Expression of a cDNA Isolated from Rat Brown Adipose Tissue and Heart Identifies the Product as the Muscle Isoform of Carnitine Palmitoyltransferase I (M-CPT I)

We set out to determine if the cDNA encoding a carni- tine palmitoyltransferase (CPT)-like protein recently isolated from rat brown adipose tissue (BAT) by Yamazaki et al. (Yamazaki, N., Shinohara, Y., Shima, A., and Terada, H. (1995) FEBS Lett. 363, 41–45) actually encodes the muscle isoform of mitochondrial CPT I (M-CPT I). To this end, a cDNA essentially identical to the original BAT clone was isolated from a rat heart library. When expressed in COS cells, the novel cDNA and our previously described cDNA for rat liver CPT I (L-CPT I) gave rise to products with the same kinetic characteris- tics (sensitivity to malonyl-CoA and K m for carnitine) as CPT I in skeletal muscle and liver mitochondria, respec- tively. When labeled with [ 3 H]etomoxir, recombinant L-CPT I and putative M-CPT I, although having approxi- mately the same predicated masses (88.2 kDa), migrated differently on SDS gels, as did CPT I from liver and muscle mitochondria. The same was true for the products of in vitro transcription and translation of the L- CPT I and putative M-CPT I cDNAs. We conclude that the BAT cDNA

Transport of long chain fatty acyl groups into the mitochondrial matrix to undergo ␤-oxidation is effected by the mitochon-drial carnitine palmitoyltransferase (CPT) 1 enzyme system. CPT I, an integral outer membrane protein, catalyzes the transfer of an acyl group from coenzyme A to carnitine, the acylcarnitine product traversing the inner membrane by means of a specific translocase. The transesterification is then reversed by CPT II, associated with the matrix face of the inner membrane. CPT I has been the focus of particular attention due to its unique inhibition by malonyl-CoA, a property of the enzyme that is central to the physiological regulation of the ␤-oxidation pathway (1,2). As a consequence, CPT I has aroused interest as a potential site for pharmacological inhibition of fatty acid oxidation in the liver in states where this process occurs at excessive rates (such as poorly controlled diabetes (3,4)) or in the ischemic heart, where elevated levels of acylcarnitines have been associated with arrhythmias (5).
Whereas CPT II appears to be the same protein in all tissues (6), CPT I exists as at least two isoforms (6,7). 2 These have been designated L-CPT I (expressed in liver and fibroblasts (9,10)) and M-CPT I (expressed in skeletal muscle (6)). The heart expresses both forms, the muscle variant becoming increasingly predominant during neonatal development in the rat (7,11,12). A fuller understanding of the tissue distribution and properties of CPT I isoforms has been impeded by the fact that although cDNAs for L-CPT I and CPT II have been cloned from both rats (9,13) and humans (10,14), the same goal has not been achieved with certainty in the case of M-CPT I.
Recently, a cDNA encoding a CPT I-like protein was isolated from rat brown adipose tissue (BAT) (15). This was obtained by a subtractive cloning strategy aimed at identifying proteins expressed in BAT, but not in white adipose tissue (WAT). The derived product was predicted to be a protein of 772 amino acids having 62.6% identity to rat L-CPT I (773 amino acids). Northern blot analysis indicated high levels of expression in BAT, skeletal muscle, and heart (15). However, the study cited left unanswered three important questions. First, does this cDNA in fact correspond to M-CPT I or to an additional member of the growing family of carnitine acyltransferases that have been characterized in recent years (16 -19)? Second, if the * This work was supported by National Institutes of Health Grant DK18573, the Chilton Foundation, and Sandoz Pharmaceuticals. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Contributed equally to this work and should be considered co-first authors.
new cDNA does encode M-CPT I with a predicted mass of 88,227 Da, which is almost identical to the value of 88,150 Da predicted for L-CPT I (9), why do the two proteins (when labeled with [ 3 H]etomoxir) migrate so differently on SDS gels (6,7)? Third, why should brown and white fat express different CPT I isoforms?
The studies outlined below leave little doubt that the CPT I expressed in rat BAT is the muscle isoform of the enzyme. They also address the question of why the liver and muscle forms of CPT I behave differently on SDS gels. Finally, the new findings allow refinement of the pattern of CPT I isoform expression in brown and white fat.

EXPERIMENTAL PROCEDURES
Animals-Male Sprague-Dawley rats were fed a standard laboratory chow (4% fat, w/w) with lighting from 10.00 to 22.00 h. For studies on BAT (interscapular), the animals were used at a body weight of 100 -120 g; in all other cases, body weight ranged from 180 to 220 g. Organs were removed from anesthetized or cervically dislocated animals between 10.00 and 11.00 h.
General Molecular Biology Methods-Standard molecular biological techniques were employed (20). cDNA clones were sequenced by the dideoxy chain termination method (21) using the Sequenase T7 DNA polymerase kit from U. S. Biochemical Corp. First strand cDNA was synthesized from total RNA by reverse transcriptase using the kit from Life Technologies, Inc.
Cloning Methodology-Two oligonucleotides were synthesized in order to amplify, by the polymerase chain reaction (PCR) method (22), a fragment of the clone reported by Yamazaki et al. (15). The forward primer, 5Ј-GTGTTCGTCTCCTGTCCAGC-3Ј, extended from nucleotides 490 to 509, and the reverse primer, 5Ј-GTTTCTTATGGGAC-CCCGTC-3Ј, from nucleotides 1886 to 1905 (bases numbered according to Ref. 15). BAT cDNA was used as template. The PCR product was verified by sequencing. The amplified fragment was radiolabeled using the random hexanucleotide method and employed to screen a rat heart library in phage gt10 (7). A positive clone was isolated, excised with NotI, and subcloned into both pBluescript SK(ϩ) and the mammalian expression vector pCMV6 (23). Sequencing of the 5Ј-end of the clone revealed that the bases encoding the first two amino acids were missing. Oligonucleotides were synthesized to amplify the 5Ј-end of the desired sequence using BAT cDNA as template. The forward oligonucleotide read 5Ј-GGGGTACCAGCTGTGCTGACTAAACC-3Ј (nucleotides 4 -20), and the reverse primer read 5Ј-CATGGGTACCATAC-CCAGTG-3Ј (nucleotides 919 -938). The PCR product was subcloned into bacteriophage M13 and sequenced. The KpnI fragments from the pBluescript SK and pCMV6 plasmids containing the incomplete cDNA clone were removed and replaced with the amplified product to give the plasmids pB-rM-CPT I and pCMV6-rM-CPT I, respectively. pCMV6-rL-CPT I encoding rat liver CPT I is the same plasmid referred to as pCMV6-CPT I in Ref. 9. pB-rL-CPT I was constructed by inserting into pBluescript SK(ϩ) the coding sequence excised from pY-CPTI-MetI (24) with the restriction enzyme EcoRI.
COS Cell Transfection-Simian COS-M6 cells were transfected with the pCMV6-rL-CPT I and pCMV6-rM-CPT I plasmids as described (9). As a control, cells were mock-transfected using water in place of plasmid DNA.
In Vitro Transcription and Translation-pBluescript plasmids were linearized 3Ј to the coding region with NotI. In vitro transcription was accomplished using T7 RNA polymerase in the Stratagene mRNA capping kit (28). Translation was performed with a rabbit reticulocyte lysate system (DuPont NEN) in the presence of [ 35 S]methionine (28).
Preparation and Treatment of Mitochondria-Mitochondria were prepared from whole tissues as described (Ref. 29, Method A), except for the inclusion of 2% fatty acid-free bovine serum albumin during homogenization of BAT and epididymal fat pads. Tissue was pooled from one to two rats (liver, heart, and hind limb muscle), three to six rats (BAT), or four rats (epididymal fat pad). Adipocytes were isolated from the combined epididymal fat pads of eight rats by the collagenase digestion method of Rodbell (30). They were then homogenized, and mitochondria were prepared as from whole fat pads.
Whole mitochondria were used directly for assay of CPT I. To allow for variation in the extent of malonyl-CoA-insensitive CPT II exposed during mitochondrial preparation, the CPT I values given refer to enzyme activity inhibited by 100 M malonyl-CoA (see "Results"). For measurement of CPT II, mitochondria were made 1% (w/v) with octyl glucoside and kept on ice for 30 min before assay, with frequent vortexing. This procedure solubilizes CPT II in active form while inactivating CPT I (31). Treatment of mitochondrial fractions with the covalent inhibitors [ 3 H]etomoxir and DNP-etomoxir were as described (11).
CPT Assay-Mitochondrial CPT activity was measured using a modified version of the assay previously described (29) in a total volume of 500 l containing 500 M L-carnitine (including 0.125 Ci of L-[ 14 C]carnitine), 50 M palmitoyl-CoA, 4 mM ATP, 0.25 mM GSH, 4 mM MgCl 2 , 40 mg/ml rotenone, 2 mM KCN, 15 mM KCl, 1% (w/v) bovine serum albumin, and 105 mM Tris-HCl (pH 7.2). Malonyl-CoA was included as indicated. Reactions were started by addition of mitochondria (in 100 l) and were allowed to proceed for 4 -10 min (rates were linear for up to 12 min). After addition of 500 l of 1.2 N HCl, the palmitoyl[ 14 C]carnitine product was extracted into 500 l of butan-1-ol. Three-hundred microliters of the upper butanol phase were back-extracted with 60 l of water, and 250 l of the final organic phase were taken for liquid scintillation counting. Protein concentration was determined by the method of Lowry et al. (32).
Materials-Sources of materials have been given previously (9 -13).

Expression of Putative M-CPT I cDNA-
Oligonucleotides were designed on the basis of the published sequence of the BAT CPT-like protein and used to amplify a portion of BAT cDNA by PCR. This was then labeled and used as a probe to screen a rat heart cDNA library, resulting in the isolation of several positive clones. The longest one was subjected to DNA sequencing, revealing complete identity to the published sequence (15). However, it was found to be short at the 5Ј-end, lacking codons for the first two amino acids (including the initiator methionine) and all of the 5Ј-untranslated region. To generate the full-length construct necessary for expression studies, PCR was again used, this time to amplify the 5Ј-end of the sequence from BAT cDNA. DNA sequencing of this product revealed only two silent mutations (or amplification errors) at codons 27 (Ala GCT 3 GCA) and 95 (Tyr TAT 3 TAC). A full-length construct was then assembled in the mammalian expression vector pCMV6 to form pCMV6-rM-CPT I as described under "Experimental Procedures." COS-M6 cells were transfected with plasmid pCMV6-rM-CPT I or pCMV6-rL-CPT I (encoding rat L-CPT I). Mock transfections contained no plasmid. Table I shows the CPT activity measured in crude homogenates of the cells 48 h after transfection in two independent experiments. pCMV6-rL-CPT I caused a 6 -7-fold induction of CPT I activity (i.e. activity measured in the absence of octyl glucoside), consistent with previous results (9). A 5-fold induction was observed with pCMV6-rM-CPT I. Both activities were substantially inhibited in the presence of 100 M malonyl-CoA. Putative M-CPT I was the more sensitive, residual activity being similar to that in untransfected cells. For assay of CPT II, the membranes were solubilized in 1% octyl glucoside. Under these conditions, no change was observed in pCMV6-rM-CPT I-transfected relative to untransfected cells, but after transfection with pCMV6-rL-CPT I, enzyme activity in the presence of octyl glucoside did rise ϳ1.5-fold. This may represent incomplete inactivation of the induced L-CPT I by the detergent or up-regulation of endogenous CPT II. A more detailed analysis of the kinetic properties of the two expressed CPTs is presented in Fig. 1. A gross difference in sensitivity to malonyl-CoA is apparent, with I 50 values (concentration needed for 50% inhibition) of ϳ8 and 0.15 M, respectively, for the recombinant L-CPT I and putative M-CPT I variants (Fig. 1A). The response of each enzyme to increasing concentrations of carnitine is shown in Fig. 1B. Expressed L-CPT I saturated rapidly (K m ϳ 25 M), whereas the other enzyme displayed a much more gradual response (higher K m ). In the case of expressed M-CPT I, the presence of endogenous COS cell CPT I (probably the liver isoform) rendered the corresponding Eadie-Hofstee plot markedly nonlinear (more than one component), preventing simple calculation of an accurate K m . It is clear, however, that the exogenous CPT in this case contributed a high K m M-CPT I-like component.
Transfected and untransfected COS cells were incubated for 6 h in medium containing 3 M [ 3 H]etomoxir to covalently label the expressed CPT I isoforms. Total cell membrane extracts were then analyzed by SDS-PAGE and subsequent fluorography ( Fig. 2A). Both types of transfected cells contained highly induced labeled bands. Labeled L-CPT I migrated more slowly than its presumed muscle-type counterpart (ϳ88 and ϳ82 kDa, respectively), mirroring the behavior of [ 3 H]etomoxir-labeled CPT I isoforms in mitochondria from rat liver and muscle (see below). Only a faint labeled band, of approximately the size of the L-CPT I expression product, was visible in untransfected cells.
To address the question of the differential mobility of the two enzymes during SDS-PAGE, we used an in vitro transcription and translation system to synthesize 35 S-labeled protein products from the two CPT I clones. These would not be subject to post-translational modification, as might occur in the whole cell (e.g. upon mitochondrial import). In this system also, the radioactive proteins migrated differently (Fig. 2B). Furthermore, the apparent sizes of the in vitro synthesized 35 S-labeled proteins were indistinguishable from those of the corresponding [ 3 H]etomoxir-labeled enzymes from rat liver and muscle mitochondria.

Analysis of CPT I Isoform Expression in Rat Adipose
Tissues-[ 3 H]Etomoxir labeling of CPT I enzymes from mitochondria from several rat tissues is shown in Fig. 3A. Lanes 1-3 illustrate the established pattern of expression of CPT I isoenzymes in rat liver (L-CPT I), skeletal muscle (M-CPT I), and heart (predominantly M-CPT I), the muscle form migrating somewhat faster than the liver type. In BAT (lane 4), only a protein of the size of M-CPT I was detected. However, in white adipose tissue, mitochondria bands of both muscle and liver sizes were seen, L-CPT I being the major component (lane 5). 3 Since the epididymal fat pad, used here as a source of white adipose tissue, contains other cell types in addition to adipocytes, it was necessary to establish whether the labeling pattern observed was a property of the adipocytes themselves. To this end, similar experiments were performed using mitochondria from purified adipocytes (Fig. 3B). Under these circumstances, the expression pattern was reversed, the M-CPT I size band now predominating. To exclude the possibility that this phenomenon was an artifact due to degradation of L-CPT I during the prolonged adipocyte preparation protocol, we employed another covalent CPT I inhibitor to confirm the identities of the two labeled proteins. This agent, DNP-etomoxir, has been shown to have a high degree of specificity for the liver isoform of CPT I and to block subsequent binding of [ 3 H]etomoxir (11). Mitochondria were prepared from adipocytes and then preincubated for 1 h in the absence or presence of 10 M unlabeled DNP-etomoxir before exposure for an additional hour to 3 M [ 3 H]etomoxir. The effect of the DNP-etomoxir preincubation is shown in Fig. 3C. The minor upper L-CPT I size band was virtually eliminated. In contrast, the intensity of the lower band was unaffected by the DNP-etomoxir preincubation, consistent with its representing the muscle isoform.
The different CPT I profiles indicated by [ 3 H]etomoxir labeling of mitochondria prepared from BAT, whole epididymal WAT, and purified white adipocytes were paralleled by differences in the kinetic properties of the enzyme from those sources. When expressed relative to mitochondrial protein (Table II), the activity of CPT I measured in liver, heart, and skeletal muscle varied over only a 2-fold range, with a proportional change in the level of CPT II, so that the CPT II/CPT I ratio remained close to unity. (It is important to note that since the K m values for the substrates of L-CPT I, M-CPT I, and CPT II are different, the relative activities measured for the different isoenzymes will depend upon experimental conditions. The present data are intended to show that wide variation exists between tissues; however, they do not necessarily reflect the molar ratios in vivo.) In BAT mitochondria, the ratio was doubled, and in WAT and purified adipocytes, it rose to 9 and 14, respectively. This observation explains the apparent resistance to complete inhibition of CPT I activity in mitochondria from whole epididymal fat pads and adipocytes isolated from them (Table II, sixth column). Whereas inhibition of Ն90% is routine with 100 M malonyl-CoA in "intact" mitochondria from liver, heart, and skeletal muscle, the value dropped to ϳ83% in BAT and to only 40 -50% in white adipose-derived preparations. It is likely that in the case of each tissue or cell type, a similar small fraction of the mitochondria becomes damaged, but that the amount of malonyl-CoA-insensitive CPT II rendered overt is exaggerated in those tissues where the CPT II/CPT I ratio is highest. Accordingly, values for CPT I shown here have been assessed as the malonyl-CoA-sensitive component of overt CPT activity.
The potency of malonyl-CoA as an inhibitor of CPT I in each mitochondrial type is shown in Fig. 4. Skeletal muscle was the most sensitive (I 50 ϭ 0.04 M) and liver the least, with an I 50 100-fold greater (Table II, fourth column). Heart and BAT mitochondria exhibited a sensitivity close to that of skeletal muscle. CPT I from whole WAT behaved more like the liver enzyme, displaying an I 50 of 1.5 M. However, the malonyl-CoA response curve of CPT I in mitochondria from purified adipocytes was clearly shifted to the left, the I 50 of 0.23 M being consistent with a predominance of M-CPT I.
The K m values for carnitine of L-CPT I and M-CPT I are ϳ30 and 500 M, respectively (29). In BAT mitochondria, the K m was found to be Ͼ400 M, close to that of the muscle isoform of CPT I (Table II, fifth column). Unfortunately, the substantial contamination of CPT I by exposed CPT II in mitochondria from whole WAT and white adipocytes precluded accurate determination of a K m for carnitine in either case.
CPT I isoform expression was also studied at the level of mRNA for each tissue or cell type (Fig. 5). Fig. 5 (A and B) shows a Northern blot analysis of poly(A ϩ ) RNA isolated from liver, heart, skeletal muscle, and BAT. The L-CPT I probe generated strong signals in liver and heart, as expected, and a weak signal in BAT, while no band was detected in skeletal muscle (Fig. 5A). The M-CPT I probe revealed expression of this isoform in heart, skeletal muscle, and BAT, but not in liver (Fig. 5B). Due to the extremely low RNA yields from purified adipocytes, total RNA was used to perform Northern analysis on both adipocytes and whole WAT to allow for a direct comparison. The band representing the liver form was stronger in WAT than in adipocytes (Fig. 5C), whereas the mRNA for the muscle isoform was found to be the more abundant species in the purified cells (Fig. 5D).

DISCUSSION
The recent cloning and expression of cDNAs encoding rat and human L-CPT I and CPT II have provided considerable insight into the structure/function relationships between the CPT isoenzymes (9,13,24,28,33,34). There are, however, a number of important but unresolved issues. Particularly intriguing is the question of why CPT I should exist in at least two isoforms with distinct kinetic properties and tissue distribution, and how these relate to whole body fuel homeostasis. Progress on this front has been hampered by the unavailability of a cDNA corresponding to the muscle enzyme. Our initial goal here, therefore, was to determine whether a candidate cDNA isolated from rat BAT (15) did in fact represent muscle CPT I. To this end, we used sequence information from the BAT cDNA to screen a rat heart cDNA library, and the CPT I-like sequence was confirmed in that tissue.
As with L-CPT I (9), COS cells transfected with the putative M-CPT I cDNA generated a CPT activity with characteristics typical of mitochondrial CPT I, i.e. it was membrane-bound and malonyl-CoA-sensitive and lost activity upon solubilization of the membranes with the detergent octyl glucoside. Malonyl-CoA response curves established that expressed putative M-CPT I was far more sensitive to the inhibitor than was the liver enzyme. Although the I 50 value observed (0.15 M) was somewhat higher than that for CPT I from skeletal muscle mitochondria (ϳ0.04 M), the difference was likely due to the presence of background endogenous CPT I activity in the COS cells, which is of the less sensitive liver type. The I 50 for the ex-  (29,35,36). Furthermore, [ 3 H]etomoxir labeling experiments established that the recombinant L-CPT I and putative M-CPT I enzymes exhibited the same differential migration on SDS-PAGE as is seen with the native proteins. Thus, when expressed in COS cells, the putative M-CPT I clone matched in every respect the characteristics of the native muscle enzyme. These findings, coupled with the tissue expression data presented by Yamazaki et al. (15) and below, provide compelling evidence that the enzyme encoded by the clone originally isolated from BAT is identical to M-CPT I.
That being so, the question arises as to why two highly homologous proteins with almost identical predicted molecular masses should display such distinct electrophoretic mobilities. To investigate this question, we generated 35 S-labeled protein from the L-CPT I and M-CPT I clones using in vitro transcription and translation. The products were found not only to run differently, but to migrate exactly with their mitochondrial equivalents. This suggests that the phenomenon is not the result of post-translational modification, but stems from an intrinsic difference in the primary sequence of the two proteins. A comparison of the cDNA-derived polypeptide molecular masses (both ϳ88 kDa) with those estimated from SDS-PAGE analysis (ϳ88 and ϳ82 kDa for L-and M-CPT I, respectively) suggests that the muscle isoform behaves anomalously.
Our final aim was to investigate the implication of the work of Yamazaki et al. (15), that whereas M-CPT I is expressed in BAT, L-CPT I is the primary isoform in WAT. An additional consideration here was the earlier work by Saggerson and Carpenter showing that CPT I in BAT mitochondria is highly malonyl-CoA-sensitive, as is CPT I in muscle (37), but that in mitochondria from purified adipocytes, CPT I exhibits a sensitivity intermediate between that of the muscle and liver enzymes (38). A more detailed interpretation of these observations is now possible. The [ 3 H]etomoxir labeling patterns of CPT I obtained using mitochondria prepared from whole rat epididymal fat pad and BAT appeared to conform to a model in which L-and M-CPT I are the primary isoforms in those tissues, respectively, and Northern blots appeared to support this notion. 4 However, when the white adipocytes were separated from the bulk of the stromal and other cell types, the M-CPT I band labeled with [ 3 H]etomoxir was seen to dominate. Moreover, kinetic analysis of CPT I in mitochondria isolated from whole fat pad and adipocytes corroborated the predomi-   5. Northern blot analysis of RNA from rat tissues. A, 5 g of poly(A ϩ ) RNA from the indicated rat tissues were analyzed as described under "Experimental Procedures" using a single-stranded liver cDNA probe. B, 8 g of poly(A ϩ ) RNA were analyzed using a singlestranded muscle cDNA probe. Twenty micrograms of total RNA from whole WAT and purified adipocytes were analyzed using a singlestranded liver (C) or muscle (D) cDNA probe. kb, kilobases. nance of M-CPT I-like activity in the purified cells, and this was also reflected at the level of the mRNAs for L-and M-CPT I. Therefore, all of the above evidence pointed to a scenario in which the muscle-type enzyme not only represents the sole species of CPT I in BAT, but is also the dominant isoform in isolated white adipocytes from the epididymal fat pat. Furthermore, when expressed relative to mitochondrial protein, the CPT I activity of white adipocyte mitochondria is seen to be 20 -40-fold lower than that of preparations from any other tissue examined (liver, heart, skeletal muscle, or BAT). This suggests that a small contamination of the adipocytes with any other cell type containing CPT I activity at a similar level to those listed and that happens to express primarily L-CPT I would result in a disproportionate effect on the overall isoform profile. Hence, the minor quantity of L-CPT I found in the white adipocyte mitochondrial preparations could well have resulted from a slight impurity and might not have been of adipocyte origin. For this reason, we cannot exclude the possibility that M-CPT I is the sole CPT I isoform expressed both in brown and white fat cells. Therefore, at this stage, the differential cloning exercise that led to the original isolation of the M-CPT I cDNA (15) should not be taken to imply that white and brown adipocytes differ fundamentally in terms of their CPT I isoform expression. 5 A picture of tissue CPT I distribution is now emerging in which both isoenzymes are to be found in a variety of tissues (L-CPT I in liver, heart, and fibroblasts; M-CPT I in skeletal muscle, heart, and white and brown adipocytes). This will be a key consideration in the design of CPT I inhibitors as potential pharmaceutical agents. The diverse tissue expression of each isoform may also help to explain the multitissue symptoms exhibited by L-CPT I-deficient patients (39 -41). No M-CPT I deficiencies have yet been described. Obviously, further studies will be needed to establish a teleological basis for the presence of M-CPT I in adipocytes, indeed, to explain the tissue-specific expression of L-and M-CPT I generally. Identification of the rat M-CPT I cDNA represents an important step in this direction.