Mitogen-activated protein kinase kinase kinase 1 ... - Semantic Scholar

Download PDF
1 downloads 0 Views 643KB Size Report
Comment
Oct 18, 2011 - MAP3K1 is therefore the nexus of an intracrine regulatory loop connecting the TGF-α/EGFR/RhoA-c-Jun and JNK-c-Jun-AP-1 pathways in ...
Mitogen-activated protein kinase kinase kinase 1 (MAP3K1) integrates developmental signals for eyelid closure Esmond Geha,1, Qinghang Menga,1, Maureen Mongana, Jingcai Wanga,b, Atsushi Takatoric, Yi Zhengd, Alvaro Pugaa, Richard A. Lange,f, and Ying Xiaa,e,2 Departments of aEnvironmental Health and eOphthalmology, University of Cincinnati Medical Center, Cincinnati, OH 45267; bSouthern Medical University, Guangzhou 510515, China; cChiba Cancer Center, Chiba 260-8717, Japan; and dExperimental Hematology and Cancer Biology and fVisual Systems Group, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 45229

Developmental eyelid closure is an evolutionarily conserved morphogenetic event requiring proliferation, differentiation, cytoskeleton reorganization, and migration of epithelial cells at the tip of the developing eyelid. Many signaling events take place during eyelid closure, but how the signals converge to regulate the morphogenetic process remains an open and intriguing question. Here we show that mitogen-activated protein kinase kinase kinase 1 (MAP3K1) highly expressed in the developing eyelid epithelium, forms with c-Jun, a regulatory axis that orchestrates morphogenesis by integrating two different networks of eyelid closure signals. A TGF-α/EGFR-RhoA module initiates one of these networks by inducing c-Jun expression which, in a phosphorylation-independent manner, binds to the Map3k1 promoter and causes an increase in MAP3K1 expression. RhoA knockout in the ocular surface epithelium disturbs this network by decreasing MAP3K1 expression, and causes delayed eyelid closure in Map3k1 hemizygotes. The second network is initiated by the enzymatic activity of MAP3K1, which phosphorylates and activates a JNK-cJun module, leading to AP-1 transactivation and induction of its downstream genes, such as Pai-1. MAP3K1 inactivation reduces AP-1 activity and PAI-1 expression both in cells and developing eyelids. MAP3K1 is therefore the nexus of an intracrine regulatory loop connecting the TGF-α/EGFR/RhoA-c-Jun and JNK-c-Jun-AP-1 pathways in developmental eyelid closure.

T

he mitogen-activated protein kinases (MAPKs) are activated through an evolutionarily conserved three-component signal transduction cascade, composed of a mitogen-activated protein kinase kinase kinase 1 (MAP3K1), a MAP2K. and a MAPK (1). In the cascade, the MAP3Ks receive signals from upstream cues and pass them downstream by phosphorylating and activating the MAP2Ks, which in turn activate the MAPKs by phosphorylation. The MAPKs can modulate transcription factor activity and gene expression, thereby regulating diverse cellular functions. MAP3K1, also known as MEKK1, is a member of the MAP3K superfamily, known to have highly cell type-specific roles. In vivo studies using mice lacking either the full-length protein (MAP3K1-null or Map3k1−/−) or its kinase domain (Map3k1ΔKD/ΔKD) have shown that MAP3K1 is involved in immune system development and function, injury repair, vasculature remodeling, and tumor progression (2). However, the most obvious function of MAP3K1 is the control of eyelid closure during fetal development (3). Mammalian eye development involves a transient closure and reopening of the eyelid (4). In mice, eyelid development begins at embryonic day 13 (E13), when the surface ectoderm folds into the lid buds. The eyelid buds continue to grow toward the center of the ocular surface and by E15 to E16.5, the epithelial cells at the tip of the eyelid start to elongate and migrate. Ultimately, the epithelium of upper and lower eyelid fuses to form a closed eyelid. Mouse eyelids remain closed between E16.5 and postnatal day 12, and thereafter the lid fusion breaks down as a www.pnas.org/cgi/doi/10.1073/pnas.1102297108

consequence of epithelial cell apoptosis at the fusion junction, resulting in open eyelids. Mice are normally born with their eyelid closed, but those impaired in embryonic eyelid closure are born with eye open-at-birth (EOB) phenotype. Based on genetic mutations in mice that lead to EOB, it has become clear that complex signal transduction processes are involved in the regulation of eyelid closure. So far, whether embryonic eyelid closure is shown depends on signals derived from WNT, Sonic hedgehog, BMP/Activin, FGF, and EGF (5– 8). In addition, eyelid closure requires the participation of a number of intracellular signaling kinases, such as MAP3K1, JNK, ROCK, and CDH1, and nuclear transcription factors, such as c-Jun, Fra-2, FOXL2, SMAD, and GRHL3 (9–17). Molecular and genetic analyses of the knockout mice have shown that the eyelid closure factors are organized into distinct signal transduction cascades, but how the different pathways crosstalk to orchestrate the closure of the eyelid has remained largely unknown. One of the well-characterized eyelid closure pathways is mediated through the EGFR. Activation of EGFR is necessary for the epithelial cells at the developing eyelid tip to migrate, leading to embryonic eyelid closure. Consequently, defective EGFR signaling in mice with genetic knockout of the EGFR itself or its ligands, TGF-α and HB-EGF, or of factors, such as FGF10, cJun, and GRHL3 that regulate ligand expression, impairs migration, resulting in the EOB phenotype. Another eyelid-closure pathway is mediated by MAP3K1 (9). Specifically, MAP3K1 was shown to active the JNK MAPKs, which in turn phosphorylate the transcription factor c-Jun in epithelial cells at the developing eyelid tip. Activation of the MAP3K1-JNK-c-Jun cascade promotes epithelial cell migration and hence, embryonic eyelid closure (9, 18). As far as embryonic eyelid closure is concerned, the EGFR and MAP3K1 signaling pathways appear to regulate a common cell activity, but the point of convergence of these pathways has not been identified (3). In the course of previous work, we found that MAP3K1 was highly expressed in the developing eyelid tip epithelial cells before eyelid closure (9). By investigating the molecular mechanisms that control MAP3K1 expression, we find that TGF-α/EGFR signals, acting through RhoA and ROCK, induce c-Jun and the c-Jun–mediated activation of the Map3k1 promoter. Once induced, MAP3K1 is re-

Author contributions: A.P. and Y.X. designed research; E.G., Q.M., M.M., J.W., and A.T. performed research; Y.Z. and R.A.L. contributed new reagents/analytic tools; E.G., Q.M., M.M., J.W., A.T., A.P., and Y.X. analyzed data; and Y.X. wrote the paper. The authors declare no conflict of interest. *This Direct Submission article had a prearranged editor. 1

E.G. and Q.M. contributed equally to this work.

2

To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1102297108/-/DCSupplemental.

PNAS | October 18, 2011 | vol. 108 | no. 42 | 17349–17354

DEVELOPMENTAL BIOLOGY

Edited* by Michael Karin, San Diego School of Medicine, University of California, La Jolla, CA, and approved September 16, 2011 (received for review February 10, 2011)

quired for activation of the JNK-c-Jun pathway and induce AP-1 responsive genes. Thus, MAP3K1 is the nexus of an intracrine regulatory loop connecting the TGF-α/EGFR/RhoA-c-Jun and JNK-cJun-AP-1 pathways in the control of eyelid morphogenesis. Results Activation of the EGF Receptor Leads to MAP3K1 Induction. The Map3k1ΔKD allele contains a bacterial β-galactosidase gene knocked-in into the Map3k1 locus, replacing the exons coding for the MAP3K1 kinase domain (19). As a result, this allele produces a MAP3K1-ΔKD-β-gal fusion protein that lacks kinase but retains β-galactosidase activity. Using fibroblasts derived from heterozygous and knockout fetuses, we established that β-gal expression in genetically modified fibroblasts can be used as a surrogate to measure endogenous Map3k1 promoter directed gene expression (Fig. S1 A and B). Before eyelid closure, MAP3K1 is highly expressed in the leading edge of the developing eyelid, which consists of clumps of rounded periderm cells covering the elongated epithelium (20) (Fig. S1C). Concomitantly, several eyelid morphogens are also abundantly expressed in the protruding tip of the developing eyelids, raising the possibility that some of the morphogenic signals induce MAP3K1 expression (5, 21, 22). To test this possibility, we treated the Map3k1ΔKD/ΔKD cells with morphogenetic factors, including: TGF-α and EGF, which activate EGFR; activin B and TGF-β1, which act through receptors of the TGF-β superfamily; FGF10, which act through a FGF receptor; and retinoic acid, which acts through RXR, and measured β-gal activity (Fig. 1A and Fig. S2A). Although both EGFR ligands, TGF-α and EGF, significantly increased β-gal expression, none of the other agents had an effect. Correspondingly, TGF-α caused a clear induction of Map3k1 mRNA in wild-type cells, paralleled by that of β-gal mRNA in Map3k1ΔKD/ΔKD cells, reaching a maximum level at 2 h and diminishing to a baseline level by 6 h (Fig. 1B). In addition, TGF-α increased luciferase expression of pMap3k1-luc, a luciferase reporter driven by a 1.9kb fragment of the Map3k1 promoter (Fig. 2A). To determine whether EGFR was involved in MAP3K1 induction, we examined Map3k1ΔKD/ΔKD cells treated with TGF-α. We observed that TGF-α caused a rapid EGFR phosphorylation and that inhibition of EGFR by AG1478 significantly reduced MAP3K1ΔKD-β-gal expression (Fig. 1 C and D). In contrast, inhibition of ERK, JNK, or NF-κB pathways did not affect MAP3K1 induction by TGF-α (Fig. S2B). Collectively, these data suggest that EGFR activation by its ligands leads to induction of the Map3k1 promoter and gene transcription, which may contribute, at least partly, to MAP3K1 expression in the developing eyelid. MAP3K1 Induction Depends on RhoA and ROCK. One of the consequences of EGFR activation is the induction of the ROCK kinases by the small GTPase RhoA (23). Significantly, the ROCK kinases are crucial for embryonic eyelid closure (12), which makes them the most likely candidates for the transduction of EGFR signals to MAP3K1 during eyelid development. To explore this possibility, we investigated the roles of RhoA and ROCK in the induction of the pMap3k1-luc in HEK293 cells. Overexpression of ROCK or the dominant-active RhoA markedly induced luciferase expression, but overexpression of the dominant-negative RhoA blocked luciferase induction by TGF-α (Fig. 2A). We further showed that expression of ROCK or the dominant-active RhoA in the Map3k1ΔKD/ΔKD cells induced surrogate β-gal expression, but conversely, pharmacological inhibition of RhoA and ROCK or genetic ablation of RhoA by adenovirus-mediated expression of Cre recombinase in the Map3k1+/ΔKDRhoAF/F fibroblasts markedly decreased surrogate β-gal induction by TGF-α (Fig. 2 B and C). The above in vitro data strongly suggest that MAP3K1 expression depends on the signaling cascades that involve TGF-α 17350 | www.pnas.org/cgi/doi/10.1073/pnas.1102297108

Fig. 1. Induction of MAP3K1 expression in fibroblasts. The Map3k1ΔKD/ΔKD fibroblasts treated with various agents for 24 h or as indicated were (A) examined for the relative β-gal activities based on protein concentration, and (C and D) analyzed by Western blotting with antibodies for (C) anti–pEGFR and (D) β-gal and β-actin. The intensity of the β-gal was compared with that of β-actin. (B) Total RNA isolated from wild-type (solid bars) and Map3k1ΔKD/ΔKD fibroblasts (white bars) treated with TGF-α for different times were subjected to real-time RT-PCR for Map3k1 and β-Gal transcripts. The fold-induction was calculated based on cycle differences (ΔCt) in comparison with mouse Gapdh and the values in control cells. Data are presented as the mean values ± SE from at least four independent experiments and statistic analyses were done by comparing to the values in cells without treatment. *P < 0.05, **P < 0.01, and ***P < 0.001.

activation of EGFR and the RhoA/ROCK pathways. The available genetic evidence implicates all of the components of these cascades in embryonic eyelid closure, except for RhoA. To fill in this gap, we generated double transgenic mice containing a RhoAflox conditional allele combined with a lens epithelium promoter-driven Cre recombinase (Le-cre) (24). The resultant RhoAF/F/Le-cre mice have RhoA deletion specifically in cells originated from ocular surface ectoderm, including epithelium of eyelid, lens, and cornea. The RhoAF/F/Le-cre mice were born with closed eyelids indistinguishable from the wild-type mice, indicating that RhoA, unlike other components of this pathway, is dispensable for mouse eyelid development. To evaluate whether RhoA might interact with MAP3K1 during eyelid development, we generated RhoA deletions in Map3k1 wild-type, heterozygous, and knockout genetic backgrounds. When examined at E17.5, the developmental stage at which eyelid closure is normally completed, all of the wild-type fetuses (n = 9) had their eyelids closed, but the Map3k1ΔKD/ΔKD fetuses (n = 5) had eyelids open whether or not RhoA was deleted. These results confirm that MAP3K1, but not RhoA, is essential for mouse embryonic eyelid closure. On the other hand, although all of the Map3k1+/ΔKD/RhoAF/F fetuses (n = 5) had their eyelids closed just like the wild-type, the Map3k1+/ΔKD/ RhoAF/F/Le-Cre fetuses (n = 4) displayed partially open eyelids (Fig. 2D). Although the eyelid openings in Map3k1+/ΔKD/RhoAF/F/ Le-Cre fetuses were different in size, they were all clearly distinGeh et al.

guishable from the closed eyelids in wild-type fetuses and less pronounced than those in Map3k1ΔKD/ΔKD fetuses (Fig. 2E). The eyelids of the Map3k1+/ΔKD/RhoAF/F/Le-Cre fetuses eventually closed and the mice were born with closed eyelids similar to those of the Map3k1+/ΔKD and wild-type mice, suggesting that RhoA loss delays, but does not abolish, embryonic eyelid closure in the Map3k1 heterozygotes. Hence, as far as eyelid development is concerned, RhoA is an accessory to MAP3K1 signaling. Our in vitro studies suggest that RhoA may crosstalk with MAP3K1 at the level of Map3k1 gene transcription. To test whether this was the case in vivo, expression of the MAP3K1ΔKD-β-gal protein was tracked by whole-mount X-gal staining of Map3k1ΔKD/ΔKD fetuses at E17.5. In the developing eyelid epithelium, the intensity of β-gal was strong in the presence of RhoA, but weaker in its absence (Fig. 2F). Additionally, we analyzed the Map3k1+/ΔKD/RhoAF/F/Le-Cre and Map3k1+/ΔKD/ RhoAF/F fetuses (n = 5) at E15.5, a developmental stage before the onset of embryonic eyelid closure. There was no difference in the size and morphology of the eyelids in these fetuses, suggesting that RhoA knockout does not affect eyelid formation and growth before eyelid closure (Fig. 2G). However, examination of the eyelid epithelial cells isolated by laser capture microdissection (LCM) revealed a 60% reduction of β-gal activity by RhoA ablation. We therefore propose that RhoA is not essential for MAP3K1 expression but contributes to it, thereby acting as a modulator of embryonic eyelid closure. Molecular Mechanism of Map3k1 Promoter Regulation. To understand the mechanism of transcriptional regulation of MAP3K1, we studied the architecture of the Map3k1 promoter. Analysis of a series of promoter deletion mutants led to the Geh et al.

identification of a narrow region between −38 and −105 bp upstream of the transcription start site that was highly responsive to TGF-α (Fig. S3A). In silico analyses of the sequences within this region revealed three putative AP-1 binding sites, located at −25, −49, and −67 bp, that were evolutionarily conserved among human, mouse, and rat (Fig. S3B). This observation is appealing because several AP-1 binding proteins, such as c-Jun and Fra-2, are known to be crucial for embryonic eyelid closure (13, 14, 16). To determine whether the AP-1 complex plays a role in Map3k1 regulation, we measured AP-1 protein expression in wild-type, Map3k1+/ΔKD, and Map3k1ΔKD/ΔKD cells after treatment with TGFα (Fig. 3A and Fig. S4A). In all of the cells examined, expression of c-Jun was significantly induced, whereas expression of other AP-1 members, including Fra-1, Fra-2, and c-Fos, and expression of Elk1, was unaffected. Regression analysis revealed a significant correlation (R2 = 0.8015) between MAP3K1 and c-Jun induction (Fig. S4B). Pretreatment of the Map3k1ΔKD/ΔKD cells with the EGFR inhibitor AG1478, a condition that prevented MAP3K1 expression, also blocked c-Jun induction by TGF-α (Fig. 3A). To assess whether c-Jun was responsible for MAP3K1 induction, we expressed a dominant-negative c-Jun in pMap3k1luc transfected HEK293 cells. Expression of the dn-c-Jun significantly decreased luciferase induction by active RhoA (Fig. 3B). Conversely, overexpression of c-Jun in the Map3k1ΔKD/ΔKD cells increased surrogate β-gal activity, suggesting that elevation of c-Jun by itself was sufficient to stimulate MAP3K1 expression (Fig. 3C). To test whether the endogenous c-Jun was required for MAP3K1 induction, we isolated fibroblasts from Map3k1+/ΔKDJunF/F double-transgenic mice. Infection of the cells with Ad-Cre ablated the c-Jun gene and, correspondingly, abolished the surrogate β-gal activity induced by TGF-α or active RhoA (Fig. 3D). In PNAS | October 18, 2011 | vol. 108 | no. 42 | 17351

DEVELOPMENTAL BIOLOGY

Fig. 2. RhoA and ROCK are involved in MAP3K1 induction. (A) HEK293 cells were transfected with pMap3k1-luc together with the expression vectors for active RhoA (V14), dominant-negative RhoA (N19), and ROCKII. The cells were treated with TGF-α as indicated and the relative luciferase activity based on the cotransfected β-gal was determined. (B) The Map3k1ΔKD/ΔKD fibroblasts were transfected with the expression vectors for active RhoA (V14) and ROCKII, and the relative β-gal activities were determined based on protein concentration. (C) Fibroblasts isolated from the Map3k1+/ΔKDRhoAF/F mice were infected with Ad-Cre or Ad-GFP. The cells were treated with TGF-α in the presence or absence of chemical inhibitors for RhoA and ROCK, as indicated. The β-gal activities were determined and calculated based on protein concentration. The E17.5 fetuses of various genotypes were photographed at 1.6× magnification (D) before and (E and F) after whole-mount X-gal staining, and (F) the sections of X-gal stained eyes (light blue) were counter-stained with hematoxyline (dark blue) and photographed at 63× magnification. Arrows point a the leading edge of the developing eyelids. (G) Tissue sections of Map3k1+/ΔKDRhoAF/F (RhoAF/F) and Map3k1+/ΔKDRhoAF/FLe-Cre (RhoAF/FLe-Cre) E15.5 fetuses were photographed (Upper), focusing on the developmental eyelids. The eyelid epithelial cells were isolated by LCM and β-gal activities were determined and calculated based on protein concentration. Data are presented as the mean values ± SE from at least four independent experiments and statistical analyses were done by comparing to the values in cells without treatment or RhoA knockout, designated as 1.

Fig. 3. c-Jun regulates the Map3k1 promoter. (A) Map3k1ΔKD/ΔKD fibroblasts were treated with TGF-α for various times in the absence or presence of the EGFR inhibitor AG1478, and were analyzed by Western blotting using antibodies, as indicated. (B) HEK293 cells, transfected with pMap3k1-luc together with plasmids for active RhoA [RhoA(V14)] and dominant-negative c-Jun, and luciferase activities were determined. (C) Map3k1ΔKD/ΔKD fibroblasts, transfected with either control or c-Jun expression plasmids, were examined for β-gal activity relative to protein concentration. (D) Fibroblasts isolated from the Map3k1+/ΔKDc-JunF/F mice were infected with Ad-Cre or AdGFP. The cells were either treated with TGF-α or transfected with active RhoA (V14), and the β-gal activities were determined and calculated. (E) Map3k1ΔKD/ΔKD fibroblasts, treated with TGF-α for various times, were analyzed by ChIP assay with antibodies to RNA polymerase II, c-Jun, AcH4, and goat IgG as a negative control. The precipitated chromatin DNA was examined by real-time PCR using primers for the proximal Map3k1 promoter region. The values represent fold-change over nonspecific IgG control. (F) HEK293 cells transfected with pMap3k1-luc were treated with histone deacetylase inhibitors sodium butyrate (NaB, 5 mM) and Trichostatin A (TSA, 5 μM) for 18 h, and were examined for luciferase activity. Fold-induction was calculated based on values in control cells and results are average of at least four independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001.

contrast, infection of the cells with control Ad-GFP did not block β-gal induction. To examine whether c-Jun was directly involved in the regulation of the Map3k1 promoter, we performed ChIP assays. In both wild-type and Map3k1ΔKD/ΔKD fibroblasts, TGF-α caused a significant increase of c-Jun binding to the Map3k1 promoter, in parallel to RNA polymerase II recruitment, indicative of promoter activation (Fig. 3E). Histone H4 acetylation, known to play a primary role in the structural changes that mediate enhanced binding of transcription factors to DNA within nucleosomes (25), was induced by TGF-α at the endogenous Map3k1 promoter. Consistent with this observation, inhibition of histone deacetylase with sodium butyrate or trichostatin A (TSA) increased luciferase expression in pMap3k1-luc transfected HEK293 cells (Fig. 3F). Taken together, the data suggest that histone deacetylase inactivation and histone acetylation provide a permissive chromatin configuration that allows c-Jun binding and RNA polymerase II recruitment to activate the Map3k1 promoter in response to TGF-α. Kinase Activity of MAP3K1 Is Required for JNK and c-Jun Phosphorylation and Maximal AP-1 Activity. Although the above data suggest

that c-Jun regulates MAP3K1 expression, our previous studies have shown that MAP3K1 is required for activation of the JNK-c-Jun cascades (9). To investigate whether MAP3K1 might be upstream of c-Jun, we examined c-Jun in wild-type and Map3k1ΔKD/ΔKD cells. Treatment of wild-type cells with TGF-α clearly induced p-c-Jun, but induction was completely abolished in the Map3k1ΔKD/ΔKD cells (Fig. 4A). On the other hand, as 17352 | www.pnas.org/cgi/doi/10.1073/pnas.1102297108

previously observed, TGF-α increased c-Jun expression in both wild-type and knockout cells. These in vitro data illustrate two distinct cellular responses to TGF-α: one being the MAP3K1independent expression of c-Jun and the other the MAP3K1dependent c-Jun phosphorylation. Because JNK-mediated c-Jun phosphorylation involves transcriptional activation of AP-1 promoters, we further asked whether the kinase activity of MAP3K1 was required for AP-1 activity. Wildtype and Map3k1ΔKD/ΔKD cells were transfected with an AP-1 binding site-driven luciferase (AP-luc) construct. Treatment of the cells with TGF-α caused a significant increase of luciferase expression in wild-type; however, it failed to do so in knockout cells, suggesting that MAP3K1 was required for optimal AP-1 activation (Fig. 4B). To evaluate whether this was the case in vivo, we generated compound transgenic mice harboring an AP-1-luciferase reporter gene in Map3k1+/ΔKD and Map3k1ΔKD/ΔKD genetic backgrounds. Examination of the E15.5 fetuses showed that the developing eyelid epithelium of heterozygotes had more abundant

Fig. 4. The kinase activity of MAP3K1 is required for activation of the JNK-cJun cascade and maximal AP-1 activity. Wild-type and Map3k1ΔKD/ΔKD fibroblasts were treated with TGF-α for 12 h and were (A) analyzed by Western blotting using antibodies as indicated, and (B) transfected with AP1-luc plasmids together with phRL-TK before treatment. Cell lysates were analyzed for firefly normalized to Renilla luciferase activity and fold-induction was calculated in comparison with untreated cells. Results represent the average of at least two experiments and eight transfections. The Map3k1+/ΔKD/AP-1-luc and Map3k1ΔKD/ΔKD/AP-1-luc (C) fetuses at E15.5 were subjected to immunofluorescence staining to detect the expression of MAP3K1-β-gal (Left, green) and luciferase (Center, red) in the developing eyelid epithelium. Nuclei were stained with DAPI (Right, blue), and (D) pups at P1 were used to isolate eyelid and skin and examined for luciferase activities. (E) RNA from wild-type and Map3k1ΔKD/ΔKD fibroblasts treated with TGF-α as indicated were analyzed by real-time PCR for Pai-1 mRNA. The relative mRNA levels were calculated based on cycle differences (ΔCt) in comparison with that of mouse Gapdh. Fold-induction was calculated in comparison with control. (F) Tissue sections from wild-type and Map3k1ΔKD/ΔKD fetuses at E16 were subjected to immunohistochemistry using anti–PAI-1. PAI-1 expression was identified in the eyelid epithelium (brown, arrows), being more abundant in wild-type than in Map3k1ΔKD/ΔKD fetuses. (Scale bars, 50 μm.) Data are presented as the mean values ± SE from at least four independent experiments. Statistical analyses were done by comparing the mean values in treated cells to those in control cells. *P < 0.05, **P < 0.01.

Geh et al.

Discussion Embryonic eyelid closure is a well-characterized morphogenetic process in the vertebrate eye and has emerged as a powerful system for deciphering signal transduction mechanisms in cell migration and morphogenesis. Based on this system, we show that MAP3K1 is the centerpiece of a signaling network that controls eyelid development. On the one hand, the TGF-α/EGFR-RhoA/ROCKc-Jun pathway converges on the Map3k1 promoter, resulting in elevated MAP3K1 expression; on the other hand, the kinase activity of MAP3K1 is required for activation of the JNK-c-Jun pathway, leading to increased AP-1 activity and target gene expression. Thus, MAP3K1 integrates developmental signals at different molecular levels to coordinate eyelid development (Fig. 5). We show that activation of the EGFR by TGF-α initiates signaling cascades involving ROCK/RhoA that lead to MAP3K1 induction. Genetic evidence indicates that TGF-α and EGFR are essential for developmental eyelid closure. Additionally, ROCKI and ROCKII, the downstream effectors of RhoA, and CDH1, the activator of the anaphase-promoting complex upstream of RhoA, have all been implicated in eyelid development, although the role of RhoA itself had not been established. Ours is a unique demonstration of a genetic interaction between RhoA and MAP3K1. Specifically, we find that RhoA is required for optimal MAP3K1 expression and that its ablation results in a 60% reduction of MAP3K1 protein in the developing eyelid epithelium. Although the 40% remaining in wild-type fetuses may be above the critical threshold level required for embryonic eyelid closure, the amount left in the hemizygotes is below this critical level and, consequently, RhoA ablation delays eyelid closure in the hemizygotes. In addition to its role in MAP3K1 expression, RhoA may also

regulate MAP3K1 activity (26), and in either case, RhoA functions as a partner of MAP3K1 signaling. Distinct from the other eyelid closure factors, however, RhoA ablation by itself does not perturb developmental eyelid closure. It is therefore likely that factors functionally redundant with RhoA exist in the EGFR/ ROCK-MAP3K1 cascades during eyelid development (Fig. 5). Activation of the EGFR-ROCK/RhoA cascades leads to induction of c-Jun (27). c-Jun has an established role in embryonic eyelid closure, as keratinocyte-specific c-Jun loss leads to EOB (13, 14). We show here that c-Jun directly binds to the Map3k1 promoter. Engagement of c-Jun with the promoter is associated with a number of epigenetic marks, characteristic of open chromatin structure and promoter activation. Based on the observations that MAP3K1 expression correlates with c-Jun protein levels in Map3k1ΔKD/ΔKD cells, where c-Jun phosphorylation is basically abolished, we suggest that MAP3K1 expression depends on the presence of the c-Jun protein but not on its phosphorylation (28). c-Jun, acting in a phosphorylation independent fashion, thus serves as a molecular bridge connecting the EGFR-ROCK/RhoA cascades to MAP3K1 expression. Once expressed, the enzyme activities of MAP3K1 are required for TGF-α to activate the JNK-c-Jun cascade. The involvement of JNK in embryonic eyelid closure is supported by genetic evidence that the Jnk1−/− Jnk2+/− mice are born with open eyelids (29). Furthermore, MAP3K1 is shown haploinsufficient for eyelid closure in Jnk1−/− and Jnk1+/−Jnk2+/− mice, suggesting the existence of a MAP3K1-JNK axis in eyelid development (10). Activation of this axis leads to phosphorylation of c-Jun at serine 63/73 (18). We show that the enzymatic activity of MAP3K1 is indeed crucial for the induction of c-Jun phosphorylation, as well as of several phospho-c-Jun–dependent events, such as AP-1 activity and PAI-1 expression. Neverthelsss, neither the transgenic mice expressing a phosphorylation site mutated c-Jun nor the Pai-1 knockout mice display EOB, raising the possibility that c-Jun phosphorylation and AP-1 activation are not critical for eyelid closure (30, 31). The MAP3K1-JNK signaling cascades must regulate developmental eyelid closure through other downstream effectors. Transient lid closure and reopening is a common morphogenetic event that also takes place in humans. Unlike in mice, however, eyelid closure and reopening in humans is accomplished in utero, making its deficiency difficult to detect. Clinical diagnosis of eyelid closure defects and identification of possible associated developmental diseases present a challenge that may have to rely on a genetic approach for resolution. In this context, the genetic mouse models with an easily traceable phenotype have led to the identification of a great number of molecular players involved in eyelid development. This information may help to decipher the “genetic codes” underlying eyelid developmental defects in humans, leading to unravel the origins of congenital developmental disorders in children. Experimental Procedures Details of antibodies, reagents, plasmids, experimental animal, immunohistochemistry, whole mount X-gal staining, LCM, preparation of mouse embryonic fibroblasts, Western blotting, transfection, reverse transcription and real-time PCR, and ChIP may be found in SI Experimental Procedures. Cloning of the Mouse Map3k1 promoter. The mouse bacterial artificial chromosome clone (RP-24–114k21) obtained from BACPAC Resources CHORI (Oakland Research Institute) was used as template for PCR to amplify the promoter region of mouse Map3k1 gene. The PCR fragment was subcloned into the promoterless pGL3-Basic vector (Promega) containing the firefly luciferase reporter (luc) gene. PCR-based cloning was performed to create 5′ deletion constructs of the Map3k1 promoter.

Fig. 5. MAP3K1 orchestrates multiple signals for eyelid closure. Graphic illustration of the MAP3K1 signaling network in the developing eyelid epithelial cells.

Geh et al.

Statistical Analysis. Statistical comparisons were performed with Student’s two-tailed paired t test and ANOVA. Values of *P < 0.05, **P < 0.01, and ***P < 0.001 were considered statistically significant.

PNAS | October 18, 2011 | vol. 108 | no. 42 | 17353

DEVELOPMENTAL BIOLOGY

luciferase expression than that of Map3k1ΔKD/ΔKD fetuses (Fig. 4C). Measurement of luciferase activity in lysates of eyelid tissue confirmed this observation (Fig. 4D). The luciferase activity in the skin tissues, however, was not affected by MAP3K1 inactivation. AP-1 activation leads to the induction of downstream target genes, one of which, the plasminogen activator inhibitor 1 (Pai1), is induced by TGF-α in a manner dependent on the JNK-AP1 axis (23). To examine whether MAP3K1 was upstream of this axis, we measured PAI-1 expression in the presence or absence of MAP3K1. Although Pai-1 mRNA was induced significantly by TGF-α in wild-type fibroblast, the induction was totally abolished in Map3k1ΔKD/ΔKD cells (Fig. 4F). Examination of the developing eyelids led to the same conclusion. PAI-1 expression was abundant in the developing eyelid epithelium of the wild-type fetuses, but it was almost undetectable in the eyelids of the Map3k1ΔKD/ΔKD fetuses (Fig. 4E). Both in vitro and in vivo findings lead us to conclude that the kinase activity of MAP3K1 is required for activation of the JNK-c-Jun pathway, induction of AP-1 activity, and PAI-1 expression during eyelid development.

ACKNOWLEDGMENTS. We thank Drs. J. Tichelaar for providing the AP-1Luc mice, A. Aronheim for c-Jun expression vectors, and C. S. Baxter for reading the manuscript. This research was supported by National

Institutes of Health Grant EY15227 (to Y.X.), and National Institute on Environmental Health Sciences Grants T32 ES07250 and F31EY019458 (to E.G.).

1. Davis RJ (1995) Transcriptional regulation by MAP kinases. Mol Reprod Dev 42: 459–467. 2. Geh GN, Jin C, Xia Y (2010) Map3k1. UCSD Nature Molecular Pages. Available at http://www.signaling-gateway.org/molecule/query?afcsid=A001512. 3. Xia Y, Karin M (2004) The control of cell motility and epithelial morphogenesis by Jun kinases. Trends Cell Biol 14(2):94–101. 4. Findlater GS, McDougall RD, Kaufman MH (1993) Eyelid development, fusion and subsequent reopening in the mouse. J Anat 183(1):121–129. 5. Luetteke NC, et al. (1993) TGF alpha deficiency results in hair follicle and eye abnormalities in targeted and waved-1 mice. Cell 73:263–278. 6. Mine N, Iwamoto R, Mekada E (2005) HB-EGF promotes epithelial cell migration in eyelid development. Development 132:4317–4326. 7. Huang J, et al. (2009) FGF-regulated BMP signaling is required for eyelid closure and to specify conjunctival epithelial cell fate. Development 136:1741–1750. 8. Gage PJ, Qian M, Wu D, Rosenberg KI (2008) The canonical Wnt signaling antagonist DKK2 is an essential effector of PITX2 function during normal eye development. Dev Biol 317:310–324. 9. Zhang L, et al. (2003) A role for MEK kinase 1 in TGF-beta/activin-induced epithelium movement and embryonic eyelid closure. EMBO J 22:4443–4454. 10. Takatori A, et al. (2008) Differential transmission of MEKK1 morphogenetic signals by JNK1 and JNK2. Development 135(1):23–32. 11. Naoe H, et al. (2010) The anaphase-promoting complex/cyclosome activator Cdh1 modulates Rho GTPase by targeting p190 RhoGAP for degradation. Mol Cell Biol 30: 3994–4005. 12. Thumkeo D, Shimizu Y, Sakamoto S, Yamada S, Narumiya S (2005) ROCK-I and ROCKII cooperatively regulate closure of eyelid and ventral body wall in mouse embryo. Genes Cells 10:825–834. 13. Li G, et al. (2003) c-Jun is essential for organization of the epidermal leading edge. Dev Cell 4:865–877. 14. Zenz R, et al. (2003) c-Jun regulates eyelid closure and skin tumor development through EGFR signaling. Dev Cell 4:879–889. 15. Uda M, et al. (2004) Foxl2 disruption causes mouse ovarian failure by pervasive blockage of follicle development. Hum Mol Genet 13:1171–1181. 16. McHenry JZ, Leon A, Matthaei KI, Cohen DR (1998) Overexpression of fra-2 in transgenic mice perturbs normal eye development. Oncogene 17:1131–1140. 17. Yu Z, et al. (2008) Grainyhead-like factor Get1/Grhl3 regulates formation of the epidermal leading edge during eyelid closure. Dev Biol 319(1):56–67.

18. Yujiri T, et al. (2000) MEK kinase 1 gene disruption alters cell migration and c-Jun NH2-terminal kinase regulation but does not cause a measurable defect in NF-kappa B activation. Proc Natl Acad Sci USA 97:7272–7277. 19. Xia Y, et al. (2000) MEK kinase 1 is critically required for c-Jun N-terminal kinase activation by proinflammatory stimuli and growth factor-induced cell migration. Proc Natl Acad Sci USA 97:5243–5248. 20. Fujii S, Hatakenaka N, Kaneda M, Teramoto S (1995) Morphogenetic study of the eyelids in NC-eob mice fetuses with an open-eyelid malformation at birth. Lab Anim Sci 45(2):176–180. 21. Tao H, Ono K, Kurose H, Noji S, Ohuchi H (2006) Exogenous FGF10 can rescue an eyeopen at birth phenotype of Fgf10-null mice by activating activin and TGFalpha-EGFR signaling. Dev Growth Differ 48:339–346. 22. Vassalli A, Matzuk MM, Gardner HA, Lee KF, Jaenisch R (1994) Activin/inhibin beta B subunit gene disruption leads to defects in eyelid development and female reproduction. Genes Dev 8:414–427. 23. Samarakoon R, Higgins CE, Higgins SP, Higgins PJ (2009) Differential requirement for MEK/ERK and SMAD signaling in PAI-1 and CTGF expression in response to microtubule disruption. Cell Signal 21:986–995. 24. Ashery-Padan R, Marquardt T, Zhou X, Gruss P (2000) Pax6 activity in the lens primordium is required for lens formation and for correct placement of a single retina in the eye. Genes Dev 14:2701–2711. 25. Vettese-Dadey M, et al. (1996) Acetylation of histone H4 plays a primary role in enhancing transcription factor binding to nucleosomal DNA in vitro. EMBO J 15: 2508–2518. 26. Gallagher ED, Gutowski S, Sternweis PC, Cobb MH (2004) RhoA binds to the aminoterminus of MEKK1 and regulates its kinase activity. J Biol Chem 279:1872–1877. 27. Marinissen MJ, et al. (2004) The small GTP-binding protein RhoA regulates c-jun by a ROCK-JNK signaling axis. Mol Cell 14(1):29–41. 28. Radler-Pohl A, et al. (1993) The activation and activity control of AP-1 (fos/jun). Ann N Y Acad Sci 684:127–148. 29. Weston CR, et al. (2003) JNK initiates a cytokine cascade that causes Pax2 expression and closure of the optic fissure. Genes Dev 17:1271–1280. 30. Providence KM, Higgins PJ (2004) PAI-1 expression is required for epithelial cell migration in two distinct phases of in vitro wound repair. J Cell Physiol 200:297–308. 31. Behrens A, Sibilia M, Wagner EF (1999) Amino-terminal phosphorylation of c-Jun regulates stress-induced apoptosis and cellular proliferation. Nat Genet 21:326–329.

17354 | www.pnas.org/cgi/doi/10.1073/pnas.1102297108

Geh et al.