Extracellular Trafficking of Myocilin in Human Trabecular Meshwork Cells*

Myocilin (MYOC) is a protein with a broad expression pattern, but unknown function. MYOC associates with intracellular structures that are consistent with secretory vesicles, however, in most cell types studied, MYOC is limited to the intracellular compartment. In the trabecular meshwork, MYOC associates with intracellular vesicles, but is also found in the extracellular space. The purpose of the present study was to better understand the mechanism of extracellular transport of MYOC in trabecular meshwork cells. Using a biochemical approach, we found that MYOC localizes intracellularly to both the cytosolic and particulate fractions. When intracellular membranes were separated over a linear sucrose gradient, MYOC equilibrated in a fraction less dense than traditional secretory vesicles and lysosomes. In pulse-labeling experiments that followed nascent MYOC over time, the characteristic doublet observed for MYOC by SDS-PAGE did not change, even in the presence of brefeldin A; indicating that MYOC is not glycosylated and is not released via a traditional secretory mechanism. When conditioned media from human trabecular meshwork cells were examined, both native and recombinant MYOC associated with an extracellular membrane population having biochemical characteristics of exosomes, and containing the major histocompatibility complex class II antigen, HLA-DR. The association of MYOC with exosome-like membranes appeared to be specific, on the extracellular face, and reversible. Taken together, data suggest that MYOC appears in the extracellular space of trabecular meshwork cells by an unconventional mechanism, likely associated with exosome-like vesicles.


Myocilin (MYOC) is a protein with a broad expression
pattern, but unknown function. MYOC associates with intracellular structures that are consistent with secretory vesicles, however, in most cell types studied, MYOC is limited to the intracellular compartment. In the trabecular meshwork, MYOC associates with intracellular vesicles, but is also found in the extracellular space. The purpose of the present study was to better understand the mechanism of extracellular transport of MYOC in trabecular meshwork cells. Using a biochemical approach, we found that MYOC localizes intracellularly to both the cytosolic and particulate fractions. When intracellular membranes were separated over a linear sucrose gradient, MYOC equilibrated in a fraction less dense than traditional secretory vesicles and lysosomes. In pulse-labeling experiments that followed nascent MYOC over time, the characteristic doublet observed for MYOC by SDS-PAGE did not change, even in the presence of brefeldin A; indicating that MYOC is not glycosylated and is not released via a traditional secretory mechanism. When conditioned media from human trabecular meshwork cells were examined, both native and recombinant MYOC associated with an extracellular membrane population having biochemical characteristics of exosomes, and containing the major histocompatibility complex class II antigen, HLA-DR. The association of MYOC with exosome-like membranes appeared to be specific, on the extracellular face, and reversible. Taken together, data suggest that MYOC appears in the extracellular space of trabecular meshwork cells by an unconventional mechanism, likely associated with exosome-like vesicles.
Myocilin (MYOC), 1 also known as trabecular meshwork inducible glucocorticoid response protein, is an acidic 504-amino acid protein. Structurally, MYOC contains at least two folding domains, an N-terminal coiled-coil and a C-terminal globular domain with significant homology to an olfactomedin module present in several different proteins (1,2). Mammalian proteins with the olfactomedin module localize to different compartments of the secretory pathway, although little is known about the function of these proteins or the olfactomedin module (3)(4)(5)(6)(7)(8)(9)(10).
Despite a broad expression pattern (1,2,7,(11)(12)(13), the function of MYOC remains unknown and its cellular distribution ambiguous. For example, MYOC has been observed in various cell types to associate with structures that are part of the secretory pathway, including endoplasmic reticulum, Golgi apparatus, and intracellular vesicles. Conversely, MYOC has also been reported to associate with mitochondria and cytoplasmic filaments (11, 14 -17). Even more unusual, MYOC appears to be secreted by some cell types, but not by others (2,18). Thus, MYOC is expressed by retinal ganglion cells, photoreceptors, and retinal pigment epithelium, but is not found extracellularly in the retina nor in conditioned medium of retinal pigment cells in culture (2,18,19). In contrast, MYOC is found in conditioned medium of trabecular meshwork (TM) cells in culture and in the aqueous humor that bathes the TM in vivo (15, 20 -22).
The unique extracellular appearance of MYOC in aqueous humor is interesting because mutations in MYOC cause glaucoma and blindness, but are not associated with any other disease process (23). Glaucoma is a disease commonly associated with elevated intraocular pressure because of decreased outflow of aqueous through the TM tissue (24 -29). One possible explanation for the broad expression pattern for MYOC but isolated pathology is that effects of mutant MYOC are tissuespecific. Thus, specificity may be achieved by differences in the normal localization (intracellular versus extracellular) of MYOC and its related function (1, 15-17, 20, 21).
In the present study, we investigate the method of release of MYOC from TM cells in an effort to better understand its function. Our results demonstrate that release of MYOC from TM cells is unlike the mode of release observed for two control proteins that use a conventional secretory mechanism. Moreover, results indicate that MYOC is released from TM cells in association with lipids, from which MYOC dissociates in the medium. We suggest that MYOC is released from TM cells on the outside surface of lipid particles that display biochemical characteristics consistent with exosomes, and which carry an exosome "marker" protein, major histocompatibility complex (MHC) class II (as HLA-DR). Our results also indicate a mechanism by which tissue-specific localization and pathology of MYOC is achieved, and suggests that, in most cells, MYOC is an intracellular protein, but is released with exosomes by the TM.

EXPERIMENTAL PROCEDURES
Cell Culture, Adenoviral Infection, and Interferon-␥ Treatment Previously characterized human trabecular meshwork (HTM) cell strains HTM26, HTM29, HTM84, HTM85, HTM86, and HTM90 were isolated and cultured in our laboratory as described (30,31). Cells (passage 2-4) were grown to confluence and maintained for at least 7 days in low glucose Dulbecco's modified essential medium (DMEM) (Invitrogen) supplemented with penicillin (100 units/ml), streptomycin (100 mg/ml), glutamine (0.29 mg/ml) solution (Invitrogen), and 10% fetal bovine serum (Hyclone Laboratories, Logan, UT) before experimentation. Prior to addition to culture medium, fetal bovine serum was centrifuged at 141,000 ϫ g max for 60 min in an SW27 rotor (Beckman Coulter, Fullerton, CA) to remove any particulate material. MCF7 cells (ATCC, Manassas, VA) were cultured as described, except that cell cultures were utilized prior to confluence (at ϳ70 -80% confluence) and were never allowed to reach confluence. For experiments analyzing recombinant proteins, HTM cell cultures were seeded, and 5-7 days later cells were infected with a replication-deficient adenovirus (AV) that contained the coding sequence for ␤-galactosidase or MYOC at a multiplicity of infection of ϳ1 (32). Cells were utilized 3-5 days postinfection. For experiments examining HLA-DR expression in HTM cell cultures and conditioned medium, confluent cells were maintained in supplemented DMEM as before, with the addition of 500 units/ml interferon-␥ (IFN␥) for at least 5 days prior to experimentation.

Cellular Fractionation
Cultures of confluent HTM cells (incubated in the presence or absence of 100 nM dexamethasone for 7 days) or subconfluent MCF7 cells were transferred to and maintained in low serum DMEM (either 1 or 0.5% fetal bovine serum) for 48 h. Conditioned medium was collected, and cells were rinsed in ice-cold phosphate-buffered saline (PBS) and placed on ice. Cells and conditioned medium were processed separately.
Cells-The cells were scraped into and lysed by a hypotonic buffer (5 mM N-ethylmaleimide, 10 mM EDTA, pH 7.45, 200 M phenylmethylsulfonyl fluoride, EDTA-free protease inhibitor mixture (Roche Diagnostics Corp.)) and homogenized using three strokes of a Dounce. Lysates were only crudely homogenized (3 strokes) to facilitate the clean separation of secretory vesicles from other cellular components. Cell homogenates were centrifuged at 1000 ϫ g for 10 min (giving nuclei and large organelles in pellet) using an FA 45-30-11 rotor (Brinkmann, Westbury, NY). Supernatant was transferred to a fresh tube and centrifuged at 100,000 ϫ g av for 60 min (giving membranes and vesicles in pellet) using a 50Ti rotor (Beckman Coulter). The resulting supernatant (considered cytosol) was recentrifuged at 100,000 ϫ g av , and pellets were resuspended in PBS and repelleted, to eliminate crossover contamination. Cytosol was combined with Laemmli sample buffer (33) (62.5 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 0.005% bromphenol blue, 5% ␤-mercaptoethanol) and incubated for 5-10 min at 100°C. Pellets were resuspended in hypotonic buffer, solubilized in Laemmli sample buffer, and incubated as described.
Conditioned Medium-The medium was centrifuged at 1,000 ϫ g for 5 min in an A4-44 rotor (Brinkmann) to remove dead cells and debris, and supernatant was combined with reducing Laemmli sample buffer and incubated for 5-10 min at 100°C.

Transmission Electron Microscopy
Confluent HTM cell monolayers were rinsed with PBS and fixed in 4% paraformaldehyde for 1.5 h, followed by 30 min in 3% glutaraldehyde in 0.1 M cacodylate buffer. Cells were washed 3 times for 10 min each in 0.1 M cacodylate buffer and postfixed in 2% osmium tetroxide in 0.1 M cacodylate buffer for 1 h. Samples were dehydrated in a graded ethanol series and embedded in Epox 812 resin. Thin sections were cut on a Reichert ultracut E ultramicrotome, stained with uranyl acetate and lead citrate, and examined on a Philips CM20 transmission electron microscope at 80 kV.

Isolation of Intracellular Membranes
Untreated confluent HTM cells were lysed, scraped, and homogenized as described above. Cell lysates were centrifuged first at 1,000 ϫ g (FA 45-30-11 rotor) for 1 min to pellet nuclei. Then, supernatant was transferred to a clean tube and centrifuged at 100,000 ϫ g av (50Ti rotor) for 60 min to isolate cellular membranes. Cellular membrane pellets were resuspended in PBS and repelleted to remove potential cytosolic contamination. Pellet was resuspended in 0.5 ml of Hanks' PBS (Invitrogen) and layered into linear sucrose gradients (0 -2 M sucrose in Hanks' PBS). Gradients were centrifuged at 38,000 ϫ g av for 3 h in a SW41 rotor (Beckman Coulter). Fractions were taken from the top of the gradient and analyzed for sucrose content with a refractometer. Samples were tested for protein distribution by silver stain and Western blot analyses.

Proteomic Analysis
Two prominent bands observed in silver-stained gels were excised and digested with trypsin. Sequence of protein fragments was identified by matrix-assisted laser desorption ionization qualitative time-of-flight mass spectrometry:mass spectrometry (Proteomics Core Facility, College of Pharmacy, University of Arizona). Protein identity and coverage was determined using SEQUEST software.

Isolation of Extracellular Membranes
Extracellular membranes were isolated from the conditioned medium of confluent HTM cells (with or without prior adenoviral infection, or with or without IFN␥ treatment) by differential centrifugation as described for the isolation of exosomes (34). Medium was centrifuged first at 10,000 ϫ g av (SW41 or SW28 rotor) for 60 min to isolate dead cells and cellular debris. Supernatant was transferred to a new tube and centrifuged at 100,000 ϫ g av for 60 min to isolate extracellular membranes using SW41 or SW28 rotors. Pellets were resuspended in PBS, repelleted (to control for non-membranous protein trapping), and solubilized in Laemmli sample buffer. Samples of conditioned medium, pre-and post-centrifugation, were combined with reducing Laemmli sample buffer. To control for sticking, post-centrifugation medium was recentrifuged at 100,000 ϫ g av (SW41 or SW28 rotors) for 60 min, resuspended in PBS, and repelleted as described. Non-visible pellets were solubilized in Laemmli sample buffer as described. Pellets and medium (and samples of purification medium as controls) were combined with Laemmli buffer, and incubated as described. For gradient isolation, purified extracellular membrane pellets were pooled and resuspended in 1 ml of 2.6 M sucrose, 20 mM Tris-HCl, pH 7.2. Linear sucrose gradients (2 to 0.25 M sucrose, 20 mM Tris-HCl, pH 7.2) were layered over membrane preparations, and gradients were centrifuged at 100,000 ϫ g av for 15 h in an SW41 rotor. Equal fractions were taken sequentially from the top of the gradient, sucrose concentrations were analyzed with a refractometer, and samples were solubilized in Laemmli buffer as described.

Metabolic Labeling
Human TM cells were seeded in 6-well culture plates at a density of 1 ϫ 10 6 cells per well. Five to 7 days later, cells were infected as described above. Three to 5 days post-infection, cells in each well were first washed with (3 ϫ 1 ml), then cultured in 1 ml of methionine/ cysteine (Met/Cys)-free DMEM (Invitrogen Corp). After 30 min, Met/ Cys-free DMEM was removed and replaced with 1 ml of Met/Cys-free DMEM containing 300 Ci/ml of [ 35 S]Met/Cys (MP Biomedicals, Irvine, CA). Following a 30-min labeling period, cells were washed with (2 ϫ 1 ml), then changed into 1 ml of DMEM containing a 10 times concentration of Met/Cys for 1, 2, 4, or 6 h.

Immunoprecipitation
After collection and placement of conditioned medium from metabolic labeling assays on ice, labeled human TM cells were placed on ice and washed once with 1 ml of ice-cold PBS, pH 7.5, and scraped into 500 l of ice-cold lysis buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 5 mM N-ethylmaleimide, protease inhibitor mixture, 200 M phenylmethylsulfonyl fluoride). Cell lysates were homogenized using 4 strokes in a glass Dounce and cleared of insoluble material by centrifugation at 12,000 ϫ g for 10 min. Supernatants and conditioned media were pre-cleared with protein A-agarose (Pierce Biotechnology) for 2 h with rotation at 4°C (or conditioned media were processed separately as described for the isolation of extracellular membranes). Beads were pelleted by centrifugation (12,000 ϫ g for 10 min), and supernatants were transferred to tubes containing 2 g of anti-green fluorescent protein or anti-MYOC IgG and incubated at 4°C. After 12 h, IgG-protein complexes were captured by addition of 50 l of protein A-agarose and incubation for 2 h with rotation at 4°C. Protein Aagarose beads were pelleted by centrifugation and washed with lysis buffer. Captured proteins were denatured by boiling in Laemmli buffer for 10 min and fractionated by SDS-PAGE. Fractionated proteins were transferred to nitrocellulose electrophoretically, and labeled proteins were visualized by exposure to x-ray film for 24 -72 h at Ϫ80°C.

Brefeldin A Treatment
Human TM cells were seeded in 6-well culture plates and infected with AV as described above. Cells were metabolically labeled with [ 35 S]Met/Cys for a 30-min period, and were either treated with 1 g/ml brefeldin A or left untreated. Cells were either lysed and lysates collected at time zero, or were cultured for 4 h in DMEM containing a 10 times concentration of Met/Cys, then cells were lysed, and conditioned medium was collected. In treated cells, brefeldin A was supplemented at 1 g/ml for the chase period. Medium and lysates were immunoprecipitated with nonspecific (anti-green fluorescent protein), anti-MYOC, or anti-alkaline phosphatase antibodies as described above.

Proteolysis Experiments
Proteolysis experiments were performed on purified extracellular membrane pellets. Conditioned medium (with dead cells and debris removed) was split equally in two, centrifuged at 100,000 ϫ g av (SW41 or SW28 rotor), and extracellular membranes were purified as previ-ously described. Pellets were resuspended in purified water containing proteinase K (7 mM Proteinase K, 10 mM EDTA, 5 mM N-ethylmaleimide) or were resuspended in purified water with 10 mM EDTA and 5 mM N-ethylmaleimide. Suspensions were incubated at 37°C for 30 min, then solubilized in Laemmli buffer and incubated at 100°C for 10 min.

RESULTS
Cellular Distribution of MYOC-Based on the apparent disparity between the localization of MYOC in HTM cells (intracellular and extracellular) versus other cell types (intracellular only), we were interested in addressing two issues: first, whether trafficking of MYOC differed from conventional secretory proteins, and, second, if dexamethasone treatment of HTM cells altered the cellular and extracellular distribution of MYOC (dexamethasone treatment uniquely induces MYOC expression in HTM cells, and thus enables positive identification of HTM cells; Fig. 1). To assess the subcellular localization of MYOC in HTM cell lysates and medium, we cultured confluent HTM cells in the presence or absence of dexamethasone, collected medium, and lysed and homogenized cells. Cells were only crudely homogenized (see "Experimental Procedures") to facilitate the clean separation of secretory vesicles from large membrane-bound organelles. MCF7 (human breast adenocarcinoma) cells also express native MYOC and were used as a control. We utilized sequential differential centrifugation to separate cellular constituents into cytosol, nuclei/large organelles, and plasma membranes/vesicles. Fractions were analyzed for the presence of MYOC by SDS-PAGE and Western blotting, and further analyzed for the presence of tissue plasminogen activator (tPA; secretory protein control), organellespecific markers (to verify fractionation), and ␤-actin (cytoplasmic marker).
Results showed that dexamethasone treatment of HTM cells increased the expression of MYOC in all fractions, and that MYOC was localized to both cytosolic and membrane fractions, plus appeared in conditioned medium. Lactate dehydrogenase assays verified that cell death was not responsible for MYOC appearance in conditioned medium (n ϭ 3; data not shown). Clean partitioning of fractions was verified using antibodies to a nuclear marker (histone deacetylase 2), a mitochondria marker (cytochrome oxidase subunit II), and a Golgi marker (Golgi 58K). In addition to marking traditional secretory vesi- ) were lysed in a hypotonic buffer and subjected to differential centrifugation, to first pellet nuclei and large organelles (LG. ORG), and then to pellet cellular membranes (MEMB). Resulting supernatant (cytosol; CYTO) was considered to contain only soluble proteins. Fractions were compared with whole cell lysate (WCL) and analyzed by SDS-PAGE/Western blotting using affinity purified antibodies specific for MYOC, anti-tPA (secretory vesicle marker), histone deacetylase 2 (HDAC2; nuclear marker), cytochrome oxidase subunit II (OxPhos; mitochondria marker), Golgi 58K (Golgi apparatus marker), and ␤ actin (loading control). Percent of total protein is indicated to demonstrate relative loading. 48-h Conditioned medium (MED) is shown as a comparison of intracellular versus extracellular protein distribution. Shown are representative blots from a total of three sets of independent experiments. cles, tPA was used to verify that cytosol was not contaminated with vesicle contents. In contrast to MYOC, tPA was found associated only with intracellular membranes and was excluded from the cytosol. Furthermore, tPA was down-regulated after dexamethasone treatment (Fig. 1A); consistent with a previous report (30).
Whereas the intracellular distribution was similar in the two cell types, MYOC was not detected in conditioned medium of MCF7 cells (Fig. 1C). Interestingly, intracellular MYOC structures in MCF7 cells label in a punctate pattern by immunofluorescence microscopy, similar to intracellular labeling of MYOC in HTM cells (data not shown) (11). Furthermore, the presence of tPA in MCF7 cell culture medium verified that the normal secretory pathway was functional, and confirmed that the extracellular localization of MYOC in HTM cells is different than another cell type expressing native MYOC. Taken together, MYOC in MCF7 cells most likely associates with a population of vesicles that is not shed in the same manner as in HTM cells.
Analysis of Intracellular Membranes-Despite a difference in cytosolic localization for MYOC and tPA, both proteins were observed to sediment with intracellular membranes. Thus we were interested in examining potential differences in intracellular membrane associations. To assess whether MYOC and tPA localized to similar or distinct populations of intracellular membranes, we floated membranes isolated from HTM cell lysates (free of nuclei and large cytoskeletal elements) into linear sucrose gradients. Fractions were analyzed by SDS-PAGE and Western blotting, and the distribution of MYOC was compared with marker proteins for various organelles ( Fig.  2A). Antibodies to proteins marking nuclei (HDAC2), mitochondria (OxPhos), endoplasmic reticulum (protein disulfideisomerase), and Golgi apparatus (Golgi 58K) recognized bands of appropriate size in the precleared fraction containing nuclei and large organelles, but not in the membranes floated into sucrose gradients (data not shown). Quantification of protein in each fraction was estimated by densitometry and represented graphically (Fig. 2B). MYOC equilibrated in a buoyant fraction, less dense than and distinct from tPA and LAMP1 (marking the lysosomal pathway). Because of the unusual buoyancy of this fraction, we sought to identify other proteins cofractionating with MYOC by mass spectrometry. With 28% amino acid FIG. 2. MYOC-associated membranes in HTM cells equilibrate at a density lighter than traditional secretory vesicles. Cellular membranes were isolated from HTM cell cultures by differential centrifugation and floated into linear sucrose gradients (n ϭ 3). Fractions were collected, and proteins were analyzed by SDS-PAGE/Western blotting (panel A). Protein content of fractions was determined by probing with antibodies specific to MYOC, tPA, lysosomal-associated membrane protein (LAMP1), and cyclophilin B (CyPB). Protein distributions were quantified by densitometry (panel B). Antibodies to nuclei, mitochondria, Golgi apparatus, and endoplasmic reticulum markers recognized bands of appropriate sizes in the precleared nuclei fraction but not in the experimental membrane fraction (data not shown). Linearity of gradients was verified by refractometry (inset, panel B). Stable monolayer cell cultures were further examined by transmission electron microscopy (panel C). Shown is an ultrathin section of a human trabecular meshwork cell containing exosome-like vesicles enclosed by multilaminar structures (arrows) (bar ϭ 500 nm). Inset shows magnification of the area outlined by a box to highlight limiting membranes of enclosed vesicles (arrowheads).
coverage, we discovered one other protein, cyclophilin B, in this buoyant fraction. Its presence in the MYOC-associated fraction was confirmed with a commercial antibody ( Fig. 2A). Cyclophilin B has been suggested to reside in the secretory pathway and function in immune processes (35,36), however, the significance of its presence in this fraction is not yet clear.
One population of vesicles displaying buoyant properties in both intracellular (in association with multivesicular bodies) and extracellular compartments are known as exosomes. To assess whether HTM cells produce such vesicles, we performed transmission electron microscopy on stable monolayers of cultured HTM cells (Fig. 2C). Interestingly, in all cells analyzed, we observed numerous multilaminar structures containing small vesicles ranging from 40 to 70 nm in size. These structures are similar in appearance to multivesicular MHC class II compartments found in antigen presenting cells known to produce exosomes (34,(37)(38)(39).
Analysis of Extracellular MYOC-Considering the observed distinction of MYOC-associated intracellular membranes from traditional secretory vesicles, the fact that both MYOC and tPA localize to the extracellular compartment, and our observation that HTM cells assemble multilaminar structures resembling multivesicular bodies, we questioned whether the extracellular appearance of these two proteins was by different mechanisms. Multivesicular bodies are known to release their lumenal contents (exosomes) into the extracellular compartment. To examine whether MYOC associated with such vesicles, conditioned medium from confluent HTM cells was processed by sequential differential centrifugation essentially as previously described for the isolation of exosomes (34,40). Whole medium was compared with extracellular membranes, pellets of dead cell/debris, and with soluble proteins (post-centrifugation medium) for the presence of MYOC and tPA (Fig. 3). Extracellular MYOC was distributed in all fractions, with the majority of MYOC soluble in the medium and a small proportion of MYOC associated with extracellular membranes. In contrast, tPA was found almost entirely soluble in the medium, with only a small amount sedimenting with dead cells and debris (likely containing secretory vesicles). Moreover, tPA was completely excluded from the extracellular membrane fraction in all trials (n ϭ 12). To limit trapping and sticking of non-membranous proteins to membranes and to the inside of the centrifuge tube, specific controls were performed; pellets were routinely purified by resuspension in PBS and by repelleting in a clean tube. Furthermore, post-centrifugation medium was recentrifuged, and any membranes pelleted by this step were analyzed for the presence of MYOC (supplementary materials Fig. 1). MYOC was absent from the repelleted post-centrifugation medium eliminating the possibility that the association of MYOC with extracellular membranes was a consequence of nonspecific sticking, and suggesting that MYOC associates specifically with extracellular membranes.
Examination of Recombinant MYOC-Based on our observations of the intracellular localization and extracellular distribution of MYOC versus tPA, we were next interested in exploring the movement of MYOC from cellular to extracellular compartments. To more clearly follow MYOC, we utilized an AV expression system to control expression of a recombinant MYOC protein. We were first interested in verifying that the expression of recombinant MYOC was comparable with the native protein. HTM cell cultures were infected with a replication-deficient AV encoding either MYOC or ␤-galactosidase. Conditioned medium was analyzed 3-5 days later, at sequential time points (6 -24 h) following medium change. Conditioned medium was precleared of dead cells, and analyzed by SDS-PAGE and Western blotting for the presence of MYOC (Fig. 4A). In MYOC AV-infected cells, recombinant MYOC appeared to accumulate in the extracellular space linearly over time. In cultures infected with ␤-galactosidase AV, native MYOC expression was not detected at any time point at film exposures tested (Fig. 4A). Because MYOC was detected as early as 6 h in conditioned medium (following media change), recombinant MYOC association with extracellular membranes was evaluated by differential centrifugation (Fig. 4B). Interestingly, the distribution of extracellular, recombinant MYOC was identical to the pattern for native MYOC seen in uninfected culture medium evaluated at 48 h (Fig. 3).
To analyze a potential early association of recombinant MYOC with extracellular membranes, HTM cell cultures were first infected with MYOC AV, and then metabolically labeled with [ 35 S]methionine/cysteine. Conditioned medium was collected at 1 h and subjected to differential centrifugation as before, and analyzed by SDS-PAGE and Western blotting (Fig.  4C). At 1 h, the extracellular distribution of radiolabeled, recombinant MYOC was unlike that seen in previous experiments (Figs. 3 and 4B). Instead, MYOC localized preferentially to debris and extracellular membrane pellets, and appeared in minimal amounts free in the medium, indicating that recombinant MYOC first reaches the extracellular compartment preferentially associated with extracellular membranes.
Trafficking of Recombinant MYOC-Based on our observation of an apparent linear accumulation of recombinant MYOC in the extracellular space, we were interested in addressing whether or not the cellular and extracellular pattern of MYOC displayed properties characteristic of a constitutive secretory protein. To address this question, HTM cell cultures were infected with MYOC AV or with an alkaline phosphatase AV engineered with a signal peptide to facilitate its constitutive secretion (secretory alkaline phosphatase; SeAP), and then metabolically labeled with [ 35 S]methionine/ cysteine. Labeled proteins were followed both intracellularly and extracellularly for 6 h. Conditioned medium was collected, and cells were lysed in a hypotonic buffer. Labeled MYOC was immunoprecipitated from lysates or medium with an anti-MYOC IgG, and labeled SeAP was immunoprecipitated with an anti-alkaline phosphatase IgG. Both proteins were analyzed by SDS-PAGE and Western blotting (Fig. 5, A  and B). As determined by phosphorimaging, the amount of radiolabeled intracellular MYOC decreased linearly over time, whereas the appearance of labeled MYOC in the extracellular space increased linearly. Interestingly, the mobility pattern of radiolabeled MYOC remained unchanged (both intracellularly and extracellularly), displaying a uniform  2). In an independent experiment (panel C), conditioned medium was collected at the earliest time point (1 h) and subjected to differential centrifugation as described. Equivalent samples were analyzed by SDS-PAGE, transferred to nitrocellulose, and exposed to film (n ϭ 3). ally modified in the ER-Golgi and is not processed as a typical secretory protein.
Effect of Blocking ER-Golgi Transport on MYOC Trafficking-The apparent dissimilarities between the processing of recombinant MYOC and SeAP led us to question whether, in HTM cells, these proteins are both similarly trafficked through the endoplasmic reticulum and Golgi apparatus. As before, we infected HTM cell cultures with either a MYOC or SeAP AV, metabolically labeled the cultures with [ 35 S]methionine/cysteine, and this time either treated the cells with 1 g/ml brefeldin A or left the cells untreated. Labeled proteins were followed for 4 h, at which time cell lysates and medium were collected, and MYOC and SeAP proteins were isolated by immunoprecipitation as before. Equivalent samples were analyzed by SDS-PAGE and Western blotting (Fig. 6, A and B). Treatment with brefeldin A halted secretion of both MYOC and SeAP at 4 h. Interestingly, the mobility pattern of the MYOC doublet remained unchanged between treated and untreated cells, suggesting that MYOC does not undergo post-translational modification in the ER-Golgi. Contrastingly, in SeAP AV-infected cells, treatment with brefeldin A completely eliminated the upper broad band, but not the lower band of the doublet, suggesting that trafficking through the Golgi and post-translational modification is required to produce the mature form of the protein.
Buoyant Properties of Extracellular Membranes-The association of MYOC with an extracellular membrane fraction led us to next question whether this fraction displayed biochemical properties similar to those described for exosomes (41,42). Despite the absence of a single marker protein found consistently in every exosome preparation from different cell types, we wanted to determine whether exosome preparations from HTM cells contain MHC class II antigens, the most common protein class found in exosome preparations (42, supplemental Table S1). MHC class II antigens are proteins involved in antigen recognition in the immune system. As the TM has been implicated to have an immune function and has been shown to express the MHC class II antigen, HLA-DR, in vivo (43,44), we sought to determine whether HLA-DR was present in extracellular membranes from HTM cells. HTM cells cultured in vitro require stimulation with IFN␥ to induce expression of HLA-DR (43,44), consistent with other cell types that shed exosomes containing MHC class II antigens (38,45). Hence, we either treated confluent HTM cell cultures with 500 units/ml of human IFN␥ for a period of at least 5 days or left the cells untreated. At 48 h after medium change, conditioned medium and cell lysates were collected, and medium was processed by differential centrifugation as before. Samples of cell lysate and extracellular membranes from treated and untreated cells were analyzed by SDS-PAGE and Western blot, and the expression of HLA-DR and MYOC was analyzed using specific antibodies (Fig. 7A). As expected, both treated and untreated cells expressed similar levels of MYOC in both cell lysate and extracellular membrane fractions; yet only IFN␥-treated cells expressed HLA-DR. Significantly, we observed that extracellular membranes from IFN␥-treated cells contained the MHC class II antigen (Fig. 7A, lane 4), consistent with reports of exosomes from other cell types (34,(37)(38)(39)(40)(45)(46)(47)(48)(49)(50)(51).
To test whether MYOC and HLA-DR are found associated with extracellular membrane populations of similar densities, sucrose gradients were layered on top of purified preparations of pooled extracellular membranes from IFN␥-treated and untreated cells (essentially as previously described for the analysis of exosomes). Gradients were ultracentrifuged overnight, and fractions were collected and analyzed for the presence of MYOC and HLA-DR. Fractions from untreated cells were also analyzed by SDS-PAGE followed by silver staining for the distribution of all membrane-associated proteins. Silver staining revealed that the majority of proteins equilibrated at a density of 1.142 g/ml (Fig. 7B). Western blotting showed that MYOC equilibrated preferentially with these proteins at a density peak of 1.142 g/ml (Fig. 7B), whereas HLA-DR was undetectable in untreated cells (data not shown). In contrast, in IFN␥-treated cells, HLA-DR (Fig. 7C) and MYOC (data not shown) equilibrated in the same fractions, having a density of 1.128 -1.147 g/ml, consistent with the reported density of exosomes in other cell types (34,37,46).
Degradation Analysis by Proteinase K-To assess the orientation of endogenous MYOC on extracellular membranes, conditioned medium was subjected to sequential differential centrifugation as before, except that prior to high-speed ultracentrifugation, medium was divided equally in half. Purified pellets were incubated in a proteinase K or a control buffer. Equivalent fractions were analyzed for the presence of MYOC (Fig. 8). Analyses indicated that MYOC was completely degraded in the presence of proteinase K, and was not degraded by a control buffer under identical conditions. This finding indicates that endogenous MYOC is localized to the outside of the extracellular membranes. ]methionine/cysteine, and chased with normal medium containing 10ϫ methionine/cysteine. Cells were either treated with brefeldin A (BFA; 1 g/ ml) for the period of the pulse and chase (ϩ), or were left untreated (Ϫ). As before, cell lysates (C) and conditioned medium (M) were collected at the time points indicated and immunoprecipitated with either nonspecific (CON; anti-green fluorescent protein), anti-MYOC (panel A), or anti-alkaline phosphatase (panel B) antibodies. Equivalent samples were analyzed by SDS-PAGE, transferred to nitrocellulose, and exposed to film (n ϭ 2).

DISCUSSION
In this study, we are first to follow nascent MYOC over time and show that MYOC, in a cell-type specific manner, exits the cell via a mechanism atypical of secretory proteins. We show that, in two different cell types, native MYOC localizes to the cytosol and intracellular membranes, suggesting that MYOC is an intracellular protein with the capacity to associate with and function in the secretory pathway. Such a distribution differed from a secreted protein, tPA, which was excluded from the cytosol and found limited to cellular fractions containing membranes. Interestingly, in both cell types tested, tPA was located extracellularly as expected; however, MYOC was only found outside of one of the cell types: HTM cells.
To examine more closely the unique extracellular appearance of MYOC in HTM cells, we utilized an adenovirus expression system in combination with a metabolic labeling protocol. We followed MYOC over time and compared the processing of nascent MYOC in HTM cells to a secreted protein, SeAP. For example, we show that the characteristic mobility pattern of MYOC does not change between intracellular and extracellular compartments over time. Our observations of a uniform extracellular doublet at 1 h, plus the similarity between mobility patterns at 0 and 1 h in cell lysates, are consistent with the idea that insufficient time has elapsed for MYOC to translocate through the ER and Golgi and undergo post-translational modification. In fact, blocking ER-Golgi transport inhibits glycosylation of SeAP controls, but has no effect on the mobility pattern of the MYOC doublet, supporting the argument that MYOC does not undergo Golgi-mediated post-translational modification. This finding is consistent with a previous study (52) and in opposition to another (53) that examined changes in the mobility pattern of MYOC following treatment with enzymes that cleave sugar groups.
Except for excluding glycosylation, the scope of the present study did not include determining experimentally the nature of the two forms of MYOC on SDS-PAGE. We speculate that the doublet may be a consequence of one of a number of possibilities: first, MYOC may be subject to another type of post-translational modification, such as phosphorylation; second, MYOC may utilize two alternate start sites (Met-1 and Met-15); third, MYOC may have two protein conformations that migrate differently on SDS-PAGE. Further research is necessary to determine the exact nature of the MYOC doublet.
Evidence of an alternative mechanism of release of MYOC from HTM cells was demonstrated in experiments showing that both native and recombinant MYOC co-purify with extracellular membranes having biochemical properties of exosomes. Our temporal characterization of extracellular MYOC illustrates that, at early time points after release, the majority of MYOC was found associated with exosome-like membranes. But, by 48 h, most of MYOC was a free protein, suggesting that MYOC was released from HTM cells bound to membranes from which it dissociates over time into the medium. Consistent with this idea, MYOC in the extracellular, exosome-like fraction from HTM cells is susceptible to protease digestion.
MYOC likely associates with the secretory pathway in a manner not requiring a functional signal sequence. Consistent with our data showing a cytosolic localization of endogenous MYOC and a mobility pattern unaffected by brefeldin A, others have shown that the putative signal sequence of MYOC is MYOC-associated membranes display a protein composition and equilibrium density typical of exosomes. Panel A, cultures of IFN␥treated (ϩ) or untreated (Ϫ) confluent HTM cells were lysed, and cell lysates and conditioned medium (48 h) were collected. Medium was subjected to differential centrifugation as before, and cell lysates (WCL) were compared with extracellular membranes (EX. MEMB) by SDS-PAGE and Western blotting for the presence of MYOC and HLA-DR. Panels B and C, equivalent volumes of pooled extracellular membranes from conditioned medium (48 h) of 10 untreated (panel B) or IFN␥treated (panel C) confluent HTM cell cultures (75 cm 2 ) were floated up a linear sucrose gradient. Equal fractions were collected, and parallel samples were analyzed by SDS-PAGE, then silver staining (panel B) or Western blotting (panels B and C) (n ϭ 3). Silver stains were visualized for all proteins, and Western blots were probed for MYOC, or for HLA-DR and MYOC (not shown). Gradient fraction densities were verified by refractometry (data not shown). The majority of proteins (including MYOC and HLA-DR) from these preparations were found between densities 1.128 and 1.147g/ml, consistent with the equilibrium density of exosomes. non-functional in HTM cells (1) and transfected cells (54). In contrast, in a cell-free system, the putative signal peptide of MYOC was either functional or nascent MYOC bound tightly with reticulocyte membranes (55). We suggest that the signal peptide is inactive and that the N-terminal coiled-coil of MYOC facilitates interactions with the secretory pathway, much like that of proteins of the SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) family. Thus, on the surface of exosomes, MYOC may function in the secretory pathway that is responsible for exosome targeting or release into the extracellular compartment. Because MYOC is a cytosolic protein that also associates with cellular membranes, we propose that it may interact via its coiled-coil with the complimentary coiled-coil of a transmembrane protein located on the exosomal surface, as part of a cytoplasmic coat. The deficiency of extracellular MYOC in other cell types may be explained by the association of MYOC with a different population of intracellular vesicles that are not released as exosomes or may be related to a lower (or lack of) production and release of exosomes in other MYOC-expressing cell types. Additional work is needed to resolve these possibilities.
The association of MYOC with extracellular membranes containing HLA-DR, an MHC class II antigen, that display exosome-like properties is the first evidence to place MYOC in a specific cellular pathway. Interestingly, no single protein or group of proteins was found consistently in exosomes from different cell types. Most exosome-associated proteins appear to be cell type-specific. Hence, exosomes are typically classified not by their protein composition but by biochemical properties, as was done in the present study. However, in two-thirds of the cell types described to release exosomes, MHC class II is a prominent component of these vesicles (34,(37)(38)(39)(40)(45)(46)(47)(48)(49)(50)(51). Our observation that MYOC colocalizes with HLA-DR on membranes with a density typical of exosomes is compelling evidence that the release of MYOC from HTM cells occurs via an exosome-mediated mechanism. Exosomes have been described in a number of cell types including B lymphocytes, reticulocytes, dendritic cells, and intestinal epithelium (34,39,40,56). In such cells, exosomes have been proposed to function in immune response processes and intercellular communication (57). Because the TM is both in direct contact with the aqueous humor that passes through the anterior chamber of the eye and positioned at a strategic location between the anterior chamber and the venous circulation (the blood-aqueous barrier), and because there is a lack of lymphatic vessels draining the intraocular compartment, it is likely that the uptake and processing of antigens may occur in the TM prior to reaching the venous blood circulation (44). An attractive possibility based on the physiology and architecture of the TM is that MYOCassociated exosomes function in the initiation of ocular immune responses that may play a role in the regulation of intraocular pressure in the normal and glaucomatous human eye. The connection between mutations in MYOC, immune function, and glaucoma is presently unknown.
Understanding the function of MYOC was of clinical importance. Mutations in MYOC link only to forms of open angle glaucoma, despite a near ubiquitous tissue expression pattern (2,14,23). Thus, the function of MYOC is likely crucial only in the eye. Given that all patients with MYOC-linked glaucoma have elevated intraocular pressure, MYOC likely has a vital function at the primary site of intraocular pressure regulation: the trabecular meshwork. Our results suggest that MYOC is released from HTM cells, but not other cells as previously reported. Furthermore, we suggest that HTM cells traffic MYOC in a unique fashion, and specifically release MYOC coincident with exosomes. Thus, we hypothesize that the cell type-specific release of MYOC by HTM cells may account for the tissue-specific nature of the pathology caused by MYOC mutations.
FIG. 8. Proteinase K treatment degrades extracellular MYOC in extracellular membrane fraction. Conditioned medium from HTM cells (48 h) was subjected to differential centrifugation as described (Fig. 3), except that before centrifugation at 100,000 ϫ g, medium was divided equally in two. Derived extracellular membrane pellets (EX. MEMB) were incubated in the absence (Ϫ) or presence (ϩ) of proteinase K (PROT. K) for 30 min at 37°C. Fractions were analyzed by Western blot and compared with post-centrifugation medium (soluble proteins; SOL. PROT). Shown is one representative experiment of three total.