Crystal Structure of the Hydroxyquinol 1,2-Dioxygenase from Nocardioides simplex 3E, a Key Enzyme Involved in Polychlorinated Aromatics Biodegradation*

Hydroxyquinol 1,2-dioxygenase (1,2-HQD) catalyzes the ring cleavage of hydroxyquinol (1,2,4-trihydroxybenzene), a central intermediate in the degradation of aromatic compounds including a variety of particularly recalcitrant polychloro- and nitroaromatic pollutants. We report here the primary sequence determination and the analysis of the crystal structure of the 1,2-HQD from Nocardioides simplex 3E solved at 1.75 Å resolution using the multiple wavelength anomalous dispersion of the two catalytic irons (1 Fe/293 amino acids). The catalytic Fe(III) coordination polyhedron composed by the side chains of Tyr164, Tyr197, His221, and His223 resembles that of the other known intradiol-cleaving dioxygenases, but several of the tertiary structure features are notably different. One of the most distinctive characteristics of the present structure is the extensive openings and consequent exposure to solvent of the upper part of the catalytic cavity arranged to favor the binding of hydroxyquinols but not catechols. A co-crystallized benzoate-like molecule is also found bound to the metal center forming a distinctive hydrogen bond network as observed previously also in 4-chlorocatechol 1,2-dioxygenase from Rhodococcus opacus 1CP. This is the first structure of an intradiol dioxygenase specialized in hydroxyquinol ring cleavage to be investigated in detail.

DNA sequencing showed that 1,2-HQDs are most closely related to catechol and chlorocatechol dioxygenases (7,11,21,31,32,43). Nevertheless, 1,2-HQDs appear to have a distinct substrate specificity and do not, or relatively slowly, convert catechol or substituted catechols, respectively (1, 2, 6, 7, 20, 22, 24, 25, 30 -32). Because 1,2-HQDs on one hand and (chloro)catechol dioxygenases on the other belong to different catabolic pathways, and correspondingly, the respective genes belong to different operons, the development of HQD substrate specificity was a very important step in the evolution of pathways for the efficient biodegradation of natural aromatic compounds as well as of xenobiotics. The analysis of the first crystal structure of a 1,2-HQD from N. simplex 3E (hereafter Ns 1,2-HQD), an enzyme that catalyzes the degradation of HQ with markedly high selectivity (30), could shed some light on the structural factors governing substrate specificity in a group of enzymes that catalyze key reactions in the biodegradation of toxic compounds.

Strain, Protein Preparation, and Sequencing of N Terminus and
Peptides-Growth of the 2,4,5-trichlorophenoxyacetic acid utilizing strain N. simplex 3E with 2,4-dichlorophenoxyacetic acid as an alternative carbon source has been reported previously (44). 1,2-HQD from N. simplex 3E was purified as reported previously (30). Tryptic peptides were isolated as described for other enzymes (45, 46) using 1 mg of 1,2-HQD for the digestion. Sequencing of the N terminus and of tryptic peptides was performed using an Applied Biosystems model 473A sequencer.
Cloning and Sequencing of the 1,2-HQD Gene-General methods for isolation and manipulation of DNA and cultivation of Escherichia coli cells were as reported previously (47). pBluescript II SK(ϩ) obtained from Stratagene was used as general cloning vector, and a T vector (48) derived from it was used for cloning PCR products. Recombinant plasmids were transformed into E. coli DH5␣, bought from Invitrogen. Genomic DNA from N. simplex 3E was prepared by the method of Wilson (49).
After cloning of the 400-bp PCR product into a T vector, giving rise to plasmid pNocSi01, sequencing of the fragment proved it to be homologous to the corresponding segments of other 1,2-HQD genes. Labeling of the 400-bp fragment by a DIG DNA Labeling and Detection Kit Nonradioactive (Roche Applied Science) was performed as described in the Roche manual. The probe was then used to detect the corresponding fragment on a Southern blot of 0.64 g of N. simplex 3E DNA digested with BamHI, PstI, SacI, and XhoI, respectively, and run on a 1% agarose gel with 1*TAE buffer (47). From a second gel, an area that corresponded in size to the hybridization signal (3 kb, SacI) was excised, and the included DNA was eluted and ligated into the dephosphorylated SacI site of pBluescript II SK(ϩ). After transformation of the ligation mixture into E. coli DH5␣, the labeled insert of pNocSi01 was used to identify clone pNocSi89 by colony hybridization.
The nucleotide sequence of the 1,2-HQD was determined by preparing subclones of pNocSi89 with the restriction enzymes SacII and XhoI. Two different 650-bp SacII restriction fragments and a 1.2-kb XhoI fragment, respectively, encode the complete sequence of the gene. For sequencing reactions, a MBI Fermentas CycleReader Auto DNA Sequencing Kit was used, with subsequent electrophoresis with a Li-cor 4200 IR 2 sequencer and analysis with the e-Seq program (version 1.2). Sequences were assembled using Staden Package version 2002.0. The sequence is available under GenBank/EMBL/DDBJ accession number AY822041. Comparisons with data-base entries were performed by using BLASTX (50). Multiple sequence alignments were created using ClustalX (version 1.8) (51).
Crystallization and Data Collection-The enzyme was crystallized at 293 K using the sitting drop vapor diffusion method from a solution containing 2.0 M ammonium sulfate, 4% polyethylene glycol 400, 100 mM Hepes pH 7.5 (52). The drops consisted of 4 l of 20 mg/ml protein solution and 6 l of reservoir solution equilibrated against 50 l of reservoir solution (Crystal Clear Strips from Molecular Dimension, Inc.).
A native data set extending to a maximum resolution of 1.75 Å was collected at the X11 beamline, EMBL, DESY, Hamburg. Data were collected using a MAR CCD165 detector at a wavelength of 0.908 Å. Crystals belong to the primitive monoclinic space group P2 1 with unit cell dimensions a ϭ 46.28, b ϭ 84.98, c ϭ 83.92 Å, ␤ ϭ 92.84°. For all data collections crystals of the native enzyme were cooled at 100 K adding 17% ethylene glycol to the mother liquor solution as cryoprotectant. Crystals suffered from damage if they were transferred in solution different from their mother solution unless they were previously cross-linked adding glutaraldehyde to the drops up to a final concentration of roughly 2% (v/v).
Metal Content Analysis-Analysis of the protein metal content was performed by using a PerkinElmer Optima 2000 Inductively Coupled Plasma AES (Atomic Emission Spectrometry) Dual Vision. The metal content analysis revealed the presence of 2 equivalents of iron ions and 1 equivalent of copper ions/mol of protein.
Structure Determination and Refinement-All molecular replacements attempts, using coordinates of known intradiol dioxygenases structures as a model, failed to provide a solution for Ns 1,2-HQD.
The structure of the enzyme was, therefore, solved by multiple wavelength anomalous dispersion (MAD) using the anomalous signal of the two catalytic irons. MAD data were collected at the BM14 beamline, ESFR, Grenoble. The data collected at three wavelengths (inflection, peak, remote) were processed and integrated with DENZO and scaled by SCALEPACK, from the HKL program suite (53).
The program SOLVE (54) was used to identify the two iron sites and SCHEME 1 for phase calculation. The 2.6 Å MAD phases were improved and extended to 2.2 Å by solvent flattening and histogram mapping using the program DM from the CCP4 program suite (55). Automatic tracing was performed initially with the program RESOLVE (56) and extended using ARP/WARP version 6.0 (57). After 200 cycles of refinement and 20 cycles of autobuilding, 532 amino acids of 586 were found and placed in 16 chains with a global connectivity index of 0.94. After this process manual intervention was required to complete the model. The model was initially refined against 2.2 Å resolution (MAD remote wavelength data set) and finally against 1.75 Å data, using the program Refmac 5.1.24 from the CCP4 program suite (55). Manual rebuilding of the model was performed using the program QUANTA (58). Solvent molecules were introduced automatically using ARP (57). Refinement resulted in R factor and R free values of 19.2 and 24.6%, respectively. Data processing and refinement statistics are summarized in Table I. The overall mean B factor of the structure after refinement was 25.68 Å 2 for chain A, 28.23 Å 2 for chain B, and 29.15 Å 2 for all atoms. Protein coordinates have been deposited in the Protein Data Bank (accession number 1TMX).
The final model is composed of residues 2-293 for chain A and 4 -293 for chain B, two Fe(III) ions, two benzoate ions, two phospholipids (C13/C17), two sulfate ions, one copper ion, one chloride ion, and 837 water molecules.
There are two disordered regions in chain B corresponding to residues 67-72 and 263-275. Electron density was missed for residues B264, B265, and B270 even for the main chain atoms, and they were not introduced in the model. Furthermore, no electron density was visible for the side chains of residues: A2 Ser, A73 Glu, B8 Glu, B71 Thr, B72 Asn, B75 Arg, B95 Asn, B264 Arg, B266 Pro, B276 Gln, and B277 Ile.
A spheric electron density in the F o Ϫ F c map at 12 was found bound to His A42 (2.12 Å) and His B42 (2.02 Å), and it could be easily attributed to a metal ion. A peak at lower height, compared with the iron ones, was present in the Harker section of the anomalous Patterson maps corresponding to the position of the metal ion. The metal should be one close to the iron as number of electrons because of the height of the peak in the F o Ϫ F c map (assuming a unitary occupancy) and because it should have an absorption edge near to the iron K edge. The coordination sphere of the metal is completed by a chloride ion (2.57 Å).
An electron density was found close to the iron in both active sites, and it was explained as a benzoate-like molecule bound to the iron in a bidentate way.
Two sulfate ions were found on the surface of the A chain, hydrogen bonded to some water molecules and side chain atoms.
Structure Analysis-The stereochemical quality of the models was assessed using the program PROCHECK (59). The Ramachandran plot is of a good quality, with 479 non-glycine and non-proline residues; among these, 438 (91.4%) are in the most favored regions, 38 (7.9%) are in the additional allowed regions, 1 (2%) (Thr B71) is in the generously allowed regions, and 2 (4%) (Glu B73 and Arg B74) are in disallowed regions.
The secondary structure was defined utilizing the DSSP data base and program (60).
Global structure superimpositions were carried out by utilizing the matching algorithm implemented into the HEX 4.2 program (61). Least squares fits of the active site regions were performed using the McLachlan algorithm as implemented in the program ProFit 2.2 (www.bioinf.org.uk/software/profit/) specifying as the fitting subset the four amino acid ligands to the catalytic iron ions (62).
Electrostatic Potentials were estimated first transforming the protein data base coordinate file into a pqr file containing partial charges and radii for each atom by using the PDB2PQR web service and then solving the second order differential Poisson-Boltzmann equation, which relates the electrostatic potential in a dielectric to the charge density using the macroscopic electrostatics with atomic details (MEAD) program package (63,64).
HQ was docked manually into the active site by first simulating the dissociation of Tyr 197 from the iron center. It was assumed that the substrates bind to the iron in a bidentate fashion and with orientations of their aromatic ring similar to those observed for catechol or 4-methylcatechol in Ac 1,2-CTD or for the benzoate ion in Ns 1,2-HQD and Rho 1,2-CCD (41,42). Slight rotations and/or tilts of the iron-bound substrate molecules did not result in changes in the amino acid residues interacting with the substrates ring substituents.

RESULTS AND DISCUSSION
Gene and Protein Sequence-Using primers directed toward regions conserved in other 1,2-HQDs, it was possible to amplify a segment of the expected size (400-bp) from genomic DNA of N. simplex 3E. The use of this fragment as a probe allowed us to clone a 3-kbp fragment of N. simplex 3E DNA into E. coli. Sequencing of two 650-bp and a 1.2-kbp region from several subclones resulted in the complete sequence of the presumed 1,2-HQD gene.
Independently from the genetic approach, the following sequences were obtained by amino acid sequencing of the N terminus and of nine tryptic peptides of the purified protein All of these peptides, except 1,2-HQD-27 (for which the sequence was of low quality), occur in the sequence predicted from the cloned N. simplex 3E gene (as well as in those from the tftH, hadC, dxnF genes), thus proving that, in fact, the gene of a 1,2-HQD was cloned and sequenced.
The most similar sequence in the data base was that of 1,2-HQD from Arthrobacter sp. strain BA-5-17 (32) (73% identical positions in the alignment of Fig. 4). The similarity to other 1,2-HQDs ranged from 42 to 69% identical positions (see alignment of Fig. 4). In contrast, the similarity to the representatives of catechol and chlorocatechol 1,2-dioxygenases given in Fig. 4 was between 22 and 30% identical positions.
Overall Structure and Linker Domain-The 1,2-HQD from the Gram-positive bacterium N. simplex 3E is a homodimeric protein with overall dimensions 110 ϫ 50 ϫ 50 Å. The statistics for data collection, phasing, and structure refinement are summarized in Table I. The final model is composed of residues 2-293 for chain A and 4 -293 for chain B, two Fe(III) ions, one copper ion, two benzoate ions, two phospholipids (C13/C17), two sulfate ions, one chloride ion and 837 water molecules.
The general topology of Ns 1,2-HQD resembles that of Ac 1,2-CTD and comprises two catalytic domains separated by a common "␣-helical zipper " motif that consists of six N-terminal helices from each subunit.
Two phospholipid molecules are located inside a large hydrophobic channel formed by the two protein monomers at the interface between the two subunits and in the center of the linker domain, with the head group directed outward into the solvent and the tail moieties pointing inward, toward each other (Fig. 1). A phosphatidylcholine molecule with two C12-13/C17 hydrophobic tails was used as a model because the absence of the electron density of the head groups did not allow determination of their precise identity, and the length of each tail was based on the length of the electron density and on the stereochemistry of known phospholipids. The presence of such phospholipids appears to be distinctive for this class of enzymes, although their possible role has still to be clarified (41,42).
The linker domain is mainly composed of three long (H1-H3) and three short (H4 -H6) ␣-helices supplied by each subunit: five helices from the N terminus of each monomer are interacting with the equivalent motif from the other subunit and with the sixth helix, which elongates from the catalytic domain ( Figs. 1 and 2). A metal ion bound to both His 42 (at 2.01 and 2.12 Å) from helices H2 of both subunits and to a chloride ion (at 2.57 Å) is shown in Fig. 1. The protein metal content analysis (see "Experimental Procedures") and the trigonal coordination geometry observed suggest that such metal ion is copper in an oxidation state I. Its location suggests a possible structural role in stabilization of the enzyme quaternary assembly for such metal ion. Fig. 2 shows the three-dimensional structural least squares Each catalytic pocket of Ns 1,2-HQD is bordered by Leu 80 and Asp 83 from helix H4; Val 107 , Phe 108 , Pro 110 , and Phe 111 from a first random coil; Trp 156 from sheet S3; Tyr 164 , the mobile iron ligand, from a second random coil; Tyr 197 from sheet S6; Ile 199 and Pro 200 from a third random coil; Arg 218 from sheet S7; His 221 and His 223 from sheet S8; His 237 from sheet S9; and Val 251 from a fourth random coil.
As shown in Fig. 3, A and B, a distinctive feature of the present structure is that each active site presents two openings. The first one located as the one observed previously in the Some of the active site residues, with the corresponding F o Ϫ F c density overlaid are depicted in Fig. 3C. The mononuclear Fe(III) ion shows a His 2 Tyr 2 coordination (Tyr 164 , Tyr 197 , His 221 , and His 223 ), typical of all intradiol ring cleaving dioxygenases (28). X-ray absorption spectroscopy data collected for the same Ns 1,2-HQD indicate that generally the native enzyme is pentacoordinated with two spheres of atoms: either two at 1.90 Å and three at 2.06 Å, or three at 1.92 Å and two at 2.08 Å (33). In the present crystal structure, a benzoate-like ion is coordinated to the iron ion in a bidentate asymmetric mode substituting the metal bound water molecule/hydroxide ion, observed in the native 1,2-CTDs, increasing the iron coordination number to 6 (Fig. 3C). An equivalent molecule has also been observed recently in the active site of Rho 1,2-CCD. In Ns 1,2-HQD the benzoate ion is stabilized by a hydrogen bond network that connects the benzoate O1 atom to Arg 218 NH1 (hydrogen-bonded further to Asp 249 ) and the benzoate O2 atom to a well ordered W16 active site water molecule (B factor ϭ 20.63) (hydrogen-bonded further to Phe 108 , Pro 110 , Phe 111 , and Trp 156 ). As observed in Rho 1,2-CCD the benzoate binding does not trigger the dissociation of Tyr 197 , although causing a conformational orientation of Arg 218 observed when substrates bind to intradiol dioxygenases, but contrarily to what observed in all 1,2-CTDs, 1,2-CCDs and 3,4-PCDs, Arg 218 , supposed to promote the substrate positioning and deprotonation, is not stabilized by a strong hydrogen bond to a Gln because this residue is replaced by His 237 , which is positioned a bit further away (35,37,38,40,41).
No convincing hypotheses can be made, at the moment, on the possible reasons for the presence of a benzoate-like molecule bound to the catalytic metal ion, although exogenous ligands have been often found bound to metal sites acting as stabilizers of the active enzyme by hampering metal ion dissociation. The molecule resembles benzoate or benzamide, which actually act as very weak competitive inhibitors for these enzymes, easily displaced by catechols or HQs. 2 In Table II the distances of the iron ligands are reported: Tyr 164 and Tyr 197 exhibit shorter bonds (1.95 and 2.01 Å, respectively) than His 221 and His 223 (2.11 and 2.24 Å, respectively), and the iron coordination sphere is completed by the benzoate-like molecule asymmetrically bound with O1 (ϳ2.13 Å) and O2 (ϳ2.59 Å).
Substrate Specificity-The catalytic mechanism of intradiol ring-cleaving dioxygenases has been proposed to operate via Fe(III) activation of the diol substrate first generating an Fe(II) semiquinone, which reacts directly with dioxygen to give a hydroperoxide intermediate, a Criegee rearrangement via acyl migration would finally generate the corresponding muconic anhydride (Scheme 2) (68). An alternative mechanism for migration of the electron-deficient acyl group, via a benzeneoxide-oxepin interconversion, has also been proposed (Scheme 2) (69).
Substrate selection and conversion are expected to be controlled mainly by the ring substituents effects on the electron density of the carbon atoms exposed to the molecular oxygen attack as well as by the interactions of ring substituents with the surrounding active site amino acidic residues.
Although a number of studies on inhibitors, substrates, and substrate analog adducts of 3,4-PCDs and Ac 1,2-CTD have revealed several important features of the mechanism of exogenous ligands binding to their active site, and the structure of 2 M. Ferraroni, A. Scozzafava, and F. Briganti, unpublished results. SCHEME 2 Rho 1,2-CCD has shed some light on the substrate selectivity of chlorocatechol-cleaving enzymes, no conclusive rationalization of the observed substrate specificities for intradiol-cleaving dioxygenases has been achieved so far. 1,2-HQDs structurally belong to the group of intradiol dioxygenases comprising 1,2-CTDs. 1,2-CTDs are generally divided into two types, I and II (70). Type I dioxygenases (1,2-CTDs) are relatively specific enzymes that primarily have catechol and often also a methylcatechol as substrate. Chlorinated catechols are not used or are used only at negligible rates. Type II enzymes (better known as 1,2-CCDs) are relatively nonspecific with a wider substrate range being able to convert chlorinated catechols more rapidly than catechol and to additionally accommodate a wide range of methyl-or methoxy-substituted catechols. Regarding substrate specificity, 1,2-HQDs seem to be more closely related to type I than to the type II enzymes (Table  III). Unfortunately, HQ was not tested as a potential substrate for most of the catechol and chlorocatechol 1,2-dioxygenases; a catechol 1,2-dioxygenase able to oxidize HQ at about half the rate (51%) of catechol was described only for T. cutaneum (71).
The Ns 1,2-HQD, subject of the present study, is remarkably substrate-selective. The conversions of HQ (K m ϭ 1.2 M, V max ϭ 55 units/mg, k cat ϭ 29 s Ϫ1 , and k cat /K m ϭ 24.2 M Ϫ1 s Ϫ1 ), 5CHQ (2.4% relative to HQ activity), and 6CHQ (5.0% relative to HQ activity) are catalyzed (30). A range of catechols and variously substituted quinols were also tested as substrates but found to only act as inhibitors for HQ turnover (K i ϳ 10 mM for catechols). Table III presents the comparison of the known substrate specificity data for 1,2-HQDs and some representatives of type I and II catechol-cleaving enzymes. The 1,2-HQDs from B. cepacia AC1100, Azotobacter sp. strain GP1, and T. cutaneum as well as the 1,2-CHQD from S. rochei 303, with respect to catechol, 3-methylcatechol, or pyrogallol conversion, show relatively high substrate specificities as the enzyme from N. simplex. HQ is the main substrate for the first four enzymes, whereas 6-CHQD is the best substrate for the S. rochei dioxygenase. 6-CHQD was also a relatively good substrate for the 1,2-HQDs of Azotobacter sp. GP1, N. simplex 3E, Wautersia eutropha JMP134, R. pickettii DTP0602, but was not tested with the fungal enzymes from T. cutaneum and P. chrysosporium. The enzyme from P. chrysosporium was found unable to convert 5-CHQ, but accepted catechol with 20% of the activity shown toward HQ. Also, the 1,2-HQD isolated from R. pickettii DTP0602 presents relatively low substrate specificity, being able to catalyze the oxidation of 3-methylcatechol and pyrogallol in addition to HQ and 6-CHQ, but it is inactive toward catechol, 3-and 4-chlorocatechol, 4-methylcatechol, protocatechuate, and 2,3-dihydroxybiphenyl (31). Furthermore, the 1,2-HQD of S. wittichii RW1 showed a high activity with catechol (7), and that of Arthrobacter sp. strain BA-5-17 was shown to catalyze both the intradiol and extradiol cleavage of catechol, although the activity toward HQ was 6.8-fold higher than that toward catechol (32).
These results evidence that substrate selectivity is a very heterogeneous issue even inside the 1,2-HQD group. It appears challenging to attempt the rationalization, at the molecular level, of the structural factors responsible for the differential substrate selectivity.
The structural alignment of the active site residues of the catecholate complex of Ac 1,2-CTD and the HQ docked in Ns 1,2-HQD is shown in Fig. 3D. The main interactions of Ac 1,2-CTD with the substrate involve the following residues (respective positions in the alignment of Fig. 4  To understand which of these residues could be mainly responsible for substrate recognition, HQ was docked into the active site of Ns 1,2-HQD. Two different orientations are likely: if the 4-OH substituent is oriented toward the internal part of the cavity (Fig. 3D) it would settle into a pocket formed by Asp 83 and Val 251 (positions 90 and 273, respectively, in Fig. 4), but if the substituent is oriented outward, it would essentially interact with Leu 80 and Pro 110 (positions 87 and 122 in Fig. 4, Leu 73 /Leu 49 and Pro 108 /Pro 77 in Ac in 1,2-CTD/Rho 1,2-CCD, respectively). This second hypothesis seems to be unlikely because identical amino acids would interact with the substrate in the different enzymes, thus not clearly accounting for their markedly different substrate specificity. Furthermore, the orientation of HQ with the 4-OH substituent toward the internal part of the cavity is equivalent to that of bound protocatechuate in 3,4-PCDs and of bound 4-methylcatechol in Ac 1,2-CTD (40, 41). A multiple sequence alignment of intradiol dioxygenases revealed several positions identical in all available 1,2-HQDs but different in one or more of the selected 1,2-CTDs or 1,2-CCDs (shaded dark gray in Fig. 4). Such positions may be conserved without a specific selection pressure, but they also may be of importance for the substrate specificity without a structural reason for it so far having been elucidated.
The presence of Asp 83 and Val 251 (positions 90 and 273 in Fig. 4), the residues expected to interact with substituents in position 4 (Pro 76 /Ala 53 and Ala 254 /Cys 224 in Ac 1,2-CTD/Rho 1,2CCD, respectively), based on structural comparisons should be mainly responsible for the selective preference for HQs. Valine residues corresponding to Val 251 (position 273 in Fig. 4) occur in all available 1,2-HQDs, but not in the (chloro-)catechol dioxygenases, a distribution supporting the conclusions drawn from structural comparisons.
Also, 3,4-PCDs are selective for a substrate carrying a hydrophilic substituent in a distal position with respect to the diol: protocatechuate. It has been shown that the carboxyl group of protocatechuate forms a hydrogen bond with Tyr 324 (substituted by Asp 83 in 1,2-HQD), and its negative charge is complemented by long range electrostatic interactions with Arg 133 , Arg 330 , and Arg 450 (Ac 3,4-PCD numbering) (40). In Ns 1,2-HQD Asp 83 could have a function similar to that of Tyr 324 in 3,4-PCDs, although in the case of HQ the substrate should not be charged at neutral pH but, Asp 83 would.
The other feature crucial for HQ selection and limited affinity for catechols is the presence of the second large active site opening caused by the substitution and spatial reallocation of residues Gly 72 and Asp 81 in Ac 1,2-CTD with Ile 79 and Ser 88 , amino acids specifically conserved in all 1,2-HQDs (positions 86 and 100, respectively in Fig. 4). Contributing to the second large active site opening is a marked shift of Pro 110 (position 122 in Fig. 4, Pro 108 in Ac 1,2-CTD). The resulting larger solvent exposition of the upper part of the cavity should electrostatically favor HQ settling and stabilization.

CONCLUSIONS
The crystal structure of Ns 1,2-HQD evidences significant differences with respect to the known 1,2-CCD and 1,2-CTD enzymes. Several secondary structure differences are noticed, but more importantly a number of residues within the active cleft and believed to be responsible for substrate selection are altered. Leu 80 , Asp 83 , Val 107 , Phe 108 , Gly 109 , Pro 110 , Phe 111 , Ile 199 , Pro 200 , Arg 218 , and Val 251 seem to be directly involved in interactions with substrates. Few amino acid differences (Asp 83 and Val 251 ) between the N. simplex enzyme which exhibits a marked selectivity toward HQ, 1,2-CCDs (Ala 53 and Cys 224 , respectively, in Rho 1,2CCD) and 1,2-CTDs (Pro 76 and Ala 254 , respectively, in Ac 1,2-CTD) as well as the extensive solvent exposure of the upper part of the catalytic site are believed to be responsible for the observed substrate selectivity differences among 1,2-HQDs and the other intradiol ring-cleaving dioxygenases.