A two-domain structure of one subunit explains unique features of eukaryotic hydratase 2.

2-Enoyl-CoA hydratase 2, a part from multifunctional enzyme type 2, hydrates trans-2-enoyl-CoA to 3-hydroxyacyl-CoA in the (3R)-hydroxy-dependent route of peroxisomal beta-oxidation of fatty acids. Unliganded and (3R)-hydroxydecanoyl coenzyme A-complexed crystal structures of 2-enoyl-CoA hydratase 2 from Candida tropicalis multifunctional enzyme type 2 were solved to 1.95- and 2.35-A resolution, respectively. 2-Enoyl-CoA hydratase 2 is a dimeric, alpha+beta protein with a novel quaternary structure. The overall structure of the two-domain subunit of eukaryotic 2-enoyl-CoA hydratase 2 resembles the homodimeric, hot dog fold structures of prokaryotic (R)-specific 2-enoyl-CoA hydratase and beta-hydroxydecanoyl thiol ester dehydrase. Importantly, though, the eukaryotic hydratase 2 has a complete hot dog fold only in its C-domain, whereas the N-domain lacks a long central alpha-helix, thus creating space for bulkier substrates in the binding pocket and explaining the observed difference in substrate preference between eukaryotic and prokaryotic enzymes. Although the N- and C-domains have an identity of <10% at the amino acid level, they share a 50% identity at the nucleotide level and fold similarly. We suggest that a subunit of 2-enoyl-CoA hydratase 2 has evolved via a gene duplication with the concomitant loss of one catalytic site. The hydrogen bonding network of the active site of 2-enoyl-CoA hydratase 2 resembles the active site geometry of mitochondrial (S)-specific 2-enoyl-CoA hydratase 1, although in a mirror image fashion. This arrangement allows the reaction to occur by similar mechanism, supported by mutagenesis and mechanistic studies, although via reciprocal stereochemistry.

Multiple sequence alignment of eukaryotic hydratase 2 has revealed a conserved region showing a motif, [YF]-X 1, 2
The crystal structure of A. caviae (R)-hydratase has been solved recently (15). The core structure is a hot dog fold, which is built up of a long and hydrophobic ␣-helix ("sausage") packed against anti-parallel ␤-sheet ("bun"). In a functional dimer, two subunits associate side by side to form an extended 10-stranded anti-parallel ␤-sheet layer. A similar hot dog fold and subunit organization is also found in Escherichia coli ␤-hydroxydecanoyl thiol ester dehydrase (FabA; dehydrase) of fatty acid synthesis type II (16). The major difference between these two enzymes is the additional loop structure in (R)-hydratase, referred to as overhanging segment (15). Despite the FabA lacking the overhanging segment housing the conserved hydratase 2 motif, the proposed catalytic residues, Asp-84 and His-70Ј of the neighboring subunit, are arranged identically with the proposed catalytic residues, Asp-31 and His-36 (from the same subunit), of (R)-hydratase proposing an equal enzyme mechanism in addition/elimination of water molecule to/from 2-enoyl/ (3R)-hydroxyacyl thioesters. Incongruously, the catalytic dyad of human hydratase 2 is proposed to form by an aspartate (Asp-510) and a glutamate (Glu-366) residues (9). Interestingly, the suggested catalytic glutamate locates in the N-terminal half of eukaryotic protein, which is absent in the prokaryotic (R)-hydratase.
Even though the prokaryotic (R)-hydratase from A. caviae and eukaryotic hydratase 2 are related enzymes, they have major differences concerning the size of the enzymes, acyl chain substrate specificities, as well as biological functions in the cell.
In the present work we show for the first time the threedimensional structures for an eukaryotic hydratase 2 (from Candida tropicalis MFE-2) as a free (apoenzyme) and enzymeproduct complex (holoenzyme). The crystal structures locate the catalytic residues and suggest a novel catalytic mechanism for (R)-specific hydratases/dehydratases. Moreover, the structures introduce novel features for quaternary structure, evolution, substrate binding mode, and stereospecificity of hydratase 2 from the fatty acid ␤-oxidation pathway.

EXPERIMENTAL PROCEDURES
Primary Phasing and Structure Refinement of SeMet-labeled CtMfe2p(dh aϩb ⌬)-The recombinant and selenomethionine (SeMet)-labeled 2-enoyl-CoA hydratase 2 part (residues 628 -906) of C. tropicalis Mfe2p (CtMfe2p(dh aϩb ⌬)) was overexpressed, purified, and crystallized as described previously (17). Primary phases for a SeMet-crystal were determined by the multiwavelength anomalous dispersion method using data measured at three wavelengths (17). Data were processed using DENZO/SCALEPACK (18), and initial selenium sites were searched for using the program SOLVE (19). The positions of eight selenium atoms were confirmed and further refined using CNS (20), leading to two additional sites. The figure of merit at the beginning of the structure determination was 0.56, and the electron density calculated with the help of these 10 sites already revealed clear secondary structure elements. The initial maps were improved by solvent-flattening calculations using a water content of 56%, after which the maps were interpretable. The building of the initial model was performed manually using O (21). Phases were extended to 1.95 Å using a combined data set, having the 100% data to 2.25 Å and the following 86% complete high resolution data to 1.95 Å. Before refinement cycles, the data were reprocessed and scaled using XDS (22). Structural refinement was initiated with the simulated annealing protocol in CNS followed by refinement cycles with REFMAC (23). The non-crystallographic symmetry was not applied in the refinement. Water molecules were added by the solvent building mode of the program ARP/wARP (24). The final refinement statistics are shown in Table I.
Preparation of CtMfe2p(dh aϩb ⌬H813Q) Crystallization Sample, Cocrystallization with Its Ligand and Structure Determination-The QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used for making the H813Q mutations in CtMfe2p(dh aϩb ⌬). The oligonucleotide pairs used for mutagenesis were 5Ј-GACAGAAAC-CCATTGCAGATTGATCCAAACTTTGC-3Ј and 5Ј-GCAAAGTTTG-GATCAATCTGCAATGGGTTTCTGTC-3Ј (the underlined letters show the nucleotide triplets resulting in the amino acid substitution, His-813 to Gln). The CtMfe2p(dh aϩb ⌬H813Q) was purified like the native protein (17) and concentrated to 12 mg/ml, and the hydratase 2 activity of the mutated variant was measured with 60 M trans-2-decenoyl-CoA as described (25). with XDS, and the structure was solved with molecular replacement using CNS and the subunit D of SeMet CtMfe2p(dh aϩb ⌬) as the model. At first, only three monomers (subunits A, B, and C) were found in the asymmetric unit of the CtMfe2p(dh aϩb ⌬H813Q) crystal, and after the first refinement cycle with REFMAC the R work and R free values were 38.5% and 42.8%, respectively. The missing regions in subunits A and B were built using O, and the disordered regions 678 -697 and 729 -732 were excluded from the subunit C. The last subunit, D, was added manually using the A-B pair of the same crystal form, by superimposing the subunit A on C. The following refinement with four subunits and 130 water molecules reduced the R work to 24.7% (R free ϭ 29.2%). At this stage, the coordinates of octanoyl-CoA (27) were used to fit omit Fourier maps of subunits A, B, and D, and after refinement the R work dropped to 22.7% (R free ϭ 27.1%). The coordinates of the ligand were completed by manually adding the two missing carbon atoms to the -end of the fatty acyl tail as well as the hydroxyl group to the ␤-carbon. The final refinement statistics as well as the data processing statistics are presented in Table I.
The structures were validated with PROCHECK (28) and WHATIF (29). The secondary structures were analyzed by using PROMOTIF (30), and the structure drawings were made with Swiss-PdbViewer (31). The figures illustrating the holoenzyme structure were drawn by using the subunit A (or A-B dimer in Fig. 1B) of CtMfe2p(dh aϩb ⌬H813Q) with the bound ligand, whereas subunit C of SeMet CtMfe2p(dh aϩb ⌬) crystal structure was used for pictures presenting the active site of the apoenzyme.

RESULTS AND DISCUSSION
The Structure Determination of CtMfe2p(dh aϩb ⌬)-The complete multiwavelength anomalous dispersion data for the recombinant, SeMet-labeled CtMfe2p(dh aϩb ⌬) were used for initial phase determination of apoenzyme of hydratase 2. The data from another crystal form obtained from the co-crystallization experiment of CtMfe2p(dh aϩb ⌬H813Q) in the presence of natural ligand, trans-2-decenoyl-CoA, were phased by a molecular replacement solution using the subunit D of the apoenzyme as a starting model. In both crystal forms, the asymmetric unit consists of two functional dimers and a continuous polypeptide chain in each subunit could be built starting at Asp-630 or Pro-631, with some variation between the subunits ( Table I). The extreme C-terminal residues containing the peroxisomal targeting signal type 1 (32) are solvent-exposed and disordered in all subunits. The positive electron density in subunits A, B, and D of the CtMfe2p(dh aϩb ⌬H813Q) crystal defined acyl groups with a hydroxyl group in the (R)-position at the ␤-carbon (see Fig. 3C below) but not a planar double-bond at C ␣ -C ␤ , revealing that the actual ligand in the holoenzyme was the hydrated product of trans-2-decenoyl-CoA, the (3R)hydroxydecanoyl-CoA (3RHDC). No density for the ligand was observed in subunit C of the holoenzyme. The presence of the mutation, H813Q, was confirmed by the electron density maps (Fig. 3C). The more detailed structural statistics of both crystal forms are summarized in Table I.
The Overall Structure of CtMfe2p(dh aϩb ⌬)-The CtMfe2p-(dh aϩb ⌬) subunit is composed of 5 well defined ␣-helices (␣1 and ␣4 -␣7) and 11 ␤-strands (␤1-␤11), which form a compact molecule with dimensions of 35 ϫ 40 ϫ 45 Å (Figs. 1A and 2A). The subunit structure can be further divided into an N-terminal domain (or N-domain, residues 631-770), a C-terminal domain (or C-domain, residues 789 -900), and an intervening bridge (residues 771-788), which connects the two domains and includes a short ␣-helix, ␣4 ( Fig. 2A). The core structure of the C-domain of the CtMfe2p(dh aϩb ⌬) consists of a 16-residue ␣-helix, ␣7, and covering ␤-strands, ␤7-␤11, forming a typical hot dog fold first identified in FabA (16). Moreover, the C-domain of the CtMfe2p(dh aϩb ⌬) contains a solvent-exposed loop structure (residues 806 -830), included by an amphipathic ␣-helix, ␣6, and an ␣-helix, ␣5, which sandwiches the ␣7 with the ␤-sheet layer. The folding of the C-domain of CtMfe2p(dh aϩb ⌬) is strikingly similar with the subunit of the A. caviae (R)-hydratase (the root mean square deviation ϭ 1.54 for C ␣ atoms) as can be proposed by a 15% overall sequence identity. The solvent-exposed loop structure, mainly composed of residues belonging to the hydratase 2 motif and referred to as the overhanging segment in A. caviae (R)-hydratase, is a unique feature for (R)specific 2-enoyl-CoA hydratases and distinguishes the C-domain of eukaryotic hydratase 2 and (R)-hydratase from the crystal structure of FabA (16). However, the similar overall fold indicates that all these enzymes/domains, utilizing trans-2enoyl/(3R)-hydroxyacyl metabolites, most probably share a common ancestor.
Interestingly, the N-domain of CtMfe2p(dh aϩb ⌬) resembles significantly the C-domain, although no clear sequence similarity can be found between the domains. If the N-domain is rotated 180°around the vertical axis perpendicular to the ␤-sheet layer, the ␤-strands, ␤1-␤5, superimpose with the ␤-strands, ␤7-␤11, of the C-domain, as do ␣1 and ␣5. Moreover the N-domain contains a solvent-exposed loop structure (residues 647-668) resembling the overhanging segment of the C-domain; although it lacks the well defined ␣-helix, a PRO-MOTIF analysis (30) revealed ␣-helical structure (␣1/2) also in this loop (Fig. 1C). The major difference between the two domains is, however, that the core helix, the "sausage" of the hot dog fold is replaced in the N-domain by a region consisting of short stretches of ␣-helices, ␣2 and ␣3, connected by a random coil structure (Figs. 1A and 2A). Therefore the features of the hot dog fold are only partially fulfilled in the N-domain of CtMfe2p(dh aϩb ⌬).
The N-domain is paired with the C-domain via ␤-strands ␤2 and ␤8 such that an extended 11-stranded anti-parallel ␤-sheet layer is formed in a CtMfe2p(dh aϩb ⌬) subunit ( Fig. 2A). This arrangement of the two domains in CtMfe2p(dh aϩb ⌬) resembles strikingly the pairing of the two subunits of FabA and especially to that of A. caviae (R)-hydratase (Fig. 2). The origin of the N-domain of the CtMfe2p(dh aϩb ⌬), which replaces the neighboring subunit when compared with prokaryotic ho-mologs, is an intriguing question. Although the N-and Cdomains of the CtMfe2p(dh aϩb ⌬) share only Ͻ10% amino acid sequence identity, both domains show in addition to the similar folds a 50% identity at the level of DNA. This suggests that they have arisen via gene duplication rather than via gene fusion of two non-related genes with subsequent structural convergence.
The dimer of CtMfe2p(dh aϩb ⌬) is elongated, with dimensions of 70 ϫ 40 ϫ 45 Å, and the dimerization is accomplished by a four-helix bundle structure where the pairwise arranged ␣-helices, ␣1 from the N-domain and ␣5 form the C-domain, from one subunit are packed against their counterparts from the other one in an anti-parallel fashion (Fig. 1B). In addition to the major contacts via four-helix bundle, the N-domain solventexposed loop and the overhanging segment are participating in the dimeric interactions (Fig. 1B). One of the strengthening contacts formed between the subunits is the salt bridge Glu-659 to Arg-804Ј. Importantly, Glu-659 corresponds to the proposed catalytic Glu-366 in the human enzyme suggesting that the role of the N-domain glutamate is rather to stabilize the folding of the hydratase 2 motif and to strengthen the dimeric interactions than to participate directly in the catalytic reaction. Despite the resemblance of the subunit structure of CtMfe2p(dh aϩb ⌬) with the homodimeric bacterial counterparts, the dimerization of the two subunits of CtMfe2p(dh aϩb ⌬) makes the quaternary fold of the eukaryotic hydratase 2 a novel feature.
Substrate Binding Mode-The bound 3RHDC molecule locates between ␤-strands ␤2 and ␤8 of the extended ␤-sheet at the interface of the N-and the C-domains (Figs. 1A and 2A). The CoA molecule is in a bent conformation, a state often found in the CoA molecules bound to proteins (33). The tunnel formed in the domain interface engulfs part of 3RHDC but not the 3Ј-phosphate ADP and half of the pantetheine moiety. The adenine ring of the bound substrate lies in a pocket surrounded  by the side chains of Arg-855 and Phe-758. The adenine amino group is hydrogen-bonded (H-bonded) to the backbone oxygen of Phe-856, and the 3Ј-phosphate group of the ribose is solventexposed and has a stabilizing salt bridge to the side chain of Lys-729. Noteworthy, all these residues interacting with the CoA moiety are highly conserved among the eukaryotic hydratase 2 proteins and in PhaJ2 Pa (Fig. 1C).
The remaining ten-carbon acyl chain of the substrate is buried in the hydrophobic cleft formed by the N-terminal ␤-strands ␤2 and ␤5 and by the N-terminal loop structure (residues 678 -697). The loop contains hydrophobic residues (Phe-676, Phe-685, Phe-692, and Leu-697), which interact with the -end of the acyl group. In apoenzyme, this loop had either high temperature factors or poorly visible electron densities indicating high flexibility (therefore referred to as flexible loop I, Figs. 1C and 2A). In addition, the N-domain of CtMfe2p(dh aϩb ⌬) contains two other flexible loops (flexible loop II, residues 726 -736; flexible loop III, residues 756 -769; Figs. 1C and 2A), which were difficult to interpret in the apoenzyme but well defined in the holoenzyme. Upon ligand binding the flexible loops move toward the overhanging segment of the C-domain excluding the hydrophobic parts of the ligand out of the solvent and completing the CoA binding pocket.
The substrate binding mode of CtMfe2p(dh aϩb ⌬), where the ligand is bound to the domain interface, is similar to that found in FabA (16). However, the crystal structure of FabA complexed with an inhibitor molecule, 3-decynoyl-N-acetylcysteamine, shows two equal binding sites in the subunit interface in contrast to only one found in CtMfe2p(dh aϩb ⌬). Docking experiments based on the FabA-inhibitor crystal structure also prompted the suggestion of two binding sites in the subunit interface of the A. caviae (R)-hydratase (15). The available space in the suggested substrate binding tunnel of (R)-hydratase is restricted partly by the rigid hot dog helix (␣4) of an adjacent subunit (Fig. 2B), and the depth of the pocket allows only the entrance of fatty enoyl-CoAs up to C 6 in length. In The 3RHDC molecule is shown as sticks and is colored as follows: carbon, gray; nitrogen, blue; oxygen, red; sulfur, yellow; phosphorus, green. B, the ribbon diagram showing an overview of the quaternary structure of the 3RHDC-complexed CtMfe2p(dh aϩb ⌬) dimer. The view of the subunit A (on the top) is obtained by a counter clockwise rotation of that in panel A by 90°around the vertical axis. The ␣-helices participating in dimerization, as well as the hot dog core helices, are labeled. The 3RHDC molecules are shown as black sticks. C, alignment of the amino acid sequence of 2-enoyl-CoA hydratase 2 part of C. tropicalis Mfe2p (Ct_H2) with the corresponding parts of Saccharomyces cerevisiae (Sc_H2), human (Hs_H2), and rat (Rn_H2) multifunctional enzyme type 2s, as well as with monofunctional PhaJ2 from P. aeruginosa (Pa_H2) and A. caviae (R)-hydratase (Ac_H2). The corresponding SwissProt identifiers are: P22414, Q02207, P51659, P97852, Q9LBK1, and O32472. The secondary structure elements of CtMfe2p(dh aϩb ⌬), ␣-helices (red lines) and ␤-strands (green arrows), are marked above the sequence alignment. Black vertical arrows indicate some functionally important residues in CtMfe2p(dh aϩb ⌬). The black line below the sequence alignment indicates the hydratase 2 motif, and the yellow lines above the sequence alignment show the regions of flexible loops I, II, and III. The Glu-627 (shown by the asterisk) of full-length C. tropicalis Mfe2p corresponds to the initial methionine of CtMfe2p(dh aϩb ⌬) construct.
contrast in the N-domain of CtMfe2p(dh aϩb ⌬) the region corresponding to the rigid hot dog helix is replaced by the short ␣-helix, ␣2, and the following highly mobile polypeptide chain (beginning of the flexible loop I, Fig. 2A). This structural feature in addition to the existence of the two other flexible loops in the N-domain of CtMfe2p(dh aϩb ⌬) explains the ability of the fungal hydratase 2 to utilize also long-chain enoyl-CoAs (C 10 -C 22 ) as substrates. Recently, kinetic and mutagenesis studies showed that, by mutating one of the two major residues blocking the substrate-binding pocket, Leu-65 (located in the hot dog ␣-helix) or Val-130, to smaller residues (Ala and Gly, respectively), the substrate acceptance of A. caviae (R)-hydratase can be raised to up to C 10 in length (34).
Reaction Mechanism-The role of the conserved aspartate in A black arrow shows the catalytic water molecule, which is H-bonded to the carboxylate oxygen of Asp-808, the N⑀ 2 atom of His-813, and the N⑀ 2 atom of Asn-810. The carbons, nitrogens, oxygens, and selenium have been colored in gray, blue, red, and brown, respectively, and the water molecules are colored in magenta. B, stereo view of the active site of CtMfe-2p(dh aϩb ⌬H813Q) complexed with 3R-HDC. Only the acyl part and ␤-mercaptoethyleneamine moiety of 3RHDC are shown. Water molecules W6, W7, and W8 give way to the substrate as it enters the binding site, whereas the first five waters remain in position. The black arrow shows the (3R)-hydroxyl group, which is added to the ␤-carbon of the acyl moiety, and the red arrow indicates the H-bond crucial for oxyanion hole formation. The (3R)-hydroxyl of the 3RHDC molecule is H-bonded to the carboxylate oxygen of Asp-808 as well as to the N⑀ 2 atoms of Gln-813 and Asn-810. The point mutation, H813Q, does not affect the folding of the active site. The atom colors correspond to the coloring in panel A, in addition to two sulfur atoms, which are shown in yellow. c, the final 2F o Ϫ F c map around 3RHDC molecule at 2.35-Å resolution (contoured at 1.0 ). The map is drawn also around the catalytic residue, Asp-808, as well as Gln813, which replaces the catalytically active His-813 in the mutant protein. A black arrow points to the electron density for the (3R)-hydroxyl group. the hydratase 2 motif (510 and 31, in human MFE-2 and A. caviae (R)-hydratase, respectively) has been shown to be crucial for hydratase 2 activity (9, 15). On the contrary, the present structure of the fungal enzyme challenges the potential catalytic role of the N-domain glutamate for eukaryotic hydratase 2 as discussed above. Instead, the location of the side chain of His-813 with respect of Asp-808 is optimal for catalysis in the apoenzyme of CtMfe2p(dh aϩb ⌬). Furthermore, the mutagenesis studies showed that the mutation H813Q reduced the enzyme activity of CtMfe2p(dh aϩb ⌬) around 15,000 times (k cat value was 0.026 s Ϫ1 with 60 M trans-2-decenoyl-CoA) compared with the wild type protein (17) without affecting the folding of the enzyme. These facts indicate that His-813 may have an important role in catalysis analogous to the corresponding histidine, His-36, in A. caviae (R)-hydratase (15).
The presence of a natural product in the crystal structure of CtMfe2p(dh aϩb ⌬H813Q) indicates the active site as well as the catalytic residues for hydratase 2. In the holoenzyme the (3R)hydroxyl group of the bound 3RHDC is H-bonded to O␦ 2 of Asp-808, N⑀ 2 of Gln-813 (introduced by site-directed mutagenesis), and N⑀ 2 of Asn-810 (Fig. 3B). After superimposing the active sites of the holoenzyme and the apoenzyme, the water molecule W6 in the apoenzyme occupies the position of the (3R)-hydroxyl group of the ligand (see Fig. 5B). Moreover, the W6 is H-bonded to O␦ 2 of Asp-808 and N⑀ 2 of His-813 and can be proposed to be the catalytic water activated by Asp-808 and His-813 in hydration reaction (Fig. 3A). The role of His-813 in water activation is enhanced by N␦ 1 of the imidazole ring donating proton via H-bond to the backbone oxygen of Ile-828 (Fig. 3A), thus creating a basic lone pair of electrons on N⑀ 2 . In the optimal configuration for the hydration reaction, the electron pair of the fourth tetrahedral position of the catalytic water, W6, would be directed toward the ␤-carbon of the substrate. Because W6 is also H-bonded to the N⑀ 2 atom of Asn-810 (Fig. 3A), this requirement is fulfilled only if the carboxylate oxygen of Asp-808 and N⑀ 2 of His-813 are both unprotonated. This being the case, activated W6 is the source of both the proton added to the ␣-carbon and the (3R)-hydroxyl group added to the ␤-carbon, as suggested for hydratase 1 reaction (35). Previously, the reaction mechanism of (R)-specific hydration/dehydration is proposed to occur via acid-base catalysis (15,16), which is now challenged by the current observations with the crystal structures of apoenzyme and enzyme-product complex of eukaryotic hydratase 2. The proposed reaction Only the catalytic residues (and Asn-810) and the components comprising the oxyanion hole are shown. The catalytic His-813 as well as the catalytic water, W6, are derived from the apo form of CtMfe-2p(dh aϩb ⌬) structure after superimposing with the holoenzyme. The side chain of Gln-813 (from site-directed mutagenesis) is colored green. The H-bonds formed between the (3R)-hydroxyl group and the protein in the complex structure are not shown. mechanism of hydratase 2 is presented in Fig. 4.
The carbonyl oxygen of the substrate is H-bonded to the backbone amide of Gly-831 and water molecule, W1 (Figs. 3B and 5B), which together form an oxyanion hole, found upon substrate binding in many CoA-ester metabolizing enzymes (36). The role of the oxyanion hole is to stabilize the kinetically very unfavorable intermediate state of a thiol ester substrate, enabling catalysis of the hydration reaction to occur (Fig. 4). The polarizing effect of Gly-831 is enhanced due to the position of this residue at the N terminus of the long hot dog ␣-helix (␣7) giving Gly-831 a positive dipole. These findings underline the importance of the long central helix of the hot dog fold. Because the core helix is replaced by the discontinuous helical region in the N-domain of CtMfe2p(dh aϩb ⌬) the loss of the other catalytic site in the subunit of eukaryotic hydratase 2 can be explained.
Interestingly, the hydrogen bonding network of the active site of CtMfe2p(dh aϩb ⌬) resembles the active site geometry of hydratase 1 (35), although in a mirror image fashion (Fig. 5). This arrangement allows the reaction to occur by a similar mechanism, although following an opposite stereochemistry. Both enzymes, hydratase 1 and hydratase 2, use an oxyanion hole for orienting the ligand and stabilizing the reaction intermediate in catalysis. The catalytic residues in each of the enzymes locate on the same side of the substrate (synchemistry) (37) but are positioned oppositely when compared with each other. As visualized in Fig. 5, the stereochemistry of the added hydroxyl group is determined by the position of the catalytic residues with respect to the ␤-carbon of the substrate. CONCLUSION 2-Enoyl-CoA hydratase 2 is the key enzyme of peroxisomal ␤-oxidation of various fatty acids and fatty acyl derivatives in mammals and straight-chain fatty acids in fungi. It is structurally unrelated to hydratase 1, which catalyzes corresponding reaction via opposite stereochemistry in peroxisomal MFE-1 and extraperoxisomal ␤-oxidation systems. In contrast to hydratase 1, hydratase 2 has the hot dog fold, first identified in E. coli ␤-hydroxydecanoyl thiol ester dehydrase and more recently also in (R)-specific 2-enoyl-CoA hydatase 2 from A. caviae PHA synthesis pathway. An intriguing feature of eukaryotic hydratase 2 is the two-domain subunit structure, which is strikingly similar with the overall structure of the homodimeric bacterial homologs. However, the N-terminal domain of eukaryotic hydratase 2, which is supposed to have arisen via gene duplication, has diverged structurally during evolution so that eukaryotic hydratase 2 has adapted the ability to use also bulky fatty enoyl-CoAs as substrates, although with a cost of one active site. The present work also strengthens the current understanding on the reaction mechanism of (R)-specific hydratases/dehydratases. Also it shows that, despite the fact that hydratase 1 and 2 lack both amino acid sequence and overall structural similarity, the organization of the active site allows the hydration to occur, although with reciprocal stereochemistry, via same mechanistic principles indicating the occurrence of functional convergence during evolution.