The small G protein Arf6 expressed in keratinocytes by HGF stimulation is a regulator for skin wound healing

The earlier step of cutaneous wound healing process, re-epithelialization of the wounded skin, is triggered by a variety of growth factors. However, molecular mechanisms through which growth factors trigger skin wound healing are less understood. Here, we demonstrate that hepatocyte growth factor (HGF)/c-Met signaling-induced expression of the small G protein Arf6 mRNA in keratinocytes is essential for the skin wound healing. Arf6 mRNA expression was dramatically induced in keratinocytes at the wounded skin, which was specifically suppressed by the c-Met inhibitor. Wound healing of the skin was significantly delayed in keratinocyte-specific Arf6 conditional knockout mice. Furthermore, Arf6 deletion from keratinocytes remarkably suppressed HGF-stimulated cell migration and peripheral membrane ruffle formation, but did not affect skin morphology and proliferation/differentiation of keratinocytes. These results are consistent with the notion that Arf6 expressed in skin keratinocytes through the HGF/c-Met signaling pathway in response to skin wounding plays an important role in skin wound healing by regulating membrane dynamics-based motogenic cellular function of keratinocytes.


Results
Wound-dependent expression of Arf6 mRNA in mouse skin keratinocytes. We have previously reported that Arf6 mRNA is abundantly expressed in epithelial cells of various mouse tissues 23 . Consistent with this report, Arf6 mRNA was highly expressed in the dorsal skin epidermis of embryonic day (E)15.5 and postnatal day (P)1 mice, although the expression level was extremely lower in P56 adult mice (Fig. 1A). Interestingly, the expression of Arf6 mRNA in the skin epidermis of adult mice was dramatically enhanced when the skin was wounded: the enhancement was clearly detectable at 2 days after wounding and sustained at least up to 7 days (Fig. 1B,C), raising a possibility that Arf6 expressed in the wounded skin epidermis functions to heal skin wounds.
Skin epidermis forms the layer structures composed of basal, spinous, granular and corneum layer 24 . Fluorescence in situ hybridization of the wounded skin epidermis for Arf6 mRNA, and immunostaining for loricrin, keratin1 and keratin5, which are marker proteins for granular, spinous and basal layer, respectively, revealed that Arf6 mRNA was expressed in the keratin5-positive basal layer, but not in other layers (Fig. 1D). In addition, some population of Ki67-positive proliferating cells at the wounded site was found to express Arf6 mRNA (Fig. 1D). These results are consistent with the notion that Arf6 is expressed in the proliferating keratinocytes in the basal layer in response to wounding, and contributes to skin wound healing.
Deletion of Arf6 from keratinocytes causes delay of skin wound healing. To address the notion described above that Arf6 expressed in skin keratinocytes in response to wounding is involved in skin wound healing, we generated conditional knockout mice specifically lacking Arf6 in skin keratinocytes (K-Arf6-cKO: K14-Cre; Arf6 flox/flox ) and analyzed skin wound healing in these mice. K-Arf6-cKO mice were born at the expected Mendelian ratio, overtly healthy and able to reproduce (data not shown) without any obvious abnormalities in their skin and hair growth ( Fig. 2A). Expression of Arf6 mRNA induced by skin wounding, which was observed in wild type of mice, was almost completely suppressed in K-Arf6-cKO mice (Fig. 2B), demonstrating the successful deletion of Arf6 from skin keratinocytes. In K-Arf6-cKO mice, as expected, closure of a full-thickness wound was delayed compared with control mice (Fig. 2C,D), supporting the notion described above.
HGF/c-Met signaling promotes the expression of Arf6 mRNA in the wounded skin. Another issue to be clarified is which signaling induces Arf6 mRNA expression in response to skin wounding. It has been reported that epidermal growth factor (EGF), basic fibroblast growth factor (bFGF/FGF2), hepatocyte growth factor (HGF) and keratinocyte growth factor (KGF/FGF7) are released from fibroblasts, mesenchymal cells and keratinocytes at the wounded skin area, and involved in wound healing 3,4 . Administration of these growth factors into the skin of wild type mice elicited the expression of Arf6 mRNA in the skin epidermis after 1-3 days of administration ( Fig. 3A-C), indicating that these growth factors have a potential to promote the Arf6 mRNA expression in the skin. To identify growth factors and their receptors physiologically responsible for the induction of Arf6 mRNA expression in the wounded skin, specific inhibitors, PD153035 for the EGF receptor 25 , SU5402 for the bFGF and KGF receptors 26 , and PHA665752 for the HGF receptor c-Met 27 , were administered to the wounded skin, and the expression of Arf6 mRNA was analyzed by in situ hybridization. The Arf6 mRNA expression induced by wounding was specifically suppressed by the c-Met inhibitor, but not by other inhibitors for EGF, bFGF and KGF receptors, in a dose-dependent manner ( Fig. 3D-G). Moreover, the treatment with HGF up-regulated Arf6 mRNA expression in primary cultured keratinocytes, and this up-regulation was suppressed by the c-Met inhibitor PHA665752 (Fig. 3H). These results suggest that HGF/c-Met signaling acts as a physiological trigger of Arf6 mRNA expression in keratinocytes at the injured skin to promote wound healing.
In support of this notion, the treatment of control mice with the c-Met inhibitor PHA665752 significantly delayed wound healing (Fig. 4A,B). However, it was found that the c-Met inhibitor still showed the small suppression of skin wound healing in K-Arf6-cKO mice (Fig. 4C,D). These results indicate that other factor(s) is also involved in the HGF/c-Met axis-induced signaling pathway of skin wound healing, although Arf6 at least in part functions as a downstream molecule in this signaling pathway. Another possibility is that other Arf isoforms are up-regulated by knockout of Arf6 or stimulation with HGF to compensate for the loss of Arf6. This is unlikely because expression levels of Arf1 and Arf5 mRNAs in primary cultured keratinocytes were not affected by Arf6 knockout and HGF stimulation, while Arf6 mRNA expression levels in control and HGF-stimulated cells were significantly decreased by the deletion of Arf6 ( Supplementary Fig. S1).
HGF-stimulated cell migration and peripheral membrane ruffle formation are impaired in Arf6-deleted keratinocytes. Migration  has been reported that Arf6 plays essential roles in migration of epithelial cells in vitro 17,18 . These reports, taken together with the result shown above that HGF/c-Met signaling is responsible for the expression of Arf6 mRNA in keratinocytes, led us to speculate that delay of skin wound healing in K-Arf6-cKO mice might be attributable to the impairment in cell migration of keratinocytes. To address this assumption, wound closure was measured by the in vitro scratch-wound healing assay with primary cultured keratinocytes prepared from control and K-Arf6-cKO newborn mice. The expression level of Arf6 was significantly diminished in primary cultured keratinocytes prepared from K-Arf6-cKO mice (Fig. 5A), demonstrating that Arf6 is successfully, but not perfectly, knocked-out. In Arf6-deficent keratinocytes, wound closure stimulated by HGF remarkably delayed (Fig. 5B,C). prepared from E15.5, P1 and P56 wild type of mice were hybridized with an antisense probe1 (left panels) and a sense probe (right panels) for Arf6 mRNA. (B) Dorsal skin sections prepared from 8 weeks old wild type of mice at 0, 2 and 7 days after skin wounding were hybridized as in (A). High magnification images of rectangular regions in the left panels are shown in the right panels. (C) Signal intensity of Arf6 mRNA of wounded area relative to that of non-wounded area were analyzed and shown as mean ± SEM. Statistical significance was assessed using one-way ANOVA with Tukey's HSD test, * P < 0.05. (D) The injured dorsal skin sections prepared from 8 weeks old wild type of mice at 2 days after wounding were visualized by fluorescent in situ hybridization for Arf6 mRNA (middle panels, red) and by immunostaining for loricrin, keratin1, keratin5, and Ki67 (left panels, green). Furthermore, it was found that HGF-stimulated peripheral membrane ruffling formation, which is essential for cell migration, was markedly suppressed in Arf6-deficient keratinocytes (Fig. 5D,E). These results strongly support the notion that delay of skin wound healing in K-Arf6-cKO mice is attributed to the defect in the peripheral membrane ruffling formation and subsequent cell migration of keratinocytes.
Skin morphology and proliferation/differentiation of keratinocytes during skin wound healing were not affected by Arf6 deletion. Other cellular events essential for skin wound healing are proliferation and differentiation of keratinocytes at the wounded skin. Formation of hyperproliferative epithelium and eschar at the wounded site were similarly observed in control and K-Arf6-cKO mice as analyzed by haematoxylin-eosin (H & E) staining (Fig. 6A). The number of proliferating keratinocytes at the wounded site, which was assessed by the BrdU staining of keratinocytes, was comparable between control and K-Arf6-cKO mice (Fig. 6B,C). Furthermore, the layer structure of the epidermis visualized by immunohistochemistry for loricrin, keratin1 and keratin5, and number of Ki67-positive cells were almost the same between control and K-Arf6-cKO mice (Fig. 6D). Thus, layer structure of the epidermis and proliferation/differentiation of keratinocytes at the wound site are not affected by deletion of Arf6, leading to the conclusion that delay of skin wound healing in K-Arf6-cKO mice is attributed to the defect in cell migration of keratinocytes caused by the impaired ruffle formation.

Discussion
The results obtained in this study provide evidence for the first time that expression of Arf6 mRNA in keratinocytes is induced via HGF/c-Met signaling when the skin is wounded, and thus upregulated Arf6 at least in part plays an important role in skin wound healing by regulating the membrane dynamics-based motogenic function of keratinocytes.
It has been reported that expression of HGF and c-Met mRNAs is elicited in keratinocytes at wound edge when the skin is injured 7 , although the HGF protein is expressed only in hair follicle mesenchyme of normal skin, but not in epidermis, and c-Met exists in neighboring hair bulb keratinocytes 28 . Similar to HGF and c-Met mRNA expression, expression of Arf6 mRNA in keratinocyte was induced in response to skin wound ( Fig. 1B-D), which was specifically inhibited by the c-Met inhibitor PHA665752 in a dose-dependent manner (Fig. 3D-G), suggesting that HGF/c-Met axis is a key signaling pathway for the induction of Arf6 mRNA expression. Although the molecular mechanism(s) through which the HGF/c-Met signaling induces Arf6 mRNA expression remains to be clarified, ERK/MAPK, PI3K and the transcription factor SP1 may coordinately play roles in the HGF/c-Met-induced Arf6 mRNA expression. This idea is consistent with the report showing that SP1 is required for the EGF-induced expression of Arf6 mRNA in the human glioblastoma cell line U87 cells 29 . The idea is further supported by the report that HGF treatment of the human keratinocyte cell line HaCaT cells increases the phosphorylation of SP1, which enhances its transcription activity, through the regulation of MAPK and PI3K 30 . Thus, it is reasonable to speculate that ERK/MAPK, PI3K and SP1 coordinately mediate the induction of Arf6 mRNA expression through HGF/c-Met axis in keratinocytes at the wounded skin. Clarification of the molecular mechanisms for HGF-dependent induction of Arf6 mRNA expression in response to skin wound could contribute to developing novel treatment protocols and drugs to enhance skin wound healing.
Although we provided evidence that Arf6 at lease in part plays an important role in wound healing of the skin in vivo, we cannot totally rule out the possibility that other factor(s) is also involved in the skin wound healing promoted by HGF. This possibility is raised by the observation that the effect of Arf6 knockout on skin wound healing was smaller than that of the c-Met inhibitor (Figs 2 and 4) and that delay of the wound healing was still  observed in K-Arf6-cKO mice (Fig. 4C,D). It has been reported that HGF/c-Met axis plays crucial roles in wound healing by regulating keratinocyte migration and proliferation 7 . The results obtained in this study demonstrate that the function of Arf6 is limited to the HGF-stimulated keratinocyte migration: Deletion of Arf6 from keratinocytes inhibited the HGF-dependent keratinocyte migration (Fig. 5B,C), but not proliferation (Fig. 6B,C). From these observations, it is plausible that Arf6 plays an important role in the motogenic cellular events of keratinocytes and other factor(s) is essential for keratinocyte proliferation in the HGF/c-Met signaling pathway of cutaneous wound healing.
Another issue to be defined is the molecular mechanisms by which Arf6 regulates the HGF-dependent membrane ruffle formation which is required for cell migration of keratinocytes. Although further experiments are required to clarify this issue, PIP5K is easily speculated to be a downstream effector of Arf6 in the signaling pathway of HGF-induced membrane ruffle formation of keratinocytes, based on the following reports: (1) PIP5K functions as a downstream effector of Arf6 in the signaling pathway for EGF-stimulated membrane ruffle formation in HeLa cells 16 ; (2) Arf6 is activated in hepatocytes and vascular endothelial cells upon HGF stimulation 21,22 ; (3) HGF stimulation of the human hepatocyte carcinoma cell line HepG2 cells activates PIP5K through Arf6 (unpublished data); (4) the versatile membrane phospholipid PI4,5P 2 produced by PIP5K reorganizes actin cytoskeleton at the plasma membrane 16 , which is a critical cellular events for cell migration 31 . PI4,5P 2 at the plasma membrane interacts with actin binding proteins such as gelsolin 32 and cofilin 33 to reorganize actin filament which is essential for membrane ruffle formation to promote cell migration. In addition, PI4,5P 2 is an essential recruiter/activator of focal adhesion (FA) components such as tailin 34 , ERM proteins 35 , and vinculin 36 to regulate the cell adhesion to extracellular matrix which is also essential for cell migration. Therefore, PI4,5P 2 generated at the plasma membrane by Arf6-PIP5K axis would facilitate the membrane ruffle formation and cell adhesion at the leading edge of the cell to promote keratinocyte migration during wound healing of the skin.
Alternative downstream molecule of Arf6 is the lipid metabolizing enzyme PLD1. Arf6 has been reported to directly activate PLD1 19 , and activated PLD1 plays an important role in the migration of epithelial cells 18 . Furthermore, we have previously demonstrated that PLD1 is essential for cytokine-induced migration of neutrophils by controlling actin cytoskeleton reorganization 37 . Thus, Arf6/PLD1 axis might be critical for the regulation of actin cytoskeleton reorganization-based motogenic function of keratinocytes during wound healing of the skin.
Trafficking of the adhesion molecule β 1 integrin between the plasma membrane and endosomes 38 , which regulates attachment/detachment of cells to/from extracellular matrix 39 , is a crucial cellular event for Arf6-mediated peripheral membrane ruffling 40 and cell migration of keratinocytes 41 . Since production of PI4,5P 2 at the plasma membrane and the recycling endosome has been implicated in the β 1 integrin trafficking, including its internalization 42 and/or recycling 43 , it is plausible that the Arf6/PIP5K axis-dependent PI4,5P 2 production at these cellular compartments regulates membrane ruffle formation and cell migration of keratinocytes through β 1 integrin trafficking. This idea is further supported by the reports demonstrating that Arf6 is involved in β 1 integrin internalization to disassemble FA during keratinocytes migration 44 and in HGF-induced β 1 integrin recycling in vascular endothelial cells to form FA and promote cell migration 22 . Thus, Arf6 could regulate cell migration by regulating multiple steps of β 1 integrin trafficking process.
In the present study, we provide insight into the physiological significance of HGF/c-Met/Arf6 signaling in the skin wound healing. In addition to the skin, other tissues, such as liver, lung and kidney, express Arf6 23,45 , suggesting that Arf6 plays an key role in wound healing-related pathophysiological phenomena in these tissues. In support of this assumption, cytohesin family members of Arf6 GEFs, which regulate Arf6 activity, are required for the HGF/c-Met signaling-dependent recovery of acute kidney injury 46 . Thus, it is of interesting to investigate pathophysiological roles of HGF/c-Met/Arf6 signaling in a wide variety of tissues.

Methods
Animals and generation of K-Arf6-cKO mice. All experiments with mice were carried out according to the Guideline for Proper Conduct of Animal Experiments, Science Council of Japan. The protocols for experiments with mice were approved by Animal Care and Use Committee, University of Tsukuba.
C57BL/6J mice were purchased from SLC, Japan. K-Arf6-cKO mice were generated by mating Arf6 flox/flox mice 22 , which were used as control for K-Arf6-cKO mice, with K14-Cre mice which express Cre after E15 47 . Ageand sex-matched 8 to 12 weeks old mice were used for in vivo experiments.

Analyses for effects of growth factors and their receptor inhibitors on Arf6 mRNA expression in the skin.
To assess the effects of growth factors on Arf6 mRNA expression in the normal skin, 300 ng of EGF (Roche), 140 ng of bFGF (Pepro Tech), 120 ng of HGF (Sigma-Aldrich) or 120 ng of KGF (Pepro Tech) in 200 μ l of growth factor-reduced matrigel matrix (BD Biosciences) was subcutaneously injected into the dorsal skin of adult mice, and skin sections were prepared at 1 or 3 days after injection.
To investigate the effects of inhibitors for growth factor receptors on the expression of Arf6 mRNA in the wounded skin of adult mice, inhibitors of EGFR (PD153035, TOCRIS), FGFR (SU5402, SANTA CRUZ) and c-MET (PHA665752, SANTA CRUZ) at the indicated dose were topically applied to the wounded skin every day after wounding, and skin sections were prepared 2 days after wounding for in situ hybridization of Arf6 mRNA.
In vivo skin wound healing assay. Full-thickness wound was generated on dorsal skin of control and K-Arf6-cKO mice using 4.5 mm leather punch (BIGMAN), and wound closure of dorsal skin was analyzed up to 12 days after the wounding. To examine the effects of the c-Met inhibitor PHA665752 on the skin wound healing in control and K-Arf6-cKO mice, mice anesthetized with isoflurane (Intervet, Inc.) were applied with 33 μ g of PHA665752 topically at the wounded skin area every day after wounding, and wound closure was analyzed up to 8 days after wounding. Digital images of the wounded skin were obtained by LUMIX DMC-L10 camera (Panasonic). Wound area was measured, and the percentage of wounded area was calculated with ImageJ (NIH).

Histological analysis and immunohistochemistry of the skin. H & E staining of the mouse skin
were carried out according to the standard method 48 . Images were obtained with Biozero BZ-8000 microscope (Keyence).
For the immunohistochemical analysis of the skin shown in Fig. 6, skin sections were washed three times with PBS, and blocked with H-PHT (1.5% heat-inactivated goat serum in 0.1% Triton X-100/PBS) for 1 hr. Sections were then incubated with anti-loricrin, anti-keratin1, anti-keratin5, or anti-Ki67 antibody at 4 °C overnight.
After washing three times with PBST (0.1% Tween-20/PBS), sections were incubated with Alexa Fluor ® 488 goat anti-rabbit IgG antibody at r.t. for 1 hr, and counterstained with DAPI. Images were obtained with Zeiss Axio Observer Z1 equipped with an AxioCam MRm (ZEISS).
Assay for proliferation of skin keratinocytes. Proliferation of skin keratinocytes in vivo was assessed using BrdU (nakalai tesque). Mice were wounded on their dorsal skin and intraperitoneally administrated with 250 mg/kg body weight of BrdU at 3 days after wounding. After 2 hr of BrdU administration, sections of the wounded skin were prepared as described above in "In situ hybridization". After sections were washed three times with PBS, endogenous peroxidase in the sections was inactivated with 3% H 2 O 2 /methanol solution at r.t. for 15 min, and subsequently antigens in the section was retrieved by heating the section in 0.01 M sodium citrate, pH 6.0, with microwave oven. The sections were washed three times with 0.1% Triton X-100/PBS, blocked with H-PHT at r.t. for 1 hr, then incubated with anti-BrdU antibody (B2531, Sigma-Aldrich) in H-PHT at 4 °C overnight. After washing three times with PBST, sections were incubated with biotinylated anti-mouse/anti-rabbit IgG antibody (BA-1400, VECTOR) in H-PHT at r.t. for 1 hr. The sections were washed three times with PBST, and incubated with ABC Reagent kit (PK-6100, VECTOR) at r.t. for 30 min. After being washed twice with PBS, sections were reacted with DAB (DAKO). Sections were also counterstained with haematoxylin for nuclear staining. Images were obtained by BZ-X710 microscope (Keyence). The number of BrdU-positive cells at wound site were measured by ImageJ (NIH).
Primary culture of keratinocytes. The skin obtained from dorsal and ventral region of control and K-Arf6-cKO newborn mice were placed on the filter paper filled with 0.25% trypsin/PBS at 4 °C overnight. Epidermis were separated from dermis, and collected in EMEM.06 (Eagle's essential medium containing 0.06 mM CaCl 2 ). After gentle shaking, cells were harvested on collagen type IV-coated dishes in keratinocyte growth medium consisting of primary fibroblast conditioned medium, low-calcium medium [EMEM.06 containing 10% chelexed fetal bovine serum], 2 ng/mL EGF (BD biosciences), 0.75 mM Aminoguanidine nitrate (Sigma-Aldrich), 10 -10 M Cholera toxin (Sigma-Aldrich), and 0.4 μ g/mL Hydrocortisone (Sigma-Aldrich), as ScienTific REPORTS | 7:46649 | DOI: 10.1038/srep46649 described previously 49 . These primary cultured keratinocytes were subjected to assays for in vitro wound-scratch, membrane ruffle formation and real-time PCR.
In vitro scratch-wound assay. Keratinocytes prepared as described above were starved in EMEM.06 (starved medium) for 16 hr. After medium was replaced to and cultured in the starved medium containing 10 μ g/ml of Mitomycin C (Sigma-Aldrich) to prevent cell proliferation 50 , a single wound line was made with scraping cells with 0.2 ml yellow pipette tip. After washing out the detached cells with PBS twice, keratinocytes were stimulated with 50 ng/ml of HGF (a kind gift from Dr. K. Miyazawa, University of Yamanashi) for 48 hr. Images of the cell culture at 0 and 48 hr after scratching were obtained by Biozero BZ-8000 microscope (Keyence). Wound closure was measured by BZ analyzer (Keyence).
Peripheral membrane ruffle formation. Primary cultured keratinocytes prepared from control and K-Arf6-cKO mice as described above were starved for 18 hr, and stimulated with 50 ng/ml of HGF for the indicated times. Cells were then fixed with 4% PFA/PBS at r.t. for 30 min, and stained with Alexa Fluor ® 488 phalloidin (Life Technologies) and DAPI (Molecular probe). Fluorescence images were acquired with Zeiss Axio Observer Z1 equipped with an AxioCam MRm (ZEISS). Length of membrane ruffle and cell periphery length in single cell were measured using ImageJ (NIH), and percentage of membrane ruffle length to cell periphery length was calculated.

Western blotting for Arf6 protein.
Protein levels of Arf6 in keratinocytes prepared from control and K-Arf6-cKO mice were analyzed by western blotting with rabbit polyclonal anti-Arf6 antibody as previously reported 23 . Tubulin in keratinocytes was also detected using anti-tubulin antibody (T6199, Sigma-Aldrich) as an internal standard. Proteins reacted with the antibodies were visualized with Chemi-Lumi One (nacalai tesque), and signals were detected by the luminescent image analyzer LAS-4000 mini (Fujifilm).
Statistical Analysis. All measurements describe mean ± SEM of at least three independent experiments.
Statistical significance was determined using Student's t-test and one-way ANOVA with Tukey's HSD (honest significant difference), Scheffe's post-hoc or Dunnett's multiple comparison tests, and values of P < 0.05 were considered as statistically significant.