JMJD3 and NF-κB-dependent activation of Notch1 gene is required for keratinocyte migration during skin wound healing

It has been shown that epigenetic regulation plays an important role in skin wound healing. We previously found that histone H3K27me3 demethylase JMJD3 regulates inflammation and cell migration in keratinocyte wound healing. In this study, we identified Notch1 as a direct target of JMJD3 and NF-κB in wounded keratinocytes using in vitro cell and in vivo animal models. We found that Notch1 is up-regulated in the wound edge and its expression is dependent on JMJD3 and NF-κB in wounded keratinocytes. We also found that Notch1 activates the expression of RhoU and PLAU gene, which are critical regulators of cell migration. Consistently, depletion or inactivation of Notch1 resulted in decreased filopodia formation, increased focal adhesion and actin stress fiber, leading to reduced keratinocyte migration and skin wound healing. Thus, our findings provide the molecular mechanism involving JMJD3/NF-κB-Notch pathway in keratinocyte wound healing.

. Up-regulation of Notch1 expression in scratch-wounded keratinocytes. (a) After HaCaT keratinocytes were scratched, transcripts of Notch1~4 and GAPDH were determined by quantitative PCR. (b) Lysates from scratch-wounded HaCaT keratinocytes were immunoblotted with anti-NICD and antiβ-ACTIN antibodies, respectively. (c) Increased nuclear NICD expression in the region of the wound edge compared to the region far from the wound edge (outward) (>1 mm). HaCaT keratinocytes after scratching were immunostained with anti-NICD antibody. Nuclei were identified using DAPI staining. Scale bar, 25 μm. (d) Up-regulation of Hes1 promoter reporter activity after scratching. However, activity of promoter reporter containing mutated RBP-J binding sites is not up-regulated by scratching.
in epidermal differentiation, proliferation, and inflammation 12,15,[28][29][30] , we further examined expression of NICD (Notch intracellular domain) of Notch1 at protein level. Similarly, Western blot analysis showed the acute up-regulation of NICD expression (Fig. 1b). We also observed increased nuclear expression of NICD in the region of the leading wound edge, while no elevated expression were detected in the region far from the wound edge as well as in confluent monolayer HaCaT keratinocytes (Figs 1c and S2). In addition, the promoter activity of Hes1 gene, a direct target of Notch, was up-regulated acutely at 2~6 hr after scratching (Fig. 1d). Mutation of RBP-J binding sites on the Hes1 gene promoter abolished the scratch-induced activation of the promoter activity.
JMJD3 activates expression of Notch1 gene in wounded keratinocytes. Since the spatiotemporal expression pattern of Notch1 is similar with that of JMJD3 in the early phase of wounded HaCaT keratinocytes 23 , we next tested whether JMJD3 regulates the Notch1 gene expression. When the transcript of JMJD3 was depleted by RNA interference, the expression of Notch1 decreased in the scratch-wounded keratinocytes at RNA and protein level (Fig. 2a). Decreased expression of nuclear NICD was observed in the region of the leading wound edge (Supplementary Fig. S3a). We also utilized GSK-J4 to inhibit the histone demethylase activity of JMJD3 31 . (c,d) Following depletion (sip65) or inactivation (BAY11-7085) of NF-κB, HaCaT keratinocytes were scratched and expression level was determined by quantitative PCR and Western blotting. (e) ChIP assay was performed in scratch-wounded HaCaT keratinocytes. The occupancy of each protein was determined with quantitative PCR at the Notch1 gene encompassing the NF-κB binding sites. ChIP using normal IgG was performed as a negative control. (f,g) The methylation level of H3K27 was measured by ChIP assay in scratch-wounded HaCaT keratinocytes treated with GSK-J4 or BAY11-7085.
GSK-J4 treatment reduced Notch1 expression (Figs 2b and S3b). JMJD3 has been shown to interact with NF-κB and up-regulate gene expression of MMP, inflammatory, and growth factor during keratinocytes wound healing 23 . Thus, we also tested the dependency of NF-κB for the Notch1 gene activation. Depletion of NF-κB p65 suppressed Notch1 gene activation and nuclear expression of NICD in wounded keratinocytes (Figs 2c and S3c). Consistently, BAY11-7085, which inhibits IκBα phosphorylation, showed same effects of NF-κB p65 depletion (Figs 2d and S3d). We also found suppression of promoter activity of Notch1 gene by depletion or inactivation of NF-κB p65 ( Supplementary Fig. S4).
To investigate whether JMJD3 and NF-κB directly activates Notch1 gene expression, we identified NF-κB binding sites on the upstream region of Notch1 gene ( Supplementary Fig. S7a) 32 . The recruitment of JMJD3 and NF-κB p65 to the upstream region of Notch1 gene was investigated in HaCaT keratinocytes after scratching by a ChIP assay. Chromatin from HaCaT keratinocytes at 0 and 2 hr after scratching were immunoprecipitated. Quantitative PCR showed that JMJD3, p50, and p65 are significantly bound to the NF-κB binding sites of the Notch1 gene (Fig. 2e). Consistent with the recruitment of JMJD3, we detected a decreased level of tri-methylated H3K27 (H3K27me3) as well as an increased level of mono-methylated H3K27 (H3K27me1) at the promoter (Fig. 2e). Inactivation of JMJD3 or NF-κB consistently showed the maintenance of the tri-methylation level of H3K27 in the upstream region of Notch1 gene (Fig. 2f,g). Notch1 regulates keratinocyte wound healing via cell migration. We investigated the effect of Notch1 depletion or inactivation on the keratinocytes migration in response to scratching. As shown in Fig. 3, wound closure was significantly delayed in Notch1-depleted HaCaT keratinocytes as compared to control siRNA-transfected keratinocytes (Fig. 3a). Transwell migration assay also showed significantly fewer migrating keratinocytes in Notch1-depleted keratinocytes (Fig. 3b). Inhibition of Notch signaling by DAPT, a γ-secretase inhibitor, resulted in delayed wound closure and suppressed keratinocyte migration (Fig. 3c,d). In contrast, over-expression of NICD in HaCaT keratinocytes accelerated wound closure as compare to control empty vector-transfected keratinocytes ( Supplementary Fig. S5).
NF-κB p65 depletion or inhibition showed the same effect on the RhoU and PLAU gene expression (Fig. 4c).
RhoU has been shown to regulate cell migration through formation of filopodia, focal adhesion, and dissolution of stress fiber in diverse cell types [40][41][42] . In addition, PLAU converts plasminogen to plasmin, which regulates cell adhesion and migration though cleavage of ECM and MMPs 43 . Therefore, we investigated whether Notch regulates formation of filopodia, focal adhesion, and actin stress fibers in the scratch-wounded HaCaT keratinocytes using cell staining with anti-vinculin antibody and phalloidin. As shown in Fig. 4d, we observed that control HaCaT keratinocytes show well developed filopodia-like protrusion, which is stained by phalloidin, at the leading edge. However, fewer filopodia were observed in DAPT, GSK-J4, and BAY11-7085-treated keratinocytes, respectively (Fig. 4d). In addition, DAPT, GSK-J4, and BAY11-7085 treatment significantly increased the number of focal adhesion stained with anti-vinculin antibody (Fig. 4d). Similarly, increased bundles of stress actin filaments were observed in DAPT, GSK-J4, and BAY11-7085-treated keratinocytes compared with control HaCaT keratinocytes after scratching (Fig. 4d). We found consistently that depletion of RhoU or PLAU results in delayed wound closure as compared to control siRNA-transfected HaCaT keratinocytes ( Supplementary Fig. S6).

Recruitment of NICD to the RhoU and PLAU gene promoters in wounded keratinocytes. Since
our results demonstrated Notch1-dependent RhoU and PLAU gene activations, we next examined the recruitment of NICD to the genes in HaCaT keratinocytes after scratching by ChIP assay. Chromatin from HaCaT keratinocytes at 0~12 hr after scratching was immunoprecipitated with anti-NICD, and anti-acetylated histone H3, respectively. Quantitative PCR was performed to determine the occupancy of proteins using oligonucleotide primers encompassing RBP-J binding site ( Supplementary Fig. S7b). Upon scratch wound, NICD significantly bound to the RBP-J binding site of RhoU and PLAU genes (Fig. 5a). We also detected an increased level of histone H3 acetylation (Fig. 5a). As expected, DAPT treatment resulted in a reduced level of NICD and acetylated H3 (Fig. 5b). In addition, the inactivation of JMJD3 and NF-κB also showed the same effects of DAPT on the level of NICD and H3 acetylation at the genes (Fig. 5c,d).
Effect of JMJD3 and NF-κB on the Notch in skin wound healing. We further investigated the dependency of JMJD3 and NF-κB in Notch1-mediated wound healing using an in vivo mouse skin wound model. We first determined the expression of Notch1~4, RhoU, and PLAU in the skin after wounding. Quantitative PCR analysis revealed that the expression of the Notch1 and 4 genes increased at 1~2 days after scratching and then decreased slowly (Figs 6a and S8a). Consistent to scratch-wounded HaCaT keratinocytes, Notch1 is expressed more abundantly than Notch4 in skin wounds ( Supplementary Fig. S8b). The expression pattern of RhoU and PLAU genes is correlated with that of Notch1 (Fig. 6a). Immunohistochemical results further showed that NICD is up-regulated and localized in the nucleus of wound edge keratinocytes, which are immunostained with anti-keratin14 (KRT14) antibody, at 1~7 days after wounding ( Supplementary Fig. S9a).
To evaluate the necessity of JMJD3 and NF-κB for the up-regulation of Notch1 expression in skin wound healing, JMJD3 or NF-κB p65 siRNA was applied topically to the skin wound. JMJD3 or p65 depletion attenuated the up-regulation of Notch1, RhoU, and PLAU expressions in wounded skin (Fig. 6b,c). We further found a decreased nuclear NICD expression in the wound edge after wounding ( Supplementary Fig. S9b,c). Notch1 depletion also resulted in decreased RhoU and PLAU gene activations (Fig. 6d). Finally, we investigated the effects of Notch1 on skin wound healing processes. Although the wound was nearly closed by 7 days after wounding in a control siRNA or DMSO-applied skin wound, we found a significantly delayed wound closure in Notch1 siRNA or DAPT-applied skin wound (Fig. 6e,f).

Discussion
Notch pathway plays an important role in epidermal development in embryo and epidermal homeostasis in adult [8][9][10][11] . Notch family genes and its ligands are expressed in various epidermal layers during development [12][13][14][15][16] . Notch1 activates expression of keratin 1 and involucrin genes, which are early keratinocyte differentiation markers 44 . Deficiency of Notch1 further showed epidermal hyperplasia by reducing p21WAF/Cip1 expression in keratinocytes 44 . Notch signaling also regulates the proliferation and differentiation of keratinocyte stem cells located in basal layer of the epidermis 45 . However, a few reports have been reported the role of Notch signaling in keratinocyte wound healing.
In this study, we found that JMJD3 and NF-κB activate Notch1 expression at the wound edge in wounded keratinocytes. Although the expression pattern of Notch 4 was similar to that of JMJD3, Notch1 is expressed more abundantly than Notch4 in wounded keratinocytes and skin (Figs 1A and S1 and S6). Similarly, Notch1 has been considered as a primary Notch receptor in epidermal differentiation, proliferation, and inflammation 12,15,[28][29][30] . Notch1 and its ligand Jagged1 are up-regulated at the wound edge in grafted human skin on the nude mouse 16 . Consistent to our findings, Notch1 is up-regulated in the epidermis including suprabasal and basal laminal keratinocytes at 1~4 days after wounding in the mouse skin 30,46 . Up-regulated Notch1 activates TNFα expression, which can induce CCL20 and CXCL13 chemokine expression, leading to the recruitment of RORγ + group3 innate lymphoid cells in wounded skin 30 . The inhibition of Notch by antisense RNA or γ-secretase inhibitor DAPT showed impaired skin wound healing due to inhibition of leukocyte infiltration, angiogenesis, and keratinocyte migration 47,48 . DAPT-treated wound also showed incomplete wound closure 46,49 .
Similar spatiotemporal expression of Notch1 and JMJD3 led us to investigate whether JMJD3 is required for Notch1 gene activation in wounded keratinocytes. In fact, depletion or inhibition of JMJD3 and NF-κB attenuated up-regulation of Notch1 gene expression. Consistently, JMJD3 and NF-κB were recruited to NF-κB binding sites on the Notch1 gene upon wounding as well as demethylation of H3K27me3. Our results are supported by requirement of NF-κB for the Notch1 gene activation 32,50 .
At the wound edge, keratinocytes begin to rearrange their cytoplasmic and membrane structures, leading to disassembly of cell-extracellular matrix and cell-cell interaction for cell migration. These events include the actin cytoskeleton-mediated regulation of lamellipodia, filopodia, and focal adhesions during the early phase of wound healing 51,52 . We found that depletion or inhibition of Notch1 results in decreased filopodia formation, increased focal adhesion and actin stress fibers, leading to delayed keratinocyte migration and wound closure. Our results further demonstrated that Notch1 directly up-regulates the expression of RhoU and PLAU gene, which are important regulators for cell migration (see below). Consistently, RhoU and PLAU have been identified as the target gene of Notch signaling 33,35,37 . RhoU is considered atypical Rho GTPase because it is likely to be predominately GTP-bound 53 . In fact, RhoU activates cell migration through formation of filopodia, loss of focal adhesion, and dissolution of stress fiber in several cell types [40][41][42][53][54][55][56] . In particular, RhoU prominently localizes to focal adhesion, as indicated by vinculin expression 40 . Depletion of RhoU results in increased number of focal adhesion and decreased cell migration in response to wounding 40 . Filopodia formation by RhoU requires interaction with Pyk2 non-receptor tyrosine kinase in the presence of Src 41 . Pyk2 plays important role in cell adhesion and focal adhesion assembly through regulation of cytoskeleton dynamics [57][58][59] . PLAU converts plasminogen to plasmin, which cleaves ECM and MMPs for cell adhesion and migration 43 . Interestingly, PLAU is shown to be expressed in leading edge of migration keratinocytes [60][61][62][63] . In addition, PLAU colocalizes with vinculin at focal adhesion sites in various cell types [64][65][66] . Stimulation of uPA by Endo180, which is 180-kD transmembrane glycoprotein and complex with uPA and uPAR, leads to increased filopodia production in breast cancer cells 67 . Inhibition of uPA further decreases cell migration [68][69][70][71][72] . Interestingly, the expression and activity of MMPs are regulated transcriptionally and post-transcriptionally in wounded keratinocytes since we identified MMP-1, MMP-2, MMP-3, MMP-9, MMP-13, and MMP-14 as direct target of JMJD3 and NF-κB 23 and PLAU-converted plasmin is shown to activate these MMPs [73][74][75][76][77][78][79][80][81][82][83] .
In conclusion, our study supports the important role of Notch1 as well as JMJD3 and NF-κB in keratinocytes wound healing. Specifically, JMJD3 and NF-κB-mediated Notch1 activation is required for the regulation of focal adhesion, filopodia, and stress fiber formation though RhoU and PLAU gene expression, leading to keratinocyte migration during the early phase of skin wound healing (Fig. 6g). Given that some genes related to Notch signaling are down-regulated in keratinocyte of venous ulcers and activation of Notch signaling in epidermal stem cells accelerate diabetic wound healing 84, 85 , our study may provide a new therapeutic intervention for the chronic skin wound.

Materials and Methods
Cell culture, animals, and chemicals. Human HaCaT keratinocytes were maintained in DMEM supplemented with 10% fetal bovine serum and antibiotics. Adult male ICR mice (8 weeks old; Samtako) were held in a temperature-controlled room (22 °C) at 55% humidity. The Committee for Experimental Animal Research at Sogang University approved the animal experiments and all methods were performed in accordance with guidelines and regulations relevant to the study. HaCaT keratinocytes were treated with 30 μM GSK-J4 (Tocris) and 10 μM BAY 11-7085 (Cayman) to inhibit JMJD3 and NF-κB, respectively. To inactivate Notch signaling, 10 μM DAPT (N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester, Sigma-Aldrich) was used as a γ-secretase inhibitor.
In vivo mouse skin wound healing. Mice were anesthetized by intraperitoneal injection of 300 μl of 1.25% Avertin (2,2,2-tribromoethanol, Sigma,-Aldrich), and full-thickness excision wounds using 4 mm biopsy punch (Kai) were made in the shaved dorsal skin. Wound tissue was harvested with 6 mm biopsy punch at the indicated time points (0, 1, 2, 7 days post wounding). After wounding, siRNA and DAPT were topically administered to the wound site. For the delivery of siRNA, 20 μl of 10 μM siRNA against mouse JMJD3 (m-063799-01-0005, Dharmacon), p65 (m-040776-01-0005, Dharmacon), Notch1 (18128, Bioneer) or control siRNA (d-001206-14-05, Dharmacon) mixed with 30% pluronic F127 gel (Sigma-Aldrich) was chilled on ice and applied to the wound cavity 86 . The efficiency of knock down of the specific gene was confirmed with real-time PCR. In addition, 20 μl of 100 μM DAPT was applied directly to the wound once daily until skin tissues were harvested 47,49 . DMSO was applied as a control. The wounds were photographed and wound sizes were analyzed using ImageJ software (NIH, Bethesda, USA). Control and experimental wound were done on the same animal. Experiments were performed with two biological repeats (n = 3/group). Data were pooled (total n = 6/group).
RT-PCR. Total RNA was extracted from HaCaT keratinocytes and mouse skin using RNAiso Plus (Takara).
First-strand cDNA synthesis from the total RNA template was performed with PrimeScript TM RT master mix (Takara). The resulting cDNAs were subjected to real-time PCR using qPCR 2x PreMIX SYBR (Enzynomics) with a Stratagen Mx3000p (Agilent Technologies). PCR conditions used to amplify all genes were 10 min at 95 °C and 40 cycles of 95 °C for 15 s, 60 °C for 40 s. Expression data were calculated from the cycle threshold (Ct) value using the ΔCt method of quantification. GAPDH was used for normalization. Oligonucleotides used for real-time PCR are listed in Supplementary Table S1.
Standard curve method for mRNA quantification. Plasmids containing each of Notch1 and 4 cDNA were used as standards for the absolute quantification. The concentration of each plasmid was measured by spectrophotometer or Ct value of real-time PCR using specific oligonucleotides for plasmid backbone. After concentration of plasmids were determined (same copies per 1 μl), plasmids were diluted (1:1, 1:10, 1:100, 1:1000, 1:10000). The diluted plasmids were amplified by specific oligonucleotides for each gene by real-time PCR. Using Ct value of each of plasmids, Notch1 and 4 standard curve were determined as according to the manufacturer's instructions of Stratagene Mx3000 P (Agilent Technologies, Germany). The absolute amount of mRNA was calcu- Over-expression of NICD. Intracellular domain of Notch1 was amplified using oligonucleotides: for ward (EcoRI) 5′-CCGGAATTCGGTGCTGCTGTCCCGCAA-3′; reverse (XhoI) 5′-CCGCTCGAGCGTTTACTTGAAGGCCTCCGG-3′ and cloned into the pCS4 expression vector containing HA epitope. HaCaT keratinocytes were transfected transiently with jetPRIME reagent (Polyplus).
Immunocytochemistry, immunohistochemistry, and histology. HaCaT keratinocytes were fixed for 15 min with 4% paraformaldehyde in PBS and permeabilized with PBST solution (0.5% Triton X-100 in PBS) for 30 min. After blocking of cells with 5% BSA in PBST solution for 1 h, cells were incubated with the anti-NICD (ab8925, Abcam) or anti-vinculin antibodies (V9131, Sigma-Aldrich) overnight at 4 °C. Antigens were detected with the secondary antibodies conjugated to TR or FITC (Sigma-Aldrich). To detect F-actin, cells were labelled with Alexa 546-phalloidin (Sigma-Aldrich). The wounded skins were immediately fixed with 4% paraformaldehyde in PBS and left overnight at 4 °C. The samples were dehydrated, embedded in paraffin and sectioned at 6 μm. The sections were de-paraffinized and boiled in Tris-EDTA buffer (pH 9.0) for 30 min, and then they were cooled at room temperature. The sections were incubated with 5% normal serum for 30 min. Subsequently, the sections were incubated with anti-NICD (ab8925, Abcam) or anti-KRT14 antibodies (ab7800, Abcam) overnight at 4 °C. After three washings with PBST, the sections were then incubated with a secondary antibody conjugated to FITC or TRITC. For histology, the sections were stained with hematoxylin and eosin (DAKO).
Promoter reporter assay. HaCaT keratinocytes were transfected transiently with Hes1 or Notch1 gene luciferase construct in conjunction with a control thymidine kinase promoter-driven renilla luciferase. For mutation of RBP-J binding sites (TGTGGGAA → TGACGCTA and TTCACACG → TAGAGTCG) on Hes1 gene promoter, PCR was performed 87 . After transfection of luciferase reporter containing gene and control renilla luciferase expression vector, cells were scratched and reporter activity was determined as described previously 88 .

Scratch wound assay.
HaCaT keratinocytes were plated on 6-well or 12-well culture plates and grown in 10% FBS containing a DMEM medium until they reached confluence. After cells were starved in the serum-free medium for 24 hr, the cell cultures were scratched using a sterile 10 μL tip. Cells were washed with serum-free medium and allowed to grow in 2% FBS containing DMEM. Cell cultures were recorded after scratching at the indicated times using the IncuCyte imaging system (Essen BioScience).
Transwell assay. HaCaT keratinocyte migration was assessed quantitatively using Transwell (Corning Costar). After keratinocytes were starved in a serum free medium for 24 hr, keratinocytes (1 × 10 5 ) were plated on the upper chamber of a 24-well Transwell plate with an 8μm pore size. After keratinocytes were incubated for 24 hr in a medium containing 10% FBS, the migrated keratinocytes were counted using the IncuCyte imaging system.

Statistical analyses.
All quantitative data are presented as mean ± S.E.M. for three independent experiments. The differences between two groups were evaluated by a paired t-test. Analysis of variance (ANOVA) was used for multiple comparisons. Significance values were *P ≤ 0.05 and **P ≤ 0.01.