Heparin-binding epidermal growth factor-like growth factor (HB-EGF) enhances reepithelialization in wounds. Estrogen is known to promote cutaneous wound repair. We examined the in vitro effects of 17β-estradiol (E2) on HB-EGF production by human keratinocytes. E2 or membrane-impermeable BSA-conjugated E2 (E2-BSA) increased HB-EGF secretion, mRNA level, and promoter activity in keratinocytes. E2 or E2-BSA enhanced in vitro wound closure in keratinocytes, and the closure was suppressed by anti-HB-EGF antibody. Activator protein-1 (AP-1) and specificity protein 1 (Sp1) sites on HB-EGF promoter were responsible for the E2- or E2-BSA-induced transactivation. Antisense oligonucleotides against c-Fos, c-Jun, and Sp1 blocked E2- or E2-BSA-induced HB-EGF transactivation. E2 or E2-BSA enhanced DNA binding and transcriptional activity of AP-1 and generated c-Fos/c-Jun heterodimers by inducing c-Fos expression. E2 or E2-BSA enhanced DNA binding and transcriptional activity of Sp1 in parallel with the enhancement of Sp1 phosphorylation. These effects of E2 or E2-BSA were not blocked by the nuclear estrogen receptor antagonist ICI-182,780 or anti-estrogen receptor-α or -β antibodies but were blocked by inhibitors of G protein, phosphatidylinositol-specific PLC, PKC-α, and MEK1. These results suggest that E2 or E2-BSA may enhance HB-EGF production via activation of AP-1 and Sp1. These effects of E2 or E2-BSA may be dependent on membrane G protein-coupled receptors different from nuclear estrogen receptors and on the receptor-mediated activities of phosphatidylinositol-specific PLC, PKC-α, and MEK1. E2 may enhance wound reepithelialization by promoting HB-EGF production in keratinocytes.
- activator protein-1
- specificity protein 1
- G protein
heparin-binding epidermal growth factor-like growth factor (HB-EGF), which belongs to the EGF family of growth factors, is synthesized as a membrane-anchored form (proHB-EGF) in epidermal keratinocytes (20, 28). When skin is mechanically or chemically injured, soluble forms of HB-EGF are released by proteolytic cleavage of proHB-EGF at the extracellular domain (28). Soluble HB-EGF activates EGF receptor on the surface of keratinocytes in a paracrine and an autocrine manner and induces both the release of soluble HB-EGF and the expression of HB-EGF gene by a positive feedback mechanism. Soluble HB-EGF induces migration and proliferation of keratinocytes, fibroblasts, and smooth muscle cells to fill the injured area and thus promotes reepithelialization and granulation tissue formation in the wound (20, 28). HB-EGF mRNA levels are increased in keratinocytes lining the wound edge (57, 61), and soluble HB-EGF is present in skin wound fluids (40, 41).
Previous studies suggest that estrogen potentiates cutaneous wound repair in a variety of ways (5–8, 46). Estrogen prevents excessive inflammation in the wound by inhibiting neutrophil influx into the wound or by inhibiting production of a proinflammatory cytokine, macrophage migration inhibitory factor, in monocytes/macrophages (8). Estrogen promotes granulation tissue formation and collagen deposition in the wound by increasing levels of transforming growth factor-β1 or tissue inhibitor of metalloproteinases and reducing those of procollagenase and prostromelysin in fibroblasts (6, 7). Estrogen stimulates angiogenesis and wound contraction by increasing synthesis of platelet-derived growth factor in monocytes/macrophages (54). Estrogen also promotes wound reepithelialization in ovariectomized rats or aged humans (5). This indicates that estrogen may promote HB-EGF production by keratinocytes in the wound. Previous studies reported that a natural estrogen, 17β-estradiol (E2), enhances HB-EGF production in rat uterine epithelium (64, 68). The precise mechanism for these effects, however, has not been defined.
It is known that E2 manifests its effects by two different mechanisms, genomic and nongenomic. The genomic mechanism is E2-bound nuclear estrogen receptor (ER)-α or -β stimulating or inhibiting gene expression by binding to estrogen response element (ERE) of target genes or by interacting with other transcription factors (9, 24). On the other hand, the nongenomic mechanism is E2 interacting with cell surface binding sites and rapidly inducing a variety of intracellular signals (36). Membrane E2-binding sites do not appear to be a unique molecule; some may be posttranslationally modified forms of nuclear ERs, and others may be structurally different from nuclear ERs (10, 45). Recent studies reported that membrane ER-α, derived from the same transcript as that of nuclear ER-α, resides in caveolae in association with caveolin-1 or -2, although in very low numbers (∼3% of nuclear ER-α) (48, 50, 51). Such membrane ER-α, when liganded with E2, dimerizes and activates various G protein α-subunits to stimulate adenylate cyclase, PLC, mitogen-activated protein kinases, and phosphatidylinositol 3-kinase (37, 51). On the other hand, we recently found (32) that E2 activates adenylate cyclase and induces cAMP signal via a G protein-coupled receptor, GPR30, on human keratinocytes. E2 also generated a signaling cascade of phosphatidylinositol-specific PLC (PI-PLC)-PKC-α-MEK1-ERK via an as yet undefined G protein-coupled receptor(s) on keratinocytes (34).
In this study, we investigated in vitro effects of E2 on HB-EGF production in human keratinocytes. We found that E2-enhanced HB-EGF transcription is dependent on the activities of the transcription factors activator protein-1 (AP-1) and specificity protein 1 (Sp1).
MATERIALS AND METHODS
E2, 17β-estradiol 6-(O-carboxymethyl)oxime:BSA (E2-BSA), 17α-estradiol, H89, and guanosine 5′-O-(2-thiodiphosphate) (GDPβS) were purchased from Sigma (St. Louis, MO). ICI-182,780 was from Wako Pure Chemical Industries (Osaka, Japan). E2, E2-BSA, and ICI-182,780 were dissolved in ethanol at 10−2 M to create stock solutions and were subsequently diluted into experimental media to yield final concentrations. The ethanol concentration used as vehicle control was 0.1%. U-73122, PD-98059, and Gö-6976 were obtained from Calbiochem (La Jolla, CA). Anti-c-Fos, -FosB, -Fra-1, -Fra-2, -c-Jun, -JunB, -JunD, -Sp1, -Sp2, -Sp3, -Sp4, and -AP-2 antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse monoclonal anti-ER-α antibody (D-12) against NH2-terminal amino acids 2–185 and rabbit polyclonal anti-ER-β antibody (D7N) against the COOH-terminal 19 amino acids were purchased from Santa Cruz Biotechnology and Zymed Labs (San Francisco, CA), respectively.
Culture of keratinocytes.
Normal human keratinocytes from sun-protected, disease-free chest skin of a 70-year-old Caucasian woman (Clonetics, Walkersville, MD) were cultured in serum-free keratinocyte growth medium (KGM; Clonetics) consisting of keratinocyte basal medium (KBM) supplemented with 0.5 μg/ml hydrocortisone, 5 ng/ml EGF, 5 μg/ml insulin, and 0.5% bovine pituitary extract. Cells in the third passage were used.
Keratinocytes (5 × 104/well) were seeded in triplicate into 24-well plates in 0.4 ml of KGM and adhered overnight, and then medium was changed to phenol red-free KBM depleted of growth supplements and incubated for 18 h. The medium was removed, and cells were incubated with E2, E2-BSA, or vehicle (control) in 0.4 ml of KBM for 24 h. HB-EGF amounts in the supernatants were assayed by ELISA using mouse monoclonal anti-human HB-EGF antibody and biotinylated anti-human HB-EGF antibody (both from R&D Systems, Minneapolis, MN) as described previously (30).
In vitro wounding assay.
This assay was performed as described previously (63). Briefly, keratinocytes were grown to confluence in 35-mm culture dishes, after which the medium was replaced with phenol red-free KBM and the cells were incubated for a further 24 h. Cell monolayers were wounded by a micropipette tip and treated with E2, E2-BSA, or vehicle (control) in the presence or absence of anti-human HB-EGF antibody (10 μg/ml) for 36 h. Wound closure was microscopically assessed, using the initial and final wound areas. Percentage of wound closure was calculated as [(initial − final)/initial] × 100.
Keratinocytes were incubated with E2 as above for 4 h, and then total cellular RNA was extracted with TRIzol reagent (Invitrogen, Rockville, MD) and reverse-transcribed to produce cDNA as described previously (32). The cDNA was thermocycled for PCR as described previously (32). Primers used were HB-EGF sense 5′-AAA GAA AGA AGA AAG GCA AGG -3′ and antisense 5′-AGA CAG ATG ACA GCA CCA CAG-3′ (35) and GAPDH sense 5′-GCA GGG GGG AGC CAA AAG GG-3′ and antisense 5′-TGC CAG CCC CAG CGT CAA AG-3′ (31). PCR was performed at 95°C for 3 min, 24 (GAPDH) or 29 (HB-EGF) cycles at 95°C for 30 s, 58°C for 30 s, and 72°C for 30 s, and finally 72°C for 3 min. PCR products were analyzed by electrophoresis on 2.5% agarose gels and stained with ethidium bromide. Densitometric analysis of the bands was performed by ATTO lane analyzer version 3 (ATTO, Osaka, Japan). The mRNA level of HB-EGF was normalized to that of GAPDH.
Plasmids and transfections.
A firefly luciferase reporter plasmid driven by human HB-EGF promoter (−2,000/+60 bp relative to the transcriptional start site) was constructed by PCR and insertion into pGL3 basic vector (Promega, Madison, WI) as described previously (53, 55) and denoted as pHB-EGF luc. Site-specific mutation of the promoters was created by multiple rounds of PCR using primers with altered bases as described previously (55). p4xAP-1-TATA-luc, p4xSp1-TATA-luc, p2xAP-1/Sp1-TATA-luc, and p4xERE-TATA-luc were constructed by inserting four copies of AP-1 or Sp1 sequences, two copies of sequences containing AP-1 and Sp1 sites from human HB-EGF promoter, or four copies of ERE from vitellogenin A2 promoter in front of the TATA box upstream of firefly luciferase reporter as described previously (1). Transient transfections were performed with FuGENE 6 (Roche, Indianapolis, IN) as described previously (18). Briefly, keratinocytes were plated in 35-mm dishes and grown to ∼60% confluence. FuGENE 6 premixed with KGM was mixed with pHB-EGF luc, p4xAP-1-TATA-luc, p4xSp1-TATA-luc, p2xAP-1/Sp1-TATA-luc, or p4xERE-TATA-luc together with herpes simplex virus thymidine kinase promoter-linked Renilla luciferase vector (pRL-TK; Promega) and incubated at room temperature for 15 min. The mixture was added to keratinocytes in KGM. After 6 h, transfected cells were washed and incubated in phenol red-free KBM for 18 h and then treated with E2, E2-BSA, or vehicle (control) in KBM. After 6 h, firefly and Renilla luciferase activities of the cell extracts were quantified by a dual-luciferase assay system (Promega). Results in each transfection were expressed as ratios of firefly to Renilla luciferase activities.
EMSA was performed as described previously (53, 55). Probes used were 32P-labeled annealed double-stranded DNA containing a putative AP-1 site (5′-CAGCAGTCAGTCACAAGGC-3′, consensus sequences underlined) or Sp1 site (5′-GGCGCCGGGCGGGGCGGA-3′) from HB-EGF promoter. Nuclear protein extracts were incubated with radiolabeled probes at room temperature for 30 min in buffer containing (in mM) 10 Tris·HCl (pH 7.5), 50 NaCl, 5 MgCl2, 1 EDTA, 1 DTT, 1 PMSF, and 125 KCl, with 10% glycerol, 0.1 mg/ml poly(dI-dC), and 0.1 mg/ml salmon sperm DNA. In antibody supershift experiments, nuclear extracts were preincubated with various antibodies for 30 min before the addition of probes. Reactions were fractionated on a nondenaturing 5% polyacrylamide gel and visualized with PhosphorImager software (Molecular Dynamics, Sunnyvale, CA).
Treatment with antisense oligodeoxynucleotides.
Antisense oligonucleotides against AP-1 or Sp1 proteins were synthesized as phosphorothioate-modified oligos corresponding to 5′-ends of mRNAs as described previously (19, 26). The oligonucleotides were c-Fos, 5′-GCGTTGAAGCCCGAGAA-3′; FosB, 5′-GGGGAAAGCCTGAAACAT-3′; Fra-1, 5′-CCCGAAGTCTCGGAACAT-3′; Fra-2, 5′-GGGATAATCCTGGTACAT-3′; c-Jun, 5′-CGTTTCCATCTTTGCAGT-3′; JunB, 5-TTCCATTTTAGTGCACAT-3′; JunD, 5′-GTAGAAGGGTGTTTCCAT-3′; Sp1, 5′-ATATTAGGCATCACTCCAGG-3′; and control scrambled, 5′-ACCGTTCGCTGTTATCTT-3′. Keratinocytes were finally transfected with 0.2 μM of the indicated oligonucleotides premixed with FuGENE 6 in KGM for 6 h. The medium was aspirated, and cells were cultured with phenol red-free KBM for 18 h and then treated with E2 in KBM. In some experiments, these antisense oligonucleotides were transfected together with pHB-EGF luc and pRL-TK. These phosphorothioate antisense oligonucleotides encapsulated with FuGENE 6 easily enter into keratinocytes and reduce target protein levels by translational arrest and/or RNase H-mediated degradation of target mRNA (14, 27, 58).
Western blot analysis.
Keratinocytes were lysed with lysis buffer [in mM: 50 HEPES (pH 7.5), 150 NaCl, 1.5 MgCl2, 100 NaF, 100 sodium orthovanadate, and 1 EGTA (pH 7.7), with 10% glycerol, 1% Triton X-100] followed by centrifugation for 20 min at 14,000 g and 4°C. The supernatant proteins were subjected to SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was blocked and incubated with anti-human c-Fos, FosB, Fra-1, Fra-2, c-Jun, JunB, JunD, or Sp1 antibodies, followed by peroxidase-conjugated secondary antibodies (Bio-Rad, Hercules, CA). The blots were developed with an enhanced chemiluminescence kit (Amersham, Arlington Heights, IL).
Flow cytometry was performed as described previously (10), using intact and permeabilized keratinocytes. The latter were prefixed with 0.5% paraformaldehyde and permeabilized with phosphate-buffered saline containing 0.05% Tween 20 and 0.5% BSA. Intact and permeabilized cells were incubated with anti-ER-α (D-12) or anti-ER-β (D7N) antibody or control IgG for 1 h and then with FITC-conjugated secondary antibodies for 45 min. The cells were postfixed with 1% paraformaldehyde and were analyzed in a FACScan (Becton Dickinson, Sunnyvale, CA) with a sample size of 10,000 cells gated on the basis of forward and side scatter. As positive controls, intact and permeabilized human umbilical vein endothelial cells (HUVEC; Clonetics) were analyzed as above.
In vivo phosphate labeling and immunoprecipitation of Sp1.
Keratinocytes were cultured on 100-mm dishes and labeled with 100 μCi/ml 32P-labeled orthophosphate (Amersham) in phosphate-free KBM for 3 h, followed by preincubation with signal inhibitors for 30 min and incubation with E2 for 30 min. The cells were lysed, and 32P-labeled Sp1 protein from cell lysate was immunoprecipitated by anti-Sp1 antibody and protein A agarose (Santa Cruz), resolved on 8% SDS-PAGE, and autoradiographed.
E2 enhanced HB-EGF secretion.
E2 dose-dependently increased HB-EGF secretion in keratinocytes, whereas the E2 stereoisomer 17α-estradiol was ineffective (Fig. 1A). Vehicle (ethanol) treatment at 0.1% did not reduce cell viability (>95% viable) or HB-EGF secretion [mean ± SE: 0.27 ± 0.03 ng/ml (n = 5) with vehicle vs. 0.26 ± 0.02 ng/ml with medium alone; P > 0.05 by paired t-test]. The stimulatory effect of E2 was manifested at 10−9 M and maximized at 10−8 M, which increased the secretion 4.1-fold over controls, whereas at 10−6 M, the stimulatory effect of E2 was weakened, and 10−5 M E2 showed no significant effect. Like E2, plasma membrane-impermeable E2-BSA enhanced HB-EGF secretion (Fig. 1B), whereas BSA alone was ineffective, indicating membrane receptor-mediated effects. The nuclear ER antagonist ICI-182,780 (10−6 M) did not counteract the E2- or E2-BSA-induced increase of HB-EGF secretion (Fig. 1B). ICI-182,780 was functional, because it blocked E2-stimulated transcriptional activity through ERE (Fig. 1C). E2-BSA did not increase ERE-dependent transcriptional activity, indicating that E2-BSA may not act on nuclear ERs. To examine the involvement of membrane ERs in E2-induced HB-EGF secretion, we analyzed whether the effects of E2 may be blocked by anti-ER antibodies that are known to interact with membrane ERs in certain cell types (51, 62). As examined by flow cytometry, anti-ER-α or anti-ER-β antibodies did not significantly interact with intact keratinocytes without permeabilization (Fig. 2, B and C), whereas only anti-ER-β interacted with permeabilized keratinocytes (Fig. 2F), indicating that antibody-detectable membrane ER-α or ER-β may not exist and only intracellular ER-β may exist in keratinocytes. Both antibodies interacted with a subpopulation of intact HUVEC with modest intensity (Fig. 2, H and I), whereas both interacted with a majority of permeabilized HUVEC with higher intensity (Fig. 2, K and L). These findings indicate that a subpopulation of HUVEC may express membrane ER-α and ER-β in modest amounts and that a majority of HUVEC may express intracellular ER-α and ER-β more abundantly, which is consistent with previous studies (51, 52). Neither anti-ER-α (D-12) nor anti-ER-β (D7N) antibody suppressed E2-induced enhancement of HB-EGF secretion (Fig. 1B). These antibodies did not appear to have access to intracellular ERs in intact cells without permeabilization because anti-ER-β did not suppress ERE-dependent transcriptional activity in keratinocytes containing intracellular ER-β alone (Fig. 1C). These results suggest that E2-induced enhancement of HB-EGF secretion may be mediated by membrane receptor(s) structurally different from classic nuclear ERs. Because 10−8 M E2 was optimal for this effect, this concentration was used in the following experiments.
E2 enhanced wound closure in monolayer keratinocytes via HB-EGF.
We further examined whether enhancement of HB-EGF secretion by E2 may contribute to in vitro wound closure in keratinocytes. Scratch-wounded monolayers of keratinocytes treated with E2 or E2-BSA displayed a significant increase in wound closure compared with vehicle controls (Fig. 3). Anti-HB-EGF antibody suppressed both basal and E2- or E2-BSA-induced wound closure, but control mouse IgG did not (data not shown), indicating the requirement of HB-EGF for both basal and E2- or E2-BSA-induced wound closure. In parallel with the results in HB-EGF secretion, ICI-182,780 did not suppress E2-induced enhancement of wound closure, and 17α-estradiol or BSA did not enhance wound closure compared with control. These results suggest that E2 or E2-BSA may enhance keratinocyte wound closure by promoting HB-EGF secretion.
E2 increased HB-EGF mRNA level.
We next examined whether E2 may alter the steady-state HB-EGF mRNA level in keratinocytes. At 4 h of incubation, E2 or E2-BSA increased HB-EGF mRNA level, whereas 17α-estradiol and BSA were ineffective (Fig. 4). ICI-182,780 and anti-ER-α or anti-ER-β antibodies did not block the E2- or E2-BSA-induced increase of HB-EGF mRNA level. Thus E2 enhanced HB-EGF production in keratinocytes at the pretranslational level, possibly via membrane receptor(s) different from classic nuclear ERs. We then examined whether E2 or E2-BSA may enhance HB-EGF promoter activity.
E2 enhanced HB-EGF promoter activity through AP-1 and Sp1 sites.
Human HB-EGF promoter contains AP-1- and CCAAT/enhancer-binding protein (C/EBP)-like sites and two closely spaced GC-rich Sp1 sites (Ref. 22; Fig. 5A). E2 or E2-BSA increased wild-type HB-EGF promoter activity (Fig. 5B), and the effects of E2 and E2-BSA were not counteracted by ICI-182,780 or anti-ER-α or anti-ER-β antibodies (data not shown). The mutation of the AP-1 site reduced the fold increase of promoter activity by E2 or E2-BSA; however, significant stimulation was still retained, indicating that an AP-1-like site may partially confer E2- or E2-BSA-induced promoter activation. The mutation of C/EBP or upper Sp1 [Sp1(I)] sites did not affect basal and E2- or E2-BSA-induced promoter activities. The mutation of the proximal Sp1 [Sp1(II)] site reduced basal promoter activity by 62% and reduced the fold increase of promoter activity by E2 or E2-BSA, indicating that Sp1(II) may confer basal promoter activity and may be involved in E2- or E2-BSA-induced promoter activation. The mutation of both AP-1 and Sp1(II) sites completely abrogated E2- or E2-BSA-induced enhancement of promoter activity. These results suggest that both AP-1 and Sp1(II) sites may be required for full activation of HB-EGF promoter by E2 or E2-BSA.
AP-1 family proteins are composed of Jun family (c-Jun, JunB, JunD) and Fos family (c-Fos, FosB, Fra-1, Fra-2) proteins (2). Fos/Jun or Jun/Jun dimers bind the consensus AP-1 (TGAG/CTCA) or related sequences on target promoters. To determine the transactivator proteins involved in E2-induced HB-EGF transcription, we examined whether antisense oligonucleotides against AP-1 components or Sp1 may block E2-induced HB-EGF transcription. c-Fos expression in keratinocytes was not constitutively detected but was induced by E2, whereas the expression of c-Jun or Sp1 was constitutive and was not altered by E2 (Fig. 5C). Treatment with antisense c-Fos, c-Jun, or Sp1 selectively reduced the respective protein levels (Fig. 5C). Antisense Sp1 suppressed both basal and E2-induced HB-EGF promoter activities, indicating the requirement of Sp1 for both basal and E2-induced HB-EGF transcription (Fig. 5D). Antisense c-Fos or c-Jun blocked the E2-induced increase of HB-EGF promoter activity, although it did not decrease basal promoter activity (Fig. 5D), indicating that c-Fos and c-Jun may be required for E2-induced HB-EGF transcription but may not be involved in basal transcription. Control scrambled oligonucleotide did not alter HB-EGF promoter activity in either the presence or the absence of E2. Other AP-1 proteins, FosB, Fra-1, Fra-2, JunB, and JunD, were constitutively expressed in keratinocytes, and the expression levels were not altered by E2 (data not shown). Treatment with antisense FosB, Fra-1, Fra-2, JunB, or JunD selectively reduced the respective protein levels; however, it did not alter HB-EGF promoter activity in either the presence or the absence of E2 (data not shown), indicating that these AP-1 components may not be involved in basal and E2-induced HB-EGF transcription. The effects of individual antisense oligonucleotides on E2-BSA-induced HB-EGF promoter activity were similar to those on E2-induced activity (data not shown).
E2 enhanced transcriptional activities of AP-1 and Sp1.
Keratinocytes were transiently transfected with luciferase reporter linked to four repeats of HB-EGF promoter-derived AP-1 or Sp1 elements in front of the TATA box, which reflect AP-1- or Sp1-dependent transcriptional activity, respectively. E2 and E2-BSA enhanced both AP-1- and Sp1-dependent transcriptional activities in keratinocytes more than twofold those of controls (Fig. 6). In addition, E2 and E2-BSA enhanced transcriptional activity through an AP-1-Sp1 composite element. ICI-182,780 and anti-ER-α and anti-ER-β antibodies did not block the E2- or E2-BSA-induced enhancement of transcriptional activities through AP-1, Sp1, or AP-1/Sp1.
E2 induced c-Fos/c-Jun binding to AP-1 site and enhanced DNA binding of AP-1 and Sp1.
We then examined whether E2 or E2-BSA may enhance DNA binding of transcription factors at AP-1 or Sp1 sites on the HB-EGF promoter. At 1 h of incubation, E2 or E2-BSA increased the amount of DNA-protein complex with AP-1-like sequences from the HB-EGF promoter 8.91-fold or 8.69-fold of controls, respectively, as examined by band density (Fig. 7A, lanes 3 and 7), indicating that E2 and E2-BSA may enhance DNA binding of transcription factors at the AP-1 site. ICI-182,780 and anti-ER-α and anti-ER-β antibodies did not block the E2- or E2-BSA-induced enhancement of transcription factor binding to this site (Fig. 7A, lanes 4–6 and 8). In nonstimulated keratinocytes, anti-c-Jun, but not anti-c-Fos, antibody supershifted the complex (Fig. 7A, lanes 10 and 11), whereas in E2 (Fig. 7A, lanes 12 and 13)- or E2-BSA (data not shown)-treated keratinocytes, both antibodies supershifted the complex. Antibodies against the other Fos family (FosB, Fra-1, Fra-2) or Jun family (JunB, JunD) proteins did not reduce or supershift the complexes by nuclear extracts from nonstimulated or E2- or E2-BSA-stimulated keratinocytes (data not shown). These results suggest that E2 or E2-BSA may enhance DNA binding of AP-1 and induce the binding of c-Fos/c-Jun heterodimers, whereas only c-Jun/c-Jun homodimers may bind the AP-1-like sequences in nonstimulated cells.
E2 or E2-BSA increased the amount of DNA-protein complex with Sp1(II) sequences from HB-EGF promoter 2.59-fold or 2.47-fold of controls, respectively, as determined by band density (Fig. 7B, lanes 3 and 7). Anti-Sp1 antibody supershifted the complexes by nuclear extracts from nonstimulated (Fig. 7B, lane 10) and E2 (Fig. 7B, lane 12)- or E2-BSA (data not shown)-stimulated keratinocytes. These results suggest that E2 or E2-BSA may enhance DNA binding of Sp1. ICI-182,780 and anti-ER-α and anti-ER-β antibodies did not block the E2- or E2-BSA-induced enhancement of Sp1 binding (Fig. 7B, lanes 4–6 and 8). Antibodies against Sp3 (Fig. 7B, lanes 11 and 13, and data not shown) or Sp2, Sp4, or AP-2 (data not shown) did not reduce or supershift the complexes by nuclear extracts from nonstimulated or E2- or E2-BSA-stimulated keratinocytes.
E2 enhanced phosphorylation of Sp1.
Because it is reported that phosphorylation of Sp1 enhances its DNA binding and/or transcriptional activity (11), we examined whether E2 may alter the phosphorylation level of Sp1. E2 and E2-BSA increased the Sp1 phosphorylation level without altering the total amount of Sp1 (Fig. 8, lanes 2 and 7). ICI-182,780 and anti-ER-α and anti-ER-β antibodies did not block E2- or E2-BSA-induced Sp1 phosphorylation (Fig. 8, lanes 3–5 and 8).
G protein, PI-PLC, PKC-α, and MEK1 were involved in E2-induced HB-EGF transcription.
We recently found that E2 induced a PI-PLC/PKC-α/MEK1/ERK signaling pathway via undefined membrane G protein-coupled receptor(s) (33) or induced a cAMP/PKA pathway via membrane G protein-coupled receptor GPR30 in keratinocytes (32). It is reported that E2 transactivates EGF receptor via membrane G protein-coupled receptors in MCF-7 breast carcinoma cells (23, 49). We thus examined which of these signals may be responsible for E2-induced HB-EGF transcription, using specific inhibitors for G protein and signaling enzymes. The G protein inhibitor GDPβS, the PI-PLC inhibitor U-73122, the PKC-α inhibitor Gö-6976, and the MEK1 inhibitor PD-98059 counteracted E2-induced increases of HB-EGF promoter activity (Fig. 9A) and AP-1 (Fig. 9B)- or Sp1 (Fig. 9C)-dependent transcriptional activities without altering basal activities and also suppressed E2-induced enhancement of DNA binding of AP-1 and Sp1 (Fig. 10). GDPβS, U-73122, Gö-6976, and PD-98059 also blocked E2-induced increases of HB-EGF secretion and mRNA level (data not shown). In parallel with transcriptional activity and DNA binding of Sp1, GDPβS, U-73122, Gö-6976, and PD-98059 suppressed E2-induced Sp1 phosphorylation (Fig. 8, lanes 9–12). GDPβS, U-73122, Gö-6976, and PD-98059 blocked E2-BSA-induced increase of HB-EGF transcription, DNA binding, or transcriptional activities of AP-1 or Sp1 or Sp1 phosphorylation (data not shown). These results suggest that E2- or E2-BSA-induced HB-EGF transcription from AP-1 and Sp1 sites may be mediated by membrane G protein-coupled receptor(s) and dependent on the receptor-triggered PI-PLC/PKC-α/MEK1/ERK pathway. On the other hand, the PKA inhibitor H89 and the EGF receptor tyrosine kinase inhibitor AG-1478 did not suppress E2 (Figs. 8–10)- or E2-BSA (data not shown)-induced enhancement of HB-EGF promoter activity, Sp1 or AP-1 transcriptional activities or DNA binding, and Sp1 phosphorylation. These results indicate that PKA or EGF receptor may not be involved in these effects of E2 or E2-BSA.
E2 enhanced HB-EGF transcription through AP-1 (−324 bp) and Sp1 (−62 bp) elements. These two elements are also responsible for oxidant stress-induced HB-EGF transcription in renal epithelial cells (53). E2 induced c-Fos expression and changed AP-1 composition from c-Jun/c-Jun homodimers to c-Fos/c-Jun heterodimers, leading to the enhancement of DNA binding and transcriptional activity at the AP-1 site. It is reported that c-Fos/c-Jun heterodimers are more stable and have higher DNA binding affinity and transcriptional activity than c-Jun/c-Jun homodimers (2). c-Fos expression by E2 was dependent on a PI-PLC/PKC-α/MEK1/ERK pathway. c-fos promoter contains a serum response element that is bound by ternary complex factors like Elk-1 or serum response factor accessory protein 1 (SAP-1) (56). It is known that ERK-mediated phosphorylation of ternary complex factor promotes its transcriptional activity (39). It is thus anticipated that phosphorylation of ternary complex factors by ERK may be involved in E2-induced c-Fos expression. Although E2 did not enhance c-Jun expression in keratinocytes, c-Jun was necessary for E2-induced HB-EGF transcription. c-Jun may be required as an anchor for c-Fos to bind DNA because c-Fos by itself cannot homodimerize and bind target sequences (2). Previous studies also reported that c-Jun is a crucial regulator of HB-EGF and/or EGF receptor and necessary for wound closure (38, 67). In mice lacking c-Jun, keratinocytes at the leading edge of the wound cannot properly activate EGF receptor, express keratin-6, activate focal adhesion kinase, or form actin stress fibers and thus cannot elongate or migrate (38). This inability is rescued by exogenous HB-EGF (38). Thus c-Jun may contribute to wound closure by promoting autocrine and paracrine loops of HB-EGF/EGF receptor.
A GC-rich element (−62 bp) on HB-EGF promoter was bound by Sp1 in keratinocytes. E2 enhanced DNA binding and transcriptional activity of Sp1 at this element. This element was also responsible for gastrin-induced HB-EGF transcription in gastric parietal cells (55). In these cells, however, this element was bound by as yet undefined zinc finger family proteins different from Sp1 (55). It is known that numerous zinc finger family proteins including Sp1 have highly conserved COOH-terminal zinc finger domains that function in DNA binding and preferentially bind GC-rich sequences (11). Zinc finger family members binding target GC-rich elements may vary with cell type, promoter context, and stimuli, which may be influenced by the sequences and relative expression or activity of individual members. E2 enhanced Sp1 phosphorylation, which correlated with enhanced DNA binding and transcriptional activity. Phosphorylation of Sp1 may dissociate repressor(s) from Sp1 such as Sp1-I or p74, which inhibit DNA binding or transcriptional activity, respectively (13, 44), or from a Sp1-bound promoter such as histone deacetylase 1 (15). Alternatively, Sp1 phosphorylation may recruit a transcriptional coactivator like p300 or enhance the interaction with basal transcriptional machinery such as dTAFII110 (25). The results with kinase inhibitors suggest that Sp1 kinase activated by E2 may be downstream of PKC-α and MEK1. One candidate is ERK, because Sp1 contains six putative ERK phosphorylation sites (42). It is also reported that ERK enhances Sp1 phosphorylation and its DNA binding (42, 43). Alternatively, Sp1 kinase may be downstream from ERK (16). Further studies should identify E2-stimulated Sp1 kinase and phosphorylation sites on Sp1.
In this study, E2-induced activation of Sp1 and AP-1 and the resultant induction of HB-EGF expression were suppressed by GDPβS. E2-BSA manifested almost the same effects as E2. These results indicate that the effects of E2 may be mediated by G protein-coupled membrane receptor(s). A series of effects of E2 were not suppressed by ICI-182,780. It is known that ICI-182,780 suppresses dimerization and DNA binding of nuclear ERs (3, 21). Thus the results with ICI-182,780 can rule out the involvement of nuclear ERs. In the literature, however, ICI-182,780 does or does not inhibit membrane ER-mediated signaling effects of E2, depending on the types of signals or target cells (4, 12, 48, 52, 66). Thus the failure in inhibition by ICI-182,780 cannot completely rule out the involvement of membrane ERs. Membrane localization of ER may change its conformation and thus hinder its interaction with ICI-182,780. In addition, anti-ER-α or anti-ER-β antibodies did not significantly detect membrane ERs in flow cytometry and did not block a series of effects of E2 in keratinocytes. However, the possible involvement of membrane ERs cannot completely be ruled out because membrane targeting of ER may sequester epitopes and interfere with detection by antibodies and the amount of membrane ERs may be less than that detectable by ordinary methods. We are now studying whether anti-ER-α or anti-ER-β antibodies targeting different epitopes may block E2-induced HB-EGF expression. We previously found (32) that human neonatal foreskin keratinocytes express mRNA of ER-β, but not that of ER-α, by RT-PCR. Our present study using flow cytometry demonstrated that keratinocytes from adult skin may not express membrane ER-α or ER-β but only express intracellular ER-β. Thornton et al. (60) also detected nuclear ER-β but not ER-α in keratinocytes of adult scalp skin by immunohistochemistry, which was consistent with our results. In contrast, Verdier-Sevrain et al. (62) recently detected both ER-α and ER-β in human neonatal foreskin keratinocytes by immunoblotting and membrane ER-α by immunocytochemistry. Such a discrepancy may be caused by different experimental conditions, different keratinocyte sources, different methods (RT-PCR vs. immunostaining), different antibody sources, and the presence or absence of phenol red in medium. In particular, levels of membrane ER-α, if any, are very low and change dynamically with cell passage, density, or cell cycle (65). Further studies should precisely examine the presence or absence of membrane ER-α or ER-β in keratinocytes, and their dynamics, with appropriate fixatives, protocols, and antibodies.
It is reported that E2 binds to G protein-coupled membrane ER-α or orphan receptor GPR30 in MCF-7 breast carcinoma cells, which leads to release of proHB-EGF on the cell surface and activation of EGF receptor by released HB-EGF (23, 49). The activation of EGF receptor is known to stimulate Sp1 phosphorylation and transcriptional activity via a Ras/Raf/MEK1/ERK1/2 pathway (16). It is also reported that PKA phosphorylates Sp1 and promotes its transcriptional activity (11). However, EGF receptor and PKA may not be involved in E2-induced Sp1 phosphorylation and transcriptional activation, at least in human keratinocytes, according to the results with kinase inhibitors (Figs. 8–10). Signals effectively linked to Sp1 activation might vary with cell types or stimuli. It is reported that E2-bound nuclear ER-α interacts with Sp1 and potentiates its DNA binding and/or transcriptional activity in MCF-7 cells (17, 47). However, such a manner of Sp1 stimulation by E2 is unlikely in keratinocytes, because the nuclear ER antagonist ICI-182,780 did not suppress Sp1 stimulation by E2 (Figs. 6 and 7).
E2 in vitro induced HB-EGF release by keratinocytes. The released HB-EGF appeared to contribute to wound closure in keratinocytes (Fig. 3). These results indicate that in vivo E2 may promote reepithelialization in skin wounds by enhancing HB-EGF production in keratinocytes. Wound reepithelialization is slowed down with aging (59), which may be related to the impaired HB-EGF production in senescent cells (35). It is reported that systemic administration of E2 reversed the impaired wound reepithelialization in postmenopausal women (7), which may be related to restoration of HB-EGF production by E2. Cutaneous wound healing requires a variety of growth factors, and the application of a single growth factor is not always effective (29). In addition to HB-EGF production in keratinocytes, E2 induces production of multiple growth factors in multiple cell types, including nerve growth factor and vascular endothelial growth factor production in macrophages (31, 33), leading to reinnervation and angiogenesis in the wound. E2 stimulates transforming growth factor-β1 production in fibroblasts (6), related to granulation tissue formation. Application of E2 to a skin wound may thus produce synergistic effects of these growth factors for repair. HB-EGF is also produced by other cell types in skin wounds, such as fibroblasts or macrophages (40). Further study should elucidate whether E2 may enhance HB-EGF production in these cell types by a mechanism(s) similar to or different from that in keratinocytes.
This work was supported in part by aid from the Foundation for Total Health Promotion and the Japanese Society for Investigative Dermatology’s Fellowship SHISEIDO Award 2004.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society