Portal de Periódicos da CAPES

Crystal Structure of Human DJ-1, a Protein Associated with Early Onset Parkinson's Disease

2003; Elsevier BV; Volume: 278; Issue: 33 Linguagem: Inglês

10.1074/jbc.m304221200

1083-351X

Xiao Tao, Liang Tong,

Alzheimer's disease research and treatments

We report the crystal structure at 1.8-Å resolution of human DJ-1, which has been linked to early onset Parkinson's disease. The monomer of DJ-1 contains the α/β-fold that is conserved among members of the DJ-1/ThiJ/PfpI superfamily. However, the structure also contains an extra helix at the C terminus, which mediates a novel mode of dimerization for the DJ-1 proteins. A putative active site has been identified near the dimer interface, and the residues Cys-106, His-126, and Glu-18 may play important roles in the catalysis by this protein. Studies with the disease-causing L166P mutant suggest that the mutation has disrupted the C-terminal region and the dimerization of the protein. The DJ-1 proteins may function only as dimers. The Lys to Arg mutation at residue 130, the site of sumoylation of DJ-1, has minimal impact on the structure of the protein. We report the crystal structure at 1.8-Å resolution of human DJ-1, which has been linked to early onset Parkinson's disease. The monomer of DJ-1 contains the α/β-fold that is conserved among members of the DJ-1/ThiJ/PfpI superfamily. However, the structure also contains an extra helix at the C terminus, which mediates a novel mode of dimerization for the DJ-1 proteins. A putative active site has been identified near the dimer interface, and the residues Cys-106, His-126, and Glu-18 may play important roles in the catalysis by this protein. Studies with the disease-causing L166P mutant suggest that the mutation has disrupted the C-terminal region and the dimerization of the protein. The DJ-1 proteins may function only as dimers. The Lys to Arg mutation at residue 130, the site of sumoylation of DJ-1, has minimal impact on the structure of the protein. Parkinson's disease (PD) 1The abbreviations used are: PD, Parkinson's disease; PIASxα, protein inhibitor of activated STAT; STAT, signal transducers and activators of transcription; DJBP, DJ-1-binding protein; GAT, glutamine amidotransferase; r.m.s., root mean square; HPII, hydroperoxidase II. is a common, progressive neurodegenerative disorder affecting roughly 1% of the population at the age of 65 (1Dawson T.M. Dawson V.L. J. Clin. Invest. 2003; 111: 145-151Crossref PubMed Scopus (206) Google Scholar). Clinically, PD generally presents with brady-kinesia, resting tremor, muscular rigidity, and postural instability. PD is a heterogeneous disease, and the majority of the cases appear to have sporadic origins. At the same time, the disorder can also be associated with specific genetic defects, especially for cases of familial PD (2Gasser T. J. Neurol. 2001; 248: 833-840Crossref PubMed Scopus (128) Google Scholar, 3Giasson B.I. Lee V.M.-Y. Neuron. 2001; 31: 885-888Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar, 4Cookson M.R. Neuron. 2003; 37: 7-10Abstract Full Text Full Text PDF PubMed Scopus (91) Google Scholar). Ten different genetic loci have been linked with familial PD, and the genes responsible for PD at these loci include the presynaptic protein α-synuclein (5Polymeropoulos M.H. Lavedan C. Leroy E. Ide S.E. Dehejia A. Dutra A. Pike B. Root H. Rubenstein J. Boyer R. Stenroos E.S. Chandrasekharappa S. Athanassiadou A. Papapetropoulos T. Johnson W.G. Lazzarini A.M. Duvoisin R.C. di Iorio G. Golbe L.I. Nussbaum R.L. Science. 1997; 276: 2045-2047Crossref PubMed Scopus (6734) Google Scholar), parkin (6Kitada T. Asakawa S. Hattori N. Matsumine H. Yamamura Y. Minoshima S. Yokochi M. Mizuno Y. Shimizu N. Nature. 1998; 392: 605-608Crossref PubMed Scopus (4231) Google Scholar), and ubiquitin carboxyl-terminal hydrolase L1 (7Liu Y. Fallon L. Lashuel H.A. Liu Z. Lansbury Jr., P.T. Cell. 2002; 111: 209-218Abstract Full Text Full Text PDF PubMed Scopus (699) Google Scholar). Most recently, it was discovered that mutations in the DJ-1 gene are linked with autosomal recessive early onset familial PD (8Bonifati V. Rizzu P. van Baren M.J. Schaap O. Breedveld G.J. Krieger E. Dekker M.C.J. Squitieri F. Ibanez P. Joosse M. van Dongen J.W. Vanacore N. van Swieten J.C. Brice A. Meco G. van Duijn C.M. Oostra B.A. Heutink P. Science. 2003; 299: 256-259Crossref PubMed Scopus (2287) Google Scholar). Two types of disruptions of this gene have been identified in PD patients. One is a deletion of several of its exons, which abolishes the production of the DJ-1 protein. The other disruption is a single point mutation, giving rise to the L166P mutant at the protein level. This mutant has a different cytoplasmic distribution and is believed to be functionally inactive. Therefore, loss of the normal function of DJ-1 appears to contribute to PD. The exact biological function of the DJ-1 protein is currently unknown. It may play a role in the oxidative stress response, and this function could be important in preventing the onset of PD (8Bonifati V. Rizzu P. van Baren M.J. Schaap O. Breedveld G.J. Krieger E. Dekker M.C.J. Squitieri F. Ibanez P. Joosse M. van Dongen J.W. Vanacore N. van Swieten J.C. Brice A. Meco G. van Duijn C.M. Oostra B.A. Heutink P. Science. 2003; 299: 256-259Crossref PubMed Scopus (2287) Google Scholar). Both α-synuclein and parkin participate in oxidative stress responses as well (9Hashimoto M. Hsu L.J. Rockenstein E. Takenouchi T. Mallory M. Masliah E. J. Biol. Chem. 2002; 277: 11465-11472Abstract Full Text Full Text PDF PubMed Scopus (173) Google Scholar, 10Hyun D.-H. Lee M. Hattori N. Kubo S.-I. Mizuno Y. Halliwell B. Jenner P. J. Biol. Chem. 2002; 277: 28572-28577Abstract Full Text Full Text PDF PubMed Scopus (160) Google Scholar). On the other hand, DJ-1 may also be associated with several other biological processes. It was first identified as an oncogene, because it can transform NIH3T3 cells in cooperation with the ras oncogene (11Nagakubo D. Taira T. Kitaura H. Ikeda M. Tamai K. Iguchi-Ariga S.M.M. Ariga H. Biochem. Biophys. Res. Commun. 1997; 231: 509-513Crossref PubMed Scopus (672) Google Scholar). The DJ-1 protein is also involved in the fertilization process in rat and mouse (12Wagenfeld A. Gromoll J. Cooper T.G. Biochem. Biophys. Res. Commun. 1998; 251: 545-549Crossref PubMed Scopus (95) Google Scholar, 13Welch J.E. Barbee R.R. Roberts N.L. Suarez J.D. Klinefelter G.R. J. Androl. 1998; 19: 385-393PubMed Google Scholar, 14Okada M. Matsumoto K.-I. Niki T. Taira T. Iguchi-Ariga S.M.M. Ariga H. Biol. Pharm. Bull. 2002; 25: 853-856Crossref PubMed Scopus (83) Google Scholar). A significant reduction in the amount of this protein on the surface of sperm makes them unable to fertilize eggs (15Wagenfeld A. Yeung C.-H. Shivaji S. Sundareswaran V.R. Ariga H. Cooper T.G. J. Androl. 2000; 21: 954-963PubMed Google Scholar). This finding suggests that the protein may be secreted under some circumstances, which has also been observed in breast cancers (16Le Naour F. Misek D.E. Krause M.C. Deneux L. Giordano T.J. Scholl S. Hanash S.M. Clin. Cancer Res. 2001; 7: 3328-3335PubMed Google Scholar). Finally, DJ-1 was identified as the regulatory subunit of a 400-kDa RNA-binding protein complex and its presence inhibits the binding of RNA by the complex (17Hod Y. Pentyala S.N. Whyard T.C. El-Maghrabi M.R. J. Cell. Biochem. 1999; 72: 435-444Crossref PubMed Scopus (173) Google Scholar). The DJ-1 protein is a positive regulator of the androgen receptor by sequestering its negative regulators PIASxα (protein inhibitor of activated STAT) or DJBP (DJ-1-binding protein) (18Takahashi K. Taira T. Niki T. Seino C. Iguchi-Ariga S.M.M. Ariga H. J. Biol. Chem. 2001; 276: 37556-37563Abstract Full Text Full Text PDF PubMed Scopus (289) Google Scholar, 19Niki T. Takahashi-Niki K. Taira T. Iguchi-Ariga S.M.M. Ariga H. Mol. Cancer Res. 2003; 1: 247-261PubMed Google Scholar). The activation of the androgen receptor might be related to the effects of this protein on fertility. PIAS proteins are SUMO-1 (small ubiquitin-like modifier-1) ligases and control their target proteins by sumoylation (20Kotaja N. Karvonen U. Janne O.A. Palvimo J.J. Mol. Cell. Biol. 2002; 22: 5222-5234Crossref PubMed Scopus (355) Google Scholar). It has been reported that PIASxα can sumoylate DJ-1 on the Lys-130 residue (18Takahashi K. Taira T. Niki T. Seino C. Iguchi-Ariga S.M.M. Ariga H. J. Biol. Chem. 2001; 276: 37556-37563Abstract Full Text Full Text PDF PubMed Scopus (289) Google Scholar). DJBP is almost exclusively expressed in the testis in humans. It negatively modulates the androgen receptor by recruiting histone deacetylases. DJ-1 can inhibit this recruitment, thereby leading to the activation of the androgen receptor (19Niki T. Takahashi-Niki K. Taira T. Iguchi-Ariga S.M.M. Ariga H. Mol. Cancer Res. 2003; 1: 247-261PubMed Google Scholar). The human DJ-1 protein contains 189 amino acid residues. It belongs to the DJ-1/ThiJ/PfpI superfamily of proteins, which are conserved in many different organisms (Fig. 1). The function of ThiJ is currently unknown, although it might be related to the biosynthesis of thiamin (21Mizote T. Tsuda M. Smith D.D.S. Nakayama H. Nakazawa T. Microbiology. 1999; 145: 495-501Crossref PubMed Scopus (41) Google Scholar, 22Taylor S.V. Kelleher N.L. Kinsland C. Chiu H.-H. Costello C.A. Backstrom A.D. McLafferty F.W. Begley T.P. J. Biol. Chem. 1998; 273: 16555-16560Abstract Full Text Full Text PDF PubMed Scopus (137) Google Scholar). PfpI is an intracellular protease that was first identified from the archaeon Pyrococcus furiosus (23Halio S.B. Blumentals I.I. Short S.A. Merrill B.M. Kelly R.M. J. Bacteriol. 1996; 178: 2605-2612Crossref PubMed Google Scholar), and it is present in most bacteria and archaea. The crystal structure of the closely related intracellular protease PH1704 from Pyrococcus horikoshii revealed the presence of a Cys-100/His-101/Glu-74′ catalytic triad (primed residue indicate a different monomer) with the active site at the interface of neighboring monomers in a hexameric oligomer (24Du X. Choi I.-G. Kim R. Wang W. Jancarik J. Yokota H. Kim S.-H. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 14079-14084Crossref PubMed Scopus (98) Google Scholar). A Cys-185/His-186/Asp-214 catalytic triad was found in a structural homolog of this protease, the heat shock protein Hsp31 from E. coli (25Quigley P.M. Korotkov K. Baneyx F. Hol W.G.J. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 3137-3142Crossref PubMed Scopus (104) Google Scholar). The amino acid sequence of this chaperone, however, shares only a 19% identity with that of PH1704. The DJ-1/ThiJ/PfpI proteins also share limited sequence and structural homology to those of the type I glutamine amidotransferase (GAT) domains, which contain a Cys-His-Glu/Asp catalytic triad (26Horvath M.M. Grishin N.V. Proteins. 2001; 42: 230-236Crossref PubMed Scopus (32) Google Scholar). Although the Cys residue in the active site of the PfpI proteases is conserved among all of the members of the DJ-1/ThiJ/PfpI superfamily, the His and Asp/Glu residues are not conserved in DJ-1 and its close homologs (Fig. 1). To determine the composition of this putative active site in DJ-1 and to obtain a better understanding for the possible molecular function of this protein, we have determined its crystal structure at a 1.8-Å resolution. We have also characterized the disease-causing L166P mutant in solution, which suggests that the DJ-1 protein may function only as a dimer. Protein Expression and Purification—The gene for human DJ-1 was subcloned into the pET24d vector and overexpressed in E. coli at 20 °C. The recombinant protein contains a hexa-histidine tag at the C terminus. After cell lysis, the soluble protein was purified by nickel-agarose affinity, anion exchange, and gel-filtration chromatography. The protein was concentrated to 30 mg/ml in a buffer containing 20 mm Tris (pH 8.0), 150 mm NaCl, 3 mm DTT, and 5% (v/v) glycerol and stored at –80 °C. There are no Trp residues in DJ-1, and the protein concentration was determined by the Bradford method. The C-terminal His tag was not removed for crystallization. For the production of selenomethionyl proteins, the expression construct was transformed into the methionine auxotroph E. coli DL41(DE3) cells. The bacterial growth was carried out in defined LeMaster media (27Hendrickson W.A. Horton J.R. LeMaster D.M. EMBO J. 1990; 9: 1665-1672Crossref PubMed Scopus (1008) Google Scholar), and the protein was purified using the same protocol as for the wild-type protein. Mutagenesis—The L166P and K130R single site mutants of human DJ-1 were created with the QuikChange kit (Stratagene) and sequenced to confirm the incorporation of the correct mutation. The mutant proteins were expressed in E. coli and purified following the same protocol as for the wild-type protein. Protein Crystallization—Crystals of the DJ-1 wild-type and mutant proteins were prepared by the vapor diffusion method at 4 °C. Initial crystallization conditions for the wild-type protein were identified by sparse-matrix screening with commercial kits (Hampton Research). The reservoir solution contained 100 mm Tris (pH 8.0) and 10–12% (w/v) polyethylene glycol 3350. The protein was at a concentration of 15 mg/ml (diluted 1:1 with 30 mm Tris (pH 8.0) from the original stock). Both hexagonal bipyramid and plate crystals appeared overnight. They were cryoprotected with a solution containing 100 mm Tris (pH 8.0) and 35% (w/v) PEG3350 and flash-frozen in liquid propane. For crystals of the K130R mutant, the reservoir contained 100 mm Bis-Tris (pH 7.0), 20% (w/v) PEG3350, and 50 mm CaCl2. The protein was at 10 mg/ml concentration pre-mixed with 100 mm Bis-Tris (pH 5.5), 25% (w/v) PEG3350, and 200 mm ammonium acetate. Crystals in the shape of thin plates were cryoprotected with a solution containing 100 mm Bis-Tris (pH 7.0) and 35% (w/v) PEG3350 and flash-frozen in liquid propane for data collection at 100 K. Data Collection and Processing—X-ray diffraction data were collected at 100 K on an ADSC (Area Detector Systems Corporation) CCD at the X4A beamline of Brookhaven National Laboratory. The diffraction images were processed and scaled with the HKL package (28Otwinowski Z. Minor W. Method Enzymol. 1997; 276: 307-326Crossref Scopus (38617) Google Scholar). A selenomethionyl single wavelength anomalous diffraction data set to 1.8-Å resolution was collected on a hexagonal bipyramid crystal. It belongs to space group P3121, with one molecule in the asymmetric unit. The plate-shaped crystal of the wild-type protein belongs to space group P21 with four dimers in the asymmetric unit. A diffraction data set to 2.2-Å resolution was collected for this crystal. The K130R mutant crystal belongs to space group C2 with a dimer in the asymmetric unit. Its diffraction data set extended to 1.7-Å resolution. The data processing statistics are summarized in Table I.Table ISummary of crystallographic informationProteinWild-typeWild-typeK130R mutantSpace groupP3121P21C2Cell parameters (a, b, c, β)75.66, 75.66, 75.4971.26, 83.66, 114.16, 100.5677.71, 78.17, 63.82, 108.88Maximum resolution (Å)1.82.21.7Number of observations237,493211,982112,333R merge (%)aR merge = ΣhΣi|Ihi - 〈Ih 〉|/ΣhΣi Ihi4.9 (20.6)9.4 (23.1)6.5 (26.4)Resolution range for refinement20-1.820-2.220-1.7Number of reflections41,32561,03136,794Completeness (%)92 (82)91 (83)93 (80)R-factorbR=∑h|Fho-Fhc|/∑hFho (%)19.0 (21.9)18.0 (19.3)19.1 (21.1)Free R-factorbR=∑h|Fho-Fhc|/∑hFho (%)21.5 (23.3)24.3 (27.5)23.1 (24.9)R.m.s. deviation in bond lengths (Å)0.0040.0050.005R.m.s. deviation in bond angles (°)1.31.31.3a R merge = ΣhΣi|Ihi - 〈Ih 〉|/ΣhΣi Ihib R=∑h|Fho-Fhc|/∑hFho Open table in a new tab Structure Determination and Refinement—The locations of four Se atoms were determined with the program SnB (29Weeks C.M. Miller R. J. Appl. Crystallogr. 1999; 32: 120-124Crossref Scopus (384) Google Scholar). Reflection phases to 1.8-Å resolution were calculated based on the single wavelength anomalous diffraction data and improved with the program SOLVE (30Terwilliger T.C. Berendzen J. Acta Crystallogr. Sec. D. 1999; 55: 849-861Crossref PubMed Scopus (3220) Google Scholar), which also automatically located 80% of the residues in the molecule. The full atomic model was built into the electron density with the program O (31Jones T.A. Zou J.Y. Cowan S.W. Kjeldgaard M. Acta Crystallogr. Sec. A. 1991; 47: 110-119Crossref PubMed Scopus (13014) Google Scholar). The structures of the monoclinic crystal form and the K130R mutant were determined by the molecular replacement method with the program COMO (32Jogl G. Tao X. Xu Y. Tong L. Acta Crystallogr. Sec. D. 2001; 57: 1127-1134Crossref PubMed Scopus (89) Google Scholar). The structure refinement was carried out with the program CNS (33Brunger A.T. Adams P.D. Clore G.M. DeLano W.L. Gros P. Grosse-Kunstleve R.W. Jiang J.-S. Kuszewski J. Nilges M. Pannu N.S. Read R.J. Rice L.M. Simonson T. Warren G.L. Acta Crystallogr. Sec. D. 1998; 54: 905-921Crossref PubMed Scopus (16979) Google Scholar). The statistics on the structure refinement are summarized in Table I. The Structure Determination—The crystal structure of wild-type human DJ-1 protein in a trigonal crystal form has been determined at 1.8-Å resolution by the selenomethionyl single wavelength anomalous diffraction method (34Hendrickson W.A. Science. 1991; 254: 51-58Crossref PubMed Scopus (1019) Google Scholar). The positions of the Se sites were determined by direct methods (29Weeks C.M. Miller R. J. Appl. Crystallogr. 1999; 32: 120-124Crossref Scopus (384) Google Scholar), which enabled the phasing of the reflections and automatic tracing of 80% of the residues in the protein (30Terwilliger T.C. Berendzen J. Acta Crystallogr. Sec. D. 1999; 55: 849-861Crossref PubMed Scopus (3220) Google Scholar). The current atomic model contains residues 2–188 of DJ-1 together with 199 water molecules. The last residue of the protein and the C-terminal hexa-histidine tag (LEHHHHHH) are not visible in the electron density map. The atomic model has excellent agreement with the experimental data with a R-factor of 19.0% (Table I). The r.m.s. deviation from ideal values in bond lengths and bond angles are 0.004 Å and 1.3°, respectively. All of the residues with the exception of Cys-106 are located in favored regions of the Ramachandran plot (data not shown). The Cys-106 residue has a strained main chain conformation (see below). We have also determined the crystal structure at 2.2-Å resolution of the wild-type protein in a monoclinic crystal form (Table I) by the molecular replacement method (32Jogl G. Tao X. Xu Y. Tong L. Acta Crystallogr. Sec. D. 2001; 57: 1127-1134Crossref PubMed Scopus (89) Google Scholar, 35Rossmann M.G. Acta Crystallogr. Sec. A. 1990; 46: 73-82Crossref PubMed Scopus (226) Google Scholar). This crystal contains four dimers of DJ-1 in the asymmetric unit, allowing us to assess the conformational flexibility of the monomer and dimer of DJ-1. The structure of the K130R mutant of human DJ-1 has been determined at 1.7-Å resolution by the molecular replacement method (Table I). This crystal form is not isomorphous to either of the native crystals. The structure shows that the K130R side chains in both monomers are involved in crystal-packing interactions with neighboring dimers in the crystal, and these interactions are different from those for the Lys-130 side chain in the wild-type crystals. Structure of the DJ-1 Monomer—The structure of the monomer of DJ-1 has the α/β-fold with 11 β-strands (β1–β11) and 8 α-helices (αA–αH) (Fig. 2A). The central β-sheet of the structure contains seven strands. The six parallel strands of this sheet are arranged similar to those in the Rossmann-fold with the distinction that the third strand is very short in the DJ-1 structure (β5 with only two residues) (Figs. 1 and 2A). Outside the central β-sheet, strands β3 and β4 form a β-hairpin structure and are involved in the dimerization of DJ-1 (see below). Strands β8 and β9 together with helix αF form a β-α-β motif, which contributes the conserved His-126 residue to the putative active site of DJ-1 (Fig. 2A). Most of the helices flank the two faces of the β-sheet (Fig. 2B). The only exception is helix αH at the extreme C terminus of DJ-1. It projects away from the rest of the protein and only contacts helices αA and αG in the monomer (Fig. 2B). Residues 181–187 in helix αH are conserved to be hydrophobic amino acids among the DJ-1 and ThiJ proteins (Fig. 1). These residues mediate the interactions with helices αA and αG as well as the dimerization of DJ-1 (see below). The Dimer of DJ-1—Gel-filtration and light-scattering studies showed that wild-type DJ-1 protein exists as dimers in solution. 2X. Tao and L. Tong, unpublished data. In the trigonal crystal of the wild-type protein, there is a monomer of DJ-1 in the asymmetric unit. The other monomer of the dimer is related by the crystallographic 2-fold symmetry axis (Fig. 3A). The αA helices of the two monomers come into close contact in the core of the dimer (Fig. 3B). At one edge of the interface, the β3-β4 hairpin is situated right next to the dimer 2-fold axis, such that a four-stranded anti-parallel β-sheet is formed across the dimer (Fig. 3A). At the other edge, the β11-αG loop interacts with residues near the C terminus of the other monomer, including those in helix αH (Fig. 3B). These interactions may also be important to stabilize the conformation of the αH helix at the C terminus of DJ-1. In the monoclinic crystal form of the wild-type protein, there are four dimers in the asymmetric unit. Structural comparisons among these dimers and the dimer in the trigonal crystal form show that the organizations of the dimers are essentially the same. The r.m.s. distance between equivalent Cα atoms of any pair of these dimers is approximately 0.35 Å. The αB helix shows conformational variability among the four dimers in the monoclinic crystal, and this helix also has weak electron density in the trigonal crystal. The organization of the dimer is also conserved in that of the K130R mutant with a r.m.s. distance of 0.3 Å for 374 pairs of equivalent Cα atoms of the wild-type and mutant dimers. The DJ-1 dimer has an extensive interface between the two monomers. A total of 1200 Å2 of the surface area of each monomer is buried at this interface, involving mostly residues that are conserved among DJ-1 proteins (Fig. 1). Our structural and solution observations therefore suggest that this dimer is likely to be a stable and conserved oligomerization state for these proteins. Interestingly, a deletion mutant of DJ-1 removing residues 178 to the C terminus (Fig. 1) is still mostly dimeric in solution (data not shown), suggesting that either the αH helix has a minor role in the dimerization of DJ-1 or the C-terminal hexahistidine tag (LEHHHHHH) is partially rescuing the dimer formation. On the other hand, the deletion mutant lacking residues 173 to the C terminus forms large aggregates of various sizes (molecular masses of 200–400 kDa) with little dimeric species in solution. Structural Homologs of DJ-1—The structure of DJ-1 bears the strongest similarity to that of PH1704 from P. horikoshii (24Du X. Choi I.-G. Kim R. Wang W. Jancarik J. Yokota H. Kim S.-H. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 14079-14084Crossref PubMed Scopus (98) Google Scholar), a member of the PfpI family, and the two proteins share 22% amino acid sequence identity. The r.m.s. distance is 1.1 Å for 137 equivalent Cα atoms between DJ-1 and PH1704. Most of the secondary structure elements are superimposable between the two proteins (Fig. 2C). However, the αH helix at the C terminus of DJ-1 does not have a structural equivalent in PH1704. Sequence comparisons suggest that all of the PfpI family members lack this helix, because they are approximately 15 residues shorter than DJ-1 at the C terminus (Fig. 1). The quaternary structure of PH1704 and DJ-1 are entirely different, despite the homology in the sequence and structure of their monomers. In contrast to the dimeric association of DJ-1, a hexameric structure roughly obeying 32 symmetry was observed in the crystal of PH1704 (24Du X. Choi I.-G. Kim R. Wang W. Jancarik J. Yokota H. Kim S.-H. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 14079-14084Crossref PubMed Scopus (98) Google Scholar). This hexamer can be considered as a trimer of dimers, and the interface of the dimer is mediated by the αA and αG helices. However, such a mode of dimerization is not possible with DJ-1 as it is blocked by the extra αH helix in DJ-1. Therefore, the DJ-1 dimer is formed by a different arrangement of the αA and αG helices of the two monomers and the αH helices also contribute to the dimerization interface (Fig. 3A). In addition, the contribution of the β3–β4 hairpin to the dimer interface is unique to DJ-1 (Fig. 3A). The dimerization of Hsp31, a structural homolog of PH1704 and DJ-1, is mediated by segments outside the PfpI/DJ-1-fold (25Quigley P.M. Korotkov K. Baneyx F. Hol W.G.J. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 3137-3142Crossref PubMed Scopus (104) Google Scholar). Searches against the Protein Data Bank with the program Dali (36Holm L. Sander C. J. Mol. Biol. 1993; 233: 123-138Crossref PubMed Scopus (3566) Google Scholar) showed that DJ-1 also shares significant structural similarity with the C-terminal domain in the catalase hydroperoxidase II (HPII) (37Bravo J. Mate M.J. Schneider T. Switala J. Wilson K. Loewen P.C. Fita I. Proteins. 1999; 34: 155-166Crossref PubMed Scopus (57) Google Scholar) as has been suggested earlier (26Horvath M.M. Grishin N.V. Proteins. 2001; 42: 230-236Crossref PubMed Scopus (32) Google Scholar). The r.m.s. distance for 136 equivalent Cα atoms between DJ-1 and this C-terminal domain (residues 600–753) is 2.0 Å. The amino acid identity for these residues is however only 12%. This domain in HPII lacks the β8–β10 insertion (Fig. 2D), which accounts for its smaller size. At the C terminus, HPII also has a helix but it is running in an opposite direction from the αH helix in DJ-1. The exact function of this C-terminal domain in HPII is not fully understood. It mediates the dimer interface in the tetramer of the enzyme and helps to form the substrate channel leading to the active site (37Bravo J. Mate M.J. Schneider T. Switala J. Wilson K. Loewen P.C. Fita I. Proteins. 1999; 34: 155-166Crossref PubMed Scopus (57) Google Scholar). This domain may also facilitate the folding of the full-length enzyme, as deletion mutants lacking this domain cannot accumulate in the cell. The dimerization of this domain in HPII uses the unique αH helix at the C terminus and is different from that in DJ-1. Structural homology between the DJ-1/ThiJ/PfpI superfamily and the GAT domains has also been suggested earlier (26Horvath M.M. Grishin N.V. Proteins. 2001; 42: 230-236Crossref PubMed Scopus (32) Google Scholar). However, the actual similarity is rather limited. The GAT domains lack the β3–β4 hairpin structure in DJ-1 but contain significantly more elaborate insertions between αE and β10. Moreover, the binding site for glutamine in GAT is blocked by the β6-αC loop in the DJ-1 structure. Therefore, it is unlikely that DJ-1 can have GAT function. The Putative Catalytic Residues of DJ-1—The exact biochemical function of the DJ-1 protein is currently not known. However, structural and sequence comparisons with homologous proteins can provide valuable hints as to the possible roles of this protein in oxidative stress response, androgen receptor regulation, and other processes. A cysteine residue is conserved among all of the members of the DJ-1/ThiJ/PfpI superfamily, and the residue has been proposed as the catalytic nucleophile for the PfpI family of intracellular proteases (24Du X. Choi I.-G. Kim R. Wang W. Jancarik J. Yokota H. Kim S.-H. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 14079-14084Crossref PubMed Scopus (98) Google Scholar). This residue is equivalent to Cys-106 in human DJ-1 (Fig. 1) and is located just after strand β7 in the structure (Fig. 2A). The main chain of this residue assumes a strained conformation with φ and ψ torsion angles of 66 and –114°, respectively, which is also observed in the structures of PH1704 and Hsp31 (24Du X. Choi I.-G. Kim R. Wang W. Jancarik J. Yokota H. Kim S.-H. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 14079-14084Crossref PubMed Scopus (98) Google Scholar, 25Quigley P.M. Korotkov K. Baneyx F. Hol W.G.J. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 3137-3142Crossref PubMed Scopus (104) Google Scholar). The structural conservation of this Cys residue in DJ-1 suggests that it may also have a catalytic role in the biochemical function of this protein, identifying this region of the structure as the putative active site of DJ-1. In both the PfpI proteases and the Hsp31 chaperone, the Cys residue is part of a Cys-His-Asp/Glu catalytic triad (24Du X. Choi I.-G. Kim R. Wang W. Jancarik J. Yokota H. Kim S.-H. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 14079-14084Crossref PubMed Scopus (98) Google Scholar, 25Quigley P.M. Korotkov K. Baneyx F. Hol W.G.J. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 3137-3142Crossref PubMed Scopus (104) Google Scholar). The second member His residue in the triad follows immediately after the Cys residue in the primary sequence of these proteins (Figs. 1 and 2C). In DJ-1, however, the residue immediately after Cys-106 is Ala (Fig. 1), signifying that there are differences between DJ-1 and the PfpI proteases in this active site. Our structures show that His-126, from the β8-αF loop and conserved among all of the DJ-1 proteins (Fig. 1), is placed near the Cys-106 side chain in the putative active site of DJ-1 (Fig. 4A). The His-126 residue could be the second member of the catalytic machinery for DJ-1. The distance between the side chains of Cys-106 and His-126 is 4.2 Å in the current structure (Fig. 4A) similar to the distance of approximately 5 Å that is found in the active site of caspases (38Watt W. Koeplinger K.A. Mildner A.M. Heinrikso