Wnt/β-catenin Signaling Pathway Promotes Transdifferentiation from Fetal Skin Fibroblasts to Keratinocyte-like Cells

Background: Wound healing is a dynamic, sequential,and complex physiological process, including a variety of cellular events, such as proliferation, adhesion, chemotaxis, and apoptosis. Skin broblasts and keratinocytes are the two most important cells involved in wound repair, and Relying on the proliferation and differentiation of keratinocytes to form epithelium to completely cover the wound is the most ideal result for wound repair, so expanding the source of keratinocytes is a huge challenge. In this study, we examined the phenomenon that fetal skin broblasts spontaneously transdifferentiated into keratinocyte-like cells in conventional culture, and evaluated the characteristics of KLCs and the potential mechanisms of the transdifferentiation process. Methods: HFF-1 were routinely cultured in ordinary DMEM medium for more than 40 days,and observed the cell morphology. The cytological properties of KLCs at the cellular and molecular levels were detected by RT-PCR, Western-blot, immunouorescence, Transwell, and cell scratch experiments.The functionality and safety of KLCs were determined through wound healing and tumorigenicity experiments. And high-throughput transcriptome sequencing (RNA-seq) was performed to explore the mechanism underlying HFF-1 transdifferentiation. Results: The transdifferentiation process started on the 25th day and was completed by the 40th day. KLCs and KCs had similar expressions at the molecular and protein levels, both functioned similarly in wound healing and were non-tumorigenic.RNA-seq revealed that the transdifferentiation process was regulated by the activation of the classical Wnt/β-catenin signaling pathway, which could shorten the process to 10 days. Conclusion: This study demonstrates that HFF-1 can spontaneously transdifferentiate into KLCs with conventional culture conditions, and the Wnt/β-catenin signaling pathway regulates the transdifferentiation process. black arrow (n=5). c Immunohistochemical staining revealed CK5-positive cells in tumor tissues in the KLCs, KCs and Hepg2 groups (n=5).

The repair of wounds depends on the division and proliferation of keratinocytes, which fuses gradually and reform the epidermis. When extensive burns occur, the structure and function of the skin and its accessory organs are severely damaged. The current therapeutics in clinical practice mainly rely on autologous or allogeneic skin grafting. However, for patients with extensive burns, the available skin donor sites are limited, which hardly meet the needs of large-area skin grafting and thus limit the patient's chances of survival. Keratinocytes (KCs) are the target cells for wound repair, and it remains a huge challenge to obtain keratinocytes on a large scale. Studies have found that KCs can be derived by transdifferentiation from embryonic stem cells (ESCs), iPSCs, adipose stem cells (ATSCs), wound resident mesenchymal cells and broblasts [8]. Bilousova et al found that Oct4, Sox2, c-Myc and Klf4 genes could be transfected with viral vectors to dedifferentiate broblasts into iPSCs, and then by incubation with a medium containing retinoic acid (RA) and a medium containing Bmp4, the induced pluripotent stem cells were further differentiated into keratinocytes. Chen et al. reported that two genes, P63 and KLF4, were transfected to directly transdifferentiate broblasts into keratinocytes [9]. Cell reprogramming provides a new idea for the source of KCs. Although gene transfection using viral vectors can induce cell transdifferentiation, this method is di cult to use in clinical practice due to toxicity, tumorigenic risk, and especially low transdifferentiation e ciency.The purpose of wound healing is to quickly restore the structure and function of the skin, but in extreme cases this will lead to scar formation. Scar tissue is a dysfunctional brous tissue [10]. Unlike wound healing in adults, perfect tissue regeneration is observed in the fetus, which is characterized by the absence of scars [11].
When we routinely cultured fetal skin broblasts, a wonderful phenomenon of cell transdifferentiation was observed, that is, HFF-1 spontaneously transdifferentiated into keratinocyte-like cells (KLCs) in a regular high-sugar DMEM medium in a time-accumulating manner, and the transformed KLCs showed similar behaviors and effects to KCs in terms of cell characteristics, cell behavior and wound healing, and were not tumorigenic.To further explore the potential mechanism of HFF-1 transdifferentiating into KLCs, we performed RNA-seq. Through analysis, we found that this time-dependent HFF-1 transformation is regulated by the Wnt/β-catenin signaling pathway. The discovery of cell transdifferentiation has opened up a new horizon for skin regeneration and scarless wound healing in severe burns.

Cell isolation and culture
Human skin tissue specimens were taken from our burn center, with ethical approval and informed consent. Human skin KCs were extracted as previously described [12]. In brief, the obtained skin tissue was cleaned with PBS (500ml,Cat#G4202-500ML, Servicebio) for three times before the subcutaneous tissue and adipose tissue was removed, and the skin tissue was trimmed to make it regular.
Subsequently, the skin tissue was placed into Dispase II enzyme (2.5mg/mL,Cat#D4693-1G,Sigma) for digestion overnight at 4°C, and the epidermis and dermis were separated the next day.
Then,trypsin/ethylenediaminetetraacetic acid (0.25%,Cat#25200072,Gibco) preheated at 37°C was used to grind and digest the skin tissue for 5 minutes, and then the cell suspension was sucked. The said trypsinization was repeated three times until the skin tissue was completely digested. The cell suspension was ltered through a 70um cell sieve and sterilized (0.22um,Cat#SLGP033RB,Millex), centrifuged at 13000rpm for 3 minutes, and washed once with PBS. Afterwards, the cells were seeded on a 6-well plate (Cat#3516S,Corning), pre-coated with type IV collagen (Cat#C6745,Merck), and routinely cultured with 1% HKGS (Cat#S0015,Gibco) using epilife medium (Cat#Mepicf500,Gibco). Fetal skin broblast cell line (HFF-1,The cell identi cation certi cate is in the supplementary material 1) was purchased from JiKai Gene (Shanghai,China) and grown in DMEM high glucose medium (Cat#SH30022.01,Hyclone) with 10% FBS (Cat#10099141C,Gibco). The third-generation cells were used in the experiment. HepG2 liver cancer cells were donated by Peking University People's Hospital and cultured under the same conditions as described above. All cells were cultured in a 5% CO 2 , 37°C incubator, and the medium was changed every three days. All cells were grown without antibiotics.

Flow Cytometry
Flow cytometry was used to quantitatively analyze HFF-1 vimentin and CK14 positive cells in the process of transdifferentiation. Simply put, the cell suspension at each time point in the HFF-1 culture process was taken xed with 4% paraformaldehyde xative (Cat#P0099,Beyotime) at 4°C for 30 min, and then the permeation buffer TritonX-100 (Cat#P0096,Beyotime) was used to in ltrate the cells and further incubated with the corresponding antibody , rabbit anti-CK14 (Cat#ab192056,Abcam) and mouse anti-Vimentin (Cat#ab8069,Abcam)overnight at 4°C. After incubation, the cells were washed to form a 200 μl cell suspension. Flow cytometry analysis of cell suspension was performed with FACScan (Beamcyte-1026,China).

RT-PCR analysis
RNA extraction and qRT-PCR detection were performed as previously described [13]. The total RNA was extracted with Trizol (Cat#15596026,Thermo Fisher), and the precipitate was washed with 75% ethanol and dissolved in distilled water without ribonuclease. The purity and concentration of total RNA were determined with an ultraviolet spectrophotometer (NanoDrop 2000,Thermo Fisher,Waltham, MA,USA). As per the instructions of FastKing gDNA Dispelling RT SuperMix (Cat#KR118, TianGen, China), a 20μl reverse transcription reaction system was prepared and incubated at 42°C for 15 min and 95°C for 3 min in a T100 thermal cycler (BIO-RAD, Hercules, CA, USA), and then stored at 4°C. Afterwards, Top Green qPCR SuperMix, Nuclease-free Water, cDNA template, and forward and reverse ribonucleic acid primers were added to form a 20μl reaction system. The ampli cation was performed in the CFX Connect realtime system (BIO-RAD,Hercules,CA,USA) as follows: 94°C for 30 s, 95°C for 5 s, 56°C for 15 s, 72°C for 10 s for 45 cycles. The CT value of each well was detected and recorded, and 2-ΔΔCt was used to normalize the gene expression in the sample to the expression of GAPDH. The primer sequence was shown in Table 1.

CCK8 assay
After the cells were trypsinized (Cat#KR11825200056, Gibco), HFF-1, KLCs and KCs were seeded in a 96well plate, each group of cells with 6 wells. After they grew adherently for 8h. 10ul CCK8 reagent (Cat#HY-K0301, MCE) was added to each well, and the cells were incubated at 37°C for 4 h before a multifunctional microplate reader (Biotek Synergy2, USA) was employed to measure the absorbance 12, 24, 36, 48hour at 450 nm wavelength.

Cell migration analysis
Scratch assay: After the cells were trypsinized, HFF-1, KLCs and KCs were seeded in a 6-well plate at a density of 5×10 5 cells/well. After overnight incubation, a 200μl pipette tip (Cat#TF200-RS, Axygen) was used to form wound scratches. Each well was cleaned with PBS for three times to remove sloughed cells.
At 0, 12, 24, and 48 hour, images were taken with a Leica optical microscope to calculate the number of migrated cells.
Transwell assay: After the cells were digested, Transwell assay was performed in a 24-well Transwell chamber (Cat#3422, Corning). In brief, the mixed cell culture medium (DMEM high glucose medium + 10% fetal calf serum + 1% EGF) was added to the lower chamber, and 200 μl of DMEM high glucose medium was added to the upper chamber after resuspending the cells. After incubation at 37°C for 24 h, the cells were taken out of the chamber, and the cells that migrated to the bottom of the chamber were stained with crystal violet (Cat#C8470, Solarbio, China), and each well was counted manually under a microscope.

Wound healing model and treatment
All animal experiments were approved by the appropriate ethics committee of our hospital. Eight-week-old male BALB/c mice (SPF Biotechnology Beijing, China) were randomly divided into four groups, each with 15 mice. As described by Wu et al [15], a full-thickness skin defect of 5mm × 5mm was created on mice's dorsum. The treatment plan was: (1) PBS group: 100ul PBS was administered into the wound surface; (2) HFF-1 group: 100ul HFF-1 (1×10 7 cells/ml) cell suspension was administered into the wound surface; (3) KLCs group: 100ul KLCs (1×10 7 cells/ml) cell suspension was administered into the wound; (4) KCs group: 100ul KCs (1×10 7 cells/ml) cell suspension was administered to the wound. The above treatment plan was repeated every day, and digital photos were taken on the 0 th , 3 rd , 6 th , and 9 th days. Wound healing rate = (original wound area-actual wound area)/original wound area × 100%.

Tumorigenicity analysis
Eight-week-old male BALB/c-Nu nude mice (SPF Biotechnology Beijing, China) were randomly divided into three groups: KLCs, KCs and Hepg2, with 5 mice randomly selected from each group. Cells from KLCs, KCs and Hepg2 groups were suspended respectively in PBS (5×10 6 cells per 100µl). We injected 100µl of this cell suspension (5×10 6 cells) into the subcutaneous tissue of nude mice. The mice were sacri ced 40 days after cell transplantation. The tumor size was measured and the tumor was dissected for histopathological examination [16].

HE staining
In brief, fresh tissues were xed in 4% paraformaldehyde for 24 h, dehydrated, para n embedded, sectioned, and H&E stained as previously described [17]. The prepared para n sections were depara nized, stained with hematoxylin-eosin (Cat# C0105M, Beyotime), and nally dehydrated and mounted. The histopathological changes were observed with an optical microscope (Nikon DS-U3).

Immunohistochemical analysis
Tissue sections were taken, depara nized with xylene, dehydrated in a decreasing ethanol gradient, and incubated with citric acid buffer. The sample was incubated overnight with mouse anti-human CK5 at 4°C, washed with PBS for three times, and then incubated with secondary antibody for corresponding specimens at room temperature for 1 hour. Subsequently, the sample was incubated with diaminobenzidine and sectioned, and re-stained with hematoxylin to obtain images.

RNA-seq analysis
Total RNA was extracted using Trizol reagent kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. RNA quality was assessed on an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) and checked using RNase free agarose gel electrophoresis. After total RNA was extracted, eukaryotic mRNA was enriched by Oligo(dT) beads, while prokaryotic mRNA was enriched by removing rRNA by Ribo-Zero TM Magnetic Kit (Epicentre, Madison, WI, USA). Then the enriched mRNA was fragmented into short fragments using fragmentation buffer and reverse transcripted into cDNA with random primers. Second-strand cDNA were synthesized by DNA polymerase I, RNase H, dNTP and buffer. Then the cDNA fragments were puri ed with QiaQuick PCR extraction kit (Qiagen, Venlo, The Netherlands), end repaired, A base added, and ligated to Illumina sequencing adapters. The ligation products were size selected by agarose gel electrophoresis, PCR ampli ed, and sequenced using Illumina Novaseq6000 by Gene Denovo Biotechnology Co. (Guangzhou, China). DNA/RNA/Small RNA/cDNA library sequencing was performed on the Illumina HiseqTM 2500/4000 by Gene Denovo Biotechnology Co., Ltd (Guangzhou, China).

DEG analysis
RNAs differential expression analysis was performed by DESeq2 software between two different groups and by edgeR between two samples [18,19]. The genes/transcripts with the parameter of false discovery rate (FDR) below 0.05 and absolute fold change≥2 were considered differentially expressed genes/transcripts.

Principal component and pathway enrichment analysis
Principal component analysis (PCA): PCA was performed with R package gmodels (http://www.rproject.org/) in this experience. PCA is a statistical procedure that converts hundreds of thousands of correlated variables (gene expression) into a set of values of linearly uncorrelated variables called principal components. PCA is largely used to reveal the structure/relationship of the samples/data.
Pathway analysis: Genes usually interact with each other to play roles in certain biological functions.
Pathway-based analysis helps to further understand genes biological functions. KEGG is the major public pathway-related database. Pathway enrichment analysis identi ed signi cantly enriched metabolic pathways or signal transduction pathways in DEGs comparing with the whole genome background [20]. The calculated p-value was gone through FDR Correction, taking FDR ≤ 0.05 as a threshold. Pathways meeting this condition were de ned as signi cantly enriched pathways in DEGs.

Statistical analysis
All data were expressed as mean ± standard deviation (SD). The t test was used for the analysis between two groups, and the one-way analysis of variance (ANOVA) was used to compare the data of three or more groups. P values < 0.05 were considered statistically signi cant. Statistical analysis was performed with the software GraphPad Prism 5.

HFF-1 transforms into KLC spontaneously in a timeaccumulated manner
When HFF-1 was grown in conventional culture medium for more than 40 days without any special inducers, we unexpectedly found that it could spontaneously transdifferentiate into KLCs. The entire transdifferentiation process started on the 25th day, marked with the "diamond-like" appearance of the cell colony, and all HFF-1 appeared "diamond-like" by the 40th day (Figure 1a, 1b). To quantify the cell types in the process of transdifferentiation, CK14-positive cells and vimentin-positive cells were evaluated by ow cytometry. The percentage of CK14-positive cells was 19.55±0.43%, 49.92±0.38%, 76.63±0.64%, and 98.10±0.66% on the 10th, 20th, 30th and 40th day, respectively; the percentage of vimentin-positive cells on the 10th, 20th, 30th and 40th days were72.42±0.48%,59.19±0.10%,22.41±0.33%, and 3.54±0.26% (Figure 1c). Immuno uorescence showed that HFF-1 transformed into KLCs in a time-accumulated manner. The number of positive cells for epidermal cell-speci c protein CK14 gradually increased [21], while the number of positive cells of broblast-speci c proteins Vimentin and a-SMA gradually decreased with time [22,23]. By the 40th day, the number of positive cells of vimentin or α-sma were signi cantly reduced and could hardly be detected (Figure 1d,1e).In short, HFF-1 transdifferentiated to KLCs gradually with time, with decreasing HFF-1 and increasing KLCS, and a entire transformation occurred on the 40th day.

Characteristics of KLCs
To explore the characteristics of KLCs after transdifferentiation, we compared KLCs with HFF-1 and KCs. From a molecular level, the Western blot results revealed that the epithelial marker protein Ecadherin [24] as well as KCs-labeled proteins, including CK5 [25], CK14 [26], CK19 [27,28] and Integrin beta1 [29] were expressed in KCs and KLCs, but not in HFF-1. Meanwhile, the HFF-1 marker protein Vimentin [22,23] was expressed in HFF-1, but not in KLCs or KCs (Figure 2a). Consistent with the Western blot results, qRT-PCR results con rmed that although the mRNA expression levels of E-cadherin and CK19 in KLCs were lower than those in KCs, they were 40 times and 20 times that in HFF-1, respectively. Interestingly, the mRNA expression level of CK14 in KLCs was 100 times that in KCs. The expression of CK14 mRNA in KLCs and KCs was much higher than that in HFF-1. In the meantime, the mRNA expression of Vimentin species in HFF-1 was higher than that in KLCs and KCs (Figure 2b).
From the cellular level, CCK8 assay showed that the proliferation rate of KCs and KLCs was signi cantly higher than that of HFF-1 at 24h, 48h and 72h after the adherent growth of the cells, and the proliferation group of KLCs was higher than that of KCs at each time point (Figure 2c). In terms of cell migration rate, according to the scratch assay results, there was no signi cant difference between HFF-1, KLCs and KCs 12 h after the establishment of the cell wound model. However, at 24 h and 48 h, the migration rate of HFF-1 was signi cantly higher than that of KLCs and KCs (Figure 2e). The Transwell assay further showed that the number of migrating cells decreased in HFF-1, KLCs and KCs (Figure 2d). These results indicated that the proliferative ability of KCs and KLCs was higher than that of HFF-1, but the migration e ciency of HFF-1 was signi cantly higher than that of KCs and KLCs.

KLCs accelerate wound healing
To explore the therapeutic effect of KLCs on wound healing after transdifferentiation, PBS, HFF-1, KLCs and KCs suspension were administered into to a full-thickness skin defect wound (5 mm) on the back of the mice, and the wound was inspected on the 3rd, 6th, and 9th days (Figure 3a). In the PBS group, the wound healing rates on the 3rd, 6th and 9th days were 51.80±0.67%, 73.80%±0.32 and 90.20±4.52%, respectively; in the HFF-1 group, the wound healing rates on the 3rd, 6th and 9th days were 60.60±0.53%, 85.40±0.69% and 92.20±3.97%, respectively; in the KLCs group, the wound healing rates on the 3rd, 6th and 9th days were 72.60±4.63%, 92.20±6.02% and 97.40±3.18%, respectively; in the KCs group, the wound healing rates on the 3rd, 6th and 9th days were 82.80±2.81%, 89.00±4.97% and 96.80±3.52%, respectively. The healing rate of each group at each time point was shown in Figure 3b. In a word, there was no signi cant difference in the therapeutic effect between KLCs and KCs, but the therapeutic effect of KLCs and KCs was much better than that of PBS and HFF-1. Compared with PBS and HFF-1, the wounds treated with KLCs and KCs shrank earlier. On the 9th day after injury, the wound healing rate in the KLCs group was even higher than that in the KCS group. The therapeutic effect of HFF-1 was also signi cantly higher than that of PBS. The results of HE staining showed that on the 3rd day the skin tissue structure of each group was severely abnormal, the epidermal layer fell off, and a large number of neutrophil in ltration was seen; the skin appendages were missing, no hair follicles, sebaceous glands and collagen bers were seen, and a large number of cellulose-like structures were seen , A large number of pathological changes such as angiogenesis. On the 6th day, the skin tissue structure of the PBS group and HFF-1 group was abnormal, there was no obvious epidermal layer structure, and a large number of neutrophil in ltration was still seen; a large number of broblasts and in ammatory cells proliferated in some areas, a large number of angiogenesis, and no dermis layer was seen Sebaceous glands, hair follicles and collagen bers can be seen with cellulose-like structures; but in the KLCs and KCs groups, the skin tissue structure is basically normal, the epidermal structure is intact, no cell degeneration is seen, most of the normal tissues can be seen under the epidermis, and a large number of skin appendages can be seen. Collagen bers are arranged closely and regularly. On the 9th day, the skin tissue structure of the PBS group and HFF-1 group was slightly abnormal, the epidermal structure was relatively complete, some normal tissues were visible under the epidermis, a large number of skin appendages were visible, collagen bers were arranged regularly, and a large number of in ammatory cells were seen in the tissue; KLCs The skin tissue structure of the KCs group and the KCs group is basically normal, the epidermal structure is intact, the collagen bers of the dermis are arranged neatly, and the skin is closer to the physiological state. In short, the wound healing of the KLCS group and the KCS group is signi cantly better than that of the PBS group and HFF-1 Group (Figure 3c). To further con rm that KLCs are closely related to wound healing, CK5 antibody was used to perform immunohistochemical staining on the wound. The results showed that after transplantation of KLCs and KCs derived from HFF-1, a row of single cells remained between the epidermis and the dermis (Figure 3d), which played a role in connection during wound healing.In summary, the degree of re-epithelialization in the KLCs and KCs groups was signi cantly better than that of the HFF-1 and PBS groups, which could substantially promote wound healing.

KLCs were not tumorigenic
Safety is an issue that cannot be ignored in the future application of transformed KLCs cells in wound healing in clinical practice. Therefore, subcutaneous tumorigenesis assay was performed in nude mice.
On the 40 th day after the injection of cells in each group, no obvious tumorigenesis was observed in the KCs and KLCs groups, but there was obvious growth of new organisms at the injection site in the Hepg2 group (control group). The tumor tissues of the KCs, KLCs and Hepg2 groups were removed and weighed ( Figure 4a). HE staining of tumor tissue in the Hepg2 group revealed a large number of neutrophil in ltration, as shown by yellow arrows; local cells were loosely distributed, and the cytoplasm was vacuolated, as shown by red arrows; a large number of tumor cells in the tissue had increased atypia and obviously irregular nuclei, as shown by black arrows, while the KLCs and KCs groups showed the hair follicle structure and did not have the above-mentioned pathological features (Figure 4b).
Immunostaining of tumors with human CK5 antibody demonstrated that there was no carcinogenicity caused by KLCs in vivo (Figure 4d). These results indicate that KLCs do not have the potential to induce tumorigenesis and are safe for in vivo administration in the future.

Identi cation of the mechanism underlying KLCs transformation
To explore the potential mechanism underlying the transformation of HFF-1 cells into KLCs, we analyzed the mRNA pro les of HFF-1 as well as HFF-1 cultured for 25 days, KLCs and KCs by RNA-seq, generated a heat map of the mRNA, and clustered the four cell sample types by PCA analysis (Figure 5a). Candidate mRNAs for dynamic transformation from HFF-1 to HFF-1-KLCs and nally to KLCs were identi ed ( Figure   5b). The results showed that the mRNA expression pro les of KLCs and KCs were similar. We analyzed the signi cantly changed mRNAs based on their dynamic expression patterns and mapped 20 clusters. Among them, 6 color clusters showed that the enriched mRNA was detected, which was statistically signi cant. We found that the gene expressions of clusters 17 and 19 were up-regulated, while clusters 0, 1, 2, and 7 were down-regulated. The p-values of up-regulated clusters 7 and down-regulated 17 were much lower than those of other clusters (Figure 5c). Therefore, clusters 7 and 17 were identi ed as the focus in our next research. KEGG enrichment analysis of the differential genes of clusters 7 and 19 respectively demonstrated that the transdifferentiation of HFF-1 into KLCs was closely related to the changes in the Wnt signaling pathway (Figure 5d). Activation of the Wnt/β-catenin pathway promoted transformation of KLCs β -catenin, GSK-3β, Axin1, APC and the TCF/LEF family of transcription factors are important members of the Wnt/β-catenin signaling pathway. LEF1 can interact with nuclear β-catenin and act as the central transcription mediator of Wnt signaling [48], and Axin1 can form β-Catenin degradation complex with APC and GSK3β to inhibit β-Catenin aggregation in the cytoplasm or entering the nuclear transcription. Based on our previous RNA-seq results, we hypothesize that the transformation from HFF-1 to KLCs is caused by the activation of the Wnt/β-catenin pathway. Western-blot results showed that with the progress of HFF-1 transformation, the levels of Axin1 protein and β-catenin phosphorylated protein increased, resulting in a decrease in the total expression of β-catenin, and also signi cantly inhibiting the levels of APC and p-GSK-3β protein.
From another perspective, the transdifferentiation of HFF-1 into KLCs can be considered as a process of mesenchymal epithelial transformation (MET), and ZEB1 is one of the most critical molecules in this process. The results revealed that β-catenin entered the nucleus and bound to the transcription factor LEF1 after its level increased, thereby increasing the level of LEF1, inhibiting the expression of ZEB1 and epithelial mesenchymal transformation (EMT), and promoting the occurrence of MET, that is, the transdifferentiation of HFF-1 into KLCs (Figure 6a, b). By immuno uorescence staining, β-catenin represented by the green uorescence was distributed in the whole cell with high intensity in the HFF-1 group. However, in the KLCs group and KCs group cell species, the uorescence intensity was reduced in the KLCs and KCs groups, and the uorescence intensity in the nucleus was much lower than that in the cytoplasm, indicating that the expression of β-catenin decreased signi cantly, and the translocation of βcatenin in the nucleus was suppressed (Figure 6c).
Subsequently, to further explore the in uence of the Wnt/β-catenin signaling pathway on the transdifferentiation of HFF-1 to KLCs, HFF-1 was treated with the effective activator of the Wnt/β-catenin signaling pathway CHIR-99021 (10μM,Cat#HY10182, MCE,China) and the inhibitor XAV939 (10μM,Cat#HY15147,MCE,China). Surprisingly, HFF-1 treated with CHIR-99021 began to transform on the 10 th day, but this transformation did not occur in the unhandled, PBS and XAV939 groups (Figure 6d). The characteristics of transformed cells were further examined at the molecular protein level (Figure 6e, f), and the results indicated that the activation of the Wnt/β-catenin pathway played a key role in the transformation of KLCs.

Discussion
Our study revealed that fetal skin broblasts can spontaneously transdifferentiate into KLCs by accumulating over time with ordinary culture conditions. On the 25th day of HFF-1 culture, it was observed that the morphology of HFF-1 cells began to change, and the total transformation rate reached 98% on the 40th day. The transformed KLCs and KCs both highly expressed cytokeratins CK5, CK14, CK19, E-cadherin and integrin β1, and did not express broblast-speci c proteins Vimentin and a-SMA. From the perspective of cell proliferation and migration, transformed KLCs and KCs have similar performance. The activation of Wnt/β-catenin signaling pathway played an important role in the transformation of HFF-1 to KLCs. The possible mechanism underlying this process was that with the activation of Wnt signaling, the stability of the degradation complex composed of β-catenin, axin1, GSK-3β, colorectal adenomatous polyp gene (APC) could be attenuated, thus preventing the degradation of βcatenin induced by phosphorylation and increasing its concentration in the cytoplasm. β-catenin then transferred into the nucleus and bound to lymphoid enhancer-binding factor-1 (LEF1) to inhibit the expression of the downstream target gene ZEB1 (a key factor in EMT) (Figure 7). The transformed KLCs were as effective as KCs in healing wounds, and were superior to PBS and broblasts. More importantly, with the treatment of KLCs and KCs, the immunohistochemical results showed hair follicle-like structures. Our future research will probe into the identi cation of these hair follicle-like structures and their sources.
The phenomenon of spontaneous transdifferentiation of fetal broblasts reported in this study differs signi cantly from previous studies [30]. For example, it has been reported that dermal broblasts can transdifferentiate into KLCs phenotype in only 6 days, accounting for a transdifferentiation rate of about 50%, which requires the combined use of inducers (p63 and KLF4) to enter dermal broblasts, and there are functional differences between KLCs and KCs after induction [31]. In the meantime, we found a report that presented similar ndings with ours, namely, adult skin broblasts underwent similar transformations (from FBs to KLCs) on the 120th day during the routine culture in vitro, which was regulated by the PI3K-AKT pathway mediated by LINC00672 [32]. The difference in transformation time may be attributed to the stronger pluripotency and plasticity of fetal skin broblasts than that of adult skin broblasts. This further shows that human skin broblasts have strong pluripotency and plasticity when cultured in vitro,In addition, RNA-seq results show that the enrichment of PI3K-AKT signaling pathway ranks lower. We believe that the possible cause is the heterogeneity of fetal skin broblasts and adult skin broblasts.Many studies have been reported on the direct induction of skin broblasts into the cell types we need through reprogramming and other methods [33,34]. Therefore, our focus is on the potential of skin broblasts to spontaneously transdifferentiate into KLCs on wound healing and scar repair.
Skin wound healing occurs in a series of overlapped but different steps, and the most ideal outcome of this process is the remodeling of the epithelium and its appendages and the restoration of the epidermal barrier function. After skin injury, the wound healing is mainly manifested with the increase of KCs proliferation. This phenomenon occurs in the proliferation area around the wound, 0.5-1.5 mm away from the wound edge, but the proliferation of KCs was not observed at the wound edge [35][36][37]. Intravital microscopy revealed that during the wound healing process, both the basal layer and the super-basal layer migrated [37,45] and the keratinocytes closer to the wound edge migrated faster, which both migrated and proliferated 0.5 mm away from the wound edge. KCs at the wound edge have a stronger migration ability than those in other parts. We hypothesized that the KCs at the wound edge may have undergone epithelial-mesenchymal transition (EMT), so that the epithelial KCs have acquired certain phenotypes of interstitial cells, which represents the ability to migrate. Re-epithelialization plays an important role in the entire wound healing process, involving the proliferation and migration of epidermal KCs around the wound. The dynamic mutual transformation of the proliferation and migration characteristics of KCs is similar to some aspects of EMT, so the healing process involves part of EMT [38]. Mesenchymal-epithelial transformation (MET) is the reverse process of EMT. In recent years, MET has been largely studied in the eld of tumor. Metastatic tumor cells need continuous EMT-MET to colonize distant organs [39], and complete the distant tumor colonization by epithelial-mesenchymalepithelial transition. Few has explored the relationship between MET and wound healing.It is noteworthy that skin broblasts and epidermal stem cells colonized in the basal layer of the skin and hair follicles need to migrate to the wound to undergo terminal differentiation during the process of wound healing.
We found that the transdifferentiation from broblasts into KLCs is a manifestation of MET, and proposed that this process is mainly regulated by the Wnt/β-catenin-LEF1-ZEB1 signaling pathway (Figure7).It may also be jointly regulated by other unknown genes or signal pathways. It is therefore believed that wound healing is the result of the combined effect of the EMT of keratinocytes and the MET of broblasts, similar to the distant colonization of tumor cells.
Scars are a common consequence of skin injury, causing various adverse effects on patients, including physical deformities and mental illness [40,41].The excessive production of connective tissue (especially collagen) during wound healing may lead to scar formation [42,43], which is closely associated with the synthesis of extracellular matrix (ECM) and the promotion of ECM remodeling by broblasts.Studies have shown that the skin damage of human fetuses in early pregnancy can be cured, forming normal skin tissues without leaving scars [44].The speci c mechanism of scarless healing of fetal skin is not yet clear, but the in vitro transdifferentiation potential of fetal skin broblasts may represent a brand-new therapeutic option.

Conclusions
In this study, we found that fetal skin broblasts can spontaneously transdifferentiate into KLCs when cultured in ordinary DMEM nutrient solution for about 40 days, and this process may be regulated by the Wnt/β-catenin signaling pathway. When the Wnt/β-catenin pathway is activated, the broblast transformation time can be shortened from 40 days to 10 days. The transformed KLCs have similar biological characteristics as skin KCs, which promote wound healing, and have no obvious tumorigenicity. Therefore, the spontaneous transdifferentiation potential of HFF-1 can provide new ideas for extensive and scarless wound healing.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Consent for publication
Not applicable

Figure 4
KLCs were not tumorigenic. a The tumorigenicity test of KLCs, KCs and Hepg2 cells in nude mice (the area indicated by the red arrow on day 0 is the cell injection site, and the area marked by the red arrow on day 40 is subcutaneous tumor on day 40 after cell injection, ***P<0.001,n=5). b HE staining of tumors in the KLCs, KCs and Hepg2 groups. KLCs and KCs group showed normal skin tissue, while there was a large number of in ltrated neutrophils in the Hepg2 group was observed, as shown by the yellow arrow; the local cells were loosely arranged, and the cytoplasm was vacuolated, as shown by the red arrow; a large number of tumor cells in the tissue increased atypia, and the nucleus was obviously irregular, as indicated by the black arrow (n=5). c Immunohistochemical staining revealed CK5-positive cells in tumor tissues in the KLCs, KCs and Hepg2 groups (n=5).

Figure 5
Identi cation of the mechanism of KLCs transformation. a PCA was used to analyze the composition of each group of samples. The more similar the sample composition is, the closer the distance re ected in the PCA chart is. The samples from different effective treatments often aggregate. b Heat maps of differentially expressed genes determined by RNA microarray analysis of HFF-1, HFF-1 (day 25), KLCs and KCs. c According to the gene expression patterns of each group of cell stages, 20 clusters were divided from the number and P value of differential genes, and the number of clusters is shown in the upper left corner. d The pathways with signi cant enrichment of differential genes were obtained and the most important biochemical metabolic pathways and signal transduction pathways involved in differential genes were determined through KEGG analysis on No. 7 and No. 17 clusters.  Molecular mechanism of HFF-1 transdifferentiating into KLCs.