Myc-dependent dedifferentiation of Gata6+ epidermal cells resembles reversal of terminal differentiation

Abstract

Dedifferentiation is the process by which terminally differentiated cells acquire the properties of stem cells. During mouse skin wound healing, the differentiated Gata6-lineage positive cells of the sebaceous duct are able to dedifferentiate. Here we have integrated lineage tracing and single-cell mRNA sequencing to uncover the underlying mechanism. Gata6-lineage positive and negative epidermal stem cells in wounds are transcriptionally indistinguishable. Furthermore, in contrast to reprogramming of induced pluripotent stem cells, the same genes are expressed in the epidermal dedifferentiation and differentiation trajectories, indicating that dedifferentiation does not involve adoption of a new cell state. We demonstrate that dedifferentiation is not only induced by wounding, but also by retinoic acid treatment or mechanical expansion of the epidermis. In all three cases, dedifferentiation is dependent on the master transcription factor c-Myc. Mechanotransduction and actin-cytoskeleton remodelling are key features of dedifferentiation. Our study elucidates the molecular basis of epidermal dedifferentiation, which may be generally applicable to adult tissues.

Fig. 1. Gata6lin⁺ cells and lin⁻ cells in wounds are indistinguishable on the basis of their transcriptomes.

a, Schematic of the experimental design (top) and representative flow cytometric plots (bottom) showing Itga6 levels in Gata6lin⁺ cells at the indicated timepoints after wounding. Light-blue lines represent gating strategy used for quantification in Extended Data Fig. 1e. b, Representative section of day 9 wounded skin showing tdTomato Gata6lin⁺ cells stained for Itga6 (green). Nuclei are visualized with DAPI staining (blue). Boxed regions are shown at higher magnification. White arrows indicate lin⁺ cells attached to the basement membrane. Scale bars, 200 µm (overview) and 40 µm (magnifications). n = 4 independent experiments. c, Left: schematic of location of epidermal populations in undamaged skin. Right: transcriptomes of control Gata6lin⁺ and lin⁻ epidermal cells were projected onto tSNE space re-analyzed from Joost et al. (2016). Note that Gata6lin⁺ cells mapped primarily to the uHF compartment. d, Gata6lin⁺ and lin⁻ cells isolated from wounds were visualized on a UMAP plot. Note that Gata6lin⁺ and lin⁻ cells were present in all six clusters. e, Detection of Cst6 and Defb6 by mRNA in situ hybridization in skin sections showing tdTomato Gata6lin⁺ cells 9 days after wounding. HFs proximal to wounds and healing IFE are shown. Boxed regions are shown at higher magnification. White arrows indicate lin⁺ cells expressing Cst6 (top) or Defb6 (bottom). Scale bars, 20 µm. Representative images from n = 3 independent experiments.

Fig. 2. Dedifferentiation occurs via reversal of the normal differentiation process.

a, Gata6lin⁺ cells were visualized on a tSNE plot. Left: Gata6lin⁺ cells mapped into three clusters, colour-coded according to unsupervised clustering. Middle: the indicated timepoints after wounding are shown. Right: pseudotime trajectory. b, Heat maps showing the correlation between the three clusters identified with Monocle and cell compartments defined in the Joost dataset. c, CellRank was used to visualize future states (arrows) on the generated tSNE plot in a. d, Heat maps showing the correlation between the three clusters and basal and suprabasal epidermal compartments defined in Human Protein Atlas version 20.0. e, Expression of the wound markers Krt6a and Krt6b, and the SD marker Gata6. f, Expression of the upper HF markers Defb6, Cst6, basal marker Itga6 and stem cell markers Lrig1 and Lgr5 is shown on the tSNE plot. Data in a–f are from two independent biological replicates per timepoint. All Gata6lin⁺ cells from the scRNA-seq data were analyzed. g, Left: schematic of the normal differentiation process (Lrig1 to Gata6, control lin⁻ cells) and the dedifferentiation process (Gata6 to Lrig1, wound lin⁺ cells). Right: transcriptional profile correlation between Gata6lin⁺ cells dedifferentiating to Lrig1⁺ stem cells (wounded skin) and Lrig1 lin⁻ cells differentiating to Gata6⁺ cells (unwounded skin). Loess regression was used for fitting. Differentially expressed genes between the two populations are represented by dots (545 genes from 112 cells, 32 differentiating and 80 dedifferentiating cells). The solid line represents the Loess regression fit. The grey area indicates the error bands. Two-sided Wilcoxon rank sum test was used to identify differentially expressed genes. Adjustments were made for multiple comparisons. h, Volcano plot showing differentially expressed genes identified in g. Genes with a fold change greater than 1.6 are shown in red.

Fig. 3. Evidence for upregulation of Myc during dedifferentiation.

a, Heat map showing expression of TFs obtained from the TRRUST database after analysis of each gene module. TFs were selected on the basis of q value <0.5 and Maron value >0.9. Gata6lin⁺ cells distributed into the different clusters obtained in Fig. 2a are represented along the horizontal axis, and gene modules are shown along the vertical axis. b, Expression of Lrig1 and Myc on the pseudotime trajectory. c, Scatter plots showing expression of Ki67 and Myc, and Lrig1 and Myc in control and wound cells. Spearman r coefficient is shown on the plots. Note that wound cells show a higher correlation between Myc and Ki67, and Myc and Lrig1 compared with control cells. d, Keratinocytes from K14MycER transgenic mice were transfected with a doxycycline (Dox)-inducible GATA6 construct or with mock lentiviruses and seeded at clonal density. Representative images of dishes showing colony formation 12 days after plating are shown. GATA6 or Myc expression was induced by addition of 2 μg ml⁻¹ Dox or 25 nM 4-OHT, respectively. One day after GATA6 induction, Myc expression was induced by adding 25 nM 4-OHT. Cells transfected with mock lentiviruses were pre-treated with or without ethanol (VC, vehicle control). e,f, Bar graphs showing number of colonies in each of the indicated conditions (e) and expression of Ivl measured by RT–qPCR (f). Data are the mean ± s.d. from three to four independent experiments. One-way ANOVA with Šidák’s multiple comparisons test was used to determine statistical significance. Source data

Fig. 4. Myc is required for wound-induced dedifferentiation of Gata6lin⁺ cells.

a, Representative sections of HFs distal and proximal to the wound site showing tdTomato Gata6lin⁺ cells stained for Lrig1 (grey) and Myc (green). Boxed regions show separation of channels. White and yellow arrows indicate Myc-expressing cells from the uHF and the IFE, respectively. Scale bar, 40 µm. n = 7 independent experiments. b,c, Bar graphs showing the percentage of Gata6lin⁺ cells expressing Lrig1 per HF (b) or Myc nuclear intensity (c). Data are the mean ± s.d. n = 7 mice (b); n = 6 mice (c). d, Skin of Myc^+/+ and Myc^−/− mice showing tdTomato Gata6lin⁺ cells expressing Myc (green) and Lrig1 (grey). Arrows indicate Gata6lin⁺ cells expressing Myc. Note that Gata6lin⁺ cells in Myc^−/− mice do not express Myc. Scale bar, 40 μm. Representative images from n = 3 independent experiments. e,f, Bar graphs showing the percentage of Gata6lin⁺ cells expressing Lrig1 per HF adjacent to wound (e) and per 0.01 mm² of healing IFE (f) in Myc^+/+ and Myc^−/− mice. Data are the mean ± s.d. n = 3 mice per group. g,h, Bar graphs showing fluorescence intensity of nuclear Myc in Gata6lin⁺ cells (g) and lin⁻ Lrig1⁺ cells present in HFs (h). Data are the mean ± s.d. n = 3 mice per group. i, Detection of Lrig1 by mRNA in situ hybridization in skin sections of Myc^+/+ and Myc^−/− mice showing tdTomato⁺ Gata6lin⁺ cells. Scale bars, 200 µm (overview) and 40 µm (magnification). Representative images from n = 3 independent experiments. j, Quantification of the number of Lrig1 RNA molecules per HF. Data are the mean ± s.d. n = 3 mice per group. Two-tailed Student’s unpaired t-test was used to determine statistical significance in b, c, e–h and j. Source data

Fig. 5. Myc-dependent hydrogel-induced dedifferentiation.

a, Skin sections of control and hydrogel-injected mice stained for YAP (top) and MAL (bottom). Scale bar, 20 µm. b, Percentage of cells showing nuclear YAP or MAL (N > C), even distribution of YAP or MAL (N = C) and cytoplasmic YAP or MAL (N < C). Data are the mean ± s.d. n = 3 mice. c, Epidermal sections of control and hydrogel-injected mice stained for Ivl and Krt14. Scale bar, 10 µm. d, Bar graph showing IFE thickness (µm) of control and hydrogel-injected mice. Data are the mean ± s.d. n = 4 mice. e, Representative skin section showing position of the injected hydrogel (dashed line). Gata6lin⁺ cells and DAPI nuclear labelling are shown. Scale bar, 500 µm. Boxed region shows magnified HF stained for Lrig1. Scale bar, 40 µm. n = 3 independent experiments. f–h, Bar graphs showing the percentage of suprabasal and basal Gata6lin⁺ cells (f), the percentage of Gata6lin⁺ cells in IFE (g) and the percentage of Gata6lin⁺ cells expressing Lrig1 per HF (h) in control, hydrogel-injected and bleomycin-treated mice. Data are the mean ± s.d. n = 3 mice. i, Skin sections showing tdTomato Gata6lin⁺ cells stained for CD45. Scale bar, 40 µm. j, Bar graph showing CD45 fluorescence intensity (a.u.) in hydrogel and bleomycin-treated mice. Data are the mean ± s.d. n = 3 mice (day 2, day 7, bleomycin); n = 4 mice (control, day 1, day 4, PBS). k, Skin sections of Myc^+/+ and Myc^−/− hydrogel-injected mice showing tdTomato Gata6lin⁺ cells stained for Lrig1. Scale bar, 40 µm. l,m, Percentage of Gata6lin⁺ cells in the IFE (l) and Gata6lin⁺ cells expressing Lrig1 per HF (m). Data are the mean ± s.d. n = 3 mice. n, Skin sections of Myc^+/+ and Myc^−/− hydrogel-injected mice showing tdTomato Gata6lin⁺ cells stained for YAP. Scale bar, 20 µm. o, Percentage of cells showing nuclear YAP (N > C), even distribution of YAP (N = C) and cytoplasmic YAP (N < C). Data are the mean ± s.d. n = 3 mice. Two-tailed Student’s unpaired t-test was used to determine statistical significance in b, l, m and o. One-way ANOVA with Šidák’s multiple comparisons test was used to determine statistical significance in d, g, h and j. Source data

Fig. 6. Effect of Myc deletion on cell stiffness and the actin cytoskeleton.

a, Representative brightfield images and Young’s modulus maps of HFs distal to wounds in Myc^+/+ and Myc^−/− mice. Boxed regions show the upper HF regions that were measured by AFM. Each square corresponds to the Young’s modulus obtained from one force-distance curve. The dashed lines mark hair shafts. The arrows indicate the tip of the cantilever. Representative images from n = 3 independent experiments. b, Epidermal cell stiffness was analyzed in HFs by AFM. Violin plots show Young’s modulus (kPa) in distal and proximal HFs of Myc^+/+ and Myc^−/− mice (deleted via the Gata6 promoter). A total of 150 measurements were analyzed per region. At least 18 HFs (9 distal and 9 proximal) from 3–4 mice per group were analyzed. c, Schematic showing Young’s modulus means (kPa) obtained in b. Note that in the absence of Myc there is no difference in stiffness between distal and proximal HFs. d, Schematic of single-cell AFM analysis and violin plots showing Young’s modulus (kPa) measurements of Myc^+/+ and Myc^−/− keratinocytes. A total of 29 Myc^+/+ cells and 27 Myc^−/− cells from three independent experiments were analyzed. e, Cells stained with phalloidin (red), anti-Myc (green) and DAPI (blue). Scale bar, 10 µm. f,g, Bar graphs show cell area (µm²) (f) and fluorescence intensity of F-actin and nuclear Myc (g) in Myc^+/+ and Myc^−/− keratinocytes. n = 12 cells per condition from 2 independent experiments. h, Detection of Cst6 by mRNA in situ hybridization in HFs of Myc^+/+ and Myc^−/− mice showing tdTomato Gata6lin⁺ cells. Arrows indicate cells in the basal layer expressing Cst6. Scale bar, 20 µm. i, Quantification of the number of cells per HF expressing Cst6 in the basal layer. Data are the mean ± s.d. n = 3 mice per group. Two-tailed Mann–Whitney test was used to determine statistical significance in b. Two-tailed Student’s unpaired t-test was used to determine statistical significance in d, f, g and i. Source data

Fig. 7. Myc-dependent actin network remodelling and cell contractility.

a, Top: skin sections of HFs distal and proximal to wounds showing tdTomato Gata6lin⁺ cells stained with phalloidin (green) and DAPI (blue). Bottom: masks were obtained using the ImageJ plugin TWOMBLI. Scale bar, 20 µm. Representative images from n = 4 independent experiments. b,c, Bar graphs show total length (µm) (b) and lacunarity (a.u.) (c) obtained in distal and proximal HFs of Myc^+/+ mice and proximal HFs of Myc^−/− mice. Data are the mean ± s.d. n = 4 mice per condition. d–g, Wounded skin sections (d) and skin sections of hydrogel-injected mice (f) showing tdTomato Gata6lin⁺ cells were stained for pMLC2 and colour-coded for signal intensity with ImageJ. Bar graphs show fluorescence intensity of pMLC2 in wounded (e) and hydrogel-injected (g) epidermis of Myc^+/+ and Myc^−/− mice. The hair shaft was excluded from the analysis and only areas containing tdTomato⁺ cells were measured. Data are the mean ± s.d. n = 3 mice per condition. Scale bars, 20 µm. h,i, Ridgeline plots showing expression of the small GTPases RhoA, Rac1 and Cdc42 (h), and the adhesion markers Lad1, Itgb1 and Tln1 (i) at days 0, 6, 9 and 11 post-wounding. Data are from two independent biological replicates per timepoint. All Gata6lin⁺ cells from the scRNA-seq data were analyzed. Two-tailed Student’s unpaired t-test was used to determine statistical significance in b, c, e and g. Source data

Extended Data Fig. 1. Flow sorting strategy.

(a) Representative section of wounded skin 9 days after wounding with a 6 mm biopsy punch. DAPI was used to visualize nuclei. Scale bar, 4 mm. n = 4 independent experiments. (b) Representative sections of unwounded and wounded skin showing tdTomato Gata6lin+ cells stained for Itga6. Nuclei are visualized with DAPI staining. Boxed regions are shown at higher magnification. White arrows indicate lin+ cells attached to the basement membrane. Scale bars, 40 µm. Representative images from n = 4 independent experiments per timepoint. (c) Bar graphs showing the percentage of suprabasal and basal Gata6lin+ cells in unwounded (day 0) and wound cells (day 6 and day 9). Data are means ± s.d. n = 4 mice (d) Flow cytometry of epidermal cells from unwounded and wounded skin of Gata6EFGPCreERT2 Rosa26-fl/STOP/fl-tdTomato43 mice buffered with GFP+ cells from CAGGS eGFP mice. GFP⁻CD45⁻, tdTomato⁺Itg6⁻, tdTomato⁺Itg6⁺, tdTomato⁻Itg6⁻, and tdTomato⁻Itg6⁺ cells were analyzed. (e) Quantification of low/mid-Itga6 and high-Itga6 Gata6lin+ cells at the indicated timepoints after wounding. n = 2 independent experiments. (f) Bar graph showing Itga6 expression (a.u) at different timepoints after wounding. n = 2 independent experiments. (g) t-SNE plot showing batch 1 and batch 2 used for single-cell RNA analysis after batch correction (left). Ridgeline plots showing the number of mRNA features and percentage of mitochondrial genes sequenced from every cell of the scRNAseq dataset (right). Data are from two independent biological replicates per timepoint. (h) Schematic of location of epidermal populations in undamaged skin based on the second level of clustering introduced in Joost et al., 2016 (left panel). Heatmaps showing the correlation between the lin+ cells and cell compartments defined in the Joost dataset (right panel). (i) Heatmap showing the most highly enriched genes in each cluster in Fig. 1d. Clusters were colour-coded along the horizontal axis. Source data

Extended Data Fig. 2. Trajectory of Gata6lin+ cells from uHF-like to IFE-like states.

(a) Expression of the upper HF markers Krt79 and Krt17, IFE markers Mt2 and Krt14, OB markers Cd34 and Postn, SG markers Mgst1 and Scd1, and IFE-D markers Ptgs1 and Krt10 is shown along the pseudotime trajectory. (b, c) Ridgeline plots showing the expression of the wound markers Krt6a and Krt6b (b), the uHF markers Defb6 and Cst6 and the basal layer markers Itga6, Krt14, Lrig1 and Lgr5 (c) at days 0, 6, 9, and 11 post-wounding. Data in (a), (b) and (c) are from two independent biological replicates per timepoint. All Gata6lin+ cells from the scRNAseq data were analyzed. (d) Sections of HFs proximal to the wound site showing tdTomato Gata6lin+ cells stained for Lrig1 (green) and Ki67 (grey) at days 1, 2, and 3 after wounding. 2 mm wounds were made to assess the early timepoints of wound healing. Scale bar, 40 µm. (e) Bar graph shows the number of Gata6lin+ cells expressing Lrig1 (light grey bars), Ki67 (dark grey bars), and the total number of lin+ cells (red bars) in HFs proximal to a wound and ctrl HFs. Data are means ± s.d. n = 3 mice per group. Two-tailed Student’s unpaired t-test was used to determine statistical significance. Source data

Extended Data Fig. 3. Pseudotransition-dependent genes along the state trajectory.

(a) Gene modules defined in pseudotime analysis visualized on a heatmap. (b) Heatmap showing smoothed expression of pseudotransition-dependent genes ordered by hierarchical clustering and maximum expression. Genes (rows) are ordered by peak expression from cluster 3 to cluster 1. (c) Ligand-receptor analysis. In the wound Gata6lin- cells were assumed to be signal sender cells and Gata6lin+ cells receiving cells. Prioritized ligands expressed by sender cells are shown on the vertical axis and predicted target genes on the horizontal axis. (d) Ligands expressed by lin- cells are shown on the vertical axis and predicted receptors expressed by Gata6lin+ cells on the horizontal axis. (e) Bar graphs showing colony area in each of the indicated conditions. Data are means ± s.d. n = 3 independent experiments (f) Bar graphs show gene expression levels of Dox-inducible human GATA6 (left) and endogenous mouse Gata6 (right) after 2 μg/ml Dox treatment. n = 2 independent experiments. (g) Keratinocytes from K14MycER mice transfected with GATA6 lentivirus were treated with 4-OHT and stained with DAPI and anti-Myc (red). Scale bar, 40 µm (h) Bar graph shows fluorescence intensity of nuclear Myc in cells ± 25 nM 4-OHT. n = 23 cells (-) and n = 21 cells (+) from 3 independent experiments. One-way ANOVA with Šidák’s multiple comparisons test was used to determine statistical significance in (e). Two-tailed Student’s unpaired t-test was used to determine statistical significance in (h). Source data

Extended Data Fig. 4. Effects of Myc depletion in Gata6lin+ cells.

(a) Top view of healing epidermis (top panels). Top, middle and bottom planes are shown. Dashed lines mark proximal and distal HFs to the wound. Boxed region indicates the HF shown in the stills (bottom panels). Stills from time-lapse in vivo recording showing expansion of tdTomato Gata6lin+ cells (red) in the HF and subsequent migration (arrows) into the IFE of a healing wildtype wound (to left of HF). Nuclei (Hoechst) and collagen are visualized in green. Scale bar, 100 µm. (b) Bar graph shows the relative area occupied by tdTomato Gata6lin+ cells in the uHF at pre-migration and post-migration timepoints. Data are means ± s.d. n = 3 mice. (c) Bar graphs showing fluorescence intensity of nuclear Myc in lin- cells present in IFE and bulge. Data are means ± s.d. n = 3 mice. (d) Skin of Myc + /+ and Myc - /- mice showing tdTomato Gata6lin+ cells stained for Ki67 and Lrig1. Scale bar, 40 µm. (e) Bar graph shows the percentage of tdTomato Gata6lin+ and lin- Lrig1+ cells expressing Ki67 per HF adjacent to a wound. Data are means ± s.d. n = 3 mice (f) Representative H&E staining of day 9 wounded epidermis. n = 4 independent experiments. (g) Bar graph showing epidermal thickness in uHFs and IFE after wounding. Data are means ± s.d. n = 3 mice (ctrl uHF, Myc + /+ wound uHF, ctrl IFE, Myc -/- wound IFE); n = 4 mice (Myc-/- wound uHF, Myc + /+ wound IFE). (h) Detection of Lgr5 and Lgr6 by mRNA in situ hybridisation in skin sections of Myc + /+ and Myc -/- mice showing tdTomato Gata6lin+ cells. Scale bars, 200 µm for the overview and 40 µm for the magnification. (i) Quantification of the number of Lgr5 and Lgr6 RNA molecules per HF. Data are means ± s.d. n = 3 mice. Two-tailed Student’s unpaired t-test was used to determine statistical significance in (b), (c), (e), and (i). One-way ANOVA with Šidák’s multiple comparisons test was used to determine statistical significance in (g). Source data

Extended Data Fig. 5. RA treatment induces dedifferentiation.

(a) Skin was treated with acetone or RA and skin sections were stained for Lrig1 with DAPI counterstain. Scale bar, 40 µm. (b) Bar graph shows epidermal thickness upon RA treatment. Data are the mean ± s.d. n = 4 mice. (c, d) Bar graphs showing the number of Gata6lin+ cells expressing Lrig1 per HF (c) and Gata6lin+ cells in the IFE (d). Data are the mean ± s.d. n = 4 mice. (e) Representative H&E staining of skin of hydrogel-injected mice. n = 4 independent experiments. (f) Skin sections showing tdTomato Gata6lin+ cells stained for Ki67. Representative images from n = 2 independent experiments. (g) Bar graph showing the percentage of epidermal Ki67+ cells per HF at the indicated timepoints after hydrogel injection. n = 2 mice. (h) Bar graph showing epidermal thickness (µm) in the uHF. Data are the mean ± s.d. n = 4 mice (ctrl, day 1, day 7), n = 3 mice (day 2 and day 4). (i) Bar graph showing expression of the inflammation markers Tnfa and Ccl2 measured by RT-qPCR. Data are the mean ± s.d. n = 3 mice. (j) Bar graph showing epidermal thickness (µm) in control and bleomycin-treated mice. Data are the mean ± s.d. n = 3 mice. (k) Bar graph shows epidermal thickness upon hydrogel injection. Data are the mean ± s.d. n = 5 mice (ctrl uHF, Myc + /+ expanded uHF); n = 6 mice (Myc-/- expanded uHF); n = 4 mice (ctrl IFE); n = 3 mice (Myc + /+ and Myc -/- expanded IFE). (l) Skin sections of Myc + /+ and Myc -/- hydrogel-injected mice showing tdTomato Gata6lin+ cells stained for MAL. Scale bar, 20 µm. (m) Bar graphs showing the percentage of cells showing nuclear MAL (N > C), even distribution of MAL in nucleus and cytoplasm (N = C), and cytoplasmic MAL (N < C). Data are the mean ± s.d. n = 3 mice. (n) Skin of Myc + /+ or Myc -/- hydrogel-injected mice showing tdTomato Gata6lin+ cells stained for Ki67. Scale bar, 20 µm. (o) Bar graph shows the number of tdTomato Gata6lin+cells expressing Ki67 per HF in the indicated conditions. Data are the mean ± s.d. n = 4 mice. One-way ANOVA with Šidák’s multiple comparisons test was used to determine statistical significance in (b), (h), and (k). Two-tailed Student’s unpaired t-test was used to determine statistical significance in (c), (d), (i), (j), (m), and (o). Source data

Extended Data Fig. 6. F-Actin remodelling is linked to dedifferentiation.

(a) Detection of Krt10 in HFs of Myc + /+ and Myc-/- mice showing tdTomato Gata6lin+ cells. Arrows indicate cells in the basal layer expressing Krt10. Scale bar, 20 µm. Representative images from n = 3 independent experiments. (b) Quantification of the number of cells per HF expressing Krt10 in the basal layer. Data are the mean ± s.d. n = 3 mice per group. (c) Skin sections of RA-treated Myc + /+ and Myc -/- mice showing tdTomato Gata6lin+ cells stained with phalloidin and DAPI (left panels). Masks were obtained using the ImageJ plugin TWOMBLI (right panels). Scale bar, 40 µm. Representative images from n = 4 independent experiments. (d, e) Bar graphs show total length (µm) (d) and lacunarity (a.u) (e) in RA-treated epidermis using TWOMBLI. Data are the mean ± s.d. n = 4 mice per group. (f) Skin sections of expanded skin of Myc + /+ and Myc -/- mice showing tdTomato Gata6lin+ cells were stained with phalloidin and DAPI (left panels). Masks were obtained using TWOMBLI (right panels). Scale bar, 20 µm. Representative images from n = 4 independent experiments. (g, h) Bar graphs show total length (µm) (g) and lacunarity (a.u) (h) in mechanically expanded epidermis using TWOMBLI. Data are the mean ± s.d. n = 4 mice per group. (i) Ridgeline plots showing the expression of the cytoskeleton regulators Arpc4, Cnbp, Myh9, Tmsb10, Twf1, and Pdlim5. Data are from two independent biological replicates per timepoint. All Gata6lin+ cells from the scRNAseq data were analyzed. Two-tailed Student’s unpaired t-test was used to determine statistical significance in (b), (d), (e), (g), and (h). Source data