Crystal Structures of Intermediates in the Dehalogenation of Haloalkanoates by L-2-Haloacid Dehalogenase

The L -2-haloacid dehalogenase from the 1,2-dichloro- ethane-degrading bacterium Xanthobacter autotrophicus GJ10 catalyzes the hydrolytic dehalogenation of small L -2-haloalkanoates to their corresponding D -2-hy- droxyalkanoates, with inversion of the configuration at the C 2 atom. The structure of the apoenzyme at pH 8 was refined at 1.5-Å resolution. By lowering the pH, the catalytic activity of the enzyme was considerably reduced, allowing the crystal structure determination of the complexes with L -2-monochloropropionate and monochloro- acetate at 1.7 and 2.1 Å resolution, respectively. Both complexes showed unambiguous electron density extending from the nucleophile Asp 8 to the C 2 atom of the dechlorinated substrates corresponding to a covalent enzyme-ester reaction intermediate. The halide ion that is cleaved off is found in line with the Asp 8 O d 1–C 2 bond in a halide-stabilizing cradle made up of Arg 39 , Asn 115 , and Phe 175 . In both complexes, the Asp 8 O d 2 carbonyl oxygen atom interacts with Thr 12 , Ser 171 , and Asn 173 , which possibly constitute the oxyanion hole in the hy- drolysis of the ester bond. The carboxyl moiety of the substrate is held in position by interactions with Ser 114 , Lys 147 , and main chain NH groups. The L -2-monochloro-propionate CH 3 group is located in a small pocket formed by side chain

C 2 atom. Several homologous L-DEXs have been found in various Pseudomonas species and in Xanthobacter autotrophicus GJ10, a bacterium that is able to degrade the xenobiotic compound 1,2-dichloroethane (1,2). This halogenated hydrocarbon is industrially produced in large quantities and is applied as a solvent and as an intermediate in the production of plastics (3). Because microorganisms that contain dehalogenases can be used in a biotechnological approach to detoxify halogenated aliphatics (4), such enzymes are a fascinating target for research. In addition, the stereospecificity of L-DEXs could make them useful for the biosynthesis of chiral 2-hydroxyalkanoic acids. Furthermore, L-2-haloacid dehalogenase is the prototypical member of a large superfamily of hydrolases, the haloacid dehalogenase (HAD) superfamily identified by Koonin and coworkers (5,6). Based on three conserved sequence motifs, the L-DEXs, epoxide hydrolases, P-type ATPases, and a variety of phosphatases are recognized as members of this superfamily. Detailed information on L-DEXs is of interest as the enzyme is the only member of the HAD superfamily that has been structurally characterized so far.
The x-ray structures of two L-2-haloacid dehalogenases have been reported, L-DEX YL from Pseudomonas sp. YL (Protein Data Bank code 1JUD (7)) and DhlB from X. autotrophicus GJ10 (Protein Data Bank code 1AQ6 (8)). The enzymes share a sequence identity of 40%, and their structures are closely related. Both enzymes have a mixed ␣/␤ core domain in a Rossmann fold with a four-helix bundle subdomain insertion. DhlB is somewhat larger, and the 21 extra residues form a two-helix excursion from the ␣/␤ core domain on the same side as the four-helix bundle. Together these helical domains provide a tight dimer interface and limit the substrate specificity of the X. autotrophicus enzyme to short substrates such as haloacetates and halopropionates (8,9).
Comprehensive biochemical data have been obtained for the Pseudomonas enzyme (1, 10, 11). 2 Asp 8 was identified as the nucleophile in the first step of the enzymatic reaction, the formation of a covalent enzyme-ester intermediate. Furthermore, these studies revealed eight more charged and polar amino acids (Thr 12 , Arg 39 , Ser 114 , Lys 147 , Tyr 153 , Ser 171 , Asn 173 , and Asp 176 ) that are involved in substrate binding and catalysis. Most of the catalytically critical residues are conserved in the HAD superfamily and they stand out from the main domain. The only exception is Arg 39 , which is provided by the four-helix bundle domain. The x-ray structures enabled a detailed discussion of the role of these residues, and in particular useful information could be extracted from a model of a bound L-2-monochloropropionate (MCPA) substrate, which was based on the position of a formate ion in the active site of DhlB (8). The conserved serine residue in motif II, Ser 114 , was pro-* This work was supported by the Netherlands Foundation for Chemical Research (SON) with financial aid from the Netherlands Organization for Scientific Research (NWO). 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 posed to bind the carboxylate moiety of the substrate, and a halide-binding cradle formed by Arg 39 , Tyr 10 , and Phe 175 was postulated. Conserved residues from motifs I and III, Thr 12 , Ser 171 , and Asn 173 , were found to interact with the nucleophile Asp 8 O␦2 atom possibly making up an oxyanion hole, and the motif III Lys 147 N was hydrogen-bonded to the O␦1 atom of the nucleophile.
These proposals were recently corroborated by crystal structures of an inactive S171A L-DEX YL mutant enzyme covalently bound to several chloroalkanoate substrates (12,13). However, structural evidence for the location of the halide ion is still lacking. To gain more insight into the abstraction of the halide ion and the reaction of the enzyme with MCPA, one of its best substrates (9), we extended our crystallographic studies on DhlB. One soaking experiment with MCPA was done at pH 8, at which the enzyme is active, to obtain information about the enzyme structure during or after processing of the substrate. Two other experiments were performed at low pH to trap the reaction intermediates of MCAA and L-MCPA in a similar way as was done previously for haloalkane dehalogenase (14). Here we present the 1.52 Å resolution structure of the unliganded enzyme and the structures of covalent ester intermediates of the native enzyme with L-MCPA and MCAA, including a bound chloride ion. The structures experimentally confirm the substrate binding model proposed earlier (8) and contribute to the detailed understanding of the reaction mechanism of the enzyme.

MATERIALS AND METHODS
Crystal Preparation-Crystals of L-2-haloacid dehalogenase were grown in the presence of sodium formate by macroseeding as described previously (15). The enzyme crystallizes in two orthorhombic space groups, C222 1 and P2 1 2 1 2 1 . For the soaking experiment the better diffracting primitive crystal form was chosen (a ϭ 56.75 Å, b ϭ 83.83 Å, c ϭ 90.81 Å, 2 mol/asymmetric unit). As formate binds in the active site of the enzyme, crystals were washed two times in fresh synthetic mother liquor (25% (w/v) polyethylene glycol 8000, 100 mM bis-Tris, pH 7.0) in an attempt to remove the formate.
One crystal was transferred to a solution of slightly higher pH, containing 25% (w/v) polyethylene glycol 8000, 100 mM bis-Tris, pH 8.0, equilibrated for 20 min at room temperature, and finally soaked for 3 h in synthetic mother liquor, pH 8.0, containing 20 mM D,L-2-monochloropropionic acid.
A second crystal was transferred to a more acidic solution of 25% (w/v) polyethylene glycol 8000, 100 mM citrate, pH 5.0, at 4°C, equilibrated for 20 min, and soaked for 45 min at 4°C in synthetic mother liquor containing 20 mM D,L-MCPA. A third, analogous experiment was done at pH 5.0 using 10 mM monochloroacetic acid (MCAA) instead of MCPA.
Refinement of the pH 8 Structure-For the MCPA pH 8 experiment, the native DhlB structure (Protein Data Bank entry 1AQ6 (8)) without the formate ions and water molecules was positioned in the cell using AMoRe (17). Initial A -weighted OMIT 2F o Ϫ F c and F o Ϫ F c electron density maps (18 -20) clearly showed density for a formate ion in the active site of both molecules, indicating that the substrate had not bound. Nevertheless, the model was refined because of the improved resolution compared with the 1AQ6 structure. The set of reflections set apart to calculate R free values (21) in the refinement of the 1.95 Å native structure was extended to cover 5% of the additional data as well. After an initial round of rigid body refinement, the model was subjected to one refinement cycle of torsion-angle dynamics (22) and subsequent cycles of positional and B-factor refinement for all atoms and refine-ment of the occupancy of residues with two conformations, all performed with CNS 0.5 (23). Geometry checks pointed out a considerable number of N-C␣-C bond angles that deviated from their ideal value (24). Hence, the CNS topology and parameter files were adjusted such that the restraint on the main chain dihedral angle was relaxed to 20% of its original value without changing the restraints on the other X-N-C-X dihedral angles (X ϭ any atom). This was done to allow for the large variability of the angle observed in atomic resolution structures (25,26). After the adjustment, no deviating bond angles were flagged anymore. However, the angle of the nucleophile Asp 8 became 163°, which is far off the normal value for a trans-peptide bond.
At all stages A -weighted 2F o Ϫ F c electron density maps were calculated and inspected with O (27) to check the agreement of the model with the data. PROCHECK (28) and WHATCHECK (29) were used to assess the stereochemical quality. Whenever necessary the model was manually adjusted in O. When the refinement gave no further decrease in R free nor any improvement in stereochemistry, it was considered completed. A final round of refinement including all data resulted in an overall crystallographic R-factor of 19.8% for data ranging from 20.0 to 1.52 Å resolution. Refinement and model statistics are given in Table I, and the quality of the electron density can be judged from Fig. 1A.
The asymmetric unit contains two protein molecules of residues 1-245 forming the DhlB dimer, 571 water molecules, and two formate ions. In molecule A, Arg 25 , Glu 79 , Thr 93 , Lys 109 , Gln 122 , Ser 172 , Ser 186 , Met 218 , Thr 222 , and Glu 225 have two conformations, and in molecule B, a second side chain position is found for Asp 21 , Glu 79 , Thr 93 , Ser 152 , Ser 172 , Met 218 , and Thr 222 . In both molecules, the loop formed by residues 204 -207 and the C termini have been remodeled, and in molecule B, the cis-peptide bond between Ala 106 and Pro 107 has been replaced by a trans-peptide bond. Furthermore, in both molecules the helix and loop formed by residues 122-132 have shifted by 0.6 Å on average, although no structural reason is apparent.
Refinement of the pH 5 Structure with MCPA Bound-For the MCPA pH 5 experiment, the 1.52 Å pH 8 structure described above was used as a starting model. Initial A -weighted 2F o Ϫ F c and F o Ϫ F c electron density maps clearly showed density extending from Asp 8 O␦1 for all MCPA atoms except for the chlorine, indicating that the substrate had bound covalently and that the chlorine atom had been cleaved off. The same refinement strategy as described above was employed. Unambiguous difference electron density enabled us to build in the covalently bound substrate (Fig. 1B). Also a chloride ion was identified in the active site on the basis of residual positive difference density around a "water molecule" with a B-factor of 2.0 Å 2 and the chemical environment. In addition, a diffraction experiment with a 2-monoiodoacetic acid substrate gave a peak of more than double intensity in the electron density map at this position (data not shown). Topology and parameter files for the covalently modified residue were generated with XPLO2D (30). At the end of the refinement, a final round including all data resulted in an overall crystallographic R-factor of 17.0% for data ranging from 20.0 to 1.70 Å resolution. Refinement and model statistics are given in Table I.
The DhlB structures with and without bound MCPA are very alike with an r.m.s. deviation of 0.24 Å for all C␣ atoms of the dimer, so it can be concluded that the overall structure of the enzyme is not influenced by the binding of its substrate. The disorder observed with the pH 8 experiment in the loop that contains residues 25-27 (see "Results" and "Discussion") could be resolved in the B molecule in the MCPA experiment as two different conformations of the main chain in this region. In molecule A, Arg 25 , Thr 93 , Lys 109 , Ser 152 , Ser 172 , Ser 186 , Met 218 , Thr 222 , and Glu 225 have two conformations, and in molecule B, a second side chain position is found for Lys 41 , Arg 54 , Glu 79 , Thr 93 , Gln 100 , Lys 109 , Ser 152 , Ser 172 , Ser 186 , Met 218 , and Thr 222 . The region from residues Ala 208 to Phe 213 is disordered as was observed in the native enzyme, pH 8 (see "Results" and "Discussion"). In the refinement of the MCAA model these characteristics were preserved.
Refinement of the pH 5 Structure with MCAA Bound-For the MCAA pH 5 experiment, the pH 5 DhlB structure with the covalent MCPA intermediate bound was used as a starting model. An initial A -weighted 2F o Ϫ F c electron density map clearly showed density extending from Asp 8 O␦1, and in the F o Ϫ F c map negative electron density was observed at the C 3 position of the MCPA, indicating that the smaller MCAA substrate had bound. The chloride ion was also found in an analogous position. A similar refinement strategy as described above for the MCPA structure was employed. Residues with alternative conformations were preserved and their occupancies were not refined. Refinement was completed with a final round including all data, resulting in an overall crystallographic R-factor of 17.1% for data ranging from 20.0 to 2.1 Å resolution. Refinement and model statistics are given in Table I. The enzyme structures complexed with MCPA and MCAA have an r.m.s. deviation of only 0.13 Å for all 490 C␣ atoms, and consequently the results discussed below refer to both molecules unless stated otherwise.
Analysis of the Structures-The structures were analyzed using programs from the CCP4 suite (31), LSQMAN (32), VOIDOO (33), and the BIOMOL package (Protein Crystallography Group, University of Groningen). The atomic coordinates and structure factor amplitudes have been deposited in the Protein Data Bank with the entry codes 1qq5, 1qq7, and 1qq6 for MCPA (pH 8), MCPA (pH 5), and MCAA (pH 5) experiments, respectively.

RESULTS AND DISCUSSION
Native Structure at pH 8 and 1.52 Å Resolution-The structure of dimeric L-2-haloacid dehalogenase at pH 8 is very similar to the previously reported 1.95 Å resolution structure (8), with an r.m.s. difference of 0.47 Å for all 490 C␣ atoms. This difference can be attributed to a few regions in the protein where the molecule has been modeled differently (see "Materials and Methods"). The disorder observed before in two loops and the C terminus (residues 25-27, 204 -206, and 243-245) is present in this structure as well. The high resolution of the data enabled the identification of alternate conformations for 17 side chains. Moreover, in both molecules residues 208 -213 were modeled in two different ways to account for ambiguities in the electron density. This region is part of the two-helix excursion (residues 193-219) that contributes significantly to the dimerization interface. Residues 208 -213 in molecule A interact with the same residues in molecule B (Fig. 2). It is surprising to see that equally good dimerization contacts can be made by two different conformations.
The A and B molecules of the dimer can be superimposed with an r.m.s. difference of 0.29 Å for 245 C␣ atoms. The major differences are found around the rebuilt residues (Pro 107 , the loop from residues 204 -207, and the C terminus). None of these differences are considered functionally relevant, and therefore all results discussed below pertain to both molecules, unless stated otherwise. The active site of the enzyme at pH 8 contains a formate ion that originates from the crystallization solution. This ion was observed in the 1.95 Å resolution structure as well, and its position was used to construct a model for the binding of an L-MCPA substrate (8). To replace the formate ion by MCPA, the crystal was washed several times. Apparently, the washing procedure was not sufficiently adequate to remove the ion.
Active Site Structure of the Enzyme-Ester Intermediates-In the experiments with MCPA and MCAA at pH 5, a substrate is covalently bound in the active site of DhlB. In these structures, it has replaced the formate ion and the water molecules that were present in the native structure. Very clearly, continuous electron density extends from the Asp 8 O␦1 atom to the C 2 atom of the substrates, whereas no density for a covalently attached chlorine atom is observed (Fig. 1, B and C). This indicates that a covalent bond has formed between the nucleophilic Asp 8 residue and the substrate and that the covalent Cl-C bond in the substrate has been cleaved. The Asp 8 O␦2 atom has a carbonyl functionality in the enzyme-ester intermediate structure. It is hydrogen-bonded to the hydroxyl group of the Thr 12 side chain and the side chain amide group of Asn 173 (Fig. 3A). Furthermore, in the MCPA covalent intermediate Ser 171 O␥ is located within a 3.3-Å distance, but it is not in the plane of the carbonyl oxygen atom lone electron pairs. In the MCAA bound structure, however, the derivatized Asp 8 side chain is rotated Ϫ30°about the C␤-C␥ bond, bringing Ser 171 O␥ much closer to the O␦2 atom and at the same time increasing the distance to Thr 12 O␥1 (Fig. 3B). This suggests a rotational freedom in the enzyme-ester intermediate, which might be used to optimize the interactions of the O␦2 atom with the enzyme in the hydrolysis step of the reaction, when a negative charge develops on the Asp 8 O␦2 atom. The negative charge of this oxyanion intermediate might be stabilized by Thr 12 , Ser 171 , and Asn 173 , which together would form an oxyanion hole with a tetrahedral coordination of the oxyanion.
A chloride ion is present close to the substrate C 2 atom, Asn 115 , and Arg 39 . It is found in line with the Asp 8 O␦1-C 2 bond, at a distance of 3.6 Å from the C 2 atom of the substrate, and the ion is close to both N atoms of Arg 39 and to the side chain amide group of Asn 115 . Both amino acid residues were shown to be catalytically essential in the L-DEX YL enzyme (1), but the chloride ion was not found in the covalent intermediate structures of this enzyme (12). Furthermore, the chloride is located in the plane of the aromatic ring of Phe 175 at a closest distance of 4.7 and 4.2 Å in the MCPA and MCAA experiments, respectively. The difference is because of the interaction of the phenyl ring with the methyl group of the MCPA substrate, which is absent in MCAA. Aromatic ring systems are known to be partially positively charged in the plane of the ring (34) and binding interactions of tryptophan and tyrosine rings with halide ions have been observed before in haloalkane dehalogenase (DhlA) (35,36). In 4-chlorobenzoyl-CoA dehalogenase the active site is surrounded by aromatic residues as well (37). Other aromatic residues near the DhlB active site, Tyr 10  functions to bind the halogen moiety of the substrate and to counterbalance the negative charge that develops on the halide during cleavage of the C-Cl bond.
The carboxylate moiety of the substrate is bound in the same position as the formate ion in the native DhlB structure. It is held in position by electrostatic interactions of its O 1 atom with the main chain NH groups of residues 9 and 10 and with the side chain of Ser 114 (Fig. 3). The other carboxylate oxygen atom is hydrogen bonded to the main chain NH group of Asn 115 and the Lys 147 ⑀-amino group. The latter interaction is facilitated by a reorientation of the lysine side chain. In the native structure, the lysine is hydrogen-bonded to Asp 176 O␦2, the main chain carbonyl oxygen atom of residue 113, and Asp 8 O␦1. The geometry of the contact with the formate ion is unfavorable for hydrogen bond formation. Because of the reorientation of Lys 147 in the MCPA-and MCAA-bound structures, its interaction with Asp 8 O␦1 has weakened. This is in accordance with the change of the functionality of the oxygen atom to an ether oxygen, which is generally a weaker hydrogen bond acceptor (38). All side chains that contribute to the binding of the carboxylate moiety were shown to be catalytically critical in L-DEX YL (1).
The structure of the complex with MCPA, the only chiral substrate to be efficiently degraded by DhlB (9), shows clear electron density for the C 3 atom of the substrate (Fig. 1B). The pocket in which the CH 3 group is located is not particularly hydrophobic as it is lined by side chain atoms of Lys 147 , Asn 173 , Phe 175 , and Asp 176 . It is shielded from the solvent by residues from helices ␣2 and ␣10, thereby limiting its size to about 75 Å 3 . This explains why the substrate specificity of DhlB is restricted to small haloalkanoates (9). The position of the small pocket also resolves the preference of the enzyme for L-substrates; as for any substrate with the methyl group and the hydrogen atom interchanged at the C 2 position, steric clashes with main chain atoms of residues 10 and 11 would occur (Fig. 4A).
The overall structures of native DhlB and the reaction intermediates are very alike (see "Materials and Methods"), demonstrating that the native enzyme is in an active conformation. All DhlB structures represent a form of the enzyme in which the active site is very compact and shielded from the solvent. This is in contrast with L-DEX YL, in which a significant movement of the Asp 8 -Ser 18 , Tyr 89 -Asp 100 , and Leu 113 -Arg 131 regions toward the active site was observed in the structures of the covalent substrate complexes of L-DEX YL compared with the wild-type enzyme, which is more open (12). The compact form of DhlB could be caused by the extra two-helix subdomain that is not present in L-DEX YL and that in part closes off the entrance to the active site (8).
Li et al. (12,13) have succeeded in trapping a covalent intermediate using L-2-monochloro-n-butyrate and the S171A mutant of L-DEX YL, which has an impaired oxyanion hole. Comparison of the DhlB and L-DEX YL covalent intermediate structures shows that the alkyl tail of the butyrate is close to the position of the chloride ion in DhlB (Fig. 4, A and B). In view of the 40% amino acid sequence identity between the two enzymes, one would expect that the catalytic residues and the   (Fig. 4C). This might be because of the L-DEX YL S171A mutation, which caused the loss of the attractive interaction between Asp 8 O␦2 and Ser 171 O␥, present in the wild-type enzyme. Further research is required to establish whether the binding modes of the covalently bound complexes of L-DEX YL with MCAA and n-butyrate are of functional relevance. Implications for the Reaction Mechanism-The dehalogenation reaction catalyzed by DhlB starts with the import of the negatively charged substrate via the cleft between the main domain and the four-helix bundle subdomain. The substrate can be either "pulled in" by an overall electrostatic dipole, which is directed along the dimer axis that is parallel to the import route (8), or via a guidance mechanism in which Arg 39 plays a major role (12). The substrate is bound through specific binding interactions of its functional groups with conserved active site residues. The structures presented here demonstrate that the reaction proceeds through a nucleophilic attack of the Asp 8 O␦1 atom on the C 2 of the substrate, resulting in the formation of a covalent enzyme-ester intermediate (Fig. 5). The salt bridge to the positively charged Lys 147 side chain reduces the pK a of the aspartate, thereby increasing its nucleophilicity. The formation of the O␦1-C 2 bond is accompanied by the cleavage of the C 2 -Cl bond. These three atoms are found in line, with the chloride ion located in a halide-stabilizing cradle formed by the side chains of Arg 39 , Asn 115 , and Phe 175 . As a result, the configuration of the other three substituents of the C 2 atom is inverted, which is in agreement with the observed inversion of configuration at the chiral center of the substrate (9).
In the next step of the reaction, the ester bond is hydrolyzed by a nucleophilic attack of a water molecule (or hydroxyl ion) on the C␥ atom of Asp 8 (10). The negative charge, which develops on the O␦2 carbonyl oxygen atom, is stabilized by an oxyanion hole formed by side chain atoms from Thr 12 , Asn 173 , and Ser 171 . The nature of the hydrolytic water molecule and its activation is still subject to speculation, as there is no such molecule present within 6 Å from the Asp 8 C␥ atom in the DhlB covalent intermediates. A movement of a part of the enzyme would be required to allow a water molecule to enter the active site. In the S171A L-DEX YL mutant one water molecule (Wn) was found near the ester bond (12), but it occupies a position near where the Ser 171 O␥ atom would be in the wild-type enzyme. However, the Asp 8 C␥ atom is accessible from this side only, as main chain atoms from residues 9 -12 prohibit the approach of a water molecule from the other side (Fig. 4A). Lys 147 and Asp 176 are possible candidates to activate a water molecule that attacks the Asp 8 C␥, as they are located close to the ester bond to be cleaved. The pK a of a lysine side chain (ϳ10) (39) is around the relatively high pH optimum of 9.5 for DhlB (9), but the different orientations of Lys 147 in the various structures point at a flexibility of this side chain that could hamper directing a nucleophilic water molecule. Moreover, the two nearby negatively charged carboxylate moieties of the substrate and Asp 176 might increase the pK a of the residue.
A comparison of the esterified nucleophile of DhlB and its immediate environment with the covalent intermediate of DhlA (14) shows that the N⑀2 atom of the His 289 , the residue which activates the hydrolytic water molecule, takes up a position in between Lys 147 N and Asp 176 O␦2, although closer to the latter (Fig. 6). Furthermore, the position of the DhlA nucleophilic water molecule is then close to Asp 176 O␦1 and Ser 171 O␥ and very near Wn in the L-2-monochloro-n-butyrate-L-DEX YL complex. This supports the suggestion from Li et al. (12) that Asp 176 and Ser 171 are essential for hydrolysis of the ester. Above pH 9, the enzyme could employ the flexibility of Lys 147 to vacate the space needed for the water, driven by a pH effect. We have not been able to obtain stable crystals at pH 9 or higher, which is indicative of structural changes around this pH. Also the dimer interface might be susceptible to change around the pH optimum, as Lys 41 and Tyr 68 of both molecules make hydrogen bonds to the main chain carbonyl groups of Leu 216 and Ala 215 , respectively, and the donor groups are only 4 Å apart. Lys 147 could also play a role in the deprotonation of Asp 176 after hydrolysis, although Tyr 153 , the other residue hydrogen bonded to Asp 176 , is another candidate for this function.
Many of the residues in the active site are conserved among the members of the HAD superfamily (5,6). The model, which was constructed for the structure and mechanism of the phosphatase and P-type ATPase members of the superfamily (40), is fully corroborated by the DhlB enzyme-ester intermediate structures. These enzymes cleave covalent bonds of phosphorylated substrates by nucleophilic attack of the motif I Asp (Asp 8 ) on the phosphorus of the substrate, resulting in the formation of a phosphoenzyme intermediate (41,42). Like in DhlB, the motif II Ser/Thr (Ser 114 ) and the motif III Lys (Lys 147 ) partly compensate the negative charge of the intermediate. Two of the three residues of the oxyanion hole in L-DEXs, Ser 171 , and Asn 173 are not found in the phosphatase and P-type ATPase members of the superfamily. This is not surprising as the proposed hydrolytic mechanism does not include the formation of an oxyanion intermediate, but instead the phosphoenzyme intermediate is hydrolyzed by attack on the phos- Only the C␣, C␤, C␥, O␦1, and O␦2 atoms of the nucleophilic aspartate residues were superimposed with r.m.s. differences for A and B of 0.19 and 0.09 Å, respectively. In B, the proposed L-DEX YL nucleophilic water molecule Wn is only 0.9 Å away from its DhlA counterpart. phorus atom (43). The negative charge that develops on the phosphoryl group was proposed to be counterbalanced by a magnesium ion (40).
Although many details of the L-DEX mechanism are now well understood, it is evident that further research is required to resolve the hydrolysis step of the dehalogenation mechanism of L-2-haloacid dehalogenase. In particular, a three-dimensional structure of the native enzyme at the pH optimum of 9.5 could provide useful information about the position of the nucleophilic water molecule, although so far we have not been able to obtain crystals that are stable at this pH.