Defining stem cell dynamics and migration during wound healing in mouse skin epidermis.

Results

Spatiotemporal proliferation and migration during WH
To define the role of cell proliferation during the regenerative stage of WH, we performed a 3 mm punch biopsy in the tail skin of adult mice and analysed the result of short-term BrdU incorporation by confocal microscopy on whole-mount epidermis at different time points during WH (Fig. 1a). Immediately after wounding, there was no increase in BrdU incorporation. However, at day 2 (D2) and even more at D4 following wounding, we found that BrdU incorporation was increased by 5-fold in a zone spanning from 500 μm to 1.5 mm from the LE, with 40% of basal cells entering into cycle during a period of 4 h (Fig. 1b). The width of the annulus of cells that proliferated around the wound progressively decreased with time (Fig. 1a,c,d). We found that epidermal cells at the LE, spanning a distance of 500 μm from the wound front, did not incorporate BrdU at any time point from D2 to D7 following wounding (Fig. 1a–c). This showed that cells at the LE of the epidermal sheath, which ensures skin regeneration, do not proliferate actively, but migrate to the centre of the wound. These results confirm the existence of a migrating LE that has been hypothesized for several decades based on the histological examination of wounded tissues and ex-vivo skin explants3. Reaching a maximal size at D4 following wounding, the size of the non-proliferating LE zone progressively decreased over time, suggesting that the specification and differentiation of LE cells occurs only during the early stage of WH (Fig. 1a,d). After D14, the wound edges fused at midline and proliferation resumed at the centre of the wound region (Fig. 1a,e).
As wound contraction contributes to wound closure13, we assessed the relative importance of epidermal regeneration and wound contraction to the overall wound repair. As de novo HF formation only occurs with more extensive wounding and at a later stage14, wound contraction was measured by the distance between the HF and the wound centre at D0 minus the same measurement at a given time point, while the newly formed epidermis was measured by the difference between the radius at D0 (1.5 mm) and the radius at any time point minus the contraction. Surprisingly, we found that the distance between HF triplets and the centre of the wound after the punch biopsy did not decrease significantly from D0 to D7, where proliferation was maximum, suggesting that proliferation is not very productive during the initial stage of wound repair (Fig. 1e,f). From D10 to D14, this distance decreased linearly in time until re-epithelialization was completed (Fig. 1e,f). At this stage the average distance between the HF triplets and the wound centre is 0.9 mm, suggesting that an epithelial regeneration contributes approximately to two-thirds of the healing process, and wound contraction (0.6 mm) is responsible for the remainder.

Cell shape and polarity during WH
The shape and size of the epidermal cells, which is the reflection of the forces that epidermal cells experience during the regeneration process, was very different depending on the wound region and the time point following wounding (Fig. 2a–c). At D0, the basal cells of the LE appeared less compacted (Fig. 2a), consistent with a relaxation in the force exerted on the wound edge. At D1, the LE cells were elongated toward the wound centre (Fig. 2a), as previously shown15, consistent with the active migration of the LE cells toward this point. At D4, basal cells far from the wound presented a regular cuboidal/hexagonal shape (Fig. 2b). The density of basal cells in the proliferative zone was increased, leading to a more compressed cell shape (Fig 2b,d). In contrast, in the non-proliferative zone, the basal cells were bigger, polarized in the same direction, and elongated along an axis perpendicular to the direction of the wound closure (Fig. 2b). This suggests that, at D4 and thereafter, the movement of the LE is a passive process possibly mediated by the proliferating cells (Fig. 2b). Consistent with this notion, blocking epidermal cell proliferation by 5-fluorouracil (5-FU), which inhibited the re-epithelialization process and WH (Fig. 2e), prevented the perpendicular polarization of the LE at D4 (Fig. 2f). These data demonstrate that the two distinct epidermal compartments, the proliferative hub and the LE, present different cell shape and polarity that change with time, likely reflecting the difference in the physical forces that these different zones experience at the different stages of WH.

Molecular signature of LE and proliferation hub during WH
To define the molecular features associated with the formation of the proliferative hub and the LE, we performed transcriptional profiling of different concentric rings of the wound using different sizes of punch biopsy and fluorescence-activated cell sorting (FACS) sorting. A first punch biopsy of 4 mm in diameter around the wound was used to enrich for LE cells (spanning two times 500 μm, the average width of the LE), and a second punch biopsy (6 mm) was enriched for the proliferative hub of the wound (Fig. 3a; Supplementary Fig. 1). We performed a third biopsy far from the wound corresponding to control normal unwounded epidermis. We performed duplicate microarrays of these three skin regions at D4 and D7 post-wounding. We found genes upregulated in both wound regions at the two different time points, which correspond to a generic wound-healing signature. This gene signature included genes regulating cell adhesion (for example, Dsc2), cytoskeleton (for example, Krt6, Krt17), inflammation (for example, Il24, Il33 S100a8/a9), cell signalling (for example, Areg, Ereg, Emb, Epgn), (Supplementary Fig. 2a) and cell cycle-related genes (for example, Ccna2, Ccnb1) (Supplementary Fig. 2b). For some genes, such as Krt6 (refs 3, 16, 17, 18, 19), in which expression was confirmed by immunofluorescence (Supplementary Fig. 2c), Il24, S100a8/a9 or the EGFR ligands, their role in the regulation of WH has previously been described20212223. In other cases, including Fscn1, Emb, Sprr1b and Sprr2h, genes were not known to be involved in skin WH.
We next defined which genes were preferentially upregulated and downregulated in the LE as compared with the proliferative hub (the LE signature) (Supplementary Fig. 2d–f). We found that α5-integrin was highly enriched in the LE signature (Supplementary Fig. 2e); consistent with a previous study that showed that α5-integrin was expressed at the LE of human skin explants ex vivo24252627 and at the LE during eyelid closure, a developmental process that involves epidermal cell migration28, reminiscent of the LE during WH. Whole-mount immunostaining confirmed the rapid upregulation of α5-integrin in the non-proliferative cells of the LE in vivo (Fig. 3b). To refine the molecular signature of the LE without contamination of proliferative cells, we isolated α5-integrin positive cells from a 4 mm punch biopsy by FACS at D4 following wounding (Supplementary Fig. 3) and performed microarray analysis in triplicates. These molecular analyses confirmed the preferential expression of many of the previously described genes expressed during WH, validating the approach used here and allowing for the first time to distinguish the spatial localization of these genes at the LE and/or in the proliferative hub. In addition, the gene signatures of the proliferative and LE cells during wounding uncover many novel genes not previously described during WH and tissue regeneration (Fig. 3c–j). Gene Ontology Enrichment (GO) analysis revealed that the genes upregulated in the LE comprised genes regulating cell adhesion, cytoskeleton organization, epidermal differentiation, cell migration and other processes involved in WH (Fig. 3c). The most upregulated genes of the LE signature were genes coding for proteins regulating cell migration including several metalloproteinases (MMPs) (Mmp9, Mmp13, Mmp1b) (Fig. 3d), whereas Timp3, an inhibitor of metalloproteinase, was the most downregulated gene (Supplementary Fig. 2f), suggesting that the high level of MMPs expressed by the cells of the LE promote the remodelling of the ECM at the wound front allowing the front cells to progress toward the centre. MMPs also help the breakdown of the hemi desmosomes that anchor the cells at the basal membrane and are therefore essential for the movement of basal cells. MMPs deletion in flies and mice results in wound-healing defects due to defective cell elongation, cytoskeleton and basal membrane remodelling as well as cell migration2930313233. The migrating zone also expressed high level of urokinase (Plau) and plasminogen activator (Plaur) (Fig. 3e), two key fibrinolytic proteins contributing to the remodelling of the blood clot during WH34. The LE also expressed high level of Ephb2 and Efnb1 (Fig. 3d), a receptor and its ligand, which have recently been shown to control WH35 as well as other genes such as Cxcr4, C5ar1, Myh9, Procr, Wnt5a, Elk3 (Fig. 3e,g,i,j), which regulate cell migration in other cellular contexts. We found that Inhibin-βa, a subunit of Activin A was overexpressed preferentially at the LE (Fig. 3j), in good accordance with in-situ hybridization of Inhibin-βa during WH36 and the previously reported role of Activin A during WH in mice373839404142.
Our LE signature encompassed many genes controlling cell adhesion, including several protocadherins (Pcdh7, Pcdhb19, Pcdhga1) (Fig. 3f), integrins (Itga5, Itga6) (Fig. 3f) and some of their ECM ligands (Fn1, Lama3, Lamb3, Lamc2) (Fig. 3h), desmosomes (Cdsn)43 and gap junction proteins (Gjb6/Cx30 and Gjb2/Cx26) (Fig. 3f). Corneodesmosin (Cdsn), a desmosome protein of the wound edge signature (Fig. 3f) was reported to be expressed using transgenic reporter mice in the wound edge and in the inner root sheath (IRS) of the hair44.
Many genes controlling cell cytoskeleton and actin remodelling including actin regulators (Fscn1, Cald1, Nav2, Fmnl2), myosin (Myo1b, Myo5b, Myh9, Tpm1, Tpm2) and tubulin (Tubb2a, Tubb3, Tubb6), were preferentially overexpressed at the LE (Fig. 3i), and Tubb6 was previously shown to be upregulated during wounding19. These genes may control the morphology, polarity, rigidity of the cell cortex and migration during WH.
Several genes that might regulate the quiescence of the LE surfaced from the microarray analysis. Gprc5a, an orphan G protein coupled receptor, acting as tumour suppressor genes in the lung by negatively regulating EGFR and Stat3 signalling454647, was strongly upregulated at the LE (Fig. 3j.). The upregulation of E2f7 (ref. 48) or Fgf18 (ref. 49), genes that promote cell quiescence in other contexts may also contribute to shut down of proliferation in the wound LE (Fig. 3g,j).

Spatiotemporal expression of the LE signature during WH
Immunostaining performed against several of these newly identified markers of the LE signature including cell adhesion, receptor and cytoskeleton proteins (Flrt2, Gprc5a, Tubb2, Myo1b, Itga5) confirmed their preferential enrichment at the LE of the wound as predicted by our microarray analysis (Fig. 4a–e). Itga5 was expressed preferentially in the basal cells of the LE (Fig. 4a). Flrt2, a repulsive guidance protein that regulates the migration of neuronal progenitors during embryonic development50 was more expressed in the cells of the wound edge than in the proliferation zone, but was not present in the normal skin epidermis (Fig. 4b). Gprc5a was expressed at the LE of the wound at D4 (Fig. 4c), and similarly to Cdsn44, Flrt2 or Tubb2, to the IRS or precortex cells of the HFs (Fig. 4b–d; Supplementary Fig. 4a–c). Myosin 1b, a tension-sensitive myosin was also expressed in all cell types of the LE (Fig. 4e), which by regulating actin foci stability, controls cell migration or repulsion. Myosin 1b was also expressed in the bulge and outer root sheath (ORS) of the HFs (Supplementary Fig. 4d). The expression of all these newly identified molecular markers of the wound LE signature decreased overtime and at D14 post-wound, when the opposite margin of epidermal cells fused together, these markers were not expressed anymore (Fig. 4a–e). These data demonstrate the transient nature of this wound LE structure.

Uncoupling proliferation and differentiation during WH
To gain further insights into the mechanisms that specify these two distinct regions, we determined whether the fate and the differentiation programme of the LE are linked to cell division, by assessing the impact of blocking cell proliferation on the fate of LE cells. Topical application of 5-FU, which strongly decreased epidermal cell proliferation, did not prevent or impair the expression of LE markers (Fig. 5a–c), demonstrating that the particular differentiation programme of the LE is specified independently of cell division.
As inflammation plays an important role in orchestrating the early step of WH1251, we assessed the impact of blocking inflammation on these two distinct epidermal compartments during re-epithelialization. Interestingly, treating the wounded mice with dexamethasone, a potent anti-inflammatory drug, reactivate proliferation in the LE without impairing its particular gene expression signature (Fig. 5d–g) demonstrating that, the LE-specific gene signature is not associated with terminal differentiation (Fig. 5f,g). These data show that glucocorticoid treatment suppresses a negative regulator of cell cycle acting on the LE. As glucocorticoid can also directly act on keratinocytes52, this negative regulator may originate either from the keratinocytes or from the inflammatory cells and the granulation tissue. Although, these data do not allow to discriminate whether the inhibition of proliferation at the LE is regulated by an intrinsic or an extrinsic mechanism, these results provide compelling evidence that the LE-specific cellular quiescence and gene expression signature can be functionally and molecularly uncoupled.

Clonal analysis of IFE SC during WH
To follow the progeny of the basal epidermal cells, and study their cellular dynamics at the single cell level, we performed clonal analysis on K14CREER/Rosa Confetti mice targeting preferentially IFE SCs12 (Fig. 6a). We administrated Tamoxifen (TAM) 14 days before wounding and analysed the respective clonal contribution during the healing process (Fig. 6b). At the end of the re-epithelialization, K14-labelled cells gave rise to long streaks of labelled cells directed toward the LE of the wound (Fig. 6c–e). With a clone merger probability estimated at roughly 5% (Fig. 6f; Supplementary Note 1), we deduced that streaks labelled with the same fluorescent protein were clonal in origin, derived from single SCs. The clonal lines were often interrupted by unlabelled cells (Fig. 6d,e), suggesting that a cycle of active SC proliferation followed by cell intercalation from neighbouring clones (or clonal fragmentation) occurs repetitively during WH. Interestingly, all of these fragmented clonal streaks originate from the proliferative hub previously described (Fig. 6g). Quantification of the clonal persistence revealed that more than 90% of IFE-labelled clones were lost during the first week following wounding (Fig. 6h), consistent with the majority of progenitors maintaining homoeostatic behaviour, leading to a progressive decrease in the labelled cell fraction125354. Importantly, this clonal dynamic contrasts with that reported for oesophagus, where repair seems to involve a switch of progenitors to a proliferative mode of division55, or following in vitro culture of human keratinocytes56. Consistent with a major increase in the rate of terminal differentiation, as measured by clonal persistence, the epidermal thickness increased during the same period (Fig. 6i).
Three-dimensional analysis of labelled clones revealed that, in contrast to clones in the control regions that are composed of stacks of cells that lie on the top of each other (Fig. 6j), at D4 K14CREER IFE SCs gave rise to basal and suprabasal cells that migrate toward the wound edge (Fig. 6k). The restriction of the basal footprint of the clone at the trailing edge suggest that marked SCs undergo predominantly asymmetric cell division, giving rise to a steady production of progenitors. Interestingly, despite the high rate of proliferation observed in the first days following wounding, at this time point, K14CREER clones are not on average bigger as compared to D0, suggesting that the effect of SC renewal is not yet observed at the population level.
By D14, most of the persisting K14CREER clones became enlarged in their basal attachment and were composed of a majority of suprabasal cells (Fig. 6l), while others formed long streaks of basal cells and suprabasal cells (Fig. 6m), suggesting that these clones produced an increased number of basal progenitors leading to basal cell expansion. Interestingly, these streaks originate from a minority (>1%) of K14CREER-induced cells and seem to be responsible for only part of the wound regeneration (for details see Supplementary Note 1).
It has been proposed that, during wounding, differentiated suprabasal cells can revert back to a progenitor state and actively contribute to repair5758. To test this possibility, we administrated TAM to InvCREER/RosaYFP that labelled suprabasal cells and rare basal cells (Supplementary Fig. 5a,b), performed a punch biopsy and analysed the contribution of labelled suprabasal cells during wounding. Ten days after wounding, most of the lineage labelled suprabasal cells had been shed from the skin surface or remained present as spinous or granular cells. Despite the high frequency of suprabasal cell labelling, we found only very rare basal cells initially targeted by the InvCREER, and these contribute minimally and transiently to the wound repair as previously described3. We found no evidence that suprabasal cells can revert back to a progenitor-like state, as the density of basal cells contributing to the wound repair is much lower compared to that of basal cells labelled at induction (Supplementary Fig. 5a,b). These data suggest that dedifferentiation of differentiated suprabasal cells does not contribute to WH in the tail epidermis.

Clonal analysis of infundibulum SC during WH
To assess whether different types of SCs arising from distinct epidermal regions present different clonal dynamics during WH, we performed clonal analysis on Lrig1CREER/Rosa Confetti mice targeting the upper HF SCs that include cells from the infundibulum, junctional zone and sebaceous gland, and that have been shown to contribute to WH8 (Fig. 7a,b). At D14, most of the Lrig1-labelled cells give rise to long streaks of progeny from the infundibulum to the LE (Fig. 7c) with the clones presenting the same fragmentation reported for the K14 tracing (Figs 6d,e and 7c). The 3D reconstruction showed that the Lrig1 clones, starting from D7, had an analogous cellular composition as the K14 with long streak of basal and suprabasal cells emanating from the infundibulum and directed toward the wound centre (Fig. 7d–f).

Similar clonal dynamic of different epidermal SCs during WH
To gain further insight into the clonal dynamics of the IFE and the HF-derived SCs populations, we used a previously validated biostatistical framework596061 to infer with high confidence the number of cells and cellular composition of clones (basal versus suprabasal cells) arising from single K14 and Lrig1CREER/Rosa Confetti-targeted cells (Fig. 8a–e; Supplementary Note 1). To define how SCs balance proliferation and differentiation in the proliferative hub surrounding the wound edge, we analysed the clonal composition of K14 and Lrig1CREER/Rosa Confetti-derived clones from D0 to D14 after wounding. From D4 to D14, the basal size of clones of both K14 and Lrig1CREER grew remarkably linearly (Fig. 8a,c). At first sight, this behaviour could indicate neutral competition, as has been reported for uninjured epidermis1262. However, adapting the standard model of balanced stochastic progenitor cell fate, the observed rate of increase in the average basal size would translate to an unreasonably fast cell cycle time of >6 h. Rather, in this case, the linear increase indicates the labelling of an asymmetrically dividing subpopulation that drives the basal expansion through the steady production of progenitors (Supplementary Note 1). On the basis of the quantitative clonal analysis, we find that the data are consistent with a proliferative hierarchical model in which, during WH, a putative SC population at the apex divide perfectly asymmetrically giving rise to self-renewing progenitors in which the frequency of symmetrical duplication is balanced by symmetrical differentiation (Fig. 8f). Indeed, the existence of a persistent (SC) and a transient (progenitor) basal subpopulation might explain the fragmentation of clonal streaks. From the modelling of the clonal dynamics, we found that such model indeed predicts not only the increase in average basal clone size (Fig. 8c) but also recapitulates the detailed distributions of basal clones sizes in the K14CREER assay (Fig. 8d; Supplementary Note 1). Notably, such a hierarchical model coincides with that inferred from the study of homoeostatic turnover of interscale epidermis, but where the proliferation rate of SCs and progenitors have been massively increased54. Surprisingly, the conserved linearity of the average basal clone size and the shape of the size distribution suggested that Lrig1CREER targets SCs belonging to the same hierarchy as that targeted by K14CREER, but where a burst of proliferative activity at the earliest stages of regeneration expands the average number of SCs in each clone (Fig. 8c–f; Supplementary Note 1). Altogether, these data suggest that, irrespective of the epidermal origin, the regenerative stage of WH involves a sustained increase in proliferative activity of a minority of SCs, while keeping the fate behaviour and lineage relationship of SCs and progenitors largely unperturbed from their homoeostatic dependences.

Discussion
Our study uncovers the clonal dynamics and individual contribution of SCs coming from different epidermal compartments during skin WH in mice. In contrast to what has been proposed for oesophagus repair and the growth of human keratinocytes in vitro5556, our data show that WH does not increase the self-renewal capacities of progenitors, but rather leads to their massive depletion as proliferation increases. The repair of the skin epidermis does not induce a change in the cellular hierarchy of SCs and progenitors, or a change in the balance between renewal and differentiation but rather involves an increase in the proliferation rate of a small population of SCs that gives rise to progenitors upon asymmetric division leading to a linear increase in the individual clone size over time. Interestingly, IFE and infundibulum SCs present very similar clonal dynamics during wound repair, despite the fact that they are recruited from different regions of the epidermis.
Our study confirms the existence of two distinct epidermal zones during wound repair; a proliferative hub composed of the IFE and HF-derived SCs and their progeny and a LE composed of non-proliferative cells3, and uncovers the timing, gene expression signature and mechanisms that specify these two distinct compartments during WH (Fig. 9). We propose that the non-proliferative LE of the wound acts as a scaffold allowing a harmonious healing process, by creating a platform secreting high level of enzymes that remodel the surrounding ECM and fibrin clot allowing tissue regeneration to progress toward the centre of the wound and protecting the SCs and their progeny from the immediate vicinity of the wound front and infection.
A similar wound margin structure with elongated migrating cells has been described during the early stages of WH following incisional wound in humans63, supporting the notion that this mode of wound repair has been conserved during evolution. Further functional studies will be needed to refine the respective role of the genes identified here in the LE signature. The high level of expression of several of these genes in patients with chronic ulcers6465 suggests that defects in the formation and/or function of this structure may induce wound-healing problems leading to chronic ulcer formation.

Methods

Mice
K14CREER transgenic mice were provided by Fuchs66. Lrig1-CreERT2 mice were a kind gift from Jensen67. Involucrin-CreERT2 were previously described12. Rosaconfetti mice were provided by Clevers68. Rosa YFP69 mice were obtained from Jackson Laboratory. All animals were mixed strains. No statistical methods were used to predetermine sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment. Mice colonies were maintained in a certified animal facility in accordance with European guidelines. The experiments were approved by the local ethical committee (CEBEA).

Targeting Confetti or YFP expression in wound experiments
For lineage tracing experiment, K14CreER/RosaConfetti, Lrig1-CreERT2/RosaConfetti and Involucrin-CRERT2/Rosa-YFP male and female adult (between 2 and 6 months old) mice were induced at 2 months with 0.03 mg g−1, 0.27 mg g−1 or 0.08 mg g−1 of TAM (Sigma-Aldrich), respectively, by intra-peritoneal (IP) injection. For K14-CreER and Lrig1-CreERT2 tracing, 2 weeks after TAM induction, mice were anesthetized (5% xylazine 10% ketamine in PBS) and circular pieces of epidermis were removed from the tail epidermis using a 3 mm diameter biopsy punch (Stiefel, Ireland). For Involucrin-CRERT2/Rosa-YFP mice the wound was performed 4 days after TAM injection to analyse the contribution of suprabasal cells. Each mouse was subjected to three different punches in the tail and at least five mice per time points were analysed.

Proliferation experiments
For BrdU experiments, CD1 male and female adult (between 2 and 6 months old) mice were wounded, injected with one single injection IP of BrdU (50 mg kg−1 in PBS) at the different time points and killed 4 h after. For the quantification at least an area of 1.5 mm2 per region per wound was analysed with Imaris software (Bitplane) to determine the percentage of BrdU positive cells.

Epidermal whole-mount and immunostaining
Pieces of skin tail surrounding the wound were incubated in PBS/EDTA (20 mM) on a rocking plate at 37 °C for 1 h. Epidermis was separated from the dermis using forceps as an intact sheet and washed two times with PBS. Pieces of epidermis were pre-fixed in 4% paraformaldehyde for 1 h at room temperature. Epidermis were rinsed two times with PBS for 5 min and conserved in PBS with 0.2% azide at 4 °C. For the immunofluorescence staining, the entire pieces of epidermis were incubated in blocking buffer (1% BSA, 5% horse serum, 0.8% Triton in PBS) for 3 h at room temperature on a rocking plate (100 r.p.m.). The samples were incubated in primary antibodies overnight at 4 °C. The primary antibodies used were the following: anti-Integrinβ4 (rat, 1:200, BD Biosciences), anti-K14 (chicken, 1:2,000, custom batch, Thermo Fischer), anti-GFP (rabbit, 1:200, Molecular Probes). Samples were then washed three times in PBS with 0.2% tween during 1 h and incubated in appropriate secondary antibodies diluted 1:400 in blocking buffer for 1 h at room temperature on the rocking plate. For BrdU staining, samples were incubated in HCl 1 M at 37 °C for 45 min, washed with PBS 0.2% tween, stained with anti-BrdU (rat, 1:200, Abcam) in blocking buffer and with appropriate secondary antibody. The following secondary antibodies were used: anti-rat, anti-chicken, anti-rabbit conjugated to AlexaFluor488 (Molecular Probes), to Rhodamine Red-X or to Cy5 (Jackson Immuno Research). Alexa488-conjugated phalloidin (Life Technologies) was used 1:200 in blocking buffer to visualize F-actin microfilaments. Nuclei were stained in Hoechst solution diluted 1:5,000 for 30 min and mounted in DAKO-mounting medium supplemented with 2.5% Dabco (Sigma). Immunostaining pictures of the whole mounts presented in the figures are representative images of at least five different experiments.

Microscope image acquisition and measurements
All pictures of section immunostaining were acquired using the Axio Imager M1 Microscope, the AxioCamMR3 or MrC5 camera and using the Axiovision software (Carl Zeiss). Acquisitions were performed at room temperature using × 20 numerical aperture (NA) 0.4 (Carl Zeiss). All confocal images were acquired at room temperature with a LSM780 confocal system fitted on an AxioExaminer Z1 upright microscope equipped with C-Apochromat × 40/1.1 or Plan Apochromat × 25/0.8 water immersion objectives (Zeiss, Iena, Germany). Optical sections 512 × 512 pixels were collected sequentially for each fluorochrome. The data sets generated were merged and displayed with the ZEN2012 software (Zeiss). 3D reconstitution images were processed using Imaris (Bitplane) software.

Histology and immunostaining on sections
Skin epidermis was removed from tailbone, embedded in OCT and kept at −80 °C. Sections of 6 μm were cut using a CM3050S Leica cryostat (Leica Mycrosystems). After fixation in 4% paraformaldehyde for 10 min at room temperature, tissues were washed three times in PBS for 5 min and incubated in blocking buffer (1% BSA, 5% Horse serum, 0, 2% Triton in PBS) for 1 h at room temperature. Primary antibodies were incubated overnight at 4 °C. Sections were rinsed three times in PBS and incubated with secondary antibodies and Hoechst in blocking buffer for 1 h at room temperature. Sections were again washed three times with PBS. The following primary antibodies were used: anti-K14 (chicken, 1:20,000, custom batch, Thermo Fischer); anti-K6 (rabbit, 1:6,000, Covance), anti-Flrt2/3 (rabbit, 1:100, Sigma); anti-Gprc5a (rabbit, 1:100, Sigma); anti-Myo1b (rabbit, 1:100, Sigma) and anti-α5-integrin (PE-conjugated rat, 1:200, BD or rabbit, 1:200, Abcam). For the anti-Tubb2 (rabbit, 1/1,000, Abcam) staining, fixation was performed in methanol at −20 °C for 2.5 min and the rest of the protocol was performed as described. The following secondary antibodies were used diluted to 1:400: anti-rabbit, anti-rat, anti-chicken conjugated to Alexa Fluor 488 (Molecular Probes), to rhodamine Red-X (Jackson Immunoresearch) or to Cy5 (Jackson Immunoresearch). Nuclei were stained in Hoechst solution (1:2,000) and slides were mounted in DAKO-mounting medium supplemented with 2.5% Dabco (Sigma). Immunostaining pictures of the skin sections presented in the figures are representative images of at least five different experiments.

Dissociation of epidermal cells and cell sorting
The dermis and epidermis were removed from the tail bone and micro dissection was performed using two different sizes of punch biopsy: one punch of 4 mm in diameter was done, around the wound of 3 mm in diameter, to remove the LE zone and one punch of 6 mm in diameter was used to remove the proliferative centre, at D4 and D7 after wound. Another piece of skin was taken, as control, from a region far from the wounded area. The two replicate samples at each time points were a pull of five CD1 mice. The samples were incubated in HBSS (Gibco) 0,25% trypsin (Gibco) at 37 °C until the epidermis was separated from the dermis (30 min). Epidermis was then incubated on a rocking plate (100 r.p.m.) at room temperature for 5 min. Basal cells were mechanically separated from the epidermis by flushing 10 times under the epidermis. Tissues were then cut in small pieces with a scalpel and incubated again for 5 min on a rocking plate (100 r.p.m.) at room temperature. Trypsin was then neutralized by adding DMEM medium (Gibco) supplemented with 2% Chelex Fetal Calf Serum (FCS) and the cells were mechanically separated by pipetting 90 times and filtrated on 70 μm filter (Falcon). Cells were incubated in 2% FCS/PBS with primary antibodies for 30 min on ice, protected from the light, with shaking every 10 min. Primary antibodies were washed with 2% FCS/PBS and cells incubated for 30 min in APC-conjugated streptavidin (BD Biosciences), on ice, with shaking every 10 min. Living epidermal cells were gated by forward scatter, side scatter and negative staining for Hoechst dye. For the first microarray analysis, basal IFE and infundibulum cells were stained using PE-conjugated anti-α6-integrin (clone GoH3; 1/200, ebioscience) and bulge cells were stained with biotinylated CD34 (clone RAM34; 1:50, BD Biosciences). Basal cells from the IFE were targeted using CD34 negative and α6-integrin positive gating. For the second microarray analysis, FITC-conjugated anti-α6-integrin (CD49f) (clone GoH3: 1:200, ebioscience) and PE-conjugated anti-α5-integrin (CD49e) (clone 5H10-27, 1:200, BD Bioscience) and biotinylated CD34 antibodies were used. The cells were sorted using CD34 negative α6-integrin positive α5-integrin positive gating. Before sorting, the cells were filtered again on 70 μm filter (Falcon). Fluorescence-activated cell sorting analysis was performed using FACSAria I at high pressure (70 psi) and FACSDiva software (BD Biosciences).

Microarray analysis
Sorted cells (300 cells per sample) were collected directly in 45 μl of lysis buffer (20 mM DTT, 10 mM Tris–HCl pH 7.4, 0.5% SDS, 0.5 μg μl−1 proteinase K). Samples were then lysed at 65 °C for 15 min and frozen. RNA isolation, amplification and microarray were performed in the Functional Genomics Core, Barcelona. cDNA synthesis, library preparation and amplification were performed as described70. Microarrays were then performed on Mouse Genome 430 PM strip Affymetrix array at IRB Functional Genomics Core (Barcelona, Spain). The data were normalized using RMA algorithm. The entire procedure was repeated in three technical independent samples. Genetic signatures were obtained by considering genes presenting a fold change greater or smaller than 2 or −2, respectively, in each replicates. The accession number for the microarray data are GEO: GSE76795 and GSE93638.

5-FU and dexamethasone experiments
For the 5-FU experiments, mice were treated shortly after wound surgery with Efudix 5% cream (Meda Pharma) applied topically on the upper part of the tail three times per day until the sacrifice. For the dexamethasone experiments, dexamethasone powder (Sigma) was resuspended at 1 mg ml−1 in ethanol 100% and diluted 5 × in sterile PBS. The mice were injected intraperitoneally once per day at the dose of 1 mg kg−1. The treatment started 2 days before the wound surgery and was sustained until the end of the experiment.

Data availability
Data supporting the findings of this study are available within the article (and its Supplementary Information files) and from the corresponding author on reasonable request. The accession number for the microarray data are GEO: GSE76795 and GSE93638.

Additional information
How to cite this article: Aragona, M. et al. Defining stem cell dynamics and migration during wound healing in mouse skin epidermis. Nat. Commun.
8, 14684 doi: 10.1038/ncomms14684 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Material

Supplementary Information
Supplementary Figures, Supplementary Note and Supplementary References