Rat Liver Carnitine Palmitoyltransferase 1 Forms an Oligomeric Complex within the Outer Mitochondrial Membrane*

Carnitine palmitoyltransferase (CPT) 1A catalyzes the rate-limiting step in the transport of long chain acyl-CoAs from cytoplasm to the mitochondrial matrix by converting them to acylcarnitines. Located within the outer mitochondrial membrane, CPT1A activity is inhibited by malonyl-CoA, its allosteric inhibitor. In this study, we investigate for the first time the quaternary structure of rat CPT1A. Chemical cross-linking studies using intact mitochondria isolated from fed rat liver or from Saccharomyces cerevisiae expressing CPT1A show that CPT1A self-assembles into an oligomeric complex. Size exclusion chromatography experiments using solubilized mitochondrial extracts suggest that the fundamental unit of its quaternary structure is a trimer. When studied in blue native-PAGE, the CPT1A hexamer could be observed, however, suggesting that under these native conditions CPT1A trimers might be arranged as dimers. Moreover, the oligomeric state of CPT1A was found unchanged by starvation and by streptozotocin-induced diabetes, conditions characterized by changes in malonyl-CoA sensitivity of CPT1A. Finally, gel filtration analysis of several yeast-expressed chimeric CPTs demonstrates that the first 147 N-terminal residues of CPT1A, encompassing its two transmembrane segments, trigger trimerization independently of its catalytic C-terminal domain. Deletion of residues 1–82, including transmembrane 1, did not abrogate oligomerization, but the latter is limited to a trimer by the presence of the large catalytic C-terminal domain on the cytosolic face of mitochondria. Based on these findings, we proposed that the oligomeric structure of CPT1A would allow the newly formed acylcarnitines to gain direct access into the intermembrane space, hence facilitating substrate channeling.

lytic C-terminal domain (11) and not within its cytosolic N terminus (12). However, the latter is essential for maintaining the high affinity malonyl-CoA-binding site (13). We have demonstrated that the N terminus of CPT1A, which contains both positive and negative determinants of malonyl-CoA sensitivity (14,15), influences the degree of malonyl-CoA sensitivity through modulations of its interaction with the C-terminal domain (16). Moreover, the existence of the two malonyl-CoAbinding sites, as well as the importance of these intramolecular N/C interactions for malonyl-CoA sensitivity, was recently supported by a three-dimensional in silico model of both the Nand C-terminal domains (17).
Although key insights into the structure-function relationships of CPT1A have emerged during the past decade, the quaternary structure of CPT1A still remains totally unknown. Previously reported hysteretic behavior and intrinsic positive cooperativity of CPT1A for palmitoyl-CoA (3, 18 -20) suggested multisubunit cooperation. Moreover, the observation of positive cooperative inhibition of CPT1A by malonyl-CoA and that malonyl-CoA could introduce sigmoidicity into the relationship between palmitoyl-CoA and CPT1A activity (3,18,21) strongly suggested allosteric behavior. Binding of malonyl-CoA could produce a conformational change that would either make the binding of a second malonyl-CoA much easier or make the binding of a second long chain acyl-CoA more difficult. Proteins exhibiting such cooperative effects are thought to be complexes of two or more subunits, raising the crucial question of whether CPT1A exists as a monomer or is assembled into an oligomeric complex once imported into the OMM.
In this study we have used three complementary approaches, namely chemical cross-linking, chromatography by gel filtration, and blue native PAGE, to study for the first time the quaternary structure of rat CPT1A. We show evidence that both CPT1A expressed in Saccharomyces cerevisiae and native protein from fed rat liver mitochondria self-assemble into an oligomeric complex and that the fundamental unit of their quaternary structure is a homotrimer. Moreover, the oligomeric state of CPT1A was unchanged by starvation and diabetes, conditions characterized by changes in malonyl-CoA sensitivity of CPT1A. Finally, by using several fusions or partially deleted CPT constructs, we demonstrated that the N-terminal domain of CPT1A can trigger its oligomerization, but the latter is limited to a trimer likely by the presence of its large catalytic C-terminal domain on the cytosolic face of the mitochondria. The functional relevance of the existence of an oligomeric CPT1A complex within the OMM is discussed.
A C-terminally Myc/His-6 tagged construct had been previously generated for expression in Pichia pastoris as follows. The PCR primers 5Ј-GCAGGTGGAGCTCTTTGAC-TTT-3Ј and 5Ј-TTTTGGGCCCCTTTTTAGAATTGATG-GTGAG-3Ј were used to amplify a region corresponding to the C terminus of CPT1A, with replacement of the stop codon with an ApaI restriction site. The product was cleaved with SacI-ApaI and used to replace the 3Ј end of CPT1A previously subcloned into pGAPZ A (Invitrogen). To generate the construct for S. cerevisiae expression, the SacI-BamHI fragment (including the region encoding the tag) was subcloned into empty pYeDP1-8/10. The remainder of CPT1A was then added as an EcoRI-SacI fragment. Each cDNA was placed under the control of the inducible GAL10 promoter present in the vector, and the constructs were used to transform S. cerevisiae. The fidelity of PCR and the quality of DNA subcloning were confirmed by DNA sequencing.
Isolation of Yeast and Rat Liver Mitochondria-Methods for yeast culture and isolation of yeast mitochondria were performed as described previously (22). Rat liver mitochondria were isolated from male Wistar rats (200 -300 g) (Centre d'Elevage et de Reproduction Janvier, Le Genest St. Isle, France) that had continuous free access to water and were either fed ad libitum on a standard laboratory chow diet (62% carbohydrate, 12% fat, and 26% protein in terms of energy) or starved for 48 h. Diabetes was induced by a single intraperitoneal injection of streptozotocin as described previously (16). All animals were kept on a light/dark cycle (light from 15:00 to 03:00 h) and were killed at 08:00 h. Rat liver mitochondria were isolated in an isolation buffer (0.3 M sucrose, 5 mM Tris-HCl, 1 mM EGTA, pH 7.4) using differential centrifugation, further purified on selfforming Percoll gradients, and resuspended in the isolation buffer as described previously (7). Protein concentration was determined by the method of Lowry with bovine serum albumin as a standard.
Chemical Cross-linking-Mitochondria were washed twice either in HSY buffer (10 mM Hepes, 0.6 M sorbitol, pH 8) (yeast mitochondria) or in HSR buffer (10 mM Hepes, 0.3 M sucrose, pH 8) (rat liver mitochondria), and finally resuspended in the respective buffer at a protein concentration of 2 mg/ml. The polar cross-linkers, sulfo-GMBS, sulfo-MBS, and sulfo-KMUS, used were dissolved in water. The apolar cross-linker, DSS, was dissolved in Me 2 SO. After addition of cross-linkers to a final concentration of either 50, 100, 200, 250, or 500 M to mitochondria (80 g), samples were incubated for 30 min at 4°C. Excess of cross-linkers was then quenched by addition of 6 l of TSY buffer (750 mM Tris-HCl, 0.6 M sorbitol, pH 8) (yeast mitochondria) or TSR buffer (750 mM Tris-HCl, 0.3 M sucrose, pH 8) (rat liver mitochondria). After incubation for 15 min at 4°C, mitochondria were recovered by centrifugation, resuspended in 0.1 M Na 2 CO 3 , pH 11.5, at a final protein concentration of 0.2 mg/ml, and incubated on ice for 30 min. After centrifugation at 177,000 ϫ g for 30 min at 4°C, integral membrane proteins were recovered in the pellet, analyzed by SDS-PAGE in a 4 -8% gradient gel, and immunoblotted.
Western Blot Analysis-Detection of proteins after blotting onto nitrocellulose was performed as described previously (22) using the ECL detection system (Pierce). The antibodies used were against the rat CPT1A (22), the mouse DHFR (BD Biosciences), the His tag (Novagen), the yeast cytochrome b 2 (gift from Prof. W. Neupert, Munich, Germany), and the rat Tom40 (gift from Prof. K. Mihara, Japan). For the generation of anti-CPT2 polyclonal antibody, peptide corresponding to the last 20 C-terminal residues of human CPT2 was synthesized, conjugated to keyhole limpet hemocyanin, and used to immunized New Zealand White rabbits (Neosystem, Strasbourg, France). The immunoblots were quantified using a Chemigenius apparatus (Syngene).
Chemicals-PCR reagents, restriction enzymes, and T4 DNA ligase were purchased from New England Biolabs (Ozyme, Saint-Quentin en Yvelines, France). Yeast culture media products were from Difco, and Zymolase 20T was from ICN Biomedicals, France. All cross-linkers were purchased from Pierce. Other chemicals were purchased from Sigma.

CPT1A Exists as an Oligomeric Complex within the OMM-
The first approach used to investigate the organization of rat CPT1A within the OMM was chemical cross-linking using intact mitochondria isolated from fed adult rat liver. After treatment with the noncleavable water-insoluble DSS (11.3 Å) cross-linker, CPT1A-containing cross-linking products were analyzed by SDS-PAGE and immunoblotting. Whereas the monomeric form of CPT1A (88 kDa) decreased as the concentration of DSS increased, both a major band corresponding to an apparent molecular mass ranging between 250 and 400 kDa and a minor band corresponding to an apparent molecular mass of ϳ100 kDa were detected (Fig. 1A). Treatment with the noncleavable water-soluble sulfo-GMBS (6.8 Å), sulfo-MBS (9 Å), and sulfo-KMUS (15.7 Å) cross-linkers also led to the generation of these two specific CPT1A adducts. In addition, another band with an apparent molecular mass of ϳ160 kDa was detected upon cross-linking with sulfo-GMBS and sulfo-MBS, which have a shorter spacer-arm length (Fig. 1A). These results suggested that within intact OMM of rat liver mitochondria, CPT1A interacts with protein(s) and may be assembled into complex(es).
To determine whether the high molecular cross-linked product could result from oligomerization of CPT1A or not, crosslinking experiments were performed using mitochondria isolated from S. cerevisiae expressing rat CPT1A. Indeed, the yeast model is not only devoid of any of the mitochondrial CPTs but also of all the mitochondrial enzymes necessary for mitochondrial LCFA oxidation because this pathway takes place in peroxisomes. Therefore, the use of the yeast heterologous expression model specifically allowed us to study the oligomeric state of rat CPT1A when expressed in this model. As shown in Fig.  1B, treatment of mitochondria isolated from yeast cells expressing rat CPT1A with the sulfo-GMBS, sulfo-MBS, and sulfo-KMUS cross-linkers resulted in the generation of a high molecular mass band (ϳ200 -360 kDa), whereas the ϳ100-kDa band could never be detected. These observations suggested that, independently of the existence of a mitochondrial LCFA oxidation "metabolon" or of other protein-protein interactions, CPT1A could form an oligomeric complex within the OMM.
Gel Filtration Chromatography Analysis of CPT1A Complex(es)-To demonstrate the existence of CPT1A complex(es), Superose 6 gel filtration chromatography experiments were performed after lysis of purified rat liver mitochondria with 1% (v/v) Triton X-100 under a low detergent to protein ratio. Internal protein control was the IMM COX IV (17 kDa), which is 1 of the 13 subunits of the cytochrome c oxidase (180 kDa) that exists as a dimer of monomers (25). Western blot analysis of the eluted fractions indicated that rat liver COX IV was eluted as a single peak ( Fig. 2A) corresponding to a molecular mass of ϳ400 kDa (Fig. 2B). These observations validated that our experimental conditions did not dissociate this well known mitochondrial complex. The native rat CPT1A was eluted as a single peak ( Fig. 2A), and the corresponding molecular mass of the detected complex was ϳ275 kDa (Fig. 2B). We next addressed whether this complex may represent an oligomeric and/or multiproteic CPT1A complex by performing gel filtration experiments using solubilized mitochondrial extracts from yeast cells expressing rat CPT1A. Internal protein control for these experiments was the yeast cytochrome b 2 (cyt b 2 ) protein (52 kDa) that is assembled into a 250-kDa complex within the IMM (26). As expected, Western blot analysis of the eluted fractions indicated that yeast cyt b 2 was eluted as a single peak ( Fig. 2A) corresponding to a molecular mass of ϳ260 kDa (Fig.  2B). As observed for the native protein, the yeast-expressed CPT1A was also fractionated as an ϳ275-kDa complex (Fig. 2,  A and B). Similar results were obtained when rat liver or yeast mitochondria were solubilized with 1% (w/v) digitonin (results not shown). Because CPT1A has a molecular mass of 88 kDa, the results obtained using the yeast-expressed CPT1A revealed that CPT1A forms a homotrimeric complex within the OMM. In the native environment, i.e. rat liver OMM, the size of the ϳ275-kDa CPT1A species ( Fig. 2A) indicated that this trimeric organization is also preserved following solubilization and gel filtration chromatography, whereas other protein-protein interactions, as suggested by our cross-linking experiments (Fig. 1A), were likely disrupted under this experimental procedure.
Native Gel Electrophoresis Analysis of CPT1A Complex(es)-To further examine the complex state(s) of CPT1A, detergentsolubilized rat liver mitochondria were subjected to BN-PAGE followed by Western blotting. Whatever the detergent (Triton X-100 or digitonin) used, the major fraction of CPT1A migrated as an ϳ550 -620-kDa complex (Fig. 3A). In addition, bands with a much lower intensity were observed and estimated at ϳ130 -150 and ϳ90 kDa, the latter most likely corresponding to the monomeric form of CPT1A. CPT1A complexes solubilized with Triton X-100 were observed to migrate faster in BN-PAGE than those solubilized with digitonin (Fig. 3A). Such  a difference between the two detergents could be explained by variation in the lipid content of the various complexes as reported previously (27). Upon solubilization of rat liver mitochondria with the denaturing SDS detergent, only the ϳ90-kDa signal could be detected but with a much higher intensity (Fig.  3A), indicating SDS-induced dissociation of CPT1A complexes. To validate our results, Tom40 (40 kDa) was used as an internal control because this membrane protein, which belongs to the OMM protein import machinery, is assembled into a ϳ400-kDa complex (28). As shown in Fig. 3A, Tom40 migrated at a molecular mass of ϳ400 kDa indicating that our experimental conditions allowed solubilization of mitochondrial complexes without their disruption.
We next addressed whether the ϳ550 -620-kDa species may represent an oligomeric and/or multiproteic CPT1A complex by performing BN-PAGE experiments using solubilized mitochondrial extracts from yeast expressing rat CPT1A. Unfortunately, a too high nonspecific background was generated using our anti-CPT1A antibody after BN-PAGE and Western blotting of the yeast extracts. To circumvent this problem, we overexpressed in S. cerevisiae a rat CPT1A harboring at its C terminus a polyhistidine tag. As reported previously (29), the presence of this tag did not alter the conformation state, the enzymatic activity, nor the malonyl-CoA sensitivity of CPT1A. 4 Mitochondria isolated from yeast cells expressing the Histagged CPT1A were then solubilized by Triton X-100 or digitonin and analyzed by BN-PAGE using an anti-histidine tag antibody. Fig. 3B clearly showed that, similarly to what observed for the native rat CPT1A, the His-tagged CPT1A was mainly recovered in a complex with an apparent molecular mass of ϳ530 -600 kDa. Moreover, the ϳ130 -150-kDa signal previously observed in rat liver mitochondria (Fig. 3A) was not detected, indicating that this complex was specific to the rat liver and probably corresponds to the association of CPT1A with other protein(s). The size of the ϳ530 -600-kDa complex observed in yeast extracts confirms CPT1A oligomerization but is suggestive of a homo-oligomer of six monomers. However, by taking into account the trimeric state observed using the gel filtration approach (Fig. 2, A and B), these results suggested that CPT1A trimers might be arranged as a dimer that likely dissociated upon gel filtration while it is stabilized in the BN-PAGE conditions known to allow separation of protein complexes under native conditions (24). This is supported by the observation in rat liver mitochondria of an ϳ130 -150-kDa CPT1A complex only in BN-PAGE (Fig. 3A) but not in gel filtration experiments (Fig. 2, A and B). In addition, although the size of the ϳ550 -620-kDa species detected in rat liver mitochondria (Fig. 3A) is also suggestive of a hexamer (possibly dimer of trimers), we cannot rule out the presence of additional proteins in the complex, as pointed out by the cross-linking experiments (Fig. 1A). Taken together, these results clearly indicate for the first time that, independently of other proteinprotein interactions, CPT1A self-assembles into an oligomeric complex and that the fundamental unit of its quaternary structure is likely a homotrimer.
CPT1A Oligomeric State Is Not Altered by Starvation and Diabetes-To determine whether physiopathological situations that are associated with changes in malonyl-CoA sensitivity are able to alter the oligomeric state of CPT1A, liver mitochondria were isolated from control fed, 48-h starved or streptozotocin-induced diabetic rats. As reported previously (16), starvation and diabetes increased CPT1A protein level and activity (results not shown) and decreased its sensitivity to malonyl-CoA inhibition, the IC 50 value for malonyl-CoA being 11.5-and 4.7-fold higher, respectively, than that measured in mitochondria isolated from fed rat liver (fed, 5.24 Ϯ 0.31; starved, 60.57 Ϯ 14.49; diabetic, 24.70 Ϯ 3.08 M). Gel filtration analysis of Triton X-100-solubilized mitochondria isolated from 48-h starved and diabetic rat livers (Fig. 4A) show similar elution profiles for CPT1A to that observed with mitochondria isolated from fed rat liver (Fig. 2, A and B), with CPT1A being fractionated as a ϳ275-kDa complex. Moreover, the CPT1A migration pattern in BN-PAGE was also similar both in liver mitochondria isolated from fed, 48-h starved and diabetic rats (Fig. 4B). Therefore, we conclude that changes in malonyl-CoA sensitivity, such as those occurring during starvation and diabetes, are not associated with a modification in the oligomeric state of CPT1A.
CPT1A Oligomerization Is Mediated by Its N-terminal Domain-We have shown previously that the N-terminal domain of CPT1A, encompassing TM1 and TM2, mediates both mitochondrial protein targeting and import into the OMM (7). Indeed, residues 123-147, located immediately downstream of TM2, function as a noncleavable matrix-targeting signal, whereas TM2 at least acts as a stop-transfer sequence that stops and anchors the translocating CPT1A into the OMM (8). Therefore, to examine the role of this N-terminal domain in CPT1A oligomerization, we used various deleted CPT1A or fusion proteins that have been reported previously to be specif-  ically imported into the OMM, their C termini being exposed to the cytosol as for the full-length CPT1A (7,8). As expected, these proteins, which are presented schematically in Fig. 5A, were all expressed in S. cerevisiae mitochondria at the predicted sizes (Fig. 5B) and were then analyzed by gel filtration experiments.
We first examined whether the N-terminal domain of CPT1A could mediate oligomerization of the mature form of CPT2 (which lacks the matrix-targeting signal and hence is incompetent for mitochondrial import). We used pOM29-CPT2, which corresponds to the fusion of the signal anchor sequence of S. cerevisiae Tom70, pOM29, to the mature form of CPT2 (Fig. 5A) as a control of dimerization because pOM29 contains a structural motif of dimerization (23). Superose 6 gel filtration experiments indicated that CPT1-(1-147)-CPT2 and pOM29-CPT2 were respectively fractionated as an ϳ288and ϳ150-kDa complex (Fig. 6A). Taking into account their respective molecular weight (88 and 70 kDa), these results indicated that the N-terminal domain of CPT1A triggers trimerization of CPT2, whereas pOM29 induces its dimerization, as expected.
We then ensured that the presence of a carnitine acyltransferase domain, i.e. CPT2, C-terminally to the first 147 N-terminal residues of CPT1A did not interfere with the formation of the complex. To this end, the N-terminal domain of CPT1A was fused to DHFR, a cytosolic monomeric protein (Fig. 5A). The elution profile of CPT1-(1-147)-DHFR (43 kDa) in a Superose 12 gel filtration column showed that this fusion protein was eluted as a single peak with a corresponding molecular mass estimated at ϳ150 kDa (Fig. 6B). Thus, we conclude that the N-terminal domain of CPT1A confers the ability to induce trimerization of nonrelated carnitine acyltransferase proteins such as DHFR.
The Second Half of the N-terminal Domain of CPT1A Is Sufficient for Oligomerization-We next wanted to determine whether each half of this N-terminal domain could play an equivalent role in CPT1A oligomerization. However, residues 1-82 (containing the cytosolic N terminus and TM1) have been shown to be unable to mediate mitochondrial protein targeting and OMM import (8), hence preventing the expression of either CPT1⌬83-148 or CPT1-(1-82)-DHFR in yeast mitochondria. Therefore, only the importance of the second half of the N-terminal domain of CPT1A could be presently examined by expressing CPT1⌬1-82 and CPT1-(97-147)-DHFR in yeast (Fig. 5, A and B). Gel filtration analysis revealed that CPT1⌬1-82 protein (82 kDa) was eluted as a single peak and that the corresponding molecular mass of the detected complex was ϳ245 kDa (Fig. 6C). Therefore, deletion of the first 82 N-terminal residues in CPT1A neither abrogated nor altered its trimerization. Conversely, fusion of the second half of the N-terminal domain of CPT1A to DHFR led to a resulting CPT1-(97-147)-DHFR protein (29 kDa) that was fractionated as an ϳ190-kDa complex (Fig. 6D), which surprisingly was indicative of a hexamer rather than a trimer. Thus, deletion of the first half of the N-terminal domain in CPT1-(1-147)-DHFR, but not in CPT1A, resulted in doubling the number of monomers able to self-assemble into an oligomeric complex. Altogether, these observations clearly indicate that the second half of the N-terminal domain of CPT1A that encompassed TM2 is sufficient to mediate oligomerization.

DISCUSSION
This study was aimed at investigating the quaternary structure of the rat CPT1A within the OMM, an issue that surprisingly has never been explored since its cDNA cloning in 1993 (5). By using chemical cross-linking, we clearly show that CPT1A in its native environment, i.e. in intact rat liver mitochondria, interacts with protein(s) and is assembled into complex(es). Further size chromatography analysis using solubilized mitochondrial extracts shows the existence of a trimeric CPT1A complex that might be arranged as a dimer, as suggested by our BN-PAGE experiments. Therefore, as most mitochondrial proteins, rat CPT1A does not function as a monomer but has to self-assemble into an oligomeric structure (likely a dimer of trimer) to become functional. Indeed, such an oligomeric organization has already been described for the rat liver short, medium, and long chain acyl-CoA dehydrogenases (tetramers, i.e. dimers of dimers) (30), the human L-3-hydroxyacyl-CoA dehydrogenase (dimer) (31), most members of the hydratase/isomerase (crotonase) superfamily (hexamers, i.e. dimers of trimers) (32), the bovine liver monoamine oxidase B (hexamer, i.e. trimer of dimers) (33), and the yeast cytochrome b 2 (tetramer) (26). Then we took advantage of the heterologous yeast expression model that allows us to study, because of the absence of fatty acid oxidation pathways in mitochondria, the oligomeric state of the yeast-expressed rat CPT1A independently of the existence of putative interactions with other proteins. This allowed us not only to confirm the oligomeric state of CPT1A but also, using several yeast-expressed chimeric CPTs, to demonstrate that the N-terminal domain (i.e. the first 147 N-terminal residues) of CPT1A encompassing its two TM segments (TM1 and TM2) mediates trimerization of either CPT2 (CPT1-(1-147)-CPT2) or nonrelated-carnitine acyltransferase such as DHFR (CPT1-(1-147)-DHFR). Moreover, the observation that deletion of the extreme N terminus plus TM1 in CPT1A did not abrogate trimerization whereas similar deletion in CPT1-(1-147)-DHFR resulted in doubling the number of assembled monomers suggested the following: (i) the second half of the N-terminal domain of CPT1A encompassing TM2 is sufficient to mediate oligomerization, and (ii) the latter is likely limited to a trimer by the presence of the large catalytic C-terminal domain of CPT1A on the cytosolic face of mitochondria. Altogether, these results demonstrated for the first time that rat CPT1A is not a monomeric enzyme but exists within the OMM as an oligomer whose fundamental structural unit is likely a trimer.
The current investigations of the quaternary structure of CPT1A provide new insights into the functionality of this key LCFA oxidation enzyme. The first functional relevance of the CPT1A oligomeric state is that it may be the clue for its positive cooperativity with respect to palmitoyl-CoA and/or its positive cooperative inhibition by malonyl-CoA (3,18,19,21), similarly to what described previously for the bovine mitochondrial glutamate dehydrogenase (dimer of trimers) (34). Moreover, according to the position of the cytosolic N terminus of CPT1A toward its catalytic site within the C-terminal domain, one could propose that this N terminus could more or less protrude from each monomer and serves as an inter-subunit communication during positive cooperativity and allosteric regulation.
The second insight concerns the changes in malonyl-CoA sensitivity of CPT1A depending on the physiopathological state of the animals. We have previously demonstrated that CPT1A adopts different conformational states that differ in their intramolecular N/C interactions, and that this determines its degree of malonyl-CoA sensitivity (16). Moreover, Zammit and co-workers (35) have recently shown that the sequence spanning the intermembrane loop-TM2 boundary determines the disposition of this TM in the membrane so as to alter the conformation of the C-terminal catalytic domain, and thus malonyl-CoA sensitivity. These observations may be mediated by changes in the physical properties of the membrane. Indeed, several studies have established that malonyl-CoA sensitivity of CPT1A is dependent both in vivo and in vitro on the molecular order of the membrane hydrophobic core and on membrane fluidity (16,29,36). This study clearly shows that the decrease in the effectiveness of malonyl-CoA as an inhibitor during starvation or streptozotocin-induced diabetes is not associated with a change in the oligomeric state of CPT1A. However, we cannot exclude that an increased membrane fluidity, which is presumed to facilitate the freedom of the relative movement of each monomer within the CPT1A trimer, may elicit a differential positioning of the monomers. This in turn could participate two-domain architecture that unexpectedly share the same backbone fold with that of both the enoyl-CoA hydratase (32), the chloramphenicol acetyltransferase (46), and the dihydrolipoyl trans-acetylase (47) monomer. All of these enzymes, as well as other acyltransferase enzymes such as the galactoside acetyltransferase (48), the serine acetyltransferase (49), and the acetyltransferase/uridyltransferase (50), are in fact trimeric enzymes. Moreover, their crystallographic structure indicated that the trimeric association forms a channel in which the catalytic reaction occurs, the active site of the enzymes being located in the subunit interface. Two CrAT (45) and COT (43) molecules have been observed in their crystallographic asymmetric unit, but the contacts among the crystallographically related dimers have been reported to be generally weak and hydrophilic in nature. Alignment sequence and in silico structural comparison of the catalytic C-terminal domain of CPT1A (excluding its N-terminal domain encompassing the two TMs) with those of CrAT and COT indicated that the active region is largely conserved (11,39). However, our previous structural modeling studies of the catalytic domain of human CPT1A (51) revealed the existence, outside the active site, of significant insertions in several of the surface loops, as compared with mouse CrAT. Therefore, in addition to the presence of TMs (in particular TM2), these secondary structural elements of CPT1A may also be important features for the establishment of tight subunit contacts in the CPT1A trimer. However, their functional exploration, as well as involvement of differences in the surface electrostatic charge and electrostatic potential distributions, requires the specific crystal structure of the CPT1A enzyme.