Localization of Myocilin to the Golgi Apparatus in Schlemm’s Canal Cells

Abstract

purpose. Biochemical and genetic evidence suggests that overexpression of or mutations in myocilin within the cells of the aqueous humor outflow pathway play a significant role in the development of steroid-induced and several other open-angle glaucomas. As a baseline to understanding the normal and pathologic function of myocilin, we determined the subcellular localization of myocilin in steroid-treated human Schlemm’s canal endothelial (SC) cells.

methods. SC cells were grown to confluence, treated with dexamethasone for 10 days, and then stained using antibodies against myocilin, tubulin, orβ -COP (a specific golgi protein) or vital stains for endoplasmic reticulum (ER) and golgi. Brefeldin A (BFA) and nocodazol (NZ) were used to disrupt the golgi or microtubules.

results. The authors found that myocilin staining was (a) always centered around the centrosome, (b) very similar to the pattern seen with NBD-ceramide, (c) was disrupted in characteristic ways by BFA and NZ and (d) showed extensive colocalization with β-COP.

conclusions. Results indicate that myocilin is localized to the golgi in SC cells. Such localization is consistent with myocilin being processed for secretion but is also consistent with sequence analysis and other data that suggest that myocilin or myocilin mutations might be targeted to the cytoplasmic face of the golgi, and under some circumstances play a role in or interfere with golgi or vesicle function. How such interference could eventually lead to open angle glaucoma is discussed.

The trabecular meshwork-inducible glucocorticoid response/myosin-ciliary stalk–related (TIGR/myocilin) protein has been implicated by biochemical and genetic evidence to be in the pathway that results in either steroid-induced or several additional open angle glaucomas. ^{1 2 3 4 5 6 7 8 9 10 11} Myocilin was initially of interest to glaucoma research because trabecular meshwork (TM) endothelial cells greatly increased their expression of the protein in response to treatment with glucocorticoids. Thus, myocilin provided a mechanistic link from TM cells to an important form of open angle glaucoma. ^{2 5} Studies to identify genes associated with diseases of the retina had named the protein myocilin because of sequence homology to myosin in the N-terminal half of the protein and because it localized to the ciliary root of the rod inner segment. ¹² Interest in myocilin increased dramatically when mutations associated with juvenile onset open angle glaucoma were linked to the area in chromosome 1 that coded for myocilin. ³ Many subsequent genetic linkage studies have demonstrated a strong relationship between mutations in myocilin, specifically in the C-terminal, olfactomedin-homology domain, with both juvenile and adult onset open angle glaucomas. ^{1 4 5 6 7 8 9 10 11 13} Yet the role that myocilin normally plays in outflow pathway cells and how mutated forms of myocilin might cause reduced outflow, increased intraocular pressure, and loss of vision have not been established. 

Using antibodies generated in the laboratory of Jon Polansky to full-length recombinant myocilin, ⁵ we have recently shown that human Schlemm’s canal endothelial (SC) cells also express myocilin in culture when exposed to dexamethasone (Dex) but that the pattern of staining is different from that seen in TM cells. ¹⁴ Steroid-treated TM cells exhibited myocilin staining throughout the cell body, whereas in SC cells myocilin staining was confined to a ribbon-like compartment near the nucleus. Because the extensive staining for myocilin in TM cells, present even in nonsteroid controls, would have overwhelmed particular localizations, we used SC cells to determine the identity of the myocilin-stained organelle. Our hypothesis was that myocilin-stained organelle was either the golgi apparatus and small vesicles or the ER. Identification of the particular compartment found to contain myocilin staining is a useful first step in understanding the functional role myocilin plays in the cells of the outflow pathway. 

Methods

SC cells were obtained from nonglaucomatous human donor tissue using methods described previously. ¹⁴ SC cell cultures of characteristic growth rate and fusiform shape at confluence were grown in Medium 199 (Gibco, Grand Island, NY) with 12% FBS, penicillin, streptomycin, and amphotericin B, at 36°C in 3% CO₂. For experiments, cultures of less than passage 8 were plated onto gelatin-coated glass coverslips in six-well plates and grown to confluence. At confluence, the serum concentration was reduced to 2%, and the cells were treated with 1 μM Dex (Sigma, St. Louis, MO) in EtOH, or EtOH control, such that EtOH was 0.1% by volume. The medium was replaced on the fifth day after Dex addition, and cells were maintained for 10 days total in the presence of Dex. After 10 days in Dex, the cells were washed briefly in PBS and fixed in 4% freshly made formaldehyde in PBS for 5 minutes, permeabilized for 5 minutes with 0.5% Triton X-100 in PBS, and washed three times in PBS with 5 minutes between washes. Primary antibodies to myocilin (anti–full-length recombinant myocilin polyclonal, kindly provided by Jon Polansky ² ), β-COP (Sigma, clone maD),β -tubulin (Sigma, clone Tub 2.1), acetylated tubulin (Sigma, clone 6-11B-1), and nonimmune rabbit or mouse serum were used, at the appropriate dilutions. After washing, goat anti-rabbit FITC and anti-mouse TRITC-conjugated secondary antibodies (affinity purified IgGs; BioSource, Camarillo, CA) and 1/1000 dilution of diamidino phenylindole (DAPI; Sigma) were used to stain the primary antibodies and DNA. 

For live cell experiments, no Dex exposure was used. Vital stains specific for the ER and golgi (ER tracker or NBD C₆-ceramide; Molecular Probes, Eugene, OR) were added to the cells for 20 to 30 minutes at 36°C, and then the coverslips were removed and quickly placed face down on a cleaned glass microscope slide. Two layers of double-sticky cellophane tape were used to provide a space between the coverglass and slide. Medium without stain was added to the space, and the chamber was then sealed with fingernail polish. The cells were then placed in a warmed (35–36°C) chamber surrounding a Zeiss Axioplan (Thornwood, NY) upright microscope and viewed with appropriate filter sets, and 25× and 100× planneofluor objectives. 

For drug addition experiments with cells that would be viewed live or fixed, brefeldin A (BFA; Sigma) was used at a final concentration of 5 μg/ml. Fifteen microliters of a 1 mg/ml stock was added to 3 ml of medium. Nocodozol (Sigma) was dissolved in DMSO at 10 mM stock, and added as a 1/1000 dilution to the cells (3 μl/3 ml). Controls had vehicle (DMSO or PBS) added at the same time and volume. Drugs were present for 30 minutes before fixation or viewing live cells. For live cell experiments, the vital stain was added 10 minutes before drug addition. All experiments were replicated three or more times with three independent SC lines (from normal donors). Control cells in the same figure are from the same experiment with cells treated with the appropriate vehicle(s). Images for both fixed and live cells were obtained using a Photometrics CH 250 cooled CCD camera using Image Processing Laboratory (Scanalytics, Inc., Billerica, MA) software. Only Figure 3 used a sharpening filter (unsharp mask; 250%, 3.2 pixel radius and two-level threshold) within Photoshop (Adobe, Inc., San Jose, CA) and only on the red channel. Color figures were merged in the IP laboratory from individual grayscale images. 

Results

To ascertain the position of myocilin staining within the cell relative to the centrosome, we doubly stained Dex-treated SC cells with antibodies to both myocilin and to acetylated tubulin. We tried several antibodies that we expected to stain the centrosomal area, such asγ -tubulin, but these did not stain our cells. However, we had noticed that SC cells, being endothelial, consistently showed a prominent primary cilium that stained with either β-tubulin or acetylated tubulin antibodies. Figure 1 shows the pattern of myocilin staining relative to the primary cilium. The left column (A, C, E, G) shows the rhodamine channel only, with the primary cilia identified with arrows. The cilia were each 4 to 6 μm long and always situated adjacent to one “end” of the each elongated nucleus. Figures 1B 1D 1F and 1H show the same cells as on the left, but with both the FITC channel and DAPI channel included. Note that myocilin typically stains only one cell per view, consistent with our earlier results that only 5% to 15% of cells stain for myocilin even after Dex treatment. ¹⁴ However, every cell that had myocilin staining showed the myocilin surrounding that cell’s primary cilium. This is what would be expected if myocilin was present in or attached to the surface of the golgi apparatus. 

Figure 2 compares vital stains for ER and golgi using ER tracker (Figs. 2A 2B) and NBD C₆-ceramide (Figs. 2C 2D) . We present ER tracker as a grayscale image because when presented in blue, it did not show sufficient detail. The left panels (Figs. 2A 2C) show lower magnification (25× objective) views of living SC cells, whereas the panels on the right show 100× views (scale bars, 50 and 10 μm, respectively). ER tracker shows very fine tubelike stings throughout the cells. These are only barely visible at low magnification but are readily apparent in 100× views. In contrast, the NBD-ceramide stain shows a relatively compact, ribbonlike compartment near the nucleus in every cell, as seen with myocilin. 

If myocilin colocalizes with the golgi, it should be dependent on intact microtubules for its localization to the peri-centrosomal region. We therefore asked whether myocilin localization would be significantly altered by nocodazol (NZ), which rapidly disassembles all but the most stable cellular microtubules. ¹⁵ Figure 3 shows an example of myocilin staining without (Fig. 3A) or after 30 minutes of 10 μM NZ (Fig. 3B) . Microtubules are shown in red. Myocilin staining was typically perinuclear in control cells, but with most of the microtubules disassembled by NZ, myocilin staining was dispersed into numerous punctate bodies or vesicles throughout the cell body. This was the case in all nocodazol-treated cells, consistent with observations in other cell types that show golgi disaggregation into vesicles after NZ treatment. ^{16 17} The primary cilium in Figure 3A was present near the myocilin stain but was out of the plane of focus. 

BFA blocks the separation of golgi membrane from the ER, ^{18 19 20} allowing golgi membrane components to disperse into the ER. ²¹ BFA treatment at 5 μg/ml for 30 minutes disrupted both golgi organization, as judged by NBD ceramide staining in live cells (Figs. 4A 4B ) and the normal pattern of myocilin staining in cells fixed after treatment (Figs. 4C 4D) . Because relatively few cells stain with myocilin, we show the myocilin staining pattern at high magnification. In contrast to the NZ results, BFA consistently caused a much less punctate, more diffuse staining pattern with myocilin than the large vesicular staining pattern of myocilin seen after NZ, consistent with dispersal of the golgi membrane into the ER. Note that in Figure 4C and to a lesser extent in 3A, the myocilin-stained compartment extended further around the nucleus than usual. However, this is consistent with ceramide staining, which shows that although the majority of staining is confined to one side of the nucleus, sometimes a small projection of golgi membrane extends part way around the nucleus. 

As a final test of the specificity of myocilin to the golgi, we stained cells with antibodies to both myocilin and to the golgi-specific protein β-COP. Figure 5 shows a comparison between myocilin (A, D, G, J) and β-COP (B, E, H, K) localization. Figures 5C 5F 5I and 5L show the merged image of FITC and rhodamine fluorescence. The cells in the last two rows (G through L) had been treated with NZ (G through I) or BFA (J through L) as described above. Although myocilin and β-COP appeared to stain identical compartments, the myocilin stain consistently appeared more punctate than did β-COP. After treatment with NZ, myocilin staining moved throughout the cell body in many distinct vesicle-like components (Figs. 5G and 5I and Fig. 3B ). β-COP did not show the same dispersed, bright vesicular staining pattern after NZ, but rather a golgi remnant was visible in most cells at the time point tested (30 minutes) along with a diffuse background staining (Fig. 5H) . Both β-COP and myocilin dispersed after BFA, consistent with both dispersing into the ER, but the regions of maximal staining sometimes did not coincide, as shown in Figure 5L . 

Discussion

Our results are consistent with localization of myocilin to the golgi apparatus. Identification of the myocilin-stained organelle as the golgi is based on (1) the location of staining centered around the microtubule organizing center (centrosome); (2) strong similarity between the pattern of myocilin staining and that seen with NBD-ceramide; (3) dispersal of myocilin staining with BFA and NZ in a manner almost identical with that seen with NBD-ceramide under the same conditions; and (4) substantial colocalization with β-COP, a protein known to localize specifically to the golgi. 

The base of the primary cilium marks the site of the centrosome, where the “minus” ends of most cellular microtubules are anchored. ^{22 23} Golgi membranes are tethered via accessory proteins to the microtubule motor protein dynein, through the dynactin protein complex. ²⁴ Dynein moves golgi membranes (and vesicles) toward minus microtubule ends and in this manner positions the golgi near the centrosome and the nucleus. ^{16 17} Thus, localization of myocilin to a compartment surrounding the base of the primary cilium strongly suggests that the myocilin-stained compartment is the golgi. Interestingly, Kubota’s original observation that myocilin was located at the root of the ciliary stalk in rod outer segments ¹² could also be related to our findings. The golgi apparatus is very large in rod cells and is generally found throughout the inner segment just underneath and surrounding the base of the ciliary stalk. ²⁵  

The pattern of NBD-ceramide staining in SC cells was very similar to that determined to be golgi in other cell types ^{26 27} and was almost identical with that seen with myocilin. The main difference was that the myocilin-stained compartment was typically less extensive than that seen with ceramide, suggesting the possibility that myocilin is restricted to a subset of the golgi. This impression is strengthened by the observation that, although there was substantial colocalization of myocilin with β-COP, they appeared to distribute differently after exposure to BFA or NZ. β-COP is thought to play a role in trafficking from the intermediate compartment of the ER to the golgi and thus predominantly stain the “cis” golgi. ^{28 29 30 31 32} The slight difference in localization of myocilin and β-COP may indicate a location of myocilin more “trans” than β-COP, that is, toward the formation of vesicles. 

What does localization of the myocilin protein to the golgi apparatus mean? Because the golgi apparatus is the site of sorting of proteins for export from the cell, localization to the golgi is consistent with the idea that myocilin protein is processed for export from the cell. ^{2 5} Similarly, the golgi is the site within which glycosylation of proteins occurs, ³³ consistent with evidence that myocilin is glycosylated before it is secreted. ⁵ Although vesicles were not prominently stained in our cells, we often did observe small vesicles staining for myocilin, consistent with reports by Stamer et al. ³⁴ Therefore, our results are consistent with the idea that myocilin overexpression due to glucocorticoid treatment could alter aqueous humor outflow by acting extracellularly. Extracellular myocilin, potentially via dimerization or oligomerization and accumulation within the outflow pathway, could clog or otherwise impede the flow of aqueous humor from the eye. ^{2 5 35}  

Our results are also consistent with an alternative mechanism of golgi localization. Recent work has shown that proteins that function specifically in the golgi are targeted there by particular amino acid sequences. ^{36 37 38} If myocilin has homology to golgi proteins, it could also be targeted there by the same mechanism. Using NIH’s PSI-BLAST software, ³⁹ we asked whether myocilin had homology to proteins known to be functionally associated with the golgi or vesicles. Table 1 presents 13 golgi or vesicle-related proteins to which myocilin had the most significant homology. Nine proteins are specifically found at the cytoplasmic face of golgi membranes, dynactin was described above as essential for golgi positioning or minus end-directed vesicle movement, and kinesin motors move vesicles and other cargo, usually toward microtubule plus ends. ⁴⁰ Most of the homology recognized by PSI-BLAST was found in short (6–8 amino acid) motifs within the N-terminal half of myocilin, but inclusion of the olfactomedin domain in the search always improved the significance of the scores obtained. Direct inspection of the human myocilin sequence disclosed three additional motifs homologous to “golgi localization domains,” and all three were in the olfactomedin domain. These sequences were located at amino acid numbers 298 (FEYDL), 369 (FPYS), and 451 (FAYD). Each of the sequences contains a tyrosine two-amino acid C-terminal to a phenylalanine, which is thought to be most critical to golgi targeting. ³⁷ Although further work will be necessary to determine if any of the identified sequences confer golgi localization, it should be noted that these sequences would act along with the motifs found in the N-terminal portions of myocilin and that even low affinity may be all that is necessary to produce localization. 

Recent evidence has shown that open angle glaucoma–related mutations in myocilin result in the protein becoming part of a triton-insoluble pool, ⁴¹ and being blocked from secretion. ⁴² Thus, it becomes likely that myocilin mutations cause pathology by acting intracellularly. A mechanism to explain pathogenesis by intracellular, mutated myocilin would be if altered myocilin was targeted to the cytoplasmic face of the golgi and vesicles and now blocked the function of normal golgi and/or motor proteins, thus interfering with golgi trafficking, sorting, and transport functions. Such interference would vary in extent and kind with the type of mutation present, but even a relatively slight interference in golgi function would likely decrease cell viability. Such a mechanism could be a factor in the best-documented evidence of cellular pathology of the outflow pathway, the decreased number of TM cells on the trabeculae in patients with primary open angle glaucoma. ^{43 44 45 46} Such loss of cellularity would expose the underlying collagen and extracellular matrix to enzymatic degradation and collapse, closing pathways for aqueous outflow and gradually raising intraocular pressure. Thus, although SC cells have proven useful in proving myocilin localization to the golgi, future studies will return to TM cells to assess the effects of myocilin mutations on cellular viability, intracellular myocilin localization and movement, and golgi and vesicle function. 

 Supported by National Institutes of Health, National Eye Institute Grant EY12172, a grant from the Glaucoma Research Foundation, and NIH Grants EY02477 and EY05722 (a core research grant from the NEI to the Duke University Eye Center).

 Submitted for publication March 7, 2000; revised May 30 and July 14, 2000; accepted July 28, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Pattern of myocilin staining in SC cells relative to the primary cilium. Immunofluorescence micrographs of acetylated tubulin (A, C, E, G) and myocilin (green), acetylated tubulin (red) and DAPI (blue) (B, D, F, H). Each left-right pair shows the same view. Arrows point to the primary cilium of each cell. Scale bar, 10 μm.

Figure 2.

 

Comparison of endoplasmic reticulum versus golgi stains in live SC cells using vital stains for ER (A, B) and golgi (C, D). Original magnification, ×25 (A, C), ×100 (B, D). Scale bars, 50 and 10 μm, respectively.

Figure 3.

 

Myocilin staining after nocodazol treatment. Immunofluorescence micrographs of SC cell in the absence (A) or presence (B) of 10 μM nocodazol for 30 minutes. Myocilin is green and β-tubulin stained microtubules are shown in red (rhodamine). Scale bar, 10 μm.

Figure 4.

 

Effect of BFA on golgi structure. SC cells were treated with vehicle control (A, C) or 5 μg/ml BFA (B, D). For live cell studies (A, B), 15 minutes after addition of BFA or control the cells were stained with NBD C₆ ceramide and observed at 37°C. Parallel control (C) or BFA-treated (D) cells were fixed after 30 minutes in BFA and stained for myocilin. Original magnification, ×25 (A, B; scale bar, 50 μm) or× 100 (C, D; scale bar, 10 μm).

Figure 5.

 

Colocalization of myocilin and β-COP. Myocilin-stained (green; A, D, G, J), β-COP–stained (red; B,E,H,K), and combined images (C, F, I, L) are shown from ×100 original magnification views. (A through F) Cell stained with no drug treatment; (G through I) cells treated for 30 minutes with nocodazol; (J through L) a cell treated with BFA. Scale bar, 10 μm.

Table 1.

 

Golgi and Microtubule Motor Proteins with Homology to Myocilin

Protein	Accession No.	% Amino Acids Identical or Conserved*	Sequence Showing Homology, †	No. of Proteins with Greater Homology, ‡	Expected Value, §
88-kDa golgi protein	AF204231	23.4	43–367	6	6e−46
Golgi-associated microtubule-binding protein	Y12490	25.4	35–243	8	5e−36
Golgin subfamily a and b	NM 004487, NM 008146.1	21.4	48–229	17	1e−19
										1e−18
364K golgi complex-protein	JC5837	24.4	32–238	23	4e−15
Golgin 67	AF163441	10.7	75–231	28	5e−14
Trans-Golgi p230	U41740	18.6	34–230	52	1e−11
256 kD golgin protein	CAA58041	16.7	45–226	52	1e−11
Golgin-245	AAC51791.1	16.7	45–226	52	7e−11
Golgin-160	P55937	23.4	34–230	72	3e−9
Dynactin (human, mouse, chick, rat), ∥	Q14203, CAA44617, P35458, P28023	20.2	29–224	9	7e−30, 1e−30, 2e−32, 8e−31
Mitotic kinesin-like protein	AAF00594.1	21.6	51–225	10	2e−22

*  % homology was calculated as the number of identical amino acids plus the number of conserved substitutions in the matched sequence divided by the 504 amino acids in full length myocilin.

†  Based on the 504 aa of myocilin, N-to-C terminal.

‡  Only distinct, non-hypothetical, non-golgi related eucaryotic proteins have been included. Myocilin itself was not included, and all olfactomedins and myocins were counted as one each.

§  Expected value is the chance that the amino acid sequence correspondence occurred by chance. Lower numbers (higher negative numbers) imply greater significance. The reported values were obtained after 1 iteration using PSI–BLAST software.

∥  This is a subset of the 11 dynactin isoforms picked up by PSI–BLAST.

The authors thank the Glaucoma Research Foundation, the North Carolina Eye and Tissue Bank, and the Old Dominion Eye Foundation for the donation of eye tissue; Jon Polansky and Albert Harris for critical reading of the manuscript, Susan Whitfield (UNC Biology) for the production of the final images; and Albert Harris, Edward Salmon, Alan Feduccia, and other members of the UNC Biology Department for providing an excellent scientific environment in which to do this work.