View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Cell Article

Mitochondrial Phosphatase PTPMT1 Is Essential for Cardiolipin

Ji Zhang,1 Ziqiang Guan,5 Anne N. Murphy,1 Sandra E. Wiley,1 Guy A. Perkins,6 Carolyn A. Worby,1 James L. Engel,1 Philip Heacock,7 Oanh Kim Nguyen,1 Jonathan H. Wang,1 Christian R.H. Raetz,5 William Dowhan,7 and Jack E. Dixon1,2,3,4,* 1Department of Pharmacology 2Department of Cellular and Molecular Medicine 3Department of Chemistry and University of California, San Diego, La Jolla, CA 92093, USA 4Howard Hughes Medical Institute, Chevy Chase, MD 20815-6789, USA 5Department of Biochemistry, Duke University Medical Center, Durham, NC 27710, USA 6National Center for Microscopy and Imaging Research, University of California, San Diego, La Jolla, CA 92093-0608, USA 7Department of Biochemistry and Molecular Biology, University of Texas-Houston Medical School, Houston, TX 77030, USA *Correspondence: [email protected] DOI 10.1016/j.cmet.2011.04.007

SUMMARY importance of PTEN in tumor biology prompted us to identify additional phosphatases similar to PTEN (Pagliarini et al., PTPMT1 was the first protein phosphatase 2004). We used the amino acid sequence surrounding the active found localized to the mitochondria, but its biological site, HCKAGKGR, which includes two invariant basic amino function was unknown. Herein, we demonstrate acids (bold) along with the catalytically essential Cys-X5-Arg that whole body deletion of Ptpmt1 in mice leads to (CX5R) motif, to search for closely related phosphatases. To embryonic lethality, suggesting an indispensable our surprise, we discovered a protein tyrosine phosphatase role for PTPMT1 during development. Ptpmt1 defi- (PTP) that localized exclusively to the inner mitochondrial membrane (Pagliarini et al., 2005). This was the first CX R ciency in mouse embryonic fibroblasts compromises 5 phosphatase found in mitochondria, and, because of its unique mitochondrial respiration and results in abnormal localization, we named it PTPMT1 (PTP localized to the Ptpmt1 mitochondrial morphology. Lipid analysis of - mitochondrion 1). This phosphatase is highly conserved and is deficient fibroblasts reveals an accumulation of present in animals, plants, and bacteria (Pagliarini et al., 2004). phosphatidylglycerophosphate (PGP) along with Mitochondria are ubiquitous and dynamic organelles that a concomitant decrease in phosphatidylglycerol. participate in crucial cellular processes in eukaryotic organisms. PGP is an essential intermediate in the biosynthetic They are responsible for the production of over 90% of cellular pathway of cardiolipin, a mitochondrial-specific ATP (Voet and Voet, 2004). Mitochondria are also the sites of phospholipid regulating the membrane integrity and fatty acid oxidation, ketone body production, heme biosyn- activities of the organelle. We further demonstrate thesis, cardiolipin metabolism, and coenzyme Q synthesis. In that PTPMT1 specifically dephosphorylates PGP addition, mitochondria are a major source of reactive oxygen species. They harbor key steps of and the in vitro. Loss of PTPMT1 leads to dramatic diminu- urea cycle and are central to the mechanisms of apoptosis (New- tion of cardiolipin, which can be partially reversed meyer and Ferguson-Miller, 2003; Voet and Voet, 2004). by the expression of catalytic active PTPMT1. Our Due to the unique localization of PTPMT1 in the inner mito- study identifies PTPMT1 as the mammalian PGP chondrial membrane, we postulated that it plays important roles phosphatase and points to its role as a regulator of in oxidative or other mitochondrial functions. cardiolipin biosynthesis. Therefore, we sought to elucidate the cellular function(s) and identify the endogenous (s) of this phosphatase. INTRODUCTION RESULTS Reversible phosphorylation is a major regulatory mechanism controlled by more than 500 and 100 phosphatases Inactivation of the Ptpmt1 Gene Leads to Embryonic (Alonso et al., 2004; Manning et al., 2002). Although the majority Lethality of kinases and phosphatases function on protein substrates, To study the physiological function(s) of PTPMT1, we genetically there are notable examples of kinases and phosphatases that engineered mice in which the Ptpmt1 gene was ablated through regulate the production and breakdown of phospholipids. For homologous recombination (Figure 1A). Subsequently, we inter- instance, phosphoinositide 3- (PI3K) and PTEN are two crossed Ptpmt1-heterozygous mice to examine the conse- that regulate the abundance of key cellular phospho- quences of PTPMT1 deficiency. Out of 173 offspring born, lipids (Maehama and Dixon, 1998; Whitman et al., 1988). The 66 were wild-type, 107 were heterozygous, and none were

690 Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

Inhibition of Mitochondrial Respiration in Cells Lacking PTPMT1 To further investigate the physiological functions of PTPMT1, we generated a floxed allele of Ptpmt1 in mice for Cre-mediated conditional knockout (Ptpmt1+/flox)(Figure 1A and Figure S1B). We next derived Ptpmt1flox/flox mouse embryonic fibroblasts (MEFs) from E13.5 embryos obtained from intercrosses of Ptpmt1+/flox mice. The cells were then infected with control adenovirus or adenovirus expressing Cre recombinase. The expression of PTPMT1 protein was almost completely ablated in the Cre-treated (KO) cells 3 days after the infection and remained negligible for at least 2 weeks in culture (Figure 1C). The level of PTPMT1 mRNA transcripts was also greatly dimin- ished in those cells (Figure S1C). MEFs lacking PTPMT1 grow slowly. Three days after infection with Cre recombinase, the proliferation rate of KO cells was reduced 25% compared to that of control adenovirus-treated cells (Flox) (Figure S2A). Inhibition of was more pronounced in KO MEFs at day 6. A colony formation assay was also performed to examine the long-term effect of PTPMT1 deficiency on cell growth. Not surprisingly, KO MEFs formed fewer and smaller colonies than Flox cells (Figures S2B and S2C). It is noteworthy that Ptpmt1 null cells were not apoptotic or necrotic (data not shown) and had cell sizes similar to Flox MEFs (Figure S2D). The position of PTPMT1 on the inner mitochondrial membrane places it within close proximity to electron transport chain complexes and enzymes of the tricarboxylic acid cycle. Therefore, we investigated the effects of PTPMT1 deletion on mitochondrial respiration and oxidative phosphorylation. Intact cell oxygen consumption rates (OCR) were measured at 13 days after deletion of Ptpmt1 (Figure 2A). The basal respira- tion rates were substantially lower in Ptpmt1 KO cells relative to that of Flox MEFs. We also assessed state 4 rates (a measure- Figure 1. Inactivation of the Ptpmt1 Gene Leads to Mouse Embry- ment of proton leakage across the inner membrane) by the onic Lethality addition of the ATP synthase inhibitor oligomycin, and maximal (A) Gene targeting strategy. Exons (blue boxes), translation initiation site (ATG), respiratory rates by the addition of the uncoupler, carbonylcya- selection gene (neomycin, neo), positions of 50 and 30 probes for Southern blot nide-r-trifluoromethoxyphenylhydrazone (FCCP). A dramatic (red bars), PCR primer (red arrows), loxP sites (blue triangles), Frt sites (green reduction in maximal respiratory rates as well as state 4 rates triangles), and restriction sites are indicated. X-XbaI, E-EcoRI. was observed in the KO cells (Figure 2A), suggesting that (B) Genotype of offspring from Ptpmt1 heterozygous intercrosses at various PTPMT1 is required for optimal mitochondrial respiration. developmental stages. (C) Derivation of Ptpmt1-deficient mouse embryonic fibroblasts. Ptpmt1flox/flox In contrast, when we concomitantly measured the cellular MEFs were infected with recombinant adenoviruses encoding GFP(–) or ATP content, an increase in ATP level was observed in GFP-Cre (+). Cells were harvested at the indicated time points postinfection for Ptpmt1-deficient MEFs. The increase was more evident after western blot analysis. dpi, days postinfection. preincubation of cells with low medium (Figure 2B), See also Figure S1. consistent with our previous report that suppression of PTPMT1 expression by small-interfering RNA (siRNA) markedly homozygous null (Figure 1B) (see Figure S1 available online for enhances ATP levels in INS-1 832/13 cells in response to low detailed characterization of mouse genotyping), indicating glucose exposure (Pagliarini et al., 2005). Mitochondrial oxida- a Mendelian ratio typical of embryonic lethality. We next con- tive phosphorylation is a major source of cellular ATP produc- ducted timed mating experiments to determine the stage of tion; however, it is known that cells can modulate the balance the embryonic lethality. Embryos from E8.5, E10.5, E14.5, and between aerobic and anaerobic (glycolytic) ATP production E17.5–18.5 gestational stages were primarily wild-type or (Warburg, 1956) and that lowering extracellular glucose heterozygous for the Ptpmt1 allele (Figure 1B). Further dissection enhances glucose transport and utilization (Kletzien and identified an average of 26% resorption sites on the uterus Perdue, 1975). The diminished mitochondrial respiratory lacking discernible embryos from all developmental stages capacity in the presence of an elevated ATP level prompted (Figure 1B), suggesting that Ptpmt1–/– embryos implanted but us to hypothesize that glycolysis is increased in Ptpmt1-defi- died prior to E8.5. Hence, our results suggest that PTPMT1 plays cient cells. Therefore, we carried out a time course analysis an indispensable role in early development. of mitochondrial respiration versus glycolysis following low

Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. 691 Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

Figure 2. Inhibition of Mitochondrial Respi- ration in Ptpmt1-Deficient Cells (A) Oxygen consumption rates (OCR) of intact MEFs cultured in regular medium (25 mM glucose) at 13 dpi. The basal rate, the state 4 rate (with 1 mM oligomycin, Oligo), and the maximal uncoupler stimulated rate (in the presence of FCCP) are shown. Rotenone (Rot) and antimycin A (Anti A) were used to block all electron flow through complexes I and III, demonstrating that non- mitochondrial oxygen consumption in these cells is negligible. Mean ± SEM from four replicates is shown. (B) Whole-cell ATP contents in MEFs (13 dpi) cultured in regular medium or preincubated with 3 mM glucose containing medium overnight. Mean ± standard deviation (SD), n = 3, **p < 0.01. (C and D) Maximal uncoupler stimulated respira- tion rates (C) and extracellular acidification rate (ECAR) rates (D) of Ptpmt1-Flox and KO MEFs preincubated with low glucose (3 mM) media overnight are obtained at the indicated time post adenoviral infection. Mean ± SEM from four repli- cates is shown. **p < 0.01, *p < 0.05. (E) At 13 dpi, crude mitochondria were isolated from Flox or KO MEFs, solubilized in b-dode- cylmaltoside, subjected to BN-PAGE, and followed by consecutive complex I and complex II in-gel activity assays. The activities of each complex are indicated by the appearance of a pink colored band. See also Figure S2.

glucose incubation. Four and 6 days after Ptpmt1 deletion, the chondria from both Flox and KO MEFs at 13 days after deletion maximal oxygen consumption rates in Ptpmt1-deficient cells of the Ptpmt1 gene. The electron transport chain complexes were not statistically different than those in Flox cells (Fig- were separated by blue-native polyacrylamide gel electropho- ure 2C). However, significant reduction of maximal respiration resis (BN-PAGE) followed by in-gel activity assays. A NADH: was observed in Ptpmt1-deficient cells 20 days later. Similar Nitrotetrazolium blue (NTB) reductase assay was first carried results were obtained in cells cultured in regular growth media out to examine the functional integrity of complex I (Figure 2E, (data not shown). To determine the glycolytic capacity of KO left panel). Under the same assay conditions, a lower band cells, we measured the rates of extracellular acidification as was also observed, which is likely to be dihydrolipoamide dehy- an indicator of glycolytic lactate production in Flox and KO drogenase (DLDH) (Yan and Forster, 2009), a component of MEFs after low glucose incubation (Parce et al., 1989; Wu multiple mitochondrial dehydrogenases. A profound reduction et al., 2007). Indeed, a substantial increase of extracellular of complex I activity was observed in cells lacking PTPMT1, acidification rates (ECAR) was observed in KO MEFs at all whereas the DLDH activity remained unchanged. Subsequently, time points, even as early as 4 days after the deletion of Ptpmt1 we determined the activity of complex II by performing a succi- (Figure 2D). Additionally, when we measured cellular lactate nate: NTB reductase assay on the same gel strip. In Ptpmt1 production levels directly using a colorimetric assay, KO null mitochondria, complex II activity was slightly decreased. MEFs displayed a marked increase of lactate at day 13 (Fig- (Figure 2E, right panel). Our results demonstrate that PTPMT1 ure S2E), when the inhibition of mitochondrial respiration was deletion profoundly inhibits the activity of complex I, consistent apparent (Figure 2A). Together, our results show that prolonged with the decreased respiratory rates we observe in those cells. loss of PTPMT1 expression leads to a profound reduction of Together, our results indicate that PTPMT1 is required for mito- mitochondrial maximal respiration, concomitant with an chondrial respiration, likely through its regulation of electron increase in cellular ATP content. The increase of ATP levels in transport chain complexes. KO cells may result from an augmentation of glycolysis, although we cannot rule out the possibility that the utilization PTPMT1 Maintains Mitochondrial Membrane Integrity of ATP was also diminished, particularly since KO MEFs grow We next examined the effect of Ptpmt1 deletion on mitochondrial at a much slower rate than Flox cells (Figures S2A–S2C). morphology using immunocytochemistry. Cytochrome c stain- To further understand the mechanism by which PTPMT1 ing of Flox MEFs revealed a typical reticular mitochondrial reduces mitochondrial respiratory capacity, we isolated mito- network that distributed in a radial manner (Figure 3A, left panel).

692 Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

(Figure 3C), such as vesicular matrix compartments (inset) with ruptured outer mitochondrial membranes (red arrow), and degradation of cristae and matrix components (red asterisk). Mitochondrial length was subsequently determined by morpho- metric analysis of electron micrograph images. Not surprisingly, mitochondria were much shorter in KO cells than those in Flox cells (Figure 3D), which further supports that loss of PTPMT1 resulted in mitochondrial fragmentation. In addition, the loss of cristae in KO MEFs was confirmed by morphometric analysis comparing the ratio of cristae/outer mitochondrial membrane (OMM) surface area in Flox and KO cells (Figure 3E). Matrix vesicular formation has been observed in other paradigms of mitochondrial damage, such as apotosis (Sun et al., 2007). However, the distortion and deficiency of cristae in KO MEFs indicated that PTPMT1 is essential for the integrity of the inner mitochondrial membrane.

Accumulation of PGP in Ptpmt1-Deficient Cells A full understanding of PTPMT1’s function requires identification of the endogenous substrate(s) of this phosphatase. Although initially characterized as a phosphoinositide phosphatase with a preference for phosphatidylinositol 5- (PI(5)P) in vitro, our results indicated that PI(5)P is not likely to be the physiolog- ical substrate of PTPMT1 (Pagliarini et al., 2004, 2005). In addi- tion, the poor in vitro activity of PTPMT1 toward proteinaceous substrates prompted us to employ an unbiased search for a lipid substrate by comparing cellular phospholipid profiles of Flox and KO MEFs. Total lipids from both cells were extracted and subjected to liquid chromatography followed by tandem mass spectrometry (MS/MS) analysis. Levels of phosphatidylethanolamine, phos- phatidylserine, and phosphatidylinositol were comparable in both Flox and KO MEFs (data not shown). The abundance of the major phosphatidylglycerophosphate (PGP) species (34:1) as well as minor forms of PGP were markedly elevated in KO cells. In contrast, levels of all PGP species were barely detect- Figure 3. PTPMT1 Maintains Mitochondrial Membrane Integrity able in Flox MEFs (Figures 4A and 4B; see Figure S3B for the (A) Ptpmt1-deficient MEFs have fragmented mitochondria. Two weeks after identification of 34:1 PGP species by MS/MS analysis). deletion of Ptpmt1, cells were stained with anti-cytochrome c antibody and In the 1950s and 1960s, Eugene Kennedy and his coworkers visualized by immunofluorescence microscopy. DNA was stained blue with characterized the biosynthetic pathways for numerous cellular 0 4 ,6-diamidino-2-phenylindole (DAPI). phospholipids including cardiolipin (Rock, 2008; Vance and (B) Electron micrographs of Flox MEFs. Normal mitochondria with lamellar Vance, 2008). In , glycerol-3-phosphate is first acyl- cristae are enlarged in the inset. The scale bar represents 1 mm. (C) EM images of defective mitochondria in KO cells. Inset shows a vesicular ated to form phosphatidic acid and then converted to cytidinedi- mitochondrion and rupture of the outer mitochondrial membrane/OMM (red phosphate diacylglycerol (CDP-DAG) by the CDP-DAG arrow). Red asterisk, loss of cristae and matrix components. synthase. Subsequent steps in cardiolipin biosynthesis take (D) Morphometric analysis of mitochondrial length. Mean ± SEM, n = 80. place in the inner membrane of mitochondria. These steps (E) Morphometric analysis of cristae/OMM surface area ratio. Mean ± SEM, involve the conversion of CDP-DAG to phosphatidylglycerol n = 52. ***p < 0.001, **p < 0.01. (PG) by the sequential action of PGP synthase and a PGP phos- phatase. Cardiolipin is subsequently produced via condensation In contrast, KO MEFs exhibited a distinct punctate staining of PG with CDP-DAG by cardiolipin synthase (Figure 4F). Most of pattern, suggesting fragmentation of mitochondria (Figure 3A, the biosynthetic enzymes in this pathway have been extensively right panel). studied except the PGP phosphatase whose identity has re- In order to further characterize this phenotype, the ultrastruc- mained elusive in mammals. Our phospholipid profiling analyses ture of the mitochondria was examined in Flox and KO MEFs by show that PGP is markedly elevated in KO MEFs, indicating that transmission electron microscopy. As shown in the electron PTPMT1 is the mammalian enzyme responsible for the conver- micrograph in Figure 3B, Flox cells had many normal, elongated sion of PGP to PG. If PTPMT1 was the phosphatase that dephos- mitochondria with abundant cristae in a lamellar architecture phorylates PGP, we would expect to find PG levels decreased in (inset). In contrast, deletion of Ptpmt1 produced many smaller KO cells. As predicted, while several molecular forms of PG were mitochondria with a variety of morphological abnormalities identified in the Flox cells (Figure 4C), we observed a substantial

Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. 693 Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

Figure 4. Loss of PTPMT1 Leads to the Accumulation of Phosphatidylglycerophos- phate In Vivo (A and B) Mass spectra of PGP in Flox (A) and KO (B) MEFs 13 days after deletion of Ptpmt1. Insets of (A) and (B) identified [M-2H]2– ions of 34:1 PGP 13 (exact mass m/z 413.239) at m/z 413.25, its C1 13 isotope at m/z 413.75, C2 isotope at m/z 414.25, and 34:2 PGP at m/z 412.24. The total number of carbons and unsaturations in the fatty acyl chains are listed to indicate the molecular species of each ion. (C and D) Mass spectra [M-H]– ions of PG and [M-2H]2– ions cardiolipin (CL) in Flox and KO MEFs, respectively. The scales of (C) and (D) are set at different levels to facilitate the observation of PG and cardiolipin ions in KO cells. (E) Ratio of 34:1 PGP to 34:1 PG at the indicated time after Ptpmt1 deletion. (F) Schematic diagram of cardiolipin biosynthesis pathway. PA, phosphatidic acid; CDS, CDP-DAG synthase; PGS, PGP synthase; CLS, cardiolipin synthase; TAZ, cardiolipin remodeling enzyme, Tafazzin. Enzymatic reactions that take place in mitochondrion are enclosed within the blue rectangle. See also Figure S3.

Ptpmt1 ablation. The ratio of 34:1 PGP/ PG was determined by mass spectrom- etry analysis (Figure 4E). Notably, an increase in the PGP/PG ratio was found in KO MEFs as early as 3 days after the deletion of Ptpmt1 (Figure 1C). Moreover, the PGP/PG ratio was augmented in a time-dependent manner in KO cells as compared to that of Flox MEFs. A similar accumulation of PGP and a reduction of PG were observed in C2C12 mouse myoblasts in which PTPMT1 was knocked down by siRNA in a separate lipid analysis (data not shown). These results indicate that PTPMT1 specifically catalyzes the conversion of PGP to PG in multiple cell types.

PTPMT1 Dephosphorylates PGP In Vitro To determine whether PTPMT1 pos- sesses PGP phosphatase activity in vitro, we enzymatically synthesized radiola- beled PGP. In brief, phosphatidyl-sn- [U-14C]glycerophosphate (14C-PGP) was produced by incubating CDP-DAG with sn-[U-14C] glycerol 3-phosphate in the presence of purified PGP synthase (PGS) from E. coli (Dowhan, 1992). An reduction in all PG species in cells lacking PTPMT1 (Figure 4D). overnight incubation resulted in the quantitative conversion of To further analyze the change in PGP and PG levels induced by starting materials to 14C-PGP as assessed by thin layer chroma- loss of PTPMT1, MEFs were harvested at 3, 6, and 13 days after tography (TLC) (data not shown).

694 Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

Figure 5. PTPMT1 Dephosphorylates Phos- phatidylglycerophosphate In Vitro (A) Conversion of 14C-labeled PGP to 14C-PG by recombinant wild-type (WT) versus catalytically inactive (C132S) PTPMT1. A trace amount of ra- diolabeled lysoPG was generated during the reaction. (B) The specific activities of PTEN, Laforin, DSP21, and VHR are compared with that of PTPMT1 using pNPP as the substrate. Mean ± SD, n = 3. (C) PGP phosphatase activity is not a general feature of protein tyrosine phosphatases. The activity of each enzyme using PGP as the substrate is indicated by the absorbance at 620 nm using malachite green reagents to measure the release of inorganic phosphate. Mean ± SD, n = 3.

Notably, our mass spectrometry anal- ysis of phospholipids revealed a marked decrease in cardiolipin (CL) content in KO MEFs 13 days after deletion of Ptpmt1 (Figures 4C and 4D). Moreover, The 14C-PGP was then incubated with purified recombinant both the content and the acyl chain composition of cardiolipin PTPMT1. The radioactive reaction products were extracted and were altered in KO cells. To further quantify the effect of PTPMT1 separated by TLC analysis. 14C-labeled PGP remained near the on cardiolipin metabolism, we next determined the steady-state origin, whereas 14C-PG migrated to the middle of the TLC plate phospholipid composition of Flox versus KO MEFs by labeling (Figure 5A), as indicated by the comigration of nonradiolabeled intact cells with 32P-orthophosphate. Two weeks after Cre PG standards (data not shown). While recombinant PTPMT1 effi- recombinase infection, cardiolipin and PG levels in KO MEFs ciently converted PGP to PG in a dose-dependent manner, the were reduced to 36% and 39% of those in Flox cells, respec- catalytically inactive mutant of PTPMT1 (C132S) failed to catalyze tively (Table 1), indicating that PTPMT1 is required for the biosyn- the conversion. Our results demonstrate that PTPMT1 dephos- thesis of both lipids. Notably, the abundance and composition of phorylates PGP in vitro. To rule out the possibility that the PGP cardiolipin remained indistinguishable in Flox and KO MEFs at 3 phosphatase activity is a general feature of the PTP superfamily, and 6 days after Ptpmt1 deletion (Figures S3C and S3D), which in we compared the activity of PTEN, laforin, DSP21, and VHR turn explained why the mitochondrial oxygen consumption rates against PGP to that of PTPMT1. The specific activity of each were not reduced in KO cells at early time points (Figure 2C). The enzyme using pNPP as a substrate is consistent with previously time-dependent loss of cardiolipin in KO cells may reflect the reported values (Figure 5B) (Denu et al., 1995; Maehama and slow turnover rate of this mitochondrial lipid, which may be influ- Dixon, 1998; Rardin et al., 2008; Worby et al., 2006). PTPMT1 enced by a combination of factors, including degradation and re- was more efficient at removing phosphate from 14C-PGP than cycling (Schlame, 2008). It is important to note that the relative the other phosphatases (Figure 5C), suggesting that the PGP abundance of phosphatidic acid, phosphatidylcholine, phos- phosphatase activity is not a common property shared by PTP phatidylethanolamine, phosphatidylserine, and phosphatidyli- superfamily members. Taken together, our results strongly nositol were not affected by the loss of PTPMT1 (Table 1). The suggest that PGP is the physiological substrate of PTPMT1. level of sphingomyelin was decreased to a lesser extent in the KO MEFs. The physiological significance of this reduction PTPMT1 Is Required for Cardiolipin Biosynthesis remains unclear. Collectively, our data demonstrate that As the conversion of PGP to PG is an integral part of the de novo PTPMT1 is required for cardiolipin biosynthesis, consistent cardiolipin biosynthetic pathway, we predict that loss of PTPMT1 with its function as the PGP phosphatase. would lead to a reduction in cardiolipin content in the cells. Car- diolipin is one of the major components of the mitochondrial The Catalytic Activity of PTPMT1 Is Essential inner membrane (Daum, 1985). It optimizes mitochondrial respi- for Cardiolipin Synthesis ration through its interactions with electron transport chain To verify that the catalytic activity of PTPMT1 is required for complexes and multiple carrier proteins including the adenine cardiolipin biosynthesis, we next examined whether wild-type transporter, as well as stabilizing the supercomplex or a phosphatase inactive mutant (C132S) of PTPMT1 is able formation of respiratory enzymes (Hatch, 1998; Houtkooper to rescue the mitochondrial defects in KO cells. One week after and Vaz, 2008; Schlame et al., 2000). As such, it is understand- deletion of Ptpmt1, MEFs were infected with recombinant able that Ptpmt1-deficient cells would have profound defects in adenoviruses encoding b-galactosidase (LacZ), wild-type, or mitochondrial respiratory capacity and inner mitochondrial mutant PTPMT1. Cardiolipin content and mitochondrial respira- membrane morphology due to cardiolipin deficiency. tion were subsequently analyzed after rescuing for 7 days. Both

Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. 695 Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

Table 1. Phospholipid Composition of Ptpmt1 Flox and KO MEFs PTPMT1 is indeed essential for cardiolipin synthesis and mito- chondrial function. Percent of Total Phospholipids Flox KO PTPMT1 Functionally Replaces Yeast PGP Cardiolipin 2.8 ± 0.2 1.0 ± 0.3*** Phosphatase Gep4 Phosphatidylglycerol 1.8 ± 0.3 0.7 ± 0.2** Putative PTPMT1 orthologs have been identified in a variety of Phosphatidic acid 1.6 ± 0.3 1.5 ± 0.4 evolutionarily distinct species, except the budding yeast Phosphatidylcholine 39.9 ± 2.0 45.1 ± 2.4 S. cerevisiae (Pagliarini et al., 2004). Osman et al. recently Phosphatidylethanolamine 28.1 ± 0.6 28.1 ± 0.5 reported that Gep4 functions as a PGP phosphatase in yeast Phosphatidylserine 11.0 ± 1.0 10.3 ± 1.4 mitochondria (Osman et al., 2010). Gep4 is a member of the haloacid dehydrogenase superfamily and is present only in fungi Phosphatidylinositol 11.2 ± 0.5 10.7 ± 0.3 and plants. Therefore, we sought to determine whether PTPMT1 Sphingomyelin 3.7 ± 0.3 2.6 ± 0.4* serves as a functional equivalent of Gep4 in yeast. After 13 days of adenoviral infection, cells were cultured in the medium FLAG-tagged Gep4, wild-type PTPMT1, or mutant PTPMT1 containing 32P (1uCi/ml) for 4 days. Total cellular phospholipids were i was first transformed into GEP4 null strain, and the expression extracted and analyzed as described in the Experimental Procedures. was confirmed by western blot analysis (Figure 6E, bottom two Data shown are the mean ± SEM for three independent experiments and are statistically analyzed with a Student’s t test. ***p < 0.001, panels). PTPMT1 contains an N-terminal mitochondrial targeting **p < 0.01, *p < 0.05 between Flox and KO MEFs. sequence (Pagliarini et al., 2005), which we hypothesized would also direct PTPMT1 to the yeast mitochondria. Indeed, our subcellular fractionation analysis indicated that PTPMT1 prefer- wild-type and mutant PTPMT1 efficiently expressed in the cells entially accumulated in the yeast mitochondria similar to Gep4 (Figure 6A), but expression of the CS mutant severely impaired (data not shown). Consistent with the observation by Osman the cell growth of KO MEFs (data not shown). Nonetheless, et al., the yeast strain that lacks GEP4 did not grow at 37Con we examined steady-state phospholipid composition. The glucose-containing medium (Figure 6E, second panel). Both abundance of cardiolipin was greatly diminished in KO MEFs Gep4 and wild-type PTPMT1 efficiently rescued the temperature expressing LacZ, compared to that of the LacZ-expressing sensitive growth defects. However, the PTPMT1 CS mutant Flox cells (Figure 6B), consistent with our earlier observation failed to complement the loss of GEP4. In addition, gep4D cells (Table 1). Interestingly, overexpression of wild-type PTPMT1 were unable to grow in the presence of ethidium bromide, an partially restored the cardiolipin content of KO MEFs. In contrast, agent that induces loss of mitochondrial DNA and defects in the cardiolipin level was further decreased in KO cells expressing cell wall (Janitor and Subı´k, 1993; Zhong et al., 2005). Notably, the CS mutant, suggesting that the catalytic activity of PTPMT1 overexpression of wild-type PTPMT1 greatly improved the is important for cardiolipin biosynthesis. Given the dramatic growth of gep4D cells in ethidium bromide containing medium, alteration of mitochondrial morphology in KO cells, it is conceiv- similar to that of Gep4-expressing cells (Figure 6E, third panel). able that overexpression of wild-type PTPMT1 may not be Our lipid analysis indicated that wild-type but not mutant able to fully reverse the structural defect and cardiolipin PTPMT1 rescued the cardiolipin deficiency of gep4D cells (Fig- deficiency in the time frame of our experiment. Moreover, the ure S4). Together, our results demonstrate that PTPMT1 abundance of PGP was simultaneously examined by TLC complements the loss of GEP4 in yeast and this rescue is depen- analysis (Figure 6C). PGP was barely detectable in Flox cells, dent on the phosphatase activity of PTPMT1. whereas a distinct PGP band was found in KO MEFs expressing LacZ, as indicated by the comigration of 14C-labeled PGP. As DISCUSSION expected, wild-type PTPMT1 alleviated the accumulation of PGP in KO cells. In contrast, overexpression of the PTPMT1 Over 100 human genes have been identified belonging to the

CS mutant in Ptpmt1 null cells substantially augmented cellular PTP superfamily, all of which contain a highly conserved CX5R PGP content, suggesting that this mutant protein may function motif in the for (Alonso et al., 2004; Tonks, as a trap to accumulate PGP. Similar ‘‘substrate-trapping’’ 2006). Most members of the PTP family are key mediators of mutants have been studied in many members of PTP family to through their of proteins stabilize and identify their endogenous targets (Tonks and and phosphoinositides (Tonks, 2006). The unique subcellular Neel, 2001). In contrast, the CS mutant had a minimal effect on localization of PTPMT1 prompted us to investigate whether it the levels of cardiolipin and PGP in Flox cells, probably because plays an important role in mitochondrial signaling. In the present the endogenous PTPMT1 can efficiently convert PGP to PG. study, we discovered that PTPMT1 dephosphorylates PGP, To rescue the mitochondrial defects in Ptpmt1-deficient cells, a glycerophospholipid, and is essential for mitochondrial we next overexpressed either wild-type or mutant PTPMT1 function through its regulation of cardiolipin biosynthesis. Our and measured mitochondrial respiration rates of those cells findings strongly suggest that phosphatases in the PTP family (Figure 6D). As anticipated, intact cell oxygen consumption can be involved in cellular processes other than signal trans- rates were much slower in LacZ-expressing KO MEFs, duction. It will be of interest to determine whether some of the compared to that of Flox cells. Wild-type PTPMT1 partially less characterized PTP family members target unorthodox reversed the inhibition of mitochondrial respiration in Ptpmt1 substrates like PTPMT1. null cells, whereas CS mutants failed to rescue the phenotype. The enzyme reaction catalyzed by the PGP phosphatase was Together, our results indicate that the catalytic activity of first characterized by Kennedy et al. in 1963 using rat and

696 Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

Figure 6. Wild-Type but Not an Active Site Mutant PTPMT1 Rescues Cardiolipin Deficiency (A) The expression of b-galactosidase (LacZ), wild- type PTPMT1 (MT1), or catalytically inactive PTPMT1 mutant (CS). (B) After labeling of MEFs with 32P-orthophos- phate, phospholipids were extracted and the percentage of cardiolipin (CL) among total phospholipids was determined by TLC analysis. Mean ± SEM from three independent ex- perimentsis shown. One-way ANOVA with a post hoc F test, ***p < 0.001, **p < 0.01. (C) The accumulation of PGP in Ptpmt1-deficient cells. For each sample, 50,000 cpm of 32P-labeled phospholipids were applied onto TLC plates and visualized by phosphoimager. (D) Oxygen consumption rates of intact MEFs cultured in regular medium. The basal rate, the state 4 rate (Oligo), and the maximal uncoupler stimulated rate (FCCP) are shown as described in Figure 2. Mean ± SEM from four replicates is shown. (E) Wild-type but not mutant PTPMT1 comple- ments the growth deficiency of yeast PGP phosphatase GEP4 null (gep4D) cells in a serial dilution spotting assay. The expression levels of FLAG-tagged Gep4 and PTPMT1 are indicated by western blot analysis. The level of Cdc2 is shown as a control. WT, wild-type strain; –, empty vector; SCD, synthetic complete medium with dextrose; EtBr, ethidium bromide. See also Figure S4 and Table S1.

and Carman, 1997). Why nature has utilized such divergent phosphatases to remove the phosphate from PGP is presently unclear, but it may reflect chicken liver mitochondria (Kiyasu et al., 1963). This reaction, fundamental differences among organisms to regulate cardioli- which is conserved in bacteria, is catalyzed by PgpA, PgpB, pin synthesis and possibly energy metabolism. and the recently identified PgpC, three bacterial enzymes that Substantial accumulation of PGP was observed in cells lacking display no homology to PTPMT1 (Icho and Raetz, 1983; Lu PTPMT1 (Figure 4B). Notably, PGP exists in cells at low abun- et al., 2011). Osman et al. recently reported that Gep4 dephos- dance (Kiyasu et al., 1963; MacDonald and McMurray, 1980). phorylates PGP in yeast (Osman et al., 2010). Gep4 is a member Little is known about the function of PGP in cells other than it is of the haloacid dehydrogenase superfamily which contains an intermediate in the cardiolipin biosynthetic pathway. a DxDx(T/V) motif in the active site and utilizes a unique Asp- However, we found that depletion of PTPMT1 not only inhibited based catalytic mechanism (Kerk et al., 2008). Gep4 orthologs cell growth (Figures S2A–S2C), but also led to an increase in exist in fungi and plant, but are not found in higher eukaryotes, cellular ATP content and glycolysis (Figures 2B and 2D). This providing little insight into the identity of the mammalian phenomenon was observed even before the onset of cardiolipin PGP phosphatase. On the other hand, PTPMT1 is one of the deficiency (Figures S3C and S3D) and inhibition of mitochondrial most widely distributed and highly conserved PTPs, with respiratory capacity (Figure 2C), implying a possible role for PGP over 60 orthologs identified in Animalia, Plantae, Protista, or other intermediates of the pathway in the regulation of aerobic and Eubacteria (Pagliarini et al., 2004), which raises the and anaerobic metabolism. It will be interesting to determine possibility that PTPMT1 orthologs also utilize PGP as their whether deletion of Ptpmt1 alters the activity of glycolytic substrates. Although a PTPMT1-like sequence is not present in enzymes or the abundance of glycolytic intermediates as a result Saccharomyces cerevisiae, our results demonstrate that of PGP accumulation. Indeed, several enzymes of the glycolytic mammalian PTPMT1 serves as a functional equivalent of Gep4 pathway have been found associated with mitochondrial in yeast (Figure 6E). PTPMT1, Gep4, PgpA, PgpB, and PgpC membranes, such as hexokinase on the outer membrane (Wil- are structurally and evolutionarily distinct enzymes with distinct son, 2003) and FAD-linked glycerol-3-phosphate dehydroge- regulatory mechanisms for their activities (Table S2)(Kerk nase on the inner membrane (Klingenberg, 1970). In addition, it et al., 2008; Kumaran et al., 2006; Pagliarini et al., 2004; Stukey is likely that the rate of energy utilization is diminished in the

Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. 697 Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

slower growing KO cells, which in turn may contribute to the over- described (Denu et al., 1995; Maehama and Dixon, 1998; Pagliarini et al., all increase of cellular ATP content. Further investigation of glyco- 2004; Rardin et al., 2008; Worby et al., 2006). 1,2-dioleoyl-sn-glycerol-3- lytic intermediates, glycolytic rates, and ATP consumption in both cytidine diphosphate (ammonium salt) and 1,2-dioleoyl-sn-glycero-3-phos- pho-(10-rac-glycerol) (sodium salt) were purchased from Avanti Polar Lipids Flox and KO MEFs is clearly warranted and is ongoing in our lab. (Alabaster, AL), -a-glycerol phosphate dipotassium salt was from Sigma, –/– DL Our results found that Ptpmt1 embryos implanted but died and sn-[U-14C] glycerol 3-phosphate disodium salt was from American Radio- prior to E8.5, suggesting an essential role for the Ptpmt1 gene labeled Chemicals (Saint Louis, MO). during early embryonic development. However, the cause of embryonic lethality is not fully understood. It is noteworthy that Mice the demand for mitochondria varies at different embryonic Generation of Ptpmt1 knockout mice is described in the Supplemental Exper- imental Procedures. All procedures were performed in accordance with the developmental stages. In the early embryo, mitochondria are National Institutes of Health (NIH) Guide for the Care and Use of Laboratory maternally inherited and their population remains relatively Animals and approved by the Institutional Animal Care and Use Committee constant until after implantation, when rapid mitochondrial of University of California San Diego. biogenesis and mitochondrial DNA replication resume in response to the increasing demand for energy production during Derivation and Adenoviral Infection of MEFs +/flox organogenesis (Cerritelli et al., 2003; Van Blerkom, 2009; Hance MEFs from E13.5 embryos from intercrossed of Ptpmt1 mice were et al., 2005; Larsson et al., 1998). Therefore, it is possible that isolated, cultured, genotyped, and immortalized. For transduction of Ptpmt1flox/flox fibroblasts with the Cre recombinase, cells were infected with PTPMT1 is required for the increased mitochondrial biogenesis adenoviruses encoding either GFP or GFP-Cre (Gene transfer vector core, that occurs in postimplantation embryos. University of Iowa). Adenoviral vectors containing untagged wild-type or Identifying the mammalian PGP phosphatase is of interest not C132S mutant of murine PTPMT1 were constructed into pAd/CMV/V5-DEST only because it is a missing step of lipid biosynthesis, but also vector (Invitrogen), and adenoviruses were generated and purified. pAd/ due to its potential implication in many cardiolipin deficiency- CMV/V5-GW/lacZ (Invitrogen) was used as the control. related diseases, such as ischemia/reperfusion injury, heart failure, aging, and neurodegeneration (Chicco and Sparagna, Western Blot and Immunofluorescence Analysis The western blots were performed with antibodies against PTPMT1 (Pagliarini 2007; Houtkooper and Vaz, 2008). It is noteworthy that the reac- et al., 2005), VDAC1 (Calbiochem), FLAG M2 (Sigma), and Cdc2 (Santa Cruz). tive cysteine in PTP family phosphatases is highly susceptible to For immunofluorescent staining, MEFs were plated on glass coverslips, fixed oxidative stress (Tonks, 2005). PTPMT1 is tethered to the inner with 3.7% formaldehyde and then incubated with anti-cytochrome c (BD mitochondrial membrane, where its activity could be regulated Biosciences) antibody and Alexa Fluor 488 goat anti-mouse-conjugated 0 by mitochondrial reactive oxygen species (ROS). Loss of two secondary antibody (Molecular Probes). Nuclei were stained with 4 ,6-diami- copies of the Ptpmt1 gene in all tissues is embryonic lethal; dino-2-phenylindole (DAPI, Molecular Probes). however, a single copy of a functional Ptpmt1 gene could result Liquid Chromatography/Tandem Mass Spectrometry Analysis in an organism being highly susceptible to episodes of oxidative Total cellular phospholipids were extracted (Bligh and Dyer, 1959) and quan- stress. Furthermore, loss of cardiolipin content in mitochondria tified with normal phase Liquid Chromatography coupled to a QSTAR XL has been well documented in cardiac ischemia/reperfusion tandem mass spectrometer (Applied Biosystems, Foster City, CA) equipped injury (Chicco and Sparagna, 2007), but the underlying molecular with an electrospray source (see the Supplemental Experimental Procedures mechanism is poorly understood. Hence, our identification of for details). Data acquisition and analysis were performed with the instru- a potentially redox-sensitive enzyme in the cardiolipin synthesis ment’s Analyst QS software. pathway raises the question of whether inactivation of PTPMT1 PGP Phosphatase Assay by mitochondrial ROS contributes to cardiolipin deficiency under 14C-labeled PGP was synthesized as described (Dowhan, 1992). The PGP those conditions. phosphatase activities of recombinant PTPMT1, PTEN, Laforin, DSP21, and In conclusion, our work has identified the substrate of PTPMT1 VHR were determined by TLC or malachite green analysis (see the Supple- by comparing phospholipid analyses coupled with mass spec- mental Experimental Procedures). trometry of lipids extracted from wild-type versus knockout cells. Phospholipid Quantification These analyses provided us with the key to understanding the Cellular phospholipids were extracted (Bligh and Dyer, 1959) and separated by biological functions of PTPMT1 within mitochondria. This TLC with Chloroform/Methanol/ammonium hydroxide/water (130:75:2:6). For approach can be applied to understanding the molecular mech- the separation of PGP, ethylacetate/2-propanol/ethanol/6% ammonia anisms underlying other alterations in phospholipid metabolism. (3:9:3:9) was used as TLC solvent (Osman et al., 2010). Individual species Finally, in addition to the discovery of a lipid biosynthetic were identified by comigration of nonradiolabeled standards. Their relative enzyme, we demonstrated that PTPMT1 functions as an impor- intensities were analyzed with the Typhoon 9410 Variable Mode Imager and tant regulator of mitochondrial physiology, which may provide ImageQuant TL software (GE Healthcare Life Science). perspectives into the etiology of many mitochondria defi- Mitochondrial Functional Analyses ciency-related diseases. Oxygen consumption (OCR) and extracellular acidification rates (ECAR) were measured with a Seahorse Extracellular Flux Analyzer XF24 (Seahorse Biosci- EXPERIMENTAL PROCEDURES ence, North Billerica, MA) (Wu et al., 2007). Forty thousand cells were plated into each well prior to the assay. OCR and ECAR were normalized by the Cell Culture and Reagents DNA content. (See the Supplemental Experimental Procedures for details.) MEFs and C2C12 mouse myoblast cells were grown in Dulbecco’s modified The data were presented as the means ± standard error of the mean (SEM) Eagle’s medium (DMEM) supplemented with 10% FBS, 100 units/ml penicillin, for four replicates. Qualitatively similar results were obtained in three separate 100 mg/ml streptomycin, and 2 mM glutamine and were cultured at 37Cin experiments. Differences in rates were analyzed by one-way ANOVA with a humidified incubator with 5% CO2. The expression and purification of a post hoc F test. For measurement of cellular ATP content, MEFs were plated recombinant PTPMT1, PTEN, laforin, DSP21, and VHR were previously and grown overnight in regular or low glucose (3 mM) media, and whole-cell

698 Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

ATP content was measured by CellTiter-Glo Luminescent Cell Viability/ATP REFERENCES Assay kit (Promega) and normalized by DNA content. Statistical analysis was performed via one-way ANOVA with a post hoc F test. Cellular lactate Alonso, A., Sasin, J., Bottini, N., Friedberg, I., Friedberg, I., Osterman, A., was measured by a colorometric assay kit (Abcam) and normalized by total Godzik, A., Hunter, T., Dixon, J., and Mustelin, T. (2004). Protein tyrosine protein amount. Differences between Flox and KO MEFs were examined for phosphatases in the human genome. Cell 117, 699–711. statistical significance with a Student’s t test. Bligh, E.G., and Dyer, W.J. (1959). A rapid method of total lipid extraction and purification. Can. J. Biochem. Physiol. 37, 911–917. Blue-Native PAGE Brachmann, C.B., Davies, A., Cost, G.J., Caputo, E., Li, J., Hieter, P., and Subconfluent MEFs were harvested and crude mitochondria were isolated by Boeke, J.D. (1998). Designer deletion strains derived from Saccharomyces differential centrifugation (Pallotti and Lenaz, 2001). Twenty micrograms of cerevisiae S288C: a useful set of strains and plasmids for PCR-mediated mitochondria were solubilized on ice with 1% b-dodecylmaltoside (Invitrogen) gene disruption and other applications. Yeast 14, 115–132. for 30 min and further separated by 4%–16% Native PAGE gel (Invitrogen) (Scha¨ gger and von Jagow, 1991; McKenzie et al., 2007). For assessment of Cerritelli, S.M., Frolova, E.G., Feng, C., Grinberg, A., Love, P.E., and Crouch, the activity of complex I, the gel was incubated in 2 mM Tris/HCl (pH 7.4) R.J. (2003). Failure to produce mitochondrial DNA results in embryonic lethality containing 0.1 mg/ml NADH and 2.5 mg/ml nitrotetrazolium blue (Zerbetto in Rnaseh1 null mice. Mol. Cell 11, 807–815. et al., 1997). Subsequently, the gel was incubated in 2 mM Tris/HCl (pH 7.4) Chicco, A.J., and Sparagna, G.C. (2007). Role of cardiolipin alterations in mito- buffer in the presence of 20 mM sodium succinate, 2 mM phenazine methosul- chondrial dysfunction and disease. Am. J. Physiol. Cell Physiol. 292, C33–C44. fate, and 2.5 mg/ml nitrotetrazolium blue for the complex II in-gel activity assay Daum, G. (1985). Lipids of mitochondria. Biochim. Biophys. Acta 822, 1–42. (Wittig et al., 2007). Denu, J.M., Zhou, G., Wu, L., Zhao, R., Yuvaniyama, J., Saper, M.A., and Transmission Electron Microscopy and Morphometric Analysis Dixon, J.E. (1995). The purification and characterization of a human dual- Cells were grown on 35 mm glass bottom culture dishes (MatTek, Ashland, specific protein tyrosine phosphatase. J. Biol. Chem. 270, 3796–3803. MA). Sample preparation and conventional EM analysis were carried out as Dowhan, W. (1992). Phosphatidylglycerophosphate synthase from described in the Supplemental Experimental Procedures. Mitochondrial Escherichia coli. Methods Enzymol. 209, 313–321. lengths and cristae/OMM surface area ratio were measured using ImageJ Hance, N., Ekstrand, M.I., and Trifunovic, A. (2005). Mitochondrial DNA (NIH). Differences between Flox and KO MEFs were examined for statistical polymerase gamma is essential for mammalian embryogenesis. Hum. Mol. significance with a Student’s t test. Genet. 14, 1775–1783.

Yeast Complementation Assays Hatch, G.M. (1998). Cardiolipin: biosynthesis, remodeling and trafficking in the Yeast strains used in this study are derivatives of S288C (Brachmann et al., heart and mammalian cells (Review). Int. J. Mol. Med. 1, 33–41. 1998). GEP4 knockout strains are obtained from Open Biosystems and verified Houtkooper, R.H., and Vaz, F.M. (2008). Cardiolipin, the heart of mitochondrial by gene-specific PCR primer pairs. The plasmids for complementation metabolism. Cell. Mol. Life Sci. 65, 2493–2506. assays in yeast were constructed by inserting the PCR products of GEP4, Icho, T., and Raetz, C.R. (1983). Multiple genes for membrane-bound phos- M. musculus PTPMT1 (wild-type), and mutant PTPMT1 (C132S) with phatases in Escherichia coli and their action on phospholipid precursors. D C-terminal FLAG tag into pRS415-GPD vector (ATCC). Subsequently, gep4 J. Bacteriol. 153, 722–730. cells were transformed with plasmids containing GEP4, PTPMT1, or empty vector. Cells were grown in the appropriate selective medium at 30Cto Janitor, M., and Subı´k, J. (1993). Molecular cloning of the PEL1 gene of mid-log phase and counted, followed by 5-fold serial dilution. Each dilution Saccharomyces cerevisiae that is essential for the viability of petite mutants. was spotted onto either YPD plates or Synthetic Completed Medium with Curr. Genet. 24, 307–312. Dextrose (SCD) in the presence of 25 mg/ml ethidium bromide, and cultured Kerk, D., Templeton, G., and Moorhead, G.B. (2008). Evolutionary radiation at the indicated temperature for 2 days. pattern of novel protein phosphatases revealed by analysis of protein data Full experimental procedures and associated references are enclosed in the from the completely sequenced genomes of humans, green algae, and higher Supplemental Experimental Procedures. plants. Plant Physiol. 146, 351–367. Kiyasu, J.Y., Pieringer, R.A., Paulus, H., and Kennedy, E.P. (1963). The biosyn- SUPPLEMENTAL INFORMATION thesis of phosphatidylglycerol. J. Biol. Chem. 238, 2293–2298. Kletzien, R.F., and Perdue, J.F. (1975). Induction of sugar transport in chick Supplemental Information includes Supplemental Experimental Procedures, embryo fibroblasts by hexose starvation. Evidence for transcriptional regula- four figures, and two tables and can be found with this article online at tion of transport. J. Biol. Chem. 250, 593–600. doi:10.1016/j.cmet.2011.04.007. Klingenberg, M. (1970). Localization of the glycerol-phosphate dehydroge- nase in the outer phase of the mitochondrial inner membrane. Eur. J. ACKNOWLEDGMENTS Biochem. 13, 247–252. We thank Ju Chen and Matthew Rardin for strains and reagents and David Kumaran, D., Bonanno, J.B., Burley, S.K., and Swaminathan, S. (2006). Pagliarini, Fred Robinson, Youngjun Kim, Hongqiang Cheng, Gloria Reyes, Crystal structure of phosphatidylglycerophosphatase (PGPase), a putative Gregory Taylor, Matthew Gentry, Mikhail Bogdanov, and members of the membrane-bound lipid phosphatase, reveals a novel binuclear metal binding Dixon laboratory for technical assistance. We thank Claudia Kent, Vincent site and two ‘‘proton wires’’. Proteins 64, 851–862. Tagliabracci, and Seema Matto for critically reading the manuscript. This Larsson, N.G., Wang, J., Wilhelmsson, H., Oldfors, A., Rustin, P., Lewandoski, work was supported by grants from the National Institutes of Health M., Barsh, G.S., and Clayton, D.A. (1998). Mitochondrial transcription factor A (DK18024 and DK18849 to J.E.D., DK054441 to A.N.M., GM56389 to W.D., is necessary for mtDNA maintenance and embryogenesis in mice. Nat. Genet. GM069338 to the LIPID MAPS Large Scale Collaborative Grant, and 18, 231–236. RR04050 to the National Center for Microscopy and Imaging Research) and the Larry L. Hillblom Foundation postdoctoral fellowship to J. Z. Lu, Y.H., Guan, Z., Zhao, J., and Raetz, C.R. (2011). Three phosphatidylgly- cerol-phosphate phosphatases in the inner membrane of Escherichia coli. Received: December 3, 2010 J. Biol. Chem. 286, 5506–5518. Revised: February 28, 2011 MacDonald, P.M., and McMurray, W.C. (1980). Partial purification and proper- Accepted: April 6, 2011 ties of mammalian phosphatidylglycerophosphatase. Biochim. Biophys. Acta Published: June 7, 2011 620, 80–89.

Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc. 699 Cell Metabolism PTPMT1 Is the Mammalian PGP Phosphatase

Maehama, T., and Dixon, J.E. (1998). The tumor suppressor, PTEN/MMAC1, sional light and electron microscopy reveals transformation of mitochondria dephosphorylates the lipid second messenger, phosphatidylinositol 3,4,5-tri- during apoptosis. Nat. Cell Biol. 9, 1057–1065. sphosphate. J. Biol. Chem. 273, 13375–13378. Tonks, N.K. (2005). Redox redux: revisiting PTPs and the control of cell Manning, G., Whyte, D.B., Martinez, R., Hunter, T., and Sudarsanam, S. (2002). signaling. Cell 121, 667–670. The complement of the human genome. Science 298, 1912– Tonks, N.K. (2006). Protein tyrosine phosphatases: from genes, to function, 1934. to disease. Nat. Rev. Mol. Cell Biol. 7, 833–846. McKenzie, M., Lazarou, M., Thorburn, D.R., and Ryan, M.T. (2007). Analysis of Tonks, N.K., and Neel, B.G. (2001). Combinatorial control of the specificity of mitochondrial subunit assembly into respiratory chain complexes using Blue protein tyrosine phosphatases. Curr. Opin. Cell Biol. 13, 182–195. Native polyacrylamide gel electrophoresis. Anal. Biochem. 364, 128–137. Van Blerkom, J. (2009). Mitochondria in early mammalian development. Newmeyer, D.D., and Ferguson-Miller, S. (2003). Mitochondria: releasing Semin. Cell Dev. Biol. 20, 354–364. power for life and unleashing the machineries of death. Cell 112, 481–490. Vance, D., and Vance, J.E. (2008). Phospholipid biosynthesis in eukaryotes. In Osman, C., Haag, M., Wieland, F.T., Bru¨ gger, B., and Langer, T. (2010). A mito- Biochemistry of Lipids, Lipoproteins and Membranes, D.E. Vance and J.E. chondrial phosphatase required for cardiolipin biosynthesis: the PGP phos- Vance, eds. (Amsterdam: Elsevier), pp. 213–244. phatase Gep4. EMBO J. 29, 1976–1987. Voet, D., and Voet, J.G. (2004). Electron transport and oxidative phosphoryla- Pagliarini, D.J., Worby, C.A., and Dixon, J.E. (2004). A PTEN-like phosphatase tion. In Biochemistry, Volume 1 (Hoboken, NJ: Wiley), pp. 797–842. with a novel substrate specificity. J. Biol. Chem. 279, 38590–38596. Pagliarini, D.J., Wiley, S.E., Kimple, M.E., Dixon, J.R., Kelly, P., Worby, C.A., Warburg, O. (1956). On the origin of cancer cells. Science 123, 309–314. Casey, P.J., and Dixon, J.E. (2005). Involvement of a mitochondrial phospha- Whitman, M., Downes, C.P., Keeler, M., Keller, T., and Cantley, L. (1988). Type tase in the regulation of ATP production and insulin secretion in pancreatic I phosphatidylinositol kinase makes a novel inositol phospholipid, phosphati- beta cells. Mol. Cell 19, 197–207. dylinositol-3-phosphate. Nature 332, 644–646. Pallotti, F., and Lenaz, G. (2001). Isolation and subfractionation of mitochon- Wilson, J.E. (2003). Isozymes of mammalian hexokinase: structure, subcellular dria from animal cells and tissue culture lines. Methods Cell Biol. 65, 1–35. localization and metabolic function. J. Exp. Biol. 206, 2049–2057. Parce, J.W., Owicki, J.C., Kercso, K.M., Sigal, G.B., Wada, H.G., Muir, V.C., Wittig, I., Karas, M., and Scha¨ gger, H. (2007). High resolution clear native elec- Bousse, L.J., Ross, K.L., Sikic, B.I., and McConnell, H.M. (1989). Detection trophoresis for in-gel functional assays and fluorescence studies of membrane of cell-affecting agents with a silicon biosensor. Science 246, 243–247. protein complexes. Mol. Cell. Proteomics 6, 1215–1225. Rardin, M.J., Wiley, S.E., Murphy, A.N., Pagliarini, D.J., and Dixon, J.E. (2008). Worby, C.A., Gentry, M.S., and Dixon, J.E. (2006). Laforin, a dual specificity Dual specificity phosphatases 18 and 21 target to opposing sides of the mito- phosphatase that dephosphorylates complex . J. Biol. Chem. chondrial inner membrane. J. Biol. Chem. 283, 15440–15450. 281, 30412–30418. Rock, C.O. (2008). Fatty acid and phospholipid metabolism in prokaryotes. In Wu, M., Neilson, A., Swift, A.L., Moran, R., Tamagnine, J., Parslow, D., Biochemistry of Lipids, Lipoproteins and Membranes, D.E. Vance and J.E. Armistead, S., Lemire, K., Orrell, J., Teich, J., et al. (2007). Multiparameter Vance, eds. (Amsterdam: Elsevier), pp. 59–96. metabolic analysis reveals a close link between attenuated mitochondrial Scha¨ gger, H., and von Jagow, G. (1991). Blue native electrophoresis for isola- bioenergetic function and enhanced glycolysis dependency in human tumor tion of membrane protein complexes in enzymatically active form. Anal. cells. Am. J. Physiol. Cell Physiol. 292, C125–C136. Biochem. 199, 223–231. Yan, L.J., and Forster, M.J. (2009). Resolving mitochondrial protein complexes Schlame, M. (2008). Cardiolipin synthesis for the assembly of bacterial and using nongradient blue native polyacrylamide gel electrophoresis. Anal. mitochondrial membranes. J. Lipid Res. 49, 1607–1620. Biochem. 389, 143–149. Schlame, M., Rua, D., and Greenberg, M.L. (2000). The biosynthesis and func- Zerbetto, E., Vergani, L., and Dabbeni-Sala, F. (1997). Quantification of muscle tional role of cardiolipin. Prog. Lipid Res. 39, 257–288. mitochondrial oxidative phosphorylation enzymes via histochemical staining Stukey, J., and Carman, G.M. (1997). Identification of a novel phosphatase of blue native polyacrylamide gels. Electrophoresis 18, 2059–2064. sequence motif. Protein Sci. 6, 469–472. Zhong, Q., Gvozdenovic-Jeremic, J., Webster, P., Zhou, J., and Greenberg, Sun, M.G., Williams, J., Munoz-Pinedo, C., Perkins, G.A., Brown, J.M., M.L. (2005). Loss of function of KRE5 suppresses temperature sensitivity of Ellisman, M.H., Green, D.R., and Frey, T.G. (2007). Correlated three-dimen- mutants lacking mitochondrial anionic lipids. Mol. Biol. Cell 16, 665–675.

700 Cell Metabolism 13, 690–700, June 8, 2011 ª2011 Elsevier Inc.