Intracellular Localization of the 3-Hydroxy-3-methylglutaryl Coenzyme A Cycle Enzymes in Liver SEPARATE CYTOPLASMIC AND MITOCHONDRIAL 3-HYDROXY-3-METHYLGLUTARYL COENZYME A GENERATING SYSTEMS FOR CHOLESTEROGENESIS AND KETOGENESIS*

Acetoacetyl-CoA thiolase and 3-hydroxy3-methylglutaryl coenzyme synthase which comprise the 3-hydroxy3-methyl-glutaryl-CoA-generating system(s) for hepatic cholesterogenesis and ketogenesis exhibit dual mitochondrial and cytoplasmic localization. Twenty to forty per cent of the thiolase and synthase of avian and rat liver are localized in the cytoplasmic compartment, the remainder residing in the mitochondria. In contrast, 3-hydroxy-3-methylglutaryl-CoA lyase, an enzyme unique to the “3-hydroxy-3-methylglutaryl-CoA cycle” of ketogenesis, appears to be localized in the mitochondrion. The small proportion, 4 to 8%, of this enzyme found in the cytoplasmic fraction appears to arise via leakage from the mitochondria during cell fractionation in that its properties, p1 and stability, are identical to those of the mitochondrial base. These results are consistent with the view that ketogenesis which involves all three enzymes, acetoacetyl-CoA thiolase, 3-hydroxy-3-methylglutaryl-CoA synthase and 3-hydroxyd-methylglutaryl-CoA lyase, occurs


Substantial mitochondrial
3-hydroxy-3-methylglutaryl-CoA lyase activity is present in all tissues surveyed, while only liver and kidney possess significant mitochondrial 3-hydroxy3-methylglutaryl-CoA synthase activity. Therefore, it is proposed that tissues other than liver and kidney are unable to generate acetoacetate because they lack the mitochondrial synthase.
3-Hydroxy-3-methylglutaryl coenzyme A (HMG-CoA), is synthesized from acetyl-CoA by a two-step reaction sequence (Scheme 1, Reactions 1 and 2) catalyzed by acetoacetyl-CoA-(AcAc-CoA)thiolase and HMGr-CoA synthase (l-4). By 1958 Rudney and Lynen and their collaborators had determined that HMG-CoA is an intermediate in two major pathways in animal liver, the synthesis of cholesterol and acetoacetate (1, 2). Based on these facts, Gould and Popjak proposed (5) that HMG-CoA is synthesized in common for both cholesterogenesis and ketogenesis, and that a branch point occurs between these two pathways at the first reaction unique to each pathway, i.e. the reduction of HMG-CoA to mevalonate and the cleavage of HMG-CoA to acetoacetate as illustrated in Scheme 1 (Reactions S and 4,respectively). This concept of a branched pathway for cholesterogenesis and ketogenesis has been generally accepted (6-9).
The work of Bucher et al. (10) on the subcellular distribution i The abbreviations used are: HMG, 3-hydroxy-3-methylglutaryl; Hepes, N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid. of HMG-CoA synthase in rat liver indicated that this enzyme was localized almost entirely in the mitochondrion; only miniscule levels of synthase activity were detected in the microsomal or cytosolic fractions.
Since cholesterogenesis from HMG-CoA appears to be primarily a cytoplasmic process in liver (8), the question is raised as to whether mitochondrially generated HMG-CoA is the major precursor for cytoplasmic cholesterol synthesis. Were this the case HMG-CoA generated in the mitochondrion could escape the high cleavage activity of HMG-CoA lyase in this organelle (11) only if the lyase were regulated. Moreover, since the inner mitochondrial membrane is impermeable to CoA derivatives (12) a regulated transport system would be needed for the HMG-acyl group to traverse this barrier. More recently the above assumptions were found to be unnecessary since HMG-CoA synthase, as well as acetoacetyl-CoA thiolase, were detected at substantial levels in the cytoplasmic compartment of liver (13)(14)(15). This suggested that liver possesses two HMG-CoA generating systems, a mitochondrial pathway for ketogenesis and a cytoplasmic pathway for cholesterogenesis. Some of the molecular and catalytic properties of mitochondrial and cytoplasmic acetoacetyl-CoA thiolases have been reported recently, and it has been established that different thiolase species function in ketogenesis and cholesterogenesis (13,14). Additional support for the existence of independent mitochondrial and cytoplasmic pathways for HMG-CoA synthesis was provided by the finding that different pools of acetyl-CoA supply ketogenesis and cholesterogenesis (16)(17)(18).
The present paper reports the subcellular distribution of the HMG-CoA cycle enzymes in chicken and rat liver and provides evidence that molecularly distinct species of HMG-CoA synthase participate in ketogenesis and cholesterogenesis.
In the accompanying papers (19,20) in chicken liver Livers from 20 &week-old white Leghorn cockerels fed ad libitum were fractionated into mitochondria (Fraction I, also containing nuclei and some unbroken cells) and cytoplasm (Fraction II) after which enzyme assays were conducted in the presence of 0.2% Triton (w/v) as described under "Experimental Procedure." HMG-CoA synthase and HMG-CoA lyase were assayed spectrophotometrically; 100 rnM glycylglycine buffer, pH 8.8, was employed for HMG-CoA synthaae assays. Protein w&s determined by the biuret method (31). "Homogenate" refers to enzymatic activity in the homogenate prior to the separation of the subcellular fractions, and this activity is taken as 100% for the calculation of percentages shown in parentheses. Enzyme activity is expressed as units (1 pmol of substrate converted/min)/g wet weight of liver & the standard deviation of the mean. Isolated mitochondria from chicken liver were prepared as described under "Experimental Procedure" except that the washed mitochondria were resuspended in 50 mM potassium phosphate buffer, pH 7.0, instead of homogenizing buffer. This mitochondrial suspension (20 mg of protein/ml) was sonicated for three 20-s periods at 4' with a Bronwill Biosonik sonicator (model Bio II) at a setting of 20 using the small probe. Sonicated mitochondria were separated into a soluble and particulate fraction by centrifugation at 126,000 X g for 1 hour at 4', after which the pellet was suspended in 50 mM potassium phosphate, pH 7.0. HMG-CoA synthase and HMG-CoA lyase were msayed spectrophotometrically; Tris(Cl-) buffer, pH 8.2, was used for the synthase assays. Acetoacetyl-CoA thiolase and malate dehydrogenase assays were as described under "Experimental Procedure." "Mitochondria" refers to the sonicated, but uncentrifuged mitochondria. Values given in parentheses are total units (micromoles of substrate converted per min) in the mitochondrial fraction/g wet weight of liver. The chicken liver and rat liver (referred to only in the text) mitochondria had acceptor control ratios of 5.7 and 7.5 respectively, with succinate as substrate. a This value and that reported earlier (13) are low relative to the activity shown in Table I. The difference is attributable to partial inactivation of thiolase by sonication.
solution contained 50 ~1 of 2-mercaptoethanol, 25% of the appropriate Ampholine, and water to a final volume of 50 ml. After adding the dense electrode solution to the electrofocusing column, a linear density gradient, prepared from the dense and light solutions with a gradient mixer, was applied to the column. Enzyme samples were either mixed with an equal volume of the dense solution and added to the column at the midpoint of the gradient or, when the volume of the sample was >2 ml, mixed with the dense solution and applied with the gradient. The light electrode solution was applied to the top of the column and focusing conducted at 4' with the voltages for the periods indicated, after which the column contents were collected fractionally and the pH of each fraction was determined immediately at 4'.

HMG-CoA Synthase
Assays-HMG-CoA synthase activity can be accurately measured in enzyme extracts free of HMG-CoA lyase by following the incorporation of [l-or 2-%]acetyl-CoA into HMG-CoA or its enzymic dehydration product, 3-methylglutaconyl-CoA.
The complete reaction mixture contains 20 pmol of Tris (Cl-), pH 8.0,20 nmol of EDTA, 40 nmol of [l-or 2J%]acetyl-CoA (2 X 106 cpm/pmol), 10 nmol of acetoacetyl-Cob, and from 0.1 to 1.0 milliunit of HMG-CoA synthase in a total volume of 0.2 ml. The reaction is initiated by addition [%]acetyl-CoA to the otherwise complete reaction mixture after a preliminary incubation for 2 min at 30". At. 2, 4, 6, and 8 min after addition of acetyl-CoA, 0.04-ml aliquots are removed and pipetted into glass vials (15 X 45 mm) containing 0.1 ml of 6 N HCl, and the acidified aliquot is taken to dryness at 95'in a forced-draft oven. Water and liquid scintillator are added and nonvolatile "C activity (as [%]HMG-CoA plus 3-[%]methylglutaconyl-CoA) is determined with a scintillation spectrometer. Under the conditions described, a linear increase in nonvolatile "C activity with time is observed until at least 70y0 of the added acetoacetyl-CoA is consumed and the rate of incorporation is proportional to the amount of synthase added as shown in Fig. 1A. When highly purified preparations of HMG-CoA synthase are employed, the sole radioactive product of the assay following alkaline deacylation of thioesters was found to be 3-hydroxy-3-methylglutaric acid as judged by paper chromatography (Whatman 3MM paper; n-amylformate-formic acid solvent ascending (32), Rp = 0.53, and 1-butanol-acetic acidwater ascending (33), Rp = 0.72). With crude preparations of HMG-CoA synthase, two radioactive products are observed, 3hydroxy-3-methylglutarate and 3-methylglutaconate, at a near equilibrium ratio of 4:l (34). The sum of the radioactivity in these two products was equal to the nonvolatile radioactivity observed in the HMG-CoA synthase assay. The 3-methylglutaconyl-CoA observed presumably arose by the action of HMG-CoA dehydrase (EC 4.2.1.18) present in the crude enzyme preparations.
It should be noted that the radiochemical assay cannot be employed for enzyme preparations containing active HMG-CoA lyase. Lyase activity of crude avian liver extracts can be quantitatively inactivated by a procedure described under "HMG-CoA Lyase Assay." For crude enzyme preparations containing HMG-CoA lyase, a spectrophotometric assay modified from that of Ferguson and Rudney (35) was generally used. The complete assay mixture contains, in a l.O-ml volume: either 100 pmol of Tris (Cl-), pH 8.0 or 8.2, or glycylglycine (Na+), pH 8.8, as indicated in the text; O.lamol of EDTA; 0.2pmol of acetyl-CoA; 50 nmol of acetoacetyl-CoA; and HMG-CoA synthase. Following a 2-min preliminary incubation at 30" of 0.5 to 5.0 milliunits of the enzyme in the assay mixture minus acetyl-CoA, the reaction is initiated by addition of acetyl-CoA. The rate of HMG-CoA synthesis is monitored by following the acetyl-CoA-dependent decrease in Aa00 nm due to the consumption of acetoacetyl-CoA.
The decrease in ANO nm is linear with time until approximately 70'% of the acetoacetyl-CoA has reacted and is proportional to enzyme concentration within the limits specified (as is shown in Fig. 1B) HMG-CoA synthase were assayed, glycylglycine (Na+), pH 8.8, was used without MgCl, where acetoacetyl-CoA has an e:$yrn = 7.8 X 108 &I-=.
HMG-CoA synthase activity is equal to one-half the rate of acetoacetyl-CoA consumption for preparations, e.g. crude mitochondrial or cytoplasmic extracts, in which acetoacetyl-CoA thiolase activity > HMG-CoA synthase activity. This correction is made to account for the CoA-dependent consumption of acetoacetyl-CoA catalyzed by thiolase, the extent of which is governed by the amount of CoA generated by HMG-CoA synthesis. As illustrated in Fig. 1, A and B, the activities determined by the radiochemical and spectrophotometric assays are in good agreement.

HMG-CoA
Lyase Assays-A new radiochemical assay for HMG-CoA lyase was developed which has the advantage of increased sensitivity over the spectrophotometric assay (36). The lyasecatalyzed conversion of [3-"C]HMG-CoA (not volatile when taken to dryness in 6 N HCl at 95'), to its volatile product, [3-i4C]acetoacetate, is measured. The reaction is initiated by the addition of 0.5 to 3.5 milliunits of HMG-CoA lyase to a reaction mixture containing 2Opmol of Tris (Cl-), pH 8.2, and 40 nmol of (R,S)-[3-"Cl-HMG-CoA (specific activity, 5 to 20 X lo6 cpm/pmol) in a total volume of 0.2 ml. After 2,4,6, and 8 min of incubation at 30°, 0.04ml diquots are transferred to glass vials containing 0.1 ml of 6 N HCl and nonvolatile 14C activity determined as described for the radiochemical HMG-CoA synthase assay. As shown in Fig. 2, A

Intracellular
Distribution of Acetoacetyl-CoA Thiolase, S-Hydroxy-S-methylglutaryl-CoA Synthase, and S-Hydroxy-S-methylglutaryl-CoA Lyase-It has been assumed and earlier studies appeared to support the primarily mitochondrial localization of acetoacetyl-CoA thiolase, HMG-CoA synthase, and HMG-CoA lyase (10, 40). Of these enzymes, acetoacetyl-CoA thiolase and HMG-CoA synthase were known to function in both ketogenesis and cholesterogenesis (1. 2), while HMG-CoA lyase was thought to participate only in ketogenesis (2). More recent reports (13)(14)(15), indicating that acetoacetyl-Coil thiolase and HMG-CoA synthase are present both in the cytoplasmic and mitochondrial compartments of liver, prompted us to undertake careful intracellular fractionation studies on the HMG-CoA cycle enzymes in chicken and rat liver. Table I shows that although most of the cellular acetoacetyl-CoA thiolase and HMG-CoA synthase are found in the mitochondrial-nuclear fraction of chicken liver, a substantial fraction, 44 and 21%, respectively, is present in the cytoplasm. It is evident from the intramitochondrial distribution studies to be discussed that > 90% of the thiolase, synthase, and lyase in Fraction I is of mitochondrial, rather than nuclear, origin. In contrast to this dual localization of the HMG-CoA synthesizing system of liver, the ketogenic enzyme, HMG-CoA lyase, appears to be exclusively localized within the mitochondrion.
Only 4% of the cellular lyase activity is found in cytoplasm, and this small percentage is probably the result of mitochondrial leakage, as will be discussed subsequently. Contamination of the cytoplasmic fraction with mitochondrial components was minimal (Table I) as estimated by the presence of the mitochondrial matrix marker enzyme, citrate synthase. Cytoplasmic contamination of the mitochondrial fraction was about 10% as judged by the distribution of the cytosolic marker, lactate dehydrogenase (Table I). A similar intracellular enzyme distribution was obtained with rat liver (results not shown) where 30 and 20% of the total cellular acetoacetyl-Cob thiolase and HMG-CoA synthase activities, respectively, were present in the cytoplasmic fraction, the remainder being mitochondrial.
Does HMG-CoA Lyase Activity Detected in Cytoplasmic Fraction Arise by Lealcage from Mitochondria during Subcellular Fractiona-tion&It has been reported that a significant fraction of the HMG-CoA lyase activity of rat and avian liver resides in the cytoplasm (41,42). However, there is disagreement on this point since the present investigation (Table I) and others (10, 43) indicate that the lyase activity of the cytoplasm constitutes only a minor fraction of the total hepatic activity, an amount readily accounted for by mitochondrial breakage during cell fractionation.
In our studies the percentage of hepatic lyase in the cytoplasm ranged from 4y0 in the fed state (Table I) to 7 to 8% in the fasted state (results not shown). Nonetheless, it was important to ascertain whether a cytoplasmic lyase exists because its presence in the cytoplasm would require that the flux of HMG-CoA into acetoacetate versus cholesterol be regulated. To determine whether HMG-CoA lyase found in the cytoplasmic fraction could have arisen by leakage from mitochondria, the isoelectric points and heat inactivation profiles of lyase activities were compared for the enzyme from the mitochondrial and cytoplasmic fractions of liver (from 4%hour fasted chickens). As shown in Fig. 3, both HMG-CoA lyase activities present in the mitochondrial and cytoplasmic fractions focus at precisely the same point, i.e. at pH 6.1. Although the identical isoelectric points of these activities could be fortuitous, this result supports the view that the lyase found in the cytoplasmic fraction is the mitochondrial enzyme. Additional evidence for this contention was obtained by comparing the heat lability of the HMG-CoA lyase activity in the cytoplasmic and mitochondrial fractions.
As illustrated in Fig. 4, both activities exhibit identical exponential activity decay at 48", the apparent half-life for inactivation being 1.4 min. Thus, several lines of evidence indicate that "cytoplasmic" HMG-CoA lyase activity and hence cytoplasmic ketogenesis are artifacts resulting from the leakage of the lyase from mitochondria.
It appears that HMG-CoA lyase, an exclusively ketogenic enzyme, is localized solely within the mitochondrion.
Intramitochondrial Localization of Acetoacetyl-CoA Thiolase, HMG-CoA Synthase, and HMG-CoA Lyase-Mitochondria from chicken liver after brief sonication were separated into soluble and particulate fractions by centrifugation at 126,000 x g for 1 hour. As shown in Table II, 78 to 97% of the HMG-CoA cycle enzyme activities were recovered in the soluble fraction (Fraction I) ; the small percentages of activity found in the particulate fraction (Table II, Fraction II) are probably due to undisrupted mitochondria. This is supported by the fact that 19% of the soluble mitochondrial matrix marker enzyme, malate dehydrogenase, is also found in this membrane fraction (Table II). Experiments conducted in an identical manner, but with rat liver mitochondria (results not shown), revealed that the HMG-CoA cycle enzymes of rat liver also appear in the soluble fraction of the mitochondrion.
These results indicate that mitochondrial acetoacetyl-Cob thiolase, HMG-CoA synthase, and HMG-CoA 3. Isoelectric focusing of mitochondrial and "cytoplasmic" HMG-CoA lyase activity from chicken liver. Isoelectric focusing was conducted as described under "Experimental Procedure." A, mitochondrial matrix from a chicken fasted for 3 days was prepared as described in Table II except that 20 mM sodium phosphate, pH 7.0, was substituted for potassium phosphate. Focusing of 6.9 units of mitochondrial HMG-CoA lyase activity (32 mg of protein) was carried out in 1% Ampholine 8141 for 70 hours at 300 volts. B, cytosol was prepared from a 3-day fasted chicken as described under "Experimental Procedure." Focusing of 1.0 unit of "cytoplasmic" HMG-CoA lyase activity (97 mg of protein) was carried out in 1% Ampholine 8141 for 66 hours at 300 volts. For both A and B, the column contents were collected in 2-ml fractions, and HMG-CoA lyase activity was determined by the radiochemical assay using [3-*'C]HMG-CoA of specific activity 8.9 x lo6 cpm/pmol. Recovery of HMG-CoA lyase activity after isoelectric focusing was 17yo for A and 670/, for B.
lyase are localized within one of the two soluble compartments of the mitochondrion, i.e. the intracristal space or the matrix. In order to determine which soluble compartment within the mitochondrion contains the HMG-CoA cycle enzymes, chicken and rat liver mitochondrial preparations were fractionated by the method of Schnaitman and Greenwalt (26) to separate the mitoplast (inner membrane-matrix vesicle) from the outer membrane and intracristal space components. The ability of this procedure to resolve these submitochondrial fract,ions from both chicken and rat liver is verified by the distribution of appropriate marker enzymes (Table III). Essentially all of the succinate dehydrogenase and citrate synthase, inner membrane, and matrix enzymes (26,44), respectively, was found to be associated with the mitoplast.
Monoamine oxidase, an accepted outer membrane marker (26, 44), also yielded the expected intramitochondrial distribution pattern although in chicken liver mitochondria, the level of activity is quite low and approached the limit of detection.
Adenylate kinase whi.ch served as intracristate marker enzyme (26, 44) was localized primarily in the fraction (Fraction I, Table III) external to the inner membranematrix vesicle. Although the basis for the apparent activation (or deinhibition) of adenylate kinase caused by fractionating chicken liver mitochonclria (Table III) is not understood, this phenomenon was observed consistently. HMG-CoA lyase activities from chicken liver. Mitochondrial matrix was prepared as described in Table II, while cytosol was prepared as described under "Experimental Procedure." Heat inactivation was conducted in 0.1 ml of 20 mM Hepes buffer, pH 7.0, with either 2.1 milliunits of mitochondrial or 2.3 milliunits of "cytoplasmic" HMG-CoA lyase activity. After incubating at 48" for the times indicated, samples were removed and placed in an ice bath and assayed for HMG-CoA lyase activity using the radiochemical assay described under "J&perimental Procedure." HMG-CoA Lyase Activity Remaining refers to per cent activity remaining as compared to samples kept at t&2'. Table III, the three HMG-CoA cycle enzymes, acetoacetyl-CoA thiolase, HMG-CoA synthase, and HMG-CoA lyase, were found primarily in the fraction containing the matrix and inner membrane components.

As illustrated in
Since these enzymes had already been shown to be soluble and not membrane-bound (Table II), it is evident that the ketogenic pathway is compartmentalized within the mitochondrial matrix. This is in basic agreement with a recent report of Chapman et al. (45) who found thiolase and HMG-CoA lyase to be localized within the matrix compartment.
However, these investigators studied neither the distribution of these enzymes between mitochondria and cytoplasm nor the localization of HMG-CoA synthase.

Intracytoplasmic
Localization of Acetoacetyl-CoA Thiolase and HMG-CoA Synthase-Cytoplasmic HMG-CoA synthase has been reported to be both a soluble and a membrane-bound enzyme in liver (15,46). To evaluate the intracytoplasmic localization of the enzymes of the HMG-CoA synthesizing system, cytoplasm was separated into cytosol and washed microsomes as described under "Experimental Procedure," and the distribution of acetoacetyl-CoA thiolase and HMG-CoA synthase between these fractions was compared with that of lactate dehydrogenase, a cytosolic marker enzyme. All of the acetoacetyl-CoA thiolase and HMG-CoA synthase activity was found in the cytosolic fraction (Table IV, Fraction 1) of both chicken and rat liver; no activity of either enzyme could be detected in the washed microsomal fraction. Thus, it appears that the intracytoplasmic location of the HMG-CoA generating system which presumably provides precursor for cholesterol synthesis, is cytosolic.
Distinct Forms of HMG-CoA Synthase in Mitochondriul and Cytoplasmic Compartments of Liver-Recently it was established (13,14) on the basis of several criteria, including isoelectric focusing, that different forms of acetoacetyl-CoA thiolase are present in the mitochondrion and cytoplasm of liver. This and the fact that HMG-CoA synthase activity is present ( CoA. It became important to determine whether the synthase activities found in the mitochondrial and cytoplasmic fractions are indeed different molecular forms of the enzyme. Our earlier work (15) indicates the existence of two cytoplasmic forms of the enzyme, i.e. HMG-CoA synthase I and II. When the cytoplasmic fraction of chicken liver is subjected to isoelectric focusing, as illustrated in Fig. 5B, two distinct peaks of activity are observed. The p1 values for these forms, 6.7 and 4.8, correspond to the p1 values obtained for the purified cytoplasmic HMG-CoA synthases I and II, respectively (20). Isoelectric focusing of the lysed mitochondrial fraction of chicken liver yielded a single peak of HMG-CoA synthase activity with an isoelectric point at pH 7.2. Of importance is the fact that the mitochondrial and cytoplasmic synthases retain their respective p1 values after purification to homogeneity (19,20). Thus, it is evident that the mitochondrial and cytoplasmic species of HMG-CoA synthase are molecularly distinct as judged by their isoelectric points. Evidence is presented in an accompanying paper (20) only tissue unable to metabolize ketones, i.e. acetoacetate and B-hydroxybutyrate, at significant rates (48). Mitochondrial acetoacetyl-CoA thiolase is involved in both the synthesis of ketones by liver as well as the utilization of ketones by extrahepatic tissues. Consistent with these physiological roles of mitochondrial acetoacetyl-CoA thiolase, this enzyme is present in all animal tissues surveyed (14). The inability of extrahepatic tissues to synthesize ketones has been attributed to the absence of mitochondrial HMG-CoA lyase or HMG-CoA synthase, or both (47). Mitochondrial HMG-CoA synthase appears to function only in ketogenesis while HMG-CoA lyase, in addition to hepatic ketogenesis, participates in leucine catabolism (36, 37). As shown in Table V, liver and kidney contain significant activities of mitochondrial HMG-CoA synthase, whereas other tissues, such as heart, ileum, brain and skeletal muscle, do not contain detectable levels of this enzyme, i.e. <O.Ol unit/g wet weight of tissue. To confirm this finding, mitochondrial HMG-CoA synthase activity was estimated using antiserum prepared against purified chicken liver mitochondrial HMG-CoA synthase (19). By the Ouchterlony double diffusion technique, precipitin lines were observed for tissues such as liver and kidney which contain >0.02 unit of HMG-CoA synthase activity/g of tissue, while no precipitin line could be detected for heart, brain, ileum, or skeletal muscle .  TABLE 3116 connection, mitochondrial HMG-CoA synthase appears to function primarily in ketogenesis, while HMG-CoA lyase participates in both hepatic ketogenesis and in the leucine catabolic pathway (37, 51). As illustrated in Table V, liver and kidney contain significant mitochondrial HMG-CoA synthase activity, whereas in other tissues, such as heart, ileum, brain, and skeletal muscle, the enzyme appears to be absent, i.e. <O.Ol unit/g wet weight of tissue. In contrast, the same tissues contained significant levels of mitochondrial HMG-CoA lyase activity (Table V). These results are in disagreement with those conducted with the rat where HMG-CoA lyase activity was not detectable in brain and skeletal muscle (47,52). Rat brain tested for HMG-CoA lyase activity as described in Table V contains 0.7 unit of the mitochondrial enzyme/g wet weight of tissue (average of four rats). This level of activity was confirmed by the direct measurement of [3-14C]acetoacetate formed from [3-14C]-HMG-CoA. The presence of HMG-CoA lyase in all extrahepatic tissues surveyed is consistent with the occurrence of leucine catabolism primarily in extrahepatic tissues (53). In view of the tissue distribution of mitochondrial thiolase. synthase, and lyase discussed above, we suggest that the ability of a tissue to synthesize acetoacetate is determined by its mitochondrial content of HMG-CoA synthase.