The Extreme C Terminus of Rat Liver Carnitine Palmitoyltransferase I Is Not Involved in Malonyl-CoA Sensitivity but in Initial Protein Folding*

We previously reported that the N-terminal domain (1–147 residues) of rat liver carnitine palmitoyltransferase I (L-CPTI) was essential for import into the outer mitochondrial membrane and for maintenance of a ma-lonyl-CoA-sensitive conformation. Malonyl-CoA binding experiments using mitochondria of Saccharomyces cerevisiae strains expressing wild-type L-CPTI or previously constructed chimeric CPTs (Cohen, I., Kohl, C., McGarry, J.D., Girard, J., and Prip-Buus, C. (1998) J. Biol. Chem. 273, 29896–29904) indicated that the N-terminal domain was unable, independently of the C-terminal domain, to bind malonyl-CoA with a high affinity, suggesting that the modulation of malonyl-CoA sensitivity occurred through N/C intramolecular interactions. To assess the role of the C terminus in malonyl-CoA sensitivity, a series of C-terminal deletion mutants was generated. The kinetic properties of (cid:1) 772–773

The carnitine palmitoyltransferase I (CPTI, 1 EC 2.3.1.21) catalyzes the rate-limiting step of mitochondrial long-chain fatty acid oxidation in all mammalian tissues that converts long-chain acyl-CoA to acylcarnitine (1). A unique feature of CPTI is its potent inhibition by malonyl-CoA, the first committed intermediate of fatty acid biosynthesis (2). This provides a mechanism for physiological regulation of ␤-oxidation in liver and other tissues and for cellular fuel sensing based on the availability of fatty acids and glucose (1,3,4). In recent years the concept has emerged that lipid disorders, such as inefficient fatty acid oxidation, may contribute in the etiology of insulin resistance, non-insulin-dependent diabetes mellitus, coronary artery disease, and other heart diseases (5)(6)(7)(8). Acute metabolic complications of uncontrolled insulin-dependent diabetes mellitus, life-threatening diabetic ketoacidosis, mainly stem from excessive fatty acid oxidation (9). An ideal design of a drug to control the CPT system in such metabolic disorders has not come into being (10), largely because the molecular mechanisms underlying CPTI inhibition by malonyl-CoA have not thoroughly been understood.
In mammalian tissues, there are two CPTI isoforms, the liver type (L-CPTI) and the muscle type (M-CPTI), encoded by distinct genes (1). The M-CPTI is ϳ30 -100-fold more sensitive to malonyl-CoA inhibition than L-CPTI. L-CPTI is a polytopic outer mitochondrial membrane (OMM) protein harboring two hydrophobic transmembrane segments (TM1 and -2) (11). Its membrane topology is such that both the N terminus (residues 1-47) and the C-terminal catalytic domain (residues 123-773) are located on the cytosolic face of mitochondria (12). The N-terminal domain (1-147 residues) of L-CPTI is not only responsible for mitochondrial targeting and import into the OMM but is also involved in maintenance of a folded active and malonyl-CoA-sensitive conformation of the enzyme (13). Recently, TM2 was shown to be essential for the correct folding of the catalytic domain of L-CPTI, suggesting that some N/C intramolecular interactions may be directly involved in the establishment and/or maintenance of the native functional conformation of L-CPTI (14).
The malonyl-CoA binding site as well as the molecular basis for malonyl-CoA inhibition remain largely unknown despite clues that have begun to emerge from functional mutagenesis studies. Deletion of the highly conserved 18 N-terminal residues of L-CPTI or mutation of Glu 3 and His 5 led to a decreased malonyl-CoA sensitivity and impairment of malonyl-CoA binding (15)(16)(17). However, whether Glu 3 participates directly to malonyl-CoA binding or allows the extreme cytosolic N terminus to interact with the catalytic C-terminal domain in order to maintain a conformational malonyl-CoA binding site is still unclear. We have previously shown that the N-terminal domain of L-CPTI cannot confer malonyl-CoA sensitivity to the malonyl-CoA-insensitive CPTII (13), which allowed two hypotheses to be formulated. The first one was that, if malonyl-CoA binds only the N terminus of L-CPTI, the CPTII component of the CPTI (1-147)-CPTII chimera must have a tertiary conformation that is unable to interact with this domain. The second hypothesis was that the C-terminal region of L-CPTI is critical for malonyl-CoA binding and that the necessary site(s) is not present in CPTII. Since this observation, different studies have indicated that (i), the different malonyl-CoA sensitivities of the CPTI isoforms result primarily from differences in their C termini (16,18,19) and (ii), the nature of the cytosolic N/C interaction of the L-CPTI is involved in the degree of malonyl-CoA sensitivity (20,21). The existence of such a physical N/C interaction was recently supported by the identification of a highly trypsin-resistant 60-kDa folded core within the catalytic C-terminal domain that is hidden in the native protein by its cytosolic N-terminal residues (14). However, whether the extreme N terminus (and particularly Glu 3 ) participates directly to malonyl-CoA binding or allows the extreme cytosolic N terminus to interact with the catalytic C-terminal domain in order to maintain a conformational malonyl-CoA binding site is still unclear.
In the present study, we examined whether the N-terminal domain of L-CPTI exhibits a high affinity for malonyl-CoA binding independently of the presence of the C-terminal domain of L-CPTI, and we investigated for the first time whether the extreme C terminus of L-CPTI, which exhibits notable differences with its M-CPTI counterpart, could be involved in the N/C intramolecular interactions that determine the degree of malonyl-CoA sensitivity. For this purpose, three C-terminal deletions of L-CPTI, ranging in size from 2 to 31 residues, were generated and expressed in Saccharomyces cerevisiae to compare their functional and conformational properties. The results outlined below reveal that the high malonyl-CoA affinity binding site does not reside within the N-terminal domain of L-CPTI and that the extreme C terminus of L-CPTI is not involved in malonyl-CoA sensitivity but plays an unexpected role in the initial folding of the enzyme.

EXPERIMENTAL PROCEDURES
Construction of Expression Plasmids-The S. cerevisiae strains expressing rat L-CPTI, pOM29-CPTII, pOM29-CPTI⌬148, and CPTI (1-147)-CPTII were obtained as described previously (13,22). Briefly, pOM29-CPTII and -CPTI⌬148 correspond to the fusion of pOM29 (a specific OMM signal anchor sequence), to the mature form of CPTII, or to L-CPTI lacking the first 148 N-terminal residues, respectively. CPTI (1-147)-CPTII is the fusion of the mature form of CPTII to the Nterminal domain of L-CPTI. Escherichia coli DH5␣ strain was used to propagate various plasmids. The yeast expression vector pYeDP1/8 -10 containing the full-length rat L-CPTI cDNA (pYeDP1/8 -10-L-CPTI) (22) was used to generate the deletion mutants. All DNA manipulations (restriction and ligation) were performed according to the instructions provided by the manufacturer's protocols for the respective enzymes. The C-terminal deletion rat L-CPTI mutants were constructed as follows.
CPTI⌬772-773-PCR was performed to copy a 336-nucleotide stretch of the 3Ј-coding sequence of the rat L-CPTI by using the 5Јprimer (5Ј-TATGTGGTGTCCAAGTAT-3Ј) located upstream the unique SalI restriction site of L-CPTI cDNA and the 3Ј-primer (5Ј-TC-CCCGCGGTTAAGAATTGATGGTGAG-3Ј) introducing a stop codon and a SacII site immediately downstream the codon encoding Ser 771 . The PCR product was cut by SalI and SacII and ligated into pYeDP1/ 8 -10-L-CPTI cut by the same enzymes. This construct encodes an L-CPTI lacking the last two C-terminal amino acids.
CPT Assay-CPT activity was assayed at 30°C as palmitoyl-L-[methyl-3 H]carnitine formed from L-[methyl-3 H]carnitine (200 M; 10 Ci/mol) and palmitoyl-CoA (80 M) in the presence of 1% bovine serum albumin (w/v) as described previously (22). Malonyl-CoA concentration varied over 0.01 to 150 M for estimation of the IC 50 value. The apparent K m for carnitine was measured at 600 M palmitoyl-CoA with 5-800 M carnitine and the apparent K m for palmitoyl-CoA at 200 M carnitine with 10 -900 M palmitoyl-CoA in the presence of a fixed molar ratio of palmitoyl-CoA/albumin (6.1:1) (22).
Malonyl-CoA binding was determined by a modified centrifugation assay as described previously (23). Isolated mitochondria (400 g of protein) from wild-type and mutants were resuspended in 0.4 ml of ice-cold medium containing 250 mM sucrose, 60 mM KCl, 10 mM Hepes (pH 7.4), 1 mM EGTA, 1% bovine serum albumin (w/v), 0.01-5 M [2-14 C] malonyl-CoA (56 mCi/mmol) (Amersham Biosciences) in the absence or presence of 200 M unlabelled malonyl-CoA. Incubations were started by addition of the mitochondria and were continued for 20 min at 0°C with gentle mixing at 4-min intervals. Bound and free malonyl-CoA were then separated by centrifugation at 15,000 ϫ g for 10 min at 4°C. The supernatants were discarded, and 0.25 ml of 1 M KOH was immediately added to the pellets. Solubilization of the mitochondria was facilitated by heating at 50°C for 30 min, after which time the contents of the tube, together with 0.8 ml of water wash, were transferred to counting vials and assayed for radioactivity after addition of 10 ml of Aquasol. The final malonyl-CoA binding values for the wild-type and mutants were corrected for background malonyl-CoA binding by the yeast control strain that carried the vector but without the CPTI insert. Saturation binding experiments were analyzed with the non-linear least squares curvefitting procedure of the EBDA/LIGAND program (Biosoft Elsevier, Cambridge, UK). After subtraction of nonspecific malonyl-CoA binding to the CPTI-free yeast control strain, data were always best fitted according to a one-site model. When needed, the validity of a one-site versus two-site model was verified using an F-test.
Assessment of the Folding State of CPTI Mutants-Proteolytic analysis of the full-length L-CPTI and mutants was performed as described previously (22) with the following modifications. Eighty g of mitochondria isolated from yeast strains expressing the different proteins were resuspended in SH buffer (0.6 M sorbitol, 20 mM Hepes-KOH, pH 7.4) at a concentration of 0.5 mg of protein/ml in the absence or in the presence of 30 g/ml trypsin. Samples were kept on ice for 30 min, and then soybean trypsin inhibitor (STI) was added in a 30-fold excess. After 10 min, mitochondria were reisolated, washed once with SH buffer containing 1 mM EDTA and 1 mg/ml STI, and then lysed directly in Laemmli buffer. Samples were analyzed by SDS-PAGE and immunoblotting.
Western Blot Analysis-Aliquots of proteins were subjected to SDS-PAGE (24) in an 8% gel. The detection of proteins after blotting onto nitrocellulose was performed as described previously (22) using the ECL detection system (Pierce) according to the supplier's instructions.
Miscellaneous and Chemicals-Protein concentration was determined by the method of Ref. 25 with bovine serum albumin as a standard. All restriction enzymes and T4 DNA ligase were purchased from Biolabs (Ozyme, Saint-Quentin en Yvelines, France). PCR reagents and T4 DNA polymerase were obtained from Invitrogen. Yeast culture media products were from Difco, and Zymolase 20T was from ICN Biomedicals, France. Others chemicals were purchased from Sigma.
Statistics-Results are expressed as means Ϯ S.E. Statistical analysis was performed using the Mann-Whitney U test.

Malonyl-CoA Sensitivity and [ 14 C]Malonyl-CoA Binding of
Chimeric CPTs-The first purpose of this study was to determine whether the N-terminal domain (the first 147 N-terminal amino acids) of L-CPTI was able to bind malonyl-CoA with a high affinity, independently of the presence of the C-terminal domain. For this purpose, we further characterized some chimeric CPTs previously generated and expressed in S. cerevisiae (13), a system devoid of endogenous CPT enzyme (22,26). CPTI (1-147)-CPTII (fusion of the mature form of CPTII to the Nterminal domain of L-CPTI) offered the possibility of directly addressing the former question. pOM29-CPTII and pOM29-CPTI⌬148 were used as negative control proteins, pOM29 being a specific OMM signal anchor sequence (27). All of these chimeric proteins were expressed in yeast at similar steadystate levels and were located at the OMM, their C terminus exposed to the cytosol (13). As previously reported (13), both pOM29-CPTII and CPTI (1-147)-CPTII were catalytically active, whereas pOM29-CPTI⌬148 retained only 8% of that observed with L-CPTI (Table I). In contrast to L-CPTI, which exhibited an IC 50 for malonyl-CoA of 0.6 M (Table I and Fig.  1), pOM29-CPTII, pOM29-CPTI⌬148, and CPTI (1-147)-CPTII were largely insensitive to malonyl-CoA whatever the malonyl-CoA concentration tested. Malonyl-CoA binding to mitochondria from yeast strains expressing L-CPTI and chimeric CPT was clearly saturable ( Fig. 2A). As shown by the Scatchard plots in Fig. 2B, saturation binding experiments were resolved into a single high affinity site (L-CPTI) or a single low affinity site (chimeric CPT). Replacement of the N-terminal domain of L-CPTI by pOM29 completely abolished the high affinity malonyl-CoA binding (Fig. 2B), increasing the K D value by 20-fold whereas the B max value decreased only slightly (Table I). In agreement with the fact that CPTII does not possess a malonyl-CoA binding domain, pOM29-CPTII exhibited a K D value 30fold higher than that of L-CPTI (Table I)  malonyl-CoA binding ( Fig. 2B and Table I). This clearly shows that the insensitivity of CPTI (1-147)-CPTII to malonyl-CoA resulted from the absence of high affinity malonyl-CoA binding. Fig. 3. In contrast to the N terminus, the C terminus exhibits a low degree of sequence conservation because no more than 13 of the last 33 residues are identical. The major difference between the L-and M-CPTI isoforms is the presence of two lysine residues at the C-terminal end of all known mammalian L-type isoforms. Secondary structure prediction methods predict that residues 747-768 of the rat L-CPTI form an amphipathic ␣-helix, as for the corresponding residues in the other CPTI isoforms. Therefore, three deletion mutants of rat L-CPTI were constructed in which the last 2 (2 lysines), 7 (downstream the ␣-helix), or 31 (encompassing the ␣-helix) C-terminal residues were deleted (Fig. 3). All these constructs, as well as the full-length L-CPTI, were expressed in S. cerevisiae.

Generation of C-terminal Deletion Mutants of Rat L-CPTI and Expression in S. cerevisiae-Amino acid sequence alignment of the last 33 C-terminal residues of CPTI isoforms from different mammalian species is shown in
Western blot analysis of mitochondria isolated from yeast cells expressing the different proteins using a polyclonal antibody directed against residues 317-430 of L-CPTI is shown in Fig. 4. All proteins were synthesized at predicted sizes and were expressed at similar steady-state levels. Subcellular fractionation experiments did not reveal any change in the mitochondrial targeting of these expressed proteins (results not shown). This strengthened the fact that only the N-terminal domain of L-CPTI is involved in its import into the OMM (13,14). Table II, ⌬772-773 and ⌬767-773 retained significant CPT activity that was 96 and 99%, respectively, of that observed for the wild-type L-CPTI. By contrast, deletion of the last 31 C-terminal residues resulted in a protein that was totally inactive (Table II), despite similar levels of protein when compared with wild-type L-CPTI (Fig. 4). Deletion of the last 2 or 7 C-terminal residues did not affect malonyl-CoA sensitivity, whatever the concentration of the inhibitor tested (Fig. 5A), and the IC 50 values for malonyl-CoA of the wild-type L-CPTI, ⌬772-773, and ⌬767-773 were similar (Table II). All active CPTI mutants exhibited normal saturation kinetics when the carnitine concentration varied relative to a fixed concentration of palmitoyl-CoA (Fig. 5B) or when palmitoyl-CoA concentration varied when the molar ratio of palmitoyl-CoA:albumin was fixed at 6.1:1 (Fig. 5C). The calculated K m for carnitine and palmitoyl-CoA of ⌬772-773 and ⌬767-773 were similar to those of the wild-type L-CPTI (Table II).

CPT Activity, Malonyl-CoA Sensitivity, and Kinetic Parameters of the C-terminal Deletion Mutants-As shown in
Assessment of the Folding State of the C-terminal Deletion Mutants-To understand the abolition of CPT activity upon deletion of the last 31 C-terminal residues, we examined the possibility that the conformational state of L-CPTI could have been affected. We have previously shown that native or yeastexpressed L-CPTI exhibited a highly folded conformation resistant to trypsin proteolysis (13,14,22). When intact mitochondria isolated from the different expressing yeast strains were incubated in the presence of trypsin, both ⌬772-773 and ⌬767-773 remained largely resistant to the protease treatment, indicating that their conformation was not dramatically affected by the deletion (Fig. 6). By contrast, ⌬743-773 was totally digested by trypsin (Fig. 6) without apparent detectable proteolytic fragments (results not shown). Thus, in the absence of the last 31 C-terminal residues, the C-terminal domain of L-CPTI did not harbor the highly folded core characteristic of the native enzyme. Therefore, we concluded that, in the case of ⌬743-773, disappearance of functional activity directly resulted from an unfolded state of the protein.

N-terminal Domain of L-CPTI and Malonyl-CoA Sensitivity-
The N-terminal domain of rat L-CPTI is not only responsible for protein targeting to the outer mitochondrial membrane but is also essential for malonyl-CoA sensitivity (13-15, 17, 20). Inhibition of L-CPTI activity by malonyl-CoA involves two sites of malonyl-CoA interaction. The first one is the low affinity site, near the catalytic acyl-CoA-binding domain (28,29) as recently modeled by Ref. 30, whereas the second site, the high affinity site, is separated from the active site and does not compete with acyl-CoA (31)(32)(33)(34). Although previous studies have shown that the N-terminal domain influences the degree of malonyl-CoA sensitivity, it has not been ascertained whether the N-terminal residues of L-CPTI contribute directly to a malonyl-CoA binding site. Using heterologous expression in S. cerevisiae, we have previously shown that fusion of the Nterminal domain of L-CPTI (1 to 147 amino acids) to the mature form of the malonyl-CoA-insensitive CPTII allowed a specific OMM targeting of the resulting protein leaving the CPTII moiety on the cytosolic face of the mitochondria, as for the catalytic domain of L-CPTI (12). This chimeric CPTI (1-147)-CPTII protein was functionally active but totally malonyl-CoAinsensitive (13). In the present study, we show that chimeric  CPTI (1-147)-CPTII does not exhibit a high affinity for malonyl-CoA binding when compared with pOM29-CPTII. These results indicate that (i), the N-terminal domain of rat L-CPTI has not the ability by itself (i.e. in the absence of the C-terminal domain) to bind malonyl-CoA and (ii), the high affinity binding site for malonyl-CoA likely involves either residues located both in the N-and C-terminal domains or only residues within the C-terminal domain. In the first hypothesis, physical N/C interactions are necessarily involved in order to constitute the malonyl-CoA binding site. Any modification within the N terminus would then be expected to alter profoundly the high affinity malonyl-CoA binding. Initial deletion mutation analysis of the conserved first 18 N-terminal residues of rat L-CPTI was in agreement with this statement (15,17), whereas further studies indicated that the mechanism of malonyl-CoA inhibition is more complicated because the cytosolic N terminus contained both positive and negative determinants of malonyl-CoA sensitivity (20). Moreover, functional characterization of several L-CPTI chimeras has predicted that malonyl-CoA binding sites were located in the cytosolic C-terminal domain (16,19). If this is true, the cytosolic N terminus should (i), stabilize the high affinity malonyl-CoA binding site through its interaction with the catalytic C-terminal domain and (ii), modulate the degree of malonyl-CoA sensitivity through conformational changes that alter these N/C intramolecular interactions, as previously suggested (19,20). C Terminus of L-CPTI and Malonyl-CoA Sensitivity-Whatever the location of the high affinity malonyl-CoA binding site within L-CPTI, it might be a conformational site that is highly sensitive to interaction(s) with the N terminus. Which part(s) of the C-terminal domain of L-CPTI do interact with the N terminus? Because the negative charge of Glu 3 has been shown to be essential for mediating the inhibitory effects of malonyl-CoA (17, 21), we asked whether the positive charges of the 2 Lys residues present at the COOH terminus of all known L-CPTI species (1) could play a role in these N/C intramolecular interactions and/or explain the discrepancies in the degree of malonyl-CoA sensitivity between the L-and M-CPTI isoforms. The kinetic properties of the L-CPTI mutants lacking the last 2 or 7 C-terminal amino acids were indistinguishable from those of the wild-type. We concluded that the two highly conserved Lys residues in the C terminus of all known L-CPTI species are essential for neither functional activity nor malonyl-CoA sensitivity and hence are not involved in a physical interaction with the N terminus of the enzyme. To further investigate the role of the C terminus of L-CPTI, we constructed two other deletion mutants, ⌬743-773 and ⌬719 -773. Among the seven independent yeast clones transfected with pYeDP1/8-⌬719 -773, no protein expression was detected. 2 This was in contrast with the three other deletion mutants (⌬772-773, ⌬767-773, and ⌬743-773) for which we observed correct protein expression in all tested yeast clones. This observation suggested that the last 55 C-terminal amino acids are critical for protein stability. Moreover, ⌬743-773 mutant was correctly imported into the OMM but was not folded, as shown by its high sensitivity to trypsin proteolysis and, hence, was functionally inactive. The effect of this deletion was unexpected from our previous observations (14). Indeed, we have shown that once imported into the OMM, L-CPTI adopts its native functional conformation that is characterized by a highly folded state resistant to trypsin proteolysis (Fig. 7, a and b). However, when the loop connecting TM1 and -2 is first cleaved by the protease, such as during the swelling procedure (Fig. 7c), possible trypsin cleavage sites occur C-terminal to Arg 595 or Arg 598 and Lys 631 or Lys 634 , and the highly folded trypsin resistant core (likely to be in part the catalytic core) is still detected after solubilization by Triton X-100 in the presence of trypsin (Fig.  7d). This indicates that the last 139 -178 C-terminal amino acids of L-CPTI can be cleaved from the native protein without impairing the folding of the remaining C-terminal domain. By contrast, if deletion of the last C-terminal amino acids occurs previous to mitochondrial protein import, such as in ⌬743-773, then the resulting imported protein is unable to reach its correct folded conformation (Fig. 7e). This observation indicates that the last 31 C-terminal amino acids, which are predicted to form an amphipathic ␣-helice, are critical for initial protein folding of the catalytic C-terminal domain of L-CPTI. Once folded, the tertiary structure of this domain is thereafter stabilized by other intramolecular interactions independently of the last 31 C-terminal residues. The functional relevance of these results is of great importance and must be kept in mind when performing functional analysis of L-CPTI mutants. Indeed, not enough caution has been taken in differentiating between whether a specific mutation within the C-terminal domain directly affects catalytic activity without drastically altering protein folding (functional determinant) or whether it induces protein unfolding (structural determinant) that secondarily decreases functional activity. During their course of folding, proteins undergo different types of structural rearrangements, ranging from local to large-scale conformational changes that lead to a series of sequential transition folding states. In this protein folding pathway, cooperative interactions of specific residues may be critical in establishing a bonding network that transiently stabilizes the intermediate conformations (35,36). For instance, point mutations can increase local flexibility or affect kinetic folding without altering the average native conformation of the protein (37)(38)(39). Such a direct effect of mutations on the degree and/or rate of protein folding independently of function could be proposed for the L-CPTI-H277A and L-CPTI-H277A/H483A mutants described recently in Ref. 30. These mutants expressed in yeast showed different sensitivity to malonyl-CoA inhibition, depending on the time after galactose induction, the kinetics of inhibition by malonyl-CoA being affected only at the time of 1 h of induction but not after 20 h. Such time-dependent behavior of the mutants could be due to a slower kinetic of folding and not necessarily be linked to the level of protein expression and/or organization within mitochondrial membrane. In conclusion, the present study indicates that the last 31 C-terminal residues of rat L-CPTI are not involved in malonyl-CoA sensitivity but constitute a "secondary structural specificity determinant" that may prevent neighboring residues from adopting an alternate protein fold by acting as a folding initiation site.