G-protein Palmitoyltransferase Activity Is Enriched in Plasma Membranes*

Heterotrimeric G proteins are covalently modified by lipids. Myristoylation of G-protein (cid:97) subunits and prenylation of (cid:103) subunits are stable modifications. In con-trast, palmitoylation of (cid:97) subunits is dynamic and thus has the potential for regulating protein function. Indeed, receptor activation of G s increases palmitate turnover on the (cid:97) subunit, presumably by stimulating deacylation. The enzymes that catalyze reversible palmitoylation of G-protein (cid:97) subunits have not been characterized. Here we report the identification of a palmitoyl-CoA:protein S -palmitoyltransferase activity that acylates G-protein (cid:97) subunits in vitro . Palmitoyltransferase activity is membrane-associated and requires detergent for solubi-lization. The preferred G-protein substrate for the enzyme activity is the (cid:97) subunit in the context of the heterotrimer. Both myristoylated and nonmyristoylated G-protein (cid:97) subunits are recognized as substrates. The palmitoyltransferase activity demonstrates a modest preference for palmitoyl-CoA over other fatty acyl-CoA substrates. Palmitoyltransferase activity is high in plasma membrane and present at low or undetectable levels in Golgi, endoplasmic reticulum, and mitochondria of rat liver. The subcellular localization of this enzyme activity is consistent with a role for regulated cycles of acylation and deacylation accompanying activation of G-protein signal transduction pathways. Signal-transducing G proteins are located on the cytoplasmic surface

Signal-transducing G proteins are located on the cytoplasmic surface of the plasma membrane where they couple receptors to intracellular effectors. Membrane association of heterotrimeric G proteins is facilitated by the covalent addition of lipids to G-protein subunit polypeptides (reviewed in Refs. 1 and 2). The carboxyl-terminal cysteine residue is prenylated and methylated on G-protein ␥ subunits. Prenylation is not required for ␤␥ complex formation, but facilitates subunit and effector interactions. G-protein ␣ subunits are fatty-acylated. Members of the mammalian G i␣ family (G i␣ , G o␣ , G z␣ , and transducin ␣) contain amide-linked myristate at the amino terminus. Transdu-cin ␣ (T ␣ ) 1 is modified heterogeneously at this site by C14:0, C14:1, C14:2, or C12:0 fatty acids. Myristoylation also facilitates subunit and effector interactions. Thioester-linked palmitate is found on most mammalian G-protein ␣ subunits. G s␣ , G q␣ , and G 12␣ , which are not N-myristoylated, are palmitoylated at one or more cysteine residues near the amino terminus. G i␣ , G o␣ , and G z␣ are palmitoylated at a cysteine residue (Cys-3) adjacent to the amino-terminal myristoylated glycine.
Reversible post-translational modification is a well-characterized mechanism for regulating protein activity. Regulatory cycles of acylation and deacylation of G-protein ␣ subunits may fit this paradigm. Dynamic acylation of G proteins has been characterized best for G s␣ . Studies of agonist-induced turnover of palmitate on G s␣ are consistent with a model where G s␣ is deacylated upon activation and dissociation from ␤␥ subunits (3)(4)(5). Deacylation may be accompanied by release of G s␣ from membranes, suggesting a potential role for this process in desensitization of G s -coupled pathways (5).
The enzymes responsible for addition and removal of palmitate from G-protein ␣ subunits have not been identified. Gprotein ␣ subunits can be deacylated in vitro by a protein palmitoyl thioesterase that has recently been purified (6). However, subsequent cloning of the cDNA encoding protein palmitoyl thioesterase and characterization of the gene product revealed that protein palmitoyl thioesterase is a secreted enzyme (7), and thus is not likely to be a physiological regulator of G-protein palmitoylation. Palmitoyltransferase (PAT) activities have been identified using a number of proteins known to be palmitoylated as substrates, including viral glycoproteins (8), p21 ras (9), and p59 fyn (10), but it is not known whether the substrate specificity of these activities extends to G-protein ␣ subunits. Purification to homogeneity and molecular cloning of palmitoyltransferase activities have not been achieved to date. Here we report the initial characterization of a palmitoyl-CoA: protein S-palmitoyltransferase activity highly enriched in plasma membranes that acylates both myristoylated and nonmyristoylated G-protein ␣ subunits in vitro. G-protein Substrates-Myristoylated recombinant G-protein ␣ subunits were purified after co-expression in Escherichia coli with S. cerevisiae N-myristoyltransferase (12). The C3A mutant of G i␣1 was constructed using site-directed mutagenesis of the rat G i␣1 coding sequence in plasmid pQE60 (Qiagen). The amino-terminal 387 nucleotides of G i␣1 were amplified in a polymerase chain reaction using a mutagenic oligonucleotide (5Ј-GAAATTAACCATGGGCGCCACACTGAGC-3Ј) as the forward primer and an oligonucleotide carboxyl-terminal to an internal SacII site as the reverse primer (5Ј-CGAGCTCCGCGGTCAT-3Ј). The resulting PCR product was digested with NcoI and SacII and subcloned into the parental plasmid replacing the wild-type amino-terminal sequence. DNA sequence analysis (13) was performed to confirm the sequence of the amplified region of DNA. Myristoylated rG i␣1 (C3A) was expressed and purified using the same procedures used for the wild type protein. rG i␣1 (nonmyristoylated) and rG s␣ were expressed and purified from E. coli (14). The bovine cDNA encoding the long form (52 kDa) of G s␣ (15) was subcloned into pQE60 as an NcoI/HindIII fragment, and the resulting plasmid was transformed into JM109 bacteria. Activity of recombinant G-protein ␣ subunit preparations was assessed by GTP␥S binding (14).
Assay for PAT Activity-A source of PAT (10 l) was diluted in 50 mM Tris (pH 8), 1 mM EDTA, 1 mM DTT, and 0.1% Triton X-100 and mixed with 10 l of G-protein ␣ (2 g) and ␤␥ (2 g) subunits that had been diluted in 50 mM NaHepes (pH 8.0), 1 mM EDTA, and 1 mM DTT before addition. The reaction was started by the addition of 30 l of [ 3 H]palmitoyl-CoA diluted in MES (pH 6.4) and DTT such that the final concentrations in the assay were 0.4 M, 100 mM, and 1 mM, respectively. Assays were incubated for 5-15 min at 30°C and terminated by the addition of 0.5 ml of 4% SDS. Proteins were precipitated by addition of 0.5 ml of 30% trichloroacetic acid, incubated at room temperature for at least 15 min, and collected on glass fiber filters. The filters were washed twice with 2-ml aliquots of 6% trichloroacetic acid containing 2% SDS, followed by four washes of 1.5 ml of 6% trichloroacetic acid. The filters were dried, and bound radioactivity was quantitated using liquid scintillation counting. Experimental data are representative of at least three individual experiments. Error bars represent the standard error based on samples assayed in triplicate. For analysis of samples by SDS-PAGE, reactions were terminated by the addition of 5-fold concentrated SDS sample buffer and boiled for 1 min before resolution on 13% SDS-PAGE. Gels were soaked for 30 min in Amplify (Amersham), dried, and exposed to film (Kodak XAR-5).
Preparation of Bovine Brain Extracts-Crude bovine brain membranes were prepared as described (18) and stored at Ϫ80°C. To prepare washed membranes, an equal volume of 200 mM Na 2 CO 3 (pH 11.0) was added to crude membranes and stirred for 30 min. Carbonateextracted membranes were collected by centrifugation for 30 min at 100,000 ϫ g. The pellet was resuspended in 100 mM Na 2 CO 3 (pH 11.0), stirred, and subjected to centrifugation for 30 min at 100,000 ϫ g. The pellet (washed membranes) was resuspended in 50 mM Tris (pH 8.0), 1 mM EDTA, and 1 mM DTT plus protease inhibitors. Washed membranes were frozen in liquid nitrogen and stored at Ϫ80°C at a protein concentration of approximately 20 mg/ml. Protein assays were per-formed by the precipitation procedure of Brown et al. (19) using the Pierce BCA kit.
Preparation of Detergent Extracts-Washed membranes were diluted to a final protein concentration of 10 mg/ml in a solution of 50 mM Tris (pH 8.0), 1 mM EDTA, 1 mM DTT, 400 mM (NH 4 ) 2 SO 4 , 10% glycerol, 2% BigChap (Calbiochem) plus protease inhibitors. The mixture was homogenized in a Potter-Elvehjem homogenizer, stirred for 60 min, and subjected to centrifugation for 60 min at 100,000 ϫ g. The supernatant (detergent extract) was collected and either used immediately or frozen in liquid nitrogen. PAT activity in the extract was usually equal to or greater than the starting activity in the homogenate.
Preparation of Partially Purified PAT-A detergent extract of bovine brain membranes was applied to a Sephacryl S-300 (Pharmacia) gel filtration column (2.6 ϫ 32 cm) and developed in 20 mM Tris (pH 7.3), 1 mM EDTA, 1 mM DTT, 0.1% Triton X-100, 10% glycerol, and 150 mM KCl (Buffer A). Fractions of 2 ml were collected and assayed for PAT activity. PAT activity eluted slightly before ␤-amylase (200,000 Da) in the included volume of the S-300 column. Fractions containing PAT activity were pooled and loaded onto a 2-ml Q-Sepharose (Pharmacia) column that had been equilibrated in Buffer A. The column was washed with 3 ml of Buffer A, and PAT activity was eluted with a 30-ml linear gradient of KCl (150 -1000 mM) in Buffer A. PAT activity eluted at a salt concentration between 200 and 400 mM KCl. Fractions of 1 ml were collected and assayed for PAT activity. Fractions containing PAT activity were pooled and stored at 4°C. This procedure resulted in approximately a 10-fold purification of PAT activity.

RESULTS AND DISCUSSION
Identification of an Enzymatic Activity That Acylates G-protein ␣ Subunits-To develop an assay for palmitoyltransferase (PAT) activity, a suitable protein substrate was required. Because most G-protein ␣ subunits can be purified easily after expression in E. coli, we chose to use the recombinant ␣ subunit as a substrate in the assay. G i␣ , G o␣ , and G z␣ are palmitoylated at Cys-3, adjacent to the myristoylated glycine (4,25,26). Although myristoylation is not required for palmitoylation (27), it appears to increase palmitoylation of G i␣ family members in vivo (4,25). Accordingly, we purified myristoylated recombinant ␣ subunits from E. coli that co-express N-myristoyltransferase with its G-protein substrate (12,28). This system provided an abundant source of substrate with an appropriately modified amino terminus to use for in vitro acylation.
PAT activity was detected in membranes and detergent extracts. Bovine brain membranes were incubated with myristoylated rG i␣1 , bovine brain ␤␥ subunits (see below), and [ 3 H]palmitoyl-CoA. Palmitate incorporation into myristoylated rG i␣1 was assessed by SDS-PAGE and fluorography. Palmitate was incorporated into myristoylated rG i␣1 incubated with bovine brain membranes, suggesting that a palmitoyltransferase activity was present (Fig. 1A, lane 1). 2 Similar results were obtained when the assay was terminated by precipitation and protein-bound radioactivity was quantitated by liquid scintillation counting (data not shown). Far less PAT activity, if any, was detected in cytosol (Fig. 1A, lane 2), nor was it released by treatment of membranes with high pH or high concentrations of salt (data not shown). However, palmitoyltransferase activity was solubilized by detergent (Fig. 1A, lane 3). The C3A 2 To confirm that the radioactivity incorporated into myristoylated rG i␣1 in vitro was palmitate, the protein was excised from the gel and treated with base to cleave thioester bonds. Hydrolysates were resolved by reversed phase HPLC, and fatty acids were identified by co-elution with fatty acid standards. The radioactivity released from myristoylated rG i␣1 was identified as palmitic acid (data not shown). These data also demonstrate that in vitro palmitoylation of the protein is alkalinesensitive, consistent with a thioester linkage. mutant of myristoylated rG i␣1 was not a substrate in the assay (Fig. 1A, lane 4), indicating that in vitro acylation is occurring at the appropriate cysteine residue in the protein.
Experiments were performed to determine if in vitro acylation of G-protein substrates was indeed enzymatic. Uncatalyzed acylation of substrate proteins incubated with palmitoyl-CoA in the absence of a source of enzyme was first reported for rhodopsin (29) and myelin proteolipid protein (30). Low levels of palmitate were incorporated into myristoylated rG i␣1 in the absence of enzyme (Fig. 1A, lane 6). Uncatalyzed acylation was more pronounced if detergent was omitted from the assay or when the protein was incubated with higher concentrations of palmitoyl-CoA (data not shown). To confirm that the activity observed in a detergent extract of membranes required a protein component, we treated the extract with trypsin. Protease treatment reduced PAT activity to background levels (Fig. 1B). Soybean trypsin inhibitor abolished the effect of trypsin. PAT was also inactivated by boiling or after treatment with SDS, a denaturing detergent (Fig. 1B). PAT activity is time-dependent; in the presence of membranes, the assay is linear to 10 min (data not shown). The activity demonstrated strict concentration dependence on the source of enzyme, G protein, and palmitoyl-CoA. At high concentrations of palmitoyl-CoA or G protein, the activity was saturable (data not shown). Taken together, these data demonstrate that in vitro acylation of myristoylated rG i␣1 is an enzymatic process.
G-protein Substrate Specificity-To analyze G-protein substrate specificity of PAT, a detergent extract of bovine brain membranes was incubated with various purified preparations of G-protein ␣ subunits in the presence or absence of ␤␥ subunits (Fig. 2). Incorporation of [ 3 H]palmitate was analyzed by SDS-PAGE and fluorography (A) or by liquid scintillation counting of protein-bound radioactivity (B). Comparison of panels A and B shows that the results obtained with both methods were essentially the same. The ␣ subunit in the context of the heterotrimer appeared to be the preferred substrate for PAT. The addition of stoichiometric amounts of ␤␥ subunits significantly increased the palmitate incorporated into myristoylated rG i␣1 (lanes 3 and 4). PAT activity toward rG i␣1 (lanes 5 and 6) and rG s␣ (lanes 7 and 8) was barely detectable in the absence of ␤␥ subunits. As expected, T ␣ was not acylated in vitro (lanes 9 and 10). Although T ␣ contains N-linked fatty acids at its amino terminus, it does not have cysteine residues near the amino terminus and is not palmitoylated in vivo (31,32).
Myristoylated rG i␣1 was a better substrate than nonmyris-FIG. 1. Enzymatic acylation of myristoylated rG i␣1 . Reactions were carried out as described under "Experimental Procedures." A, aliquots of 20 g of bovine brain membranes (Mb) (lane 1) or cytosol (Cyto) (lane 2) protein were assayed for PAT activity in the presence of myristoylated rG i␣ and bovine brain ␤␥ subunits. Bovine brain membranes were extracted with 2% BigChap, and an aliquot of the extract (XT) (5 g) was assayed in the presence (lane 3) or absence (lane 5) of myristoylated rG i␣ plus ␤␥ or myristoylated rG i␣ C3A plus ␤␥ (lane 4). The samples were subjected to SDS-PAGE and fluorography. Exposure time was 1 day. B, bovine brain detergent extract (4 g) was pretreated at 30°C for 1 h with 3.2 mg/ml trypsin (column 2) or 3.2 mg/ml trypsin plus 10 mg/ml soybean trypsin inhibitor (column 3). Trypsin digestion (column 2) was terminated by the addition of 10 mg/ml soybean trypsin inhibitor prior to the PAT assay. Detergent extract (4 g) was heattreated at 100°C for 1 min (column 4) or preincubated in 0.05% SDS (column 5) before the PAT reaction. PAT assays were performed using myristoylated rG i␣ and bovine brain ␤␥ subunits as substrate. Samples were subjected to filter binding and liquid scintillation counting.
FIG. 2. G-protein substrate specificity of PAT activity. G-protein ␣ subunits (2 g) were incubated with bovine brain detergent extract in the presence or absence of bovine brain ␤␥ subunits (2 g) (lanes 2, 4, 6, and 8) or T ␤␥ subunits purified from bovine retina (lane 10). G-protein ␣ subunits assayed were: myristoylated rG i␣ (lanes 3 and  4), nonmyristoylated rG i␣ (lanes 5 and 6), rG s␣ (lanes 7 and 8), and T ␣ purified from bovine retina (lanes 9 and 10). Assays were analyzed by SDS-PAGE and fluorography as described under "Experimental Procedures" and exposed to film for 1 day (A), or trichloroacetic acid-SDS precipitates of the reactions were collected on glass fiber filters and quantitated by liquid scintillation counting (B). toylated rG i␣1 (compare lanes 3 and 4 with lanes 5 and 6). The effect of myristoylation on PAT activity was also observed when rG i␣2 and myristoylated rG i␣2 were assayed (data not shown). PAT activity is not limited to myristoylated substrates. G s␣ is not a myristoylated protein, but is palmitoylated at Cys-3 (4). rG s␣ in the presence of ␤␥ subunits was palmitoylated in vitro, although not as efficiently as myristoylated rG i␣1 (Fig. 2B,  lanes 7 and 8). rG s␣ (C3A) was not a substrate for PAT activity in vitro (data not shown).
The substrate specificity of PAT activity in vitro correlates well with our understanding of G-protein palmitoylation in vivo. G s␣ , G i␣ , and G o␣ are substrates for PAT; the site of palmitoylation in vitro and in vivo appears to be Cys-3. Myristoylated G i␣1 is the optimal substrate for palmitoylation in vivo. In mammalian cells, expression of a mutant G i␣1 lacking the myristoylation site (G2A) results in a protein that is almost entirely cytosolic and has undetectable levels of [ 3 H]palmitate incorporation. If ␤␥ subunits are co-expressed with G i␣1 (G2A), a small fraction of G i␣1 (G2A) is found associated with membranes and a low level of palmitoylation is observed (27). In the in vitro assay, myristoylated ␣ subunits were better substrates than those lacking myristate. However, a moderate level of acylation of nonmyristoylated G i␣1 was observed in the presence of ␤␥ subunits. Thus, these characteristics are similar to what is observed in vivo.
The role of G-protein ␤␥ subunits in substrate affinity for PAT may be to provide a mechanism for substrate presentation to PAT. ␤␥ may facilitate targeting of the ␣ subunit to the membrane, allowing it to become acylated by PAT. Although prenylation of G protein ␥ subunits is required for membrane association of the ␤␥ complex (1, 2), it is not required to support in vitro acylation of myristoylated rG i␣1 . Mutation of the prenylated cysteine residue to serine (C68S) in the ␥2 subunit yields a nonprenylated ␥ that heterodimerizes with the ␤1 subunit (16). The mutant ␤␥ (␤␥C68S) binds to myristoylated rG i␣1 , forming a heterotrimer (33) that is acylated in vitro with efficiency similar to that of the wild type heterotrimer (data not shown). These data suggest that ␤␥ provides more than a hydrophobic anchor to bind to a membrane (or detergent micelle) containing PAT. The palmitoylation site is contained within the amino-terminal region of the ␣ subunit, which is known to directly interact with ␤␥ subunits (33)(34)(35). Indeed, ␤␥ binding changes G ␣ amino-terminal structure (33), perhaps making it a better substrate for PAT.
Myristoylation may also facilitate access to PAT. Studies with acylated peptides and model membranes have demonstrated that a myristoyl moiety is not sufficient for high affinity interaction of the acylated peptide with phospholipid vesicles (36). However, even a transient interaction of a myristoylated protein with the membrane may allow the protein to be recognized by PAT and become palmitoylated. The dually acylated protein will then have a high affinity for membranes. Most members of the Src family of protein-tyrosine kinases have an amino-terminal motif of a myristoylated glycine followed by a palmitoylated cysteine (37)(38)(39). Palmitoylated Src family kinases require prior myristoylation to be palmitoylated both in vivo (39) and in vitro (10). Although we have shown that myristoylation is not an absolute requirement for recognition of the ␣ subunit by the enzyme, the presence of myristate increases in vitro acylation of G-protein subunits. Further purification and characterization of PAT activity is required to resolve whether the same PAT activity acylates G-protein ␣ subunits and Src family kinases.
Fatty Acyl-CoA Substrate Specificity-To determine the fatty acyl chain length that PAT prefers as substrate, unlabeled acyl-CoAs of varying chain length and saturation were tested for inhibition of a partially purified preparation of PAT. Unlabeled palmitoyl-CoA competed the best for PAT activity, suggesting that palmitate is the biologically relevant fatty acid for the enzyme (Fig. 3A). C14 and C18 chain length acyl-CoAs competed almost as well as C16. Unsaturated acyl-CoAs were as effective as saturated forms of similar chain length. Inhibition of PAT activity by competing acyl-CoAs was concentrationdependent (Fig. 3B). Although not as potent an inhibitor as C14 and C16 fatty acyl-CoAs, arachidonoyl-CoA does compete with palmitoyl-CoA. Hallak et al. (40)   of the enzyme is similar to that reported by Berthiaume and Resh (10).
Subcellular Localization of PAT-Although palmitoylation of proteins occurs both in intracellular compartments and at the plasma membrane (41), dynamic acylation of G-protein ␣ subunits is likely to occur at the plasma membrane. To determine if PAT activity is present in plasma membranes, subcellular fractions of rat liver were prepared and analyzed. Specific activity was highest in plasma membranes and present at significantly lower levels in Golgi and mitochondria (Fig. 4A). PAT activity was absent in membranes enriched in endoplasmic reticulum (ER). In the absence of exogenously added G protein, plasma membrane and Golgi PAT activities were reduced to background levels (Fig. 4B, first lane and data not shown). PAT activity in mitochondria is due to an endogenous mitochondrial protein (Fig. 4B, last lane) that migrates at approximately 45 kDa. The mitochondrial protein is not recognized by immunoblot analysis with P-960 (42), an anti-peptide antibody that recognizes G o␣ , G i␣ , G s␣ , T ␣ , and G z␣ . Marker enzyme assays of the fractions illustrated that each membrane fraction was substantially enriched in the appropriate enzyme (Fig. 4A). Mitochondrial contamination of the plasma membrane fraction was evident but was not likely to contribute significantly to PAT activity in assays of plasma membrane. The 45-kDa mitochondrial protein substrate was not detected in assays of plasma membrane analyzed by SDS-PAGE and fluorography (Fig. 4B, first lane). Enrichment of PAT activity in the plasma membrane was not an artifact of palmitoyl-CoA substrate depletion in the other membranes. Levels of palmitoyl-CoA, assayed by TLC (43), were not significantly reduced in Golgi, ER, or mitochondrial fractions during the time course of the assay (data not shown).
Studies of receptor-stimulated turnover of palmitate on G s␣ are consistent with the following model (3)(4)(5). In the basal state, the ␣ subunit is palmitoylated and associated with ␤␥ subunits. Upon ligand binding to the receptor, the ␣ subunit becomes activated and dissociates from ␤␥ subunits. The ␣ subunit in its GTP-bound form is a substrate for a protein palmitoyl thioesterase and becomes deacylated. Deactivation of ␣ by GTP hydrolysis results in its reassociation with ␤␥ subunits and coincides with reacylation of the ␣ subunit. These studies do not discriminate between palmitoylation occurring before or after inactive ␣ binds to ␤␥ subunits. Because PAT activity prefers the ␣ subunit in the context of the heterotrimer as a substrate, we suggest that palmitoylation occurs after reassociation of the subunits. Enrichment of PAT activity in plasma membranes suggests that the subcellular localization of this enzyme allows for rapid reacylation of the G protein at the plasma membrane and does not require the G protein to cycle to an intracellular compartment for reacylation. FIG. 4. Subcellular localization of PAT in rat liver membranes. A, rat liver membranes were fractionated as described under "Experimental Procedures" and assayed for PAT activity. Heterotrimers of myristoylated rG i␣1 G203A and ␤␥C68S were used as substrates in the assay. The G203A substitution in E. coli-derived G i␣1 increases its ability to bind to ␤␥C68S (44) which has reduced affinity for G-protein ␣ subunits (16). (Note, however, at the concentrations of G-protein subunits in the PAT assay, both wild type G i␣1 and G i␣1 G203A form heterotrimers with ␤␥C68S.) The assay was performed as described except that the reaction was carried out using 15 g of membrane protein and the incubation time was 5 min. Specific activities for PAT activity are expressed as pmol/min/mg. Total PAT activities (pmol/min) in the subcellular fractions were: plasma membrane, 1.2 pmol/min; Golgi, 0.19 pmol/min; endoplasmic reticulum, none detected; and mitochondria, 14.8 pmol/min (primarily due to the 45-kDa mitochondrial protein). Specific activities of the marker enzymes are expressed as nmol of product/min/mg. B, SDS-PAGE analysis of PAT activity in subcellular fractions of rat liver. In an independent fractionation, smooth (SER) and rough (RER) endoplasmic reticulum were prepared as described (20), as well as plasma membrane (PM), cytosol (C), and mitochondria (M). Note the 45-kDa band migrating above G i␣ in the assay of mitochondrial PAT activity.