Myocilin Mediates Myelination in the Peripheral Nervous System through ErbB2/3 Signaling*

Background: Myocilin, a secreted glaucoma-associated protein, is detected in the sciatic nerve, but its function there is not clear. Results: Myocilin null mutation leads to defects in the myelination of sciatic nerve acting through the ErbB2/ErbB3 receptor. Conclusion: Myocilin is a novel player in sciatic nerve myelination. Significance: This is the first demonstration of myocilin involvement in myelination. The glaucoma-associated gene, myocilin, is expressed in ocular and non-ocular tissues including the peripheral nervous system, but its functions in these tissues remain poorly understood. We demonstrate that in sciatic nerve, myocilin is expressed in Schwann cells with high concentrations at the nodes of Ranvier. There, myocilin interacts with gliomedin, neurofascin, and NrCAM, which are essential for node formation and function. Treatment of isolated dorsal root ganglion cultures with myocilin stimulates clustering of the nodal proteins neurofascin and sodium channel Nav1.2. Sciatic nerves of myocilin null mice express reduced levels of several myelin-associated and basal membrane proteins compared with those of wild-type littermates. They also demonstrate reduced myelin sheath thickness and partial disorganization of the nodes. Myocilin signaling through ErbB2/3 receptors may contribute to these observed effects. Myocilin binds to ErbB2/ErbB3, activates these receptors, and affects the downstream PI3K-AKT signaling pathway. These data implicate a role for myocilin in the development and/or maintenance of myelination and nodes of Ranvier in sciatic nerve.

The glaucoma-associated gene, myocilin, is expressed in ocular and non-ocular tissues including the peripheral nervous system, but its functions in these tissues remain poorly understood. We demonstrate that in sciatic nerve, myocilin is expressed in Schwann cells with high concentrations at the nodes of Ranvier. There, myocilin interacts with gliomedin, neurofascin, and NrCAM, which are essential for node formation and function. Treatment of isolated dorsal root ganglion cultures with myocilin stimulates clustering of the nodal proteins neurofascin and sodium channel Na v 1.2. Sciatic nerves of myocilin null mice express reduced levels of several myelin-associated and basal membrane proteins compared with those of wild-type littermates. They also demonstrate reduced myelin sheath thickness and partial disorganization of the nodes. Myocilin signaling through ErbB2/3 receptors may contribute to these observed effects. Myocilin binds to ErbB2/ErbB3, activates these receptors, and affects the downstream PI3K-AKT signaling pathway. These data implicate a role for myocilin in the development and/or maintenance of myelination and nodes of Ranvier in sciatic nerve.
Myocilin is a secreted glycoprotein that belongs to the family of olfactomedin domain-containing proteins (1). The MYOCILIN (MYOC) gene is heavily expressed in ocular drainage structures, such as the trabecular meshwork, which is responsible for a maintaining proper balance of intraocular pressure. It is well established that mutations in the MYOC gene can lead to glaucoma, a major cause of blindness. MYOC mutations are found in Ͼ10% of juvenile open-angle glaucoma and in 3-4% of patients with adult onset primary open-angle glaucoma (2)(3)(4)(5)(6). The most severe mutations in MYOC lead to intraocular pressure elevation. Available data suggest that wild-type myocilin is not required for the physiological regulation of intraocular pressure (7)(8)(9). In particular, Myoc knock-out mice do not develop glaucoma and appeared to be grossly normal (8). However, more careful analyses of Myoc null mice showed that these mice exhibit a moderate reduction in the amount of dystrophin-associated syntrophin and reduced levels of phospo-Akt in skeletal muscle compared with wild-type littermates (10). Patients with glaucoma who carry MYOC mutations are not known to develop abnormal non-ocular phenotypes even though the MYOC gene is expressed in several non-ocular tissues (4,6,11,12).
The sciatic nerve is one of the known sites of extraocular myocilin expression (13). Another olfactomedin domaincontaining protein, gliomedin, is expressed by myelinating Schwann cells and is essential for molecular assembly at the nodes of Ranvier in sciatic nerve and for clustering of nodal components including voltage-gated Na ϩ channels at heminodes during myelination (14 -17). Gliomedin interacts with several immunoglobulin family cell adhesion molecules including NrCAM and two isoforms of neurofascin, NF155 and NF186, in an olfactomedin domain-dependent fashion (14,15). Although clustering is critical for saltatory conduction of action potentials along myelinated axons, adult gliomedin null mice do not show any obvious neurological abnormalities and exhibit normal nerve conduction.
The aim of the present study was to characterize the role of myocilin in sciatic nerve. We demonstrate that myocilin is concentrated at the nodes of Ranvier, where it interacts with several nodal proteins. Myocilin induces clustering of the Na V 1.2 sodium channel and NF186 in dorsal root ganglion (DRG) 4 axons in the absence of glial cells. In sciatic nerve in vivo, myocilin co-localizes with and is able to physically bind the epidermal growth factor receptors ErbB2/ErbB3, resulting in phosphorylation (activation) of the receptors. Finally, the sciatic nerves of Myoc null mice show changes in the organization of axon bundles and partial disorganization of the nodes. These treated with glial growth factor (10 g/ml) and forskolin (5 M) to expand the Schwann cell population. On day 7 in culture, cells were trypsinized, washed, and plated onto polylysinecoated cover glasses in 2-well chamber slides. Three days after plating, the medium was removed, and cells were washed 2-3 times with PBS. Cells were fixed with fresh 3.7% formaldehyde for 10 min at room temperature and permeabilized with 0.1% Triton X-100 in PBS for 5 min. After blocking with 5% BSA in PBS at room temperature for 1 h, cells were incubated with anti-myocilin polyclonal (1:500 dilution) and anti-P0 monoclonal antibody (1:500 dilution) at room temperature for 1 h. Cells were washed with PBS-Tween 20 and then incubated with Alexa-488 or 594 conjugated secondary antibodies for 30 min. An Axioplan 2 fluorescent microscope and Zeiss700 confocal microscope (Carl Zeiss MicroImaging, Inc.) were used to detect fluorescence. The images were processed with Adobe Photoshop Elements 2.0 (Adobe Inc.).
Western Blot Analysis-Sciatic nerves were dissected and homogenized using a Dounce homogenizer in radioimmune precipitation assay buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, and 0.25% sodium deoxycholate) containing a mixture of protease inhibitors (Roche Applied Science) at 4°C. Homogenates were centrifuged for 20 min at 21,000 ϫ g at 4°C. To analyze laminin levels in sciatic nerve of P22 and adult mice, a membrane protein extraction kit (Thermo Scientific) was used. Briefly, sciatic nerves were first washed with a cell wash solution and then homogenized in a permeabilization buffer at 4°C. Homogenate was centrifuged at 16,000 ϫ g for 15 min at 4°C. The supernatant containing cytosolic proteins was removed, and the pellet was solubilized in 8 M urea-containing solubilization buffer by up and down pipetting with subsequent incubation for 30 min at 4°C. This extract was centrifuged for 15 min at 16,000 ϫ g at 4°C, and the supernatant was used for subsequent analysis. Equal amounts of protein from each supernatant (measured with the BCA Protein Assay; Bio-Rad) were size-fractionated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted. For quantification, the chemiluminescent signals were captured using an 8-megapixel scientific-grade CCD camera (Flurochem M. Santa Clara, CA), and the signal intensities were quantified and analyzed using AlphaView software (Proteinsimple, Santa Clara, CA). All Western blot experiments were performed at least three times using independent animals or pairs of littermates.
Binding Assays-AP-tagged fusion protein expression constructs were transfected into HEK293 cells to generate conditioned medium (CM)-containing AP-fusion proteins. The culture medium was changed to the fresh serum-free medium at 24 h after the transfection, and CM was harvested 24 -48 h later, filtered through a 0.22-m filter, and stored at Ϫ80°C until use. Absolute concentration and integrity of AP-tagged myocilin was determined by Western blotting using samples with known amounts of purified myocilin. NIH3T3 cells were transfected with ErbB2, ErbB3, or vector plasmids and incubated with AP-myocilin containing CM for 90 min at room temperature 48 h after transfection. Cells were washed 5 times and fixed by treatment with 60% acetone, 3% formaldehyde, and 20 mM HEPES, pH 7.5, for 30 s, and surface binding was visualized using the AP substrate 5-bromo-4-chloro-3-indolyl phosphate following the manufacturer's instructions (Genhunter Corp.). The images of stained cells were obtained with a dissection microscope (Zeiss, STEMI SV-11). For quantitative analysis of the activity of cell-bound AP, 1-Step TM PNPP (Pierce) was added to the fixed cells, and the absorbance at 405 nm in the supernatant was measured at 405 nm using a microplate reader (Bio-Rad, Model-680).
Coimmunoprecipitation-Sciatic nerve lysates were cleared by centrifugation at 16,000 ϫ g for 15 min and were immunoprecipitated with antibodies against myocilin, ErbBs, gliomedin, or NrCAM at 4°C overnight and then with protein-A agarose (Roche Applied Science) at room temperature for 1 h. Bound proteins were eluted from the beads by boiling in SDS-PAGE sample buffer and analyzed by Western blotting using the indicated antibodies. HEK293 cells were transiently transfected with myocilin-FLAG, gliomedin-His, or NF186-Fc using Lipofectamine 2000 (Invitrogen) and seeded in 6-well culture dishes. Cells were washed with PBS and lysed in lysis buffer 48 h after transfection. Cleared lysates were subjected to immunoprecipitation with anti-FLAG and anti-His antibodies (Sigma) and Protein-G magnetic beads (Dynal, Invitrogen). Immuno-precipitates were analyzed by Western blotting using indicated antibodies.
Tissue Preparation for Immunofluorescence-Sciatic nerves, spinal cords, and brains were embedded in optimal cutting temperature compound and frozen in acetone cooled with dry ice. Mice were anesthetized using a lethal dose of ketamine/xylazine (1:10) injected intraperitoneally. Anesthetized animals were perfused with a fixative containing 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, and the dissected tissues were fixed for 30 min in the same solution, cryoprotected by infiltration in 20% sucrose overnight, and embedded in optimal cutting temperature compound. Cryostat sections (5-10 m thick) were thaw-mounted on Super-Frost Plus glass slides (Fisher) and stored at Ϫ20°C. Tissue sections were fixed in 4% PBS-buffered paraformaldehyde and permeabilized with 0.3% Triton X-100 and 3% goat serum in PBS. Samples were incubated with primary antibodies diluted in PBS containing 3% goat serum at 4°C overnight. After washing with PBS for 3 times, samples were incubated with Alexa Fluor 488 goat antirabbit or Alexa Fluor 594 goat anti-mouse secondary antibody for 1 h at room temperature. Samples were mounted using Vectashield mounting medium (Vector), and images were taken using a Zeiss LSM700 confocal microscope with a 40ϫ objective lens. Images were scanned using ZEN software (Zeiss) and quantified using NIH image J software.
Electron Microscopy-Mice were anesthetized using a lethal dose of ketamine/xylazine (1:10) injected intraperitoneally. Anesthetized animals were perfused with a fixative containing 4% paraformaldehyde, 2.5% glutaraldehyde, 0.13 M NaH 4 PO 4 , and 0.11 M NaOH, pH 7.4. Perfused tissues were fixed in PBSbuffered 2.5% glutaraldehyde and 0.5% osmium tetroxide, dehydrated, and embedded into Spurr's epoxy resin. Ultrathin sections (90 nm) were made, double-stained with uranyl acetate and lead citrate, and viewed in a JEOL JEM 1010 (Tokyo, Japan)  transmission electron microscope equipped with digital imaging capabilities. The g ratio was determined by dividing the circumference of an axon (without myelin) by the circumference of the same axon including myelin. Six wild-type and seven Myoc null mice were analyzed. 12 and 20 nodes were found and analyzed in the sciatic nerves of 5 wild-type and 12 Myoc null 2-month-old mice.
Phosphoreceptor Tyrosine Kinase Antibody Array-HEK293 cells grown on 25T flasks were treated with control or myocilincontaining medium for 30 min. Cells were lysed in the lysis buffer (20 mM Tris-HCl, pH 8.0, 137 mM NaCl, 10% glycerol, 2 mM EDTA, 1 mM sodium orthovanadate) containing a protease inhibitor mixture (Roche Applied Science). Equal amounts (200 g) of cell lysates were added to human phosphoreceptor tyrosine kinase (RTK) antibody arrays containing 42 different anti-RTK antibodies (R&D systems, Minneapolis, MN). Information about anti-RTK antibodies spotted on the filters and the location of each spot can be found at the company's website. The antibody array was performed according to manufacturer's instructions. The signal was detected with Pierce ECL Western blotting Substrate by chemiluminescence. Intensities of identified spots were quantified using NIH ImageJ software.
Clustering-Clustering experiments were performed as described (23). In brief, DRG explants were cultured as above for 2 weeks on slides coated with 100 g/ml poly-D-lysine and 10 g/ml laminin (BD Biosciences). The neurons were then incubated with purified myocilin protein with or without antimyocilin polyclonal antibody for 30 min, washed once with neurobasal medium, and grown for an additional 48 h in culture medium before fixing and staining. Quantification of nodal protein clustering was performed in three separate experiments by counting the number of NF186 and Na v 1.2 staining sites in several 320-m square fields.
TUNEL Assay-The TUNEL assay was performed using the apoBrdU TUNEL assay kit (Invitrogen). Briefly, 1 ϫ 10 4 isolated Schwann cells were seeded per well on two well chamber slides. Cells were washed in equilibration buffer, incubated with a TdT enzyme in a humidified chamber at 37°C for 1 h, washed, and then incubated in the dark with fluorescein-488-conjugated anti-BrdU at room temperature for 30 min. The washed specimens were counterstained with DAPI, and images were collected using Zeiss LSM700 fluorescent confocal microscope.
Statistical Analysis-In the experiments where multiple samples were compared, statistical significance was determined using a one-way analysis of variance test.

RESULTS
Myocilin Interacts with Several Nodal Proteins within the Sciatic Nerve-Expression of the Myoc gene is reportedly detectable in rat sciatic nerve starting from P15 (13). We detected low levels of myocilin protein in mouse sciatic nerve as early as P3, the earliest analyzed stage, with expression levels increasing to reach approximately adult levels by P21, a time when myelination is almost completed (24) (Fig. 1). In transverse sections of adult mouse sciatic nerve, myocilin was detected mainly in Schwann cells as evidenced by co-localization of myocilin immunofluorescence with that of glial-associated proteins laminin ␣2, S100␤, and MBP (Fig. 2, A-C). Cultured Schwann  cells isolated from P5 wild-type but not Myoc null mice also expressed myocilin (Fig. 2, D and E). Immunofluorescent labeling of longitudinal sections of sciatic nerve confirmed that myocilin was located preferentially in Schwann cells, with individual cells showing different intensity of staining. Because myocilin is a secreted protein interacting with NrCAM and NF186 (see below) that are expressed in neurons, partial overlapping of myocilin immunofluorescence with neuronal marker neurofilament H immunofluorescence was also observed (Fig. 3). Myocilin was concentrated at the nodes of Ranvier where it co-localized with gliomedin, a known nodal marker, but not with caspr, a known paranodal marker ( Fig.  4A and C) (14). Specificity of myocilin immunostaining was confirmed using sciatic nerve from Myoc null mice that demonstrated the absence of myocilin staining, whereas gliomedin and caspr staining was not significantly disturbed (Fig. 4, B and D).
Co-localization of myocilin and gliomedin at sciatic nerve nodes of Ranvier prompted us to explore whether myocilin physically interacts with gliomedin and/or its receptors, NF186 and NrCAM. Myocilin can be proteolytically cleaved after Arg 226 , resulting in two fragments of about 22-and 35-kDa molecular mass (25). Thus, we analyzed protein-protein interactions with full-length myocilin and its proteolytic fragments. Protein-protein interactions were first assessed using CM of transiently transfected HEK293 cells, as all constructs that we used encoded well secreted modified proteins (17,19). Both myocilin and its N-terminal fragment (Myocilin-⌬C) but not the C-terminal fragment containing the olfactomedin domain (Myocilin-⌬N) interacted with gliomedin, NF186, and NrCAM ( Fig. 5A-E). The N-terminal fragment of myocilin interacted more strongly with gliomedin than full-length myocilin (Fig. 5,  A and B). Interaction of myocilin with neurofascin, NrCAM, and gliomedin in physiological conditions was confirmed using lysates of adult mouse sciatic nerve (Fig. 6). Nodal localization of myocilin and its interaction with several nodal proteins suggested that myocilin may play a role in the organization and/or function of the nodes of Ranvier in the sciatic nerve.
Sciatic Nerves of Myoc Null Mice Exhibit Myelin Abnormalities and Partially Disorganized Nodes of Ranvier-To explore whether the absence of myocilin affects the structural organization of the sciatic nerve, including the nodes of Ranvier, we The two lower panels show the levels of tested proteins in CM before immunoprecipitation. 5% of CM used for immunoprecipitation was loaded. The myocilin-⌬C construct in many cases (see panels B-E) produced two well separated bands due to differential glycosylation. In some cases (see panel A), glycosylation was less pronounced). The black dot marks bands corresponding to Myocilin-⌬C; * marks bands corresponding to myocilin-⌬N.  first evaluated the expression levels of several proteins by Western blotting and by immunofluorescence. Among the proteins tested, the overall levels of MBP, S100␤, P0, laminin ␣2, laminin ␤1, laminin ␥1, integrin ␤1, ␣-dystroglycan, ␤-dystroglycan, Krox-20, and NF186 were reduced in sciatic nerves of P22-P25 Myoc null mice compared with wild-type littermates (Fig. 7, A-D). Among laminin chains, laminin ␣2 was reduced more dramatically than laminin ␤1 and laminin ␥1. At the mRNA level, only mRNA encoding laminin ␣2 was reduced in the sciatic nerve of Myoc null mice compared with wild-type samples, whereas the levels of laminin ␤1 and laminin ␥1 mRNAs did not show statistically significant changes (Fig. 7F). Reduced levels of MBP, a major myelin component, were observed at different stages of sciatic nerve myelination (P14-P21) in Myoc null mice compared with wild type (Fig. 7E). In adult Myoc null and wildtype mice, the differences in the levels of MBP and tested laminin chains became less pronounced than in young animals, with only laminin ␣2 and ␥1 chains showing statistically significant differences (Fig. 8).
Immunostaining of sciatic nerve sections confirmed the results of Western blot analysis; changes in the intensity of MBP staining between Myoc null and wild-type mice were more pronounced at early postnatal ages compared with adult animals, and the difference in the intensity of laminin ␣2 staining was more pronounced than such differences for ␤1 and ␥ laminins (Fig. 9, A-D).
It has been previously reported that laminin-deficient Schwann cells failed to radially sort and myelinate axons (26 -28). Therefore, we compared myelination and sorting of axons in the sciatic nerves of wild-type and Myoc null mice. There were dysmyelinated areas in the sciatic nerves of 2-month-old Myoc null mice compared with wild-type animals (Fig. 10, A  and B). In the dysmyelinated areas, unmyelinated axons were grouped in bundles in unsorted manner (arrows in Fig. 10, A  and B). Immature Schwann cells were also observed with high frequency in Myoc null sciatic nerves (arrowheads in Fig. 10, A  and B). Axons had a lower average diameter in the sciatic nerves of Myoc null mice compared with wild-type animals (Fig. 10E). In the subset of Myoc null Schwann cell/axon units that appeared to be relatively normal, the myelin sheaths were thinner, with lower electron density than those of wild-type mice (Fig. 10, A-C) with g ratios (inner axon diameter/total fiber diameter of myelinated axons) of 0.75 Ϯ 0.01 (n ϭ 540 for Myoc null) versus 0.63 Ϯ 0.07 (n ϭ 350 for wild-type) (p Ͻ 0.01) (Fig.  10F). The reduction in myelin thickness was observed in axons of different sizes ranging from 1 to 7 m (Fig. 10F). Finally, the basal membrane of myelinated axons appeared fuzzy in most Myoc null samples and in some areas was disrupted as compared with the basal membrane in wild-type littermates (Fig.  10D). These differences may be connected with a lower density of membrane components in Myoc null mice.
Significant differences between wild-type and Myoc null mice were also observed in the nodal region. 20 and 12 nodes were analyzed by electron microscopy in Myoc null and wildtype nerve, respectively. Although all analyzed nodes in wildtype mice appeared morphologically normal, 15 of 20 analyzed nodes in Myoc null mice exhibited abnormalities of varying degrees (Fig. 11, A-C). Schwann cell microvilli normally have a   relatively regular, radial arrangement around the centrally located axon (Fig. 11A). In contrast, most of the nodes were partially disorganized in sciatic nerves of Myoc null mice (Fig.  11, B and C). The microvilli of Myoc null mice often ran parallel to, or averted away from, the axon, and their degeneration was observed in some areas (Fig. 11, B and C). It should be pointed out that Schwann cells isolated from Myoc null mice were more prone to apoptosis than those isolated from wild-type mice (Fig.  12). This may also contribute the degenerative processes observed in the nodes of Myoc null mice. In summary, our data suggest that myocilin might play an important role in sciatic nerve myelination and confirm its role in the organization and integrity of the nodes in the sciatic nerves.
Myocilin May Regulate Myelination through Epidermal Growth Factor Receptor Signaling-To identify the signaling pathway activated by myocilin, we employed human RTK antibody arrays. HEK293 cells were treated with control or myocilin-containing CM for 30 min and lysed. Equal amounts of lysates were added to the RTK antibody arrays, which were analyzed as suggested by the manufacturer. Among 42 different RTKs analyzed, only epidermal growth factor receptor (ErbB1) appeared activated by ϳ5-fold upon treatment with lysate from myocilin-treated cells compared with control cells (Fig. 13). It should be mentioned that ErbB2 and ErbB3 were also spotted at the RTK antibody arrays, but the level of their expression in HEK293 cells was very low compared with ErbB1 as judged by our gene expression array analysis (29).
It is now well established that signaling through ErbBs plays a critical role in myelination in the PNS (30,31). There are four ErbBs in mammals. Available data suggest that ErbB1, ErbB2, and ErbB3 may be expressed at similar levels in adult mouse sciatic nerves, whereas ErbB4 is not expressed in the sciatic nerve (32,33). In the sciatic nerve, ErbB2 and ErB3 are localized   preferentially in Schwann cells (32,33) with ErbB2 showing preferential localization to the nodal region of the myelinating Schwann cells, with individual cells showing different intensity of staining (34). These patterns of staining were very similar to the myocilin staining pattern that we observed (Figs. 2 and 4). Indeed, immunofluorescent labeling of sciatic nerve transverse sections with myocilin and ErbB2 or ErbB3 demonstrated colocalization of these proteins (Fig. 14, A and B).
To elucidate whether the defects in sciatic nerve myelination observed in Myoc null mice may be explained by modulation of ErbB signaling, we analyzed the effects of myocilin on ErbB signaling. It is known that heterodimers of ErbB2 and ErbB3 are critical for Schwann cell differentiation (35), so our analysis focused on these receptors. First, we sought to determine whether myocilin physically interacts with ErbB2/ErbB3 complexes. NIH3T3 cells were transfected with various receptor constructs and then treated with CM from cells expressing myocilin fused to AP or CM from cells expressing free AP (19). Myocilin bound stronger to cells expressing both ErbB2 and ErbB3 than to cells expressing either ErbB2 or ErbB3 alone (Fig.  15A). The affinities of myocilin for ErbB2, ErbB3, and their combination were estimated by measuring binding levels of increasing concentrations of the myocilin-AP protein to ErbBexpressing COS7 cells. The calculated K d were 27.8 Ϯ 0.4, 23.7 Ϯ 0.7, and 17.3 Ϯ 0.2 nM for ErbB2, ErbB3, and their combination, respectively (Fig. 15, B-D). In vivo binding of myocilin to ErbB2 and ErbB3 was confirmed by immunoprecipitation using sciatic nerve extracts (Fig. 15, E-H). Phosphorylation of ErbB2 at Tyr-1221 was diminished in sciatic nerves of Myoc null mice and increased in transgenic mice expressing elevated levels of myocilin compared with wild-type mice, although the total levels of ErbB2 were similar in the sciatic nerves of these lines (Fig. 15, I-J). Several major signaling pathways are stimulated upon activation of ErbB2/ErbB3. We tested whether reduced ErbB2 phosphorylation in Myoc null mice is associated with attenuated activation of one of such pathway, PI3K-AKT. Among proteins tested, the levels of PI3K␤ and p-AKT were reduced in Myoc null mice compared with wild-type litter-mates, whereas the levels of PI3K␣ and PI3K regulatory subunit p85 did not change (Fig. 16, A and B). PI3K activity is antagonized by PTEN (36), which inhibits myelination in Schwann cells together with Dlg1 (37). The levels of both Dlg1 and PTEN were increased in sciatic nerve of Myoc null mice compared with wild-type littermates (Fig. 16, A and B). On the basis of these results, we concluded that myocilin is able to signal through ErbBs, and this signaling may be related to myelination in the sciatic nerve.
Myocilin Induces Clustering of Nodal Components-Because myocilin interacts with several nodal components, we evaluated whether myocilin, similar to gliomedin (14), induces clustering of nodal proteins in isolated DRG explant cultures. DRGs, which had been permitted to regenerate axons for 5 days, were treated with myocilin for 30 min at room temperature, washed with basal media, and then incubated for an additional 24 h at 37 ºC. This treatment resulted in axonal clustering of NF186 and sodium channel Na v 1.2 (Fig. 17, A-F). Myocilin co-localized with Ͼ80% of NF186 clusters (p Ͻ 0.01). Preincubation of myocilin with polyclonal antibody against myocilin for 30 min at room temperature before DRG exposure significantly reduced the clustering effect of myocilin (Fig. 17, B and  F). Myocilin-induced clustering of nodal components requires signaling through ErbB1 and ErbB2, as the addition of GW2974, a dual inhibitor of these receptor tyrosine kinases, dramatically reduced the clustering effect of myocilin (Fig.  17D). We concluded that myocilin, similar to gliomedin, induces clustering of nodal components and that signaling through ErbB receptors contributes to this process.

DISCUSSION
Although a connection between mutations in the MYOC gene and glaucoma is well established, the role of wild-type myocilin is poorly understood. Our recent observations indicate that myocilin plays a role in several non-ocular tissues including skeletal muscle (10). In skeletal muscle, myocilin interacts with ␣-syntrophin and is part of the dystrophin-associated protein complex. Overexpression of myocilin in trans- genic mice leads to a redistribution of some proteins in the complex and increases its stability. In particular, the level of the active form of ␣-dystroglycan and laminin binding to ␣-dystro-glycan were markedly increased in muscle of transgenic mice as compared with wild-type littermates. Conversely, Myoc null mice demonstrated a moderate reduction in the amount of syn-  1000 dilution), and protein complexes were immunoprecipitated (IP) using protein G beads, eluted from the beads, separated by SDS-PAGE, and then probed with antibodies against ErbB2, ErbB3, or myocilin (1:1000). Immunoprecipitation with rabbit IgG was used as a control. The lower panel shows Western blot (IB) of lysates (10% input) before immunoprecipitation. I and J, ErbB2 phosphorylation in the sciatic nerves of wild-type, Myoc knock-out, and Myoc transgenic mice is shown. Note that larger amount of total protein was loaded, and longer exposure was used in I to detect phosphorylation of ErbB2 in Myoc null mice. trophin associated with dystrophin compared with wild-type littermates, which led to a reduction in the level of the active form of ␣-dystroglycan (10). Although we did not perform careful morphological or behavioral studies, adult Myoc null mice were viable, fertile, and did not demonstrate visible abnormalities including gross tremor (8,10).
Here we demonstrate that myocilin is involved in sciatic nerve myelination, interacts with and induces clustering of several nodal proteins, and participates in nodal integrity in the PNS. In the sciatic nerve, myocilin is actively expressed in Schwann cells but not in neurons. The absence of myocilin expression in axons reported here together with published results (13) argues against the idea that a reduced average diameter of axons in the sciatic nerve of Myoc null mice compared with wild-type littermates can be explained by neuron-autonomous effects of myocilin loss of function. Another olfactomedin domain-containing protein, gliomedin, has also been detected at PNS nodes where it was concentrated in Schwann cell microvilli (14). Unlike myocilin, which is secreted, gliomedin is a membrane-bound protein. However, the extracellular portion of gliomedin containing the olfactomedin domain may be cleaved in a furin-dependent manner and released into the extracellular space (17). Myocilin co-localizes with gliomedin in the nodal region of sciatic nerves and interacts with gliomedin via the N-terminal domain of myocilin that lacks the olfactomedin domain. Similar to gliomedin, myocilin interacts with NF186 and NrCAM and induces clustering of neurofascin and sodium channels in cultured dorsal root ganglion neurons in the absence of myelinating cells. Our data suggest that myocilin interacts with gliomedin, NF186, and NrCAM directly, but we cannot exclude a possibility that other unidentified proteins are necessary for such interactions and protein complex formation. The genetic elimination of gliomedin (16) similar to genetic elimination of myocilin (8) does not lead to abnormal gross phenotypes. We suggest that myocilin and gliomedin may perform complimentary functions in the PNS, although the effects of knocking out myocilin appear to be more severe than for gliomedin. It has been demonstrated that gliomedin null mice display no overt neurological abnormalities and exhibit normal nerve conduction. These mice formed compact PNS myelin that was indistinguishable from their wild-type littermates but exhibited disorganization and impaired attachment of Schwann cell microvilli to the nodal axolemma (16). It has been proposed that Schwann cells govern the assembly of PNS nodes by two overlapping independent processes that cooperate to provide reciprocal backup systems to ensure fast conduc- tion in myelinated nerves (16). Here we show that sciatic nerves of Myoc null mice have reduced levels of several myelin and basement membrane proteins including MBP, P0, and laminin 2 (␣2, ␤1, and ␥1 chains), thinner myelin sheaths, defective radial sorting of axons, and abnormal nodes of Ranvier when compared with wild-type littermates. We speculate that, similar to skeletal muscle, elimination of myocilin leads to a reduced stability of the dystrophin glycoprotein complex in Schwann cells (38), resulting in reduced levels of ␣and ␤-dystroglycans (Fig. 7, A and B). It has been previously shown that specific ablation of dystroglycan in Schwann cells led to profound nodal changes including reduced sodium channel density and disorganized microvilli (39). At the same time, the sciatic nerve of dystroglycan null mice did not show a significant reduction in the levels of laminin ␣2, ␤1, and ␥1 chains compared with wildtype mice (39). Therefore, reduced levels of laminin ␣2, laminin ␤1, and laminin ␥1 in Myoc null sciatic nerve compared with wild-type samples cannot be explained solely by the reduction in the levels of their dystroglycan receptor. Changes in the laminin 2 level may contribute to the observed defects in the axon myelination and Schwann cell differentiation in Myoc null mice (Figs. 10 and 11). It has been shown that Schwann cellspecific ablation of laminin ␥1 resulted in depletion of all other laminin chains in these cells. At P28, mutant sciatic nerve showed large bundles of unsorted axons with some Schwann cells located outside of axon bundles. Mutant Schwann cells lacked a continuous basal lamina (27). These defects were more severe than those observed in dy2J/dy2J or dy3K/dy3K mutants (27,40). Null mutation of laminin ␣2 chain in dy3K/dy3K mice resulted in elevated levels of laminin ␣4 chain, a component of laminin-8, in the sciatic nerve, which can partially compensate for the loss of laminin-2 (40). At the same time, mice hypomorphic for laminin ␥1 subunit expression demonstrated Schwann cell axonal sorting defects and continuous basal membranes with reduced levels of structural components (41). Defects in axon myelination in the sciatic nerve of Myoc null mice were less dramatic than the ones that were observed in the sciatic nerve after ablation of individual laminin chains and were more similar to ones observed in mice hypomorphic for laminin ␥1. It is also interesting to note that hindlimb muscle mass was gradually reduced in laminin ␥1 hypomorphic mice (41). Although we did not carefully analyze hindlimb muscle in Myoc null mice, we previously reported that overexpression of myocilin in hindlimbs led to an increase of their mass and average size of muscle fibers (10).
In the PNS, axon-derived factor neuregulin 1 (Nrg1) is one of the key factors regulating myelination (33,42). Nrg1 acts by activating a family of tyrosine kinase ErbBs. ErbB1 binds Nrg1 only after dimerization with ErbB4. ErbB2 does not bind Nrg1 or any known ligand with high affinity but is able to transmit the Nrg1 signal after dimerization with ErbB3 or ErbB4. ErbB3 lacks tyrosine kinase activity and transmits signals only after dimerization with ErbB2. Among homodimers, only ErbB4 dimers are able to transmit Nrg1 signals. Our data demonstrate that myocilin may also signal through the ErbB family. Because the level of ErbB4 is very low in sciatic nerve, we studied the interaction of myocilin with ErbB2-3. Similar to Nrg1, myocilin binds ErbB2/ErbB3 heterodimers, but unlike Nrg1 it also binds ErbB2 alone. K D values for myocilin binding to ErbB3 or ErbB2/ ErbB3 complex were higher than the corresponding reported values for Nrg1; reported K D values for Nrg1 binding to COS7 cells transfected with ErbB3 ranged from 1.9 to 11 nM (43,44). Myocilin binding leads to the activation of ErbB signaling as evident by the phosphorylation of Tyr-1221 in ErbB2. At present, we do not know whether myocilin competes with Nrg1 for binding to the same sites in ErbBs. Disruption of Nrg1 of Nrg1 or its receptors (ErbB2/ErbB3) leads to the nearly complete loss of Schwann cells followed by the death of motor and sensory neurons that they support (45)(46)(47)(48), whereas disruption of Myoc by null mutation reduces myelination but does not eliminate Schwann cells. Nevertheless, cultured Myoc null Schwann cells are more prone to apoptosis than those isolated from wild-type mice. It has been reported that laminin knock-out reduces phosphorylation of ErbB receptors in Schwann cells (27). In Myoc null mice, although decreased levels of laminins may contribute to a reduced phosphorylation of ErbB2, the reduction of laminin levels results from myocilin elimination. Moreover, overexpression of myocilin in transgenic mice led to an elevated phosphorylation of ErbB2 (Fig. 15J), supporting a primary role of myocilin in the regulation of ErbB signaling. It has been reported that disruption of neurofascin and gliomedin at nodes of Ranvier precedes demyelination in experimental allergic neuritis (49). These alterations correlated with the presence of serum antibodies against neurofascin and gliomedin suggest that autoantibodies against nodal proteins might contribute to disease progression in inflammatory demyelinating pathologies (49). It would be interesting to investigate whether similar changes occur with myocilin in patients with peripheral neuritis and various neuropathies. It would be also useful to test possible functional and structural defects in PNS of people with MYOC-related glaucoma, as these patients have not been systematically tested. Previously, we demonstrated that myocilin may serve as a modulator of the Wnt signaling pathway (19).
Here we show that myocilin may modulate ErbB signaling. Our data introduce myocilin as a novel player in sciatic nerve myelination and suggest that myocilin is a multifunctional protein that may have differing functions in ocular and non-ocular tissues.