Structures of Human Cytosolic NADP-dependent Isocitrate Dehydrogenase Reveal a Novel Self-regulatory Mechanism of Activity*

Transcription

1 THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 279, No. 32, Issue of August 6, pp , by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A. Structures of Human Cytosolic NADP-dependent Isocitrate Dehydrogenase Reveal a Novel Self-regulatory Mechanism of Activity* Received for publication, April 19, 2004, and in revised form, May 24, 2004 Published, JBC Papers in Press, June 1, 2004, DOI /jbc.M Xiang Xu, Jingyue Zhao, Zhen Xu, Baozhen Peng, Qiuhua Huang, Eddy Arnold, and Jianping Ding From the Key Laboratory of Proteomics, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai , China, the State Key Laboratory for Medical Genomics, Shanghai Institute of Hematology, Ruijin Hospital, Shanghai Second Medical University, Shanghai , China, and the Center for Advanced Biotechnology and Medicine and Rutgers University Department of Chemistry and Chemical Biology, Piscataway, New Jersey Isocitrate dehydrogenases (IDHs) catalyze the oxidative decarboxylation of isocitrate to -ketoglutarate, and regulation of the enzymatic activity of IDHs is crucial for their biological functions. Bacterial IDHs are reversibly regulated by phosphorylation of a strictly conserved serine residue at the active site. Eukaryotic NADP-dependent IDHs (NADP-IDHs) have been shown to have diverse important biological functions; however, their regulatory mechanism remains unclear. Structural studies of human cytosolic NADP-IDH (HcIDH) in complex with NADP and in complex with NADP, isocitrate, and Ca 2 reveal three biologically relevant conformational states of the enzyme that differ substantially in the structure of the active site and in the overall structure. A structural segment at the active site that forms a conserved -helix in all known NADP-IDH structures assumes a loop conformation in the open, inactive form of HcIDH; a partially unraveled -helix in the semiopen, intermediate form; and an -helix in the closed, active form. The side chain of Asp 279 of this segment occupies the isocitrate-binding site and forms hydrogen bonds with Ser 94 (the equivalent of the phosphorylation site in bacterial IDHs) in the inactive form and chelates the metal ion in the active form. The structural data led us to propose a novel self-regulatory mechanism for HcIDH that mimics the phosphorylation mechanism used by the bacterial homologs, consistent with biochemical and biological data. This mechanism might be applicable to other eukaryotic NADP-IDHs. The results also provide insights into the recognition and specificity of substrate and cofactor by eukaryotic NADP-IDHs. * This work was supported in part by National Natural Science Foundation of China Grants , , and ; 863 Hi-Tech Program Grants 2001AA235071, 2001AA233021, and 2002BA711A13; and Chinese Academy of Sciences Grant KSCX1- SW-17 (to J. D.). This work was also supported in part by National Institutes of Health Grants AI and GM (to E. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The atomic coordinates and structure factors (codes 1T09 and 1T0L) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ ( To whom correspondence should be addressed: Key Laboratory of Proteomics, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, 320 Yue- Yang Rd., Shanghai , China. Tel.: ; Fax: ; jpding@sibs.ac.cn Isocitrate dehydrogenases (IDHs) 1 comprise a family of enzymes that catalyze oxidative decarboxylation of isocitrate into -ketoglutarate in the Krebs cycle. Regulation of the enzymatic activity of IDHs is crucial for their biological functions. Eukaryotic cells express two distinct classes of IDHs that utilize either NAD or NADP as their cofactors and serve diverse biological functions. NAD-dependent IDH is localized to the mitochondrial matrix and is well known for its central role for energy production in the Krebs cycle. NADP-dependent IDHs (NADP- IDHs, EC ) are primarily located either in mitochondria (1) or cytoplasm (2). In addition to their potential catabolic role in the Krebs cycle, both mitochondrial and cytosolic NADP- IDHs are shown to play an important role in cellular defense against oxidative damage as a source of NADPH (3 5). Moreover, all eukaryotic cytosolic IDHs contains a type 1 peroxisomal targeting sequence at their C terminus (6) that is sufficient to direct proteins into peroxisomes (7). Cytosolic IDHs have been found in peroxisomes of yeast, human, and rat liver cells and are shown to be required for the -oxidation of unsaturated fatty acids as a provider of NADPH inside peroxisomes (8 12). In contrast, Escherichia coli as well as other bacterial cells contain only a single NADP-IDH (EcIDH) whose structures and functions have been studied extensively (13 17). EcIDH functions at a critical juncture between the Krebs cycle and the glyoxylate cycle, and regulation of its activity controls the substrate flux between the two reactions (18). The activity of EcIDH is regulated by the phosphorylation of Ser 113, which is located at the isocitrate-metal ion-binding site (13, 14, 16, 19). The phosphorylation reaction is reversibly catalyzed by a bifunctional enzyme, IDH kinase/phosphatase (IDH-K/P) (18). This regulatory mechanism has been proposed to apply to Bacillus subtilis NADP-IDH (BsIDH), which exhibits extensive sequence and structural similarities with EcIDH (20, 21). The regulatory mechanism of mammalian NADP-IDHs is still unclear. Although mammalian and bacterial NADP-IDHs share less than 20% sequence identity, the structure of porcine heart mitochondrial NADP-IDH (PmIDH) shows an overall structure very similar to its bacterial counterparts (22). Structure-based sequence alignments indicate that mammalian NADP-IDHs have a strictly conserved serine at a position 1 The abbreviations used are: IDH, isocitrate dehydrogenase; BsIDH, B. subtilis NADP-IDH; EcIDH, E. coli NADP-IDH; HcIDH, human cytosolic NADP-IDH; PmIDH, porcine mitochondrial NADP-IDH; IDH- K/P, IDH kinase/phosphatase; NADP-IDH, NADP-dependent IDH; RMSD, root mean square deviation; MES, 2-morpholinoethanesulfonic acid; NCS, noncrystallographic symmetry. This paper is available on line at

2 Regulatory Mechanism of Human Cytosolic IDH equivalent to Ser 113 of EcIDH (6, 22). Those results led to the suggestion that mammalian NADP-IDHs might also contain a phosphorylation site and are possibly regulated by kinases and phosphatases like their bacterial homologs (22). We report here the crystal structures of human cytosolic NADP-IDH (HcIDH) in complex with its cofactor, NADP (designated as binary complex thereafter), and in complex with NADP, the substrate isocitrate, and Ca 2 (designated as quaternary complex thereafter). Structural comparisons of HcIDH with other NADP-IDHs reveal that HcIDH is likely to utilize a unique mechanism to self-regulate its activity that mimics the phosphorylation mechanism used by its bacterial homologs. The structural data also provide insights into the recognition and specificity of substrate and cofactor by eukaryotic NADP-IDHs. EXPERIMENTAL PROCEDURES Protein Expression and Purification The plasmid pet-22b( ) containing the recombinant HcIDH gene fused with a His 6 tag at the C terminus was expressed in the E. coli BL21(DE3) strain. E. coli cells transformed with the vector were grown overnight at 37 C in 100 ml of LB media containing ampicillin (0.1 mg/ml). The overnight culture was used to inoculate 2 liters of LB medium and grown to an A 600 nm of 0.6. After 12 h of expression induced with 1 mm isopropyl- -D-thiogalactopyranoside at 20 C, the cells were collected by centrifugation at 4,000 g and suspended in 40 ml of lysis buffer containing 20 mm Tris HCl, ph 7.4, 1%Triton X-100, 500 mm NaCl, 5 mm -mercaptoethanol, and 1 mm phenylmethylsulfonyl fluoride. The cells were lysed on ice by sonication, and cell debris was precipitated by centrifugation at 15,000 g. HcIDH was purified by affinity chromatography using a nickel-nitrilotriacetic acid-agarose column (Qiagen). The lysis extract was first loaded on the column and then washed with a washing buffer (20 mm Tris HCl, ph 7.4, 30 mm imidazole, and 500 mm NaCl) to elute nonspecific binding proteins. The target protein was eluted finally with an elution buffer (20 mm Tris HCl, ph 7.4, 200 mm imidazole, and 500 mm NaCl). The protein was concentrated to 15 mg/ml in a storage buffer (20 mm Tris HCl, ph 7.4, and 100 mm NaCl). HcIDH consists of 414 amino acids with a calculated molecular mass of 46.7 kda. Dynamic light scattering and gel filtration analyses indicate that the protein is homogeneously dispersed and forms a homodimer in solution in both the presence and absence of substrate (data not shown). Biochemical assays indicate that the enzyme is active with a Michaelis constant (K m ) of 12.4 M and a maximum velocity of 41.9 mol/min/mg. Enzymatic Activity Assay The IDH activity of HcIDH was assayed spectrophotometrically by measuring the reduction of NADP to NADPH, which has an extinction coefficient of 6220 M 1 cm 1 at 340 nm. The standard reaction buffer (1 ml) consists of 20 mm Tris HCl, ph 7.4, 0.1 mm NADP, 4 mm dl-isocitrate, and 2 mm MnSO 4. The reaction was initiated by the addition of 20 l of HcIDH at a concentration of 0.05 mg/ml or its appropriate dilution. The reaction rate was calculated as A 340 /min. Crystallization and Diffraction Data Collection Crystallization was conducted using the hanging drop vapor diffusion method. Crystals of the HcIDH-NADP complex were grown at 4 C in a drop with equal volumes of the protein solution containing 15 mg/ml HcIDH in the storage buffer and the reservoir solution (100 mm MES, ph 6.5, and 12% polyethylene glycol 20,000). Crystals of the HcIDH-NADP-isocitrate-Ca 2 complex were grown at 20 C in a drop with equal volumes of the protein solution containing 15 mg/ml HcIDH, 10 mm dl-isocitrate, 10 mm CaCl 2, and 10 mm NADP and the reservoir solution containing 100 mm MES, ph 5.9, and 20% polyethylene glycol 6,000. The diffraction data for the binary complex were collected from a flash-cooled crystal at 100 K using the F2 beamline at the Cornell High Energy Synchrotron Source with a wavelength of Å and were processed, integrated, and scaled together with HKL2000 (23). Diffraction data for the quaternary complex were collected from a flash-cooled crystal at 100 K using an in-house Rigaku R-Axis IV and CuK radiation (wavelength of Å) and were processed and scaled together using CrystalClear (24). The diffraction data statistics are summarized in Table I. Structure Determination The structure of the HcIDH binary complex was solved with the molecular replacement method as implemented in CNS (25) using the crystal structure of PmIDH (22) as the search model. Cross-rotation function search and subsequent translation function search using the monomeric PmIDH produced two outstanding peaks, corresponding to two HcIDH monomers in the asymmetric unit. Structure refinement was carried out using CNS. Model building was performed with O (26). The free R factor was calculated with 5% of the data, and a bulk solvent correction was applied in all refinements. Strict 2-fold noncrystallographic symmetry (NCS) constraints were used in the early stage of refinement, but the two monomers in the asymmetric unit were refined independently in the later stage of refinement because of the conformational differences. In the initial difference Fourier maps there was strong residual electron density at the active site of each monomer, and the position and shape of the electron density match NADP very well. Because HcIDH is an NADP-IDH, we modeled this piece of electron density as NADP. Because no NADP was added during the purification process and in the crystallization solution, the bound NADP must have copurified together with the enzyme from the expression system. Electron density peaks in difference Fourier maps at a height of above 2.5 were assigned as water molecules if they had reasonable geometry in relation to hydrogen bond donors or acceptors from amino acids or other water molecules, and their B factors did not rise above 60 Å 2 during subsequent refinement. The structure of the HcIDH quaternary complex was also solved with the molecular replacement method. We employed the monomeric and dimeric PmIDH and the monomeric and dimeric HcIDH-NADP complex as the search models, respectively. The dimeric PmIDH model produced the best results, which revealed two clear and outstanding solutions in cross-rotation function and translation function searches, corresponding to two HcIDH homodimers related by a pseudo 2-fold NCS in the asymmetric unit. Application of strict 4-fold NCS constraints could not decrease both the R and free R factors to below 30%. Therefore, NCS restraints were applied in the final stage of refinement. NADP, isocitrate, and Ca 2 at the active site have well defined electron density in the initial difference Fourier maps and were included in the refinement after the R factor was reduced to 26.5%. Water molecules were added gradually in the model, and refinement was based on the criteria described above. RESULTS AND DISCUSSION Overall Structures of HcIDH The structure of the binary complex contains two HcIDH molecules in an asymmetric unit that forms an asymmetric homodimer. The final structure model contains two full HcIDH polypeptide chains each consisting of residues 1 414, two NADP molecules, and 154 water molecules. The region consisting of residues in subunit A has well defined electron density (Fig. 1A), whereas its counterpart in subunit B is partially disordered and has discontinuous electron density. The structure of the quaternary complex contains four HcIDH molecules per asymmetric unit that form two homodimers related by a pseudo 2-fold NCS. The final structure model contains four full-length HcIDH molecules, four NADP molecules, four isocitrate molecules, four Ca 2 ions, and 914 water molecules. The region consisting of residues in all four subunits of the quaternary complex is well ordered and has good electron density (Fig. 1B). A summary of the structure refinement and the final model statistics is given in Table I. The HcIDH monomer has a secondary structure fold and topology similar to other dimeric NADP-IDHs whose structures are known, specifically EcIDH (13, 14), BsIDH (20), and PmIDH (22), even though these enzymes share varying sequence identity ( 16 17% between mammalian and bacterial NADP-IDHs and 69% between HcIDH and PmIDH) (Figs. 2 and 3). Like other NADP-IDHs, HcIDH consists of three domains: a large domain, a small domain, and a clasp domain. The large domain comprises residues and and has a typical Rossmann fold. The small domain contains residues and and forms an / sandwich structure. The clasp domain ranges from residues 137 to 185 and folds as two two-stranded anti-parallel -sheets stacked on each other. The large and small domains are joined together by a -sheet (Fig. 2). There are two clefts flanked on each side of the -sheet.

3 33948 Regulatory Mechanism of Human Cytosolic IDH FIG. 1.Representative SIGMAA-weighted 2 F o F c maps (1 contour level) at the active site. A, subunit A in the structure of the binary complex. B, subunit A in the structure of the quaternary complex. The final coordinates of the structures are shown as ball-and-stick models. The large, deep cleft is formed by the large and small domains of one subunit and the small domain of the adjacent subunit and comprises the active site (Fig. 2). The active site cleft is hydrophilic in nature and is exposed to solvent and accessible to substrate and/or cofactor. It consists of the NADP-binding site and the isocitrate-metal ion-binding site. The small, shallow cleft is formed between the large and small domains of the same subunit and is located at the back of the large cleft (designated as the back cleft) (Fig. 2). In contrast to bacterial IDHs in which the back cleft is largely hydrophobic (13, 20), the small cleft in HcIDH is hydrophobic at the base and highly hydrophilic at the rim. This feature is also observed in PmIDH. The back cleft has been proposed to have functional importance (13). Our results suggest that the back cleft is involved in the regulation of conformational changes of HcIDH and may affect the function of the enzyme (see discussion later). The clasp domain functions as an interlock in all dimeric NADP-IDH structures that holds the two subunits together to form the catalytic active site (Fig. 2). Structural and sequence comparisons indicate that the clasp domain varies substantially in primary sequence and differs in three-dimensional structure among NADP-IDHs from different species (Figs. 2 and 3). In both HcIDH and PmIDH structures, the clasp domains of the two subunits in the homodimer interwind together to form two layers of four-stranded anti-parallel -sheets. However, the clasp domains of the bacterial enzymes form two anti-parallel -helices and a four-stranded anti-parallel -sheet (13, 20). These differences can serve as indicators for phylogenetic and evolutionary analyses. Like other dimeric NADP-IDHs, HcIDH exists as a homodimer in solution in both the absence and the presence of substrate determined by dynamic light scattering and gel filtration analyses (data not shown) and forms a homodimer in the crystal structures. Similar to EcIDH, BsIDH, and PmIDH, the dimeric HcIDH has two active sites, each composed of amino acid residues from both subunits of the homodimer (see discussion later), suggesting that HcIDH most likely functions as a homodimer. The intersubunit relationship and interface contacts in the HcIDH quaternary complex are similar to those in the structures of EcIDH, BsIDH, and PmIDH. Dimerization is mediated by extensive hydrophobic and hydrophilic interactions primarily between the two clasp domains and between four -helices of the small domains ( 9 and 10 of each subunit) (Figs. 2 and 3). The dimer interface buries 3703 Å 2 solvent-accessible surface of each monomer, corresponding to 18.5% of the surface area of the monomer. The buried surface area in EcIDH, BsIDH, and PmIDH is 3278 Å 2 (17.3%), 3610 Å 2 (18.5%), and 3404 Å 2 (17.7%), respectively. The large buried surface area in the dimer interface suggests that these dimeric enzymes are very stable. However, in the HcIDH binary com-

4 TABLE I X-ray diffraction data and refinement statistics Binary complex Quaternary complex Diffraction data statistics Resolution (Å) ( ) a ( ) Space group P P2 1 a, b, and c (Å) 82.7, 82.7, , 86.7, ( o ) Completeness (%) 99.8 (99.8) 95.7 (95.7) R merge (%) b 10.3 (38.6) 8.5 (8.5) Observed reflections 275, ,244 Unique reflections 30,445 71,234 I/ (I) 10.2 (6.3) 7.8 (2.9) Redundancy 9.1 (7.5) 3.1 (2.9) Mosaicity Refinement statistics Total reflections 30,393 70,684 Working set 28,864 67,062 Free R set 1,529 3,622 Subunits/asymmetric unit 2 4 Total protein atoms 6,556 13,124 Total ligand atoms Total solvent atoms Average B factor (Å 2 ) Residues (A), (B) 15.5 NADP 32.9 (A), 27.9 (B) 15.8 Isocitrate 34.6 R factor c Free R factor c RMSD from ideal geometry Bonds (Å) Angles ( ) Ramachandran plot Core (%) Allowed (%) Generously allowed (%) Luzzati atomic positional error a The numbers in parentheses refer to the highest resolution shell. b R merge hkl i I i (hkl) i I(hkl) / hkl i I i (hkl). c R factor F o F c / F o. Regulatory Mechanism of Human Cytosolic IDH plex the two subunits of the homodimer adopt different conformations. Helix 10 of the small domain is completely unwound and intrudes into the active site in subunit A; in subunit B it is partially unwound and is extended into subunit A. Helix 9 in both subunits is expanded outwards (Figs. 2 and 3). The dimer interface buries 2867 Å 2 solvent-accessible surface of each monomer, corresponding to 13.8% of the surface area of the monomer. Compared with the quaternary complex, the dimer interface in the binary complex is less compact and probably less stable. Isocitrate-Metal Ion-binding Site In the structure of the HcIDH quaternary complex, an isocitrate molecule and a Ca 2 ion are found to bind at the active site of each subunit. The isocitrate substrate interacts with Thr 77, Ser 94, Arg 100, Arg 109, Arg 132, Tyr 139, and Asp 275 of one subunit; Lys 212, Thr 214, and Asp 252 of the adjacent subunit; 3 water molecules; NADP; and Ca 2. Most of these residues are strictly conserved in all NADP-IDHs, except Thr 214, which is invariant in eukaryotic NADP-IDHs but is replaced with an Asn in bacterial enzymes. The -carboxylate of isocitrate forms hydrophilic interactions with the side chains of Arg 100, Arg 109, Arg 132, and Asp 275 and the carbonyl group of the nicotinamide moiety of NADP. Its hydroxyl group makes hydrogen bonding interactions with the side chain of Asp 275 and the side chains of Lys 212 and Asp 252 of the adjacent subunit. The -carboxylate group has hydrophilic interactions with the side chains of Arg 100, Arg 132, Tyr 139, and Asp 275, and the side chain of Lys 212 of the adjacent subunit. The -carboxylate group forms hydrogen bonds with the side chains of Thr 77 and Ser 94, the side chain of Thr 214 of the adjacent subunit, and the 2 -OH group of the nicotinamide ribose of NADP. In addition, one -carboxylate oxygen and the hydroxyl group of isocitrate chelate the Ca 2 ion. NADP-IDHs require a divalent metal ion, usually Mn 2 or Mg 2, for their enzymatic activity. In all structures of NADP- IDHs bound with a divalent metal ion, the residues involved in chelating the metal ion are strictly conserved. Substitution of Ca 2 for Mg 2 has very little effect on the structure of the active site with only slight changes of the coordination geometry of the metal ion in EcIDH. This substitution decreases drastically the enzymatic rate of EcIDH but does not significantly change the relative binding affinities of the substrate and the metal ion (15, 17). In protein structures Mn 2 and Mg 2 prefer to have six ligands, and Ca 2 prefers to have seven or eight ligands (27). In our HcIDH quaternary complex, Ca 2 is coordinated by one -carboxylate oxygen and the hydroxyl group of isocitrate, the main chain carbonyl oxygen, and the side chain O -1 atom of Asp 275, the side chain O -1 atom (and possibly O -2 atom) of Asp 279, the side chain O -2 atom of Asp 252 of the adjacent subunit, and one water molecule and adopts a distorted pentagonal bipyramid coordination geometry. This geometry is similar to that observed in the Ca 2 - bound EcIDH structure but slightly different from the ideal octahedral geometry observed in the Mg 2 - or Mn 2 -bound NADP-IDH structures (17, 22). NADP-binding Site Although the conformation of the active site is different in the two subunits of the homodimer in the binary complex, NADP is bound at the same location of the active site and interacts with almost the same set of residues of the large domain. The adenine moiety of NADP has both hydrophobic and hydrophilic interactions with the side chains of His 309, Val 312, Arg 314, and His 315 and both side chains and main chains of Thr 327 and Asn 328 (Fig. 4). The main chain amide nitrogen and carbonyl oxygen of Asn 328 form hydrogen bonds with the N-1 and N-6 atoms of the adenine, respectively.

5 33950 Regulatory Mechanism of Human Cytosolic IDH FIG. 2. Overall structures of HcIDH. A, ribbon diagrams of the HcIDH structures viewing in perpendicular to the pseudo 2-fold NCS axis in the binary complex (left panel) and in the quaternary complex (right panel). The color codes for the three domains of HcIDH are: large domain in green, small domain in blue, and clasp domain in red. The bound NADP (in magenta) and isocitrate (in gold) are shown as ball-and-stick models, and Ca 2 appears as a cyan sphere. B, ribbon diagrams of the HcIDH structure viewing along the pseudo 2-fold NCS axis in the binary complex (left panel) and in the quaternary complex (right panel). C, the electrostatic surface representations showing the active site cleft and the back cleft viewing from the active sites of subunit A (left panel) and of subunit B (middle panel) in the binary complex and the active site in the quaternary complex (right panel). The bound NADP and isocitrate are shown as ball-and-stick models. The side chains of Arg 314 and His 315 form salt bridges with the 2 -phosphomonoester group. The adenine ribose moiety has only hydrophobic interactions with the side chain of Leu 288. The first phosphate group forms several hydrogen bonds with the main chains of Gly 310, Thr 311, and Val 312 ; the second phosphate group forms a hydrogen bond with a conserved water molecule. The nicotinamide moiety of NADP also forms both hydrophobic and hydrophilic interactions with surrounding residues. The endocyclic oxygen atom of the nicotinamide ribose forms two hydrogen bonds with the main chain amide nitrogen and the side chain (O atom) of Thr 311. Its 2 -OH group makes hydrogen bonding interactions with the main chain amide nitrogen and the side chain (O ) ofthr 77, and its 3 -OH group forms hydrogen bonding interactions with the main chain amide nitrogen of Thr 77 and the side chain (N and N -2) of Arg 82. The amide group of the nicotinamide moiety forms hydrogen bonds with the main chain and side chain of Thr 75 and the side chains of Lys 72 (N ) and Asn96 (N -2), and its carbonyl group forms one hydrogen bond with the side chain (N -2) of Asn 96. Because of the conformational difference at the active site, the side chain of Asp 279 mimics the isocitrate substrate and makes hydrophobic interactions with the nicotina-

6 Regulatory Mechanism of Human Cytosolic IDH FIG. 3. Structure-based sequence alignment of HcIDH with other NADP-IDHs. The structures used in comparisons are PmIDH, EcIDH, and BsIDH. Invariant residues are highlighted as shaded red boxes, and conserved residues are indicated by open red boxes. The secondary structure of HcIDH in the quaternary complex is placed above the alignment. The alignment was drawn with ESPript (35). mide moiety of NADP in subunit A; the corresponding residue in subunit B is distant to NADP. In the quaternary complex, the residues involved in interactions with NADP are similar to those in the binary complex. The major difference occurs at Arg 314 (Fig. 4). In this structure, the side chain of Arg 314 spans the top of NADP, and its side chain N -1 and N -2 atoms form salt bridges with the side chain atoms (O -1 and O -2) of Asp 253 and make hydrogen bonding interactions with the side chains of Arg 249 (N ) and Gln 257 (O -1) of the adjacent subunit. The side chain (N -1) of Arg 314 also forms a hydrogen bond with the 3 -OH of the adenine ribose moiety. In addition, the 2 -phosphomonoester group of NADP forms two extra hydrogen bonds with the side chains of Gln 257 (O -1) and Lys 260 (N ) of the adjacent subunit. NADP Recognition and Specificity HcIDH is the first eukaryotic NADP-IDH whose structure has been determined both in complex with NADP and in complex with NADP, isocitrate, and a metal ion. Comparison of these structures with other NADP-IDH structures can reveal detailed information about the specific residues involved in recognition of NADP in eukaryotic NADP-IDHs. NADP and NAD are differentiated only by the 2 -phosphomonoester group of NADP, and the residues that have direct interactions with this group are considered to be the potential specificity determinants of NADP. In contrast to the strict conservation of the residues involved in substrate and metal ion binding in all dimeric NADP-IDHs, the residues involved in interactions with NADP, especially these interacting with the 2 -phosphomonoester group of NADP, vary considerably between eukaryotic and bacterial NADP-IDHs (Figs. 3 and 4). In the structure of the HcIDH binary complex, only residues Arg 314 and His 315 make interactions with the 2 phosphomonoester group, suggesting that Arg 314 and His 315

7 33952 Regulatory Mechanism of Human Cytosolic IDH FIG. 4. Structure of the NADP-binding site. A, comparison of the NADP-binding site in the binary complex (in silver) and the quaternary complex (in green). NADP, isocitrate, and Ca 2 are colored as in Fig. 2. The side chain of Arg 314 is shown in slate blue (in the binary complex) or in red (in the quaternary complex). Residues of the adjacent subunit in the quaternary complex are shown in blue. B, close-up of the electrostatic surface at the NADP-binding site in the binary complex (left panel) and the quaternary complex (right panel). are the primary specificity determinants of NADP in HcIDH. Because these two residues are completely conserved in eukaryotic NADP-IDHs (6), it is possible that all eukaryotic NADP-IDHs use the same two residues to recognize NADP. The corresponding residues in PmIDH are Arg 314 and His 315, respectively (Fig. 3). Mutation of His 315 by a glutamine in PmIDH increases the K m for NADP by 40-fold, suggesting that it is involved in interaction with NADP (28). In bacterial NADP-IDHs, the equivalent positions are occupied by two different, but conserved, residues corresponding to Lys 344 and Tyr 345 in EcIDH and Lys 350 and Tyr 351 in BsIDH, respectively. In the NADP-bound EcIDH structures, the 2 -phosphomonoester group of NADP has interactions with the side chains of Tyr 345, Tyr 391, and Arg 395, which are suggested to be the specificity determinants of NADP in EcIDH; the side chain of Lys 344 is either disordered or has no contact with NADP (14, 16). In addition to participating in NADP recognition and specificity determination, Arg 314 appears to have additional functional role(s) in the enzymatic reaction. Compared with the binary complex, substantial conformational changes occur in both the local structure of the active site and the overall structure, and the large and small domains of one subunit and the small domain of the adjacent subunit are brought together much closer to each other in the quaternary complex (see discussion later) (Figs. 2 and 5). Meanwhile, the side chain of

8 Regulatory Mechanism of Human Cytosolic IDH FIG. 5. Conformational differences at the active site. A, subunit A of the HcIDH binary complex. B, subunit B of the HcIDH binary complex. C, subunit A of the HcIDH quaternary complex. D, structure of the PmIDH-isocitrate-Mn 2 complex (Protein Data Bank code 1LWD). E, structure of the EcIDH-NADP-isocitrate-Ca 2 complex (Protein Data Bank code 1AI2). F, structure of the BsIDH-citrate complex (Protein Data Bank code 1HQS). Isocitrate (ICT, in gold), NADP (in magenta), and the side chains of several important residues (in green) are shown with ball-and-stick models, and metal ions appear as spheres (in cyan). Arg 314 is oriented differently, enabling it to wrap around NADP and interact with Asp 253, Gln 257, and Arg 249 from the small domain of the adjacent subunit via both salt bridges and hydrogen bonds (Fig. 4). These interactions appear not only to prevent NADP from dissociation but also to bring the structural elements and amino acids involved in interacting with the substrate, metal ion, and cofactor in precise positions so that the catalysis can occur efficiently and effectively. These interactions have not been observed in any other NADP-IDH structures, and the residues involved are highly conserved in eukaryotic NADP-IDHs but different in bacterial NADP-IDHs. These results suggest that eukaryotic and bacterial NADP-

9 33954 Regulatory Mechanism of Human Cytosolic IDH IDHs use a different set of amino acids to recognize and to interact NADP and might have acquired their NADP specificity differently during evolution. Conformational Differences between Different Structures In the structure of the quaternary complex, the four monomers in the asymmetric unit adopt almost identical conformations except for minor differences in the clasp domain and in a few surface-exposed regions. The RMSD values of the superposition of any pair of subunits are in the range of Å for 411 C atoms. In the structure of the binary complex, the conformations of the two monomers of the homodimer differ substantially from each other and from that in the quaternary complex. Superposition of the two subunits yields an RMSD value of 1.44 Å for 398 C atoms excluding the region of residues Structural comparisons reveal that these three conformationally distinct subunits differ in both the local structure of the active site and the overall structure, although their secondary structures are very similar. A structural segment (residues ) at the active site (we designate it as the regulatory segment) folds differently in the three conformers (Fig. 5). In the quaternary complex, it forms an -helix ( 10) similar to that observed in all other NADP-IDH structures. However, this conserved -helix is completely unwound into a loop in subunit A of the binary complex and is partially unwound such that the C-terminal portion (residues ) retains an -helical conformation, whereas the N-terminal portion (residues ) is unraveled into a coiled coil in subunit B. The overall conformational difference between the two subunits of the binary complex is localized to a hinge movement of the clasp domain relative to the rest of the structure. Superposition of the clasp domain itself exhibits an RMSD value of 1.14 Å for 49 C atoms; superposition of the remaining part yields an RMSD of 0.69 Å for 349 C atoms. The hinge point is the region of residues that forms a loop connecting the small and clasp domains. This region exhibits relatively high B factor in subunit A and has weak electron density in subunit B, suggesting significant flexibility. The equivalent region in the quaternary complex has well defined electron density and forms a one-turn -helix ( 1) similar to that observed in other NADP-IDH structures (Fig. 3). It is conceivable that the high mobility of this region is correlated with the conformational difference between the clasp domain in the two conformers. The overall conformational difference between the two conformers in the binary complex and the conformer in the quaternary complex is manifested by a hinge movement of the clasp and small domains relative to the large domain. The hinge point is located in the region of residues ( 4) in the middle of the central -sheet joining the small and large domains. Superposition of the large domain by itself shows a fairly good fit (an RMSD value of 0.61 Å for 229 C atoms); superposition of the small and clasp domains yields an RMSD of 1.09 Å for 159 C atoms (excluding residue regions and ). Distinct Conformations May Represent Different Biologically Relevant Enzymatic States The two HcIDH complex structures contain three different conformational states of the enzyme. The overall conformation of HcIDH in the quaternary complex resembles that in the closed form of EcIDH, BsIDH, and PmIDH structures in their complexes with substrates. The active site cleft adopts a compact or closed conformation, and the width of the active site entrance is 12.8 Å (defined as the distance between residues 76 and 250 of the adjacent subunit or their equivalents in other NADP-IDH structures), which is comparable with that in the closed forms of EcIDH (14.0 Å), BsIDH (13.9 Å), and PmIDH (13.7 Å) (Fig. 2). In the meanwhile, the back cleft assumes an open conformation with a width of 16.7 Å (defined as the distance between residues 199 and 342 of the large domain or their equivalents in other NADP-IDH structures), which is also similar to that in the PmIDH structure (15.5 Å). The widths of the back clefts in the closed forms of EcIDH and BsIDH are 12.1 and 12.8 Å, respectively, which are slightly narrower presumably because of the structural differences nearby, especially at the N termini of these bacterial enzymes (Fig. 3). Because the closed form of EcIDH is considered to be the active form and the quaternary complex of HcIDH is a pseudo-michaelis complex, we therefore propose that the closed conformer of HcIDH also represents the active form of the enzyme. The overall conformation of subunit A in the binary complex is more similar to the open form of EcIDH in the absence of ligand, which is regarded as the inactive form (16), than to the closed form of HcIDH as well as the closed forms of EcIDH, BsIDH, and PmIDH. The active site cleft assumes a widely open conformation, and the opening to the active site is 21.2 Å, which is comparable with that in the open form of EcIDH (21.3 Å) (Fig. 2). As a concerted conformational change, the back cleft adopts a closed conformation with a width of 10.9 Å (the width of the back cleft is 9.1 Å in the open form of EcIDH). We propose that the open conformer of HcIDH may represent the inactive form of the enzyme. The conformation of HcIDH in subunit B of the binary complex is between the open form and the closed form of HcIDH. The width of the active site cleft (18.8 Å) is wider than that in the substrate-bound NADP-IDH structures but is narrower than that in the substrate-absent NADP-IDH structures. The width of the back cleft (11.7 Å) is also intermediate between those in the active form and inactive form of NAPD-IDHs. We propose that this semi-open form of HcIDH might represent an intermediate form during the transition from the inactive form to the active form. Distinct conformational states have been observed in a number of enzymes, and the differences range from movement of domains and fragments to specific structural elements on the surface. Conformational changes could be correlated with the biological functions of the enzyme. They could also be caused by crystal packing or by artificial crystallization conditions (such as excessive amount of substrate and metal ion, salt, specific solvent, or different ph). In both HcIDH complex structures, the HcIDH molecules have weak crystal packing contacts, and each domain of HcIDH appears to be able to undergo conformational change easily. The distinct conformations of the two subunits of the homodimer in the binary complex are not caused by cofactor binding because the bound NADP is copurified with the enzyme in preparation. Moreover, the regulatory segment is located inside the protein core, which should not be directly affected by protein-protein interaction. Therefore, we suggest that the conformational differences of the regulatory segment in the three conformers of HcIDH are likely related to the biological functions of the enzyme, instead of an artifact, and that the distinct conformations of HcIDH observed in the two complex structures are likely biologically relevant and may represent different enzymatic states in the catalytic reaction. A Self-regulatory Mechanism of HcIDH Regulation of the enzymatic activity of IDHs plays an important role in their biological functions. The activity of EcIDH (and most likely other bacterial IDHs) is reversibly regulated by phosphorylation of Ser 113 or its equivalent at the active site (14, 19, 20). Although it has been suggested that mammalian NADP- IDHs might also contain a putative phosphorylation site (22), no evidence of phosphorylation of any mammalian NADP-

10 Regulatory Mechanism of Human Cytosolic IDH FIG. 6. Schematic diagram of the conformational changes during the regulation of HcIDH activity. The competitive binding of isocitrate (and metal ion) induces the conformational changes. IDH has been reported to date. Comparisons of the crystal structures of HcIDH complexes with other NADP-IDH structures lead us to propose a novel self-regulatory mechanism for HcIDH. Our working model postulates that HcIDH may utilize the unfolding of the regulatory segment and the interactions of Asp 279 with Ser 94 to inactivate the enzyme and the competitive binding of isocitrate to induce the refolding of the regulatory segment and the overall conformational changes and, consequently, to activate the enzyme (Figs. 5 and 6). In the open, inactive form of HcIDH, the regulatory segment adopts a loop conformation and protrudes into the isocitrate-binding site. The side chain of Asp 279 occupies the exact position where the isocitrate substrate takes and forms hydrogen bonds with the side chain of Ser 94 (the equivalent of the phosphorylation site in EcIDH) as well as the side chains of Thr 77 and Asn 96, possibly blocking the access of isocitrate to the active site by a combination of steric hindrance and electrostatic repulsion (Figs. 2 and 5). This mechanism appears to mimic the inactivation of EcIDH by the phosphorylation of Ser 113. In the semiopen form of HcIDH, the regulatory segment adopts a partially unwound conformation, and the side chain of Asp 279 forms hydrogen bonding interactions with the side chains of Gln 277 and Arg 132. The fact that two subunits of a homodimer assume two distinct conformations in one crystal structure suggests that the open conformer and the semi-open conformer of HcIDH likely coexist in equilibrium in solution and are energetically more stable than the closed conformer in the absence of substrate. The conformational transition between the two conformers might occur easily with only a small free energy barrier. As the concentration of isocitrate (and metal ion) reaches to a certain level, isocitrate likely binds dominantly at the active site that would probably break the hydrogen bonding interactions of Asp 279 with Ser 94, Thr 77, and Asn 96 and expel Asp 279 and the regulatory segment away. We propose that this competitive binding of substrate induces the refolding of the regulatory segment back into an -helix and the conformational change of the overall structure, including the closing down of the active site cleft and the opening up of the back cleft. As a result, the enzyme adopts the closed, active form in which the side chain of Asp 279 forms hydrogen bonds with the Ca 2 ion to participate in the catalytic reaction. The newly formed hydrogen bonding interactions between isocitrate and the surrounding residues will likely compensate for the energy costs of the loss of hydrogen bonding interactions between Asp 279 and the surrounding residues and the transition of the enzyme from the open form to the closed form. The numerous hydrophobic and hydrophilic interactions of the -helical regulatory segment with the neighboring structural elements ( 9 of the same subunit and 9 and 10 of the adjacent subunit) at the dimer interface might also contribute to the energy costs. Similar conformational change from the open form to the closed form has been observed in EcIDH, and this transition has been suggested to be necessary during the phosphorylation of EcIDH (16). After the completion of catalysis and the release of the -ketoglutarate product and NADPH, the enzyme may either reload NADP and isocitrate to continue the catalytic reaction or change its conformation back to the open, inactive form. The transition of the closed form into the open form could proceed in two ways. One possibility is that the regulatory segment itself initiates the unwinding action to change from the -helical conformation into the loop conformation that consequently triggers the overall conformational change. The occurrence of this event relies on the assumption that the open, inactive form of HcIDH is energetically more stable than the closed form in the absence of isocitrate. Another possibility is that the interactions between the residues in the back cleft trigger the conformational changes of the large and small domains of the same subunit and the closing down of the back cleft. This further induces the conformational change at the active site cleft, including the unfolding of the regulatory segment and the opening up of the active site cleft, leading to the formation of the open, inactive form of the enzyme. These possibilities are not mutually exclusive and instead likely correlate with each other. Supporting evidence for this mechanism comes from both biochemical and crystallographic data. Kinetic studies of bovine cytosolic IDH showed a distinct lag time in enzymatic reaction and a transition of the enzyme from the inactive form to the active form (29). Preincubation with isocitrate and metal nearly abolished the lag time and increased the rate constant for the transition by 10-fold. Spectroscopic data also indicate that substrate binding induces conformational change and activates the enzymatic activity (30). Enzymatic assay showed that the dissolved solution of the HcIDH binary complex crystals retained full enzymatic activity equivalent to that of the freshly purified enzyme (data not shown). Soaking of the binary complex crystals with 5 mm isocitrate and 5 mm Ca 2 (or Mn 2 )

11 33956 Regulatory Mechanism of Human Cytosolic IDH 2 X. Xu and J. Ding, unpublished data. for 20 min yielded a structure in which the overall conformations of the two subunits of the homodimer are similar to that in the binary complex, whereas the regulatory segment in both subunits becomes disordered. 2 Soaking with a higher concentration of isocitrate (50 mm) for 1 2 h yielded crystals that had no obvious difference in morphology but diffracted x-ray very poorly (below 5 Å resolution with smeared diffraction spots). Increasing the concentration of isocitrate to 100 mm in the soaking solution led to the complete dissolution of the crystals within 20 min. These results suggest that soaking with increasing concentrations of isocitrate induced conformational changes in the regulatory segment and the overall structure and changed the crystal lattice packing. The proposed regulatory mechanism can readily explain some biochemical and biological data. Both mammalian cytosolic and mitochondrial NADP-IDHs can be modified by chemical reagents, such as nitric oxide, N-ethylmaleimide, and 4-hydroxynonenal that react easily with protein sulfydryl groups, leading to the inactivation of the enzymes in both in vitro and in vivo studies (31 33). This inactivation can be prevented by the addition of the isocitrate substrate or partially reversed by a thiol. Nitric oxide can cause damaging effects on antioxidant enzymes in vivo and lead to the perturbation of the cellular antioxidant defense system. It is found that nitric oxide forms an S-nitrosothiol adduct on Cys 379 of PmIDH (32). Cys 379 of HcIDH and PmIDH is located in helix 14 and is far away from the active site and not involved in the binding of substrate, metal ion, or cofactor (Fig. 5). The equivalent residue in bacterial enzyme is a leucine (corresponding to Leu 396 in EcIDH and Leu 402 in BsIDH, respectively) that could not form adduct with these reagents. In the substrate-bound HcIDH and PmIDH structures, Cys 379 is situated next to the C terminus of the regulatory segment, and its side chain forms a hydrophilic interaction with the main chain carbonyl group of Gly 286.Inthe HcIDH binary complex, Cys 379 is 5 Å away from Gly 286 because of the unfolding of the regulatory segment. It is partially exposed to solvent and is accessible to and can be modified by nitric oxide (or other chemical reagents). We propose that formation of the S-nitrosothiol adduct on Cys 379 is likely to prevent the refolding of the regulatory segment from the loop conformation in the inactive form to the -helical conformation in the active form, consequently inactivating the enzymatic activity. A thiol reagent can reduce the modified Cys 379 back to the reduced state and thus can restore the activity of the enzyme. The addition of substrate induces conformational change of the enzyme from the inactive form to the active form in which the interactions of Cys 379 with Gly 286 and other surrounding residues might prevent it from reacting with nitric oxide (or other chemical reagents). Biological Implication Structural studies of HcIDH have led us to propose a novel self-regulatory mechanism for HcIDH that is in marked contrast to the phosphorylation mechanism of bacterial IDHs. Considering the high degree of homology of HcIDH and bacterial IDHs at the three-dimensional level, the significant conservation of amino acid residues at the active site, and the strict conservation of Ser 94 and Asp 279 or their equivalents in HcIDH and bacterial IDHs, it is very striking that they utilize different mechanisms to regulate their enzymatic activities. It is possible that HcIDH and bacterial IDHs have evolved divergently from a common ancestral protein or that HcIDH evolves its activity regulatory mechanism from the phosphorylation mechanism of bacterial IDHs. This evolution might be correlated with its specific function(s) and/or the specific physiological environment where it exerts the function(s), for example, the specific biological functions of HcIDH in peroxisomes. Search of the NCBI databases using BLAST (34) did not identify any putative IDH-K/P or IDH-K/P homologs in humans and other eukaryotes. The lack of IDH-K/P in eukaryotes might force the enzyme to develop an alternative mechanism to control its activity. Because eukaryotic cytosolic IDHs, in particular mammalian cytosolic IDHs, display a high degree of amino acid conservation, it is very likely that this regulatory mechanism may apply to other eukaryotic or at least other mammalian cytosolic IDHs. Moreover, because eukaryotic cytosolic and mitochondrial IDHs share 70% sequence identity and high structural homology, it is possible that this regulatory mechanism is also applicable to other mitochondrial NADP-IDHs, such as PmIDH. However, if the proposed regulatory mechanism of HcIDH is correlated with its functional roles in peroxisomes, this mechanism will not apply to mitochondrial NADP-IDHs because they could not enter into the peroxisomes because of the lack of the type 1 peroxisomal targeting sequence motif. Structure determination of PmIDH in the absence of isocitrate and metal ion will reveal whether the region equivalent to the regulatory segment of HcIDH would adopt an alternative loop conformation and will eventually resolve this issue. It will have broader biological implication if this divergent evolution of structure and function of isoenzymes from different species can be found in other enzyme families. Acknowledgments We thank the staff members at the Cornell High Energy Synchrotron Source for support in data collection. REFERENCES 1. Haselbeck, R. J., and McAlister-Henn, L. (1993) J. Biol. Chem. 268, Jennings, G. T., Sechi, S., Stevenson, P. M., Tuckey, R. C., Parmelee, D., and McAlister-Henn, L. (1994) J. Biol. Chem. 269, Jo, S. H., Son, M. K., Koh, H. J., Lee, S. M., Song, I. H., Kim, Y. O., Lee, Y. S., Jeong, K. S., Kim, W. B., Park, J. W., Song, B. J., Huh, T. L., and Huhe, T. L. (2001) J. Biol. Chem. 276, Lee, S. M., Koh, H. J., Park, D. C., Song, B. J., Huh, T. L., and Park, J. W. (2002) Free Radic. Biol. Med. 32, Kim, S. Y., and Park, J. W. (2003) Free Radic. Res. 37, Nekrutenko, A., Hillis, D. M., Patton, J. C., Bradley, R. D., and Baker, R. J. (1998) Mol. Biol. Evol. 15, Gould, S. J., and Collins, C. S. (2002) Nat. Rev. Mol. Cell Biol. 3, Henke, B., Girzalsky, W., Berteaux-Lecellier, V., and Erdmann, R. (1998) J. Biol. Chem. 273, van Roermund, C. W., Hettema, E. H., Kal, A. J., van den Berg, M., Tabak, H. F., and Wanders, R. J. (1998) EMBO J. 17, Geisbrecht, B. V., and Gould, S. J. (1999) J. Biol. Chem. 274, Minard, K. I., and McAlister-Henn, L. (1999) J. Biol. Chem. 274, Yoshihara, T., Hamamoto, T., Munakata, R., Tajiri, R., Ohsumi, M., and Yokota, S. (2001) J. Histochem. Cytochem. 49, Hurley, J. H., Thorsness, P. E., Ramalingam, V., Helmers, N. H., Koshland, D. E., Jr., and Stroud, R. M. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, Hurley, J. H., Dean, A. M., Sohl, J. L., Koshland, D. E., Jr., and Stroud, R. M. (1990) Science 249, Stoddard, B. L., Dean, A., and Koshland, D. E., Jr. (1993) Biochemistry 32, Finer-Moore, J., Tsutakawa, S. E., Cherbavaz, D. R., LaPorte, D. C., Koshland, D. E., Jr., and Stroud, R. M. (1997) Biochemistry 36, Mesecar, A. D., Stoddard, B. L., and Koshland, D. E., Jr. (1997) Science 277, LaPorte, D. C. (1993) J. Cell. Biochem. 51, Dean, A. M., and Koshland, D. E. J. (1990) Science 249, Singh, S. K., Matsuno, K., LaPorte, D. C., and Banaszak, L. J. (2001) J. Biol. Chem. 276, Singh, S. K., Miller, S. P., Dean, A., Banaszak, L. J., and LaPorte, D. C. (2002) J. Biol. Chem. 277, Ceccarelli, C., Grodsky, N. B., Ariyaratne, N., Colman, R. F., and Bahnson, B. J. (2002) J. Biol. Chem. 277, Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, Pflugrath, J. W. (1999) Acta Crystallogr. Sect. D Biol. Crystallogr. 55,