Alteration of Skin Wound Healing in Keratinocyte-Specific Mediator Complex Subunit 1 Null Mice

MED1 (Mediator complex subunit 1) is a co-activator of various transcription factors that function in multiple transcriptional pathways. We have already established keratinocyte-specific MED1 null mice (Med1epi−/−) that develop epidermal hyperplasia. Herein, to investigate the function(s) of MED1 in skin wound healing, full-thickness skin wounds were generated in Med1epi−/− and age-matched wild-type mice and the healing process was analyzed. Macroscopic wound closure and the re-epithelialization rate were accelerated in 8-week-old Med1epi−/− mice compared with age-matched wild-type mice. Increased lengths of migrating epithelial tongues and numbers of Ki67-positive cells at the wounded epidermis were observed in 8-week-old Med1epi−/− mice, whereas wound contraction and the area of α-SMA-positive myofibroblasts in the granulation tissue were unaffected. Migration was enhanced in Med1epi−/− keratinocytes compared with wild-type keratinocytes in vitro. Immunoblotting revealed that the expression of follistatin was significantly decreased in Med1epi−/− keratinocytes. Moreover, the mitogen-activated protein kinase pathway was enhanced before and after treatment of Med1epi−/− keratinocytes with activin A in vitro. Cell-cycle analysis showed an increased ratio of S phase cells after activin A treatment of Med1epi−/− keratinocytes compared with wild-type keratinocytes. These findings indicate that the activin-follistatin system is involved in this acceleration of skin wound healing in 8-week-old Med1epi−/− mice. On the other hand, skin wound healing in 6-month-old Med1epi−/− mice was significantly delayed with decreased numbers of Ki67-positive cells at the wounded epidermis as well as BrdU-positive label retaining cells in hair follicles compared with age-matched wild-type mice. These results agree with our previous observation that hair follicle bulge stem cells are reduced in older Med1epi−/− mice, indicating a decreased contribution of hair follicle stem cells to epidermal regeneration after wounding in 6-month-old Med1epi−/− mice. This study sheds light on the novel function of MED1 in keratinocytes and suggests a possible new therapeutic approach for skin wound healing and aging.


Introduction
The wound healing process is divided into three phases: an inflammatory phase, a proliferative phase and a remodeling phase [1,2]. The inflammatory phase occurs immediately after injury. Tissue damage initially causes the disruption of vascular vessels and extravasation, followed by the production of a temporary platelet plug and a fibrin clot which stops bleeding and supplies a transient anchorage for subsequently infiltrating inflammatory cells. Next, during the proliferative phase, which occurs several days after tissue damage, keratinocytes and endothelial cells proliferate and migrate to the wound, resulting in re-epithelialization and angiogenesis. Finally, in the remodeling phase, some fibroblasts are stimulated by macrophages to differentiate into myofibroblasts, causing wound contraction. During this phase, production of the extracellular matrix, including collagen, proteoglycan and fibronectin, is increased, which results in the formation of a mature scar [3,4]. All of these events require the orchestrated efforts of different types of cells. Failure in any of these phases of the wound healing process can lead to chronic wounds, hypertrophic scars and/or wound-related tumor formation [5].
Mediator complex subunit 1 (MED1) is integrated into the Mediator complex as a coactivator of various transcription factors, including nuclear receptors, p53 and BRCA1 [6,7]. MED1 has also been reported to play critical roles in regulating hair cycling and epidermal proliferation [8]. Previously, we established keratinocyte-specific MED1-null (Med1 epi2/2 ) mice and characterized the roles of MED1 in regulating the proliferation of keratinocytes and the maintenance of hair follicle bulge stem cells [9]. In this study, we investigated the process of wound healing in Med1 epi2/2 skin and analyzed the underlying mechanisms, including the activin-follistatin system and epithelial stem cells.

Results
Skin wound healing is accelerated in 8-week-old Med1 epi2/2 mice To study the effect of MED1 depletion in keratinocytes on the skin wound healing process, we created full-thickness circular excisional wounds on the backs of 8-week-old Med1 epi2/2 mice and wild-type (Med1 +/+ ) mice and observed the healing process through days 1 to 7 after injury ( Figure 1A). Macroscopic evaluation revealed that wound closure of Med1 epi2/2 mice was significantly accelerated on day 3 after injury compared with wildtype mice ( Figure 1B, p,0.05). Next, we performed skin biopsies at these wound sites on days 1, 3 and 5 after injury and evaluated the skin wound healing process microscopically ( Figure 1C). Hematoxylin and eosin (H&E) staining of wound sites indicated that re-epithelialization after wounding was significantly enhanced in Med1 epi2/2 mice on days 3 (p,0.01) and 5 (p,0.05) compared with wild-type mice ( Figure 1D).
Migrating epithelial tongues are elongated and the proliferation of keratinocytes is accelerated in 8-week-old Med1 epi2/2 mice To investigate the mechanism(s) underlying the accelerated wound healing in 8-week-old Med1 epi2/2 mice, we next compared the lengths of migrating epithelial tongues and observed a significant elongation in Med1 epi2/2 mice on days 1 (p,0.01) and 3 (p,0.01) after injury ( Figure 2A). Moreover, Ki67 immunostaining in the aforementioned period clearly showed that the number of Ki67-positive keratinocytes was increased at the transitional epidermis and the epithelial tongues were longer in 8-week-old Med1 epi2/2 mice on days 1 (p,0.01) and 3 (p,0.05) after injury compared with those in age-matched wild-type mice ( Figure 2B and C), indicating the acceleration of keratinocyte proliferation by Med1 knockout. We have previously reported that Ki67-positive proliferating keratinocytes in unwounded skin of 8week-old Med1 epi2/2 mice were 1.57 times more frequently observed than in wild-type mice [9]. The number of Ki67-positive proliferating keratinocytes in the wounded epidermis was increased by 2.56-fold in 8-week-old Med1 epi2/2 mice compared with wild-type mice on day 1 after injury ( Figure 2C). These findings suggest that enhanced keratinocyte migration and Figure 1. Skin wound healing is accelerated in 8-week-old Med1 epi2/2 mice. A: Representative macroscopic views of skin wounds on days 1, 3, 5 and 7 after wounding in 8 week old wild-type and Med1 epi2/2 mice. Full-thickness wounds (4 mm in diameter) were made on the middle of the backs of mice to synchronize tension and wound healing was monitored by taking digital photographs. Note the acceleration of wound healing in Med1 epi2/2 mice. B: Evaluation of wound closure by morphometrical analysis of the wound areas. The % of the wound area to the initial area was calculated from the photographs. N = number of mice; n = number of measurements. Bars = means 6 SE. *P,0.05. C: Representative histological view of skin wound healing on day 3. Arrowheads and arrows indicate original wound edges and re-epithelialized leading edges, respectively. Scale bar = 500 mm. D: Time-course of changes of the re-epithelialization ratio after wounding in wild-type and Med1 epi2/2 mice. The % re-epithelialization was calculated by measuring the distance between the leading edges and the width between original wound edges as described in the Materials and Methods. N = number of mice; n = number of sections. Bars = means 6 SE. *P,0.05, **P,0.01. doi:10.1371/journal.pone.0102271.g001 proliferation contribute to the acceleration of skin wound healing in 8-week-old Med1 epi2/2 mice.
As wound contraction also significantly contributes to the wound healing process, the distance between the original wound edges was microscopically measured to precisely evaluate the contraction of wounds in Med1 epi2/2 skin. The original wound edges were determined as the start sites of re-epithelialization. As shown in Figure 2D, there was no significant difference in wound contraction between Med1 epi2/2 and wild-type mice. Further, because myofibroblasts play pivotal roles in granulation and scar formation as well as in wound contraction, we investigated dermal myofibroblasts in the wound sites on days 5 and 7 after wounding ( Figure 2E). Myofibroblasts were identified by staining for a-SMA. a-SMA-positive myofibroblasts were similarly distributed in 8week-old Med1 epi2/2 and wild-type mice ( Figure 2F), suggesting that granulation and scar formation were not affected in 8-weekold Med1 epi2/2 mice.
Follistatin expression is decreased and the MAPK pathway is activated in Med1 epi2/2 keratinocytes in vitro Our previous microarray study comparing gene expression profiles between Med1 epi2/2 and wild-type keratinocytes [9], which is deposited in the GEO repository (http://www.ncbi.nlm. nih.gov/geo) under the accession number GSE35406, revealed that the expression of follistatin is significantly suppressed in Med1 epi2/2 keratinocytes, while the expression of activin, a target of follistatin, as well as activin receptors is not altered. Consistent with this previous data, the expression of follistatin in Med1 epi2/2 keratinocytes was significantly decreased compared with wild-type keratinocytes ( Figure 3A). In the activin-follistatin system crucial for wound repair [10,11], follistatin sequesters and inhibits activin. On the other hand, activin secreted from keratinocytes and fibroblasts during the wound healing process [12,13] activates the MAPK pathway in keratinocytes, influencing their proliferation as well as their migration [13][14][15][16][17]. These facts prompted us to study whether the MAPK pathway is activated in Med1 epi2/2 keratinocytes. The phosphorylation of JNK as well as ERK was enhanced in Med1 epi2/2 -derived keratinocytes ( Figure 3A) compared with wild-type keratinocytes but the phosphorylation of p38 was not apparently enhanced in Med1 epi2/2 keratinocytes (data not shown). These results indicated that endogenous activin secreted from keratinocytes in vitro can robustly activate the MAPK pathway in an autocrine manner in Med1 epi2/2 keratinocytes, where follistatin expression was decreased. Next, to test the effect of endogenous activin secreted from keratinocytes on the migration of Med1 epi2/2 keratinocytes, we performed an in vitro wound healing assay in medium without growth factors ( Figure 3B). Under these conditions, Med1 epi2/2 keratinocytes showed enhanced motility at 24 h, 48 h and 72 h after wounding compared with wild-type keratinocytes (Figure 3C), indicating that endogenous activin secreted from keratinocytes in vitro may activate migration more intensely in Med1 epi2/2 keratinocytes, conceivably because their expression of follistatin was decreased.

JNK phosphorylation in Med1 epi2/2 keratinocytes is augmented by exogenous activin A in vitro
It has been reported that activin A is mainly secreted from dermal fibroblasts and acts on keratinocytes in a paracrine manner, contributing to skin homeostasis, wound healing and hair cycling [13]. Accordingly, several reports have suggested that exogenous activin can enhance the proliferation and migration of keratinocytes by activating the MAPK pathway [15][16][17]. Therefore, we next examined the activation of JNK in Med1 epi2/2 keratinocytes by exogenous activin A. As shown in Figure 3D, activin A caused an immediate and transient JNK phosphorylation, which was detectable at 10 min after the treatment and was reduced to the basal level at 30 min in Med1 epi2/2 and in wildtype keratinocytes. The peak level as well as the basal level of phosphorylation of JNK in Med1 epi2/2 keratinocytes was Keratinocytes were cultured in KBM to form confluent monolayers and then were serum deprived for 24 h. The cells were subsequently incubated with mitomycin C (0.5 mg/ml) for 2 h and were then scratched with a p200 pipette tip, followed by incubation in KBM for 72 h. The cells were analyzed by phase contrast microscopy and were photographed at the indicated time points. C: Quantification of the number of migrating keratinocytes. The number of cells which had migrated into the wounded space at the indicated time points was counted microscopically and related to the wounded area. Bars = means 6 SE. *P,0.05, **P,0.01. D: Western blot analysis of JNK phosphorylation after treatment with activin A (5 ng/ mL). The level of JNK phosphorylation reached the highest value at 10 min after activin A treatment in wild-type and in Med1 epi2/2 keratinocytes cultured in KBM. Note that both the peak and the basal level of JNK phosphorylation were enhanced in MED1 epi2/2 keratinocytes compared with wild-type keratinocytes. E: Quantification of JNK phosphorylation after administration of activin A in Med1 epi2/2 and wild-type keratinocytes (n = 3). Bars = means 6 SE. *P,0.05, **P,0.01. F: Cell cycle analysis of Med1 epi2/2 and wild-type keratinocytes. Cultured in KBM with acitivin A (5 ng/mL), Med1 epi2/2 keratinocytes showed an increased ratio of S phase cells compared with wild-type keratinocytes (left) while no difference was observed when those cells were cultured in KGM (right). Bars = means 6 SE. **P,0.01, N.S., not significant. doi:10.1371/journal.pone.0102271.g003 augmented compared with wild-type keratinocytes ( Figure 3E), suggesting that exogenous as well as endogenous activin A enhances JNK phosphorylation in Med1 epi2/2 keratinocytes because of their decreased expression of follistatin.

Exogenous activin A increases the percentage of Med1 epi2/2 keratinocytes in S-phase
We next asked if exogenous activin A also influences the cell cycle of Med1 epi2/2 keratinocytes. To optimize the readout, the cells were cultured in KBM and starved for 24 h, were subsequently treated with activin A for 24 h and then were finally subjected to cell cycle analysis. The results showed that the Sphase percentage of activin A-treated Med1 epi2/2 keratinocytes cultured in KBM was significantly higher than activin A-treated wild-type keratinocytes ( Figure 3F, left, p,0.01). On the other hand, cell cycle phases were similar in activin A-treated Med1 epi2/ 2 and wild-type keratinocytes when cultured in keratinocyte growth medium (KGM) containing numerous growth promoters ( Figure 3F, right). This is probably because the excess growth promoters in the KGM masked the endogenous activin A effect.
Together these data suggest that the wound healing acceleration in 8-week-old Med1 epi2/2 mice could be ascribed to the alteration of follistatin-activin balance in the wound sites, which activates MAPK signaling and keratinocyte proliferation and migration.
Skin wound healing in old Med1 epi2/2 mice is delayed Next, to assess the skin wound healing process in older Med1 epi2/2 mice, we performed wound healing assays in 6month-old wild-type and Med1 epi2/2 mice. As demonstrated in Figure 4A and 4B, the wound healing process was significantly delayed on days 1 (p,0.01), 3 (p,0.05) and 5 (p,0.01) after injury in 6-month-old Med1 epi2/2 mice, compared with wild-type mice. In line with this observation, the re-epithelialization ratio was significantly decreased on days 1 (p,0.05) and 3 (p,0.01) after injury in 6-month-old Med1 epi2/2 mice ( Figure 4C). The lengths of migrating epithelial tongues were correspondingly decreased on days 1 (p,0.05) and 3 (p,0.05) after injury in 6month-old Med1 epi2/2 mice ( Figure 4D), while no significant . Skin wound healing is delayed in 6 month old Med1 epi2/2 mice. A: Representative macroscopic views of skin wounds on days 1, 3, 5 and 7 after wounding in 6-month-old wild-type and Med1 epi2/2 mice. Full-thickness wounds (4 mm in diameter) were made on the middle of the back skins of mice and wound healing was monitored by taking digital photographs. Note that wound healing in 6-month-old Med1 epi2/2 mice was significantly delayed compared with the age-matched wild-type mice. B: Evaluation of wound closure by morphometrical analysis of the wound areas in 6-month-old wild-type and Med1 epi2/2 mice. The % of the wound area to the initial area was calculated from the photographs. N = number of mice; n = number of measurements. Bars = means 6 SE. **P,0.01. C: Re-epithelialization ratio on days 1, 3 and 5 after wounding in 6-month-old wild-type and Med1 epi2/2 mice. The % re-epithelialization was calculated as mentioned above. N = number of mice; n = number of sections. Bars = means 6 SE. *P,0.05, **P,0.01. D: The lengths of migrating epithelial tongues were measured on days 1, 3 and 5 after injury in 6-month-old wild-type and Med1 epi2/2 mice. N = number of mice; n = number of measurements. Bars = means 6 SE. *P,0.05, **P,0.01. E: Evaluation of distance between original wound edges in 6-month-old wild-type and Med1 epi2/2 mice on days 1, 3 and 5 after injury. N = number of mice; n = number of sections. Bars = means 6 SE. N.S., not significant. doi:10.1371/journal.pone.0102271.g004 difference was observed in the wound contraction between 6month-old wild-type and Med1 epi2/2 mice ( Figure 4E). Compared with age-matched wild-type mice, Ki67-positive proliferating keratinocytes were decreased in 6-month-old Med1 epi2/2 mice per area of transitional epidermis and the epithelial tongue in wound sites on days 1 (p,0.01), 3 (p,0.01) and 5 (p,0.01) after the injury (Figure 5A and B). Immunohistochemical staining for follistatin revealed no difference in the expression of follistatin between 8-week-old and 6-month-old Med1 epi2/2 mice (data not shown), suggesting that a mechanism(s) other than the activinfollistatin system could influence this impediment of wound healing in 6-month-old Med1 epi2/2 mice.
There is less contribution of BrdU-positive label retaining cells in hair follicles to cutaneous wound healing in old Med1 epi2/2 mice It has been reported that, after epidermal injury, hair follicle stem cells give rise to short-lived transient amplifying cells, which migrate into the wound epithelium and promote the epidermal regeneration [18][19][20][21][22][23][24][25][26][27]. In our previous study, CD34-positive and keratin 15-positive hair follicle bulge stem cells decreased in Med1 epi2/2 mice after several months of age, resulting in sparse hair in older Med1 epi2/2 mice [9]. Therefore, we hypothesized that the delay of skin wound healing in the older Med1 epi2/2 mice can be attributed to the possible reduction of hair follicle stem cells. To investigate the contribution of hair follicle stem cells to skin wound healing in old Med1 epi2/2 mice, we performed a BrdU pulse-labeling experiment in 6-month-old Med1 epi2/2 mice and age-matched wild-type mice, 2 months before the wound creation. The analysis of BrdU label retaining cells in hair follicles adjacent to the wounds demonstrated that the number of BrdU-positive follicular slow-cycling cells was decreased in 6-month-old Me-d1 epi2/2 mice compared with age-matched wild-type mice ( Figure 5C, p,0.01), indicating a significant depletion of hair follicle bulge stem cells in 6-month-old Med1 epi2/2 mice, which corresponds to our previous study [9]. Furthermore, BrdU-positive label retaining cells that had migrated into the epidermis adjacent to the wounds were detected in hair follicles of 6-month-old wildtype mice, while no such migrating cells were detected in 6-monthold Med1 epi2/2 mice ( Figure 5C). This observation suggested that there is a distinct contribution of hair follicle bulge stem cells to the epidermal regenerative process in 6-month-old wild-type mice, but not in 6-month-old Med1 epi2/2 mice. On the other hand, the number of label retaining cells in hair follicles in 8-week-old Med1 epi2/2 mice was comparable with age-matched wild-type mice ( Figure 5D).

Discussion
In the present study, we investigated the effects of MED1 depletion in the epidermis on cutaneous wound healing in Med1 epi2/2 mice. Our results provide the first evidence that cutaneous wound healing is accelerated in 8-week-old Med1 epi2/2 mice compared with age-matched wild-type mice (Figure 1). The 8-week-old Med1 epi2/2 mice demonstrated a rapid re-epithelialization due to enhanced epidermal proliferation as well as migration in the wound sites but that was not due to the wound contraction ( Figure 2). Although MED1 is known to function as a co-activator of nuclear receptors, such as PPAR, RXR and VDR, there has been no report demonstrating the accelerated wound healing phenotype in PPAR-KO mice, RXR-KO mice or VDR-KO mice [28][29][30][31]. Interestingly, Med1 epi2/2 keratinocytes show a significantly decreased expression of follistatin, a potent inhibitor of activin, with significantly increased MAPK activity compared with wild-type keratinocytes ( Figure 3).
Activins, members of the TGF-b superfamily, are disulfidelinked dimeric proteins comprised of two b subunits. Three different forms of activin, homodimeric activin A (bAbA), homodimeric activin B (bBbB) and heterodimeric activin AB (bAbB), have been identified. Activins bind to heteromeric complexes of transmembrane receptor serine/threonine kinases, type I (ACVR1, 1B and 1C) and type II (ACVR2A and 2B) activin receptors [13], mediating their biological roles including the regulation of proliferation, differentiation, apoptosis, metabolism, homeostasis, immune function, endocrine function and wound repair in many tissues [32]. Follistatins, antagonists of activins, are soluble extracellular proteins consisting of varying molecular weight isoforms due to alternative splicing at the 39 end of the mRNA [13,[33][34][35]. The most common isoforms of follistatin consist of 288 and 315 amino acids (FS288 and FS315, respectively). Follistatins have a higher affinity to the activin b subunits than the activin receptors [36][37][38] and inhibit the action of activins by two distinct mechanisms, as follows: 1) Membranebound follistatin FS288 has a high affinity to cell surface bound heparin sulfate, which causes the follistatin/activin complex to be internalized and subjected to lysosomal degradation. 2) The circulatory form of follistatin FS315, which contains a C-terminal acid tail, binds to activin and prevents binding to its receptors [13]. Although the precise distribution of the components of activin signaling in normal skin is uncertain, it is likely that the activinbA subunit is expressed in dermal cells while the activinbB subunit is expressed in proliferating keratinocytes at the wound edge and in the migrating epithelial tongue after injury [12]. On the other hand, follistatin mRNA is expressed mainly in the dermis and at low levels in the epidermis [12,13].
Several studies using transgenic and knockout mice have clearly suggested the critical involvement of activins and follistatins during cutaneous wound healing. It has been reported that transgenic mice over-expressing the activinbA chain in keratinocytes showed an acceleration of the skin wound healing process with increased keratinocyte proliferation, hyperthickening of the tongue epithelium and excessive scar formation after skin injury [39,40]. Additionally, it has been reported that mice without follistatin expression in keratinocytes (Fst null mice) show enhanced keratinocyte proliferation in the tail epidermis resulting in a thicker epithelium at the wound edge without excessive scarring after skin injury [41].
Previously it was reported that activins mediate wound repair after injury through the MAPK signaling pathway [15][16][17]. It has been reported that the blockade of JNK signaling by a JNKspecific inhibitor significantly suppresses keratinocyte proliferation at the wound site and subsequently delays wound closure [15,16]. In our model, Med1 epi2/2 keratinocytes exhibit decreased follistatin expression and an increased activity of the MAPK pathway with or without the existence of exogenous activin A in vitro. Moreover, migration is enhanced in Med1 epi2/2 keratinocytes in vitro without exogenous activin A, while exogenous activin A elicits an increase in the percentage of Med1 epi2/2 keratinocytes in S-phase (Figure 3). After injury, 8-week-old Med1 epi2/2 mice show accelerated cutaneous wound healing without excessive granulation tissue formation (Figure 2). These findings suggest that in Med1 epi2/2 mice, the loss of follistatin expression in keratinocytes enhances the biological activity of activin secreted from keratinocytes and/or dermal fibroblasts and thus constitutively activates the MAPK signaling pathway in the epidermis, resulting in rapid wound healing just like in Fst null mice.
Several reports have suggested that MED1 and nuclear receptors are involved in regulating follistatin expression in various tissues. Necela et al. reported that activation of PPARc downregulates the expression of follistatin mRNA through dimerization with RXR in intestinal epithelial cells [42]. Matsumoto et al. Figure 6. Proposed model of altered cutaneous wound healing in Med1 epi2/2 mice. In Med1 epi2/2 keratinocytes, the expression of follistatin is decreased. Consequently, activin, which is not sequestered by follistatin, is increased and activates the MAPK signaling pathway in keratinocytes in endocrine and/or paracrine manners. As a result, the proliferation and migration of Med1 epi2/2 keratinocytes are enhanced, contributing to rapid cutaneous wound healing process. However, as the number of hair follicle bulge stem cells decreases in old Med1 epi2/2 mice, the promoting effect of follistatin down-regulation on epidermal regeneration is overcome, resulting in impaired cutaneous wound healing. doi:10.1371/journal.pone.0102271.g006 reported that follistatin shows higher levels of expression in normal livers after partial hepatectomy but not in MED1-deficient livers, using transgenic mice [43]. Therefore, it is conceivable that MED1 depletion in keratinocytes has a direct and/or indirect effect on the expression of follistatin in our model, although the details are yet to be elucidated.
Accumulating evidence indicates that there are multiple populations of epithelial stem cells locating in different parts of the epidermis [19,20,23,24,[44][45][46][47][48][49][50][51][52][53][54][55][56][57]. They maintain normal skin homeostasis by regenerating the distinct epithelial cell lineages in the distinct parts of the epidermis as well as contribute to wound healing upon injury by recruiting undifferentiated progenitor cells to the wounded epidermis [21,[23][24][25][26]. Hair follicle stem cells do not normally contribute to epidermal homeostasis. However, after epidermal injury, hair follicle stem cells give rise to short-lived transient amplifying cells which are recruited into the wounded epidermis, facilitating the epidermal regeneration [27]. Hair follicles contain several populations of epithelial stem cells characterized by distinct expression patterns of stem cell markers, including CD34 and keratin 15 [19,20]. Previously, we reported that the numbers of hair follicle stem cells which express CD34 and keratin 15 are reduced in Med1 epi2/2 mice from a few months to one year after birth, which suggests that MED1 plays a distinct role in the maintenance of hair follicle stem cells [9]. Correspondingly, skin wound healing in 6-month-old Med1 epi2/2 mice is significantly delayed with decreased numbers of Ki67-positive proliferating keratinocytes compared with age-matched wild-type mice (Figures 4, 5). The analysis of BrdU-positive label retaining cells further showed decreased numbers of hair follicle bulge stem cells migrating into the epidermis adjacent to the wound sites in 6month-old Med1 epi2/2 mice, while no apparent change in 8-weekold Med1 epi2/2 mice, compared with age-matched wild-type mice ( Figure 5). These findings indicate an impaired skin wound healing process due to the lack of CD34-positive and/or keratin 15positive epithelial stem cells, which counteracts the positive effect of follistatin down-regulation on the wound healing in 6-monthold Med1 epi2/2 mice (shown schematically in Figure 6). To our knowledge, such a phenotype, in which cutaneous wound healing is accelerated in adolescence and is retarded in the elderly due to depletion of hair follicle stem cells, has not been previously reported and therefore is specific.
Although the precise mechanism by which MED1 depletion participates in the activin-follistatin system in keratinocytes and in hair follicle stem cell maintenance remains unclear, and the possibility that other factors involved in the wound healing process of Med1 epi2/2 skin can not be excluded, our findings shed light on a novel function of MED1 and offer possible new therapeutic approaches to target MED1 in the epidermis for cutaneous wound healing and aging.

Animals
The generation of Med1 epi2/2 mice, in which Med1 is disrupted under control of the keratin 5 promoter, was described elsewhere [9]. All animal studies were performed according to protocols approved by the Institutional Animal Care and Use Committee at Osaka University. Mice that were used for wound healing study were housed appropriately as previously described [58]. Briefly, Mice were raised under light/dark (12-h/12-h) cycles and fed adlibitum amount of standard chow and water according to the Institutional Animal Care and Use Committee at Osaka University. Mice were observed daily by the investigators and treated appropriately when displayed any signs of discomfort or illness by the facility veterinary.

Wound Creation and Macroscopic Examination
Full-thickness wounds were made using a sterile biopsy punch with a diameter of 4 mm (NIPRO, Osaka, Japan) on the middle dorsal shaved telogen skin of Med1 epi2/2 mice and wild-type (Med1 +/+ ) littermates at either 8 weeks or 6 months of age. Mice were administered sodium pentobarbital with or without sevoflurane anesthesia before wounding. The wounds were left uncovered and the animals were housed in separate cages. Wound healing was macroscopically monitored by digital photography at the indicated time points. The wound areas (percentage of wound area relative to the original wound) were calculated using the following formula: Relative open wound area (%) = [Open area on the indicated time point/Original wound area]6100.

Immunohistochemistry
Immunohistochemical staining was performed as previously described [9]. In brief, 5 mm thick paraffin sections were deparaffinized and autoclaved in 10 mM sodium citrate (pH 6.0) for 15 min at 121uC to retrieve epitope structures. After washing in TBS-T (Tris-buffered saline with 1% Tween 20), the sections were treated with H 2 O 2 and endogenous peroxidase activity was blocked. Specimens were then blocked with Protein Block Serum-Free (Dako, Glostrup, Denmark), incubated with rabbit polyclonal anti-Ki67 IgG (1:500; Leica Microsystems, Buffalo Grove, IL), and mouse monoclonal anti-a-SMA IgG (1:100; Dako) overnight at 4uC followed by incubation and visualization with a ChemMate ENVISION/HRP kit (Dako). Immunohistochemical staining for BrdU was performed using a BrdU In-Situ Detection kit (BD Bioscience, New Jersey, US) according to the manufacturer's protocol.

Analysis of Re-Epithelialization and Wound Contraction
The width of each wound and the distance of the traversed epithelium were measured in H&E-stained sections at the indicated time points. The percentage of re-epithelialization was calculated according to the following formula: [distance of the minor axis covered by epithelium]/[distance of the minor axis between original wound edges]6100. The original wound edges were determined as the start sites of re-epithelialization (See Figure 1C). Wound contraction was estimated by measuring the distance of the minor axis between the original wound edges.

Analysis of Cell Proliferation and Granulation Tissue
Ki67-positive cells were counted in the transitional epidermis and the epithelial tongue [2] of wounds and were related to the area of the same part of epidermis. The area of wound epidermis was determined using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Myofibroblasts were identified by immunostaining of a-SMA in the granulation tissue and the stained area was determined by planimetric image analysis using ImageJ software.

Isolation and Culture of Keratinocytes
Isolation and culture of keratinocytes was performed as previously described [9]. In brief, skins of newborn mice were derived after the mice had been sacrificed with excess anesthesia. Derived newborn mice skins were then treated with dispase and trypsin to separate the epidermis from the dermis. Isolated keratinocytes were then seeded on type I collagen coated dishes, and were cultured in CnT07 conditioned culture medium (KGM, CELLnTEC, Bern, Switzerland). For each experiment, keratinocytes were used as a primary culture or after one passage.

Keratinocyte migration assay
For a wound healing assay, keratinocytes derived from skins of newborn Med1 epi2/2 mice and wild-type (Med1 +/+ ) littermates were cultured in KBM and allowed to form confluent monolayers. After serum starvation for 24 h, keratinocytes were incubated with an S-phase cell cycle blocker mitomycin C (0.5 mg/ml) for 2 h. After straight scratch wounds were made with a p200 pipette tip, the suspended cells were removed washing with PBS and then incubated in KBM for 72 h. The number of cells which migrated into each wounded space was counted microscopically at the noted time intervals and related to the wounded area.

Cell cycle analysis
For cell cycle analysis, keratinocytes derived from skins of newborn Med1 epi2/2 mice and wild-type (Med1 +/+ ) littermates were seeded (2.5610 5 ) and cultured in KBM for 24 h to synchronize the cell cycles. Then, keratinocytes were treated with activin A (5 ng/mL) and BrdU (10 mM) and cultured in KBM for 24 h and harvested. The cell cycle of keratinocytes was analyzed by FACS CantoII(BD Biosciences) using a BD Pharmingen BrdU Flow kit (BD Biosciences) according to the manufacturer's protocol.

BrdU labeling procedures
BrdU labeling of slow-cycling cells was performed as previously described [59]. Briefly, for BrdU labeling in 6-month-old mice, 4month-old Med1 epi2/2 mice and wild-type (Med1 +/+ ) littermates were intraperitoneally injected with BrdU (50 mg per g body weight) twice daily for 5 d and then conventional club hair plucking was performed. Eight weeks after plucking, wound creation and subsequent skin biopsy were performed and followed by immunohistochemistry for BrdU as described above. The number of BrdU-positive cells was counted in hair follicle sections with bulge region in 6-month-old Med1 epi2/2 mice and agematched wild-type mice. For BrdU labeling in 8-week-old mice, neonatal mice were subcutaneously injected with BrdU (50 mg per g body weight) twice daily for 3 d from the third day after birth. After eight weeks, conventional club hair plucking and subsequent skin biopsy were performed and followed by immunohistochemistry for BrdU. The number of BrdU-positive cells was counted in hair follicle sections with bulge region in 8-week-old Med1 epi2/2 mice and age-matched wild-type mice.

Statistical Analyses
An unpaired t-test was used to determine statistical significance when the values were normally distributed. An F-test was used to test if the variances are equal. When variances were significantly different according to the F-test, an unpaired t-test with Welch's correction was used.