Artigo Acesso aberto Revisado por pares

Structural Model of the Catalytic Core of Carnitine Palmitoyltransferase I and Carnitine Octanoyltransferase (COT)

2001; Elsevier BV; Volume: 276; Issue: 48 Linguagem: Inglês

10.1074/jbc.m106920200

ISSN

1083-351X

Autores

Montserrat Morillas, Paulino Gómez‐Puertas, Ramón Roca‐Tey, Dolors Serra, Guillermina Asins, Alfonso Valencia, Fausto G. Hegardt,

Tópico(s)

Mitochondrial Function and Pathology

Resumo

Carnitine palmitoyltransferase I (CPT I) and carnitine octanoyltransferase (COT) catalyze the conversion of long- and medium-chain acyl-CoA to acylcarnitines in the presence of carnitine. We propose a common three-dimensional structural model for the catalytic domain of both, based on fold identification for 200 amino acids surrounding the active site through a threading approach. The model is based on the three-dimensional structure of the rat enoyl-CoA hydratase, established by x-ray diffraction analysis. The study shows that the structural model of 200 amino acids of the catalytic site is practically identical in CPT I and COT with identical distribution of 4 β-sheets and 6 α-helices. Functional analysis of the model was done by site-directed mutagenesis. When the critical histidine residue 473 in CPT I (327 in COT), localized in the acyl-CoA pocket in the model, was mutated to alanine, the catalytic activity was abolished. Mutation of the conserved alanine residue to aspartic acid, A381D (in CPT I) and A238D (in COT), which are 92/89 amino acids far from the catalytic histidine, respectively (but very close to the acyl-CoA pocket in the structural model), decreased the activity by 86 and 80%, respectively. The Km for acyl-CoA increased 6–8-fold, whereas the Km for carnitine hardly changed. The inhibition of the mutant CPT I by malonyl-CoA was not altered. The structural model explains the loss of activity reported for the CPT I mutations R451A, W452A, D454G, W391A, del R395, P479L, and L484P, all of which occur in or near the modeled catalytic domain. Carnitine palmitoyltransferase I (CPT I) and carnitine octanoyltransferase (COT) catalyze the conversion of long- and medium-chain acyl-CoA to acylcarnitines in the presence of carnitine. We propose a common three-dimensional structural model for the catalytic domain of both, based on fold identification for 200 amino acids surrounding the active site through a threading approach. The model is based on the three-dimensional structure of the rat enoyl-CoA hydratase, established by x-ray diffraction analysis. The study shows that the structural model of 200 amino acids of the catalytic site is practically identical in CPT I and COT with identical distribution of 4 β-sheets and 6 α-helices. Functional analysis of the model was done by site-directed mutagenesis. When the critical histidine residue 473 in CPT I (327 in COT), localized in the acyl-CoA pocket in the model, was mutated to alanine, the catalytic activity was abolished. Mutation of the conserved alanine residue to aspartic acid, A381D (in CPT I) and A238D (in COT), which are 92/89 amino acids far from the catalytic histidine, respectively (but very close to the acyl-CoA pocket in the structural model), decreased the activity by 86 and 80%, respectively. The Km for acyl-CoA increased 6–8-fold, whereas the Km for carnitine hardly changed. The inhibition of the mutant CPT I by malonyl-CoA was not altered. The structural model explains the loss of activity reported for the CPT I mutations R451A, W452A, D454G, W391A, del R395, P479L, and L484P, all of which occur in or near the modeled catalytic domain. carnitine palmitoyltransferase carnitine octanoyltransferase polymerase chain reaction Carnitine palmitoyltransferase I (CPT I1; EC 2.3.1.21) and carnitine octanoyltransferase (COT; EC 2.3.1.137) facilitate the transport of long and medium chain acyl-CoA in mitochondria and peroxisomes, respectively. Both carnitine acyltransferases facilitate the generation of energy by β-oxidation of fatty acids in the organella in which they are present. Mammalian tissues express two different isoforms of CPT I, a liver isoform (L-CPT I) (1Esser V. Britton C.H. Weis B.C. Foster D.W. Mc Garry J.D. J. Biol. Chem. 1993; 268: 5817-5822Abstract Full Text PDF PubMed Google Scholar, 2Britton C.H. Schultz R.A. Zhang B. Esser V. Foster D.W. McGarry J.D. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 1984-1988Crossref PubMed Scopus (125) Google Scholar) and a heart/skeletal muscle isoform (M-CPT I) (3Yamazaki N. Shinohara Y. Shima A. Yamanaka Y. Terada H. Biochim. Biophys. Acta. 1996; 1307: 157-161Crossref PubMed Scopus (102) Google Scholar, 4Yamazaki N. Shinohara Y. Shima A. Terada H. FEBS Lett. 1995; 363: 41-45Crossref PubMed Scopus (110) Google Scholar). As an enzyme that catalyzes the first rate-limiting step in β-oxidation, CPT I is tightly regulated by its physiological inhibitor malonyl-CoA. In regulating CPT I, malonyl-CoA confers the ability to signal to the cell the availability of lipid and carbohydrate fuels (5McGarry J.D. Brown N.F. Eur. J. Biochem. 1997; 244: 1-14Crossref PubMed Scopus (1322) Google Scholar). CPT I has a critical metabolic role in general metabolism in heart, liver, and β-cells of the pancreas and is a potential target for the treatment of metabolic disorders involving diabetes and coronary heart disease. Peroxisomal COT is also inhibited in physiological conditions by malonyl-CoA (6A'Bhaird N.N. Ramsay R.R. Biochem. J. 1992; 286: 637-640Crossref PubMed Scopus (23) Google Scholar) but to a lesser extent than CPT I. Other enzymes of the family, which are not regulated by malonyl-CoA, are CPT II, which catalyzes long-chain acyl-CoA in the mitochondria, and carnitine acetyltransferase, which has acetyl-CoA as substrate.These enzymes have recently generated much interest, especially the spatial organization of CPT I in the mitochondrial outer membrane. CPT I is an integral membrane protein, and both the N and C termini project to the cytosol, since it has two trans-membrane segments within the first 130 N-terminal residues of its primary sequence (7Fraser F. Corstorphine C.G. Zammit V.A. Biochem. J. 1997; 323: 711-718Crossref PubMed Scopus (122) Google Scholar). Interaction between amino acids from the N and C termini may determine the kinetic characteristics of the enzyme, not only in the inhibitory effect of malonyl-CoA but also in the catalytic activity (8Jackson V.N. Cameron J.M. Fraser F. Zammit V.A. Price N.T. J. Biol. Chem. 2000; 275: 19560-19566Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). COT appears not to be an integral protein of peroxisomes. Sequence alignment between CPT I and COT shows that COT lacks the first 152 amino acid residues, which indicates that it has no transmembrane regions. However, the two show high sequence identity, which suggests a common genetic origin (9van der Leij F.R. Huijkman N.C. Boomsma C. Kuipers J.R. Bartelds B. Mol. Genet. Metab. 2000; 71: 139-153Crossref PubMed Scopus (85) Google Scholar).Although several attempts have been made to identify the malonyl-CoA site, few data are available on the structure of the catalytic site of carnitine acyltransferases. It has been proposed that a histidine residue is critical in catalysis (10Schmalix W. Bandlow W. J. Biol. Chem. 1993; 268: 27428-27439Abstract Full Text PDF PubMed Google Scholar). Site-directed mutagenesis experiments have demonstrated an essential catalytic role for histidine 372 in CPT II (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar) and the homologous histidine 327 in COT (12Morillas M. Clotet J. Rubı́ B. Serra D. Asins G. Ariño J. Hegardt F.G. FEBS Lett. 2000; 466: 183-186Crossref PubMed Scopus (19) Google Scholar). However the role of a histidine in catalysis has been questioned for both CPT I and COT (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar, 14Nic a'Bhaird N. Yankovskaya V. Ramsay R.R. Biochem. J. 1998; 330: 1029-1036Crossref PubMed Scopus (11) Google Scholar). Dai et al. observed that overexpressed CPT I in yeast treated with diethylpyrocarbonate did not decrease the enzyme activity at variance with wild type rat mitochondria, although differences in the folding of CPT I in yeast and in rat liver could lead to an alternative interpretation.Other amino acids residues have been also implicated in catalysis. Mutation of several conserved arginines (Arg388, Arg451) and tryptophans (Trp391, Trp452) comprised between amino acids 381 and 481 of CPT I decreased enzyme activity (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar). These authors suggested that this segment could be the putative palmitoyl-CoA binding site of CPT I. In this fragment, mutants D376A and D464A in rat CPT II were completely inactive (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar). As to the carnitine binding site, arginine 505 in beef COT, which lies outside the proposed palmitoyl-CoA binding site, has been implicated (15Cronin C.N. Eur. J. Biochem. 1997; 247: 1029-1037Crossref PubMed Scopus (33) Google Scholar).In this study, we propose three-dimensional models of the catalytic site comprising amino acids 368–568 in CPT I and 226–417 in COT. These models were made using an integrative approach of several threading procedures taking into account different parameters such as solvatation potentials, contacts, environment-specific substitution tables, and structure-dependent gap penalties. The fold recognition analysis using these procedures showed a common template for the catalytic site of the carnitine/choline acyltransferase family of proteins, corresponding to the structure of the enoyl-CoA hydratase enzyme (Protein Data Bank entry 2dub; Ref. 16Engel C.K. Kiema T.R. Hiltunen J.K. Wierenga R.K. J. Mol. Biol. 1998; 275: 847-859Crossref PubMed Scopus (102) Google Scholar); this motif was also found in CPT I and COT. The three-dimensional models for CPT I and COT are practically identical and show a common architecture for the catalytic site. Site-directed mutagenesis indicated that CPT I His473 and COT His327 are the catalytic residues. These histidines are located near the thioester bond in the acyl-CoA, which is broken in catalysis. Mutation of CPT I A381 and COT A238, located close to the catalytic histidine decreased the enzyme activity by 80–86% without modifying the sensitivity to malonyl-CoA.DISCUSSIONDespite efforts in biochemical characterization of carnitine palmitoyltransferases and their related enzymes, their mode of action is not completely understood, probably due to the lack of any structural characterization of the catalytic site. In the absence of an appropriate crystallized reference, some bioinformatics tools can be applied to obtain a structural model able to approximately address some important questions related to these proteins' activity.The threading or "remote homology design" is a three-dimensional structure prediction technique useful when there is not enough sequence similarity of the input sequence and a known three-dimensional structure and, therefore, the "homology modeling" is not applicable. The process adapts the sequence to different known foldings and evaluates the fitting. The meaning of "fitting" varies from one threading program to others: secondary structure coincidence, similar accessibility, or solvatation energy, etc. Methods of protein fold recognition attempt to detect similarities between protein three-dimensional structure that are not accompanied by any significant sequence similarity. There are many approaches, but the unifying theme is to try and find folds that are comparable with a particular sequence. Unlike sequence-only comparison, these methods take advantage of the extra information made available by three-dimensional structural data.To detect structural homologies between CPT I and COT and any other three-dimensional representation, we used an integrative approach of two programs, THREADER2 (22Jones D.T. Miller R.T. Thornton J.M. Proteins. 1995; 23: 387-397Crossref PubMed Scopus (76) Google Scholar) and FUGUE threading server (23Shi J. Blundell T.L. Mizuguchi K. J. Mol. Biol. 2001; 310: 243-257Crossref PubMed Scopus (1080) Google Scholar). Whereas the first one uses solvatation potentials and predicted contacts, the latter performs a fold recognition analysis using structural environment-specific substitution tables and structure-dependent gap penalties. The integration of both methodologies revealed that the three-dimensional fold of the central site of all acyl-CoA transferases can be structured in the same way as the enoyl-CoA hydratase monomer (Protein Data Bank entry 2dub, chain E). What was more important was to observe that CPT I and COT had nearly identical structural models for the central region, which putatively contains the catalytic site (amino acids 368–567 of CPT I and amino acids 226–417 of COT). The predicted secondary structure for the 200 amino acids that putatively contains the palmitoyl-CoA or decanoyl-CoA binding region consists of 6 α-helices and 4 β-sheets.Additional support of this fold as template is based on the fact that these proteins bind very similar ligands, all of them acyl-CoA derivatives; the crystal structure of the enoyl-CoA hydratase monomer includes precisely a molecule of octanoyl-CoA, used as inhibitor in this case, the natural substrate of COT. Enoyl-CoA hydratase (Protein Data Bank code 2dub), also known as crotonase, belongs to the enoyl-CoA hydratase/isomerase family. It is a homohexameric enzyme, located at the mitochondrial matrix. It catalyzes the second step in the mitochondrial fatty acid β-oxidation pathway, transforming the 3-hydroxyacyl-CoA into trans-2(or 3)-enoyl-CoA plus H2O.The model predicts that although CPT I Ala381(Ala238 in COT) is 92 amino acids away from the CPT I His473 (His327 in COT) it is very close to the catalytic histidines. The important decrease in activity (14–20% of residual activity with respect to the wild type) after mutation CPT I A381D or COT A238D confirms the function of these alanines and supports the three-dimensional model. The marked modification ofKm for acyl-CoA of the mutants supports the role of these alanines in locating the substrates to the catalytic site.In addition, these alanine residues had been implicated in CPT and COT activity by a tree determinant study of the complete alignment of the carnitine-choline acyltransferase family of proteins. In a protein family alignment, the positions considered as tree determinants, or subfamily conserved residues, are usually accepted to be implicated in key catalytic activities, being responsible for the different substrate specificities or enzymatic activities of the different subfamilies of the alignment (37Casari G. Sander C. Valencia A. Nat. Struct. Biol. 1995; 2: 171-178Crossref PubMed Scopus (346) Google Scholar, 38Pazos F. Sanchez-Pulido L. Garcı́a-Ranea J.A. Andrade M.A. Atrian S. Valencia A. Lundh D. Olsson B. Narayanan A. Biocomputing and Emergent Computation. World Scientific, Singapore1997: 132-145Google Scholar). A clear example of this type of analysis was made previously in this same family of proteins related to the carnitine versus choline affinity (44Cronin C.N. J. Biol. Chem. 1998; 273: 24465-24469Abstract Full Text Full Text PDF PubMed Scopus (25) Google Scholar). The structure-based alignment of the template sequence with all the other transferases sequences used in the extensive threading procedure can also be used to build three-dimensional models of all the other members of the family in the future.The group of Woldegiorgis (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar) suggested that the region comprised between 381–481 could be the putative palmitoyl-CoA binding site. Moreover, the abolition of activity after mutation of H473A strongly suggests that this His473 is the catalytic site. McGarry and co-workers (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar), after mutagenesis of homologous histidine in CPT II, proposed that this was the catalytic site. This was confirmed in a previous study in COT, in which mutant H327A abolished the catalytic activity (45Morillas M. Clotet J. Rubı́ B. Serra D. Ariño J. Hegardt F.G. Asins G. Biochem. J. 2000; 351: 495-502Crossref PubMed Google Scholar). Dai et al. (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar) questioned whether these histidines were the catalytic sites after the observation that chemical modification of mitochondria from yeast strains expressing L-CPT I and M-CPT I by diethylpyrocarbonate had no effect on catalytic activity. It is possible that expressed CPT I in yeast mitochondria makes this histidine inaccessible for diethylpyrocarbonate modification. Brown et al. (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar) suggested that a charge-relay system involving this His coupled with an Asp extracts a proton from the C-3 hydroxyl group of carnitine, allowing for nucleophilic attack of the resulting oxyanion of the carbonyl group of the acyl-CoA thioester. Site-directed mutagenesis experiments in carnitine acetyltransferase supports this model, which tends to exclude a modified enzyme intermediate from the reaction pathway (46Cronin C.N. Biochem. Biophys. Res. Commun. 1997; 238 (; Correction (1997) Biochem. Biophys. Res. Commun.247, 803–804): 784-789Crossref PubMed Scopus (18) Google Scholar).The question of whether the substrates of carnitine acyltransferases bind to the same locus as malonyl-CoA has been subject of much discussion. Under this view, malonyl-CoA could be a competitive inhibitor of palmitoyl-CoA as substrate in CPT I. Whereas mutation of A381D (in the middle of the catalytic channel) strongly decreases catalytic activity, it does not modify the inhibition to malonyl-CoA in the range 1–200 μm in CPT I and in COT. These specific mutants behave similarly to the mutants reported in Ref. 13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar, since in most of them the capacity of malonyl-CoA to inhibit the mutated enzymes is maintained. On the contrary, natural mutant P479L (43Brown N.F. Mullur R.S. Subramanian I. Esser V. Bennett M.J. Saudubray J.M. Feigenbaum A.S. Kobari J.A. Macleod P.M. McGarry J.D. Cohen J.C. J. Lipid Res. 2001; 42: 1134-1142Abstract Full Text Full Text PDF PubMed Google Scholar), located in a domain that is facing toward the middle of the substrate-binding channel, has decreased sensitivity to malonyl-CoA, whereas the CPT I activity is not severely decreased (21.6% residual activity). Therefore, Ala381, although located in the model near Pro479, appears not to mediate the malonyl-CoA inhibitory effect, suggesting fine interactions in the amino acids involved in the binding of malonyl-CoA.Mutation of residues described previously also supports the model: amino acids Trp391, Arg451, Trp452 (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar), and Asp454 (42Ijist L. Mandel H. Oostheim W. Ruiter J.P. Gutman A. Wanders R.J. J. Clin. Invest. 1998; 102: 527-531Crossref PubMed Scopus (78) Google Scholar), because they are in the channel in which substrates are fitted in the catalytic event, and the amino acids Arg388 and Arg395(13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar), located in the neighborhood of the catalytic channel, because the change in charge probably disrupts the delicate charge environment. Mutant L484P (43Brown N.F. Mullur R.S. Subramanian I. Esser V. Bennett M.J. Saudubray J.M. Feigenbaum A.S. Kobari J.A. Macleod P.M. McGarry J.D. Cohen J.C. J. Lipid Res. 2001; 42: 1134-1142Abstract Full Text Full Text PDF PubMed Google Scholar) is also present at the end of the catalytic channel, which confirms the absence of activity. The similar trace of the amino acid backbone of enoyl-CoA hydratase (determined by x-ray), the location in the model of amino acids previously shown as important in the catalytic event and the functional location of alanines, predicted to be placed at least 4 Å from the catalytic histidine in CPT I and COT confirm the model. This model will facilitate in the future the studies of interaction of the substrates (palmitoyl-CoA or octanoyl-CoA) or the physiological inhibitor, malonyl-CoA, with CPT I and COT and their role in the physiological regulation of fatty acid β oxidation. Carnitine palmitoyltransferase I (CPT I1; EC 2.3.1.21) and carnitine octanoyltransferase (COT; EC 2.3.1.137) facilitate the transport of long and medium chain acyl-CoA in mitochondria and peroxisomes, respectively. Both carnitine acyltransferases facilitate the generation of energy by β-oxidation of fatty acids in the organella in which they are present. Mammalian tissues express two different isoforms of CPT I, a liver isoform (L-CPT I) (1Esser V. Britton C.H. Weis B.C. Foster D.W. Mc Garry J.D. J. Biol. Chem. 1993; 268: 5817-5822Abstract Full Text PDF PubMed Google Scholar, 2Britton C.H. Schultz R.A. Zhang B. Esser V. Foster D.W. McGarry J.D. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 1984-1988Crossref PubMed Scopus (125) Google Scholar) and a heart/skeletal muscle isoform (M-CPT I) (3Yamazaki N. Shinohara Y. Shima A. Yamanaka Y. Terada H. Biochim. Biophys. Acta. 1996; 1307: 157-161Crossref PubMed Scopus (102) Google Scholar, 4Yamazaki N. Shinohara Y. Shima A. Terada H. FEBS Lett. 1995; 363: 41-45Crossref PubMed Scopus (110) Google Scholar). As an enzyme that catalyzes the first rate-limiting step in β-oxidation, CPT I is tightly regulated by its physiological inhibitor malonyl-CoA. In regulating CPT I, malonyl-CoA confers the ability to signal to the cell the availability of lipid and carbohydrate fuels (5McGarry J.D. Brown N.F. Eur. J. Biochem. 1997; 244: 1-14Crossref PubMed Scopus (1322) Google Scholar). CPT I has a critical metabolic role in general metabolism in heart, liver, and β-cells of the pancreas and is a potential target for the treatment of metabolic disorders involving diabetes and coronary heart disease. Peroxisomal COT is also inhibited in physiological conditions by malonyl-CoA (6A'Bhaird N.N. Ramsay R.R. Biochem. J. 1992; 286: 637-640Crossref PubMed Scopus (23) Google Scholar) but to a lesser extent than CPT I. Other enzymes of the family, which are not regulated by malonyl-CoA, are CPT II, which catalyzes long-chain acyl-CoA in the mitochondria, and carnitine acetyltransferase, which has acetyl-CoA as substrate. These enzymes have recently generated much interest, especially the spatial organization of CPT I in the mitochondrial outer membrane. CPT I is an integral membrane protein, and both the N and C termini project to the cytosol, since it has two trans-membrane segments within the first 130 N-terminal residues of its primary sequence (7Fraser F. Corstorphine C.G. Zammit V.A. Biochem. J. 1997; 323: 711-718Crossref PubMed Scopus (122) Google Scholar). Interaction between amino acids from the N and C termini may determine the kinetic characteristics of the enzyme, not only in the inhibitory effect of malonyl-CoA but also in the catalytic activity (8Jackson V.N. Cameron J.M. Fraser F. Zammit V.A. Price N.T. J. Biol. Chem. 2000; 275: 19560-19566Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). COT appears not to be an integral protein of peroxisomes. Sequence alignment between CPT I and COT shows that COT lacks the first 152 amino acid residues, which indicates that it has no transmembrane regions. However, the two show high sequence identity, which suggests a common genetic origin (9van der Leij F.R. Huijkman N.C. Boomsma C. Kuipers J.R. Bartelds B. Mol. Genet. Metab. 2000; 71: 139-153Crossref PubMed Scopus (85) Google Scholar). Although several attempts have been made to identify the malonyl-CoA site, few data are available on the structure of the catalytic site of carnitine acyltransferases. It has been proposed that a histidine residue is critical in catalysis (10Schmalix W. Bandlow W. J. Biol. Chem. 1993; 268: 27428-27439Abstract Full Text PDF PubMed Google Scholar). Site-directed mutagenesis experiments have demonstrated an essential catalytic role for histidine 372 in CPT II (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar) and the homologous histidine 327 in COT (12Morillas M. Clotet J. Rubı́ B. Serra D. Asins G. Ariño J. Hegardt F.G. FEBS Lett. 2000; 466: 183-186Crossref PubMed Scopus (19) Google Scholar). However the role of a histidine in catalysis has been questioned for both CPT I and COT (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar, 14Nic a'Bhaird N. Yankovskaya V. Ramsay R.R. Biochem. J. 1998; 330: 1029-1036Crossref PubMed Scopus (11) Google Scholar). Dai et al. observed that overexpressed CPT I in yeast treated with diethylpyrocarbonate did not decrease the enzyme activity at variance with wild type rat mitochondria, although differences in the folding of CPT I in yeast and in rat liver could lead to an alternative interpretation. Other amino acids residues have been also implicated in catalysis. Mutation of several conserved arginines (Arg388, Arg451) and tryptophans (Trp391, Trp452) comprised between amino acids 381 and 481 of CPT I decreased enzyme activity (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar). These authors suggested that this segment could be the putative palmitoyl-CoA binding site of CPT I. In this fragment, mutants D376A and D464A in rat CPT II were completely inactive (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar). As to the carnitine binding site, arginine 505 in beef COT, which lies outside the proposed palmitoyl-CoA binding site, has been implicated (15Cronin C.N. Eur. J. Biochem. 1997; 247: 1029-1037Crossref PubMed Scopus (33) Google Scholar). In this study, we propose three-dimensional models of the catalytic site comprising amino acids 368–568 in CPT I and 226–417 in COT. These models were made using an integrative approach of several threading procedures taking into account different parameters such as solvatation potentials, contacts, environment-specific substitution tables, and structure-dependent gap penalties. The fold recognition analysis using these procedures showed a common template for the catalytic site of the carnitine/choline acyltransferase family of proteins, corresponding to the structure of the enoyl-CoA hydratase enzyme (Protein Data Bank entry 2dub; Ref. 16Engel C.K. Kiema T.R. Hiltunen J.K. Wierenga R.K. J. Mol. Biol. 1998; 275: 847-859Crossref PubMed Scopus (102) Google Scholar); this motif was also found in CPT I and COT. The three-dimensional models for CPT I and COT are practically identical and show a common architecture for the catalytic site. Site-directed mutagenesis indicated that CPT I His473 and COT His327 are the catalytic residues. These histidines are located near the thioester bond in the acyl-CoA, which is broken in catalysis. Mutation of CPT I A381 and COT A238, located close to the catalytic histidine decreased the enzyme activity by 80–86% without modifying the sensitivity to malonyl-CoA. DISCUSSIONDespite efforts in biochemical characterization of carnitine palmitoyltransferases and their related enzymes, their mode of action is not completely understood, probably due to the lack of any structural characterization of the catalytic site. In the absence of an appropriate crystallized reference, some bioinformatics tools can be applied to obtain a structural model able to approximately address some important questions related to these proteins' activity.The threading or "remote homology design" is a three-dimensional structure prediction technique useful when there is not enough sequence similarity of the input sequence and a known three-dimensional structure and, therefore, the "homology modeling" is not applicable. The process adapts the sequence to different known foldings and evaluates the fitting. The meaning of "fitting" varies from one threading program to others: secondary structure coincidence, similar accessibility, or solvatation energy, etc. Methods of protein fold recognition attempt to detect similarities between protein three-dimensional structure that are not accompanied by any significant sequence similarity. There are many approaches, but the unifying theme is to try and find folds that are comparable with a particular sequence. Unlike sequence-only comparison, these methods take advantage of the extra information made available by three-dimensional structural data.To detect structural homologies between CPT I and COT and any other three-dimensional representation, we used an integrative approach of two programs, THREADER2 (22Jones D.T. Miller R.T. Thornton J.M. Proteins. 1995; 23: 387-397Crossref PubMed Scopus (76) Google Scholar) and FUGUE threading server (23Shi J. Blundell T.L. Mizuguchi K. J. Mol. Biol. 2001; 310: 243-257Crossref PubMed Scopus (1080) Google Scholar). Whereas the first one uses solvatation potentials and predicted contacts, the latter performs a fold recognition analysis using structural environment-specific substitution tables and structure-dependent gap penalties. The integration of both methodologies revealed that the three-dimensional fold of the central site of all acyl-CoA transferases can be structured in the same way as the enoyl-CoA hydratase monomer (Protein Data Bank entry 2dub, chain E). What was more important was to observe that CPT I and COT had nearly identical structural models for the central region, which putatively contains the catalytic site (amino acids 368–567 of CPT I and amino acids 226–417 of COT). The predicted secondary structure for the 200 amino acids that putatively contains the palmitoyl-CoA or decanoyl-CoA binding region consists of 6 α-helices and 4 β-sheets.Additional support of this fold as template is based on the fact that these proteins bind very similar ligands, all of them acyl-CoA derivatives; the crystal structure of the enoyl-CoA hydratase monomer includes precisely a molecule of octanoyl-CoA, used as inhibitor in this case, the natural substrate of COT. Enoyl-CoA hydratase (Protein Data Bank code 2dub), also known as crotonase, belongs to the enoyl-CoA hydratase/isomerase family. It is a homohexameric enzyme, located at the mitochondrial matrix. It catalyzes the second step in the mitochondrial fatty acid β-oxidation pathway, transforming the 3-hydroxyacyl-CoA into trans-2(or 3)-enoyl-CoA plus H2O.The model predicts that although CPT I Ala381(Ala238 in COT) is 92 amino acids away from the CPT I His473 (His327 in COT) it is very close to the catalytic histidines. The important decrease in activity (14–20% of residual activity with respect to the wild type) after mutation CPT I A381D or COT A238D confirms the function of these alanines and supports the three-dimensional model. The marked modification ofKm for acyl-CoA of the mutants supports the role of these alanines in locating the substrates to the catalytic site.In addition, these alanine residues had been implicated in CPT and COT activity by a tree determinant study of the complete alignment of the carnitine-choline acyltransferase family of proteins. In a protein family alignment, the positions considered as tree determinants, or subfamily conserved residues, are usually accepted to be implicated in key catalytic activities, being responsible for the different substrate specificities or enzymatic activities of the different subfamilies of the alignment (37Casari G. Sander C. Valencia A. Nat. Struct. Biol. 1995; 2: 171-178Crossref PubMed Scopus (346) Google Scholar, 38Pazos F. Sanchez-Pulido L. Garcı́a-Ranea J.A. Andrade M.A. Atrian S. Valencia A. Lundh D. Olsson B. Narayanan A. Biocomputing and Emergent Computation. World Scientific, Singapore1997: 132-145Google Scholar). A clear example of this type of analysis was made previously in this same family of proteins related to the carnitine versus choline affinity (44Cronin C.N. J. Biol. Chem. 1998; 273: 24465-24469Abstract Full Text Full Text PDF PubMed Scopus (25) Google Scholar). The structure-based alignment of the template sequence with all the other transferases sequences used in the extensive threading procedure can also be used to build three-dimensional models of all the other members of the family in the future.The group of Woldegiorgis (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar) suggested that the region comprised between 381–481 could be the putative palmitoyl-CoA binding site. Moreover, the abolition of activity after mutation of H473A strongly suggests that this His473 is the catalytic site. McGarry and co-workers (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar), after mutagenesis of homologous histidine in CPT II, proposed that this was the catalytic site. This was confirmed in a previous study in COT, in which mutant H327A abolished the catalytic activity (45Morillas M. Clotet J. Rubı́ B. Serra D. Ariño J. Hegardt F.G. Asins G. Biochem. J. 2000; 351: 495-502Crossref PubMed Google Scholar). Dai et al. (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar) questioned whether these histidines were the catalytic sites after the observation that chemical modification of mitochondria from yeast strains expressing L-CPT I and M-CPT I by diethylpyrocarbonate had no effect on catalytic activity. It is possible that expressed CPT I in yeast mitochondria makes this histidine inaccessible for diethylpyrocarbonate modification. Brown et al. (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar) suggested that a charge-relay system involving this His coupled with an Asp extracts a proton from the C-3 hydroxyl group of carnitine, allowing for nucleophilic attack of the resulting oxyanion of the carbonyl group of the acyl-CoA thioester. Site-directed mutagenesis experiments in carnitine acetyltransferase supports this model, which tends to exclude a modified enzyme intermediate from the reaction pathway (46Cronin C.N. Biochem. Biophys. Res. Commun. 1997; 238 (; Correction (1997) Biochem. Biophys. Res. Commun.247, 803–804): 784-789Crossref PubMed Scopus (18) Google Scholar).The question of whether the substrates of carnitine acyltransferases bind to the same locus as malonyl-CoA has been subject of much discussion. Under this view, malonyl-CoA could be a competitive inhibitor of palmitoyl-CoA as substrate in CPT I. Whereas mutation of A381D (in the middle of the catalytic channel) strongly decreases catalytic activity, it does not modify the inhibition to malonyl-CoA in the range 1–200 μm in CPT I and in COT. These specific mutants behave similarly to the mutants reported in Ref. 13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar, since in most of them the capacity of malonyl-CoA to inhibit the mutated enzymes is maintained. On the contrary, natural mutant P479L (43Brown N.F. Mullur R.S. Subramanian I. Esser V. Bennett M.J. Saudubray J.M. Feigenbaum A.S. Kobari J.A. Macleod P.M. McGarry J.D. Cohen J.C. J. Lipid Res. 2001; 42: 1134-1142Abstract Full Text Full Text PDF PubMed Google Scholar), located in a domain that is facing toward the middle of the substrate-binding channel, has decreased sensitivity to malonyl-CoA, whereas the CPT I activity is not severely decreased (21.6% residual activity). Therefore, Ala381, although located in the model near Pro479, appears not to mediate the malonyl-CoA inhibitory effect, suggesting fine interactions in the amino acids involved in the binding of malonyl-CoA.Mutation of residues described previously also supports the model: amino acids Trp391, Arg451, Trp452 (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar), and Asp454 (42Ijist L. Mandel H. Oostheim W. Ruiter J.P. Gutman A. Wanders R.J. J. Clin. Invest. 1998; 102: 527-531Crossref PubMed Scopus (78) Google Scholar), because they are in the channel in which substrates are fitted in the catalytic event, and the amino acids Arg388 and Arg395(13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar), located in the neighborhood of the catalytic channel, because the change in charge probably disrupts the delicate charge environment. Mutant L484P (43Brown N.F. Mullur R.S. Subramanian I. Esser V. Bennett M.J. Saudubray J.M. Feigenbaum A.S. Kobari J.A. Macleod P.M. McGarry J.D. Cohen J.C. J. Lipid Res. 2001; 42: 1134-1142Abstract Full Text Full Text PDF PubMed Google Scholar) is also present at the end of the catalytic channel, which confirms the absence of activity. The similar trace of the amino acid backbone of enoyl-CoA hydratase (determined by x-ray), the location in the model of amino acids previously shown as important in the catalytic event and the functional location of alanines, predicted to be placed at least 4 Å from the catalytic histidine in CPT I and COT confirm the model. This model will facilitate in the future the studies of interaction of the substrates (palmitoyl-CoA or octanoyl-CoA) or the physiological inhibitor, malonyl-CoA, with CPT I and COT and their role in the physiological regulation of fatty acid β oxidation. Despite efforts in biochemical characterization of carnitine palmitoyltransferases and their related enzymes, their mode of action is not completely understood, probably due to the lack of any structural characterization of the catalytic site. In the absence of an appropriate crystallized reference, some bioinformatics tools can be applied to obtain a structural model able to approximately address some important questions related to these proteins' activity. The threading or "remote homology design" is a three-dimensional structure prediction technique useful when there is not enough sequence similarity of the input sequence and a known three-dimensional structure and, therefore, the "homology modeling" is not applicable. The process adapts the sequence to different known foldings and evaluates the fitting. The meaning of "fitting" varies from one threading program to others: secondary structure coincidence, similar accessibility, or solvatation energy, etc. Methods of protein fold recognition attempt to detect similarities between protein three-dimensional structure that are not accompanied by any significant sequence similarity. There are many approaches, but the unifying theme is to try and find folds that are comparable with a particular sequence. Unlike sequence-only comparison, these methods take advantage of the extra information made available by three-dimensional structural data. To detect structural homologies between CPT I and COT and any other three-dimensional representation, we used an integrative approach of two programs, THREADER2 (22Jones D.T. Miller R.T. Thornton J.M. Proteins. 1995; 23: 387-397Crossref PubMed Scopus (76) Google Scholar) and FUGUE threading server (23Shi J. Blundell T.L. Mizuguchi K. J. Mol. Biol. 2001; 310: 243-257Crossref PubMed Scopus (1080) Google Scholar). Whereas the first one uses solvatation potentials and predicted contacts, the latter performs a fold recognition analysis using structural environment-specific substitution tables and structure-dependent gap penalties. The integration of both methodologies revealed that the three-dimensional fold of the central site of all acyl-CoA transferases can be structured in the same way as the enoyl-CoA hydratase monomer (Protein Data Bank entry 2dub, chain E). What was more important was to observe that CPT I and COT had nearly identical structural models for the central region, which putatively contains the catalytic site (amino acids 368–567 of CPT I and amino acids 226–417 of COT). The predicted secondary structure for the 200 amino acids that putatively contains the palmitoyl-CoA or decanoyl-CoA binding region consists of 6 α-helices and 4 β-sheets. Additional support of this fold as template is based on the fact that these proteins bind very similar ligands, all of them acyl-CoA derivatives; the crystal structure of the enoyl-CoA hydratase monomer includes precisely a molecule of octanoyl-CoA, used as inhibitor in this case, the natural substrate of COT. Enoyl-CoA hydratase (Protein Data Bank code 2dub), also known as crotonase, belongs to the enoyl-CoA hydratase/isomerase family. It is a homohexameric enzyme, located at the mitochondrial matrix. It catalyzes the second step in the mitochondrial fatty acid β-oxidation pathway, transforming the 3-hydroxyacyl-CoA into trans-2(or 3)-enoyl-CoA plus H2O. The model predicts that although CPT I Ala381(Ala238 in COT) is 92 amino acids away from the CPT I His473 (His327 in COT) it is very close to the catalytic histidines. The important decrease in activity (14–20% of residual activity with respect to the wild type) after mutation CPT I A381D or COT A238D confirms the function of these alanines and supports the three-dimensional model. The marked modification ofKm for acyl-CoA of the mutants supports the role of these alanines in locating the substrates to the catalytic site. In addition, these alanine residues had been implicated in CPT and COT activity by a tree determinant study of the complete alignment of the carnitine-choline acyltransferase family of proteins. In a protein family alignment, the positions considered as tree determinants, or subfamily conserved residues, are usually accepted to be implicated in key catalytic activities, being responsible for the different substrate specificities or enzymatic activities of the different subfamilies of the alignment (37Casari G. Sander C. Valencia A. Nat. Struct. Biol. 1995; 2: 171-178Crossref PubMed Scopus (346) Google Scholar, 38Pazos F. Sanchez-Pulido L. Garcı́a-Ranea J.A. Andrade M.A. Atrian S. Valencia A. Lundh D. Olsson B. Narayanan A. Biocomputing and Emergent Computation. World Scientific, Singapore1997: 132-145Google Scholar). A clear example of this type of analysis was made previously in this same family of proteins related to the carnitine versus choline affinity (44Cronin C.N. J. Biol. Chem. 1998; 273: 24465-24469Abstract Full Text Full Text PDF PubMed Scopus (25) Google Scholar). The structure-based alignment of the template sequence with all the other transferases sequences used in the extensive threading procedure can also be used to build three-dimensional models of all the other members of the family in the future. The group of Woldegiorgis (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar) suggested that the region comprised between 381–481 could be the putative palmitoyl-CoA binding site. Moreover, the abolition of activity after mutation of H473A strongly suggests that this His473 is the catalytic site. McGarry and co-workers (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar), after mutagenesis of homologous histidine in CPT II, proposed that this was the catalytic site. This was confirmed in a previous study in COT, in which mutant H327A abolished the catalytic activity (45Morillas M. Clotet J. Rubı́ B. Serra D. Ariño J. Hegardt F.G. Asins G. Biochem. J. 2000; 351: 495-502Crossref PubMed Google Scholar). Dai et al. (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar) questioned whether these histidines were the catalytic sites after the observation that chemical modification of mitochondria from yeast strains expressing L-CPT I and M-CPT I by diethylpyrocarbonate had no effect on catalytic activity. It is possible that expressed CPT I in yeast mitochondria makes this histidine inaccessible for diethylpyrocarbonate modification. Brown et al. (11Brown N.F. Anderson R.C. Caplan S.L. Foster D.W. McGarry J.D. J. Biol. Chem. 1994; 269: 19157-19162Abstract Full Text PDF PubMed Google Scholar) suggested that a charge-relay system involving this His coupled with an Asp extracts a proton from the C-3 hydroxyl group of carnitine, allowing for nucleophilic attack of the resulting oxyanion of the carbonyl group of the acyl-CoA thioester. Site-directed mutagenesis experiments in carnitine acetyltransferase supports this model, which tends to exclude a modified enzyme intermediate from the reaction pathway (46Cronin C.N. Biochem. Biophys. Res. Commun. 1997; 238 (; Correction (1997) Biochem. Biophys. Res. Commun.247, 803–804): 784-789Crossref PubMed Scopus (18) Google Scholar). The question of whether the substrates of carnitine acyltransferases bind to the same locus as malonyl-CoA has been subject of much discussion. Under this view, malonyl-CoA could be a competitive inhibitor of palmitoyl-CoA as substrate in CPT I. Whereas mutation of A381D (in the middle of the catalytic channel) strongly decreases catalytic activity, it does not modify the inhibition to malonyl-CoA in the range 1–200 μm in CPT I and in COT. These specific mutants behave similarly to the mutants reported in Ref. 13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar, since in most of them the capacity of malonyl-CoA to inhibit the mutated enzymes is maintained. On the contrary, natural mutant P479L (43Brown N.F. Mullur R.S. Subramanian I. Esser V. Bennett M.J. Saudubray J.M. Feigenbaum A.S. Kobari J.A. Macleod P.M. McGarry J.D. Cohen J.C. J. Lipid Res. 2001; 42: 1134-1142Abstract Full Text Full Text PDF PubMed Google Scholar), located in a domain that is facing toward the middle of the substrate-binding channel, has decreased sensitivity to malonyl-CoA, whereas the CPT I activity is not severely decreased (21.6% residual activity). Therefore, Ala381, although located in the model near Pro479, appears not to mediate the malonyl-CoA inhibitory effect, suggesting fine interactions in the amino acids involved in the binding of malonyl-CoA. Mutation of residues described previously also supports the model: amino acids Trp391, Arg451, Trp452 (13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar), and Asp454 (42Ijist L. Mandel H. Oostheim W. Ruiter J.P. Gutman A. Wanders R.J. J. Clin. Invest. 1998; 102: 527-531Crossref PubMed Scopus (78) Google Scholar), because they are in the channel in which substrates are fitted in the catalytic event, and the amino acids Arg388 and Arg395(13Dai J. Zhu H. Shi J. Woldegiorgis G. J. Biol. Chem. 2000; 275: 22020-22024Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar), located in the neighborhood of the catalytic channel, because the change in charge probably disrupts the delicate charge environment. Mutant L484P (43Brown N.F. Mullur R.S. Subramanian I. Esser V. Bennett M.J. Saudubray J.M. Feigenbaum A.S. Kobari J.A. Macleod P.M. McGarry J.D. Cohen J.C. J. Lipid Res. 2001; 42: 1134-1142Abstract Full Text Full Text PDF PubMed Google Scholar) is also present at the end of the catalytic channel, which confirms the absence of activity. The similar trace of the amino acid backbone of enoyl-CoA hydratase (determined by x-ray), the location in the model of amino acids previously shown as important in the catalytic event and the functional location of alanines, predicted to be placed at least 4 Å from the catalytic histidine in CPT I and COT confirm the model. This model will facilitate in the future the studies of interaction of the substrates (palmitoyl-CoA or octanoyl-CoA) or the physiological inhibitor, malonyl-CoA, with CPT I and COT and their role in the physiological regulation of fatty acid β oxidation. We are grateful to Dr. V. Zammit, who kindly provided the CPT I antibodies. We are also grateful to Robin Rycroft of the Language Service for valuable assistance in the preparation of the English manuscript.

Referência(s)