ADAM17 Transactivates EGFR Signaling during Embryonic Eyelid Closure

Abstract

Purpose.: During mammalian embryonic eyelid closure ADAM17 has been proposed to play a role as a transactivator of epidermal growth factor receptor (EGFR) signaling by shedding membrane bound EGFR ligands. However, ADAM17 also sheds numerous other ligands, thus implicating ADAM17 in additional molecular pathways. The goal of this study was to experimentally establish the role of ADAM17 and determine ADAM17-mediated pathways essential for the embryonic eyelid closure.

Methods.: Wild-type (WT) and woe mice, carrying a hypomorphic mutation in Adam17, were evaluated using H&E and scanning electron microscopy. Expressions of ADAM17, EGFR, and the phosphorylated form EGFR-P were evaluated using immunohistochemistry. BrdU and TUNEL assays were used to evaluate cell proliferation and apoptosis, respectively. In vitro scratch assays of primary cultures were used to evaluate cell migration. Clinical and histologic analyses established if the hypermorphic Egfr^Dsk5 allele can rescue the woe embryonic eyelid closure.

Results.: woe mice exhibited a failure to develop the leading edge of the eyelid and consequently failure of the embryonic eyelid closure. Expression of ADAM17 was identified in the eyelid epithelium in the cells of the leading edge. ADAM17 is essential for epithelial cell migration, but does not play a role in proliferation and apoptosis. EGFR was expressed in both WT and woe eyelid epithelium, but the phosphorylated EGFR-P form was detected only in WT. The Egfr^Dsk5 allele rescued woe eyelid closure defects, but also rescued woe anterior segment defects and the absence of meibomian glands.

Conclusions.: We provide in vivo genetic evidence that the role of ADAM17 during embryonic eyelid closure is to transactivate EGFR signaling.

Introduction

ADAM17 belongs to a disintegrin and metalloproteases (ADAMs) family of type I transmembrane Zn²⁺-dependent proteases that play a role in the proteolytic ectodomain release or “shedding” of membrane-tethered precursors. ¹ The role of ADAM17 has been shown to be a sheddase of various proteins, including cytokines and cytokine receptors, growth factors, and adhesion molecules. In fact, as many as 76 proteins have been shown to be substrates for ADAM17 shedding activity. ² With such functional diversity of substrates, ADAM17 has been implicated in many developmental and disease mechanisms. ³ Analysis of ADAM17 function in vivo showed that most Adam17 ^−/− mice die between embryonic day (E) 17.5 and birth, ^4,5 but some Adam17 ^−/− mice survive until weaning. ⁴ Phenotypically, Adam17 ^−/− mice exhibit heart valve abnormalities; eyelids open at birth (EOB); and defects in maturation of epithelia from lung, skin, and mammary glands, as well as waviness of the fur and vibrissae. ^4,5 Recently, an Adam17 c.C794T substitution was identified in a mouse termed waved with open eyelids (woe) that resulted in aberrant Adam17 splicing and severely reduced ADAM17 sheddase activity. ⁶ Although severely reduced, the ADAM17 sheddase activity was sufficient to reduce the severity of cardiac valve phenotypes associated with the absence of ADAM17 function, thus allowing woe mice to survive into adulthood. ^4–6 In addition to mild cardiac abnormalities, woe mice exhibit EOB and wavy fur phenotypes ⁶ similar to those of Adam17 ^−/− mice. ^4,5 The survival of woe mice into adulthood facilitated the identification of previously unknown phenotypes associated with ADAM17 deficiency such as ocular anterior segment abnormalities and the absence of meibomian glands. ⁶  

The presence of the EOB phenotype in woe and Adam17 ^−/− mice provided convincing evidence that ADAM17 is critical for proper embryonic eyelid closure; however, the precise role of ADAM17 during this process has not yet been established. Embryonic eyelid closure is common to all mammals where, following formation, eyelids migrate across the cornea, fuse together, and ultimately reopen. ⁷ Closed embryonic eyelids serve as a protective barrier by preventing premature exposure of the developing ocular structures to the environment. Both epidermal growth factor receptor (EGFR) signaling and transforming growth factor beta (TGFβ)/activin-mediated MAP3K1/JNK/cJUN signaling pathways have emerged to be essential in facilitating eyelid epithelial cell migration and proper embryonic eyelid closure. Failure of the embryonic eyelid closure is present in mice that carry mutations resulting in abolished or severely reduced EGFR signaling. ^8–17 Similarly, mice with mutations in genes encoding members of the TGFβ/activin-mediated MAP3K1/JNK/cJUN signaling also exhibit EOB phenotypes. ^18–22 Although both signaling paAlthough both signaling pathways are involved in epithelial cell migration, it remains unclear how these two signaling pathways are coordinated. ADAM17 has been proposed as a sheddase of EGFR ligands during embryonic eyelid closure and, consequently, a transactivator of EGFR signaling. ⁹ However, in cancer cells, ADAM17 has been implicated in the regulation of TGFβ. ²³ Therefore, it is possible that during embryonic eyelid closure, in addition to shedding EGFR ligands and transactivation of EGFR signaling, ADAM17 may be also regulating TGFβ and, consequently, transactivating TGFβ/activin-mediated MAP3K1/JNK/cJUN signaling. Therefore, ADAM17 may play a critical role in coordinating these two molecular pathways. 

As a part of this study, our goal was to experimentally clarify the role of ADAM17 during embryonic eyelid closure. In addition, our goal was to establish the molecular pathways and mechanisms governed by ADAM17 during this process. Our results showed that ADAM17 is indispensable for the formation of the leading edge at the tips of developing eyelids just prior to the initiation of eyelid closure. In addition, we demonstrate that ADAM17 is essential for cell migration, but not for proliferation or apoptosis. Finally, we provide genetic evidence that ADAM17 regulates embryonic eyelid closure through direct transactivation of the EGFR signaling pathway. 

Materials and Methods

Mice, Clinical Evaluation, and PCR Genotyping

C57BL/6J and woe mice were maintained as previously described. ⁶ Dsk5 mice were provided by David Threadgill, PhD, from the University of North Carolina. All mice were used with strict adherence to the guidelines set forth in the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. For clinical evaluation, mouse eyes were evaluated as previously described. ^24,25 The Egfr^Dsk5 and Adam17^woe alleles were genotyped as previously described. ^6,26  

Histology, Immunohistochemistry, and Electron Microscopy

Embryonic day 0.5 (E0.5) was defined as the morning of the day that a vaginal plug was first observed. Embryonic heads were collected, fixed, embedded in paraffin, serially sectioned, and stained with hematoxylin and eosin (H&E) as previously described. ⁶ Antigen retrieval was performed by boiling samples in a microwave oven for 10 minutes in sodium citrate (10 mM, pH 6.0). Endogenous peroxidase activity was quenched using 1% H₂O₂ in 100% methanol (40 minutes, room temperature) followed by 3 × 5-minute washes in PBS and 2 × 10-minute washes in 0.3% Triton/PBS. For immunostaining, a commercial immunodetection kit (Vector Mouse on Mouse [M.O.M.] Immunodetection Kit; Vector Laboratories, Burlingame, CA) was used with anti-ADAM17 (Abcam, Cambridge, MA), anti-EGFR (Cell Signaling Technology [CST], Beverly, MA), and anti-EGFR-P (CST) antibodies at 4°C O/N following manufacturer's protocol. A commercial ready-to-use reagent (VECTASTAIN Universal Elite ABC Kit; Vector Laboratories) was used following the manufacturer's protocol. Tyramide-Cy3 was then added to the horseradish peroxidase using a tyramide signal amplification (TSA)-cynanine 3 kit (NEL 744; PerkinElmer, Inc., Waltham, MA) at a dilution of 1:50 for 3.5 minutes, and counterstained with 4′,6-diamidino-2-phenylindole (DAPI; Life Technologies, Carlsbad, CA). Slides were then mounted and photographed with a digital camera (Nikon DS-Fi1 Camera; Nikon Instruments Inc., Melville, NY) on a digital microscope (Nikon Eclipse 80i; Nikon Instruments Inc.). Tissues for electron microscopy were fixed in a 2% glutaraldehyde solution and then processed for scanning electron microscopy as previously described ²⁷ and photographed with a scanning electron microscope (XL30; FEI/Philips, Tustin, CA). 

Isolation and Culture of Keratinocytes, Dermis Fibroblasts, and Mouse Embryonic Fibroblasts

Primary mouse embryonic fibroblasts (mEFs) were generated and cultured as described previously. ⁶ Isolation and culture of keratinocytes and fibroblasts were performed following a previously established protocol. ²⁸ The keratinocytes were cultured in K-SFM medium supplemented with bovine pituitary extract, and 10 ng/mL EGF, 1.3 mM calcium, or 0.05 mM calcium, and 1% penicillin–streptomycin (Life Technologies). Dermal fibroblasts were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and 1% penicillin–streptomycin (Life Technologies). 

BrdU Incorporation and TUNEL Assay

Mice were injected intraperitoneally at E15.5 with 100 μg/g body weight of 5-bromo-2′-deoxyuridine (BrdU) and euthanized 2 hours later. Embryo heads were collected, fixed in 4% paraformaldehyde (PFA), paraffin-embedded, and sectioned using standard protocols. Immunostaining for BrdU was carried out using a commercial staining kit (Zymed BrdU Kit; Life Technologies) following the manufacturer's protocol. The numbers of BrdU-positive cells in the eyelids were counted as previously described. ²⁹ The means and SEM were calculated and deemed significant if P < 0.05 as determined by a Student's t-test. To measure the proliferation rate in cell culture, WT and woe keratinocytes or WT and woe fibroblasts were grown to 80% confluency and incubated with 50 μg/mL BrdU at 37°C in 5% CO₂ for 4.5 hours, fixed in 70% cold methanol, and immunostained using a commercial kit (Zymed BrdU Kit; Life Technologies) following the manufacturer's protocol. The percentage of BrdU-positive cells as compared with total cell number was then calculated and deemed significant if P < 0.05, as determined by a Student's t-test. 

A TUNEL assay was done with a commercial detection kit (ApopTag Peroxidase In Situ Apoptosis Detection Kit; EMD Millipore Corp., Billerica, MA) following the manufacturer's protocol. Following TUNEL assay slides were counterstained with hematoxylin. TUNEL-positive cells were counted. The means and SEM were calculated and considered significant if P < 0.05, as determined by Student's t-test. 

In Vitro Scratch Assay

WT and woe mEFs or keratinocytes were seeded onto 12-well culture dishes, scratched with a 200 μL pipette tip, and cultured in medium containing no supplements, or containing medium + TGFα (10 ng/mL), or containing medium + TGFβ₁ (10 ng/mL). Images were taken immediately after scratching and 48 hours later. The number of cells that migrated into the gap was counted in three random fields at 50 different positions. ^28,30 Differences were deemed significant if P < 0.05, as determined by an unpaired t-test. For F-actin immunostaining, cells were rinsed twice with PBS, fixed with fresh methanol-free 3.7% PFA for 10 minutes, permeabilized with 0.1% Triton-X for 5 minutes, blocked with 1% bovine serum albumin for 30 minutes, and incubated with phalloidin conjugated to Alexa Fluor 488 (Life Technologies). DNA was counterstained with DAPI (Life Technologies), and photographed with a digital camera on a digital microscope (Nikon DS-Fi1 and Nikon Eclipse 80i, respectively; Nikon Instruments Inc.). 

Results

ADAM17 Is Essential for the Formation of the Leading Edge during Embryonic Eyelid Closure

As the initial step in determining the role of ADAM17 during embryonic eyelid closure, we set out to morphologically examine the embryonic eyelid closure defect in woe mice. At E13.5, the primitive eyelids formed and no obvious morphologic differences were noted between WT and woe mice (Figs. 1A, 1E). By E15.5, the leading edges formed and started migrating toward each other in WT mice (Fig. 1B); in contrast, the leading edge failed to form in woe mice (Fig. 1F). In WT mice, the leading edges met and formed the eyelid closure by E16.5 (Fig. 1C), whereas in woe mice the eyelids remained wide open, indicating failure of eyelid closure (Fig. 1G). Embryonic eyelid closure was complete by E18.5 in the WT mice (Fig. 1D), whereas in woe mice the eyelids remained open (Fig. 1H). These findings indicated that the initial defect observed in woe mice was failure of the formation of the leading edge observed at E15.5. Thus, we sought to further investigate morphologic differences between WT and woe eyelids at this developmental time point using scanning electron microscopy (SEM). Eyelids of E15.5 WT mice exhibited fully formed leading edges that were migrating across the cornea (Fig. 2A); higher magnification showed the presence of rounded peridermal cells present at the rims of the leading edges (Fig. 2A′). In contrast, E15.5 woe eyelids were fully open, indicating the absence of formation of even rudimentary leading edges (Fig. 2B); higher magnification confirmed this absence of the leading edge, and identified a few rounded peridermal cells at the rims of woe eyelids (Fig. 2B′). By E16.5, eyelid closure was completed in WT mice (Fig. 2C), and the eyelid junction was formed with flattened cells present at the eyelid junction (Fig. 2C′). At E16.5, woe eyelids remained open and morphologically similar in appearance to the woe eyelids observed at E15.5 (Fig. 2D), although more peridermal cells were present at the woe eyelid rims at E16.5 (Fig. 2D′) as compared with woe eyelid rims at E15.5 (Fig. 2B′). 

Figure 1. 

 

H&E staining of eyelids from WT (top) and woe (bottom) mice. At E13.5, no obvious morphologic differences were observed between WT (A) and woe (E) mice. At E15.5, the leading edge formed in WT mice (B), although in woe mice the leading edge did not form (F). By E16.5, the leading edges met and formed a junction in WT mice (C), whereas eyelids in woe mice remained open (G). Further histologic analysis of woe mice at E18.5 (H) showed the failure of embryonic eyelid closure, in contrast to WT mice that exhibited completed embryonic eyelid closure (D). Scale bars, 50 μm.

Figure 2. 

 

Scanning electron micrographs showing absence of the leading edge formation in woe embryos. In E15.5 WT mice (A), the formed leading edges were migrating toward each other. The accumulation of rounded peridermal cells was evident at the tips of the leading edges (A′). In contrast, woe embryos (B) exhibited the absence of even rudimentary leading edges, with only a few rounded peridermal cells present at the rims of the eyelids (B′). By E16.5 in WT embryos (C), the leading edges met, forming the junction (C′), whereas in woe mice (D) the eyelids remained open with a few rounded peridermal cells present at the eyelid margins (D′). Scale bars: 200 μm (A–D); 50 μm (A′–D′).

ADAM17 Expression in the Developing Eyelid

At E13.5, ADAM17 was expressed in the palpebral epithelia, although weak ADAM17 expression was observed in both bulbar epithelial and mesenchymal cells of the eyelid (Fig. 3). At E15.5, ADAM17 was highly expressed in the cells of the leading edge, in addition to the palpebral epithelia of the eyelid (Fig. 3). By E16.5, ADAM17 remained highly expressed in the palpebral epithelia of the eyelid as well as in the cells at the eyelid junction (Fig. 3). By P0.5, ADAM17 was expressed only in the epidermis of the eyelid (Fig. 3). 

Figure 3. 

 

ADAM17 expression in the developing eyelid. In the E13.5 eyelid, ADAM17 (green) was expressed in the palpebral epidermis. At E15.5, in addition to the palpebral epidermis, ADAM17 is also highly expressed in the cells of the leading edge. As the eyelids fuse at E16.5, ADAM17 remains expressed in the palpebral epidermis as well as in the cells of the eyelid junctions. By P0.5, ADAM17 is expressed only in the palpebral epidermis. DAPI (blue) was used as a nuclear stain. Scale bars, 50 μm.

ADAM17 Does Not Contribute to Cell Proliferation or Cell Death during Embryonic Eyelid Closure

We examined cell proliferation in WT and woe eyelids at E15.5 by measuring BrdU incorporation. There was no significant difference between the number of BrdU-positive cells within the epidermis or dermis of WT or woe eyelids (Figs. 4A, 4B). To further confirm whether ADAM17 regulates cell proliferation, we isolated primary keratinocytes and dermal fibroblasts from WT and woe mice at P0.5 and cultured them in the presence of BrdU. These results showed that there was no significant difference between WT and woe in the number of BrdU-positive keratinocytes or dermal fibroblasts (Figs. 4C, 4D). We also examined if ADAM17 may play a role as a regulator of apoptosis during embryonic eyelid closure. However, neither WT nor woe E15.5 eyelids exhibited any presence of TUNEL-positive cells (not shown). These results indicate that ADAM17 is not involved in cell proliferation or apoptosis during embryonic eyelid closure. 

Figure 4. 

 

Proliferation in WT and woe embryonic eyelids. BrdU incorporation in nuclei of cells from E15.5 WT and woe upper lids (UL) and lower lids (LL) (A). BrdU-positive (brown) cells were observed in both epidermis and dermis. The number of BrdU-positive cells did not differ between WT and woe E15.5 eyelids (B). The percentage of BrdU-positive cells was calculated from three WT and three woe embryos. No difference in proliferation was observed in keratinocytes (KC) or fibroblasts (FB) from WT and woe mice (C). Quantification of BrdU-positive cells from (C) is shown in (D). Data represent the mean ± SEM. Scale bars, 50 μm. N.S., not significant.

ADAM17 Is Involved in Cell Migration

As the initial step in evaluation of involvement of ADAM17 during cell migration, we performed an in vitro scratch assay in WT and woe mEFs. Previous studies have shown that ADAM17 is expressed in mEFs. ⁶ Fewer woe mEFs migrated into the gap 48 hours after the scratch as compared with WT mEFs (Figs. 5A, 5B). During embryonic eyelid closure, only the epidermal layer is involved in eyelid fusion, ¹² with keratinocytes as the major constituent of eyelid epidermis. Therefore, we evaluated whether keratinocytes from woe mice exhibited a defect in cell migration similar to the one observed in woe mEFs. The results showed that woe keratinocytes also exhibited fewer migrating cells following the scratch when compared with keratinocytes from WT mice (Fig. 5B). Next, we investigated the filopodia formation by staining mEFs and keratinocyte cells with phalloidin. In WT mEFs and keratinocytes, filopodia were abundantly present (Fig. 5C). In contrast, fewer filopodia were present in woe mEFs and keratinocytes that appeared to be shorter in length (Fig. 5C). TGFα is a well-established substrate of ADAM17³; therefore, we hypothesized that exogenous addition of TGFα may rescue the cell migration defect. Exogenous addition of TGFα indeed rescued the migration defect identified in both woe mEFs (Fig. 5D) and keratinocytes (not shown). Interestingly, exogenous addition of TGFβ also rescued the cell migration defect in woe mEFs (Fig. 5D) and keratinocytes (not shown). 

Figure 5. 

 

ADAM17 is essential for cell migration. Confluent monolayers of mEFs shown in (A) and primary epidermal keratinocytes (not shown) cultured in medium with no growth factors were subjected to in vitro scratch assays. Photographs were taken immediately and 48 hours following the scratch (A). Quantification of cells that migrated into the scratch is summarized in (B). Both mEFs and keratinocytes were stained with phalloidin conjugated to Alexa Fluor 488 for F-actin (green) and DAPI for nuclei (blue). Filopodia were abundantly present in WT mEFs and keratinocytes, but not in woe mEFs and keratinocytes (C). Exogenous addition of TGFα (10 ng/mL) or TGFβ (10 ng/mL) rescued the cell migration defect in woe mEFs (D) and woe keratinocytes (not shown). Scale bars, 50 μm. Asterisks indicated that the difference in cell migration between WT and woe mEFs and keratinocytes was significant (P < 0.05).

ADAM17 Activates EGFR Signaling during Embryonic Eyelid Closure

We set out to investigate if ADAM17 mediates EGFR signaling during embryonic eyelid closure. Immunohistochemical analysis identified expression of EGFR (Fig. 6A) and phosphorylated EGFR-P (Fig. 6C) in WT palpebral epithelia and in the leading edge cells. In woe eyelids EGFR was also expressed in the palpebral epithelia (Fig. 6B), but expression of phosphorylated EGFR-P was severely reduced (Fig. 6D). These findings suggested that at E15.5 just prior to the eyelid closure, ADAM17 facilitates EGFR phosphorylation. Next, we evaluated if ADAM17 mediates only EGFR signaling in vivo or if ADAM17 additionally mediates other signaling pathways essential for embryonic eyelid closure. Dsk5 is a hypermorphic mutation in the Egfr gene, resulting in constitutively active EGFR. ²⁶ Dsk5 mice exhibit normal eyelid closure and ocular development (not shown). Our results revealed that in woe mice (Figs. 7A, 7E), the presence of a single Egfr^Dsk5 allele resulted in partial eyelid closure, although the eyelid junction did not appear to be fully formed (Fig. 7B); histologic analysis showed that in these mice the eyelid junction was formed at the eye periphery (not shown), but not at the center of the eyelid (Fig. 7F). Mice homozygous for both woe and Egfr^Dsk5 alleles exhibited fully closed eyelids (Fig. 7C), similar in appearance to that of WT mice (Fig. 7D). Histologic analysis confirmed fully formed eyelid junctions in these mice (Fig. 7G), which were morphologically indistinguishable from the eyelid junctions formed in WT mice (Fig. 7H). 

Figure 6. 

 

EGFR and EGFR-P expression in E15.5 eyelids. EGFR is expressed in the palpebral epidermis in the eyelids in WT (A) and woe (B) mice; in WT eyelids EGFR is also expressed in the cells of the leading edge (A). Expression of phosphorylated EGFR-P is also identified in the palpebral epidermis and in the cells of the leading edge of the WT eyelids (C). In contrast, in woe epidermis expression of EGFR-P is severely reduced (D). Scale bars, 50 μm.

Figure 7. 

 

The Egfr^Dsk5 allele rescues the woe EOB phenotype. Newborn woe mice exhibit the EOB phenotype shown in (A); H&E sections of P0.5 woe mice confirm the failure of embryonic eyelid closure (E). Newborn mice carrying a homozygous woe allele and a single Egfr^Dsk5 allele exhibit less severe EOB phenotype (B) than that of woe mice (A). H&E sections of P0.5 mice carrying a homozygote woe allele and a single Egfr^Dsk5 allele show the failure of eyelid closure, although the eyelids appear to be closer together with keratinized tissue filling in the interpalpebral aperture (F). Newborn mice carrying homozygote woe and homozygote Egfr^Dsk5 allele (C) exhibit eyelids that are similar in appearance to that of WT mice (D). Histologic analysis confirms that eyelid closure in mice carrying homozygote woe and homozygote Egfr^Dsk5 alleles, as shown in (G), did not differ morphologically from the eyelid junction observed in WT newborn mice (H). Scale bars, 25 μm.

Egfr^Dsk5 Also Rescues Anterior Segment Gland Defects and the Absence of Meibomian Gland Observed in woe Mice

Clinically, adult woe mice exhibited small eyes with opaque and vascularized corneas (Fig. 8A); histologic analysis identified abnormalities of the corneal epithelium, corneal stroma, the absence of Descement's membrane, and anterior synechiae (Fig. 8E). Clinical evaluation of eyes from woe mice carrying a single Egfr^Dsk5 allele also identified smaller eyes with corneal opacification and neovascularization (Fig. 8B), although the corneal opacity phenotype appeared to be less severe (Fig. 8B) when compared with woe mice carrying both WT Egfr alleles (Fig. 8A). Histologic analysis of woe mice carrying a single Egfr^Dsk5 allele revealed a hypercellular corneal stroma with vascular spaces distorting the corneal lamellae (Fig. 8F). The corneal epithelium in these mice was intact, although slightly irregular, whereas the corneal endothelium (Fig. 8F) and other ocular structures were within normal limits (not shown). Clinical and histologic analysis of woe mice carrying two Egfr^Dsk5 alleles (Figs. 8C, 8G) identified eyes that were clinically within normal limits and morphologically similar in appearance to those of WT eyes (Figs. 8D, 8H), indicating a rescue of the phenotypes associated with ADAM17 deficiency observed in woe mice. 

Figure 8. 

 

The Egfr^Dsk5 allele rescues the woe anterior segment dysgenesis phenotype. Clinical evaluation of woe mice at P28 identified severe corneal opacification and corneal neovascularization (A). Histologic analysis of woe eyes at P28 confirmed severe anterior segment abnormalities, including highly disorganized corneal epithelium, stroma and endothelium, the absence of Descement's membrane, and anterior synechiae as shown in (E). Clinical evaluation of homozygote woe mice carrying a single Egfr^Dsk5 allele also revealed corneal opacities (B) that were less severe than those in woe mice (A). Histologic analysis of homozygote woe mice carrying a single Egfr^Dsk5 allele showed some irregularities of the corneal epithelium, a hypercellular corneal stroma, and abnormalities of the corneal lamellae (F). In woe mice carrying two Egfr^Dsk5 alleles, clinical (C) and histologic (G) analyses identified eyes that were clinically (D) and histologically (H) similar in appearance to that of WT eyes. Scale bars, 25 μm.

We previously identified the absence of the meibomian gland in woe mice. ⁶ Histologic analysis showed that meibomian glands in mice carrying both woe and Dsk5 alleles did not differ from WT mice (not shown). 

Discussion

The main goal of this study was to experimentally establish the role of ADAM17 during embryonic eyelid closure. Even though numerous proteins have been shown to be substrates for ADAM17 sheddase activity, we provide in vivo genetic evidence that the role of ADAM17 during embryonic eyelid closure is to directly transactivate EGFR signaling. EGFR is a cell surface tyrosine kinase receptor that, upon ligand binding, dimerizes and undergoes autophosphorylation events that elicit downstream signaling. Our immunohistochemical analysis demonstrated that prior to the eyelid closure ADAM17 facilitates EGFR phosphorylation. Additionally, the hypermorphic Egfr^Dsk5 allele fully rescued the woe EOB phenotype. Egfr^Dsk5 mice carry a point mutation in Egfr that results in p.Leu863Gln substitution within a three-residue β-strand that helps stabilize the EGFR activation loop and, consequently, results in constitutive EGFR transactivation. ²⁶ Functional analysis showed that the Egfr^Dsk5 allele rescues phenotypes in wavy 2 (wa2) mice carrying a hypomorphic Egfr mutation, further confirming that Egfr^Dsk5 is a hypermorphic Egfr allele. ²⁶ Although Egfr^Dsk5 mice exhibit defects in skin pigmentation, hair texture, and epidermis, ²⁶ they exhibit normal eyelid closure as well as normal ocular development. The eyelid phenotypes in newborn pups carrying both woe and Egfr^Dsk5 alleles were indistinguishable from eyelid phenotypes in WT mice, thereby establishing the role of ADAM17 during embryonic eyelid closure as a transactivator of EGFR signaling. However, it has been recently shown that EGFR signaling induces cJUN-mediated activation of MAP3K1 during embryonic eyelid closure, which in turn activates the JNK–cJUN pathway. ³¹ This suggests that ADAM17 may be involved in regulating the MAP3K1-mediated JNK/cJUN pathway, but only indirectly via its transactivation of EGFR signaling. 

EGFR ligands include TGFα, amphiregulin, HBEGF, betacellulin, epiregulin, and epigen. ³² Although membrane-bound EGFR ligands can facilitate juxtacrine or autocrine signaling, paracrine EGFR signaling facilitated by ligand shedding is most likely critical for embryonic eyelid closure. ^1,33,34 It has been proposed that complete embryonic eyelid closure requires cooperative binding of several EGFR ligands to the EGF receptor. ⁹ This comes out of the observation that mice carrying null alleles of Egfr exhibit a fully penetrant EOB phenotype. ^15,17,35 In contrast, Hbegf ^−/− and Tgfα ^−/− mice exhibit an incompletely penetrant EOB phenotype. ⁹ Interestingly, mice carrying both Hbegf and Tgfα null alleles exhibit the EOB phenotype with increased penetrance as compared with each null genotype independently, although double Hbegf and Tgfα null allele mice still exhibit an incompletely penetrant EOB phenotype. ⁹ Mice carrying triple null alleles of Egf, amphiregulin, and Tgfα also exhibit an incompletely penetrant EOB phenotype. ¹⁰ It has been shown that in tissue culture ADAM17 preferentially sheds epiregulin, Tgfα, amphiregulin, and HBEGF, ^36,37 whereas ADAM10 preferentially sheds EGF and betacellulin. ³⁷ Taken together it is unclear, in addition to HBEGF and TGFα, which additional EGFR ligands are involved in vivo during embryonic eyelid closure and if ADAM17 is the only sheddase that transactivates EGFR signaling during this process. The fully penetrant EOB phenotype observed in woe and Adam17 ^−/− implies that ADAM17 is probably the primary sheddase involved in EGFR ligand shedding, although this requires further investigation. 

Results from our study show that ADAM17 is indispensable for formation of the leading edge and epithelial cell migration. This is consistent with previous findings that the role of EGFR signaling is in the formation of the leading edge and epithelial cell migration. ⁹ Eyelid closure is a process that shares specific morphogenetic steps with wound healing as well as dorsal closure in Drosophila that include the formation of a leading edge of epithelial cells, formation of filopodia and actin stress fibers, cell migration, and reepithelization. ³⁸ The leading edge is a structure composed of epithelial cells that protrudes from the tip of the embryonic eyelids just prior to the cell migration. ⁷ As the leading edges migrate toward each other, rounded peridermal cells accumulate on the surface of the leading edges. ⁷ Our results show that ADAM17 is highly expressed in the developing eyelid epithelium prior to formation of the leading edge, as well as in the cells of the leading edge. The absence of even rudimentary leading edge structures in woe embryonic eyelids resembles the absence of leading edges observed in eyelids from Fgf10 ^−/− and Fgfr2 ^−/− mice that exhibit EOB phenotypes. ^12,39 FGF10 expressed in the developing eyelid mesenchyme, via its receptor FGFR2 expressed in the eyelid epithelium, directly regulates EGFR signaling by regulating expression of TGFα. ¹² However, in the skin it has been shown that EGFR-mediated keratinocyte migration is dependent on FGFR2 stimulation of ADAM17. ⁴⁰ Therefore, we propose that during embryonic eyelid closure, in addition to regulating expression of TGFα, FGF10/FGFR2 signaling may also be directly regulating ADAM17 sheddase activity. Our results also show that ADAM17 facilitates epithelial cell migration as well as filopodia formation, which is consistent with its role as an EGFR transactivator. It should be noted, however, that the exogenous addition of TGFβ rescued the woe cell migration defect. Previous studies showed that in vitro activation of either TGFα-mediated EGFR signaling or TGFβ/activin-mediated MAP3K1/JNK/cJUN signaling was sufficient to rescue the cell migration defect in cultured Map3k1 ^−/− keratinocytes. ¹⁸ Our results also show that activation by TGFβ or TGFα in vitro can rescue the cell migration defect in woe mEFs and keratinocytes. However, in vivo, the presence of both signaling pathways is required for embryonic eyelid closure and the absence of either signaling pathway results in EOB. 

In addition to rescuing the EOB phenotype, the hypermorphic Egfr^Dsk5 allele rescued the anterior segment defects and the absence of meibomian glands previously reported for the woe mice. ⁶ Mice with a mutation in Tgfa also exhibit anterior segment defects and the absence of the meibomian gland, ⁸ further implicating the role of EGFR signaling during anterior segment and meibomian gland development. Furthermore, overexpression of either Tgfα or Egfr in the lens results in the failure of corneal mesenchymal cells to differentiate into corneal endothelial cells, thus leading to severe anterior segment defects. ⁴¹ It has been shown that TGFα can chemoattract migrating mesenchymal cells ⁴² that lead to the formation of the corneal stroma, corneal endothelium, and associated drainage structures. ⁴³ The role of EGFR during meibomian gland development is unclear. Meibomian glands form around P0.5 from the differentiating epithelial cells near the inner surface of the lid margin. ⁷ Proper anterior segment development, as well as proper meibomian gland development observed in Egfr^Dsk5 rescued woe mice, may be due to rescued ADAM17-mediated EGFR function in these two tissues. Alternatively, the normal ocular and meibomian gland phenotypes in Egfr^Dsk5 rescued woe mice may simply be a result of the rescued eyelid closure defect. The role of the ADAM17-mediated EGFR signaling during anterior segment and meibomian gland development at this point still remains elusive and requires further clarification. 

Acknowledgments

The authors thank Benjamin Feiner at the Medical College of Wisconsin for his assistance with the genotyping of woe and Dsk5 alleles, and Doug Holmyard at the Mount Sinai Hospital, Toronto, Canada for his assistance with scanning electron microscopy. 

Footnotes

 Supported by the National Eye Institute/National Institutes of Health Grant EY18872 (DJS).

Footnotes

⁴  These authors contributed equally to the work presented here and should therefore be regarded as equivalent authors.

Footnotes

 Disclosure: E.L. Hassemer None; B. Endres, None; J.A. Toonen, None; A. Ronchetti, None; R. Dubielzig, None; D.J. Sidjanin, None