EGFR marks a subpopulation of dermal mesenchymal cells highly expressing IGF1 which enhances hair follicle regeneration

The skin harbours transcriptionally and functionally heterogeneous mesenchymal cells that participate in various physiological activities by secreting biochemical cues. In this study, we aimed to identify a new subpopulation of dermal mesenchymal cells that enhance hair follicle regeneration through a paracrine mechanism. Integrated single‐cell RNA sequencing (scRNA‐seq) data analysis revealed epidermal growth factor receptor (EGFR) as a marker of distinct fibroblast subpopulation in the neonatal murine dermis. Immunofluorescence staining and fluorescence‐activated cell sorting (FACS) were used to validate the existence of the cell population in Krt14‐rtTA‐H2BGFP mouse. The difference of gene expression between separated cell subpopulation was examined by real‐time PCR. Potential effect of the designated factor on hair follicle regeneration was observed after the application on excisional wounds in Krt14‐rtTA‐H2BGFP mouse. Immunofluorescence staining demonstrated the existence of dermal EGFR+ cells in neonatal and adult mouse dermis. The EGFR+ mesenchymal population, sorted by FACS, displayed a higher expression level of Igf1 (insulin‐like growth factor 1). Co‐localisation of IGF1 with EGFR in the mouse dermis and upregulated numbers of hair follicles in healed wounds following the application of exogenous IGF1 illustrated the contribution of EGFR+ cells in promoting wound‐induced hair follicle neogenesis. Our results indicate that EGFR identifies a subpopulation of dermal fibroblasts that contribute to IGF1 promotion of hair follicle neogenesis. It broadens the understanding of heterogeneity and the mesenchymal cell function in skin and may facilitate the potential translational application of these cells.


| INTRODUC TI ON
Mesenchymal cells are the most abundant cell types in the skin dermis, and actively participate in skin homeostasis and repair.
However, the heterogeneity and functional diversity of mesenchymal cells complicate our insights into their physiological role in skin.
Over the past decade, a number of studies have uncovered several dermal mesenchymal cell lineages that have distinct functions in skin homeostasis and diseases. [1][2][3] Cell lineage tracing analysis has revealed that CD26 and DLK1 are markers for papillary fibroblasts in the upper dermis and reticular fibroblasts in the lower dermis, respectively, each of which displays distinct activities in skin repair and regeneration. [1][2][3] Fibroblasts derived from the lower dermis move upward during wound healing and exhibit profibrotic activities. 3,4 Indeed, subsets of dermal fibroblasts with profibrotic activities have been identified, and shown to have distinct features, including ADAM12 + PDGFRα + cells 5 and engrailed-1 + mesenchymal cells. 6 On the other hand, studies have implied that certain dermal fibroblast subsets enhance skin regeneration, such as Hic1 + cells, 7 SM22α + cells. 8 Dermal mesenchymal cells play diverse functions in skin homeostasis and wound repair by secreting extracellular matrix (ECM) molecules and soluble factors, and excessive ECM synthesis is found to inhibit hair follicle regeneration after wound healing. 9 Dermal fibroblasts play a pivotal role in hair follicle regeneration during wound repair. Dermal fibroblast-conditioned medium has recently been shown to induce hair regeneration, 10 suggesting that certain fibroblasts enhances hair regeneration through a paracrine mechanism. Insulin-like growth factor 1 (IGF1) plays a vital role in tissue homeostasis and regeneration via autocrine/paracrine/endocrine activities: it enhances skeletal regeneration and brain repair after injury in mice. 11,12 In the skin, IGF1 promotes hair follicle growth. 13,14 Several studies have shown that immune cells such as M2 macrophages and also resident epidermal T cells secrete IGF1 to enhance wound healing and hair neogenesis. 15,16 Several dermal fibroblast cell lines reportedly produce high levels of IGF1, 17-19 but do not do not appear to correspond to known fibroblast subpopulations in vivo. Thus, while heterogeneous dermal fibroblasts are the major cell type to mediate skin regeneration after wounding, the subpopulation that produces higher levels of IGF1 has not yet been clearly determined.
We set out to identify dermal mesenchymal cells crucial for hair follicle regeneration. The surge of scRNA-seq technology application has resulted in a large amount of publicly available scRNA-seq data.
Since neonatal skin heals with higher levels of hair follicle regeneration after wounding, 20 we decided to start with the re-analysis of scRNA-seq data of neonatal mouse skin, which were reported earlier. 20 After screening a collected database of mouse surface markers and secreting factors, we identified a mesenchymal cell subpopulation expressing higher levels of Egfr + and Igf1 + , which was potentially involved in hair follicle regeneration. Immunofluorescence analysis confirmed the presence of EGFR + dermal fibroblasts in neonatal and adult dermis, and their co-expression of IGF1. In addition, sorted EGFR + dermal fibroblasts showed higher expression levels of Igf1 than EGFR − dermal fibroblasts. Moreover, application of IGF1 in skin wounds enhanced hair follicle regeneration in mice. Thus, our results identify that EGFR + dermal fibroblasts potentially mediate skin regeneration likely through release of IGF1.
Skin GFP expression was induced by applying 0.2 mg/mL tetracycline (dissolved in 70% ethanol). Prior to topical application of tetracycline, adult mice had their dorsal hair removed by sequential application of hair clipper and hair removal cream. Twenty-four hours later, tetracycline treated dorsal skin was ready for experiments. The animals were maintained in cages in a temperature-controlled environment, with a 12-h light-dark cycle and free access to food and fresh water. were ready for use.

| IGF1 injection and wound model
Post Day 2 neonatal Krt14-rtTA-H2BGFP littermates were anesthetized by isoflurane inhalation (RWD Life Science Co., Shenzhen, China). Two symmetrical full thickness skin wounds were created on dorsal skin with a 2-mm-diameter skin biopsy punch as previously described. 39 From post-injury Day 7 to Day 14, each mouse received a daily injection of 50 μL PBS or murine IGF1 (MCE, HY-P7070) solution (dissolved in PBS at 1 ng/μl) in the two wounds, respectively (beneath the scab into the wound bed tissue). Skin tissues at the wound site were collected at post-injury Day 15 and subjected to AP staining, and skin tissues obtained at Day 21 were subjected to immunofluorescence staining.
Seven-week-old adult mice of the same breed and near body weight were anesthetized by intraperitoneal administration of 1% sodium pentobarbital (35 mg/kg). Mouse dorsal hair were completely removed by sequentially shaved with hair clipper, depilated with hair removal cream. Then, the dorsal skin was cleansed with wet cotton swab and sterilized with iodine solution and 75% ethanol, respectively. One 1-cm-diameter disinfected biopsy punch was used to create 1-cm diameter full-thickness wound on each mouse back.
Following wounding, mice were individually housed in plastic cages.
Wounds were let to heal and left with no additional dressing.

| scRNA-seq data analysis
Gene-by-cell matrix from repeated samples was merged by Seurat To eliminate redundancy enrichment results, online tools REVIGO with default parameters was used for visualising. 42

| Fluorescence-activated cell sorting (FACS)
Dermal single-cell suspensions were prepared as described in 2.3. Data analysis was performed using Flowjo software (Treestar).  Table S2.

| Immunohistochemistry
Mice were sacrificed to harvest the desired skin areas. The skin was  were used for EGFR + cells sorting according to manufacturer direction.

| Alkaline phosphatase (AP) activity staining
According to the manufacturer's protocol, AP staining was per-

| Statistical analysis
GraphPad Prism 8.0 software was used to carry out statistical analysis of all experiments. All data were presented as means ± SEM. Paired Student's t-tests were applied to compare the difference of hair follicle numbers between two groups. Unpaired Student's ttests were performed to compare expression levels between the indicated two groups. Values of p < 0.05 were indicative of statistically significant differences.

| A distinctive EGFR + mesenchymal cell population identified in neonatal mouse skin via reanalysis of scRNA-seq data
To dissect the heterogeneity of fibroblasts in mouse skin, we reanalysed scRNA-seq data from samples of P2 neonatal mouse skin. 20 A total of 7786 sequenced cells and 18,876 genes from three libraries met the qualification threshold. We annotated with collected markers (Table S1, Figure 1A), extracted cells that expressed dermal cell marker genes and clustered them at proper resolution ( Figure 1B). Fibroblast cluster 4 (Fibro4) simultaneously expressed Egfr and Igf1 ( Figure 1C,D), indicating that an EGFR marked subtype highly expresses Igf1. Dermal cells were then classified into EGFR + and EGFR − cells, which were transcriptionally distinct from dermal papilla (DP) cells ( Figure 1E). Genes that were upregulated in the EGFR + cell population were subject to gene ontology enrichment analyses; results showed that, besides gene ontology terms related to connective tissue development and epithelial cell proliferation, GO terms that include wound healing and extracellular matrix organisation were also enriched, suggesting the possible function of EGFR + cells in the wound healing process ( Figure 1F).

| EGFR + dermal mesenchymal cells highly express Igf1
To establish the existence of EGFR + cells in the neonatal murine dermis, we used immunofluorescence staining. As with the data analysis results, a non-epithelial subpopulation stained positive for EGFR in neonatal murine skin (Figure 2A

| Igf1-expression colocalizes to EGFR + dermal mesenchymal cells in normal and wounded adult skin
We next used immunofluorescence staining to detect the expression of EGFR in adult mouse dorsal skin. In adult pdgfra-GFP mice, dermal fibroblasts are labelled by GFP. Immunofluorescence staining for EGFR detected cells double positive for GFP and EGFR in the dermis ( Figure 3A). In Krt14-rtTA-H2BGFP adult mice, partial nonepithelial cells were positively stained with EGFR which distributed across the dermis ( Figure 3B). It was reported that during wound healing, IGF1 signalling mainly functions during the granulation and epithelialisation stage. 21 We stained IGF1 in adult wounds. IGF1 + cells were mainly distributed in the lower layer of the wound bed in Day 7 wounds ( Figure 3C), and some of them were colocalized to EGFR + cells ( Figure 3D). The results indicated that the EGFR + mesenchymal cells participated in wound healing by highly expressing Igf1.

| IGF1 enhances wound-induced hair neogenesis
Next, exogenous IGF1 was tested if it facilitates wound-induced hair neogenesis (WIHN). In vivo application of exogenous IGF1 was con-  Figure 4B). Considering that the neogenic hair follicles in healed wounds have irregular directions, it was barely possible to have every hair follicle in complete shape in one frozen section slice. Each circular structure formed by the KRT14 + cells in the wound bed was considered as one hair follicle ( Figure 4C). Paired Student t-test statistics was employed to investigate the discrepancy of hair follicle numbers between two groups ( Figure 4D). The number of hair follicles that grew in the IGF1 treated group was significantly larger than the control group at post wound Day 21 ( Figure 4C,D).
The results indicate that exogenous IGF1 promoted WIHN.

| DISCUSS ION
Despite the presence of diverse subpopulations of fibroblasts in the dermis, the functional heterogeneity of these cells remains largely unknown. A number of studies have used scRNA-seq to identify subpopulations in skin from humans and mice. 3,22 We tried to take advantage of the amounted scRNA-seq data to identify dermal mesenchymal population that is important for hair follicle regeneration.
Together with experimental studies, we identified that EGFR marks a subpopulation of dermal mesenchymal cells highly expressing IGF1 which enhances hair follicle regeneration.
Mesenchymal cells play a variety of roles primarily by secreting ECM and soluble substances. As soluble factors such as growth factors and cytokines are critical components of stem cell niches, here we have focused on identifying mesenchymal stem cells that express high levels of secreted factors. Using a two-marker screening approach, we have documented that cell surface protein EGFR marks a mesenchymal subpopulation that highly expressed Igf1 in the dermis of neonatal and adult mice. Human dermal fibroblasts reportedly stain positive for EGFR. 23 Our results display that EGFR is also distributed across the murine dermis (Figures 2A and 3A), similar to its distribution in human skin (The Human Protein Atlas database). In addition, EGFR is expressed in the human fibroblast cell line Hs68, and the level of EGFR expression decreases with cell aging, 24 which might lead to its reduced responsiveness to epidermal growth factor (EGF). Furthermore, human dermal fibroblasts produce IGF1, 25,26 yet but their molecular identity remained unknown. We have demonstrated that EGFR + dermal fibroblasts express higher levels of IGF1 than EGFR − fibroblasts.
Our findings point to the possible roles of EGFR + cells in the human dermis. To obtain EGFR + mesenchymal cells, we employed a negative selection strategy, as per earlier studies, 27 in which GFP and FITC + cells, including GFP + epidermal cells, CD45 + immune cells and CD31 + endothelial cells were removed ( Figure 2B). It seemed that the strength of EGFR + signal arrows pointed to was much weaker than that of the surrounding epidermal part (Figure 2A), which could be the result of the fact that signal strength from a single mesenchymal cell did not match that of concentrically aligned epidermal cells. It could also be EGFR expression in the mesenchymal cell was significantly lower than that in the epithelial cells. We learned from the dot plot (showing in Figure 1C) that, in addition to Fibro4, the designated EGFR + cell population, other populations with high Igf1  Figure 4C,D). This is consistent with earlier studies on the roles of IGF1 in promoting WIHN. 16 Also, besides being present in epidermal cells, IGF1-R is found in fibroblast-like cells that are dispersed throughout the uppermost regions of the dermis, 26 where the fibroblasts are believed to play a supportive role in hair follicle regeneration. 1 In addition, the IGF1-R-related signal is involved in protecting cells from apoptosis during wound healing, 35 potentially contributing to hair follicle regeneration. In Figure 3C, the number of IGF1 + cells is greater than the number of EGFR + cells, suggesting that besides EGFR + mesenchymal cells, other cells such as M2 macrophage 16 that are located in wounds also likely secrete IGF1.
Our study has several limitations. (i) The characterisation of EGFR + mesenchymal cells both in vitro and in vivo can be improved.
However, after we removed CD45 + immune cells and CD31 + endothelial cells, notably high expression of Col1a1, a mesenchymal marker, and the co-localisation of EGFR and VIM provided sufficient evidence to identify the EGFR + cells as mesenchymal cells.
(ii) Experiments on loss-of-function analysis of EGFR + dermal fibroblasts support their role in hair follicle regeneration. Mouse skin in which epidermal EGFR was depleted displayed lesions similar to skin rashes seen in human, 36  However, the yield of sorting EGFR + dermal cells is very low, so a large number of mice would be needed to obtain a sufficient number of the EGFR + dermal cells. In addition, we used recombinant IGF1 to induce hair follicle regeneration, since the bioactivity of recombinant IGF1 is similar to that of endogenous IGF1 38 ; however, recombinant IGF1 may not fully simulate the function of IGF1 that is released by dermal fibroblasts.
Our finding of the dermal EGFR + mesenchymal cell subpopulation that highly expresses IGF1 and enhances cutaneous regeneration has increased the current understanding to mesenchymal cell biology. Mesenchymal cells exist widely in many tissues in our body.
Therefore, in future studies it is of interest to investigate whether some of them also express IGF1 and have an enhanced role in tissue haemostasis and regeneration. In addition, our study has suggested a therapeutic potential of the EGFR + mesenchymal cells in wound repair. On the other hand, it is also of significance to examine whether a similar population of mesenchymal cells exists in tumours and their activities in tumorigenesis and progression.
In summary, we have used a double marker screening approach in combination with animal experiments to identify an EGFR + mesenchymal population that expresses high levels of Igf1 in neonatal and adult mouse dermis, which are potentially involved in hair follicle regeneration.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors indicate no potential conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data underlying this article will be shared on reasonable request to the corresponding author.