The salicylate 1,2-dioxygenase as a model for a conventional gentisate 1,2-dioxygenase: crystal structures of the G106A mutant and its adducts with gentisate and salicylate



M. Ferraroni, Dipartimento di Chimica ‘Ugo Schiff’, Università di Firenze, Via della Lastruccia 3, I-50019 Sesto Fiorentino FI, Italy

Fax: +39 055 457 3333

Tel: +39 055 457 3342



The salicylate 1,2-dioxygenase (SDO) from the bacterium Pseudaminobacter salicylatoxidans BN12 is a versatile gentisate 1,2-dioxygenase (GDO) that converts both gentisate (2,5-dihydroxybenzoate) and various monohydroxylated substrates. Several variants of this enzyme were rationally designed based on the previously determined enzyme structure and sequence differences between the SDO and the ‘conventional’ GDO from Corynebacterium glutamicum. This was undertaken in order to define the structural elements that give the SDO its unique ability to dioxygenolytically cleave (substituted) salicylates. SDO variants M103L, G106A, G111A, R113G, S147R and F159Y were constructed and it was found that G106A oxidized only gentisate; 1-hydroxy-2-naphthoate and salicylate were not converted. This indicated that this enzyme variant behaves like previously known ‘conventional’ GDOs. Crystals of the G106A SDO variant and its complexes with salicylate and gentisate were obtained under anaerobic conditions, and the structures were solved and analyzed. The amino acid residue Gly106 is located inside the SDO active site cavity but does not directly interact with the substrates. Crystal structures of G106A SDO complexes with gentisate and salicylate showed a different binding mode for salicylate when compared with the wild-type enzyme. Thus, salicylate coordinated in the G106A variant with the catalytically active Fe(II) ion in an unusual and unproductive manner because of the inability of salicylate to displace a hydrogen bond that was formed between Trp104 and Asp174 in the G106A variant. It is proposed that this type of unproductive substrate binding might generally limit the substrate spectrum of ‘conventional’ GDOs.


Structural data are available in the Protein Data Bank databases under the accession numbers 3NST, 3NWA, 3NVC


gentisate 1,2-dioxygenase


salicylate 1,2-dioxygenase


Aerobic biodegradation of aromatic compounds usually occurs via the formation of a few central intermediates, mainly (substituted) catechol(s), protocatechuate(s) or gentisate(s). Catabolism of these compounds proceeds through the enzymatic opening of their aromatic systems, catalyzed by ring-fission dioxygenases. These enzymes cleave the carbon–carbon bond of vicinal diols such as catechol(s) and (homo)protocatechuates either between two hydroxyl groups [intradiol or ortho-cleaving dioxygenases (Class I)] or adjacent to them [extradiol or meta-cleaving dioxygenases (Class II)] [1]. These reactions result in the formation of either dicarboxylic acids (substituted muconates) in the case of ortho-cleavage pathways or monocarboxylic acids carrying an additional aldehyde function (or rarely keto-group) in the case of extradiol cleavage reactions [2]. Dioxygenases that cleave aromatic compounds with two hydroxyl groups in the para-orientation, as found in (substituted) gentisates or hydroquinone(s), or a single hydroxy-group with a carboxylate residue in ortho-position, like 1-hydroxy-2-naphthoate, salicylate or 5-nitrosalicylate [3-6] have been recently classified as Class III dioxygenases because of their structural and functional differences with respect to classic intradiol and extradiol enzymes [7].

Class III dioxygenases include gentisate dioxygenases, 1-hydroxy-2-naphthoate dioxygenases, homogentisate dioxygenases and 3-hydroxyanthranilate 3,4-dioxygenases. They are bicupins with two germin-like beta-barrel domains and usually contain a single metal active site within the N-terminal cupin domain [8].

An unusual gentisate 1,2-dioxygenase (GDO) has been isolated from the bacterium Pseudaminobacter salicylatoxidans BN12. This enzyme not only catalyzed the ring fission of gentisate, but also oxidized a broad range of monohydroxylated substrates including 1-hydroxy-2-naphthoate and (substituted) salicylate(s) [4, 9]. It was therefore designated as salicylate 1,2-dioxygenase (SDO) because cleavage of aromatic substrates by ‘conventional’ GDOs usually requires additional electron-donating ring substituents.

The crystallographic structures of the SDO from P. salicylatoxidans BN12 [10, 11], a GDO from Escherichia coli O157:H7 [12] and a GDO from Silicibacter (=Ruegeria) pomeroyi [13] have been solved. They are tetrameric proteins that share similar bicupin foldings. The enzymes from P. salicylatoxidans and E. coli contain one Fe(II) center coordinated to three histidines in the N-terminal cupin folding. By contrast, in the GDO from S. (R.) pomeroyi, each subunit contains two ferrous centers which are located in the two homologous N- and C-terminal cupin domains.

Recently, adducts of the SDO with its substrates gentisate, salicylate and 1-hydroxy-2-naphthoate have been prepared under anaerobic conditions. The structures were subsequently solved and analyzed, revealing the mode of substrate binding into the catalytic site and allowing the identification of residues that are important for the catalytic mechanism [7].

The aim of this study was to investigate the molecular basis for the unique ability of SDO to oxidatively cleave a wide range of monohydroxylated aromatic compounds. This was attempted by comparing the wild-type enzyme with a mutant variant (G106A) that demonstrated the properties of a ‘conventional’ GDO and converted only gentisate with a high catalytic efficiency; almost no activities were found with salicylate and 1-hydroxy-2-naphthoate. The kinetic characteristics and crystallographic structures of the G106A SDO mutant and its adducts with gentisate and salicylate were determined in order to shed light on the extraordinary catalytic versatility of the SDO from P. salicylatoxidans and to understand the limitations imposed on ‘conventional GDOs’ not accepting monohydroxylated aromatic compounds as substrates.

Results and Discussion

Identification of amino acid residues that could determine the substrate specificity of the ring-fission dioxygenase from P. salicylatoxidans

The SDO from P. salicylatoxidans has been shown to convert gentisate, 1-hydroxy-2-naphthoate and salicylate [4]. By contrast, ‘conventional’ GDOs do not convert salicylate and demonstrate no or only residual activity with 1-hydroxy-2-naphthoate [14, 15]. The 1-hydroxy-2-naphthoate dioxygenase from Nocardioides sp. KP7 does not convert gentisate and salicylate [16].

Amino acid sequence alignments demonstrated a high degree of similarity between the SDO from P. salicylatoxidans and GDOs from other bacteria, such as some Rhodococcus strains (~ 70% sequence identity) [17, 18], Corynebacterium glutamicum ATCC 13032 (58% sequence identity) [19] or Pseudomonas alcaligenes NCIB 9867 (31% sequence identity) [20]. In addition, significant homology to the 1-hydroxy-2-naphthoate dioxygenases from Nocardioides sp. KP7 and Arthrobacter phenanthrenivorans was observed (29–31% sequence identity) [16, 21].

In the crystal structure of the wild-type enzyme, His119, His121 and His160 coordinated the catalytically active ferrous iron ion [10]. Furthermore, combination of the sequence alignments and structural models showed that the catalytic cavity of the SDO from P. salicylatoxidans shared many amino acid residues with ‘conventional’ GDOs, such as Leu38, Gln108, Ala125, Arg127, His162, Trp172, Asp174 and Leu176 (Fig. 1).

Figure 1.

Amino acid sequence alignments of salicylate dioxygenase and other known gentisate dioxygenases: Rhodococcus sp. NCIMB 12038 GDO (ADT78164.1), Rhodococcus jostii RHA1 GDO (ABG93677.1), Rhodococcus opacus GDO (ABH01038.1), Pseudaminobacter salicylatoxidans SDO (AAQ91293.1), Corynebacterium glutamicum ATCC 13032 GDO (NP_602217.1), Pseudomonas alcaligenes NCIB 9867 (xlnE) GDO (AAD49427.1), Xanthobacter polyaromaticivorans GDO (BAC98955.1), Escherichia coli O157:H7 GDO (BAB36453.1), Sphingomonas sp. RW5 GDO (CAA12267.1), Ralstonia (Pseudomonas) sp. U2 GDO (AAD12619.1), Polaromonas naphthalenivorans CJ2 GDO (AAZ93402.1), Polaromonas naphthalenivorans CJ2 (nagI′) (AAZ93397.1), Polaromonas naphthalenivorans CJ2 (nagI3) GDO (ABM37760.1), Pseudomonas alcaligenes NCIB 9867 GDO (hbzE) (ABD64513.1), Silicibacter (Ruegeria) pomeroyi DSS-3 GDO (AAV97252.1), Haloferax sp. D1227 GDO (AAQ62856.1), Haloarcula sp. D1 GDO (AAQ79814.1). Alignment positions identical in all sequences are marked with a blue background. Mutated amino acid residues are within a red box. Codes in parentheses refer to the NCBI protein accession.

Sequence alignments showed preferential clustering of the SDO from the Gram-negative α-proteobacterium P. salicylatoxidans with GDOs from Gram-positive bacteria, such as Rhodococcus strains or C. glutamicum. This was also indicated by the ‘conservation’ of certain amino acid residues in the catalytic cavity among these enzymes, for example, Met46, Glu82, Trp104 and Phe189.

Several SDO variants were rationally designed in order to convert the ‘broad substrate-range’ SDO from P. salicylatoxidans into a ‘conventional’ GDO. This was initially based on sequence differences between the enzyme from P. salicylatoxidans and the ‘conventional’ GDO from C. glutamicum. The M103L, G106A, G111A, R113G, S147R and F159Y SDO mutants were produced and the respective recombinant enzymes analyzed in cell extracts at fixed substrate concentrations (0.5 mm each) for the conversion of gentisate, 1-hydroxy-2-naphthoate and salicylate.

No significant differences from the wild-type enzyme were found for the variants M103L, G111A, R113G, S147R and Fp159Y. In light of the structural information obtained to date on the enzyme, we can explain the kinetic properties of these variants because the mutated residues are quite distant from the active site. However, the mutation to alanine of Gly106, a residue located in the active site, although not directly interacting with the substrate, generated a SDO variant that converted gentisate, but completely lost the ability to convert 1-hydroxy-2-naphthoate and showed no detectable activity for salicylate.

The G106A variant was purified and analyzed further. It cleaved gentisate with a slightly decreased activity (Vmax) of 110 U·mg−1 compared with wild-type SDO (180 U·mg−1), but showed a higher affinity to this substrate than the wild-type SDO (KM: Gly106Ala, 90 μm; SDOwt, 170 μm). The purified enzyme did not convert salicylate or 1-hydroxy-2-naphthoate. 5-Fluorosalicylate and 5-aminosalicylate were oxidized by the variant G106A with rather low Vmax values of 0.23 and 0.2 U·mg−1, respectively (Table 1). Comparative values for the wild-type enzyme were ~ 3 and 4 U·mg−1.

Table 1. Kinetic parameters and substrate specificities of SDO wild-type and G106A mutant variant. n.d., no detectable activity in the spectrophotometric enzyme assays
KMm)kcat (s−1)kcat/KM (s−1·m−1) × 103KMm)kcat (s−1)kcat/KM (s−1·m−1) × 103
Gentisate167133800 (100%)11090.30818.68 (100%)
1H2NC14024168 (21%)n.d.n.d.n.d.
Salicylate170.950 (6.2%)n.d.n.d.n.d.
5-Aminosalicylate2162.813 (1.6%)5240.20.3 (< 1%)
5-Methylsalicylate7882.12.6 (< 1%)n.d.n.d.n.d.
5-Fluorosalicylate472.450 (6.3%)4380.20.3 (< 1%)

The pronounced effect of the G106A mutation on the substrate specificity was rather surprising, because structural analysis of the wild-type enzyme and the substrate complexes suggested no direct interaction of the relevant amino acid residue with the aromatic substrates. Therefore, the crystal structures of the SDO G106A mutant and its anaerobic adducts with gentisate and salicylate were solved to reveal the molecular factors responsible for the differences in substrate specificity observed with respect to the wild-type enzyme.

The FoFc electron density maps of the G106A mutant clearly showed the electron density corresponding to an alanine side chain substituting the glycine residue in position 106. From a superposition of the model with the coordinates of wild-type SDO, no appreciable differences in the overall 3D structure were ascertained. The structure of the enzyme active site was not modified by the point mutation, apart from the region close to the mutated residue (Fig. 2). In fact, the presence of Ala106 induces a shift in the Trp104 side chain, which assumed the conformation usually seen when a substrate is bound to the active site of the wild-type enzyme [7]. Asp174 was also displaced from its resting position forming a hydrogen bond with the side chain of Trp104 (Asp174 OD1–Trp104 NE1 3.2 Å). Furthermore, a glycerol molecule, utilized for crystal protection at cryoscopic temperatures (see 'Material and methods'), is found bound to the active metal ion.

Figure 2.

Stereoview of the least-square superposition of the active site cavity of G106A mutant (pale blue) and wild-type SDO (atoms colored as element type, i.e. carbon, in gray; oxygen, red; nitrogen, blue; sulfur, yellow; iron, brown) structures.

The crystallographic structure of the G106A SDO mutant adduct with gentisate presented the substrate molecule coordinated to the iron(II) ion in a bidentate mode through the 2-hydroxy group and one of the carboxyl oxygens, analogous to the model obtained for the native SDO in complex with gentisate, salicylate and 1-hydroxy-2-naphthoate and as suggested by spectroscopic studies and docking calculations on GDOs from other sources (Fig. 3) [7, 11, 22, 23]. Superposition of the G106A mutant and the wild-type SDO structures in complex with gentisate exhibited only minor differences between both active sites (Fig. 3). The carboxyl oxygen, which is coordinated to the iron ion, also forms a hydrogen bond with nitrogen NE of Arg83. The other carboxyl oxygen of the substrate, which is not coordinated to the iron, interacts with NE2 of His162, OE1 of Gln 208 and NH2 of Arg127. The aromatic ring of the substrate also interacts with the side chains of residues Leu38, Ile178, Leu176, Met46 and Trp104. The enzyme–gentisate complex model showed that the 5-phenolate group of gentisate is hydrogen bonded to OD2 of Asp174 and NE1 of Trp104 at a distance of 2.58 and 2.85 Å, respectively. The side chains of Arg127, His162 and Arg83 undergo significant displacement from their original positions, as seen in the substrate adduct structures of wild-type SDO. Accordingly, the same conformational changes caused by the movement of Arg83 towards the gentisate molecule were observed: closure of the loop corresponding to residues 75–85, which restricts access to the cavity, follwed by stabilization of loop 192–198 and the N-terminal region (residues 5–14). These results demonstrated that replacement of a glycine residue by a slightly larger alanine residue did not alter the protein structure significantly, especially as this glycine residue is not involved in substrate binding or catalysis. This was also in good agreement with the observation that the wild-type enzyme and the G106A variant oxidized gentisate with rather similar catalytic constants (Table 1).

Figure 3.

Stereoview of the least-square superposition of the active site cavity of G106A mutant (blue) and wild-type SDO (atoms colored as element type) structures both complexed with gentisate.

Subsequently, the SDO G106A crystals were soaked with salicylate under anaerobic conditions in order to obtain the structure of the corresponding complex. In the structure of the G106A adduct, the salicylate moiety is bound in a bidentate way to the ferrous ion but only with its carboxyl group (Fig. 4). Thus, the salicylate molecule is bound by this enzyme variant in a different way compared with gentisate and the substrate complexes of the wild-type enzyme (Figs 4 and 5). As a consequence, the C–C cleavage between the aromatic C atoms carrying the carboxyl group and the hydroxyl group cannot be catalyzed by G106A. The carboxyl group substituent of salicylate is not coplanar with the aromatic ring, as usually found in all other SDO substrate complexes, probably because of active site constraints. The 2-hydroxyl group forms hydrogen bonds with Asp174, Arg127 and one solvent molecule. The structure also revealed that the hydrogen bond between NE1 of Trp104 and OD2 of Asp174 observed in the uncomplexed G106A structure is maintained. As a consequence of the missing 5-hydroxyl substituent in salicylate, Asp174 protrudes into the cavity possibly affecting the binding of salicylate in two ways: (a) the size reduction of the active site cavity that forces the salicylate in close proximity to the iron ion; (b) the formation of a hydrogen bond between OD2 of Asp174 and the 2-hydroxyl group of salicylate (2.3 Å). In addition, the short contact (2.5 Å) between the OD1 of Asp174 and the C3 carbon of salicylate has the right geometry to be considered as a C–H–O hydrogen bond [24, 25]. All these effects cause or at least support unproductive binding of salicylate to the ferrous iron ion.

Figure 4.

Stereoview of the least-square superposition of the active site cavity of G106A mutant (atoms colored as element type) and wild-type SDO (pink) structures both complexed with salicylate, showing the different positioning of salicylate and the differential displacements of Asp174 and Arg127.

Figure 5.

Stereoview of the least-square superposition of the active site cavity of G106A mutant complexes with salicylate (brown) and gentisate (atoms colored as element type). A difference Fourier electron-density map calculated with coefficients (Fobs, gent-Fobs, sal) is superimposed on the adduct coordinates.

Furthermore, structural analysis demonstrated that after binding of salicylate to the G106A variant, typical conformational changes (e.g. the movement of His162, Arg127, Arg83 and closure of loop 75–85) that occur after the binding of salicylate or gentisate to the wild-type enzyme (or also gentisate to the G106A variant) were not observed.

The different mode of salicylate binding between the wild-type enzyme and the G106A variant could also be deduced from spectrophotometric enzyme assays. Kinetic analysis of the wild-type enzyme demonstrates that salicylate was converted with a rather low Vmax of ~ 1.1 U·mg−1 protein (compared with a Vmax of 178 U·mg−1 with gentisate as the substrate), but that salicylate was bound very tightly to the enzyme (KM = 17 μm) [22]. The G106A variant had completely lost the ability to convert salicylate, but the binding of salicylate could still be demonstrated in inhibition experiments when the conversion of gentisate (0.02–0.5 mm) was assayed in the presence of increasing amounts of salicylate (1–5 mm). These experiments demonstrated that ~ 1 mm of salicylate was necessary to reduce the gentisate conversion rate to half that observed in the absence of the inhibitor. Similar inhibition constants have previously been described for ‘conventional’ GDOs for the inhibition of the gentisate turnover by salicylate [23]. This might indicate that salicylate is also bound in these enzymes in a similar nonproductive way as described here for the G106A variant.

The hydrogen bond between Trp104 and Asp174 and the protrusion of Asp174 into the active site cavity might also explain the observed lack of activity of the G106A mutant with the substrates 1-hydroxy-2-naphthoate (or 5-methylsalicylate) because these substrates are also lacking a hydrophilic substituent in the 5-position and cannot bind in the mutant active site in a productive manner because of their inability to form a hydrogen bond with Trp104 and Asp174. The alignment of various GDO sequences revealed that in ‘conventional’ GDOs from Gram-positive bacteria (e.g. R. opacus, C. glutamicum or Streptomyces sp.) there is a conserved coupling of a tryptophan residue in position 104 (in the numbering of the SDO sequence) with an alanine in position 106 (Fig. 1). Interestingly, the wild-type SDO (with a Trp–Gly pair) seems to be the only exception to this rule. Therefore, a single point mutation of a residue not directly interacting with the substrates, but able to displace a key residue (Trp104) in the active site of SDO, converts this enzyme with its unusual and broad substrate range into a ‘conventional’ GDO with a rather restricted substrate specificity.

In the remaining GDO sequences from Gram-negative bacteria and even Archaea thus far determined, residue 104 is a tyrosine and residue 106 is a glycine. A tyrosine in position 104 could have the same function as Trp104, forming a hydrogen bond with the 5-phenolate of gentisate. Nevertheless, other mutations are present in the active sites: Ala85 in SDO is a Val in this type of GDOs, whereas Met46 can be a Val or a Ile.

This forms part of a series of structural studies intended to investigate the unique catalytic properties of SDO [7, 10, 22]. Previously, we analyzed the enzyme variant A85H, which oxidizes 1-hydroxy-2-naphthoate, but does not convert gentisate and demonstrates only residual activities for salicylate. The crystal structure of the A85H variant shows that His85 points into the active site and forms several hydrogen bonds with residues Trp104, Gln108 and Asp174, thus impeding the correct binding of gentisate. Furthermore, the loop containing this residue shifts away from the active site and thus enlarges the substrate-binding cavity, favoring the binding of bulkier substrates such as 1-hydroxy-2-naphthoate [22].

Comparison of these results with those obtained here suggests that changes in the positioning of the residues Trp104 and Asp174 and modifications in the network of hydrogen bonds connecting these residues are of pivotal importance for the discrimination of substrates, either carrying or missing a hydroxyl group in the position corresponding to the 5-hydroxy group of gentisate.

The mutation of other residues, such as Met46 and Leu38, resulted in variants which preferentially oxidized 1-hydroxy-2-naphthoate but still demonstrated residual activities with gentisate. This is reflected in the structural model by the observation that Met46 and Leu38 undergo unspecific interactions with the aromatic rings of the substrates. Therefore, changes in these residues mainly result in enzyme variants that demonstrate an altered discrimination between mono- and biaromatic substrates.

Moreover, Met46 and Leu38 are localized in the N-terminal region of each subunit which penetrates another subunit and covers the active site of the latter. Presumably this region is involved in stabilization of the tetrameric holoenzyme. Hence, it has been proposed that their alteration can influence enzyme stability and also cause an increased solvent accessibility in the active site. This could result in the oxidation and loss of the catalytic Fe2+ ion and a reduced enzyme activity.


The SDO variant G106A was unable to convert salicylate and 1-hydroxy-2-naphthoate, but maintained the ability to cleave gentisate. Nevertheless, salicylate could still bind to the G106A variant and act as an inhibitor of gentisate conversion. This suggested that the unique characteristic of the SDO from P. salicylatoxidans might be caused by the ability of this dioxygenase to bind salicylate to the catalytic active Fe(II) ion in the same bidentate fashion as gentisate.

The crystal structures of the adducts clearly demonstrated the different binding modes of salicylate and gentisate to the enzyme variant G106A. The substitution of a glycine with an alanine residue, although occurring far from the substrate, induces the salicylate ion to bind to the iron center only with a carboxylate group with a 180° rotated ring with respect to gentisate (or salicylate in the wild-type enzyme). We propose that, as in the enzyme variant G106A, the GDOs from other organisms that do not convert salicylate also bind salicylate in this unproductive conformation.

Moreover, from our structural investigation, we have learned that in order to modulate the substrate specificity of GDOs and related enzymes for future engineering designs, the upper part of the cavity (i.e. the region including Asp174, Glu108, Trp104 and surrounding residues) should be preferentially modified.

Material and methods

Expression and purification of the enzyme variants and measurement of the enzyme activities

These steps were performed as previously described [22].


The enzyme was crystallized at 277K using the sitting-drop vapor diffusion method. The crystallization conditions were the same as those used for crystallization of the wild-type enzyme in complex with substrates [7]: 8% poly(ethylene glycol)10000 and 0.1 m Tris/HCl pH 8.0. A volume of 0.2 μL of a 0.1 m calcium chloride solution was added directly to the drops (formed by 1.0 μL of protein and 0.8 μL of crystallization solution). The crystal quality was improved using the seeding technique: drops were equilibrated against 100 μL of crystallization solution for 24 h and then crystal nuclei were introduced by adding 0.3 μL of a seeding solution diluted 1 : 100.

Crystals of this form are orthorhombic and belong to the space group I222 with unit cell dimensions a = 73 Å, b = 86 Å, c = 168 Å and one molecule per asymmetric unit (VM = 3.28 Å3·Da−1 and solvent content 62.2%).

Isolation of substrates complexes

Substrates complexes of the G106A mutant enzyme were prepared under anaerobic conditions using a glove box (Unilab, MBraun, Garching, Germany). Substrate solutions were prepared dissolving the substrates in the crystallization solution. All of the solutions were deoxygenated by bubbling nitrogen for a few minutes and then equilibrated inside the glove box for ~ 2 h. Crystals were transferred in the glovebox and then washed with the crystallization solution. Adducts were prepared by soaking the crystals for 2 h in a crystallization solution containing 20 mm of salicylate or gentisate. The crystals were then flash-frozen inside the glovebox in a solution obtained by adding 30% (v/v) glycerol to the mother liquor as cryoprotectant. Once transferred outside the glovebox, crystals were kept under liquid nitrogen. The presence of oxygen inside the glovebox was monitored: the amount of oxygen varied between 2 and 30 p.p.m. during the experiments.

Data collection and refinement

Data collection for the crystals of the G106A mutant and its substrate complexes with gentisate and salicylate were performed at the EMBL beamline X13 (DESY, Hamburg, Germany) using a MARCCD 165 mm detector at a wavelength of 0.8123 Å and at a temperature of 100 K.

Data were processed using mosflm [26] and scaled using scala from the ccp4 program suite [27]. The statistics for the data collections are summarized in Table 2.

Table 2. Summary of data collection and atomic model refinement statistics. Values in parentheses are for the highest resolution shell
Data collection
Space GroupI222I222I222
Unit cell (a, b, c) (Å)a = 76.15, b = 86.13, c = 166.94a = 76.50, b = 86.89, c = 166.83a = 74.27, b = 86.98, c = 167.63
Limiting resolution (Å)2.40 (2.53–2.40)2.0 (2.11–2.0)2.45 (2.58–2.45)
Unique reflections21 100 (3080)37 295 (5337)20 120 (2913)
R sym 0.107 (0.50)0.065 (0.414)0.079 (0.477)
Multiplicity4.2 (4.2)4.2 (3.5)3.4 (2.9)
Completeness overall (%)98.4 (99.3)98.7 (98.0)99.1 (99.5)
<I/σ(I)>12.4 (2.8)15.0 (1.8)11.7 (1.6)
Resolution range (Å)30.0–2.430.0–2.030.0–2.45
Unique reflections, working/free20,011/107337,276/187220,084/1022
Rfactor (%)15.1615.8319.33
Rfree (%)21.9120.2127.24
Nonhydrogen atoms313634282945
Water molecules343506173
R.m.s.d. bonds (Å)0.0160.0170.018
R.m.s.d. angles (°)1.7221.5631.877
Average B-factor
All atoms35.3031.8544.73
Favored region333352328
Allowed region10815
Outlier region212
PDB ID code 3NST 3NW4 3NVC

FoFc maps were calculated using the refined coordinates of salicylate 1,2-dioxygenase without all the heteroatoms and after 10 cycles of refinement, indicated a clear density for the side chain of the mutated residue. Furthermore, Fo − Fc electron density maps also revealed a clear density on the iron ion for the two ligands that were introduced at the last stages of the refinement and refined at unitary occupancy.

The models were refined using refmac5 [28]. Manual rebuilding of the models was performed using coot [29]. Solvent molecules were introduced automatically using arp [30]. Data refinement statistics are summarized in Table 2. The stereochemistry of the final model was analyzed with rampage [29].

The final model is composed of residues 5–367 of subunit A, 506 water molecules, one Fe(II) ions for the complex with gentisate and of residues 15–193, 198–367 of subunit A, one Fe(II) ion, 343 and 173 for the mutant and the mutant complexed with gentisate, respectively.


We acknowledge the Italian MIUR PRIN 2009 funding. We also acknowledge the ‘European Community – Access to Research Infrastructure Action of the Improving Human Potential Programme to the EMBL Hamburg Outstation, contract number: HPRI-CT-1999-00017’.