Production of phosphatidylinositol 5-phosphate by the phosphoinositide 3-phosphatase myotubularin in mammalian cells.

MTM1, the gene encoding myotubularin (MTM1), is mutated in the X-linked myotubular myopathy (XLMTM), a severe genetic muscular disorder. MTM1 is a phosphoinositide phosphatase hydrolyzing phosphatidylinositol 3-phosphate (PtdIns(3)P) in yeast and in vitro. Because this lipid is implicated in the regulation of vesicular trafficking, we used established cell lines from XLMTM patients to evaluate whether the lack of endogenous MTM1 expression could affect PtdIns(3)P labeling patterns. Our results showed that the vesicular trafficking related to early endosomes was not significantly affected in the XLMTM cell lines compared with control cells. However, in addition to PtdIns(3)P, we found that MTM1 can hydrolyze phosphatidylinositol 3,5-bisphosphate both in vitro and in mammalian cells. Using a mass assay, we demonstrated that the product generated is phosphatidylinositol 5-phosphate (PtdIns(5)P), a recently discovered phosphoinositide, the function of which is still unknown. In L6 myotubes overexpressing MTM1, hyperosmotic shock induced an increase in the mass level of PtdIns(5)P that was reduced by 50% upon overexpression of the MTM1 inactive mutant D278A. These data demonstrate for the first time a role for MTM1 in the production of PtdIns(5)P in mammalian cells, suggesting that the lack of transformation of phosphatidylinositol 3,5-bisphosphate into PtdIns(5)P might be an important component in the etiology of myotubular myopathy.

MTM1 1 is a 66-kDa ubiquitous protein that is found mutated in the X-linked genetic disorder myotubular myopathy (XLMTM). The disease, which affects 1 of 50,000 newborn males, is characterized by muscle weakness and hypotonia at birth. In the most severe cases, it leads to the death in the first weeks of life because of respiratory failure. More than 198 different mutations have been found in the MTM1 gene, mainly leading to the absence or the expression of a truncated version of the protein (1). The gene encoding MTM1 was cloned in 1996 (2), and its sequence shows the presence of the HCSDGWDRT motif, which fits the conserved consensus sequence CX 5 R of the dual specificity phosphatase/phosphotyrosine phosphatase (DSP/PTP) family. However, studies on MTM1 activity showed that its preferred substrate is not a tyrosine-phosphorylated protein but the phosphoinositide PtdIns(3)P (3,4), thereby identifying MTM1 as a lipid phosphatase. Thus, MTM1 is a 3-phosphatase but exhibits a different substrate specificity compared with the tumor suppressor PTEN that hydrolyzes preferentially PtdIns(3,4,5)P 3 (5). MTM1 could therefore act as a regulator of the highly active phosphoinositide (PI) metabolism in cells (6), where interconversions of the eight PIs described so far is controlled by a set of specific kinases and phosphatases (7). PIs are minor phospholipids of the cell membrane; however, they play a major role of second messengers in diverse cellular functions such as proliferation, apoptosis, cytoskeletal remodeling, and vesicular trafficking (7,8). The way PIs act is via targeting of proteins to specific membrane locations by binding to specific protein domains such as the pleckstrin homology, phox, Fab1p/ YOTB/Vac1p/EEA1 (FYVE), epsin N-terminal homology, and band 4.1/ezrin/radixin/moesin domains.
MTM1 is highly conserved during evolution and is the founding member of the myotubularin-related proteins (MTMRs) family, which today counts 14 members. In addition to MTM1, other members of the family are linked to genetic diseases. MTMR1 is a target for aberrant splicing in the congenital myotonic dystrophy, cDM1 (9), and mutations in MTMR2 have been associated to the Charcot-Marie-Tooth disease type 4B1 (CMT4B1), a demyelinating neuropathy (10). Two members of the phosphatase-dead MTMR subgroup have also been implicated in defects or disease, respectively. MTMR5/SBF1 was shown to play a role in spermatogenesis in mice (11), and MTMR13/SBF2 was recently found mutated in the CMT4B2, another type of Charcot-Marie-Tooth disease (12,13).
Thus, although the cellular function of the MTM family members appears essential, it still remains elusive. Recent studies (14) on the knock-out MTM1 mice proposed a role for the phosphatase in the maintenance of muscle fiber state rather than in myogenesis to explain the physiopathological mechanism of the XLMTM. As a 3-phosphatase hydrolyzing PtdIns(3)P, MTM1 was proposed to play a regulatory role in vesicular trafficking, and in some cases, overexpression of the phosphatase indeed leads to the disappearance of PtdIns(3)P endosomal punctate labeling (15). However, more studies are necessary to definitely link MTM1 to a physiological role in vesicular trafficking.
To address this point, we investigated whether endogenous MTM1 could affect the pool of PtdIns(3)P in early endosomes by studying established cell lines from patients suffering from myotubular myopathy. Surprisingly, our results show that in the XLMTM cell lines, EEA1, a marker for early endosome, and PtdIns(3)P labeling patterns are not significantly affected compared with normal cell lines. Therefore, we then further investigated MTM1 cellular function and specificity toward other PIs. It has been shown recently that two members of the MTM family, namely MTMR3 overexpressed in the yeast Saccharomyces cerevisiae and the mouse MTMR2 in an in vitro assay, can use PtdIns(3,5)P 2 as a substrate in addition to PtdIns(3)P, leading to the production of PtdIns(5)P (16, 17), a newly described PI (18). Here we show that both MTM1 and MTMR1 are also able to hydrolyze PtdIns(3,5)P 2 in vitro, in addition to PtdIns(3)P. Moreover, we demonstrate that MTM1 is a source of PtdIns(5)P in vitro and in vivo, pointing to a role for the phosphatase in the production of this recently described PI in mammalian cells. Altogether, our data suggest that MTM1 could affect a minor pool of PtdIns(3)P and that the production of PtdIns(5)P, through its action as a PtdIns(3,5)P 2 phosphatase, might be essential to its cellular function.

EXPERIMENTAL PROCEDURES
Cloning, Antibodies, and Proteins Expressions-FLAG-MTM1 construct was obtained by PCR amplification of human MTM1 cDNA and subcloned into a pcDNA3-modified vector with an N-terminal FLAG tag. All the other constructs were already described (9,19). The 1G6 monoclonal antibody was raised against an N-terminal peptide (amino acids 13-32) of human MTM1 (20), and the EEA1 mouse antibody (clone 14) was purchased from BD Biosciences. The cDNA for the GST-2XFYVE probe interacting with PtdIns(3)P was a gift of Dr. H. Stenmark, Institute for Cancer Research, Oslo, Norway (21). The recombinant GST-2XFYVE protein was expressed in DH5␣ Escherichia coli strain, bound to a glutathione-agarose affinity column and eluted in 50 mM Tris-HCl, pH 8.0, with 25 mM reduced glutathione. The GST-2XFYVE fusion protein was then biotinylated using the biotin tag micro-biotinylation kit (Sigma) following the manufacturer's conditions, and the unreacted biotin was removed with Centricon YM10 (Millipore, Bedford, MA). PGEX-PtdIns(5)P 4-kinase type II␣ was provided by Dr. N. Divecha (The Netherlands Cancer Institute, Amsterdam, The Netherlands). The recombinant GST-PtdIns(5)P 4-kinase type II␣ was expressed in the BL21-CodonPlus E. coli strain (Stratagene) and purified on glutathione-agarose beads (Amersham Biosciences). Recombinant adenoviral genomes carrying the wild type human hMTM1 or the substrate trap hMTM1-D278A cDNAs were generated by homologous recombination as described in Chaussade et al. (22).
Cell Culture, Transfections, and Viral Infections-COS-7, HEK293, and L6 myoblasts cells were maintained at 37°C with 5% (v/v) CO 2 in Dulbecco's modified Eagle's medium (DMEM) containing 10% (v/v) fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin (Invitrogen). L6 differentiation to myotubes was induced by culturing cells for over 5 days in DMEM containing 2% (v/v) fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin before adenovirus infection. Adenoviruses were stored in PBS, 10% (v/v) glycerol at Ϫ20°C, and viral titer of stocks was Ͼ10 8 plaque-forming units/ml. Infections of the L6 myotubes were performed at a multiplicity of infection of 10. After 16 -20 h of incubation in the presence of viral particles, the medium was changed, and cells were cultured for 24 -72 h. Hyperosmotic shock was performed by incubating the cells in medium containing 0.9 M NaCl for 10 min at 37°C. For transfection experiments, COS-7 or HEK293 cells were seeded in 6-well plates and were transfected using Effectene (Qiagen) with 0.3 g of plasmid DNA according to the manufacturer's instructions. 24 h after transfection, cells were washed once in PBS and used for immunoprecipitation experiments. Jurkat cells were cultivated in RPMI plus 10% heat-inactivated fetal bovine serum and transferred into 12-well plates after nucleofection of 1 g of plasmid DNA/10 6 cells using AMAXA nucleofector kit V (Amaxa Biosystems). 24 h after nucleofection, cells were washed once in PBS and used for PtdIns(5)P mass measurement assay. Established cell lines from XLMTM patients were already described elsewhere (23)  In Vitro Phosphatase Activity-Cells were lysed in buffer A (20 mM Tris, pH 7.4, 150 mM NaCl, 4 mM EDTA, 1% Triton X-100 plus protease inhibitor mixture). Lysates were incubated with anti-FLAG M2 antibody (Sigma) or 1G6 MTM1-specific antibody plus protein A-coated agarose beads (Amersham Biosciences) for 3 h at 4°C. The immunocomplex was washed several times in buffer A and then in 50 mM ammonium acetate, pH 6.0, and assayed for phosphatase activity in 50 mM ammonium acetate plus 2 mM DTT, pH 6.0, with either di-C8-NBD or di-C16-NBD fluorescent phosphoinositides (Echelon Research Laboratories) according to Taylor and Dixon (24). Lipids were extracted according to Blight and Dyer (25) with addition of HCl to a final concentration of 0.4 N, separated on Silica Gel G60 TLC, and visualized under UV light. Phosphatase activity was also quantified by release of free phosphate using the Biomol Green (Biomol, Plymouth Meeting, PA) malachite green binding assay adapted for 96 wells according to the manufacturer's protocol.
PtdIns(5)P Mass Measurement-Total lipids were extracted according to Blight and Dyer (25) in the presence of HCl to a final concentration of 0.4 N and separated on Silica gel G60 TLC plate in the CH 3 Cl/ CH 3 OH/NH 3 (9:7:2) solvent system. Monophosphorylated PIs were scraped, eluted from the silica, and assayed for PtdIns(4,5)P 2 formation in vitro using the recombinant-specific PtdIns(5)P 4-kinase type II␣ and [␥-32 P]ATP as described by Morris et al. (26). Generated [␥-32 P]PtdIns(4,5)P 2 was separated by TLC and detected by a Phospho-rImager system, and the associated radioactivity was counted with a ␤ scintillation counter. Quantification of the mass amount of generated PtdIns(5)P was performed as published previously (26,27).
Phosphoinositides Analysis-L6 differentiated myotubes were incubated overnight with 1 Ci of [ 32 P]orthophosphate in phosphate-free DMEM. Lipids were extracted according to Blight and Dyer (25) with addition of 0.4 N HCl (final concentration) and resolved on Silica G60 TLC in CH 3 Cl/CH 3 COCH 3 /CH 3 OH/CH 3 COOH/H 2 O (80:30:26:24:14, v/v). The relevant lipids were scraped off, deacylated, and analyzed by HPLC on a Partisphere 5 SAX column (Whatman) and quantified on a continuous flow in-line scintillation detector (Beckman Instruments) as described previously (7,27). The elution positions were determined by using appropriate standards.
Immunofluorescence Microscopy-Fibroblasts and myoblasts were grown on glass coverslips in DMEM ϩ 10% fetal calf serum and minimum Eagle's medium ϩ 25% 199 medium ϩ 20% fetal calf serum, respectively. The protocol used for biotinylated GST-2XFYVE as a probe for PtdIns(3)P is a modification of the original protocol described by Gillooly et al. (21) that was described in Kim et al. (15). Briefly, after washing in PBS, cells were fixed with 4% paraformaldehyde overnight at 4°C, which permeabilizes most of the cells. All incubations and washes were then done in PBS at room temperature. The cells were washed and the free aldehyde groups were quenched for 15 min with 50 mM NH 4 Cl, and fixed cells were blocked in PBS ϩ 5% bovine serum albumin for an hour. Cover glasses were then incubated with mouse anti-EEA1 antibody (1:100) and biotinylated GST-2XFYVE recombinant protein (40 g/ml) for 1 h. After washing, coverslips were incubated for 1 h with Alexa fluor 488 goat anti-mouse and Alexa fluor 594 streptavidin (Molecular Probes). The slides were washed three times and mounted in anti-fading solution (5% propyl galate and 80% glycerol in PBS), and the signals were analyzed by fluorescence microscopy.

Localization of PtdIns(3)P and EEA1 in Cell
Lines from XLMTM Patients-We and others have shown previously that MTM1 hydrolyzes PtdIns(3)P in vitro and in yeast (3,4) and that its activity is specific for the 3-O-phosphate position of monophosphoinositides in vitro (19). PtdIns(3)P was implicated in intravesicular trafficking regulation in yeast and in mammalian cells (21). Therefore, we addressed the physiological role of MTM1 in the regulation of the PtdIns(3)P levels by comparing its localization in normal and XLMTM patients cell lines. In the control cell line, there is a good colocalization of PtdIns(3)P detected by the biotinylated 2X-FYVE probe, with one of its target, the early endosome marker EEA1, labeled by a specific antibody (Fig. 1,  A-C). A similar PtdIns(3)P labeling pattern was observed in the H31 fibroblasts cell line deleted for the entire MTM1 gene, as well as in fibroblasts cell lines G95-320 and CQ85 established from two patients affected during infancy by a myopathy resembling XLMTM but with no mutations in the MTM1 gene and by a centronuclear myopathy, respectively. Centronuclear myopathies are histopathologically related myopathies with a later onset and an autosomal inheritance. In the 3 cell lines expressing normal levels of MTM1 (control, G95-320, and CQ85) and in the MTM1-deleted cell line H31 (23), there is no modification of PtdIns(3)P labeling pattern (Fig. 1, A, D, G, and J), which still largely colocalizes with the EEA1 marker ( Fig. 1, C, F, I, and L). We also monitored PtdIns(3)P pools in established myoblasts cell lines from XLMTM patients (Fig. 2), a cell system more relevant to the severe muscular phenotype usually observed in XLMTM patients. In those cell lines, a decrease in MTM1 level compared with normal myoblasts (23), and two of them were derived from patients presenting severe phenotypes (EG84 and DM92), has been observed by immunoprecipitation using the 1G6 antibody. Our results show the colocalization of PtdIns(3)P with the EEA1 marker in the 3 myoblast cell lines from XLMTM patients (Fig. 2, F, I, and L) and the control myoblasts (Fig. 2C). As a control, Fig.  2, M-O, shows the delocalization of PtdIns(3)P by wortmannin treatment in normal myoblasts. Altogether, these data suggest that the lack of expression of endogenous MTM1 in fibroblasts or myoblasts does not significantly affect the pool of PtdIns(3)P on EEA1-decorated membranes.
MTM1 Is a PtdIns(3)P and PtdIns(3,5)P 2 3-Phosphatase in Vitro-The observations from XLMTM patient cell lines suggested that endogenous MTM1 may act on a discrete pool of PtdIns(3)P. This seems reasonable because the decrease in PtdIns(3)P levels upon MTM1 overexpression never exceeded more than 10 -15% 2 in the different cell lines that we have tested (COS-7, HEK 293, L6, C2C12, and Jurkat). This value is in agreement with the results reported by Taylor et al. (4) and suggests other possible functions for MTM1. As shown recently that MTMR2 and the mouse homologue of MTMR3 can also hydrolyze PtdIns(3,5)P 2 , we addressed this possibility for MTM1. We thus proceeded with the in vitro characterization of MTM1 phosphatase lipid substrate specificity using short side acyl chains (C6), fluorescent PIs, and immunoprecipitated MTM1 (or its close homologue MTMR1) overexpressed in mammalian cells. In addition, human MTMR2 and MTMR3 were used as controls. Fig. 3A shows that all four MTM family members can use PtdIns(3,5)P 2 as a substrate, in addition to PtdIns(3)P. None of the 5 other PIs tested are substrates for the phosphatases. MTM1 and MTMR1 seem as potent as MTMR2 and MTMR3 to hydrolyze PtdIns(3,5)P 2 and PtdIns(3)P, leading to the complete hydrolysis of the substrates after 30 min of incubation at 30°C (Fig. 3A).
In order to show that the 3-phosphatase activity of MTM1 measured in vitro is specific, we used a phosphatase-inactive MTM1 mutant, where the cysteine residue in the phosphatase catalytic center has been mutated into serine (C375S). Fig. 3B shows that the hydrolysis of PtdIns(3)P and PtdIns(3,5)P 2 is specific for MTM1 intrinsic phosphatase activity. We next measured the hydrolysis efficiency in function of fluorescent lipid concentration in the in vitro phosphatase assay using MTM1 and MTMR2. We show that the phosphatase reaction reached saturation at 1 g of fluorescent PtdIns(3)P or PtdIns(3,5)P 2 in 8 min at 30°C for both MTM1 and MTMR2 (Fig. 3C).
Because naturally occurring PIs in cell membranes are long acyl chains species, the use of C6 side chains PIs in our assays 2 B. Payrastre and H. Tronchère, unpublished observations. might be of concern. To clarify this point, we performed in vitro experiments using C16-PIs. The results presented in Fig. 4A show that the specificity of MTM1 is kept independently of the acyl chain length. Moreover, by using a malachite green assay to quantify the amount of free phosphate released during MTM1 assay, we found that both C16-PtdIns(3)P and C16-PtdIns(3,5)P 2 were hydrolyzed in a comparable manner (Fig. 4B).

MTM1 Produces PtdIns(5)P in Vitro-
We next checked whether dephosphorylation of PtdIns(3,5)P 2 by MTM1 and family members generates PtdIns(5)P. We took advantage of the use of the PtdIns(5)P 4-kinase type II␣, which phosphorylates PtdIns(5)P on position 4 of the inositol ring, to produce PtdIns(4,5)P 2 (26). First, the specific substrate selectivity of PtdIns(5)P 4-kinase type II␣ toward PtdIns(5)P was verified by using other possible substrates like PtdIns(3)P and PtdIns(4)P.  1 and 4), MTM1 (lanes 2 and 5), or the C375S MTM1 mutant (lanes 3 and 6) was expressed in COS-7 cells, immunoprecipitated, and tested for phosphatase activity by using C6-NBD-PtdIns(3)P (lanes 1-3) or C6-NBD-PtdIns(3,5)P 2 (lanes 4 -6). Lipids were separated on TLC, and fluorescence was registered under a UV table. C, immunoprecipitates of MTM1 or MTMR2 were tested in vitro for their phosphatase activity with 0.5, 1, or 1.5 g of C6-NBD-PtdIns(3)P (‚) or C6-NBD-PtdIns(3,5)P 2 (f). Percent of hydrolysis was reported. Fig. 5A demonstrates the strong specificity of the kinase for PtdIns(5)P. The PtdInsP 2 formed in this reaction was identified as being exclusively PtdIns(4,5)P 2 by HPLC analysis (data not shown). MTM1, MTMR2, and MTMR3 immunoprecipitated from overexpressed COS-7 cells where first tested for their 3-phosphatase activity by using C16-NBD-PtdIns(3,5)P 2 (Fig.  5B). The PtdInsPs generated were scraped off, recovered from silica, and used as substrate for the recombinant PtdIns(5)P 4-kinase type II␣ in the presence of [␥-32 P]ATP. Fig. 5B shows that the product generated by MTM1 and MTMRs from PtdIns(3,5)P 2 is indeed PtdIns(5)P, as reflected by the generation of [␥-32 P]PtdIns(4,5)P 2 in the kinase assay. Overall, these results demonstrate that MTM1 overexpressed in mammalian cells can transform PtdIns(3,5)P 2 into PtdIns(5)P when tested in vitro. MTM1 Produces PtdIns(5)P in Vivo-We then checked whether MTM1 could increase PtdIns(5)P mass levels in vivo when expressed in mammalian cells. Jurkat cells were used because they have a relatively high basal amount of PtdIns(3,5)P 2 . 3 Cells were nucleofected with MTM1 or empty vector, using the Amaxa recommended protocol for this cell type. In these conditions, the efficiency of transfection reached 68 Ϯ 4%, as measured by FACS analysis of GFP-transfected cells (data not shown). Overexpression of MTM1 detected by Western blot on whole cell extracts is shown in Fig. 6A. Total lipids were extracted, and the mix of monophosphorylated PIs (PtdIns(3)P ϩ PtdIns(4)P ϩ PtdIns(5)P) were scraped off and submitted to the PtdIns(5)P 4-kinase type II␣ in vitro kinase assay to specifically phosphorylate PtdIns(5)P. As a positive control, HeLa cells were transfected by GFP-tagged IpgD, a bacterial phosphatase known to produce PtdIns(5)P (27). As shown in Fig. 6B, MTM1 overexpression induces an increase of [␥-32 P]PtdInsP 2 over mock-nucleofected cells. HPLC analysis of the [␥-32 P]PtdInsP 2 produced shows the presence of only one peak identified as PtdIns(4,5)P 2 using PIs standards. To obtain a more quantitative evaluation, the mass amount of PtdIns(5)P was determined by scraping the [␥-32 P]PtdIns(4,5)P 2 spot and comparing the incorporated radioactivity to a calibration curve obtained using increasing amounts of di-C8-PtdIns(5)P. Results show that MTM1 is able to increase the PtdIns(5)P level in Jurkat cells from a basal level of 37 to 77 pmol/10 7 cells (Fig.  6C). In agreement with the in vitro PtdIns(3,5)P 2 3-phosphatase activity of MTM1, these data indicate that the enzyme can produce PtdIns(5)P in mammalian cells.
MTM1-dependent PtdIns(5)P Production upon Osmotic Shock-PtdIns(3,5)P 2 levels were shown to be modulated by hyper-osmotic shock in yeast and in mammalian cells. We therefore investigated the effect of hyper-osmotic shock on PtdIns(5)P production by MTM1 in the L6 myoblasts. Cells allowed to differentiate to myotubes and MTM1, or the inactive substrate trapping D278A MTM1 mutant, were overexpressed using an adenovirus system, leading to almost 100% of infection efficiency (22). Mass levels of PtdIns(5)P were measured using the in vitro kinase assay after 2 days of infection and a short 0.9 M NaCl osmotic shock. Data show that expression of the inactive D278A MTM1 mutant in cells reduces by 50% the intracellular level of PtdIns(5)P compared with the expression of wild type MTM1 (Fig. 7A), indicating that MTM1 activity is required. In the same experimental conditions, levels of [ 32 P]PtdIns(3,5)P 2 were measured by HPLC after labeling of the total pool of PIs with [ 32 P]orthophosphate. HPLC profiles (Fig. 7B) show that [ 32 P]PtdIns(3,5)P 2 level is more important in the cells overexpressing the D278A MTM1 mutant compared with wild type (624,060 versus 475,440 cpm). DISCUSSION Mutations in the MTM1 gene are associated with the Xlinked myotubular myopathy, a severe genetic muscle disorder. The majority of the mutations leads to an absence or a decreased protein level. However, beside the catalytic cysteine (Cys-375), the mutation S376N abrogates the in vitro enzymatic activity (4) but does not affect the protein level in patient cells (23), confirming that the 3-phosphatase activity is essential for the physiological role of MTM1. In mammalian cells, PtdIns(3)P comes from PtdIns through the action of a class III 3 B. Payrastre, unpublished observations. FIG. 4. Substrate specificity of MTM1 in vitro using long acyl chains PIs. A, empty FLAG vector or FLAG-tagged MTM1 were overexpressed in HEK 293 cells, immunoprecipitated with anti-FLAG antibody, and used for phosphatase activity in vitro by using different C16-NBD-PtdIns as indicated. Lipids were separated on TLC, and fluorescence was registered under a UV table. B, immunoprecipitates from cells overexpressing MTM1 were resuspended in a total of 50 l of phosphatase assay buffer. 10 and 20 l of immunoprecipitate were tested using either C16-PtdIns(3)P or C16-PtdIns(3,5)P 2 , and phosphate release was measured by a Malachite green binding assay. Assays were performed in respect to linearity for substrate concentration and time (1 g of substrate, 8 min of incubation at 30°C).
PI-3 kinase, the homologue of the yeast vps34, and its constitutive level is relatively low compared with the abundant PtdIns(4)P. The most studied function of PtdIns(3)P is the regulation of intravesicular trafficking through binding to FYVE-or phox homology-containing early endosomal proteins, such as EEA1 (28) or sorting nexin 3 (29).
We analyzed the localization of PtdIns(3)P in fibroblasts and myoblasts from the established cell lines of XLMTM patients, where expression of MTM1 is strongly decreased versus normal cells (23). Binding of the 2X-FYVE probe to PtdIns(3)P demon-strated a colocalization with the early endosomal marker EEA1 independently of MTM1 expression, suggesting that the primary function of endogenous MTM1 is not to regulate bulk constitutive vesicular trafficking. It has been reported that overexpressed MTM1 can dephosphorylate the endosomal pool of the PtdIns(3)P (15,22). However, it cannot be excluded that this observation results from the massive levels of overexpressed phosphatase that could bypass in vivo regulation.
This still raises the question of the physiological role of MTM1 as a 3-phosphatase. Several hypotheses could explain FIG. 5. MTM1 phosphatase activity toward PtdIns(3,5)P 2 generates PtdIns(5)P in vitro. A, the recombinant PtdIns(5)P 4-kinase type II␣ was controlled for its specificity by using di-C16-PtdIns(3)P, -PtdIns(4)P, or -PtdIns(5)P as substrates in the in vitro kinase assay in the presence of [␥-32 P]ATP. B, empty vector, MTM1, MTMR2, and MTMR3 expressed in COS-7 cells were immunoprecipitated and tested in vitro for phosphatase activity using C16-NBD-PtdIns(3,5)P 2 . The spot corresponding to PtdIns was eluted and tested in the in vitro kinase assay using the recombinant PtdIns(5)P 4-kinase type II␣ and [␥-32 P]ATP. Lipids were separated by TLC, and generated [␥-32 P]PtdIns(4,5)P 2 was detected by a PhosphorImager system. the lack of remodeling of the endosomal pool of the PtdIns(3)P by endogenous MTM1. First, MTM1, which displays a very high specific activity in vitro, is likely to be tightly regulated in vivo by interaction with proteic partners. Such interactions could either modulate its activity or its proper localization by sequestering the phosphatase away from the total PtdIns(3)P pools. Indeed it has been shown recently that MTM family members could form heterodimers. For example, MTM1 can interact with MTMR12/3-PAP, which affects its localization (30); MTMR5 can interact with MTMR2 but not MTM1 (31); and MTMR7 can bind to MTMR9 (32). Second, it is likely that MTM1 affects only a minor pool of PtdIns(3)P in vivo, because even in cells overexpressing MTM1, the level of PtdIns(3)P affected by the phosphatase is never above 20% of total PtdIns(3)P (see Refs. 4, 22, and this paper). Finally, it is pos-sible that its hydrolysis activity on PtdIns(3)P cannot totally account for the function of MTM1. Indeed, although first described as a specificity of MTMR3 (16), hydrolysis of PtdIns(3,5)P 2 has been also found for the mouse homologue of MTMR2, another member of the MTM family (17).
We report that MTM1 and MTMR1 use PtdIns(3,5)P 2 as well as PtdIns(3)P in vitro and in mammalian cells. These observations suggest that hydrolysis of PtdIns(3,5)P 2 is a common feature of this family of lipid 3-phosphatases. While this paper was in preparation, Schaletzky et al. (33) showed that recombinant MTM1 can hydrolyze PtdIns(3,5)P 2 in vitro and in yeast. Therefore, this still raises questions regarding the functional redundancy of the family members, because MTM1 and MTMR1 to R3 all show the same substrate specificity (as shown in this paper), the same cytoplasmic localization (19), FIG. 6. MTM1 produces PtdIns(5)P in vivo in Jurkat cells. Jurkat cells were nucleofected with 1 g of pcDNA-FLAG or pcDNA-FLAG-MTM1 plasmid DNA/10 6 cells using AMAXA nucleofector kit V (Amaxa Biosystems). 24 h post-nucleofection, cells were washed once in PBS and tested for MTM1 expression by Western blotting using the specific 1G6 anti-MTM1 antibody (1:5000) (A). B, in vitro kinase assay using the recombinant PtdIns(5)P 4-kinase type II␣. The area corresponding to PtdInsP 2 migration was scraped off, deacylated, and analyzed by using an HPLC system. The same experimental setup in which bacterial phosphatase IpgD has been overexpressed is shown as a positive control. C, quantification of the PtdIns(5)P mass amounts in the transfected cells. and a ubiquitous expression (34). However, discrete differences can be noted. When compared with MTM1 and MTMR1, the cytoplasmic gradient of MTMR2 and MTMR3 is more concentrated around the nucleus (9,15,19). Moreover MTM1 expression increases in differentiated muscle cells, although MTMR2 is found predominantly in central and peripheral nervous systems by in situ hybridization (9,15,17,35).
Our data demonstrate that of the seven PIs tested in the in vitro phosphatase assay after overexpression in cells, only PtdIns(3)P and PtdIns(3,5)P 2 are substrate for the enzymes. Like PtdIns(3)P, constitutive PtdIns(3,5)P 2 levels are low in cells. PtdIns(3,5)P 2 levels can be increased in response to UV radiation stress (36) and were shown to maintain vacuole sorting and integrity in yeast, where its levels are considerably increased under hyperosmotic shock (37). In this paper, we show for the first time that MTM1 is able to produce PtdIns(5)P FIG. 7. PtdIns(5)P-mediated production upon hyperosmotic shock in L6 myotubes. L6 myotubes were infected with adenovirus GFP or adenoviruses expressing wild type (WT) and D278A MTM1 mutant as described under "Experimental Procedures." After 2 days, the cells were treated for 10 min with 0.9 M NaCl. A, total lipids were then extracted, and PtdIns(5)P was measured by using the in vitro mass assay. B, the cells were labeled overnight with [ 32 P]orthophosphate before the osmotic shock, and the lipids were separated on TLC, and the different class of PIs were analyzed by HPLC and compared with known standards peaks: peak 1, PtdIns(3,5)P 2 ; peak 2, PtdIns(3,4)P 2 ; and peak 3, PtdIns(4,5)P 2 . Black arrow indicates the PtdIns(3,5)P 2 peak. The results were normalized to PtdIns(4,5)P 2 levels and are representative of two independent experiments. through its action on PtdIns(3,5)P 2 in vitro and in mammalian cells. The presence of PtdIns(5)P in mammalian cells was only recently reported mainly because its low amount and the difficulty to separate this PI from PtdIns(4)P by conventional HPLC techniques. The PtdIns(5)P level was shown to increase under stress in Chlamydomonas moewusi (38) and in mammalian cells after thrombin stimulation of blood platelets (26). In these two reports, the way of synthesis of the PI is unknown. PtdIns(5)P is believed to partially account for the production of PtdIns(4,5)P 2 in fibroblasts (18) by a PtdIns(5)P 4-kinase type II. It has been shown that PtdIns(5)P can be produced by phosphorylation of PtdIns by PIKfyve, the mammalian orthologue of the yeast fab1 kinase (39). Moreover, we described recently (27) the production of PtdIns(5)P by the Shigella flexneri enzyme IpgD, a PtdIns(4,5)P 2 4-phosphatase during infection of the host cell by the bacterium. PtdIns(5)P was also detected in the nucleus and shown to bind to PHD finger protein domains and therefore regulate nuclear response to DNA damage (40). By using a specific mass assay (26), we showed that in mammalian cells, MTM1 could account for the production of PtdIns(5)P. Indeed, in Jurkat cells, overexpression of MTM1 doubles the cellular mass amount of PtdIns(5)P. Moreover, upon osmotic shock, we clearly observed a decrease in PtdIns(5)P amounts in L6 cells that are expressing the D278A MTM1 mutant compared with wild type MTM1. This catalytic inactive mutant is defined as a substrate trapping mutant, and we hypothesize that it would have the capacity to compete with the endogenous MTM1 for PtdIns(3,5)P 2 binding. Indeed, we observed a 50% decrease of PtdIns(5)P upon D278A mutant expression in L6 cells. Moreover, the decrease in PtdIns(3,5)P 2 observed after labeling of the metabolic PI pools is dependent on the enzymatic activity of the phosphatase because it is observed only in the presence of MTM1 and not with the inactive mutant D278A. Thus, these data suggest that the 3-phosphatase MTM1 could also account for the production of PtdIns(5)P in vivo, through hydrolysis of PtdIns(3,5)P 2 , and that this PIs could transduce some of its cellular functions.
As discussed above, PtdIns(5)P is a recently discovered PI, and its function is still under investigation. Schaletzky et al. (33) have proposed that PtdIns(5)P is a specific activator of MTM1 activity by inducing oligomerization of MTM1 in vitro. Clearly, more studies are now necessary to determine the impact of the modification of PtdIns(3)P, PtdIns(3,5)P 2 , and PtdIns(5)P levels in the etiology of the genetic disease.