/GAPDH = 0.0064, Col1a1/GAPDH = 0.0002, Col3a1/GAPDH = 0.0001. (E) qPCR of ITGBL1, COL1A1, and ACTA2 expression in HDF stimulated with IL1β for 1 day (n = 3 independent cell cultures). P values (ITGBL1/GAPDH): C vs. 0.01 pg/mL = 0.0003, C vs. 0.1 pg/mL <0.0001, C vs. 1 pg/mL = 0.0001, C vs. 10 pg/mL = 0.0036, C vs. 100 pg/mL = 0.0363. P values (ACTA2/GAPDH): C vs. 0.01–100 pg/mL: <0.0001. (F) qPCR analysis of ITGBL1, ACTA2, COL1A1, and COL3A1 expression in HDF stimulated with TGFβ1 (100 pg/mL) and IL1β (100 pg/mL) for 1 day (n = 6 independent cell cultures). P values: ITGBL1/GAPDH <0.0001, ACTA2/GAPDH <0.0001, COL1A1/GAPDH <0.0001, COL3A1/GAPDH <0.0001. (G) A model summarizing the interplay of signaling factors within dermal fibroblasts. Data information: All values represent the mean. Unpaired t-test (B, D, F) and one-way ANOVA followed by Dunnett’s multiple comparisons test (control vs. sample) (A, C, E) were used to generate the indicated P values; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. C control. Source data are available online for this figure.

Figure 7. Indispensable nature of Itgbl1 in the formation of granulation tissue and support of fibrogenesis through myofibroblast differentiation-mediated antagonism of TGFβ1 and IL1β signaling.

(A) Schematic diagram of the genomic construction of Itgbl1^Tom/Tom mice. (B) Localization of Itgbl1-positive cells (red) at intact skin and wound sites in Itgbl1^Tom/Tom mice. Ep epithelial cells. Scale bars, 50 μm. (C) IHC of ACTA2 at the skin wound site on Day 7 post injury in Itgbl1^Tom/Tom mice. The white and empty arrowheads indicate ACTA2 and Itgbl1 double-positive cells and Itgbl1 single-positive cells, respectively. Scale bars; (C) 100 μm, (C–I) 50 μm. (D) Localization of macrophages (green) and Itgbl1-positive cells (red) at the skin wound site of Itgbl1^Tom/Tom::LysM^EGFP/EGFP mice on Day 7 post injury. Scale bar, 50 μm. (E) Schematic diagram of the genomic construction of Itgbl1^{CreERT2/CreERT2}::ROSA26^GRR/GRR mice. (F–H). Localization of Itgbl1-positive cells (red) on intact skin (F), and at Days 7 (G) and 14 (H) post injury. Arrowheads indicate the wound margin. Scale bars: (H) 500 μm; (F) 100 μm; (G-I, G-II, H-I) 50 μm. (I) Schematic diagram of the genomic construction of Itgbl1^flox/flox::Tagln^Cre/Cre mice. (J) qPCR analysis of Itgbl1 cKO in wound sites on Day 7 post injury confirming that Itgbl1 cKO mice were significantly decreased in an expression of Itgbl1 (n = 8). P value: Control vs. Itgbl1 cKO <0.0001. (K) Representative photographic images of the gross appearance of excisional wounds in control (Itgbl1^flox/flox) and Itgbl1 conditional knockout (cKO, Itgbl1^flox/flox::Tagln^Cre/Cre) mice. (L) The proportion of the wound area remaining open at each time point relative to the initial wound area (Control, n = 8 wounds; Itgbl1 cKO, n = 16 wounds). P value: Day 0 vs. Day 7 <0.0001. (M) H&E staining of wound site on Days 7 and 14 post injury (wound margin [arrowheads]). Scale bars: 500 μm. (N) Measurement of granulation tissue area on Days 7 and 14 post injury in control (Day 7; n = 6, Day 14; n = 7) and Itgbl1 cKO mice (Day 7; n = 7, Day 14; n = 8). P values: Day 7 = 0.0094, Day 14 = 0.0100. (O) IL1β signaling, from the proliferation phase to the maturation phase, is involved in the transformation of activated myofibroblasts and contributes to normal mature scar formation. Data information: All values represent the mean ± SD (L). Two-way ANOVA followed by Sidak’s multiple comparisons test (L) and unpaired t-tests (J, N) were used to generate the indicated P values; **P < 0.01, ****P < 0.0001.

Figure EV1. Identification of top ten marker genes in the macrophage subclusters.

Bubble heatmap of macrophage subclusters (M1–M6) and their respective marker genes.

Figure EV2. Identification of top 10 marker genes in the fibroblast subclusters.

(A) Bubble heatmap of fibroblast subclusters (F1–F8) and their respective marker genes. (B) UMAP plot (left) of expression levels of Acta2 (top), Tagln (middle), and Spp1 (bottom) across 40,024 cells from Days 3, 7, and 14 post injury (left panel). Violin plot (right) showing expression levels of these genes in each fibroblast subcluster. F1: n = 4265; F2: n = 3849; F3: n = 2218; F4: n = 1421; F5: n = 1369; F6: n = 977; F7: n = 804; F8: n = 532.

Figure EV3. Inference with the cell–cell communication network on Day 3 post injury.

(A) Circle plot of signaling pathways. (B) Heatmap analysis of the roles of the representative signaling pathways in the aggregated cell–cell communication network. (C) Alluvial plot of the outgoing signaling patterns of secreting cells, demonstrating the correspondence between the inferred latent pattern and cell groups, as well as the signaling pathways. The thickness of the flow indicates the contribution of the cell group or signaling pathway to each latent pattern. The height of each pattern is proportional to the number of its associated cell groups or signaling pathways. Outgoing patterns reveal how the sender cells coordinate with each other, as well as how they coordinate with certain signaling pathways to drive communications. (D) Bobble plot of the outgoing signaling patterns of secreting cells. (E) Incoming signaling patterns of target cells, showing how the target cells coordinate with each other and with certain signaling pathways to respond to incoming signals. (F) Bobble plot of the incoming signaling patterns of target cells.

Figure EV4. Inference with the cell–cell communication network on Day 7 post injury.

(A) Circle plot of signaling pathways. (B) Heatmap analysis of the roles of the representative signaling pathways in the aggregated cell–cell communication network. (C) Alluvial plot of the outgoing signaling patterns of secreting cells, demonstrating the correspondence between the inferred latent pattern and cell groups, as well as the signaling pathways. The thickness of the flow indicates the contribution of the cell group or signaling pathway to each latent pattern. The height of each pattern is proportional to the number of its associated cell groups or signaling pathways. Outgoing patterns reveal how the sender cells coordinate with each other, as well as how they coordinate with certain signaling pathways to drive communications. (D) Bobble plot of the outgoing signaling patterns of secreting cells. (E) Incoming signaling patterns of target cells, showing how the target cells coordinate with each other and with certain signaling pathways to respond to incoming signals. (F) Bobble plot of the incoming signaling patterns of target cells.

Figure EV5. Inference with the cell–cell communication network on Day 14 post injury.

(A) Circle plot of signaling pathways. (B) Heatmap analysis of the roles of the representative signaling pathways in the aggregated cell–cell communication network. (C) Alluvial plot of the outgoing signaling patterns of secreting cells, demonstrating the correspondence between the inferred latent pattern and cell groups, as well as the signaling pathways. The thickness of the flow indicates the contribution of the cell group or signaling pathway to each latent pattern. The height of each pattern is proportional to the number of its associated cell groups or signaling pathways. Outgoing patterns reveal how the sender cells coordinate with each other, as well as how they coordinate with certain signaling pathways to drive communications. (D) Bobble plot of the outgoing signaling patterns of secreting cells. (E) Incoming signaling patterns of target cells, showing how the target cells coordinate with each other and with certain signaling pathways to respond to incoming signals. (F) Bobble plot of the incoming signaling patterns of target cells.