Native substrates in the corresponding sites
The structure of a TGR in complex with a substrate/inhibitor in the thiol/disulphide redox active pocket (TR active cavity) has not yet been determined. The TR domain of TGR has similarities with TRs and GRs and exhibits TR and GR enzymatic activities. Since the GDS binding pocket close to the FAD-binding site (thiol/disulphide redox active centre) has been proposed to play a major role in the reduction of GDS in GR, TR and smTGR (19,20), GDS was docked into the TR active cavity of smTGR and sjTGR, based on the structure of the human GR-GDS complex (PDB ID: 2GRT) (18). It was found that GDS was embedded into the dimer interface, near the FAD-binding site and the disulphide motif (Cys154-Cys159). The binding site for GDS in the TR active cavity of TGR was observed to be composed of residues from two subunits. In smTGR, GDS was surrounded by Ser117, Ala121, Lys124, Val155, Ile160, Lys163, His204, Leu208, Tyr212, Ile446 and Arg450 from one subunit and Lys506′, Leu508′, His571′, Pro572′, Thr573′, Cys574′, Glu576′, Thr577′ and Thr580′ from another subunit. As observed in docking results reported by Sharma et al (20), GDS in the active site formed hydrogen bonds with Tyr212, Lys124 and Arg450, residues that are also conserved in GR. GDS also formed hydrogen bonds with Leu508′, His571′, Thr573′ and Glu576′ from the other subunit.
In sjTGR, GDS occupies the same position as in smTGR, except that Leu120 replaces the Lys124 that interacts with GDS in smTGR (Fig. 2A). The remaining residues that form contacts with GDS in sjTGR are the same as those described for smTGR. The estimated free energy of binding of GDS to smTGR is −2.80 kcal/mol, compared with −3.39 kcal/mol for binding to sjTGR (Tables II and III), indicating that GDS binding to sjTGR is more stable than GDS binding to smTGR.
One reduced glutathione (GSH) was found to bind in the active site of the Grx domain of the smTGR template (PDB ID 2×99) (9), surrounded by three segments of the polypeptide, residues Lys25-Tyr30, Thr71-Pro73 and Asp84-Glu86. It was also found that Gln60 could form a hydrogen bond with GSH. As a control, GSH was re-docked into the Grx domain of smTGR and it was found that the docked molecule completely overlapped with the GSH conformation observed in the X-ray structure. As with smTGR, GSH in sjTGR also made contacts with the conserved residues Lys25, Cys28, Gln60, Thr71, Val72, Pro73, Asp84 and Ser85 (Fig. 2B). In sjTGR, residues Phe30 and Lys86 replace Tyr30 and Gln86 of smTGR, respectively. These replacements resulted in an increase in free binding energy from −4.14 kcal/mol in smTGR to −3.17 kcal/mol in sjTGR (Tables II and III), indicating that the Grx domains of the two TGRs have different binding capacities for GSH.
NADPH functions as an electron donor in the electron transport chain of TGRs. An X-ray structure of smTGR in complex with NADPH (9) showed the NADPH was surrounded by residues Lys162, Pro264, Ile292-Ala298, Glu300, Met315, Arg317-Ile319, Leu321, Arg322, Pro349, Glu390-Arg393, Gln440, Leu441, Thr472 and Phe474. The present docking results showed that NADPH could re-dock into its initial position with an estimated free binding energy of −10.17 kcal/mol. In sjTGR, the docked NADPH interacts with the conserved residues Lys162, Pro264, Gly293-Ala298, Lys300, Arg317-Ile319, Leu321, Arg322, Ala390-Arg393, Gln440, Leu441 and Thr472-Phe474 (Fig. 2C). Although the NADPH binding site appears to be conserved among TGRs, the estimated free binding energy with sjTGR (−6.59 kcal/mol) was significantly higher than that with smTGR.
The present data suggested that sjTGR and smTGR have different binding capacities for their native substrates, despite the fact that they share high sequence identity and conserved residues in their binding sites. Our previous kinetic analysis (21) showed that the Km value for sjTGR with GDS (49.55 µM) was lower than that for smTGR (71.5 µM), whereas the Km values for sjTGR with NADPH (21.426 µM) and GSH (1,698 µM) were higher than those for smTGR (13.7 and 248.6 µM, respectively). This indicates that sjTGR has a higher binding capacity for GDS than does smTGR, while the binding capacities of sjTGR for GSH and NADPH were lower than those of smTGR, in agreement with our conclusion. These results are concordant with those of the present study, which suggest that sjTGR and smTGR have different binding capacities for their native substrates, despite sharing high sequence identity and conserved residues in their binding sites.
TR active site cavity as the effective binding site for inhibitors
Although the six selected compounds inhibit smTGR in vitro (4,12,13,22), the mechanism by which that happens remains unknown. The present study has focused on whether these molecules can directly gain access to the three main functional sites of TGR. The binding scores of the best-docked conformations of ligands in smTGR and sjTGR are listed in Tables II and III, respectively.
In smTGR, the NADPH-binding site and TR active cavity (GDS binding site) exhibit lower estimated free binding energies for the compounds than does the Grx domain (Table II). The free binding energy shown by one of the controls, NADPH, was much lower than the free binding energy shown by all the test compounds.
In sjTGR, the NADPH-binding site and the TR active cavity exhibited better binding capacities for the test compounds than the Grx domain (Table III). In each site, the binding capacity of sjTGR for the test compounds was generally marginally weaker than that of smTGR, apart from compound 5, which exhibited a better free binding energy for the TR active cavity of sjTGR. The most marked differences in binding capacities were observed in the Grx active site, with an increase of 1.21 kcal/mol in average binding energy for each test compound.
It is not sufficient, however, to estimate the free binding energy of single test molecules alone. In order to clarify whether compounds can competitively inhibit the enzymatic activity of TGR, the free binding energy between test molecules and native substrates was compared. The ratios of the free binding energies of inhibitors to those of native substrates (docking ratios) were used to evaluate the inhibitory efficiency of the compounds.
The docking ratios of the test molecules in smTGR are shown in Fig. 3A. All docking ratio values in the TR active cavity were >2.3, higher than those in the Grx domain or the NADPH-binding site. For sjTGR, the values of the docking ratio in the TR active centre were >1.9 (Fig. 3B). These values were generally lower than those for smTGR, indicating that the compounds were less effective sjTGR inhibitors. The data suggested that it is necessary to design effective inhibitors for the specific TGRs from different Schistosoma species.
The docking ratios in the TR active cavity of sjTGR and smTGR were both higher than those in the Grx domain or the NADPH-binding site, suggesting that these compounds may exert their inhibitory effects mainly through interactions in the TR active cavity, with smaller effects on the Grx domain and NADPH-binding site. The TR active cavity may thus be the favoured binding site for these compounds, which may effectively prevent the native substrate from accessing this binding site and disrupting the electron transfer in the TGRs. The following section focuses on the binding mode of compounds in the TR active cavity.
Binding mode of compounds in the active site of the GR/TR domain
Oxadiazole 2-oxides
Simeonov et al (12) identified 39 molecules with an IC50 <10 µM from a quantitative high-throughput screen (qHTS) of 71,028 compounds. The IC50 values of the oxadiazole 2-oxides (compounds 1 and 2) against smTGR in the initial qHTS test were 8.7 and 9.2 µM, respectively, although compound 2 was much more active (IC50=0.02 µM) in the TGR assays than compound 1 (IC50=1.8 µM) (11). According to Sayed et al (13), however, the IC50 values of compounds 1, 2 and 3 against GR activity were 0.32, 0.51 and 12.5 mM, respectively.
The present docking results showed that the oxadiazole 2-oxides occupy a similar position at the dimer interface in the two TGRs (Fig. 4A–C), making contacts with Ile160, Lys163 His204 (from one subunit) and Phe505′, Lys506′, Leu507′, Glu508′, His571′-Cys574′ and Thr577′ (from another subunit).
It was found that compound 1 adopted the same conformation in both smTGR and sjTGR (Fig. 4A) and had similar free binding energies (6.63 and 6.45 kcal/mol, respectively). The cyano group was in close proximity to Pro572′ (both 2.8 Å in the two TGRs), the N-atom of the oxadiazole formed a hydrogen bond with Leu508′ (both 3.0 Å) and the phenyl ring made an arene-H interaction with Lys506′ (both 4.2 Å). These results indicate that compound 1 may also inhibit sjTGR and are consistent with the conclusion of Sayed et al (13) that this compound is also active against Schistosoma japonicum.
Compound 2 has two large acetylthiophene groups in the oxadiazole ring. In smTGR, the carbonyl O of one acetylthiophene group and the O-atom of the oxadiazole formed hydrogen bonds with Lys506′ and Cys574′, respectively, stabilising its conformation in the active site. The best docking conformation of compound 2 in sjTGR, however, was different from that in smTGR. The O-atom of the oxadiazole ring formed a hydrogen bond with Lys506′ and N-O with Thr577′ of sjTGR (Fig. 4B).
The compound 3 molecule exists in a rigid plane. It was found to bind in the same position at the dimer interface of the TGRs by interacting with the Leu160 and Lys163 residues and peptide fragments 505′–508′ and 571′–577′ in the other chains (Fig. 4C). We cannot explain why Sayed et al (13) found compound 3 to be less active against GR and TR than compound 1, since the free binding energy of compound 3 was better than that of compound 1 in both TGRs.
Phosphinic acid amide
N-(benzothiazol-2-yl-phenyl-phosphoryl)-1,3-thiazol-2-amine (compound 4) has been identified as an inhibitor of smTGR and has also been found to be effective against the parasite (11,12). Simeonov et al (12) indicated that the role of the benzothiazole group is critical for the inhibitory activity.
In the present case, the phosphinic acid amide accessed the same pocket as the oxadiazole 2-oxides (Fig. 4A–C). Both in smTGR and sjTGR, the phosphinic acid amide formed hydrogen bonds with the carboxyl group of Glu576′. The benzothiazole ring reached out to the redox couple (Cys154-Cys159) and was surrounded by Cys154, Val155, Ile160, Leu208, Thr442 and Ile446. Two peptide segments of the TGRs, 505′–508′ and 572′–577′, stabilised the thiazole group of compound 4 in the TGRs through van der Waals forces and hydrophobic interactions (Fig. 4D).
Isoxazolone
Compound 5, the isoxazolone dimethyl 2-[3-bromo-6-oxo-6H-anthra (1,9-cd)isoxazol-5-yl] malonate, adopted a similar conformation in both smTGR and sjTGR. The binding pocket for isoxazolone is mainly composed of residues from two subunits and involves residues 160–163, 204, 208, 212, 442, 446, 505′–509′ and 571′–577′ (Fig. 4E). The ligand is stabilised in the active site of the proteins, mainly through van der Waals forces and hydrophobic interactions. In the TGRs, the isoxazolone exhibited a high binding score, indicating that it may be an effective inhibitor. This finding is in agreement with the data presented by Simeonov et al (12), in which the isoxazolone exhibited a low IC50 against smTGR in the qHTS. The isoxazolone exhibited the lowest free binding energy of all the compounds with sjTGR and its binding score for docking into sjTGR was even higher than that for docking into smTGR, therefore suggesting that isoxazolone may also be active against sjTGR.
Phosphoramidite
Compound 6, the phosphoramidite 6-methyl-2-(naphthalen-1-yloxy)-2,3-dihydro-1,3,2-diaza-phosphinin-4(1H)-one 2-oxide, was identified as an inhibitor of smTGR by both the qHTS and the TGR assay (12). As with the other compounds, the best binding site of the phosphoramidite, both in smTGR and in sjTGR, involved Ile160, Lys163, His204, Phe505′–Leu508′ and Pro572′–Thr577′. Lys506′ and His571′ were within hydrogen bonding distance of the ligand (Fig. 4F). The estimated free binding energies of the phosphoramidite were −7.66 kcal/mol with smTGR and −7.44 kcal/mol with sjTGR (Tables II and III). These data suggest that the phosphoramidite may also be active against Schistosoma japonicum.
Conclusion
In order to explore the interaction mechanism of these compounds, six previously identified TGR inhibitors were docked into the NADPH binding site, the TR active cavity and the Grx active site of smTGR, as determined by the X-ray structure (Schistosoma mansoni), and into the structure of sjTGR (Schistosoma japonicum), as constructed by homology modelling. The present results indicate that the favourite binding site of inhibitors is the TR active cavity surrounded by an FAD-binding site, a redox-active Cys154/Cys159 pair from one subunit and a C-terminal peptide from the other subunit. Although these inhibitors are structurally quite different, the best binding positions of all the compounds in smTGR and sjTGR were similar in the GDS binding site. Two peptide fragments from another subunit, Phe505′–Leu508′ and Pro572′–Thr577′, played a critical role in the interactions with the inhibitors. The data also suggest that the inhibitors are generally marginally less effective against sjTGR than against smTGR, implying that it would be necessary to design inhibitors specific for the TGRs from different Schistosoma species. An enhanced understanding of the binding mechanisms of potential inhibitors of Schistosoma TGRs would facilitate structure-based ligand design.