|Home | About | Journals | Submit | Contact Us | Français|
Tissue morphogenesis during development is regulated by growth factors and cytokines, and is characterized by constant remodeling of extracellular matrix (ECM) in response to signaling molecules, for example, growth factors, cytokines, and so forth. Proteoglycans that bind growth factors are potential regulators of tissue morphogenesis during embryonic development. In this study, we showed that transgenic mice overexpressing biglycan under the keratocan promoter exhibited exposure keratitis and premature eye opening from noninfectious eyelid ulceration due to perturbation of eyelid muscle formation and the failure of meibomian gland formation. In addition, in vitro analysis revealed that biglycan binds to TGF-α, thus interrupting EGFR signaling pathways essential for mesenchymal cell migration induced by eyelid epithelium. The defects of TGF-α signaling by excess biglycan were further augmented by the interruption of the autocrine or paracrine loop of the EGFR signaling pathway of HB-EGF expression elicited by TGF-α. These results are consistent with the notion that under physiological conditions, biglycan secreted by mesenchymal cells serves as a regulatory molecule for the formation of a TGF-α gradient serving as a morphogen of eyelid morphogenesis.
In mammals, eyelid morphogenesis is characterized by the closure and fusion of the eyelid during development, which involves two major cellular events: the differentiation of ocular surface ectoderm to corneal, conjunctival, eyelid epidermal and glandular (lacrimal and meibomian) epithelia, and the migration and differentiation of periocular mesenchymal cells for the formation of eyelid stroma (Kao and Liu, 2003). Mouse eyelid formation begins at embryonic day 12 (E12), with folds of surface ectoderm adjacent to the developing eye. As it grows, the eyelid fold extends over the developing cornea, and periocular mesenchymal cells invade the eyelid stroma. The closure of the eyelid happens between E15.5 and E16.5, when the tips of the superior and inferior eyelid epidermis elongate toward the center of the eye and eventually cover the corneal surface (Findlater et al., 1993; Harris and Juriloff, 1986). Following epithelial fusion, the eyelid stroma grows by continued invasion and proliferation of mesenchymal cells, and a fusion line is formed between the superior and inferior eyelid epithelium to close the eye; thus, completing the formation of the eyelid protective barrier over the cornea. The mouse eyelids remain closed until 12–14 days postnatal. This process resembles the epithelial sheet migration of dorsal closure during Drosophila development (Glise and Noselli, 1997; Zhang et al., 2003), closure of choroid fissure of optic cup of vertebrates (Kurita et al., 2004; Macdonald and Wilson, 1996), and closure of fusion of chorioamniotic folds over platypus embryos (Hughes and Hall, 1998) and those of birds (Gilbert, 2003).
The eyelid serves as a protective barrier for the formation of the conjunctival sac and the development of the cornea and lens (Findlater et al., 1993; Harris and Juriloff, 1986). In normal mouse eye development, the commencement of eyelid formation coincides with the rapid growth of the eyeball beginning at E13.5. The formation of lid muscles, that is, orbicularis oculi, levator palpebral, and tarsal muscles, is essential for eyelid functions, which provide the adequate strength required to withstand the pressure exerted by the rapid growth of the eye. In addition, the tarsal plate at the end of the tarsal muscles provides the scaffold for the formation of meibomian glands that derive from epithelium at the tip of eyelids (Jester et al., 1989). The meibomian gland is a lipid-secreting gland that plays a pivotal role for the maintenance of tear film function. For example, meibomian gland dysfunction resulting from ectodermal dysplasia in human manifested severe exposure keratitis (Bron and Tiffany, 1998).
As in many other tissues, eyelid morphogenesis is also governed by bidirectional mesenchyme–epithelium interactions via growth factors, cytokines, and extracellular matrix (ECM) components during embryonic development. This is best illustrated by the eye open at birth phenotype (EOB) due to impaired epithelial cell migration in mouse strains in which the genes mediating signaling pathways of EGF, EGF receptors, and TGF-β superfamily members are ablated, for example, TGF-α, EGFR, cJun, MEKK 1, activin βB, and so forth (Li et al., 2003; Luetteke et al., 1993; Sibilia and Wagner, 1995; Vassalli et al., 1994; Zenz et al., 2003; Zhang et al., 2003). Thus, it is likely that genetic perturbations that impair epithelium migration can potentially cause eyelid malformation. However, it remains unknown whether mutation of these genes may impair the functions of periocular mesenchymal cells essential for eyelid morphogenesis.
Interestingly, failure in eyelid closure caused by spontaneous autosomal recessive mutations has been reported in several mouse strains, such as eye-open-at-birth (eob), lidgap-Gates (lgGa), open-eyes (oe), and gaping lids (gp) (Fujii et al., 1995; Juriloff et al., 1996; Stein et al., 1967; Teramoto et al., 1988). However, in each strain, the mutant gene responsible for EOB has not been identified (Juriloff et al., 1996, 2000; Kelton and Rauch, 1968). The obvious implication is that other molecular and cellular mechanism(s) besides epithelium migration may also be involved in eyelid morphogenesis. One of the many possibilities is a role for extracellular matrix components in eyelid morphogenesis, for example, proteoglycans, collagens, fibronectin, or laminin.
Proteoglycans belonging to the small leucine-rich repeat proteoglycan (SLRP) family, for example, decorin, biglycan, and fibromodulin, have been implicated as important molecules that bind and modulate the activities of growth factors and cytokines via their glycosaminoglycan moieties and core proteins (Kao and Liu, 2002; Iozzo, 1999; Reed and Iozzo, 2002). Biglycan contains two side chains of chondroitin sulfate/dermatan sulfate (CS/DS) glycosaminoglycans and is widely expressed by fibroblasts of interstitial connective tissues, mesothelial cells, and epithelial cells (Bianco et al., 1990; Dobra et al., 2000; Funderburgh et al., 1998). Like other members of the SLRP family, biglycan serves as a component of ECM and binds to a variety of other proteins, including growth factors such as TGF-β, TNF-α, and WISP-1 (Desnoyers et al., 2001; Hausser et al., 1994; Hildebrand et al., 1994); extracellular matrix proteins such as collagens I and V, fibronectin (Bidanset et al., 1992; Schmidt et al., 1987; Schonherr et al., 1995; Wiberg et al., 2003), and α-amyloid (Snow et al., 1995), apo E and apo B, α-dystroglycan, and phospholipase A2 type II (Bowe et al., 2000; Olin et al., 2001; Sartipy et al., 1998), and serum proteins such as heparin cofactor II (Fukushima et al., 1993; Hildebrand et al., 1994; Whinna et al., 1993). These various binding activities may account for the ability of biglycan to exert diverse functions in many tissues. For example, it has been suggested that biglycan may serve as a neurotrophic factor for the survival of neocortical neurons and participates in the regulation of neuronal growth, differentiation, and repair (Junghans et al., 1995; Koops et al., 1996). In addition, injection of biglycan into the brain apparently promoted the facilitation of learning (Huston et al., 2000). A recent study of rat neurons and C6 glioma cells revealed that a population of biglycan targets to the nucleus and may be involved in the regulation of neuronal cell division (Liang et al., 1997). Even though there is no conclusive evidence available, it is very likely that biglycan may also bind other growth factors, for example, TGF-α, activin βB, and so forth. Thus, biglycan is ideally situated to play a critical role in modulating the functions of growth factors in situ, and it is, therefore, conceivable that biglycan may play an important role in morphogenesis and tissue homeostasis.
We have created transgenic Kera-Bgn mouse lines using a 3.2-kb keratocan promoter (Liu et al., 2000), in which migrating periocular mesenchymal cells overexpress biglycan during embryonic development. The Kera-Bgn transgenic mice exhibited noninfectious eyelid ulceration causing premature eye opening and severe exposure keratitis. In this study, we present evidence illustrating that excess biglycan perturbs the formation of eyelid muscles and the migration of periocular mesenchymal cells by sequestering TGF-α secreted by eyelid epithelium during eyelid morphogenesis of mouse embryonic development.
The coding region of mouse biglycan cDNA was generated by RT-PCR using total RNA isolated from adult mouse corneas. The first strand cDNA was synthesized by AMV-reverse-transcriptase with random primers and was used to prepare the full-length double-stranded cDNA with a pair of HindIII linker primers (5′-mBgn: 5′-cccaagcttccaccatgtgtcccctgtggc-3′ and 3′-mBgn: 5′-cccaagcttggcaaccactgcctctac-3′). The double-stranded cDNA was cloned into the HindIII site of the mouse Kera promoter cassette Kerapr3.2-saBGHpA (Liu et al., 2000). The fidelity of the biglycan cDNA was verified by sequence analysis. The Kera3.2-Bgn/BGHpA minigene (5.2 kb) was isolated by NotI and KpnI digestion and microinjected into mouse-fertilized eggs by the transgenic core facility of the Cincinnati Children Hospital Research Foundation. Transgenic founders were identified by PCR of tail DNA using a primer set [Kera9 (+): 5′-ctaacaccagccacaggact-3′ and BGHpA (−): 5′-actagaaggcacagtcgagg-3′] flanking the biglycan cDNA.
All animal procedures were approved by the University of Cincinnati Animal Care and Use Committee. Timed mating procedure was used for obtaining mouse embryos at various developmental stages. Wild-type female mice were mated with a Kera-Bgn male at 9:00 p.m. and the next morning at 9:00 a.m. the females were examined for a coital plug, and 3:00 p.m. in the afternoon was defined as E0.5 of gestation when the evidence of copulation was detected. Cesarean-derived embryos at various developmental stages (E12.5 to 18.5) were genotyped by polymerase chain reaction with limb or tail DNA.
Total RNA was isolated from the corneas of the embryonic day 18.5 (E18.5) transgenic and wild-type littermates using TRI-reagent® (Molecular Research Center, Cincinnati, OH) (Liu et al., 1994), and 10 μg of each total RNA was subjected to Northern hybridization with 32P-labeled cDNA probes of mouse biglycan and glyceralde-hyde-3-phosphate dehydrogenase (GAPDH) as described previously (Liu et al., 1998).
The embryos and adult eye tissues were fixed with 4.0% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4 at 4°C for 24 h, dehydrated, embedded in paraffin, and processed for histological examination. Histology of the eyes of the embryos was examined by light microscopy after hematoxylin and eosin and PAS staining.
For routine immunostaining, 5-μm paraffin sections were deparaffinized with xylene and rehydrated with graded ethanol. The sections were stained as described below and immune reactivity was visualized by Vecta stain ABC kit (Vector Laboratories), indirect alkaline phosphatase, or indirect immunofluorescence methods (Vector Laboratories and Molecular Probes). Primary antibodies used in this study were against: EGFR, pEGFR (Tyr1173), c-Jun, and pc-Jun (Santa Cruz, Santa Cruz, CA); TGF-α (Labvision); α-smooth muscle actin (Sigma); and biglycan (Kresse et al., 2001). The nuclei were counterstained with methyl green for immunohistochemistry using alkaline phosphatase conjugate, and propidium iodide (PI) or DAPI (Sigma) for immunofluorescence.
The mouse embryos were fixed in 2.5% glutaraldehyde and 1% tannic acid prepared with 0.1 M phosphate buffer, pH 7.4. After postfixation with 1% osmic acid, the specimens were dehydrated with routine ethanol series. Following treatment with isoamyl acetate, the specimens were dried by the critical point method and then coated by platinum in an ion sputtering device. The prepared samples were examined with a scanning electron microscope (H-2000, Hitachi) (Zhang et al., 2003).
The plasmids of mouse biglycan and TGF-α cDNA cloned in pBluescript SK+ vector (Stratagene, CA) were linearized by EcoRI and SalI digestions, and T-7 and T-3 RNA polymerases were used to synthesize digoxigenin-conjugated antisense and sense riboprobes, respectively (Roche). Five-micrometer-thick paraffin sections were deparaffinized and processed for in situ hybridization with riboprobes of mouse biglycan and TGF-α, as previously reported (Kao et al., 1996; Saika et al., 2000, 2001). The sections were dehydrated, mounted, and observed under light microscopy after methyl green counterstaining.
Filter membranes (2 mm in diameter) were cut from a Durapore® filter (pore size 0.45 μm; Millipore). To isolate epithelium, the membranes were pressed on top of the eyelid and head skin and immediately removed from the tissues. The membranes were then used in cell migration assays as described below (Hayashi et al., 2004).
The eyelid epithelium of wild-type E13.5 embryos was removed by impression cytology, and the remaining eyelid tissue was dissected under a stereomicroscope. The E13.5 embryo eyelid stromal tissues were washed in PBS three times and minced. Tissues were then cultured at 37°C in DMEM supplemented with 10% fetal calf serum in an incubator equilibrated with 95% air and 5% CO2 for 3 days. Cells from explant cultures were harvested with trypsin (1 mg/ml) and EDTA (5 mM) treatment, and subcultured on p100 cell culture dishes (Corning, NY) for three more days. Fresh medium was added to the culture overnight, prior to harvesting the cells for migration study. The culture condition used was unfavorable for the growth of epithelial cells, thus epithelial cells were absent in the culture as judged by a phase contrast microscope.
To measure cell migration, E13.5 eyelid and head epithelial tissues were prepared using impression cytology and were put on the bottom of the well containing 600 μl of serum-free DMEM (Gibco) of a 24-well plate. Cultured mesenchymal cells were harvested by treatment with trypsin-EDTA and washed two times with serum-free DMEM medium. Cells were suspended at 5 × 105 cells/ml in serum-free DMEM. One hundred microliters of cell suspension was seeded in each Transwells® (Costar) that was placed into a 24-well plate containing epithelium. Blocking antibody (2 μg/ml) and bovine tendon biglycan (1 or 10 μg/ml) were included in the culture medium of upper (Transwells®) or lower chambers (24-well plate), and incubated at 37°C for 4 h. Cotton swabs were used to remove cells from the upper surface of the Transwells®. The hematoxylin-stained migratory cells at the undersurface of Transwells® were counted with an inverted microscope. Each experimental condition was tested in quadruplicate in individual experiments, and the number of migratory cells found on the undersurface of Transwell® membranes of control experiments consisting of wild-type mesenchymal cells in the upper chamber and wild-type eyelid epithelial tissues in the lower chamber was defined as 100% migration index. Each experiment was repeated four times and statistical analyses were performed using unpaired t test.
Pregnant female mice were sacrificed by CO2 asphyxiation on E13.5. Fetuses were removed from the uterus and dissected in PBS, and the limbs were removed for genotype identification by PCR. Biglycan extracts were prepared with an alkaline extraction method modified by Bowe et al. (2000) and Casar et al. (2004). Briefly, 10 eyelid tissues were homogenized in 50 mM Tris–NaOH, pH 12.0, containing 250 U/ml of Benzonase nuclease (Novagen, WI) and protease inhibitor cocktail (Sigma, MO), and maintained under agitation over night at 4°C. Then homogenates were centrifuged 30 min at 15,000 × g in 4°C, and supernatants were adjusted to pH 8.0 with 1 M Tris–HCl pH 6.0 and added Na acetate pH 8.0 (final concentration 100 mM), and centrifuged 10 min at 15,000 × g at 4°C. The supernatants were incubated with or without 0.5 U/ml chondroitinase ABC at 37°C for 1 h. Protein content in specimens was measured by BCA protein assay kit (Pierce, IL). For SDS-PAGE, 5 μg of protein of each sample was subjected to Western blot analysis as described below.
To prepare specimens for immune precipitation and Western blot (IP-Western), eyelid tissues were lysed in buffer solutions containing 50 mM Tris–HCl (pH 7.4), 250 mM NaCl, 25 mM NaF, 5 mM EDTA, 1 mM PMSF, 0.1% Nonidet P-40 in the presence of a protease inhibitor cocktail (Sigma) at 4°C for 24 h. Tissue lysates were clarified by centrifugation at 14,000 rpm, 4°C for 10 min. To remove nonspecific binding of protein to protein G-Sepharose, lysates were incubated with protein G-Sepharose (Amersham Biosciences) for 1 h. Slurries were spun down, and an aliquot of the supernatant was incubated with rabbit anti-TGFα, anti-TGFβ, anti-EGF, anti-activinβB, anti-HB-EGF, or anti-EGFR antibodies at 1 μg/ml IgG (Santa Cruz) at 4°C overnight. The antigen–antibodies complexes were precipitated with protein G-Sepharose conjugate. Immunoprecipitated proteins were electrophoresed on SDS-PAGE and blotted onto polyvinylidene difluoride membrane (Immobilon-P; Millipore Corp.). Immunoblots were treated with chondroitinase ABC, blocked with 5% dry milk in TBST (0.1% Triton X-100, 0.15 M NaCl, 50 mM Tris–HCl, pH 7.4), and probed with anti-biglycan antibody. After incubation with goat anti-rabbit IgG HRP conjugate, immunoblots were developed by the enhanced chemiluminescence system (Amersham Biosciences). Supernatant of eyelid homogenate in 50 mM Tris–HCl (pH 7.4), 150 mM NaCl, 10 mM EDTA, 0.5% sodium deoxycholate, 1 mM PMSF, 1% Nonidet P-40, 0.1% SDS, protease inhibitor cocktail was subjected to Western blot analysis to identify growth factors, receptors, and c-Jun as previously described (Saika et al., 2000).
Developing embryos were evaluated at days 15.0, 15.5, 17.0, and 17.5 of gestation. Samples were fixed in 4% paraformaldehyde in PBS, and developing eyes and surrounding lids removed and stained with the following combinations of probes: (1) FITC-conjugated mouse anti-α-smooth muscle actin (α-SMA) and rhodamine phalloidin, or (2) FITC-conjugated phalloidin and propidium iodide. Samples were then mounted on mylar bottom plates (25 μm thick) (Invitrogen) or further dissected to include just the developing lid tissues and mounted on glass slides with PBS/glycerol (1:1). Samples were viewed using a Leica SP2 confocal microscope with argon/krypton and helium/neon lasers providing excitation frequencies of 488, 514, 543, and 633 nm. Three-dimensional data sets were collected from each sample and reconstructed using the Leica three-dimensional software package including maximum intensity projections for identifying localization of actin and α-SMA.
A 3.2-kb murine keratocan promoter was used to create transgenic mice that overexpress biglycan by migrating periocular mesenchymal cells during eyelid development (Liu et al., 2000). Two independent transgenic Kera-Bgn mouse lines (C65-8 and C65-14) were obtained via micro-injection of fertilized mouse eggs with a minigene containing a 3.2-kb keratocan promoter and a mouse biglycan cDNA (Fig. 1A). Due to the exposure keratitis observed in postnatal transgenic mice (Fig. 1C), embryonic corneas were used to determine the expression level of biglycan in corneas by Northern hybridization. As shown in Fig. 1B, at E18.5 the transgenic mice have a higher level of biglycan mRNA in the cornea than that of wild-type littermates. Both transgenic lines had the same phenotype of premature eye opening due to noninfectious ulceration at postnatal day 1 (P1) or P2 and suffered severe exposure keratitis as they grew older (Fig. 1C). Histology examinations revealed that the transgenic mice at P21 exhibited the lack of meibomian glands, malformation of the tarsal plate, epithelial hyperplasia with the formation of a cornified envelope and the presence of goblet cells on the peripheral corneal surface, and vascularization and invasion of inflammatory cells in the corneal stroma, suggesting conjunctivalization of the corneal surface.
To examine whether the phenotype observed is a primary defect caused by excess biglycan, we checked the eyelid development of transgenic mouse strains by histology and scanning electron microscopy. In Kera-Bgn mice, eyelid malformation started at E13.5 (Fig. 2A). Kera-Bgn transgenic mouse eyelid closure was delayed by about half a day in comparison to wild-type littermates. After eyelid closure, the eyelid stroma was thinner and there were fewer stromal cells than in wild-type littermates. By scanning electron microscopy study, the elongation at the leading edge of the Kera-Bgn mice epithelium was irregular compared to the wild-type littermates. Ulceration of the eyelid fusion area was already apparent in E17.5 Kera-Bgn mice (Fig. 2B). During embryonic development, Kera-Bgn mice showed normal expression patterns of Krt12 (keratin 12 gene) by corneal epithelium and Krt1/Krt10 by eyelid epidermal epithelium similar to that of wild-type mice (data not shown). This observation is consistent with the notion that changes of corneal surface observed in postnatal transgenic mice are secondary to the pathogenesis of exposure keratitis.
As shown in Fig. 1C, the Kera-Bgn mice at P21 exhibited lack of the tarsal plate. To examine whether excess biglycan impairs the formation of eyelid muscles, for example, orbicularis oculi, levator palpebrae, and tarsal muscles, confocal microscopy using Phalloidin Rhodamine (F-actin) and anti-α-smooth muscle actin (α-SMA) antibodies was performed during the development of wild type and Kera-Bgn mice.
Early development of orbicularis oculi muscles (green) was detected at the lateral and medial margins of the lids by anti-α-SMA staining. By E15.5 (Fig. 3A, top-left), rapid development of the orbicularis oculi muscles was noted with muscle fibers extending from the lateral to medial insertions of the muscle to completely circumscribe the eyelid margins. Muscle development appeared to parallel lid closure and the formation of presumptive tarsal plate (double headed arrows) that excluded any developing muscles fibers. Closure of the lid margins at E17.0 was characterized by prominent development of the orbicularis oculi muscle with muscle fibers running almost vertically alongside of the presumptive tarsal plate (Fig. 3A, bottom-left, arrowheads). At the time of eyelid fusion, prominent muscle bundles from the orbicularis oculi ran directly along the presumptive tarsal plate, vertically from the lateral to the medial insertion. At E17.5, development of the levator palpebrae and tarsal muscles underlying the palpebral conjunctiva was detected and characterized by anti-α-SMA staining of developing muscles running vertically across the orbicularis oculi toward the developing tarsal plate (Fig. 3B, top-left, arrows and bottom left, arrowheads).
Initial development of the orbicularis oculi muscle at the medial and lateral margins of the eyelid was first detected at E15.5 with staining by anti-α-SMA (Fig. 3A, top-right). Staining of the developing muscle was noted to be at least 12 h behind that observed for wild-type littermates. α-SMA-stained muscle bundles failed to develop circumferentially around the developing eyelid margin (Fig. 3A, bottom-right). Additionally, the presumptive tarsal plate never appeared to become compacted and appeared much wider than that observed in lids from wild-type mice. It should be noted that muscle bundles never developed running parallel to the presumptive tarsal plate and that the margins of the developing lid mesenchyme never fused but remained separated by epithelium covering the surface of the eye (Fig. 3A, bottom-right, asterisk). Evaluation from the palpebral conjunctival side of the lid also showed prominent interruption in the developing orbicularis oculi muscle bundle (Fig. 3B, top-right, arrow), and the development of the levator palpebrae and tarsal muscles was delayed and showed more irregular development of muscle fibers (Fig. 3B, top-right; bottom-right, arrowheads) compared to the wild-type lid (Fig. 3B, top left, arrows; bottom left, arrowheads).
Western blot analysis was used to determine whether there was more biglycan present in the eyelid of transgenic mice. Biglycan extracted from wild type and transgenic mice was treated with or without chondroitinase ABC prior to PAGE-SDS. As shown in Fig. 4A, more biglycan was recovered from tissue extracts prepared from transgenic mice than the wild type at E13.5 and E14.5. The biglycan core protein is detected in molecular size ranges from about 46 kDa to larger than 250 kDa. The high molecular weight bands are sensitive to chondroitinase ABC digestion, indicating the biglycan exists as a CS/DS-proteoglycan having various degrees of posttranslational modifications.
In situ hybridization reveals that in wild-type mice, mesenchymal cells expressing biglycan first appeared at the base of the nasal eyelid fold at E13.5 and invaded into the eyelid stroma at E14.0 and spread along a path from the base to the tip of eyelid stroma at E14.5 (Fig. 4B, left panels). In Kera-Bgn mice, the migrating mesenchymal cells expressing biglycan also appeared at the base of the lid (Fig. 4B, top right, arrow heads). Interestingly, in contrast to wild-type mice, in Kera-Bgn mice stronger hybridization signals were detected, suggesting more mesenchymal cells or the mesenchymal cells expressing a higher level of biglycan mRNA appeared concentrated at the eyelid tip at E14.0 (Fig. 4B, middle-right), and at the tip and base of the eyelid at E14.5 (Fig. 4B, bottom right). These observations of the distribution of biglycan-expressing cells in transgenic mice implicate that the initial wave of mesenchymal cell migration is not compromised, but the continued mesenchymal cell migration is perturbed by excess biglycan expression during development.
As shown in Fig. 2, the eyelids of Kera-Bgn mice had thinner stroma and fewer cells as compared to wild-type mice, we investigated the possibility that this morphological abnormality in the presence of excess biglycan could derive from decreased cell proliferation, increased apoptosis, or impaired mesenchymal cell migration. To examine these three hypotheses, we first determined cell proliferation activities by BrdU incorporation and apoptosis by TUNEL assay, and found there was no difference in BrdU incorporation and cell apoptosis between wild type and Kera-Bgn mice (data not shown). Thus, the eyelid thinning and decrease of eyelid stromal cells may be due to the perturbation of mesenchymal cell migration rather than altered cell proliferation and apoptosis in transgenic mice.
To examine the possibility that excess biglycan might perturb cell migration in developing eyelids, a two-chamber culture system was designed in which eyelid epithelium obtained by impression cytology was placed in the lower chamber, and eyelid mesenchymal cells were seeded on a Transwell® membrane in the upper chamber. The epithelium and mesenchymal cells were co-cultured at 37°C in a CO2 incubator for 4 h. The number of mesenchymal cells that migrated toward the epithelium and across the transmembrane pores was determined as described in Materials and methods. When eyelid epithelial tissues were used, many eyelid mesenchymal cells migrated to the opposite side of the Transwell® membrane as shown in Fig. 5A. In the first series of experiments, the eyelid epithelium was prepared from mouse embryos at different gestations E13.5, 14.5, and 15.5 by impression cytology. It was found that the eyelid epithelium from E13.5 embryos produced the highest chemoattractive activity on mesenchymal cells (data not shown). E13.5 embryos were then used in all the following experiments.
To examine whether biglycan might impair mesenchymal cell migration, biglycan (1 and 10 μg/ml) was included in the culture medium of either the lower or upper chamber. Fig. 5B demonstrates that biglycan added in the lower chamber containing eyelid epithelium effectively blocked the migration of mesenchymal cells from the upper chamber toward the epithelium placed in the lower chamber, whereas biglycan added in the upper chamber did not perturb mesenchymal cell migration. This observation is consistent with the notion that biglycan may bind to and abolish the activity of chemoattractive factor(s) secreted by the eyelid epithelium.
It has been demonstrated that the perturbation of EGFR and activin βB signaling pathways leads to the eye open phenotype in mouse strains. Thus, we tested the possibility that biglycan might bind to and sequester growth factors, for example, TGF-α, EGF, TGF-β, and activin βB, secreted by eyelid epithelium. Neutralizing antibodies to such growth factors were included in the culture medium. As shown in Fig. 5C, addition of anti-TGF-α antibodies, but not anti-EGF or anti-TGF-β antibodies, in the lower chamber strongly inhibits cell migration induced by E13.5 wild-type epithelium. Addition of anti-activin βB antibodies produced a moderate inhibition of cell migration. Neutralizing antibodies of TGF-α, TGF-β, EGF, and activin βB present in the upper chamber did not perturb mesenchymal cell migration. The presence of neutralizing antibodies against EGFR in the upper chamber inhibits eyelid stromal cell migration, suggesting that eyelid mesenchymal cell migration is mediated via the TGF-α/EGFR signaling pathway.
As mentioned above, excess biglycan may sequester TGF-α and abolish the TGF-α-induced periocular mesenchymal cell migration by disrupting the gradient of the growth factor in eyelid stroma morphogenesis during embryonic development of Kera-Bgn mice. To verify the existence of such a growth factor–biglycan complex in developing eyelids, proteins were extracted from eyelids of E13.5 mouse embryos and incubated with antibodies against TGF-α, TGF-β, activin βB, EGF, HB-EGF, and EGFR. The potential antigen–antibody complexes were precipitated by protein G-Sepharose conjugates and subjected to Western blot analysis with rabbit anti-biglycan antibodies. As shown in Fig. 6, only anti-TGF-α antibodies co-precipitated biglycan from eyelid extracts, but not anti-activin βB, TGF-β, EGFR, EGF, or HB-EGF antibodies. The results support the suggestion that biglycan does bind to TGF-α.
The results described above indicate that excess biglycan can sequester TGF-α and block the growth factor from serving as a chemoattractant for mesenchymal cell migration during eyelid morphogenesis. Thus, it is predictable that overexpression of biglycan by migrating periocular mesenchymal cells may alter the distribution of TGF-α in developing eyelid and consequently perturb the signaling pathways, for example, activation of EGFR and AP 1 transcription factors. To examine these possibilities, we examined expression levels and localizations of TGF-α, EGFR and phosphorylated EGFR (pEGFR), HB-EGF, and cJun and phosphorylated cJun (pcJun) in the developing eyelids of E13.5 Kera-Bgn and wild-type mice. Western blot analysis showed similar levels of active TGF-α and EGFR; an increase of total TGF-α; decreases of cJun and HB-EGF; and reduction of pEGFR in Kera-Bgn mice compared to wild-type mice (Fig. 7A). In situ hybridization revealed that TGF-α was highly expressed by epidermal epithelium of eyelid and migrating periocular mesenchymal cells of both Kera-Bgn and wild-type mice (Fig. 7B). However, immunohistochemistry with anti-TGF-α indicates that in transgenic mice the growth factor was unevenly distributed and more abundant in the upper one-third portion of the eyelid adjacent to epidermis, whereas in wild-type mice the growth factor was widely distributed and extended through more than two-thirds of the developing eyelid at E13.5. This pattern of TGF-α distribution is consistent with the overexpression of biglycan by mesenchymal cells of Kera-Bgn mice, as shown in Fig. 4. Furthermore, the expression of HB-EGF was suppressed (Fig. 7B) and the activation of EGFR (phosphorylation) was down-regulated in the eyelids of Kera-Bgn mice (Fig. 7C). As shown in Fig. 7D, excess biglycan caused a slight decrease of c-Jun protein, but strongly inhibited the activation of c-Jun as shown by a decrease of pcJun in the eyelids of Kera-Bgn mice.
In the present studies, we have characterized the Kera-Bgn transgenic mice that exhibit premature eye opening and severe exposure keratitis characterized by epithelial hyperplasia and the formation of a cornified envelope at the corneal surface. These defects are secondary to the mal-development of eyelids with thin stroma, lack of tarsal plates and meibomian glands, and perturbed maturation of lid orbicularis oculi, and levator palpebrae and tarsal muscles. In contrast in wild-type mice, compaction of the presumptive tarsal plate directly coincided with the development of muscle fibers running parallel to the presumptive tarsal plate, presumably associated with the developing pars palpebralis portion of the orbicularis oculi muscle that inserts into the medial ligament. These findings taken together suggest that development of lid and associated lid muscles plays a critical role in lid closure and fusion. Furthermore, excessive biglycan expression appears to interfere with critical EGFR signal transduction events involving TGF-α.
Tension within the muscle exerted during development might provide mechanical forces that keep the lids closed during development. Additionally, it should be noted that the eye extends out beyond the surface of the lids early in development. As the muscles develop, contraction of the muscle would tend to pull the lids across the eye and help push the eye farther back into the orbit.
Data concerning the Kera-Bgn mice suggest that eyelid open at birth may be related to a major defect in the development of eyelid muscles. Failure of the normal development of this muscle might lead to a failure in the fusion of the lids due to the mal-approximation of the mesenchymal portion of the developing eyelids. As the eye grows, the failure of the muscles to align parallel to the tarsal plate might lead to structural failure as the growing eyeball pushes forward. At some point during development, the eyeball would then push open the lids resulting in the eyelid open at birth phenotype. The malformation of tarsal muscle in Kera-Bgn mice fails to provide a scaffold for the formation of meibomian glands. The lack of meibomian glands contributes to the severe exposure keratitis observed in Kera-Bgn mice resembling the pathogenesis of human diseases of meibomian glands dysfunction from ectodermal dysplasia (Bron and Tiffany, 1998).
However, the molecular and cellular mechanism(s) whereby excess biglycan impairs lid muscle formation remains unknown. During avian eye development, it has been suggested that eyelid muscles may be also derived from periocular mesenchymal cells that give rise to stromal cells of many ocular tissues, for example, cornea, iris, ciliary body, and eyelids (Noden, 1986). It remains unknown whether malformation of orbicularis oculi, levator palpebrae, and tarsal muscles of Kera-Bgn mice is due to impaired mesenchymal cell migration or perturbed differentiation of mesenchymal cells, or both. It has been demonstrated that biglycan binds to α-dystroglycan and that in muscle biglycan is detected at both synaptic and nonsynaptic regions. Furthermore, biglycan expression is upregulated in muscle from dystrophic mdx mice (Bowe et al., 2000). Thus, our observation of malformation of lid muscles in transgenic mice is consistent with the notion that upregulated biglycan expression may account for the pathogenesis of some muscular dystrophies. Further studies are needed to elucidate the molecular and cellular mechanism(s) of eyelid muscle development, and whether the mal-development of eyelid muscles is directly caused or is mediated by the perturbation of EGFR signaling pathways due to excess biglycan.
Epithelium sheet migration induced by growth factors or cytokines plays a pivotal role in tissue morphogenesis during embryonic development. For example, TGF-β signaling is essential for dorsal closure in Drosophila (Glise and Noselli, 1997). Genetic alterations in many knockout and transgenic mouse strains that exhibit an eye open at birth phenotype are characterized by impaired epithelium sheet migration (Carroll et al., 1998; Li et al., 2003; Luetteke et al., 1993; Sibilia and Wagner, 1995; Vassalli et al., 1994; Zenz et al., 2003; Zhang et al., 2003). Many signal transduction pathways, including activin βB, TGF-α/EGFR, MEKK1, and c-Jun, are involved in eyelid closure. In addition, mutations of molecules mediating cell migration, for example, integrins, fibronectin, and so forth, are also exemplified by the eye open at birth phenotype. It has been demonstrated that mutation of these genes also causes defective cell migration of cultured epithelial cells in vitro.
In the present studies, our data demonstrate for the first time that TGF-α signaling also plays an important role in regulating the migration of periocular mesenchymal cells into the eyelid stroma during development. Mesenchymal cell migration also appears to be independent of TGF-β signal transduction, as shown in our previous studies of TGF-β knockout mice that do not exhibit mal-development of the eyelid (Saika et al., 2001). However, inhibition of cell migration by activin βB-neutralizing antibodies is consistent with the eye open at birth phenotype observed in activin βB knockout mice (Vassalli et al., 1994), suggesting that activin βB may also serve as a chemoattractant to periocular mesenchymal cells during eyelid morphogenesis.
Interestingly, the present studies provide evidence that EGFR signaling pathways play a pivotal role in periocular mesenchymal cell migration. The disruption of the TGF-α gradient necessary for mesenchymal cell migration by excess biglycan is accompanied by diminished phosphorylation of EGFR and cJun and the suppression of HB-EGF (Fig. 7). Thus, excess biglycan perturbs the autocrine or paracrine loops of the EGFR signaling pathways required for normal lid morphogenesis (Fig. 7E). EGF receptor signaling is essential for eyelid morphogenesis as evidenced by spontaneous mutations or knockout mouse strains in which this signaling pathway is interrupted. For example, ablation and spontaneous mutation of TGF-α and ablation of EGFR elicit the eye open at birth phenotype (Luetteke et al., 1993; Mann et al., 1993; Reneker et al., 1995). The TGF-α signaling is probably mediated by MAP kinase pathways. Recently, it has been shown that conditional ablation of cJun in the epidermis leads to the eye open at birth phenotype. In vitro analysis using cultured epithelial cells revealed that the lack of these gene products is associated with impaired epithelial cell migration and interruption of the autocrine and paracrine loops of EGFR signaling pathways (Li et al., 2003; Zenz et al., 2003). It is of interest to note that ablation of MEKK 1 also elicits an eye open at birth phenotype. In vitro examination of cultured epithelial cells isolated from skin of mice lacking MEKK 1 kinase activities indicates that MEKK 1 mediates the activin βB signaling pathways (Zhang et al., 2003).
In summary in the present studies, we have demonstrated for the first time that excess biglycan perturbs eyelid muscle formation that can in part explain the eye open at birth phenotype of Kera-Bgn mice. Also, excess biglycan in the eyelid stroma of transgenic mice sequesters TGF-α and disrupts the formation of the TGF-α gradient that is necessary for guiding mesenchymal cell migration and consequently interrupts the EGFR signaling loops necessary for eyelid morphogenesis (Fig. 7E). Thus, it is intriguing to suggest that the ability of biglycan to bind TGF-α allows this CS/DS-proteoglycan to serve as a regulatory molecule for the formation of a TGF-α gradient in eyelid morphogenesis during normal embryonic development.
The studies were in part supported by NIH grants: EY 11845, EY 13755, EY 12486, EY 09368, EY13215, and EY07348; Research To Prevent Blindness (RPB); the Ohio Lions Eye Research Foundation; the Shriners of North America; the Canadian Institutes for Health Research; and the Japan Eye Bank Association. WWYK and JJJ are the recipients of the RPB Senior Scientific Investigator Award. The authors thank Dr. Ying Xia, University of Cincinnati, for her suggestions in designing the cell migration assay.