Positional Stability and Membrane Occupancy Define Skin Fibroblast Homeostasis In Vivo

Abstract

Fibroblasts are an essential cellular and structural component of our organs. Despite several advances, the critical behaviors that fibroblasts utilize to maintain their homeostasis in vivo have remained unclear. Here, by tracking the same skin fibroblasts in live mice, we show that fibroblast position is stable over time and that this stability is maintained despite the loss of neighboring fibroblasts. In contrast, fibroblast membranes are dynamic during homeostasis and extend to fill the space of lost neighboring fibroblasts in a Rac1-dependent manner. Positional stability is sustained during aging despite a progressive accumulation of gaps in fibroblast nuclei organization, while membrane occupancy continues to be maintained. This work defines positional stability and cell occupancy as key principles of skin fibroblast homeostasis in vivo, throughout the lifespan of mice, and identifies membrane extension in the absence of migration as the core cellular mechanism to carry out these principles.

Keywords: aging; cell behaviors; fibroblasts; intravital imaging; mesenchyme; skin homeostasis.

Figure 1.. Fibroblasts Are Stable in Unperturbed Tissue and following the Loss of Neighboring Fibroblasts

(A) Revisits of skin dermis. Column 3 is overlay of day 0 and +2 weeks. n = 3 mice. Quantification is shown in (B) (+2). Scale bar, 10 μm. (B) “Identical” (r = 1.0) is the coefficient of an image compared to itself. +2, coefficient of day 0 versus +2 weeks (as in A). Re., coefficient between two remounts of the same location (pos ctrl). Ra., random images at same depth (neg ctrl) (see STAR Methods). Positive correlation means a high overlap between signals. Zero correlation means a random overlap. Error bars are SD. n = 3 mice for 2-week data. n = 4 regions for remount. n = 3 regions for random. (C) Experimental approach and two possible outcomes. (D) Revisits of single-cell laser ablation (see STAR Methods). Damage is visible by photo-bleached square. Three individually ablated nuclei are shown (pink dots) and are absent by 1 day later. Scale bar, 5 μm. (E) Revisits within the upper dermal layer before and after laser ablation (pink dots). +1 day, the ablated cells are absent. +2 weeks, the remaining nuclei remain largely stable in position, the depleted region is not filled in, and the fibroblast cell number remains reduced. Quantification is shown in (H) (laser, upper). n = 3 mice. Scale bar, 20 μm. (F) Revisits within the lower dermal layer before and after laser ablation. +2 weeks, remaining nuclei remain largely stable in position, similar to equivalent depth in (A) and (B). Quantification is shown in (H) (laser, lower). n = 3 mice. Scale bar, 20 μm. (G) Revisits using genetically inducible ablation. Following low-dose tamoxifen, random cells (pink nuclei) express DTA and die. Before tam, upper dermis before tamoxifen. +2 weeks and +3 weeks, the same area at respective time point. Some recombination occurs without tamoxifen at a basal rate (see STAR Methods), and some cells continue to be lost between +2 weeks and +3 weeks (red only cells in merge). Quantification is shown in (H) (DTA, upper). n = 3 mice. Scale bar, 20 μm. (H) Image correlation coefficients of data in (E)–(G) between +1 day and +2 weeks (laser), or +2 weeks and +3 weeks (DTA), in the upper and lower dermis. Image correlation coefficients are similar to image correlation coefficients in unperturbed tissue (B). Error bars are SD. n = 3 mice for each bar.

Figure 2.. Fibroblast Membrane Dynamics Maintain High Occupancy of Dermal Space

(A) Revisits of individually membrane-GFP-labeled fibroblasts in live mouse paw dermis (grayscale) (see Figure S2A for non-grayscale). Images are sums of ~5 μm of Z stacks. n = 3 mice. XY scale bars, 20 μm. XZ scale bars, 5 μm. (B) Frames from time-lapse of membrane-GFP upper dermal fibroblast (full movie: Video S2). Minutes (′) within the time-lapse are written in bottom right of orange boxed insets. Scale bar, 20 μm. n = 3 movies of 3 mice. (C) Observations of (A) and (B). Primary projections (*) are stable over time, but secondary projections (arrowhead) are dynamic. (D) Image of fibroblast nuclei (top), nuclei + membrane (middle) and inset (bottom) in upper dermis. PDGFRα-H2BGFP labels all fibroblast nuclei. Membrane labeling is induced by CreER recombination of mTmG in 98.2% ±1.0% of cells (Figure S2B). Images are max projections of 10 μm. n = 3 mice. Quantification is shown in (G). Scale bar, 20 μm. Scale bar, 20 μm (inset). (E) Images of fibroblast nuclei (top), nuclei + membrane (middle) and inset (bottom) in Rac1^+/− (left) and Rac1^−/− (right) fibroblasts 3–4 months after high-dose tamoxifen. n = 3 mice for Rac1^+/− and n = 3 mice for Rac1^−/−. Quantification is shown in (G). Scale bar, 20 μm. Scale bar, 20 μm (insets). (F) Density of nuclei between Rac1^+/− and Rac1^−/−. Data are sampled from regions ~500 × 500 μm. Error bars are SD. n = 3 mice for each bar. (G) Membrane coverage between Rac1^+/− and Rac1^−/−, computed by thresholding GFP signal (see STAR Methods). p = 0.0101 between Rac1^+/− and Rac1^−/−. Error bars are SD. n = 3 mice for each bar.

Figure 3.. Fibroblast Membrane Extensions Provide Compensation Mechanism for Loss of Neighbors

(A) Experimental approach to test cell membrane response to cell loss. (B) Laser ablation and revisits of high-dose tamoxifen membrane labeling. Laser ablated nuclei (pink dots, before ablation). +2 weeks, ablated area remains depleted of nuclei, but filled with membrane. Membrane occupancy before versus +2 weeks is similar (insets 1 and 2). Quantification is shown in (C). Scale bar, 20 μm. Inset scale bars, 10 μm. (C) Ratio of area of green signal (membrane-GFP + nuclear (H2B) GFP) at +2 weeks to before ablation (93%). Error bars are SE. n = 3 mice. (D) Short revisits following ablation of unlabeled membrane cells near to membrane-GFP-labeled cells. Arrowheads indicate membrane extensions into the cell-depleted space. (E) Laser ablation of membrane-unlabeled cells (pink dots) and revisits of labeled neighbors. Membranes from neighboring cells extend into the ablated region up until +5 days. Membranes remain present at +3 weeks. n = 3 mice. Scale bar, 20 μm. (F) Revisits of Rac1^+/− (top) and Rac1^−/− (bottom) following laser ablation at 2 weeks after high-dose tamoxifen (maximal recombination of both Rac1 and mTmG). Rac1^+/− reoccupies the ablated region similar to wild-type (E), while Rac1^−/− fails to do so. Insets (far right column) show details of the region highlighted at +3 weeks. n = 3 mice for Rac1^+/− and n = 3 mice for Rac1^−/−. Quantification is shown in (G). Scale bar, 20 μm. Scale bar, 20 μm (insets). (G) Membrane occupancy before and after ablation of a 120 × 120-μm region centered on the ablations in Rac1^+/− or Rac1^−/− fibroblast mice. n = 3 mice for each column. p = 0.0049. Error bars are SD.

Figure 4.. Fibroblasts Incorporate Membrane Extension with Proliferation and Migration following Larger Damage

(A) Nuclei and surrounding collagen is laser ablated (pink nuclei + 3 lightning bolts), causing proliferation and migration behaviors, and concurrent membrane extensions are scored. (B) Fibroblast elimination + collagen-damaging ablations (dashed circles and second harmonic signal at top) and revisits. Middle and bottom rows are both green channel, but brightness is adjusted to show stronger nuclear GFP, or fainter membrane-GFP. Quantification is shown in (C). n = 15 ablations in 4 mice. Scale bar, 10 μm. Arrowheads, proliferation or migration. Arrows, membrane extension. (C) Proliferation and/or migration (8 of 15), and membrane extensions (15 of 15) following collagen-damaging ablations (Abl.) or non-collagen-damaging controls (No Abl.). n = 15 ablations and 4 non-collagen-damaging controls across 4 mice. (D) Revisits following widespread genetic ablation of fibroblasts (DTA). Near complete loss of the upper fibroblasts occurs by +1 week following high dose of tamoxifen. By +2 weeks, the upper dermal layer partially recovers in cell number, but sub-epithelial cell number does not recover even at +6 weeks. Quantification is shown in (F). n = 3 mice. Scale bar, 20 μm. Epi, epidermis. (E) XZ projections in high DTA with revisits highlighting fibroblast membrane. Before ablation (left), cell bodies are visible in the sub-epithelial region. After high-dose tamoxifen (center and right), sub-epithelial region is still empty of cell bodies. At +3 weeks, vertical membrane projections extend from deeper cells and contact epidermis. Quantification is shown in (G). n = 28 cells in 3 mice. Scale bar, 10 μm. Epi, epidermis. (F) Before ablation, fibroblast density varies by depth between ~13 and 30 cells per (100 μm)². +2 weeks, fibroblast cell density at 30 μm is largely recovered (yellow). Cell density at 10 μm (orange) and 20 μm (light gray) remains unrecovered at +6 weeks. n = 3 mice. (G) At +3 weeks, membrane-GFP-labeled cells at approximately 30-μm depth were scored for presence of membrane extensions reaching to epidermis (85%). Error bars are SD. n = 28 cells across 3 mice.

Figure 5.. Fibroblast Stability Is Maintained in Non-remodeling Hairy Skin and Lost during Remodeling

(A) Diagrams are XZ projections, images are XY projections at 40 μm (non-remodeling) and 50 μm (remodeling) depths. Non-remodeling hairy tissue does not change over 2 weeks. In remodeling hairy tissue, hair follicles elongate over 2 weeks. HF, hair follicle. Hair follicles are outlined in orange. Full Z stacks: Videos S4 and S5. Scale bars, 50 μm. (B) Revisits in homeostatic non-remodeling hairy skin. A wide field of view is shown with green (PDGFRα-H2BGFP) and red (membrane-tdTomato) (column 1). Day 0, +2 weeks and merge is region highlighted by orange box in column 1. n = 3 mice. Quantification is shown in (G) (Non-rem. hairy, +2). Scale bar, 50 μm (column 1). Scale bar, 20 μm (columns 2–4). (C) Laser ablation (pink dots) and revisits. +1 day, ablated cells are absent. +2 weeks, remaining nuclei are largely stable, and the cell number remains reduced. Quantification is shown in (G) (Non-rem. hairy, Abl.). n = 3 mice. Scale bar, 20 μm. (D) Revisits following ablation with membrane-GFP fibroblast labeling. Membranes from neighboring labeled cells extend into the ablated region. n = 3 mice. Scale bar, 20 μm. (E) Revisits in remodeling hairy skin. Fibroblast nuclei presented in grayscale (Day 0) or merged (middle and right). n = 3 mice. Quantification is shown in (G) (Rem. hairy, +2). Scale bar, 20 μm. (F) Merge of +1 day and +2 weeks following ablation in remodeling hairy skin. Little overlap of fibroblast nuclei is observed. n = 3 mice. Quantification is shown in (G) (Rem. hairy, Abl.). Scale bar, 20 μm. (G) Image correlation coefficients of data represented in (B) (Non-rem. hairy, +2), (C) (Non-rem. hairy, Abl.), (E) (Rem. hairy, +2), and (F) (Rem. hairy Abl.). Compare to Figures 1B and 1H. Non-/rem, non-/remodeling. Re, remount. +2, 2 weeks unperturbed. Abl, ablation. Non-rem. hairy positive control correlation coefficient (0.671) is lower than remounts in non-hairy skin (Figure 1B, 10 μm, Re.) (0.905), indicating that a lower maximum correlation is expected for this tissue. Rem. hairy remount positive control correlation coefficient is even lower (0.360), nearing the limitations of this analysis approach. Error bars are SD. n = 3 mice for 2-week time course data. n = 3 regions for remounts.

Figure 6.. Fibroblast Stability Results in Progressive Accumulation of Localized Cell Loss during Aging that Is Compensated by Membrane

(A) Non-hairy paw skin fibroblast nuclei in 2-month-old mice (column 1) and ≥16-month-old mice (column 2). In Voronoi diagrams (top: 2 months old, bottom: ≥16 months old), polygons are color coded by size. n = 5 mice for each age. Scale bars, 100 μm. (B) Number of cell areas in Voronoi diagrams exceeding 900 μm², sampled from 1,000 × 1,000 × 10-μm regions in the upper dermis (~2,500 cells). One sample per mouse. n = 5 mice (2 months old). n = 4 mice (8–12 months old). n = 5 mice (≥16 months old). p = 0.0499 (2 months versus 8–10 months). p = 0.0114 (2 months versus ≥ 16 months). Error bars are SD. (C) Cell density across randomly sampled 200 × 200 × 40-μm regions of skin dermis from mice in each age cohort. p = 0.0018 (2 months versus 8–10 months). p = 0.0016 (2 months versus ≥16 months). Error bars are SD. n = 3 mice per age group. (D) Revisits of skin across 5 months, beginning in skin already containing gaps. Red-only signal (Merge) is from nuclei that disappeared during the 5-month period. Cells disappeared in localized clusters of cell loss, rather than uniformly throughout the skin. Insets (column 4) highlight examples of localized clusters of cell loss that occurred during (top, 1) and before (bottom, 2) the 5-month period. Quantification is shown in Figure S6E. Scale bar, 100 μm. Scale bar, 50 μm (insets). (E) Image of naturally occurring localized cluster of cell loss, highlighting membrane-GFP fibroblast labeling in grayscale. Insets highlight a normal nuclei organization region (top, 1) and a nuclei depleted region (bottom, 2). Quantification is shown in (F). Scale bar, 50 μm. Scale bar, 10 μm (insets). (F) Membrane occupancy in 12-month-old mice between normal nuclei and gap regions, computed by thresholding the GFP signal (see STAR Methods). Compare to Figure 2G. Error bars are SD. n = 3 mice for each bar. (G) Membrane occupancy across randomly sampled 200 × 200 × 10-μm regions of mice containing gaps. Compare to Figure 2G. Error bars are SD. n = 3 mice. (H) Model of cell loss during aging. Cells are lost in localized clusters that are similar in size and shape to our laser ablation assays. Similar to after laser ablation, the nuclei neighboring physiological cell losses remain stable in position but project new membrane extensions into the region.