Myotubularin Regulates the Function of the Late Endosome through the GRAM Domain-Phosphatidylinositol 3,5-Bisphosphate Interaction*

Myotubularin and related proteins constitute a large and highly conserved family possessing phosphoinositide 3-phosphatase activity, although not all members possess this activity. This family contains a conserved region called the GRAM domain that is found in a vari-ety of proteins associated with membrane-coupled processes and signal transduction. Mutations of myotubularin are found in X-linked myotubular myopathy, a severe muscle disease. Mutations in the GRAM domain are responsible for this condition, suggesting crucial roles for this region. Here, we show that the GRAM domain of myotubularin binds to phosphoinositide with the highest affinity to phosphatidylinositol 3,5-bisphos-phate (PtdIns(3,5)P 2 ). In patients with myotubular my- opathy, mutations in the myotubularin permeabilized with 0.5% Triton X-100, and immunostained with various antibodies. For EGF stimulation, after starvation, cells were incubated with EGF (50 ng/ml, Invitrogen) at 37 °C for the indicated times. For treatment with wortmannin, cells were incubated with wortmannin (100 n M . Sigma) for 15 min before EGF stimulation or fixation. Coverslips were examined using confocal microscopy (Bio-Rad). In Vitro Phosphatase Assay— Phosphatase assays were carried out by FLAG-tagged myotubularin immunoprecipitated from transfected COS-7 cells. Reactions were conducted at 37 °C in a buffer (50 m M ammonium acetate, pH 6.0, 100 m M NaCl, and 2 m M 1,1,1-trichloro- 2,2-bis( p -chlorophenyl)ethane), and phosphate release was quantified by using malachite green assay as described previously (33). Electron Microscopy— Conventional electron microscopy was performed as described previously (34). Briefly, COS-7 cells grown on plastic coverslips were fixed with 2.5% glutaraldehyde in 0.1 M cacody-late buffer, pH 7.4, for 2 h. After washing in the buffer, cells were post-fixed in 1% OsO 4 for 60 min, washed in distilled water, incubated with 50% ethanol for 10 min, and stained with 2% uranyl acetate in 70% ethanol for 2 h. The cells were further dehydrated with a graded series of ethanol and embedded in epoxy resin. Ultra-thin sections were dou- bly stained with uranyl acetate and lead citrate. of

In eukaryotic cells, D3-phosphorylated phosphoinositides such as phosphatidylinositol 3-phosphate (PtdIns3P) 1 play key roles in the vesicular trafficking through direct interaction with phosphoinositide-binding domains such as the PH domain, FYVE finger domain, and PX (Phox) domain found in effector proteins that control vesicular trafficking (1)(2)(3). Previous studies have revealed that PtdIns3P binding is essential for the recruitment/activation of these effector proteins at unique membrane sites (4,5). PtdIns(3,5)P 2 was one of the phosphoinositide species identified recently in both yeast and mammalian cells (6). PtdIns(3,5)P 2 is thought to be involved in osmotic stress responsiveness and essential for the maintenance of vacuole size and homeostasis in yeast (7). Recently, it was reported that PtdIns(3,5)P 2 is necessary for late endosomal trafficking in yeast (8). However, the mechanisms for cellular PtdIns(3,5)P 2 regulation are unknown. Intracellular levels of these phosphoinositide species are strictly regulated by enzymes that dephosphorylate at the D3-position of the inositol ring. Myotubularin and its related proteins (myotubularinrelated proteins; MTMRs) constitute a large and highly conserved subfamily of dual specific phosphatases that were recently revealed to be phosphoinositide 3-phosphatases (9 -11). Among those proteins, myotubularin is encoded by the MTM1 gene, which is mutated in X-linked myotubular myopathy (12), whereas MTMR2 is associated with neurodegenerative disorder Charcot-Marie-tooth disease type 4B (13). Myopathy patients have severe hypotonia at birth, and most of them die from hypoventilation within the first months of life (14). This disease is characterized by the presence of disorganized skeletal muscle fibers that contain centrally located nuclei, resembling myotubes. Recent work has reported that myotubularin is essential for skeletal muscle maintenance but not for myogenesis in mice (15). However, the cellular functions of myotubularin are still obscure. MTMRs also share a conserved region of about 70 amino acids, called the GRAM domain, which has also been found in glucosyltranferases, GTPase-activating proteins of the Rab small GTPases, and other adaptor proteins associated with membrane-coupled processes (16). Some mutations in the GRAM domain of myotubularin cause X-linked myotubular myopathy, suggesting the importance of this domain (17)(18)(19). A very recent study reported that the GRAM domain of MTMR2 binds to PtdIns(3,4,5)P 3 , PtdIns(3,5)P 2 , and PtdIns5P An ELISA phosphoinositide binding assay with GST fusion proteins was performed. a, the schematic structure of myotubularin. b, each well of a microtiter plate was coated with 2.0 g of total phospholipid containing 10% of the indicated phospholipid and then overlaid with GST, GST fusions of phospholipase C-␦1 PH, Akt/PKB PH, and myotubularin GRAM. Amounts of bound proteins were calculated as described under "Experimental Procedures." Results from three independent experiments are represented as mean values Ϯ S.E. (error bars). PtdIns4P, phosphatidylinositol 4-phosphate; PtdIns5P, phosphatidylinositol 5-phosphate. c, membrane overlay assay. Binding of either GST-myotubularin or GST to immobilized phosphoinositides was assessed by overlay assay using PIP arrays (Echelon Bioscience). d, the ELISA assay was carried out using GST-GRAM domains of MTMR2 and VRP. e, liposome binding assay with the GRAM domain. Liposomes(PC/PE ϭ 4/1; total 100 g) containing various amounts of PtdIns(3,5)P 2 were mixed with GST-GRAM. Proteins in the supernatant (s) and precipitate (p) were visualized by SDS-PAGE and stained with Coomassie Brilliant Blue (upper figure). Binding of GST-GRAM domains of myotubularin, MTMR2, and VRP to liposomes containing 5% of PtdIns or PtdIns(3,5)P 2 (lower figure). Data are representatives of three independent experiments. f, comparison of wild type and mutants of myotubularin GRAM in phosphoinositide binding.  (20), suggesting the importance of its binding for membrane targeting of MTMR2. However, the role of phosphoinositide binding to the myotubularin GRAM domain in membrane trafficking and myotubular myopathy has not been elucidated.
Here we show that the GRAM domain of myotubularin strongly binds to PtdIns(3,5)P 2 . This binding is abolished by point mutations found in the GRAM domain of myotubularin in myopathy patients, indicating that interaction with PtdIns (3,5)P 2 is essential for full function of this phosphatase. We further demonstrate that overexpression of myotubularin inhibits epidermal growth factor receptor (EGFR) trafficking from the late endosome to the lysosome and induces large endosomal vacuoles. Significantly, GRAM domain mutants, as well as phosphatase inactive mutants seen in myopathy patients, do not induce these effects. Thus, our data indicate that through interaction with PtdIns(3,5)P 2 , myotubularin functions in late endosomal trafficking and vacuolar formation.
ELISA Lipid Binding Assay-All synthetic phosphoinositides with C16 fatty acids were purchased from Cell Signals Inc., whereas other phospholipids (PA, PC, PE, PS and PI) were obtained from Sigma. Lipid vesicles (PE/PC ϭ 1/1; total 2 g) containing the indicated ratio (weight percentage) of phospholipids resuspended in ethanol were coated on a 96-well microtiter plate (lmmulon 2;Dynatech) and dried at room temperature. The wells were then blocked with phosphate-buffered saline containing 5% bovine serum albumin before incubation with the respective glutathione S-transferase (GST) fusion proteins (1.0 g/ml) for 45 min. After washing with phosphate-buffered saline containing 0.05% Tween 20, bound protein was detected by reaction of the substrate, orthophenyl enediamine dihydrochloride, with glutathione conjugated to peroxidase (sigma). The colorimetric reaction was measured at 492 nm in an ELISA plate reader (Bio-Rad).
Lipid Overlay Assay-PIP arrays were purchased from Echelon Bioscience Inc. In brief, arrays were incubated overnight at 4°C with GST fusion protein (0.5 g/ml) in TBST (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.1% Tween 20) with fatty acid-free bovine serum albumin (Sigma). Membranes were washed and incubated with monoclonal anti-GST antibody (Sigma). After washing, membranes were incubated with horseradish peroxidase-conjugated anti-mouse antibody and visualized using an ECL kit (Amersham Biosciences) Measurement of PtdIns(3,5)P 2 Levels in EGF-stimulated COS-7 Cells-COS-7 cells were cultured in serum-deprived medium for 24 h and then in orthophosphate-free Dulbecco's modified Eagle's medium for 15 min. After radiolabeling with 5 mCi of [ 32 P]orthophosphate for 2 h, cells were stimulated with 100 ng/ml human recombinant EGF (Invitrogen) for 0, 20, 40, 60, and 120 min. Cells were washed three times with ice-cold phosphate-buffered saline, and phosphoinositides were extracted with the addition of 1.5 ml of methanol followed by the same volume of chloroform:methanol (1:2, v/v). After brief sonication, the same volume of chloroform and then 1 N HCl were added. The organic phase was pooled and dried under nitrogen gas. The phosphoinositides were deacylated with methylamine reagent (10.7% methylamine, 45.7% methanol, 11.4% butanol) at 53°C for 1 h. The samples were dried in SpeedVac and resuspended in 250 l of sterile water. 250 l of butanol reagent (butanol/ethyl ether/formic acid ϭ 20/4/1, v/v/v) was added to each sample. The samples were vortexed and centrifuged at 12,000 rpm for 5 min. The aqueous phase was dried in SpeedVac, dissolved in 10 mM (NH 4 ) 2 PO 4 , pH 3.8, and separated by high performance liquid chromatography on a Partisphere strong anion exchange column (Whatman, Clifton, NJ). The column was developed with a gradient of 1 M (NH 4 ) 2 PO 4 , pH 3.8. The gradient started at 1% for 5 min, 1-20% over 44 min, 20 -50% over 3.8 min, and remained at 50% for 8 min, and the flow rate was 1.0 ml/min. The radioactivity was quantified with a Flow Scintillation Analyzer (PerkinElmer Life Sciences).
Transfections-COS-7 cells were transfected by either the calcium phosphate method or Lipofectin using LipofectAMINE (Invitrogen) and examined 24 h after transfection. Transfected cells on coverslips were fixed in 3.7% formaldehyde, permeabilized with 0.5% Triton X-100, and immunostained with various antibodies. For EGF stimulation, after starvation, cells were incubated with EGF (50 ng/ml, Invitrogen) at 37°C for the indicated times. For treatment with wortmannin, cells were incubated with wortmannin (100 nM. Sigma) for 15 min before EGF stimulation or fixation. Coverslips were examined using confocal microscopy (Bio-Rad).
Electron Microscopy-Conventional electron microscopy was performed as described previously (34). Briefly, COS-7 cells grown on plastic coverslips were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.4, for 2 h. After washing in the buffer, cells were post-fixed in 1% OsO 4 for 60 min, washed in distilled water, incubated with 50% ethanol for 10 min, and stained with 2% uranyl acetate in 70% ethanol for 2 h. The cells were further dehydrated with a graded series of ethanol and embedded in epoxy resin. Ultra-thin sections were doubly stained with uranyl acetate and lead citrate.

RESULTS AND DISCUSSION
The GRAM Domain Binds to PtdIns(3,5)P 2 -We examined the possibility that the GRAM domain of myotubularin and related proteins could bind to phosphoinositides. We used an ELISA assay where phospholipids composed of phosphatidylethanolamine (PE), phosphatidylcholine (PC), and each phosphoinositide were coated on ELISA plates and overlaid with a GST fusion protein of interest. We first confirmed that GST alone showed no affinity to any of the phospholipids in this assay (Fig. 1a). Specific binding was observed between the PH domain of phospholipase C-␦1 and PtdIns(4,5)P 2 (21,22), as well as between the PH domain of Akt/PKB with PtdIns(3,4)P 2 and PtdIns(3,4,5)P 3 (23,24), indicating the validity of this assay (Fig. 1a). The GRAM domain of myotubularin was found to bind to the phospholipid containing PtdIns(3,5)P 2 with the highest affinity (Fig. 1a). PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 were also favored to some degree. Furthermore, PH domains did not bind PtdIns(3,5)P 2 at all in this assay, confirming that the binding detected for the GRAM domain was not an artifact. Interaction between the GRAM domain and PtdIns(3,5)P 2 was also examined by overlay assay where each phosphoinositide were spotted on the membrane without any basal lipid as in the ELISA assay. Although the GRAM domain still bound strongly to PtdIns(3,5)P 2 , binding to other phosphoinositides was also observed (Fig. 1b). In this assay, GST alone did not bind to any phosphoinositide (Fig. 1b). We also carried out the same ELISA assay using GRAM domains from MTMR2 and VRP and found they also bound to PtdIns(3,5)P 2 most strongly (Fig. 1c). Moreover, to confirm this interaction in a membranous environment, we performed a liposome binding assay. Although the GRAM domain of myotubularin, MTMR2, and VRP did not co-sediment liposomes containing PtdIns, co-sedimentation between liposomes containing PtdIns(3,5)P 2 and the GRAM domain was observed (Fig. 1d). These findings indicate that the evolutionally conserved GRAM domain is designed for binding to PtdIns(3,5)P 2 .
Among GRAM-containing proteins, myotubularin is the product of the MTM1 gene, the gene responsible for X-linked myotubular myopathy. Four cases of missense mutation in the GRAM domain (V49F, R69C, L70F, L87P) have been reported to be associated with the disease (18,19). We generated these point mutants and compared their phosphoinositide binding abilities. As shown in Fig. 1e, the binding ability was significantly reduced for the V49F, a mutation that is associated with a more severe phenotype. Mutations of R69C, L70P, or L87P, FIG. 2. GRAM-PtdIns(3,5)P 2 binding is essential for translocation of myotubularin to the late endosomal compartment after EGF stimulation. a, COS-7 cells were transfected with GFP-myotubularin. After starvation, cells were incubated with EGF (50 ng/ml) for the indicated time periods. After the cells were fixed, the intracellular localization of GFP-myotubularin and EGFR was visualized by immunofluorescence. Yellow indicates co-localization of both proteins. b, increased PtdIns(3,5)P 2 content in response to EGF stimulation. COS-7 cells were labeled with [ 32 P]orthophosphate and stimulated with EGF. Lipid labeling values were corrected to a reference value of 2.0 ϫ 10 7 cpm of total 32 P-labeled lipid in each experiment. Data are representatives of three independent experiments. c, overexpression of PIKfyve recruits myotubularin to the perinuclear area, where they co-localize. Yellow indicates co-localization of both proteins. d, the localization of GFP-myotubularin mutants at 40 min after EGF stimulation was compared with that of EGFR. Both mutants showed cytosolic patterns but no co-localization with EGFR. e, quantitative representation of EGF dependent recruitment of myotubularin to the late endosomal compartments. After treatment with EGF for 40 min, cells that changed the localization of the indicated proteins to the late endosomal compartments were counted and represented as percentages relative to all cells observed (total of Ͼ50 cells) from three independent experiments. which are associated with a milder phenotype, reduced the binding to a lesser extent. These findings suggest that phosphoinositide binding ability of the GRAM domain is crucial for functions of myotubularin in vivo.
The Essential Role of the GRAM Domain Is in Translocation to the Late Endosome-Because our results redefined myotubularin as both a phosphoinositide 3-phosphatase and a PtdIns(3,5)P 2 -interacting protein, and because previous studies have indicated essential roles for PtdIns3P and PtdIns(3,5)P 2 in endosomal trafficking (3,25), we next exam- FIG. 3. Overexpression of myotubularin impairs EGFR trafficking from the late endosome to the lysosome in a GRAM domaindependent manner. a, COS-7 cells were transfected with pEFBOS-myotubularin (wild type (WT) and mutants). Cells were stimulated with EGF for 180 min and fixed. They were stained with anti-Myc antibody (green) and anti-EGFR antibody (red). Note that cells that overexpress wild type myotubularin show red spots corresponding to internalized EGFR even after they were degraded in non-expressing cells (arrows). b, the quantitative representation of a. The cells that still retained the undegraded internalized EGFR at 180 min after EGF stimulation were counted and indicated as a percentage in the whole cells. c, similar data were obtained from Western blot analysis. ined the possibility that myotubularin is involved in intracellular vesicular trafficking. The localization of the GFP-tagged form of myotubularin was monitored after EGF stimulation in COS-7 cells. In non-stimulated cells, EGFR retained at the plasma membrane, whereas distribution of GFP-myotubularin is mainly cytosolic (Fig. 2a). 20 min after EGF stimulation, EGFR internalization was observed by the significant co-localization with the EEA1-positive early endosome (not shown). GFP-myotubularin, however, was still distributed throughout the cytosol at this time point (Fig. 2a), indicating that myotubularin is not involved in the EGFR trafficking from the plasma membrane to the early endosome. However, after 40 min of EGF stimulation, myotubularin was observed to translocate together with the internalized EGFR (Fig. 2a) to the late endosomal compartment shown by its partial colocalization with LAMP1, a marker protein for the late endosome/lysosome (not shown). We also confirmed that cellular PtdIns(3,5)P 2 levels were increased in response to EGF stimulation and peaked at 40 min after EGF stimulation (Fig. 2b), indicating the possibility that translocation of myotubularin is dependent on interactions between the GRAM domain and PtdIns(3,5)P 2 . Moreover, to clarify whether intracellular localization of myotubularin is affected by PtdIns(3,5)P 2 in vivo, we overexpressed PIKfyve, a phosphoinositide 5-kinase shown to produce PtdIns (3,5)P 2 both in vivo and in vitro (26). As shown in Fig. 2c, overexpression of PIKfyve recruited cytosolic myotubularin to the perinuclear area where they co-localized, suggesting that the localization of myotubularin is determined by PtdIns (3,5)P 2 . To evaluate the importance of interaction between the GRAM domain and phosphoinositides, we further examined the effect of mutations in myotubularin on its translocation. Importantly, myotubularin ⌬GRAM in which the whole GRAM domain is deleted did not translocate to the late endosomal compartment after EGF treatment (Fig. 2d). The same result was observed with myotubularin V49F, in which the mutated GRAM domain was unable to bind PtdIns(3,5)P 2 (Fig. 2d). This indicates that the lack of PtdIns(3,5)P 2 binding ability by a point mutation is comparable with the loss of the entire GRAM domain. Moreover, cell pretreatment with the PI3-kinase inhibitor wortmannin prevented the recruitment of myotubularin to this compartment subsequent to EGF stimulation (Fig.  2e), supporting the idea that the myotubularin translocation is dependent on a D3-phosphoinositide.
Myotubularin Negatively Regulates EGFR Degradation-The translocation of myotubularin to the EGFR-positive late endosomal compartment suggested its roles in the degradation pathway of EGFR. To assess the effect of myotubularin on EGFR trafficking, we overexpressed myotubularin in COS-7 cells with the pEF-Bos vector that carries a powerful promoter derived from the upstream region of the gene for the transcription factor EF-1␣ (27). In myotubularin-overexpressing cells, EGFR displayed significant co-localization with the EEA1-positive early endosome 20 min after EGF stimulation (data not shown), indicating that myotubularin does not affect EGFR trafficking from the plasma membrane to the early endosome. At 180 min after EGF stimulation, EGFR was degraded in the lysosome in non-expressing cells as the red spots disappeared (Fig. 3a, arrows). However, EGFR still remained undegraded in cells that overexpressed myotubularin (Fig. 3, a and b), suggesting a negative regulation of EGFR degradation. Importantly, this effect requires the 3-phosphatase activity of myotubularin since the phosphatase-negative mutant G378R (10) did not affect the EGFR degradation. Furthermore, the GRAMdomain mutant (V49F) showed normal degradation seen in surrounding non-expressing cells (Fig. 3, a and b). We also confirmed that myotubularin negatively regulates EGFR deg-radation using Western blot analysis. As shown in Fig. 3c, degradation of EGFR was suppressed in myotubularin WT -expressing cells as compared with V49F-or G378R-expressing cells. These data strongly indicate that myotubularin 3-phosphatase functions as a negative regulator of EGFR trafficking from the late endosome to the lysosome for degradation and that phosphoinositide binding through the GRAM domain is required for the function of myotubularin in vivo.
Myotubularin Regulates Vacuolar Formation-In non-overexpressing cells, the EGF-EGFR complex seemed to have been sorted into the lumen of the late endosomes, forming the multivesicular bodies (28). We found that myotubularin overexpression induced abnormal, large vacuolar vesicles in about 40% of transfected cells and that EGFRs were retained at the limiting membranes on large vacuoles (Fig. 4a). Electron microscopy analysis showed that these large vacuoles include comparatively few internal vesicles as compared with vacuoles of control cells. (Fig. 4b). These observations suggest that induction of enlarged vacuoles by myotubularin might be due to inhibition of vacuolar membrane recycling or a defect of invagination on limiting membranes. More importantly, we found that overexpression of myotubularin V49F did not cause strong formation of the large vacuole. (Fig. 4, a and c). The vacuolation phenotype was not so marked in cells expressing the phosphatase-negative mutant G378R (Fig. 4, a and c), suggesting the importance of dephosphorylation of phosphoinositide substrates for the effect. We found that myotubularin could dephosphorylate PtdIns(3,5)P 2 (apparent K m value of PtdIns(3,5)P 2 ϭ 17 M) as well as PtdIns3P (K m ϭ 39 M) (not shown). Dephosphorylation of PtdIns(3,5)P 2 by myotubularin was described recently (29). These findings suggest that myotubularin translocates to the PtdIns(3,5)P 2 -rich membrane via its GRAM domain and effectively dephosphorylates PtdIns(3,5)P 2 on the limiting membrane of the late endosomal compartments to negatively regulate the vacuolar formation.
Our findings that the GRAM domain of myotubularin bind PtdIns(3,5)P 2 with high affinity and specificity are consistent with the notion that the GRAM domain of myotubularin 3-phosphatase family functions to bind their substrate PtdIns(3,5)P 2 , facilitating the dephosphorylation of the phosphoinositide. Moreover, point mutations in the GRAM domain, reported in severe cases, eliminate this binding, indicating that GRAM-PtdIns(3,5)P 2 interaction is crucial for functions of myotubularin in vivo. In yeast, the formation of PtdIns(3,5)P 2 from PtdIns3P is catalyzed by Fab1p, a PtdIns3P-5-kinase necessary for maintenance of normal vacuolar morphology and functions (7). Fab1P as well as PIKfyve, a mammalian homologue of Fab1p, are reported to localize to late endosome/multivesicular bodies (30), suggesting that localization of PtdIns(3,5)P 2 may also be to this compartment. According to this hypothesis, myotubularin probably localizes to the late endosome via interaction with PtdIns(3,5)P 2 . In addition, recent studies have revealed that PIKfyve suppressed the vacuolar enlargement in mammalian cells, indicating that it functions to maintain the proper size of vacuolar organelles (31). Thus, our data strongly suggest that myotubularin physiologically functions in the late endosomal trafficking and vacuolar morphology by controlling PtdIns(3,5)P 2 and show the indispensable roles of the GRAM domain in its cellular functions. Recently, Ent3p, a yeast epsin N-terminal homology (ENTH) domain-containing protein, has been found to bind PtdIns(3,5)P 2 through its ENTH domain and function in the late endosome (32), indicating that this lipid plays important roles in sorting of membrane proteins at late endosomal compartment throughout yeast to mammals. However, the precise mechanism of how PtdIns(3,5)P 2 regulates sorting of membrane proteins remains to be solved in future.