The role of residue Thr249 in modulating the catalytic efficiency and substrate specificity of catechol-2,3-dioxygenase from Pseudomonas stutzeri OX1 Loredana Siani 1, *, Ambra Viggiani 1, *, Eugenio Notomista 1 , Alessandro Pezzella 2 and Alberto Di Donato 1 1 Dipartimento di Biologia Strutturale e Funzionale, Universita ` di Napoli Federico II, Napoli and CEINGE-Biotecnologie Avanzate S.c.ar.l., Italy 2 Dipartimento di Chimica Organica e Biochimica, Universita ` di Napoli Federico II, Italy Several bacteria are capable of using aromatic hydro- carbons as growth substrates [1–4]. The remarkable range of substrates that can be metabolized endows these microorganisms with the potential for bioremedi- ating environmentally dangerous substances such as benzene, toluene, xylene isomers, and polycyclic aro- matic hydrocarbons and their derivatives [5–8]. Because of their toxicity, several of these compounds are in the US Environmental Protection Agency prior- ity pollutant list (http://www.epa.gov). For example Keywords bioremediation; dioxygenase; enzyme kinetics; protein expression; Pseudomonas stutzeri Correspondence A. Di Donato, Dipartimento di Biologia Strutturale e Funzionale, Universita ` di Napoli Federico II, Via Cinthia, I-80126 Napoli, Italy Fax: +39 081 676710 Tel: +39 081 679143 E-mail: didonato@unina.it *These authors contributed equally to this work (Received 27 January 2006, revised 28 March 2006, accepted 4 May 2006) doi:10.1111/j.1742-4658.2006.05307.x Bioremediation strategies use microorganisms to remove hazardous sub- stances, such as aromatic molecules, from polluted sites. The applicability of these techniques would greatly benefit from the expansion of the cata- bolic ability of these bacteria in transforming a variety of aromatic com- pounds. Catechol-2,3-dioxygenase (C2,3O) from Pseudomonas stutzeri OX1 is a key enzyme in the catabolic pathway for aromatic molecules. Its specif- icity and regioselectivity control the range of molecules degraded through the catabolic pathway of the microorganism that is able to use aromatic hydrocarbons as growth substrates. We have used in silico substrate dock- ing procedures to investigate the molecular determinants that direct the enzyme substrate specificity. In particular, we looked for a possible molecular explanation of the inability of catechol-2,3-dioxygenase to cleave 3,5-dimethylcatechol and 3,6-dimethylcatechol and of the efficient clea- vage of 3,4-dimethylcatechol. The docking study suggested that reduction in the volume of the side chain of residue 249 could allow the binding of 3,5-dimethylcatechol and 3,6-dimethylcatechol. This information was used to prepare and characterize mutants at position 249. The kinetic and regio- specificity parameters of the mutants confirm the docking predictions, and indicate that this position controls the substrate specificity of catechol-2,3- dioxygenase. Moreover, our results suggest that Thr249 also plays a previ- ously unsuspected role in the catalytic mechanism of substrate cleavage. The hypothesis is advanced that a water molecule bound between one of the hydroxyl groups of the substrate and the side chain of Thr249 favors the deprotonation ⁄ protonation of this hydroxyl group, thus assisting the final steps of the cleavage reaction. Abbreviations C2,3O, catechol-2,3-dioxygenase; DHBD, 2,3-dihydroxybiphenyl-1,2-dioxygenase; DHND, 1,2-dihydroxynaphthalene dioxygenase; DHpCD, 2,3-dihydroxy-p-cumate dioxygenase; DMC, dimethylcatechol; ECD, extradiol ring cleavage dioxygenase; HPCD, 3,4-dihydroxyphenylacetate (homoprotocatechuate)-2,3-dioxygenase; IBX, o-iodoxybenzoic acid; 3-MC, 3-methylcatechol; 4-MC, 4-methylcatechol; PH, phenol hydroxylase; THTD, 2,4,5-trihydroxytoluene-5,6-dioxygenase; ToMo, toluene ⁄ o-xylene monooxygenase. FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS 2963 long-term exposure of humans to benzene, toluene and xylene could cause damage to the central nervous sys- tem, liver and kidneys, chromosomal aberrations and cancer [9–12]. Extradiol ring cleavage dioxygenases (ECDs) are Fe(II)-dependent enzymes that catalyze a crucial ring- opening step in the catabolic pathways of microorgan- isms capable of growing on aromatic compounds [13–15]. ECDs cleave ortho-dihydroxylated aromatic rings by catalyzing the addition of two atoms from molecular oxygen at one of the C–C bonds adjacent to the diol (metacleavage; Fig. 1) to produce nonaromatic molecules that eventually enter central metabolic path- ways [13,14]. ECDs comprise five evolutionarily related subfamilies [16] that include catechol-2,3-dioxygenase (C2,3O; EC 1.13.11.2) (subfamily 1), 2,3-dihydroxybi- phenyl-1,2-dioxygenase (DHBD; EC 1.13.11.39) and 1,2-dihydroxynaphthalene dioxygenase (DHND, sub- family 2), 3,4-dihydroxyphenylacetate-2,3-dioxygenase (HPCD; EC 1.13.11.15) (subfamily 3), 2,3-dihydroxy- p-cumate-3,4-dioxygenase (DHpCD, subfamily 5) and 2,4,5-trihydroxytoluene-5,6-dioxygenase (THTD, sub- family 6). Even though the range of substrates that can be oxidized by ECDs is broad, each enzyme in the family displays restricted substrate specificity and reg- ioselectivity. ECDs belonging to subfamilies 1, 2 and 3 cleave catechols substituted at positions 3 and ⁄ or 4 at the bond adjacent to the diol and proximal to the sub- stituent, as shown in Fig. 1A [17–23]. ECDs belonging to subfamilies 5 and 6, such as DHpCD and THTD, catalyze the transformation of 3,6-disubstituted and 4,5-disubstituted catechols, respectively [24–26], and exhibit high regioselectivity by cleaving the bond prox- imal to the alkylic group of the substrates as shown in Fig. 1B,C [24–26]. The size of the substitutent that can be accommodated by a subfamily varies. For instance, C2,3Os can cleave catechols with small substituents at positions 3 and 4, such as 3-methylcatechol (3-MC) and 3,4-dimethylcatechol (3,4-DMC) [17,18], whereas enzymes belonging to subfamily 2 act on catechols with large substituents at the same positions [19–21,27] (Fig. 1A). The complete degradation of aromatic molecules is initiated by monooxygenases and dioxygenases, which produce dihydroxylated compounds in the upper metabolic pathways [28,29]. These diols are cleaved subsequently by ECDs. Since monooxygenases and dioxygenases usually exhibit a wide range of substrate specificity, they produce several dihydroxylated prod- ucts, some of which are not always substrates for ECDs and cannot be degraded further. As a conse- quence, ECDs represent the gate that controls the flow of molecules entering the lower metabolic pathways [14,28,29], by reducing the range of aromatic com- pounds that can be used by microorganisms as growth substrates. Thus, enhancement of the catabolic poten- tial of ECDs would represent a valuable tool for bio- remediation strategies by widening the number of substrates that can be consumed by bacteria that depend on these enzymes for the utilization of specific aromatic substrates as their primary source of carbon and energy. Pseudomonas stutzeri OX1 is an ideal model organ- ism for these studies, since it can utilize benzene, tolu- ene, and o-xylene, but not m-xylene and p-xylene, as sole sources of carbon and energy [30]. Two NADH- dependent monooxygenases—toluene ⁄ o-xylene mono- oxygenase (ToMO) and phenol hydroxylase (PH)—act sequentially in the microorganism [31] to convert aro- matic hydrocarbons to the corresponding catechols. These are cleaved by a C2,3O that is nearly identical to the well-characterized enzyme from Pseudomonas putida MT2 [18,32]. ToMO and PH are able to convert o-xylene as well as m-xylene and p-xylene to 3,4-DMC, 3,5-DMC and 3,6-DMC, respectively (unpublished results). However, P. stutzeri C2,3O can cleave only 3,4-DMC effectively [32], allowing this product to be Fig. 1. Scheme of the reactions catalyzed by extradiol ring cleavage dioxygenases (ECDs). Reactions catalyzed by (A) catechol-2,3-dioxy- genases (C2,3Os), 2,3-dihydroxybiphenyl-1,2-dioxygenases (DHBDs) and 3,4-dihydroxyphenylactetate-2,3-dioxygenase (HPCD), (B) by 2,3-dihydroxy-p-cumate dioxygenases (DHpCDs), and (C) by 2,4,5- trihydroxytoluene-5,6,dioxygenases (THTDs). Thr249 in catechol-2,3-dioxygenase function L. Siani et al. 2964 FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS further metabolized through the lower pathway. This is not possible in the case of 3,5-DMC and 3,6-DMC, because of the very low activity of C2,3O towards these compounds [32]. Thus, the restricted specificity of C2,3O is the primary metabolic determinant that limits the ability of P. stutzeri OX1 to efficiently grow on xylene mixtures. Moreover, the inability of P. stut- zeri C2,3O to cleave 3,5-DMC and 3,6-DMC also has an adverse effect on the metabolism of the micro- organism, since the NADH consumed by the mono- oxygenase-catalyzed hydroxylations of m-xylene and p-xylene cannot be restored by the lower pathway reac- tions. This inefficiency results in a loss of metabolic reducing power when P. stutzeri OX1 grows on xylene mixtures. An understanding of the molecular determi- nants that control the substrate specificity of P. stutzeri C2,3O offers an opportunity to develop molecular strategies aimed at adjusting the active site pocket of C2,3O to control the products of the enzyme-catalyzed reaction. Such adjustment could enhance the ability of the microorganism to grow on substituted aromatic compounds. Here, we report a study of the molecular determinants of C2,3O substrate specificity carried out by in silico substrate docking procedures followed by the preparation and characterization of mutants at position 249. Our findings indicate that Thr249 partici- pates in the control of substrate specificity and plays a previously unsuspected role in catalysis. Results Modeling of (di)methylcatechols in the active site of C2,3O C2,3Os from P. putida MT2 and P. stutzeri OX1 have nearly identical C-terminal catalytic domains, except for a single conservative substitution of leucine for valine at position 225 in the P. stutzeri enzyme. Since this substitution is 14 A ˚ from the catalytic iron atom, it is likely that the active sites of the two C2,3Os are structurally identical and that the crystal structure of P. putida MT2 C2,3O (PDB accession code, 1mpy [33]) would serve as an accurate model for investi- gating the interactions of docked methylcatechols and dimethylcatechols with the C2,3O substrate-binding pocket. The structures of two ECDs, DHBD from Pseudo- monas KKS102 (1eim [34]), and HPCD from Brevibac- terium fuscum (1q0c [35]), crystallized in their active Fe(II) form with the substrate bound to the catalytic metal, were used as templates for initial positioning of catechols in the active site of C2,3O. The available data suggest that the two structures (Fig. 2A,B) repre- sent the catalytically competent enzyme–substrate com- plex [34,35]. First, the catalytic C2,3O iron atom and three ligands (His154, His214, Glu265) were superim- posed on the corresponding atoms of DHBD (His145, His209, Glu260) and HPCD (His155, His214, Glu267). After superimposition of the active site atoms of C2,3O on the corresponding atoms of DHBD and HPCD, r.m.s.d. values were 0.35 A ˚ and 0.24 A ˚ , res- pectively. Then, a (substituted) catechol molecule was superimposed on the corresponding atoms of dihydroxy- biphenylacetate or dihydroxyphenylacetate to obtain two models of a catechol–C2,3O complex, named 1 and 2, respectively, in which the geometric parameters of the metal center atoms are very similar to those found in the DHBD and HPCD structures. The two models were inspected to find close molecular contacts between the catechol ring and the residues surrounding the binding pocket. The two complexes were very sim- ilar. In both structures, the largest contacts were found between the plane of the substrate ring and the plane of the imidazole ring of residue His246, which make p contacts. However, it should be noted that in complex 2, based on the HPCD structure, the average distance between the two interacting rings (3.0 A ˚ ) is lower than that measured in complex 1 (3.6 A ˚ ). The same distance is 3.6 A ˚ in the DHBD complex and 3.5 A ˚ in the HPCD complex (Fig. 2A,B). No other close molecular contacts were found in the two models. Given the high similarity between the two complex models, complex 1, based on the DHBD structure, was selected for further analyses. Owing to changes in the conformation of the back- bone structures in C2,3O, the side chain of His246 is shifted towards the substrate, resulting in a larger overlap between the stacked rings. Moreover, the side of the substrate ring opposite to His246 faces the edge of the Phe191 side chain (Phe186 in DHBD, Trp192 in HPCD) (not shown). The contacts between the edge of the dihydroxylated substrate ring and the active site pocket are probably involved in the determination of substrate specificity. Inspection of the substrate CH atoms at positions 3 and 4 reveals that they point towards small cavities, indicated as subsites 1¢ and 2¢ in Fig. 2C, which are defined by residues Ile204, Phe302, Ile291 and Leu248. Although the volume of subsite 2¢ is smaller than that of subsite 1¢, these cavit- ies are large enough to accommodate methyl substitu- ents at positions 3 and 4, as verified by the docking of 3-MC, 4-MC and 3,4-DMC. A model of the complex between C2,3O and 3,4-DMC is depicted in Fig. 2C. In HPCD, in contrast to what is observed in the model of the C2,3O complex, the cavity corresponding to subsite 2¢ is larger and contains two arginine residues L. Siani et al. Thr249 in catechol-2,3-dioxygenase function FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS 2965 that interact with the carboxylate group of homoproto- catechuate (Figs 1A and 2B). In DHBD, this region is open to the solvent, thus allowing for the binding of larger substituents (Fig. 2A). The CH atoms of the substrate ring at positions 5 and 6 point towards the backbone of Leu248 and the side chain of Thr249, respectively (Fig. 2C). Appar- ently, the close contacts between these two residues and the edge of the substrate ring could prevent bind- ing of 3,6-DMC and 3,5-DMC, as shown in Fig. 2- D,E,F. Thus, the binding of 3,5-DMC and 3,6-DMC to the active site of C2,3O could be possible if the con- formation of the active site changes with respect to the one observed in the crystal structure of C2,3O upon binding of the dimethylcatechols. Since the CH atoms at position 5 of the substrate ring point towards the backbone carbonyl group of Leu248, replacement of the side chain at this position would not be able to create space for accommodating a methyl group at position 5 (Fig. 2D,E,F). The CH atoms at position 6, however, contact the side chain of residue Thr249. The tightest substrate–enzyme contacts were located between the CH at position 6 and the methyl group of the Thr249 side chain. In the four protomers of C2,3O, the Thr249 side chain shows the same orientation, probably due to a hydrogen bond between the OH group of Thr249 and the oxygen atom of the Leu248 carbonyl group (the two oxygen atoms are at 2.7 A ˚ distance). A 180° rotation along the Ca–Cb bond would minimize the interaction between the side chain and the substrate bound in the putative productive conformation. However, it would also prevent formation of the hydrogen bond between the Thr249 side chain and the backbone. A reduction in the volume of this side chain might provide room for housing a methyl substituent at this position and allow for the binding of 3,6-DMC or 3,5-DMC, as depicted in Fig. 2D,E. Fig. 2. Scheme of the active sites of 2,3- dihydroxybiphenyl-1,2-dioxygenase (DHBD), 3,4-dihydroxyphenylactetate-2,3-dioxygenase (HPCD) and catechol-2,3-dioxygenase (C2,3O). (A) DHBD from Pseudomonas sp. KKS102 with 2,3-dihydroxybiphenyl bound (PDB code 1eim). (B) Brevibacterium fuscum HPCD (PDB code 1q0c) with homoproto- catechuate bound. (C,D) Pseudomonas putida C2,3O (PDB code 1mpy) with 3,4- dimethylcatechol (3,4-DMC) or 3,6-DMC, respectively, docked in the active site. Schemes in (E) and (F) illustrate the active site of P. putida C2,3O with 3,5-DMC docked in the active site in two different ori- entations. Arrows indicate groups at distan- ces between 3 A ˚ and 4.2 A ˚ . Arrows in bold indicate groups at distances less than the sum of the van der Waals’ radii. Hydrogen bonds are shown as dotted lines. The schemes of the side chains are shown only when the side chain makes the closest con- tact between the residue and the substrate. Lines with round ends indicate stacking between the ring of the substrate and His240 in DHBD (A), His248 in HPCD (B), and His246 in C2,3O (C–F). Thr249 in catechol-2,3-dioxygenase function L. Siani et al. 2966 FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS Based on the above observations, residue Thr249 was substituted in silico with valine, serine, alanine and glycine, and the molecular contacts of docked 3,6-DMC and 3,5-DMC were reinspected. Mutation T249V does not allow for a reduction of the steric hin- drance with dimethylated substrates. Indeed, the valine rotamer, which better fits the C2,3O active site, has one of the methyl groups at approximately the same position as the Thr249 methyl group that approaches the substrate. On the contrary, the progressive reduc- tion of the side chain of residue 249 caused by muta- tion of threonine to serine, alanine and glycine creates a new cavity (subsite 3¢) adjacent to CH atoms at posi- tion 6, resulting in the reduction of steric hindrance between a methyl group at this position and the pro- tein. The reduction of steric hindrance caused by mutations was estimated by measuring the radius of the largest sphere that can be fitted to the active site, using as center of the sphere the coordinates of the carbon atom of the methyl group at position 6 of 3,6- DMC bound as shown in Fig. 2D. The radius increa- ses from 0.76 A ˚ —measured for wild-type C2,3O—to 0.98, 1.25 and 1.91 A ˚ for T249S C2,3O, T249A C2,3O, and T249G C2,3O, respectively. As the radius of a methyl group is 1.9–2 A ˚ , it can be predicted that the ability of the mutants to bind dimethylcatech- ols in a productive conformation should increase pro- gressively, reaching its maximum in mutant T249G C2,3O. Kinetic parameters and regioselectivity of wild-type and mutated C2,3O To investigate the influence of the side chain of residue Thr249 of C2,3O on the cleavage of 3,5-DMC and 3,6- DMC, the catalytic properties of mutants were studied. Based on the results of docking studies, three mutants were produced by site-directed mutagenesis: T249S C2,3O, T249A C2,3O, and T249G C2,3O. All of the mutated proteins were active on catechol (Table 1), and had an iron content similar to that of wild-type C2,3O. The regioselectivity of the wild-type and mutant C2,3Os were determined by incubating them with 3-MC or 3,5-DMC, and analyzing the cleavage prod- uct by NMR after extraction with ethyl acetate. For all of the C2,3O variants, no aldehydic hydrogen was detected in the product when 3-MC was used as a sub- strate, indicating that other possible products of ring cleavage distal to the methyl group, if present, were below the detection limit (less than about 0.6–0.5% of the cleavage product). On the other hand, when 3,5- DMC was used as a substrate, the 1 H spectrum of the product showed a signal at d 9.44, a value consistent with that of an aldehydic hydrogen for a conjugate aldehyde. Moreover, no signal that could be assigned to hydrogen atoms of the product of ring cleavage proximal to the methyl group at position 3 was ever found at the expected field. This indicates that the cleavage of 3,5-DMC is distal (‡ 99.0%) to the methyl group at position 3 (Fig. 3). Thus, the analysis above leads to the conclusion that 2-hydroxy-6-oxohepta-2,4-dienoic acid and 2-hydroxy- 3,5-dimethyl-6-oxohexa-2,4-dienoic acid (Fig. 3) are the sole or main products of 3-MC and 3,5-DMC ring cleavage, respectively (the NMR spectra of the clea- vage products are shown in Supplementary Fig. 1). The kinetic parameters of wild-type C2,3O were determined on purified 3,5-DMC and 3,6-DMC (Table 1). The K m values were found to be 74 lm and 21 lm, respectively, which are approximately 50 and 14 times higher than that measured on catechol. More- over, the k cat values were found to be very low, 0.36 s )1 for 3,5-DMC and 0.66 s )1 for 3,6-DMC. These values are about 0.2–0.5% of that measured on catechol (180 s )1 ). Therefore, the low reactivity of Fig. 3. Possible extradiol cleavage reactions for (A) 3,6-dimethyl- catechol (3,6-DMC), (B), 3-methylcatechol (3-MC) and (C) 3,5- dimethylcatechol (3,5-DMC) (C). L. Siani et al. Thr249 in catechol-2,3-dioxygenase function FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS 2967 C2,3O from P. stutzeri towards 3,5-DMC and 3,6- DMC seems to depend on both weak binding and slow catalysis. Figure 4 shows the kinetic parameters determined at pH 7.5 using catechol, 3-MC, 3,5-DMC and 3,6-DMC as a function of the radius of the largest sphere that can be fitted to the active site of wild-type and mutated C2,3O as described in the previous section. This is a direct measurement of the volume of subsite 3¢ (Fig. 5), and hence of the ability of the enzyme to bind 3,5-DMC and 3,6-DMC in an orientation similar to those of catechol and 3-MC. The K m values of catechol and 3-MC (Table 1) show a regular, progressive increase as the volume of subsite 3¢ increases (Fig. 4A). In contrast, the K m values on 3,5-DMC and 3,6-DMC decrease with the increase of the volume of subsite 3¢. Assuming that the kinetics of C2,3O follow the Michaelis–Menten relationship, these results indicate that a reduction of the volume of resi- due 249 increases the affinity of the enzyme for 3,5- DMC and 3,6-DMC and decreases the affinity for smaller substrates. Mutations at position 249 result in large and partly unexpected variations in the k cat values (Table 1). For the smaller substrates, catechol and 3-MC, the T249S mutation has little or no effect on the catalytic con- stants, whereas replacement of the threonine residue with an alanine or a glycine residue causes a significant reduction of the k cat values with respect to those meas- ured for the wild-type enzyme; approximately four-fold and 20-fold for catechol and 3-MC, respectively. On the other hand, the behavior of the mutants is very different in the case of dimethylcatechols. The T249S mutation causes an increase in the k cat values on dime- thylcatechols with respect to the wild-type enzyme. In the case of 3,5-DMC, the k cat value is about eight times higher than that of the wild-type enzyme. On the contrary, mutations T249A and T249G have no signifi- cant effect on the catalytic constants measured for Fig. 4. Catalytic parameters of wild-type and mutant catechol-2,3- dioxygenases (C2,3Os) measured at pH 7.5 are shown as functions of the radii of subsite 3¢ shown in Fig. 5 (radii are: 0.76, 0.98, 1.25 and 1.91 A ˚ for wild-type, T249S, T249A and T249G C2,3O, respect- ively). Filled circles, catechol; open circles, 3-methylcatechol (3-MC); filled triangles, 3,6-dimethylcatechol (3,6-DMC); open trian- gles, 3,5-DMC. For clarity in (B), the k cat ⁄ K m values on catechol and 3-MC and the values on 3,5-DMC and 3,6-DMC are reported on dif- ferent scales—on the left and on the right, respectively. Table 1. Kinetic parameters of wild-type and mutated catechol-2,3-dioxygenase. Substrate Residue at position 249 Thr Ser Ala Gly K m (lM) Catechol 1 ± 0.09 22.5 ± 2 37 ± 3 63.6 ± 5 3-MC 3.8 ± 0.4 11.5 ± 1 14.3 ± 1 26.6 ± 3 3,5-DMC 73.8 ± 6 57.5 ± 4 38.1 ± 3 23.7 ± 2 3,6-DMC 21.5 ± 2 9.7 ± 1 5.5 ± 0.6 7.4 ± 0.6 Catechol 180 ± 11 170 ± 10 48 ± 3 47.7 ± 3 k cat (s )1 ) 3-MC 118 ± 10 60.5 ± 6 6.3 ± 0.6 2.6 ± 0.3 3,5-DMC 0.36 ± 0.04 2.65 ± 0.18 1 ± 0.08 0.4 ± 0.03 3,6-DMC 0.66 ± 0.07 1.2 ± 0.1 0.4 ± 0.03 0.23 ± 0.02 Catechol 180 ± 27 7.5 ± 1 1.3 ± 0.18 0.8 ± 0.11 k cat ⁄ K m (lM )1 Æs )1 ) 3-MC 31.0 ± 5.8 5.3 ± 0.9 0.45 ± 0.07 0.1 ± 0.02 3,5-DMC 0.005 ± 0.0009 0.05 ± 0.007 0.026 ± 0.004 0.016 ± 0.002 3,6-DMC 0.03 ± 0.006 0.13 ± 0.02 0.074 ± 0.013 0.021 ± 0.003 Thr249 in catechol-2,3-dioxygenase function L. Siani et al. 2968 FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS 3,5-DMC and cause a small decrease in the catalytic constants on 3,6-DMC. Discussion Bioremediation techniques are based on the use of microorganisms to remove hazardous substances, such as aromatic molecules, from polluted areas [5,6]. The expansion of the catabolic potential of these bacteria would greatly improve the applicability of these tech- niques, by increasing the number of molecules that can be metabolized by the microorganisms. ECD specificity and regioselectivity control the range of molecules that can be degraded through the catabolic pathways of bacteria capable of using aromatic hydrocarbons as growth substrates [14,28,29]. Knowledge of the molecular determinants that direct their substrate specificity is essential to tailor their active site to transform a wider range of substrates, hence widening the ability of the microorganism to grow on aromatic compounds. Members of the different subfamilies of ECD cata- lyze the oxidative cleavage of a very wide range of dihydroxylated aromatic substrates, ranging from the simple ring of catechol to multiple substituted catech- ols and polycyclic molecules [17–23]. Despite differ- ences in their specificity, the catalytic residues seem to be very well conserved. Six residues of the active site are completely conserved [36,37]: the three ligands to the catalytic metal (His154, His214, Glu265 in P. stut- zeri C2,3O), two histidines that have been suggested to act as acid–base catalysts (His199 and His246), and Tyr255, which is responsible for the correct positioning of the substrate [18,34,38]. The structures of two DHBDs, an Fe 2+ -dependent HPCD and an Mn 2+ -dependent HPCD are available in their reduced, active forms with the substrate bound to the active site [34,35]. In each of these structures, the substrate is bound similarly to both the catalytic metal and the conserved residues in the active site pocket. One of the substrate hydroxyl groups is posi- tioned near the conserved tyrosine residue and is closer to the metal atom than the other hydroxyl group [34,35]. Available data suggest that the hydroxyl group facing the conserved tyrosine is in the anionic form [34,35]. To shed light on the specificity of the enzyme for dimethylcatechols, the information above was used to construct models of the complexes between P. stutzeri C2,3O, a member of subfamily 1 ECDs, and different substrates. The models of the complexes indicate that the orien- tation of the substrate in the active site pocket of the C2,3O is very similar to that observed in the structure of the DHBD and HPCD complexes. A closer compar- ison of the X-ray structures and of our models of C2,3O with bound catechols reveals that the residues interacting with the first hydroxyl group that is strongly coordinated to the metal atom, and those interacting with the two faces of the substrate ring, are conserved. The polypeptide regions that contact the edge of the ring, however, are variable in the different proteins. Thus, it is likely that the determinants of sub- strate specificity reside in these regions. The model of C2,3O with catechol bound in the act- ive site pocket reveals the presence of two small sub- sites, 1¢ and 2¢ (Fig. 2C), facing positions 3 and 4 of the substrate ring. The volume of these cavities is large enough to accommode methyl substituents at positions 3 and 4, thus providing a molecular scaffold to sup- port C2,3O binding and cleavage of 3,4-DMC. Subsite 2¢, which is adjacent to position 4, is slightly smaller Fig. 5. Scheme of possible binding of 3-methylcatechol (3-MC), 3,5-dimethylcatechol (3,5-DMC) and 3,6-DMC to catechol-2,3- dioxygenase (C2,3O) active site. (A,B) Binding of 3-MC and 4-MC, respectively, to the active site of wild-type Pseudomonas stutzeri C2,3O. (C,E) Two possible orientations for the binding of 3-MC to the active site of T246G C2,3O. (D,F) Binding of 3,5-DMC and 3,6-DMC, respectively, to the active site of T246G C2,3O. L. Siani et al. Thr249 in catechol-2,3-dioxygenase function FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS 2969 than subsite 1¢ facing position 3 of the catechol ring (Fig. 2C). This difference could explain why 4-substi- tuted catechols are more inactivating substrates than 3-substituted catechols [17,18]. The region of C2,3O adjacent to positions 5 and 6 of the catechol ring pro- vides no space for the binding of substituents at these positions. This may suggest a structural basis for the fact that C2,3O is not able to cleave catechols with substituents at positions 3,5, or 3,6. (Fig. 2D,E,F). In fact, a strong 20–70-fold decrease in the affinity of P. stuzeri C2,3O for 3,5-DMC and 3,6-DMC with respect to unsubstituted catechol and to 3-DMC is found (Table 1). The region facing positions 5 and 6 of the catechol ring is mainly formed by a loop containing residues 246–249 in C2,3O (240–243 in DHBD and 248–251 in HPCD) (Fig. 2). Multiple alignments of ECDs show that this loop is well conserved within each subfamily (Supplementary Fig. 2). The consensus sequence of the loop is H-G-(L ⁄ I ⁄ V ⁄ F)-T in C2,3Os, H-(T ⁄ A ⁄ S ⁄ P)- N-D in DHBDs and H-G-(V ⁄ I ⁄ L)-S in HPCDs. Despite the differences in their primary structures, the three different types of loop tightly contact positions 5 and 6 of the substrate ring in a very similar fashion (Fig. 2). It should be noted that none of the members of the C2,3O, DHBD or HPCD subfamilies have been reported to cleave catechols with substituents at both positions 3 and 5 or 3 and 6. Members of the DHpCD subfamily, on the other hand, have been reported to cleave 3,6-substituted catechols. This sub- family has the loop consensus sequence H-P-(P ⁄ T)-S. Unfortunately, there is no available structure for any member of the DHpCD subfamily that could provide insight into the contacts between the loop residues and the substrate. Thus, the structure of 2,3-dihydrox- y-p-cumate-3,4-dioxygenase from P. putida F1, a member of the DHpCD subfamily, was modeled with the substrate bound in the active site using the struc- ture of DHBD from Burkholderia cepacia LB400 (1kmy [38]), as a template. We found (data not shown) that the loop containing residues 235–238 of DHpCD, with the sequence H-P-P-S, can assume a conformation that easily accommodates the carboxy- late group of the aromatic substrate dihydroxy-p-cu- mate, whereas the isopropyl group of the substrate can be housed in a cavity corresponding to subsite 1¢ of C2,3O (Fig. 2C). Moreover, the model indicates that the carboxylate group can hydrogen bond to Ser238 of the loop (data not shown). The model also suggests that the active sites of other ECDs could be enlarged to accommodate 3,6-disubstituted catechols by inducing small changes to the loop 246–249 (C2,3O numbering). The modeling studies of the C2,3O complexes indi- cate that the active site of this enzyme can accommo- date one methyl group from 3,5-DMC or 3,6-DMC in subsites 1¢ or 2¢, but not a second, because of the dif- ferent structure of loop 246–249 of C2,3O (subsite 3¢) with respect to that of the homologous loop 235–238 of DHpCD. Thus, the steric hindrance between the second methyl group and loop 246–249 could force the dimethylated substrate to bind in an orientation that is not suitable for efficient catalysis. This hypothesis is supported by the low affinity and low catalytic effi- ciency of wild-type C2,3O on 3,5-DMC and 3,6-DMC and by the results we have obtained from the study of C2,3O Thr249 mutants. The K m values in Fig. 4A indicate, as expected, that the apparent affinity of dimethylcatechols for C23O increases as the steric hindrance at position 249 decrea- ses. Moreover, the 3,6-DMC K m values for wild-type and mutant C2,3O are lower than those measured for 3,5-DMC and are in agreement with the models shown in Fig. 5D,F. Figure 5F shows that the two methyl groups of 3,6-DMC are housed in subsites 1¢ and 3¢, whereas in the model of Fig. 5D, the methyl groups of 3,5-DMC are housed in subsites 2¢ and 3 ¢. The smaller volume of subsite 2¢ compared to subsite 1¢ may explain the lower affinity of wild-type and mutant C2,3O for 3,5-DMC with respect to 3,6-DMC. Interestingly, the progressive decrease of the dimen- sion of the residue 249 side chain also causes an increase in the K m values for catechol and 3-MC (Fig. 4A). In the case of the smaller side chain, in mutant T249G C2,3O, the K m values are 63 and seven times higher, respectively, than those measured for the wild-type enzyme, suggesting that residue Thr249 might make an energetic contribution to substrate binding. Thr249 could contribute to substrate binding either through van der Waals’ contacts as described in Results, or a through a hydrogen bond network, dis- cussed later in this section. Thr249 mutants also give information on factors that control the regioselectivity of C2,3O. 3-MC might be cleaved at two different bonds (Fig. 3), yielding two different extradiol cleavage products. All known ECDs belonging to subfamilies 1, 2 and 3 catalyze only the proximal cleavage [17,19–23] (Fig. 3). It has been reported that this regioselectivity could depend either on the reactivity of the substrate or on the asymmetry of the active site that forces the binding of the sub- strate in the monoanionic form [38–41]. The decrease in K m values of mutant T249G C2,3O on 3,5-DMC and 3,6-DMC could indicate that the T249G mutation is successful in opening a new subsite (subsite 3¢) for methyl binding. Thus, the presence of a new cavity in Thr249 in catechol-2,3-dioxygenase function L. Siani et al. 2970 FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS the active site pocket of mutant T249G C2,3O should allow for the binding of 3-MC in two different orienta- tions, i.e. with the methyl group housed in subsite 1¢ or in subsite 3¢ (Fig. 5C,E). As reported in Results, the formation of the distal cleavage product (Fig. 3) has never been observed, either with T249G C2,3O or with the other two mutants. These data suggest that the regioselectivity of the cleavage of 3-MC is proximal, independently of the orientation of the substrate in the binding site. Thus, the regioselectivity of cleavage would be mainly controlled by the reactivity of the substrate. This could explain the finding that wild-type C2,3O and its mutants cleave 3,5-DMC only at the bond proximal to the methyl group at position 5. This regioselectivity could indicate that the methyl group at this position is more activating than the methyl group at position 3. This latter hypothesis is reinforced by the observation that the k cat value of P. stutzeri C2,3O for 4-MC is two times higher than the k cat value for 3-MC [18]. Figure 5B,D show that 4-MC and 3,5- DMC could bind in the active sites of wild-type and T249G C2,3O, respectively, with a similar orientation. Thus, the methyl group at position 4 of 4-MC is geo- metrically and chemically equivalent to the methyl group at position 5 of 3,5-MC. Consequently, it could be the reactivity of a substrate that bears a methyl sub- stituent at an equivalent position—i.e. position 4 in 4-MC and position 5 in 3,5-DMC—that controls the regioselectivity of the extradiol cleavage we have observed in the case of 3,5-DMC. Finally, the data reported in Table 1 show that resi- due 249 also strongly influences the k cat values. More- over, the variations observed in the k cat values are significantly larger than those in the K m values; as a consequence, the k cat and k cat ⁄ K m values show similar trends as a function of steric hindrance at the 3¢ sub- site (Table 1 and Fig. 4B). Mutation T249S increases the k cat and k cat ⁄ K m val- ues on 3,5-DMC and 3,6-DMC, with respect to those measured using the wild-type enzyme (Table 1 and Fig. 4B). This effect could depend on the relief of the steric hindrance in the binding of dimethylcatechols at the active site, which, in turn, might favor a more suit- able orientation of the substrate for catalysis. How- ever, mutations T249A and T249G cause instead small variations in k cat and k cat ⁄ K m values (Table 1) despite the fact that their K M values on 3,5-DMC and 3,6- DMC would suggest improved binding with respect to the wild-type enzyme. Moreover, mutation T249S has little or no effect on the k cat values on catechol and 3-MC (Table 1), whereas mutations T249A and T249G reduce by four times the k cat values on catechol and 20 times those measured on 3-MC (Table 1). These latter data are quite intriguing, and they sug- gest that the hydroxyl group of Thr249 could play an unsuspected role in catalysis. Its direct involvement in the catalytic mechanism is unlikely, given the distance (4 A ˚ or greater) between the oxygen atom of the Thr249 side chain and the groups of the substrate directly involved in the reaction. An analysis of the active sites of DHBD structures in the presence and in the absence of substrates and of the structure of P. putida C2,3O suggests a possible hypothesis. In the DHBD–substrate complex, a water molecule is bound between the carboxylate group of Asp243 and the hydroxyl group of the substrate (Supplementary Fig. 3) [34]. A solvent molecule is also present in the active site of each protomer of the P. putida C2,3O structure [33], bound to the hydroxyl group of Thr249 at a position equivalent to that of the water molecule bound to DHBD residue Asp243. Our modeling stud- ies show that binding of the substrate to the active site of P. putida C2,3O does not displace the water mole- cule, which can bridge the hydroxyl group of residue Thr249 and one of the hydroxyl groups of the sub- strate, as in the DHBD–substrate complex (Supple- mentary Fig. 3). Moreover, in the model of C2,3O with 3,6-DMC and 3,5-DMC bound to the active site, the water molecule contacts the methyl group located in the 3¢ site (about 3 A ˚ between the oxygen atom and the carbon atom of the methyl group). As a conse- quence, the possible removal of the water molecule due to mutations T249A and T249G should not make a significant contribution to the decrease in steric hin- drance between the substrate and the active site. This observation and the reduced catalytic efficiency of T249A C2,3O and T249G C2,3O with respect to the wild-type enzyme and to T249S C2,3O would strongly suggest that the bridging solvent molecule plays an important role in catalysis. Experimental procedures Materials and general procedures All chemicals were of the highest grade available and were from Amersham Pharmacia Biotech (Amersham, UK), Promega (Madison, WI, USA), New England Biolabs (Bev- erly, MA, USA), Sigma (St Louis, MO, USA), or Appli- Chem GmbH (Darmstadt, Germany). SDS ⁄ PAGE was carried out according to the method of Laemmli [43]. Protein concentration was determined colori- metrically with the Bradford reagent [44], using bovine serum albumin as a standard. Total iron content and Fe(II) content were determined colorimetrically by complexation with Ferene S [45]. L. Siani et al. Thr249 in catechol-2,3-dioxygenase function FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS 2971 Bacterial strains and plasmids Escherichia coli strain BL21(DE3) and plasmid pET22b(+) were purchased from Novagen (Madison, WI, USA). Plas- mid DNA purifications were performed by using the Qiagen purification kit (Quiagen, Valencia, CA, USA). Bac- terial transformation was carried out according to the method of Sambrook et al. [46]. The construction of recombinant plasmid pET22b(+) DXN ⁄ C2,3O used for the expression of wild-type P. stutzeri C2,3O and the preparation of C2,3O mutants is described elsewhere [18]. Construction of the expression vectors coding for mutant C2,3Os Mutant C2,3Os were produced by the Kunkel method [47], starting from plasmid pET22b(+)DXN ⁄ C2,3O. The sequences of the mutagenic oligonucleotides for T249S, T249A and T249G were 5¢-GTCTTGCCGTGACTGAG GCCGTGG-3¢,5¢-TCTTGCCGTGAGCGAGGCCGTGG C-3¢ and 5¢-GGTCTTGCCGTGGCCGAGGCCGTGG-3¢, respectively. The clones harboring the desired mutations were identified by DNA sequencing and named pET22b(+)DXN ⁄ (T249S)-C2,3O, pET22b(+)DXN ⁄ (T249A)- C2,3O, and pET22b(+)DXN ⁄ (T249G)-C2,3O. The DNA sequences of the three clones were verified by sequencing. Expression and purification of C2,3Os Wild-type and mutant C2,3O were expressed in E. coli strain BL21(DE3), transformed with the appropriate expression vector, purified and analyzed for quality as des- cribed previously [18]. C2,3Os were stored at ) 80 °C under a nitrogen atmosphere. Synthesis and characterization of 3,5-DMC and 3,6-DMC Synthesis of 3,5-DMC and 3,6-DMC was achieved by a modification of the procedure described by Pezzella et al. [48]. o-Iodoxybenzoic acid (IBX) was freshly prepared from 2-iodobenzoic acid as already described [49]. Solid IBX (2.5 equivalents) was added to a solution of 2,4-dimethylphenol or 2,5-dimethylphenol (200 mg) in CHCl 3 ⁄ MeOH 3 : 2 v ⁄ v (40 mL) at ) 25 °C. A yellow–orange color developed and the mixture was stirred for 24 h. Methanolic NaBH 4 (15 mg in 1 mL) was then added at ) 25 °C with vigorous stirring until the color disappeared (usually within 5 min). Excess NaBH 4 was removed by mild acidification with acetic acid (200–500 lL). The mixture was then washed five times with equal volumes of a saturated NaCl solution con- taining 10% sodium dithionite buffered at pH 7.0 with sodium phosphate. Evaporation of the organic layer even- tually yielded 3,5-DMC or 3,6-DMC, which could be separ- ated by preparative TLC (benzene ⁄ ethyl acetate ⁄ acetic acid 1 : 1 : 0.01) on silica. 1 H (13C) NMR spectra of products were recorded at 400.1 (100.6) MHz using a Bruker DRX ) 400 MHz instru- ment fitted with a 5 mm 1 H ⁄ broadband gradient probe with inverse geometry. Impurities were below 1 H-NMR detection limits. Spectral data of 3,5-DMC Pale brown powder. UV(MeOH): k max 281 nm. ESI(–) ⁄ MS m ⁄ z: calculated for C 8 H 9 O 2 [M–H + ] 137.061, determined 137.060. 1 H-NMR (CDCl 3 ), d (p.p.m.) of selected signals: 2.19 (s, 3H, CH 3 ), 2.21 (s, 3H, CH 3 ), 6.51 (s, 2H), 6.53 (s, 2H). Spectral data of 3,6-DMC Pale brown powder. UV(MeOH): k max 280 nm. ESI(–) ⁄ MS m ⁄ z: calculated for C 8 H 9 O 2 [M–H + ] 137.061, determined 137.060. 1 H-NMR (CDCl 3 ), d (p.p.m.) of selected signals: 2.22 (s, 6H, CH 3 ), 6.61 (s, 2H). On the basis of 1 H-NMR and ESI ⁄ MS data, it was possible to confirm the structures of 3,5-DMC and 3,6-DMC. Indeed, in the case of 3,5- DMC, the presence of two aromatic signals at slightly dif- ferent shifts, given the shielding effect of the OH group at position 1, which is positioned para and ortho to hydrogen 3 and hydrogen 5, respectively, is consistent with the struc- ture of catechol. In the case of 3,6-DMC, the 1 H-NMR spectrum features only one aromatic and one methyl group signal, as expected based on the symmetry of the molecule. Also in this case, observed shifts are in agreement with those predicted on the basis of the structure. Determination of regioselectivity on 3-MC and 3,5-DMC 3-MC or 3,5-DMC were added to a solution containing 0.1 mgÆmL )1 of wild-type or mutated C2,3O in 50 mm Tris ⁄ HCl, pH 7.0, at 200 lm final concentration. After 5 min at 25 °C, the reaction was stopped by acidification to pH 4.0 with H 3 PO 4 at 4 ° C, saturated with NaCl, and extracted with ethyl acetate (3 · 100 mL). Evaporation of the organic layer eventually furnished a pale yellow oil that was directly characterized by 1 H-NMR (solvent CDCl 3 ). About 95–100% of 3-MC and 70–80% of 3,5-DMC were converted, yielding 60–80 lmol of products. The determin- ation of the structures of the cleavage products was done only on the basis of the hydrogen atoms bound to sp 2 carbon atoms (see Supplementary Fig. 1 for details). The signals of hydrogen atoms of methyl groups were not con- sidered, as they do not allow us to discriminate distal and Thr249 in catechol-2,3-dioxygenase function L. Siani et al. 2972 FEBS Journal 273 (2006) 2963–2976 ª 2006 The Authors Journal compilation ª 2006 FEBS [...]... superimpose the structure of C2,3O on the crystallographic complexes of DHBD and Thr249 in catechol-2,3-dioxygenase function HPCD with their respective substrates Then, pymol was used to manipulate the complexes in order to manually minimize the number of nonbonded atoms at distances less than the sum of their van der Waals’ radii Catechol and (di)methylcatechols were assumed to bind in the monoanionic,... [16] A model of the DHpCD from P putida F1 (accession code Q51976) was generated using swiss pdb viewer and the Swiss Model server [50] The crystal structure of DHBD from B cepacia LB400 (1kmy [38]) served as the template The alignment between the sequences of the two proteins was extracted from the multiple alignment of ECDs Acknowledgements The authors are indebted to Dr Matthew H Sazinsky, Northwestern... (1983) The catechol dioxygenases Adv Inorg Biochem 5, 167–199 Laemmli U (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4 Nature 227, 680–685 Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding Anal Biochem 72, 248–254 Newman LM & Wackett LP (1995) Purification and. .. catecholate form The phenolate oxygen was placed on the same side as the conserved catalytic residue Tyr255 [18,33] The aromatic ring was rotated using the iron atom as rotation center in order to search for conformations that fit the van der Waals’ volume of the substrate in the active site Coordinates for the (di)methylcatechols were generated by the programs cs chemdraw pro and chem3d pro (Cambridge Soft Corporation,... carbon chain bearing an electron-withdrawing group No signal that can be attributed to an aldehydic group is visible at the expected fields (10.0–9.0 p.p.m.) In (B), the signal of the H4 proton is shifted at lower fields with respect to the signal of the corresponding hydrogen atom in the cleavage product of 3-MC, due to the releasing effect of the methyl groups No signal that can be attributed to H3 and H5... protons of the proximal cleavage product is visible at the expected fields (6.4–5.5 p.p.m.) The signals at 6.523 and 6.500 p.p.m correspond to the two aromatic hydrogen atoms of the substrate (H4 and H6 of 3,5-DMC) Fig S2 Multiple alignment of selected extradiol ring cleavage dioxygenases (ECDs) The loop 246–249 (Pseudomonas putida MT2 numbering) is highlighted Accession numbers of the sequences included in. .. catechol docked into the active site Hydrogen bonds are shown as green dotted lines ˚ Distances are expressed in A W indicates the water molecules bound in the active sites The active site of DHBD from Burkholderia cepacia LB400 (PDB code: 1kmy) is very similar to that of Pseudomonas sp KKS102 enzyme shown in (A) 2976 Table S1 Accession numbers of the sequences included in the alignment shown in Fig S2 This... Corporation, Cambridge, MA, USA) and energy minimized before docking the compounds into the active site The largest spheres that can be fitted to the active site of wild-type and mutant C2,3O were created using the program caver (http://loschmidt.chemi.muni.cz/caver/ index.php) Sequence alignments of ECDs and homology modeling of DHpCD Multiple sequence alignments of ECDs were prepared and analyzed as described... Galli E & Barbieri P (1996) Cloning of the genes for and characterization of the early stages of toluene catabolism in Pseudomonas stutzeri OX1 Appl Environ Microbiol 62, 3704–3711 32 Arenghi FL, Berlanda D, Galli E, Sello G & Barbieri P (2001) Organization and regulation of meta cleavage pathway gene for toluene and o-xylene derivative degradation in Pseudomonas stutzeri OX1 Appl Environ Microbiol 67,... KN & Harayama S (1994) Substrate specificity of catechol 2,3dioxygenase encoded by TOL plasmid pWWO of Pseudomonas putida and its relationship to cell growth J Bacteriol 176, 6074–6081 18 Viggiani A, Siani L, Notomista E, Birolo L, Pucci P & Di Donato A (2004) The role of conserved residues H246, H199 and Y255 in the catalysis of catechol 2,3dioxygenase from Pseudomonas stutzeri OX1 J Biol Chem 279, 48630–48639 . side chain and the groups of the substrate directly involved in the reaction. An analysis of the active sites of DHBD structures in the presence and in the absence of substrates and of the structure. The role of residue Thr249 in modulating the catalytic efficiency and substrate specificity of catechol-2,3-dioxygenase from Pseudomonas stutzeri OX1 Loredana Siani 1, *,. different substrates. The models of the complexes indicate that the orien- tation of the substrate in the active site pocket of the C2,3O is very similar to that observed in the structure of the DHBD and