The structure of Trametes versicolor glutathione transferase Omega 3S bound to its conjugation product glutathionyl‐phenethylthiocarbamate reveals plasticity of its active site

Abstract

Trametes versicolor glutathione transferase Omega 3S (TvGSTO3S) catalyzes the conjugation of isothiocyanates (ITC) with glutathione (GSH). Previously, this isoform was investigated in depth both biochemically and structurally. Structural analysis of complexes revealed the presence of a GSH binding site (G site) and a deep hydrophobic binding site (H site) able to bind plant polyphenols. In the present study, crystals of apo TvGSTO3S were soaked with glutathionyl‐phenethylthiocarbamate, the product of the reaction between GSH and phenethyl isothiocyanate (PEITC). On the basis of this crystal structure, we show that the phenethyl moiety binds in a new site at loop β₂‐α₂ while the glutathionyl part exhibits a particular conformation that occupies both the G site and the entrance to the H site. This binding mode is allowed by a conformational change of the loop β₂‐α₂ at the enzyme active site. It forms a hydrophobic slit that stabilizes the phenethyl group at a distinct site from the previously described H site. Structural comparison of TvGSTO3S with drosophila DmGSTD2 suggests that this flexible loop could be the region that binds PEITC for both isoforms. These structural features are discussed in a catalytic context.

Short abstract

PDB Code(s): 6HPE;

Abbreviations

AbGSTO1

Alternaria brassicicola glutathione transferase Omega 1

G site

glutathione binding site

GS‐

glutathionyl‐

GSH

glutathione

GS‐PEITC

glutathionyl‐phenethylthiocarbamate

GST

glutathione transferase

GSTO

glutathione transferase Omega

H site

hydrophobic binding site

ITC

isothiocyanate

PEITC

phenethyl isothiocyanate

TvGSTO

Trametes versicolor glutathione transferase Omega

Introduction

Glutathione transferases (GSTs) are ubiquitous phase‐II detoxification enzymes, which catalyze transfer of glutathione (GSH) to various xenobiotics and also exhibit noncatalytic ligandin functions.1 They form a superfamily divided into several classes. On a structural point of view, most GSTs are homodimeric. Each monomer is composed of an N‐terminal thioredoxin domain, which bears the GSH binding site (G site) and a C‐terminal all‐helical domain, which bears a hydrophobic binding site (H site).2 In wood‐decaying fungi, some GST classes have expanded by multiplying the genes encoding these enzymes.3 These classes are the GSTs Ure2p,4 the GSTs FuA,5 and the GSTs Omega (GSTOs).6 In a recent study, we showed that GSTOs from the white‐rot fungus Trametes versicolor (TvGSTOs) could fix various wood polyphenols without any chemical reaction.7, 8 This suggests transport or sequestration roles, similarly to plant GSTs.9

Molecules from the isothiocyanate (ITC) family are found among the substrates that are accepted by GSTs and in particular by some TvGSTOs.1, 7 ITCs are formed during the glucosinolate metabolism in plants from the Brassicaceae family.10 These sulfur‐containing molecules show promising antibacterial11 and antifungal properties.12 The well‐studied phenethyl isothiocyanate (PEITC) also displays interesting anticancer properties.13 Transcriptomic analyses suggested the presence of high levels of GSTs after exposure of the ascomycete fungus Alternaria brassicicola to ITC.14 The corresponding proteins belong to GSTs FuA, GTT, GSTO, and Ure2p classes. Interestingly, mutants that lack GSTs AbGSTO1 or AbUre2pB1 were found to be hypersensitive to ITC and showed an impaired aggressiveness against Brassica oleracea.15 GSTs reversibly conjugate ITCs with GSH to yield the corresponding dithiocarbamates (Fig. 1).16, 17 Up to the present study, no structure of a complex between a fungal GST and an ITC derivative was available in the Protein Data Bank. Crystal structures of human GSTA1 and GSTP1 in complex with the reaction product of GSH and PEITC (glutathionyl‐phenethylthiocarbamate, GS‐PEITC) allow determining the product binding site for these isoforms.18 While the glutathionyl group is stabilized at the G site mainly constituted by residues of the thioredoxin domain, the PEITC group is stabilized at the H site formed by the all‐helical domain. Different conformations of the PEITC moiety were identified in GSTA1 and GSTP1. These human GST isoforms possess a catalytic tyrosine, but fungal GSTOs exhibit either a catalytic serine or a catalytic cysteine.19 Given the many differences between these GSTs, a comparative analysis between the structures of their complexes with GS‐PEITC was needed considering the numerous biological properties of ITCs.

Figure 1.

Conjugation of phenethyl isothiocyanate with glutathione catalyzed by GST.

Trametes versicolor is a basidiomycete, which contains 16 putative genes of the GSTO class. One of them, TvGSTO3S, has a catalytic serine and conjugates GSH with PEITC.7 This reaction is inhibited by plant polyphenols that bind TvGSTO3S active site. In the present study, we determined the structure of TvGSTO3S in complex with its conjugation product GS‐PEITC. Our results provide structural insights into the binding properties of TvGSTO3S active site toward GS‐PEITC. They could differ from the ones of human GSTA1 and GSTP1. Indeed, the PEITC moiety does not bind at the H site like in human isoforms but rather in the vicinity of the loop β₂‐α₂ at the thioredoxin domain.

Results and Discussion

TvGSTO3S:GS‐PEITC overall structure

The crystal structure of TvGSTO3S in complex with the glutathionyl adduct of phenethyl isothiocyanate (GS‐PEITC) was determined by molecular replacement using the apo structure (PDB code 6F43) as a search model and refined to 1.45 Å resolution (PDB code 6HPE, Table 1). The protein was crystallized in the P2₁2₁2₁ space group (Fig. S1) with two subunits that constitute the canonical GST dimer in the asymmetric unit. Due to insufficient electron density, the residues 1–3 from both monomers A and B were not included in the final model. Electron density peaks were observed in the active sites and allowed modeling of the GS‐PEITC molecules in both monomers. An additional electron density peak corresponding to an unidentified ligand was detected at the bottom of the active site. The signal of this small molecule was already observed in the structures of TvGSTO3S apo (PDB code 6F43) and in complex with GSH (PDB code 6F4B) but not in the structures of the protein bound to glutathionyl derivatives that most likely displace this unidentified molecule, for example, in TvGSTO3S:GS‐phenylacetophenone (GS‐PAP, PDB code 6F51) and in TvGSTO3S:GS‐hexane (PDB code 6F4K, Table S1).7

Table 1.

Data Collection and Refinement Statistics

TvGSTO3S:GS‐PEITC
Diffraction data
Diffraction source	ID30B, ESRF
Detector	PILATUS 6M‐F
Wavelength (Å)	0.97625
Unit‐cell parameters
a, b, c (Å)	50.3, 104.0, 106.5
α = β = γ (°)	90
Space group	P212121
Resolution range (Å)	46.74–1.45 (1.49–1.45)
Total No. of reflections	345,308 (22,987)
No. of unique reflections	96,095 (7033)
Average redundancy	3.6 (3.3)
Mean I/σ(I)	14.3 (1.6)
Completeness (%)	96.5 (96.3)
R merge	0.04 (0.62)
R meas	0.05 (0.73)
CC1/2	1.00 (0.55)
Refinement
Resolution range (Å)	46.74–1.45 (1.47–1.45)
R work/R free	0.16/0.18 (0.25/0.36)
No. of protein atoms	7554
No. of ligand atoms	110
No. of waters	431
Average B factor (Å2)	25.6
Model quality
RMSZ bond lengths	0.56
RMSZ bond angles	0.73
Ramachandran favored (%)	98.8
Ramachandran outliers (%)	0.4
Molprobity rotamer outliers (%)	1.0
Molprobity clashscore	1.55
Molprobity score	0.91
PDB code	6HPE

R _merge = ∑_hkl∑_i|I_i(hkl) − 〈I(hkl)〉|/∑_hkl∑_iI_i(hkl).

R _meas = ∑_hkl{N(hkl)/[N(hkl) − 1]}^1/2 ∑_i|I_i(hkl) − 〈I(hkl)〉|/∑_hkl∑_iI_i(hkl).

R _work = ∑_hkl||F_obs| − |F_calc||/∑_hkl|F_obs|.

CC_1/2 is the correlation coefficient of the mean intensities between two random half‐sets of data.20 Five percent of reflections were selected for R _free calculation. RMSZ: root‐mean‐square Z‐score.21 The Molprobity clashscore is the number of serious clashes per 1000 atoms.22 The Molprobity score is a log‐weighted combination of the clashscore, percentage Ramachandran not favored, and percentage bad side chain rotamers.22 Values in parentheses are for highest resolution shell.

TvGSTO3S:GS‐PEITC structure shares the same overall features as the previously solved apo TvGSTO3S structure.7 It exhibits the GST canonical dimer where the N‐terminal thioredoxin domain (β₁α₁β₂α₂β₃β₄α₃) of one monomer cross‐interacts with the C‐terminal all‐helical domain (α₄α₅α₆α₆′α₇α₈α₉) of the second one and vice versa [Fig. 2(a)]. TvGSTO3S displays some of the structural specificities of the Omega class (GSTO) also present in human24 and insects25 GSTO structures. These features include an open dimer and an additional helix α₉ near the active site. However, TvGSTO3S like its five TvGSTOS isoforms (Fig. S2) has singular properties for a GSTO because it lacks the extended N‐terminal tail and the catalytic cysteine at the N‐terminal end of helix α₁, replaced by the serine residue S15. We showed recently that this serine interacted with the S_γ atom of GSH in the TvGSTO3S:GSH structure (PDB code 6F4B) with a probable catalytic role26 in GSH‐conjugation toward 1‐chloro‐2,4 dinitrobenzene (CDNB), PEITC, and benzyl‐isothiocyanate (BITC, Table S2).7

Figure 2.

TvGSTO3S:GS‐PEITC crystal structure. (a) TvGSTO3S:GS‐PEITC overall structure. The dimer is shown in cartoon representation. Thioredoxin domain and all‐helical domain are colored cyan and pale yellow, respectively. GS‐PEITC is represented as sticks and spheres and colored by atom type (green, C; red, O; blue, N; and yellow, S). (b) 2mFo‐DFc composite omit map contoured at 1.5 σ around GS‐PEITC in TvGSTO3S active site calculated with PHENIX.23 (c) Structural superimposition of TvGSTO3S:GS‐PEITC monomer (cyan and pale yellow) and TvGSTO3S apo monomer (light gray) highlighting loop β₂‐α₂ displacement upon GS:PEITC binding.

TvGSTO3S has a new binding site adapted for PEITC moiety binding

Soaking experiments of apo TvGSTO3S crystals with GS‐PEITC successfully resulted in ligand binding at TvGSTO3S active site (Fig. 2) with conformational changes with respect to the apo structure (RMSD of 0.615 Å for all C_α atoms). Previously, we showed that TvGSTO3S has a GSH binding site (G site) formed at the thioredoxin domain and a large and deep hydrophobic site (H site) found at the all‐helical domain [Fig. 3(a)].7 This H site has strong affinity for the hydrophobic moieties of glutathionyl derivatives (GS‐phenylacetophenone, GS‐hexane) and polyphenols (hydroxybenzophenones and flavonoids). Polyphenols efficiently inhibit the transfer of GSH to PEITC in the micromolar range.7 The crystal structure of the complex between TvGSTO3S and the product glutathionyl‐dinitrobenzene was previously solved (PDB code 6F4F). This product was bound at the active site with stabilization of the dinitrobenzyl moiety at the entrance to the H site.

Figure 3.

Structural comparison of TvGSTO3S complexes. (a) TvGSTO3S active site bound to GSH colored in green and inhibitor 2,4‐dihydroxybenzophenone colored in orange (K _i of 2 μM, PDB code 6F66).7 Active site residues are labeled and represented as sticks. Polar contacts are materialized as dashed lines. (b) TvGSTO3S active site bound to its product GS‐PEITC (this work). The PEITC moiety occupies a new site while the (glycyl) carboxyl of GSH is flipped toward the H site. (c) Structural superimposition of TvGSTO3S:GS‐PEITC monomer (cyan and pale yellow) and TvGSTO3S:2,4‐dihydroxybenzophenone monomer (light gray). GSH, GS‐PEITC, and 2,4‐dihydroxybenzophenone are colored, respectively, in yellow, green, and orange.

In the present structure, GS‐PEITC occupies both the G site, the entrance to the H site, and a new site located in the region between strand β₂ and helix α₂. This loop β₂‐α₂ (D36‐D45) adopts a new conformation in the ligand‐bound structure and probably moves upon GS‐PEITC binding [Fig. 2(c)]. Displacement of this loop (~3 Å movement at residue R41) opens a hydrophobic slit (side chains from residues V37, I39, M43, P44, and F47), which welcomes the phenethyl moiety of GS‐PEITC with an additional polar contact between the carbonyl main chain of K54 and the PEITC amide group. These hydrophobic residues are well conserved in all TvGSTO isoforms and also in the human isoform HsGSTO1 (Fig. S2) even if the GSH‐transferase activity toward PEITC was not investigated yet for this isoform. HsGSTO1 has little transferase activity with most GST substrates such as CDNB.24

The peculiar position of the phenethyl moiety hinders the canonical binding of GSH in the G site. The glycyl moiety has a different position from that observed in the previously solved TvGSTO3S:GSH structure (Fig. 3). In TvGSTO3S:GS‐PEITC structure, the glutathionyl C‐terminal end is located at the entrance to the H site whereas in the TvGSTO3S:GSH structure this group is in the G site and is stabilized by the side chain of K55 from the loop β₂‐α₂. These two glutathionyl conformations arise from different values adopted by the main torsional angles of the cysteinyl residue (φ and ψ angles of +79° and −122° for GS‐PEITC in the present TvGSTO3S:GS‐PEITC complex; −103° and +127° for GSH in TvGSTO3S:GSH complex, respectively). In the TvGSTO3S:GSH structure, the side chain of the catalytic serine S15 is hydrogen‐bonded to the S_γ atom of GSH.7 In the TvGSTO3S:GS‐PEITC structure, this interaction no longer exists since S15 forms a H‐bond that stabilizes the main chain carbonyl group of the glutathionyl cysteine residue. The glutathionyl S_γ atom is found at a distance of 5.6 Å from the S15 hydroxyl group. This distance questions a possible conformational change of GSH upon reaction with the substrate PEITC and release of the product GS‐PEITC. Mainly, rotations around the main torsional angles of the cysteinyl moiety are sterically possible and could lead to a conformational change. This suggests that some flexibility is permitted to TvGSTO3S active site.

Comparison of TvGSTO3S:GS‐PEITC structure with similar complexes

The crystal structures of complexes between GS‐PEITC and the human isoforms GSTA1 (PDB code 5JCU) and GSTP1 (PDB code 5JCW) were previously determined.18 We compared them to the structure of TvGSTO3S:GS‐PEITC (Fig. 4). Binding of GS‐PEITC in GSTA1 and GSTP1 occurs in the active site with the glutathionyl moiety at the G site similarly to free GSH in both isoforms (PDB codes 1PKW and 5GSS for GSTA1:GSH and GSTP1:GSH, respectively). In each complex with GS‐PEITC, the PEITC moiety is stabilized at the H site (all‐helical domain), located between helices α₄ and α₉ in GSTA1 and along helix α₄, and oriented toward the solvent in GSTP1.18 On the contrary, TvGSTO3S rather interacts with the PEITC group at the thioredoxin domain, via the new position of the loop β₂‐α₂. For the human isoforms, no such structural changes were observed after comparison with the apo or GSH‐bound structures.18 Our results indicate that the fungal Omega isoform TvGSTO3S has a new PEITC moiety binding site and we observed a particular conformation adopted by the glutathionyl moiety of GS‐PEITC, which suggests differences in binding properties when compared with the human isoforms.

Figure 4.

Structural comparison of human and T. versicolor GST:GS‐PEITC complexes. Subunits (left, TvGSTO3S, PDB code 6HPE; center, GSTA1, PDB code 5JCU; right, GSTP1, PDB code 5JCW) are represented in cartoon and colored cyan (thioredoxin domain) and pale yellow (all‐helical domain). GS‐PEITC is colored by atom type (green, C; red, O; blue, N; and yellow, S). Residues surrounding the PEITC moiety are shown as sticks.

Analysis of TvGSTO3S apo structure using PDBeFold27 indicated that its closest structural homologs are plant Tau GSTs (RMSD of 1.79 Å for 197 aligned C_α with wheat GSTU, PDB code 1GWC) and insect Delta GSTs (RMSD of 1.65 Å for 176 aligned C_α with Drosophila melanogaster DmGSTD2, PDB code 5F0G). Members of both classes were shown to catalyze the conjugation of GSH with molecules from the ITC family.28, 29 The structure of DmGSTD2 was recently solved and its substrate binding properties were based on a molecular dynamics (MD) study.29 DmGSTD2 actively transfers GSH to various ITC molecules presumably with four Gly residues constituting the catalytic motif. MD results suggest a particularly flexible active site that binds GSH and PEITC in a cooperative manner. This flexibility is notably indicated by high B‐factors with respect to the mean B‐factor at the region of helix α₂.29 Interestingly, our results about TvGSTO3S structure suggest a movement of the adjacent loop β₂‐α₂ in the fungal homolog upon GS‐PEITC binding. Superimposition of apo TvGSTO3S and apo DmGSTD2 structures indicates an RMSD of 1.65 Å (176 superimposed C_α) (Fig. S3). The region of the loop between strand β₂ and helix α₂ shows good superimposition between both structures. In addition, DmGSTD2 shares some of the residues that constitute the PEITC moiety binding site in TvGSTO3S (Fig. S3). Altogether, the previous work and this study suggest that this loop could be involved in PEITC binding for both TvGSTO3S and DmGSTD2.

Implications of the structural features of TvGSTO3S on its catalysis

In this study, we report the crystal structure of TvGSTO3S complexed to its conjugation product GS‐PEITC, which binds to the active site. Previously we reported the inhibition of TvGSTO3S by polyphenols. These small molecules compete for the H site with the substrate PEITC.7 Thus, this substrate is thought to bind to the H site, which is located in the vicinity of the GSH binding site. Based on the TvGSTO3S:GS‐PEITC crystal structure, we show that the PEITC moiety is rather localized in a new site at the loop β₂‐α₂. In the following lines, we discuss the biological relevance of the observed product binding mode in a catalytic context.

First possibility: Binding of GS‐PEITC at the β ₂ ‐α ₂ loop could have been a crystallization artifact. In this scenario, in solution PEITC would bind in the H site similarly to the aromatic moiety of GS‐PAP in the complex TvGSTO3S:GS‐PAP7 and similarly to GS‐PEITC in the complexes GSTA1:GS‐PEITC and GSTP1:GS‐PEITC.18 The observed artifact could have been obtained due to particular experimental conditions. One of the following reasons could have prevented this normal binding mode: (i) The unidentified ligand present in TvGSTO3S H site could have hindered the fixation of GS‐PEITC in this site. Nevertheless, it seems unlikely since soaking with other GSH adduct (GS‐PAP and GS‐hexane) at similar or lower concentrations resulted in displacement of this unidentified ligand. (ii) The TvGSTO3S crystallization buffer (acetate pH 5.8) could have interfered with the correct binding of GS‐PEITC. This is probably not a convincing reason since TvGSTO3S is still active at this pH (data not shown) and similar pH was used for crystallization of the complex for GSTA (pH 6.0).18 (iii) At last, soaking TvGSTO3S crystals with the product GS‐PEITC as an alternative to co‐crystallization could have influenced the final result (see Materials and Methods).

Second possibility: Binding of GS‐PEITC at the β ₂ ‐α ₂ loop is not an artifact and reflects the position of the product in the active site at the end of the reaction. In this case, the observed binding mode could correspond to the position of GS‐PEITC at the end of the reaction. This hypothesis suggests that a slight displacement of the loop β₂‐α₂ and a conformational change of GS‐PEITC occurred, via rotations around the NH‐C_α and C_α‐C bonds (φ and ψ angles, respectively) of the glutathionyl cysteine residue. Localization of the PEITC moiety in the new site could facilitate product release. Interestingly, a topological rearrangement that involves the region of helix α₂ upon product release was proposed for Delta GSTs,30 which are close structural homologs of TvGSTO3S (see above). Furthermore, recent simulation studies on DmGSTD2 show that the equivalent region of the loop β₂‐α₂ participates at least partially to the active site flexibility.29 The structural comparison of TvGSTO3S with DmGSTD2 revealed similarities in the loops β₂‐α₂, which could be the regions that allow stabilization of GS‐PEITC in these isoforms. Interestingly, the equivalent loop in the structure of Phi‐class GST‐I from Zea mays moves upon lactoyl‐GSH binding and was proposed to be involved in an induced‐fit mechanism.31

Both hypotheses converge to the same conclusion. TvGSTO3S has an extended active site that occupies both the thioredoxin domain (G site and loop β₂‐α₂ site) and the all‐helical domain (H site). This active site shows flexibility and adaptability to its ligands through conformational changes of the loop β₂‐α₂ as observed in the present study. Among GST superfamily, this region could be a common structural feature that participates to the active site dynamics, as previously suggested.30

Materials and Methods

Protein expression and purification, crystallization, and soaking conditions

TvGSTO3S (accession number in the JGI database: Tv48691) was produced by heterologous expression in Escherichia coli as described previously.6 In brief, a recombinant plasmid that bears the TvGSTO3S sequence with a C‐terminal hexahistidine tag was obtained by gene synthesis. Protein expression was performed at 37°C in transformed E. coli Rosetta2 (DE3) pLysS strain (Novagen) and induced with 0.1 mM isopropyl β‐D‐1‐thiogalactopyranoside (IPTG). Protein purification was performed on a (Ni²⁺‐nitrilotriacetate)‐agarose resin connected to an ÄKTA purifier FPLC system (GE Healthcare). TvGSTO3S was concentrated and dialyzed in a 30 mM Tris‐EDTA pH 8.0 buffer before crystallization assays. Crystallization trials were set up at 4°C using the microbatch under oil method as previously reported.7 TvGSTO3S was crystallized by mixing 1 μL of protein (13 mg/mL) with 3 μL of commercial solution (Wizard Classic, Molecular Dimensions) consisting in 30% (w/v) PEG 400, 0.2M calcium acetate in 0.1M pH 4.5 acetate buffer (final pH of 5.8). Orthorhombic crystals (200 μm in the largest dimension, Fig. S1) appeared after 3–4 days.

In order to get the crystal structure of TvGSTO3S bound to the product GS‐PEITC, we first tested the protocol that was previously used by Kumari and coworkers,18 however, this strategy failed. It consisted in co‐crystallizing the enzyme with GSH and an excess of PEITC to obtain a product‐enzyme complex. In our case, success came when TvGSTO3S apo crystals were soaked with the prepared product. GS‐PEITC formed spontaneously17 by a 1‐hour incubation of 10 mM GSH with 100 mM PEITC in a 30 mM pH 8.0 Tris‐EDTA buffer, with an expected GS‐PEITC concentration of 10 mM. Crystals of the complex TvGSTO3S:GS‐PEITC were obtained by soaking apo TvGSTO3S crystals in their mother liquor supplemented with 10% of the GS‐PEITC mixture. After 1‐day soaking, crystals were cryoprotected by a quick soaking in their mother liquor containing 20% glycerol and cryocooled in liquid nitrogen.

Data collection, refinement, and structure validation

Preliminary X‐ray diffraction experiments were performed on an Agilent SuperNova diffractometer with CCD detector. Data collection up to 2.5 Å resolution allowed evaluation of GS‐PEITC presence in TvGSTO3S active site. High‐resolution diffraction experiments were carried out on the ESRF beamline ID30B (Grenoble, France). TvGSTO3S crystals diffracted up to 1.45 Å. Data set was indexed and integrated with XDS and then scaled and merged with XSCALE.32 The structure was solved by molecular replacement with a search model based on apo TvGSTO3S (PDB code 6F43)7 with all ligands and water removed. Composite omit map calculated with PHENIX23 confirmed the presence of GS‐PEITC bound to TvGSTO3S active site. Restraint file for GS‐PEITC was generated with phenix.elbow.33 The structure of the complex was manually adjusted with COOT34 and refined with PHENIX.23 Refinement statistics are summarized in Table 1. Structure validation was carried out with MOLPROBITY35 and the PDB validation server (http://validate.wwPDB.org). The figures presented in this article were prepared with PYMOL (Schrödinger, LLC). Coordinates and structure factors have been deposited in the Protein Data Bank under accession code 6HPE.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Table S1 TvGSTO3S structures discussed in this study

Table S2. Kinetic parameters of TvGSTO3S toward CDNB, PEITC and BITC

Figure S1. Crystals of TvGSTO3S soaked with GS‐PEITC

Figure S2. Sequence alignment of T. versicolor GSTOS and human GSTO1

Figure S3. Structural comparison of T. versicolor GSTO3S with D. melanogaster GSTD2