HB-EGF promotes epithelial cell migration in eyelid development

The epidermal growth factor (EGF)-ERBB signaling network includes multiple ligands of the EGF/neuregulin superfamily and four related tyrosine kinase receptors, which include EGF receptor (EGFR)/ERBB1 and the ERBB receptors(ERBB2-4) (Holbro and Hynes,2004). The EGF family ligands that bind EGFR include EGF,transforming growth factor α (TGFα), amphiregulin (AR),heparin-binding EGF-like growth factor (HB-EGF), epiregulin (EPR),betacellulin (BTC) and epigen (Harris et al., 2003).

The importance of ERBB signaling during early mouse development has been demonstrated by gene targeting studies(Holbro and Hynes, 2004). Disruption of the EGFR locus results in embryonic or perinatal lethality,depending on the genetic background(Miettinen et al., 1995; Sibilia and Wagner, 1995; Threadgill et al., 1995). The phenotypes observed in these mice suggest several physiological roles for EGFR, including epithelial development. Newborn homozygous null EGFR mice exhibit immature development of epithelial cells in the skin, lung,gastrointestinal tract, tooth and eyelid(Miettinen et al., 1995; Sibilia and Wagner, 1995; Threadgill et al., 1995).

Eyelid closure represents a typical model for epidermal development. Although TGFβ/activin signaling is known to be essential for this process, EGFR signaling also is required(Xia and Karin, 2004). EGFR-deficient mice have an open eye at birth (EOB) phenotype(Miettinen et al., 1995; Sibilia and Wagner, 1995; Threadgill et al., 1995), and mice lacking the EGFR ligand TGFα occasionally exhibit an EOB phenotype(Luetteke et al., 1993).

HB-EGF is a member of the EGF family of growth factors that binds to and activates EGFR and ERBB4 (Elenius et al.,1997; Higashiyama et al.,1991). HB-EGF is synthesized as a type I transmembrane protein(proHB-EGF) and, like other EGF family members(Massague and Pandiella,1993), is cleaved at the juxtamembrane domain, resulting in the shedding of soluble HB-EGF (sHB-EGF)(Goishi et al., 1995). sHB-EGF is a potent mitogen and chemoattractant for a number of cell types(Raab and Klagsbrun, 1997),while proHB-EGF acts as a juxtacrine growth factor that signals to neighboring cells in a non-diffusible manner (Iwamoto and Mekada, 2000). HB-EGF has been implicated in a number of physiological and pathological processes(Raab and Klagsbrun, 1997). Importantly, analysis of HB-EGF-null mice has shown that HB-EGF is a crucial factor for proper heart development and function(Jackson et al., 2003; Iwamoto et al., 2003), and for skin wound healing (Shirakata et al.,2005).

Here, we show a novel role for HB-EGF in the process of eyelid closure. Our data indicate that HB-EGF functions synergistically with TGFα in leading edge formation during eyelid closure, by promoting epithelial sheet migration through activation of the EGFR-ERK signaling cascade.

Generation of HB-EGF-null mice (HB^del) and the knock-in mice expressing the uncleavable mutant form of proHB-EGF (HB^uc) was previously described (Iwamoto et al.,2003; Yamazaki et al.,2003). These mice were maintained on a mixed background of C57BL/6J, ICR and CBA. TGFα-null mice and waved 2 mice were purchased from Jackson Laboratory. TGFα-null mice were maintained on a background of C57BL/6J, while waved 2 mice were maintained on a mixed background of C57BL/6JEi and C3H/HeSnJ.

Hematoxylin/eosin staining and lacZ detection were performed as previously described (Iwamoto et al.,2003). The rates of eyelid closure and the eyelid formation were measured by microscopy as indicated in Fig. 2C and in Fig. 4A,respectively.

E15.5 pregnant mice were injected with BrdU (Nakalai) (100 μg/g of body weight). Two hours after injection, embryos were harvested, fixed with 4% PFA,dehydrated and embedded in paraffin. Sections (4 μm) were stained with anti-BrdU monoclonal antibody (ABCAM). The BrdU-positive cells in serial sections were measured by microscopy.

Embryos were fixed with 4% PFA, washed with PBS and incubated for overnight with FITC-phalloidin (Molecular Probes), as previously described(Zhang et al., 2003).

Mouse anti-mouse HB-EGF ectodomain monoclonal antibody (clone 4D9) was prepared by immunizing HB^del/del mice with an abdominal injection of the recombinant ectodomain of mouse HB-EGF, prepared by the baculovirus expression system. Lymphoid node cells from the immunized mice were fused with P3U1 myeloma cells, as previously described(Iwamoto et al., 1991), and the hybridoma producing an antibody reacting to mouse HB-EGF was selected. Purified 4D9 mAb was biotinylated. Rabbit anti-phosphorylated EGFR antibody,anti-EGFR antibody, anti-phosphorylated ERK antibody and anti-phosphorylated JNK antibody were purchased from Cell Signaling Technology. For immunohistochemistry, embryos were fixed by 4% PFA, dehydrated and embedded in paraffin. Sections (5 μm) were incubated with each antibody in the blocking solution Block Ace (Dainihon Seiyaku), permeabilized with 0.1% Triton X-100,and then incubated with Alexa fluorophore-labeled streptavidin or Alexa fluorophore-labeled secondary antibodies (Molecular Probes). After incubation,sections were washed with PBS and observed by fluorescence microscopy.

Statistical significance was assessed with the Student's t-test. A value of P<0.05 was considered statistically significant.

A mutant allele of HB-EGF was previously generated by replacing exons 1-3 of the HB-EGF locus with loxP-flanked HB-EGF cDNA linked to the lacZgene (Iwamoto et al., 2003). Removal of the HB-EGF cDNA by Cre-mediated recombination generates a null allele (HB^del) and leaves the lacZ gene under the control of the HB-EGF promoter, allowing us to use this allele to examine the expression pattern of HB-EGF during eyelid formation.

Eyelid formation occurs between embryonic day 11.5 (E11.5) and E17, and can be broken down into five steps. First, from E11.5 to E14.5, in the primitive eyelid region, the ectoderm layer surrounding the mesoderm undergoes increased cell proliferation as well as morphological changes to form the immature eyelid. Here, we refer to this immature eyelid structure as the `eyelid root'(Fig. 1K). During root formation, no HB-EGF expression is detected(Fig. 1A,F,K). Next, the leading edge begins to extend from the tip of root epithelium at E15.0. HB-EGF expression is detected exclusively in the tip of the leading edge at this stage (Fig. 1B,G,L). Between E15.0 and E16.0, the leading edge extends, followed by root extension over the cornea. During this step, HB-EGF continues to be expressed at the tip of the leading edge (Fig. 1C,H,M). At E16.0, the two leading edges from both sides of the eyelid root meet and fuse at the center of eye. At this stage, HB-EGF is still expressed at the contact point of the leading edges (Fig. 1D,I,N). Finally, after E16.5, the root epithelium migrates towards the contact point of the leading edges. At this stage, HB-EGF expression is decreased but remains at the contact point of the leading edges(Fig. 1E,J,O). After eyelid closure is complete, HB-EGF expression is greatly diminished (data not shown). The observed pattern of HB-EGF expression during eyelid closure suggests that HB-EGF might play a fundamental role in eyelid development.

To investigate the role of HB-EGF in eyelid closure, we examined eyelid closure in HB^del/del embryos and their heterozygous littermates. At E15.5, eyelid closure was obviously delayed in HB^del/del embryos compared with heterozygous littermates(Fig. 2A,B). To analyze this phenotype more quantitatively, the progression of eyelid closure was compared between HB^del/del and HB^del/+ embryos during this stage. As lacZ expression was detected at the margin of extending leading edge of the eyelid in both genotypes, the progression of eyelid closure was determined by measuring the long and short axis of the lacZ-stained ellipse (Fig. 2C). Along both axes, the average progression of eyelid closure in HB^del/delembryos was approximately three times lower than in HB^del/+ embryos(Fig. 2D). However, by birth,eyelid closure was completed and showed no abnormalities in mice of either genotype (Fig. 6C, Fig. 7G). These results indicate that HB-EGF contributes to the progress of eyelid closure.

HB-EGF is first synthesized as a membrane-anchored form (proHB-EGF), and the soluble form (sHB-EGF) is subsequently released from the cell surface by ectodomain shedding (Goishi et al.,1995). ProHB-EGF acts as a juxtacrine growth factor that signals to neighboring cells in a non-diffusible manner(Iwamoto and Mekada, 2000). To investigate which form of HB-EGF is involved in eyelid closure, we analyzed knock-in mice expressing an uncleavable form of proHB-EGF (HB^uc)(Yamazaki et al., 2003). As was the case with HB-EGF-null embryos, the average progression of eyelid closure in HB^uc/uc embryos was approximately one-third that in HB^uc/+ embryos (Fig. 2E), indicating that the soluble form of HB-EGF is required for the progress of eyelid closure.

The process of eyelid closure coordinates both cell proliferation and migration. We therefore examined the cell proliferation during eyelid closure in HB^del/del embryos and their heterozygous littermates at E15.5 by measuring BrdU incorporation. BrdU-positive cells were mainly detected in the eyelid dermis of the root region, but not in epidermal cells at the leading edge (Fig. 3A-D). There was no significant difference between the number of BrdU-positive cells in HB^del/+ and HB^del/del eyelids(Fig. 3E). These results indicate that HB-EGF is not involved in cell proliferation during eyelid closure, and may instead play a role in cell migration.

Two primary morphological changes occur during eyelid formation: extension of the root region (root formation) and extension of the leading edge (leading edge formation). We therefore compared the progression of total eyelid formation, root formation and leading edge formation in HB^del/+ and HB^del/del embryos (Fig. 4A).

The average progression of all three processes was significantly slower in HB^del/del embryos than in HB^del/+ embryos(Fig. 4B), consistent with the previously observed delay in eyelid closure in HB^del/del embryos(Fig. 2D). The progression of total eyelid formation, leading edge formation and root formation were∼2.0-, 4.2- and 1.6-fold slower, respectively, in HB^del/delembryos compared to HB^del/+ embryos. Thus, the reduction in the progression of leading edge formation contributed more strongly than the reduction in the progression of root formation to the overall delay in eyelid formation in HB^del/del embryos. These results support the possibility that HB-EGF functions predominantly in the formation of the leading edge during eyelid closure.

The role of HB-EGF in leading edge extension during eyelid closure strongly suggests that HB-EGF functions in epithelial cell sheet migration. Previous studies have shown that migrating epithelial cells exhibit actin reorganization (Zhang et al.,2003). During eyelid closure, F-actin in the form of actin cables are detected at the margin of the migrating epithelial sheet, and radial F-actin fibers align with the axis of extension of the leading edge(Xia and Karin, 2004). To examine whether epithelial cell migration during eyelid closure is normal in HB^del/del embryos, we used FITC-phalloidin staining as a marker for actin polymerization.

At E15.0, although no overt morphological differences were detected in the eyelids of HB^del/+ and HB^del/del embryos, actin cable formation at the margin of the eyelid epithelium was lower in HB^del/del embryos than in HB^del/+ embryos(Fig. 4C, parts a,c,e,g). Moreover, at E15.5, when the leading edge is extending, progressive eyelid closure was accompanied by both actin cable formation and radial F-actin fiber formation in HB^del/+ eyelids(Fig. 4C, parts b,d), while in HB^del/del embryos, eyelid closure was retarded and both actin cable and radial F-actin fiber formation were quite low(Fig. 4C, parts f,h). These results suggest that HB-EGF plays a role in epithelial cell sheet migration during leading edge formation by promoting F-actin polymerization.

To investigate the molecular mechanism underlying HB-EGF-mediated eyelid closure, we performed immunohistochemistry for HB-EGF protein in the developing eyelid using an anti-HB-EGF ectodomain monoclonal antibody. As shown in Fig. 5C,E, HB-EGF protein was detected broadly in the leading edge of HB^del/+eyelids, although HB-EGF mRNA was at the tip of the leading edge(Fig. 1, Fig. 5A). As expected, no HB-EGF protein was detected in HB^del/del eyelids(Fig. 5D,F), although the lacZ expression pattern resembled that of HB^del/+ embryos(Fig. 5B). These results indicate that sHB-EGF is secreted at the tip of the leading edge, and diffuses only in this region.

EGFR is a receptor for HB-EGF(Higashiyama et al., 1991) and is essential for eyelid closure (Miettinen et al., 1995; Sibilia and Wagner, 1995; Threadgill et al., 1995). To investigate whether HB-EGF activates EGFR during eyelid closure, we examined HB^del/+ and HB^del/deleyelids for phosphorylated EGFR. In HB^del/+ embryos, phosphorylated EGFR protein was detected predominantly in the extending leading edge(Fig. 5G), while total EGFR protein was detected more broadly in the epithelial cells of the leading edge as well as the root region (Fig. 5I). Phosphorylated EGFR was detected in the same region as HB-EGF(Fig. 5E,G). By contrast, the level of phosphorylated EGFR was drastically reduced in HB^del/delembryos, especially in the leading edge(Fig. 5H), while total EGFR protein appeared normal (Fig. 5I,J). These results suggest that HB-EGF activates EGFR in the leading edge and is restricted to this region.

Next, we used antibody staining to examine the phosphorylation state of ERK, a major downstream effector of EGFR signaling(Jorissen et al., 2003). In HB^del/+ embryos, phosphorylated ERK protein was detected predominantly in the extending leading edge(Fig. 5K), as was phosphorylated EGFR (Fig. 5G). By contrast, the level of phosphorylated ERK was decreased in HB^del/del eyelids, especially in the leading edge(Fig. 5L). Thus, activation of ERK in the leading edge is correlated with HB-EGF-mediated EGFR activation during eyelid closure.

JNK signaling has also been implicated in eyelid closure (Grose et al.,2003; Li et al., 2003; Weston et al., 2004; Zenz et al., 2003; Zhang et al., 2003). Moreover,the transcription factor Jun, a target of JNK signaling, has been shown to be necessary for HB-EGF expression in keratinocytes(Zenz et al., 2003). No difference in the level of phosphorylated JNK was detected between HB^del/+ and HB^del/del eyelids(Fig. 5M,N), suggesting that JNK signaling is not downstream of HB-EGF in leading edge formation.

These data strongly suggest that secreted sHB-EGF at the tip of the leading edge of the eyelid diffuses and locally activates EGFR and a downstream ERK signaling cascade during eyelid closure.

Since EGFR phosphorylation was decreased in leading edge of the eyelid in HB^del/del embryos, we used ZD1839, an EGFR-specific kinase inhibitor (Tanno et al., 2004; Von Pawel, 2004), to examine whether loss of EGFR function would phenocopy loss of HB-EGF. We administrated ZD1839 to E14.5 wild-type embryos in utero, and harvested and analyzed them 24 hours later. Embryos with reduced EGFR activity had a similar phenotype to HB^del/del and HB^uc/uc embryos (data not shown). In ZD1839-treated embryos, eyelid formation was significantly impaired owing primarily to defects in leading edge formation, but BrdU incorporation was normal, indicating that EGFR activity does not promote proliferation of dermal cells in the root region. In addition, formation of actin cables and radial F-actin fibers was drastically decreased in ZD1839-treated embryos. These results indicate that from E15.0 to E16.0, EGFR activity is correlated with HB-EGF activity during leading edge formation. EGFR signaling does not appear to regulate cell proliferation during leading edge formation, but probably instead functions to control epithelial cell sheet migration.

In order to determine whether EGFR acts as a receptor for HB-EGF in eyelid closure, we used a hypomorphic EGFR mutant, waved 2, and tested for a genetic interaction with HB-EGF. In waved 2 mice, the kinase activity of EGFR is decreased to less than 10% that of wild-type EGFR, owing to a point mutation in the kinase domain (Fowler et al.,1995; Luetteke et al.,1994). Although HB^del/del and waved 2(wa2/wa2) mice did not exhibit an EOB phenotype,HB^del/+; wa2/wa2 (66.6%) and HB^del/del; wa2/wa2 (100%) mice did(Fig. 6). These results suggest that eyelid closure relies on dose-dependent HB-EGF-EGFR signaling, and that EGFR mediates HB-EGF-controlled eyelid closure.

TGFα, another EGFR ligand, also has been reported to be a critical factor in eyelid closure (Luetteke et al.,1993; Wong et al., 2003). As expression of TGFα is regulated by HB-EGF in an autocrine manner in keratinocytes(Hashimoto et al., 1994; Piepkorn et al., 1998), we examined whether the defects in eyelid closure in HB^del/del embryos resulted from downregulation of TGFα expression. TGFα mRNA was expressed in the tip of the leading edge in both HB^del/+ and HB^del/del eyelids, and no significant difference in the level of TGFα expression was detected between HB^del/+ and HB^del/del eyelids (data not shown), indicating that the expression of TGFα is not affected by the presence or absence of HB-EGF.

To clarify the functional relationship between HB-EGF and TGFα in eyelid closure, we investigated this process in TGFα-null mice. As was the case with HB-EGF null mice, the BrdU incorporation in TGFα mutant eyelids appeared normal (Fig. 7A-C). We next compared the progression of eyelid closure in Tgfa^+/- and Tgfa^-/- embryos. The average progression of total eyelid formation, root formation and leading edge formation were significantly lower in Tgfa^-/- embryos than in Tgfa^+/- embryos (2.0-, 3.0- and 1.8-times,respectively, Fig. 7D), with the progression of total eyelid formation in Tgfa^-/-embryos similar to that of HB^del/del embryos(Fig. 4B, Fig. 7D). Thus, as with the HB-EGF mutants, the reduction in the progression of leading edge formation contributed more strongly to the reduced progression of total eyelid formation than did root formation in TGFα embryos.

Finally, to test for genetic interactions between HB-EGF and TGFα in eyelid closure, we examined double mutants of TGFα and HB-EGF. As shown in Fig. 7E-G, both HB^del/del; Tgfa^+/- and HB^del/+; Tgfa^-/- embryos exhibited similar frequencies of EOB (10%and 11%, respectively), while HB^del/del and Tgfa^-/- single homozygotes did not have an EOB phenotype. Moreover, homozygous double mutants (HB^del/del; Tgfa^-/-) had an even higher frequency of EOB (50%). These results strongly suggest that TGFα and HB-EGF contribute equally to leading edge formation and function synergistically in this process.

Here, we demonstrate a novel role for HB-EGF in eyelid development. Our major findings are as follows: (1) HB-EGF is expressed only at the tip of the leading edge of the migrating epithelial cell sheet during eyelid formation;(2) HB-EGF promotes epithelial cell migration, but not cell proliferation,during leading edge formation; (3) the secreted form of HB-EGF activates EGFR,resulting in the activation of ERK, which becomes localized to the leading edge; and (4) HB-EGF functions synergistically with TGFα in this process.

We demonstrate here using HB-EGF-null embryos that HB-EGF-mediated activation of EGFR promotes epithelial cell sheet migration, but not cell proliferation, during eyelid closure. Among our conclusions are the following:(1) In HB-EGF-deficient embryos, eyelid closure was significantly retarded;(2) the level of cell proliferation was not significantly different between HB^del/del and HB^del/+ eyelids; (3) retardation of eyelid formation in HB^del/del embryos was mainly due to defects in leading edge extension rather than root region extension; (4) the level of cell migration, judged by the detection of actin bundle formation in epithelial cells of the leading edge, was decreased in HB^del/del eyelids; (5)phosphorylated EGFR was dramatically decreased in the leading edge lacking HB-EGF; (6) inhibition of EGFR activity by the kinase inhibitor ZD1839 phenocopied the loss of HB-EGF; and (7) reduction of EGFR activity in mice lacking both HB-EGF and EGFR function accelerated the defect of eyelid closure.

Eyelid closure occurs between E15.5 and E16.5, with the eyelid epidermis eventually fusing to form a closed eyelid that covers the ocular surface and serves as a protective barrier that is crucial for normal eye development(Findlater et al., 1993; Harris and Juriloff, 1986). Initiation of eyelid formation requires the proliferation of epithelial cells,and impaired cell migration leads to a complete loss of eyelids(Li et al., 2001). Our results indicate that HB-EGF and EGFR signaling is not required for cell proliferation, but is required for cell migration during eyelid closure.

Recently, we demonstrated that in skin wound healing, HB-EGF promotes epithelial cell sheet migration, but not cell proliferation, in the process of re-epithelialization (Shirakata et al.,2005). In adult mice, HB-EGF is not usually expressed in the skin,but wounding induces expression of HB-EGF at the migrating leading edge of the wound. The parallels in both the domain of expression and function of HB-EGF between eyelid formation and skin wound healing suggest that HB-EGF functions to promote cell motility, but not proliferation in epithelial keratinocytes.

In this study, we present evidence that activation of EGFR by HB-EGF is important for eyelid formation. Although HB-EGF can also bind to and activate ERBB4 receptor, ERBB4 expression is not detectable in the murine epidermis(Kiguchi et al., 2000; Xian et al., 1997) or the human epidermis (Plowman et al.,1993), ruling out a role in eyelid closure. By contrast, both EGFR knockout mice (Miettinen et al.,1995; Sibilia and Wagner,1995; Threadgill et al.,1995) and the spontaneous EGFR mutant mice velvet(Du et al., 2004) showed an EOB phenotype. However, it remained unclear whether EGFR functions in cell proliferation or migration. In the present study, we show that EGFR is essential for epithelial cell sheet migration and promotes leading edge formation during eyelid closure process. These results suggest that the defects in eyelid closure resulting from loss of EGFR activity are mainly caused by retarded epithelial sheet migration.

ERK MAPK (ERK1/2) is a key component in the RAS-MAPK signaling cascade that is activated by a multitude of RTKs. Here, we show that ERK phosphorylation is decreased in HB-EGF mutant eyelids. Consistent with our results, cell migration has been reported to involve activation of ERK in primary cultured keratinocytes (Zhang et al.,2003). Moreover, another recent study has demonstrated that wound-induced activation of ERK1/2 propagates in cell sheets with the cell sheet movement, and that this movement is dependent on the activation of ERK2(Matsubayashi et al., 2004). Thus, together with our study, these findings suggest that ERK is the key signaling component for promoting epithelial cell migration.

JNK signaling, which can be activated by TGFβ/activin(Xia and Kao, 2004; Xia and Karin, 2004; Zhang et al., 2003), may also play a role in eyelid closure. Mice that lack MEKK1 showed defects in leading edge formation, owing to loss of JNK activation(Zhang et al., 2003). In addition, mice that lack the JNK target JUN showed defects in leading edge formation (Grose, 2003; Li et al., 2003; Zenz et al., 2003). Furthermore, mutant Jnk1^-/-Jnk2^+/-pups exhibited an EOB phenotype (Weston et al., 2004). Interestingly, it has been reported that expression of HB-EGF depends on JUN activation in keratinocytes(Zenz et al., 2003). However,in our experiments, there was no difference in JNK activation between HB^del/+ and HB^del/del animals. Thus, JNK may function as an upstream factor for HB-EGF-EGFR-ERK signaling pathway in eyelid closure.

TGFα is also known to be required for normal eyelid closure, with TGFα deficient mice occasionally exhibiting an EOB phenotype(Luetteke et al., 1993; Wong, 2003). However, the TGFα-null mice used in the present study did not show an EOB phenotype. This difference could be due to differences in the genetic background of our mice and the previously reported mice. Although it has been reported that TGFα functions upstream of HB-EGF signaling pathway involved in eyelid closure (Hayashi et al.,2005), our data suggest that TGFα and HB-EGF function at equal levels and synergistically in eyelid closure, based on the following findings: (1) expression of TGFα was not affected by HB-EGF expression;(2) TGFα-null mice phenocopy HB-EGF mutants, with defects in epithelial sheet migration but not proliferation; (3) mice that are homozygous null for HB-EGF and heterozygous for a TGFα null mutation showed a variably penetrant EOB phenotype; likewise, mice that are homozygous null for TGFα and heterozygous for a HB-EGF null mutation had a similar phenotype, while both single null mutant mice failed to show any EOB phenotype; and (4) HB-EGF and TGFα doubly homozygous null mice had dramatically increased penetrance of the EOB phenotype. However, in contrast to the complete penetrance of the EOB phenotype in EGFR-null mice(Miettinen et al., 1995; Sibilia and Wagner, 1995; Threadgill et al., 1995), only 50% of these double homozygotes were born with EOB. This suggests that other EGFR ligand(s) might also be involved in eyelid closure. Interestingly, it has been reported that triple null mice lacking EGF, amphiregulin, and TGFαexhibited a more severe EOB phenotype than did TGFα single null mice. However, even in these triple null mice, the EOB phenotype is not fully penetrant (Luetteke et al.,1999). This indicates that the EGFR ligands HB-EGF, TGFα,EGF and amphiregulin may cooperate in eyelid closure.

We show using the HB^del/+lacZ reporter allele that HB-EGF is specifically expressed at the tip of the leading edge of the migrating epithelial sheet, and only between E15.0 and E16.5. By contrast,HB-EGF protein was detected throughout the leading edge of HB^del/+developing eyelids. Moreover, embryos with an uncleavable mutant version of proHB-EGF (HB^uc/uc) displayed defects in eyelid closure. Together,these findings indicate that sHB-EGF, but not proHB-EGF, functions in eyelid closure, and that ectodomain shedding of proHB-EGF is essential for this process, as is the case in cardiomyocytes and in the valvulogenesis(Yamazaki et al., 2003).

Previous cell culture studies indicate that ectodomain shedding of HB-EGF is regulated by multiple signaling cascades(Izumi et al., 1998; Prentzel et al., 1999; Takenobu et al.,2003; Umata et al.,2001). Interestingly, activation of RTKs stimulates HB-EGF shedding via the RAS-MAPK pathway (Umata et al., 2001). Thus, activation of EGFR by HB-EGF may induce shedding of HB-EGF, forming an autocrine positive feedback loop. As shown in the present study, HB-EGF, activated EGFR and activated ERK were colocalized in the leading edge during eyelid closure, consistent with the existence of a feedback loop that stimulates HB-EGF shedding.

Among the proteases (convertases) that might induce ectodomain shedding of HB-EGF, ADAM17/TACE, a member of the ADAM family metalloproteases, is a prime candidate for functioning in this process in vivo(Lee et al., 2003). ADAM17-null mice had enlarged cardiac valves resembling those of HB-EGF-null mice (Jackson et al., 2003)and also exhibited an EOB phenotype (Sahin et al., 2004). ADAM17 has also been shown to be the sheddase for TGFα (Lee et al., 2003). These results suggest that ADAM17 may function in eyelid closure as a sheddase for HB-EGF and TGFα.

Although the precise mechanism is still unclear, we propose the following model for the mode of function of HB-EGF in mouse eyelid closure(Fig. 8). Before E15.0, the leading edge is not yet formed, and HB-EGF expression is not yet detected(Fig. 8A). At E15.0, the leading edge starts to extend, and the expression of HB-EGF appears in a few cells at the tip of the leading edge (Fig. 8B). Between E15.0 and E16.0(Fig. 8C), HB-EGF is constitutively expressed at the tip of the extending leading edge. On the surface of the cells located in this region, ectodomain shedding of proHB-EGF may be mediated by ADAM17 directly or indirectly, and liberated sHB-EGF diffuses throughout the leading edge region. sHB-EGF then binds to and activates EGFR, activating the ERK pathway. This results in actin polymerization and promotion of epithelial cell sheet migration. TGFαalso functions synergistically with HB-EGF, possibly in a similar manner to HB-EGF.

Detection of the endogenous proteins is important for the analysis of growth factor function in vivo; however, it is very difficult to obtain suitable antibodies to detect endogenous proteins in mouse tissues. In the present study, we were able to show the pattern of HB-EGF protein localization in the leading edge of eyelids using a monoclonal anti-mouse HB-EGF mAb (4D9),which was generated by immunizing HB-EGF null mice with recombinant mouse HB-EGF protein. Although many previous reports have examined HB-EGF localization using other anti-HB-EGF antibodies, this represents the first clear and specific detection of endogenous HB-EGF protein in mouse tissue. Thus, our new antibody will be a very useful tool for the analysis of HB-EGF function in mice in vivo.

In conclusion, in the present study we have demonstrated for the first time that HB-EGF signaling contributes to eyelid development. HB-EGF functions synergistically with TGFα in leading edge formation during eyelid closure, by promoting epithelial sheet migration through activation of the EGFR-ERK signaling cascade. We have demonstrated also that gene-expression,processing and protein localization of HB-EGF are strictly regulated temporally and spatially during eyelid closure process. Molecular mechanisms regulating these processes are remained unclear. Future study will be necessary to solve these issues.

We thank I. Ishimatsu and T. Yoneda for technical assistance. This work was supported by Grant-in-Aid from the Ministry of Education, Culture, Sports,Science, and Technology (14580696 for R.I. and 16207014 for E.M.).