Food Chemistry xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Food Chemistry journal homepage: www.elsevier.com/locate/foodchem
Molecular modelling approach to evaluate poisoning of topoisomerase I by alternariol derivatives Luca Dellafiora a, Chiara Dall’Asta b, Gabriele Cruciani c, Gianni Galaverna b, Pietro Cozzini a,⇑ a
Molecular Modeling Laboratory, Department of Food Science, University of Parma, Parco Area delle Scienze 95/A, 43125 Parma, Italy Department of Food Science, University of Parma, Parco Area delle Scienze 95/A, 43125 Parma, Italy c Laboratory for Chemometrics and Cheminformatics, Chemistry Department, University of Perugia, Via Elce di Sotto 8, I-06123 Perugia, Italy b
a r t i c l e
i n f o
Article history: Received 28 April 2014 Received in revised form 16 February 2015 Accepted 17 February 2015 Available online xxxx Keywords: Molecular modelling Alternaria metabolites Alternariol Poisoning Topoisomerase I In silico prediction
a b s t r a c t Alternaria species are widespread microfungi the secondary metabolites of which may accumulate in crops and enter into food production chain. Among them, the ‘‘emerging mycotoxin’’ alternariol and alternariol-methyl ether arouse concern due to evidences of toxicity. In particular, the disruption of topoisomerases leads to genotoxic outcomes. Metabolic modifications may drastically reduce toxic potency by enhancing clearance and/or by preventing interaction with the pharmacological targets. However, the metabolic activation may occur as well. For this reason, understanding the role of metabolised forms is paramount for the in-depth comprehension of adverse effects on living organisms, thus providing a more informed scenario for risk assessment. Regardless that a wealth of alternariol metabolites and derivatives has been identified, most have not been tested with respect to topoisomerases. Consequently, their effects in living organism are not yet well understood. Unfortunately, experimental analysis is challenging and time-consuming. With the aim of analysing a wide array of alternariol metabolites and derivatives, we presented an effective framework based on a straightforward in silico procedure. Interestingly, several metabolites were predicted to be poisons, strongly suggesting the need for further experimental trials and their inclusion in future risk assessment studies. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Alternaria is a common genus of microfungi able to infect a wide variety of crops worldwide, such as fruits, grains, and vegetables (Logrieco, Moretti, & Solfrizzo, 2009), causing serious losses due to post-harvest decay. In addition, Alternaria spp. may produce a wide range of toxins that accumulate in crops and enter the food production chain. The main sources of Alternaria toxins in the human diet are cereals and related products, tomato and sauces, sunflower seeds and oil, fruits, beer and wine (EFSA, 2011a). Alternaria genus is able to produce 71 known mycotoxins and phytotoxins, but only a few have been reported to occur at significant levels in food, mainly at the pre-harvest crop stage (Logrieco et al., 2009). Among them, the dibenzo-a-pyrone mycotoxins alternariol (AOH) and alternariol-methyl ether (AME) (Fig. 1A) deserve
Abbreviations: AOH, alternariol; AME, alternariol-methyl ether; ALT, altenuene; iALT, isoaltenuene; LBVS, ligand based virtual screening; Topo, topoisomerase; Topo I, topoisomerase I; Topo II, topoisomerase II. ⇑ Corresponding author. Tel.: +39 0521 905669; fax: +39 0521 905556. E-mail address: [email protected] (P. Cozzini).
particular attention. Although still unregulated, these emerging mycotoxins should be considered in future risk assessment owing to their toxicological relevance (Marin, Ramos, Cano-Sancho, & Sanchis, 2013). Indeed, AOH is associated with a range of potential adverse health effects having fetotoxic, teratogenic, genotoxic and mutagenic effects in in vitro assays (Fehr et al., 2009). Notably, the poisoning of topoisomerase (Topo) enzymes, which are key enzymes for the modulation of DNA topology, are related to the genotoxic effects of AOH and AME (Fehr et al., 2009), making Topo an important target for toxicology assessment. The European Food Safety Authority (EFSA) (EFSA, 2011a) recently stated the need for further data on toxicodynamics and toxicokinetics for the most toxicologically relevant Alternaria toxins. Additional evidences of genotoxicity for most of them – including AOH, AME and their metabolites – are, therefore, strongly recommended. In this respect, the in-depth analysis of metabolites action is paramount for better understanding the role of parent mycotoxins in living organisms, particularly with respect to hazard identification. Although many metabolites and modified forms have been identified so far, a careful evaluation of genotoxic effects exerted through Topo poisoning enzymes is still missing. This is
http://dx.doi.org/10.1016/j.foodchem.2015.02.083 0308-8146/Ó 2015 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Dellafiora, L., et al. Molecular modelling approach to evaluate poisoning of topoisomerase I by alternariol derivatives. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.02.083
2
L. Dellafiora et al. / Food Chemistry xxx (2015) xxx–xxx
Fig. 1. Chemicals structures of molecules under investigation. The structural features shared by dibenzo-a-pyrone toxins (A) and flavonoids (B) are coloured in blue. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
crucial to define an informed scenario for risk assessment. Unfortunately, unbiased analysis trough in vitro assays is challenging due to the high number of ‘‘hits’’ and lack of commercially available standards. In this context, the upstream priority setting of compounds allows the implementation of expensive and time consuming experimental trials only for established candidates. In this context, in silico screening using a three-dimensional molecular modelling approach can be a cost-effective choice. Such an approach was chosen among the wealth of computational techniques since it is strongly recommended when the mode of action under investigation requires degrees of ligand-target recognition (Gleeson et al., 2012). In fact, whilst Topo enzymes allow DNA relaxation by introducing double- (Topo II) or single-strand breaks (Topo I) (Khadka & Cho, 2013), poisoning of these enzymes requires stabilisation of the DNA–protein–toxin ternary complex (i.e. the so defined cleavable complex) through base-stacking interactions. The toxicant intercalation extends the distance between 1 (upstream) and +1 (downstream) base-pairs until the DNA religation is prevented. Thereby, causing Topo enzymes to damage DNA (Pommier, 2013). Whereas Topo II introduces double-strand breaks, and involves two simultaneous catalytic reactions at two distinct catalytic sites, Topo I has a unique catalytic site and introduce single-strand breaks (Staker et al., 2002). Due to the lack of straightforward procedures or commercial software to handle simultaneous docking simulations in distinct catalytic sites, analysis focused on Topo I. On the basis of a poisoning mechanism, computational analysis searched for interactions between the molecules under investigation and the poisoned Topo I-DNA binary complex by providing an index of the ternary complex formation. In the event of appreciable interaction, molecules were predicted to be Topo I poisons. In the present paper, we presented a framework for the assessment of dibenzo-a-pyrone toxins (Fig. 2) that involved a validated knowledge-based computational procedure based on ligand-based virtual screening (LBVS) and docking simulations. A total of 49 untested compounds (including phase-I and -II metabolites and some AOH modified forms) collected from the literature and then assessed hierarchically for their potential poisoning activity. A series of strong candidates for future assessment of bioactivity were identified. 2. Materials and methods The computational protocol was based on a two-steps protocol: (i) rapid pre-screening of compounds using LBVS followed by (ii)
slower docking simulation and re-scoring procedures to evaluate finely the interaction with the poisoned Topo I–DNA binary complex. 2.1. Data collection Data were collected considering three well-defined set – a training-set, a validation-set, and a query-set – composed as reported below. Training-set: In order to obtain a fit-for-purpose model, the training-set should contain molecules with structures and chemistry related closely to the class under investigation. In this respect, one has to keep in mind that AOH and AME share a polyphenolic structure with a limited number of conjugated rings. Further, metabolic modifications mainly consist of hydroxylations, methylations, reduction/oxidation of carbon–carbon bonds or oxygen atoms, and conjugation with glycosides or sulphate groups. Despite the wealth of studies on Topos poisons, only a few addressed metabolic effects exerted by molecular scaffolds similar to dibenzo-a-pyrone toxins. The assessment of a wide set of flavonoids proposed by Webb was an exception (Webb & Ebeler, 2004). Along with the comparable structural and chemical properties granted by the presence of polyphenolic core of flavonoids (Fig. 1B), the authors addressed the effect on poisoning activity caused by structural modifications mostly attributable to hydroxylations, methylations, glycosylations and reduction/oxidation of carbon–carbon bonds. A total of 20 compounds (8 active and 12 inactive) were selected. Selection criteria were based on data quality, as compounds with misleading nomenclature or experimental data lacking standard deviation were excluded. Specifically, the inactive compounds were chrysin, 30 ,40 -dihydroxy-flavone, luteolin-40 -O-glucoside, gossypetin, morin, tamarixetin, rutin, quercitrin, (+)-taxifolin, silibinin, naringenin and hesperitin. Whereas the active molecules were apigenin, luteolin luteolin-7-O-glucoside, orientin, fisetin, kaempferol, quercetin and myricetin (further details are reported in supplementary materials, Table S1). Validation-set: The validation-set originated from poisoning data for AOH and its derivatives available currently in the literature (Fehr et al., 2009). In particular, AOH showed poisoning activity at a concentration of 50 lM, whilst AME, altenuene (ALT) and isoaltenuene (iALT) were devoid of activity up to 100 lM. Query-set: A set of 49 AOH derivatives and analogous was collected from the literature. The most relevant forms in living organisms were included, with a particular focus on the phase-I and
Please cite this article in press as: Dellafiora, L., et al. Molecular modelling approach to evaluate poisoning of topoisomerase I by alternariol derivatives. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.02.083
L. Dellafiora et al. / Food Chemistry xxx (2015) xxx–xxx
3
Fig. 2. The procedural workflow. Red box highlights methodological steps. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
phase-II human metabolites. A number of relevant co-occurring forms were considered as well (Table 1) (Altemöller & Podlech, 2009; Burkhardt, Wittenauer, Pfeiffer, Schauer, & Metzler, 2011; Pfeiffer, Eschbach, & Metzler, 2007; Pfeiffer, Herrmann, Altemöller, Podlech, & Metzler, 2009; Pfeiffer, Schebb, Podlech, & Metzler, 2007; Pfeiffer, Schmit et al., 2009). 2.2. Computational protocol Pharmacophore models: The ligand-binding site was defined using the Flapsite tool in the FLAP software (Fingerprint for Ligand and Protein) developed by Molecular Discovery Ltd (http://www.moldiscovery.com/) (Baroni, Cruciani, Sciabola, Perruccio, & Mason, 2007), whilst the GRID algorithm (Goodford, 1985) was used to investigate the corresponding pharmacophore space. The DRY probe was used to describe the potential hydrophobic interactions, whilst the sp2 carbonyl oxygen (O) and the neutral flat amino (N1) probes were used to describe the hydrogen bond acceptor and donor capacity of the target, respectively. All images were obtained using the software PyMol version 1.4.1 (http:// www.pymol.org). Ligand based virtual screening (LBVS). Virtual screening (VS) was performed using a ligand based LDA (Linear Discriminant Analysis) model using FLAP software. LBVS allowed the rapid pre-screening of all compounds under analysis (
Contents lists available at ScienceDirect
Food Chemistry journal homepage: www.elsevier.com/locate/foodchem
Molecular modelling approach to evaluate poisoning of topoisomerase I by alternariol derivatives Luca Dellafiora a, Chiara Dall’Asta b, Gabriele Cruciani c, Gianni Galaverna b, Pietro Cozzini a,⇑ a
Molecular Modeling Laboratory, Department of Food Science, University of Parma, Parco Area delle Scienze 95/A, 43125 Parma, Italy Department of Food Science, University of Parma, Parco Area delle Scienze 95/A, 43125 Parma, Italy c Laboratory for Chemometrics and Cheminformatics, Chemistry Department, University of Perugia, Via Elce di Sotto 8, I-06123 Perugia, Italy b
a r t i c l e
i n f o
Article history: Received 28 April 2014 Received in revised form 16 February 2015 Accepted 17 February 2015 Available online xxxx Keywords: Molecular modelling Alternaria metabolites Alternariol Poisoning Topoisomerase I In silico prediction
a b s t r a c t Alternaria species are widespread microfungi the secondary metabolites of which may accumulate in crops and enter into food production chain. Among them, the ‘‘emerging mycotoxin’’ alternariol and alternariol-methyl ether arouse concern due to evidences of toxicity. In particular, the disruption of topoisomerases leads to genotoxic outcomes. Metabolic modifications may drastically reduce toxic potency by enhancing clearance and/or by preventing interaction with the pharmacological targets. However, the metabolic activation may occur as well. For this reason, understanding the role of metabolised forms is paramount for the in-depth comprehension of adverse effects on living organisms, thus providing a more informed scenario for risk assessment. Regardless that a wealth of alternariol metabolites and derivatives has been identified, most have not been tested with respect to topoisomerases. Consequently, their effects in living organism are not yet well understood. Unfortunately, experimental analysis is challenging and time-consuming. With the aim of analysing a wide array of alternariol metabolites and derivatives, we presented an effective framework based on a straightforward in silico procedure. Interestingly, several metabolites were predicted to be poisons, strongly suggesting the need for further experimental trials and their inclusion in future risk assessment studies. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Alternaria is a common genus of microfungi able to infect a wide variety of crops worldwide, such as fruits, grains, and vegetables (Logrieco, Moretti, & Solfrizzo, 2009), causing serious losses due to post-harvest decay. In addition, Alternaria spp. may produce a wide range of toxins that accumulate in crops and enter the food production chain. The main sources of Alternaria toxins in the human diet are cereals and related products, tomato and sauces, sunflower seeds and oil, fruits, beer and wine (EFSA, 2011a). Alternaria genus is able to produce 71 known mycotoxins and phytotoxins, but only a few have been reported to occur at significant levels in food, mainly at the pre-harvest crop stage (Logrieco et al., 2009). Among them, the dibenzo-a-pyrone mycotoxins alternariol (AOH) and alternariol-methyl ether (AME) (Fig. 1A) deserve
Abbreviations: AOH, alternariol; AME, alternariol-methyl ether; ALT, altenuene; iALT, isoaltenuene; LBVS, ligand based virtual screening; Topo, topoisomerase; Topo I, topoisomerase I; Topo II, topoisomerase II. ⇑ Corresponding author. Tel.: +39 0521 905669; fax: +39 0521 905556. E-mail address: [email protected] (P. Cozzini).
particular attention. Although still unregulated, these emerging mycotoxins should be considered in future risk assessment owing to their toxicological relevance (Marin, Ramos, Cano-Sancho, & Sanchis, 2013). Indeed, AOH is associated with a range of potential adverse health effects having fetotoxic, teratogenic, genotoxic and mutagenic effects in in vitro assays (Fehr et al., 2009). Notably, the poisoning of topoisomerase (Topo) enzymes, which are key enzymes for the modulation of DNA topology, are related to the genotoxic effects of AOH and AME (Fehr et al., 2009), making Topo an important target for toxicology assessment. The European Food Safety Authority (EFSA) (EFSA, 2011a) recently stated the need for further data on toxicodynamics and toxicokinetics for the most toxicologically relevant Alternaria toxins. Additional evidences of genotoxicity for most of them – including AOH, AME and their metabolites – are, therefore, strongly recommended. In this respect, the in-depth analysis of metabolites action is paramount for better understanding the role of parent mycotoxins in living organisms, particularly with respect to hazard identification. Although many metabolites and modified forms have been identified so far, a careful evaluation of genotoxic effects exerted through Topo poisoning enzymes is still missing. This is
http://dx.doi.org/10.1016/j.foodchem.2015.02.083 0308-8146/Ó 2015 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Dellafiora, L., et al. Molecular modelling approach to evaluate poisoning of topoisomerase I by alternariol derivatives. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.02.083
2
L. Dellafiora et al. / Food Chemistry xxx (2015) xxx–xxx
Fig. 1. Chemicals structures of molecules under investigation. The structural features shared by dibenzo-a-pyrone toxins (A) and flavonoids (B) are coloured in blue. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
crucial to define an informed scenario for risk assessment. Unfortunately, unbiased analysis trough in vitro assays is challenging due to the high number of ‘‘hits’’ and lack of commercially available standards. In this context, the upstream priority setting of compounds allows the implementation of expensive and time consuming experimental trials only for established candidates. In this context, in silico screening using a three-dimensional molecular modelling approach can be a cost-effective choice. Such an approach was chosen among the wealth of computational techniques since it is strongly recommended when the mode of action under investigation requires degrees of ligand-target recognition (Gleeson et al., 2012). In fact, whilst Topo enzymes allow DNA relaxation by introducing double- (Topo II) or single-strand breaks (Topo I) (Khadka & Cho, 2013), poisoning of these enzymes requires stabilisation of the DNA–protein–toxin ternary complex (i.e. the so defined cleavable complex) through base-stacking interactions. The toxicant intercalation extends the distance between 1 (upstream) and +1 (downstream) base-pairs until the DNA religation is prevented. Thereby, causing Topo enzymes to damage DNA (Pommier, 2013). Whereas Topo II introduces double-strand breaks, and involves two simultaneous catalytic reactions at two distinct catalytic sites, Topo I has a unique catalytic site and introduce single-strand breaks (Staker et al., 2002). Due to the lack of straightforward procedures or commercial software to handle simultaneous docking simulations in distinct catalytic sites, analysis focused on Topo I. On the basis of a poisoning mechanism, computational analysis searched for interactions between the molecules under investigation and the poisoned Topo I-DNA binary complex by providing an index of the ternary complex formation. In the event of appreciable interaction, molecules were predicted to be Topo I poisons. In the present paper, we presented a framework for the assessment of dibenzo-a-pyrone toxins (Fig. 2) that involved a validated knowledge-based computational procedure based on ligand-based virtual screening (LBVS) and docking simulations. A total of 49 untested compounds (including phase-I and -II metabolites and some AOH modified forms) collected from the literature and then assessed hierarchically for their potential poisoning activity. A series of strong candidates for future assessment of bioactivity were identified. 2. Materials and methods The computational protocol was based on a two-steps protocol: (i) rapid pre-screening of compounds using LBVS followed by (ii)
slower docking simulation and re-scoring procedures to evaluate finely the interaction with the poisoned Topo I–DNA binary complex. 2.1. Data collection Data were collected considering three well-defined set – a training-set, a validation-set, and a query-set – composed as reported below. Training-set: In order to obtain a fit-for-purpose model, the training-set should contain molecules with structures and chemistry related closely to the class under investigation. In this respect, one has to keep in mind that AOH and AME share a polyphenolic structure with a limited number of conjugated rings. Further, metabolic modifications mainly consist of hydroxylations, methylations, reduction/oxidation of carbon–carbon bonds or oxygen atoms, and conjugation with glycosides or sulphate groups. Despite the wealth of studies on Topos poisons, only a few addressed metabolic effects exerted by molecular scaffolds similar to dibenzo-a-pyrone toxins. The assessment of a wide set of flavonoids proposed by Webb was an exception (Webb & Ebeler, 2004). Along with the comparable structural and chemical properties granted by the presence of polyphenolic core of flavonoids (Fig. 1B), the authors addressed the effect on poisoning activity caused by structural modifications mostly attributable to hydroxylations, methylations, glycosylations and reduction/oxidation of carbon–carbon bonds. A total of 20 compounds (8 active and 12 inactive) were selected. Selection criteria were based on data quality, as compounds with misleading nomenclature or experimental data lacking standard deviation were excluded. Specifically, the inactive compounds were chrysin, 30 ,40 -dihydroxy-flavone, luteolin-40 -O-glucoside, gossypetin, morin, tamarixetin, rutin, quercitrin, (+)-taxifolin, silibinin, naringenin and hesperitin. Whereas the active molecules were apigenin, luteolin luteolin-7-O-glucoside, orientin, fisetin, kaempferol, quercetin and myricetin (further details are reported in supplementary materials, Table S1). Validation-set: The validation-set originated from poisoning data for AOH and its derivatives available currently in the literature (Fehr et al., 2009). In particular, AOH showed poisoning activity at a concentration of 50 lM, whilst AME, altenuene (ALT) and isoaltenuene (iALT) were devoid of activity up to 100 lM. Query-set: A set of 49 AOH derivatives and analogous was collected from the literature. The most relevant forms in living organisms were included, with a particular focus on the phase-I and
Please cite this article in press as: Dellafiora, L., et al. Molecular modelling approach to evaluate poisoning of topoisomerase I by alternariol derivatives. Food Chemistry (2015), http://dx.doi.org/10.1016/j.foodchem.2015.02.083
L. Dellafiora et al. / Food Chemistry xxx (2015) xxx–xxx
3
Fig. 2. The procedural workflow. Red box highlights methodological steps. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
phase-II human metabolites. A number of relevant co-occurring forms were considered as well (Table 1) (Altemöller & Podlech, 2009; Burkhardt, Wittenauer, Pfeiffer, Schauer, & Metzler, 2011; Pfeiffer, Eschbach, & Metzler, 2007; Pfeiffer, Herrmann, Altemöller, Podlech, & Metzler, 2009; Pfeiffer, Schebb, Podlech, & Metzler, 2007; Pfeiffer, Schmit et al., 2009). 2.2. Computational protocol Pharmacophore models: The ligand-binding site was defined using the Flapsite tool in the FLAP software (Fingerprint for Ligand and Protein) developed by Molecular Discovery Ltd (http://www.moldiscovery.com/) (Baroni, Cruciani, Sciabola, Perruccio, & Mason, 2007), whilst the GRID algorithm (Goodford, 1985) was used to investigate the corresponding pharmacophore space. The DRY probe was used to describe the potential hydrophobic interactions, whilst the sp2 carbonyl oxygen (O) and the neutral flat amino (N1) probes were used to describe the hydrogen bond acceptor and donor capacity of the target, respectively. All images were obtained using the software PyMol version 1.4.1 (http:// www.pymol.org). Ligand based virtual screening (LBVS). Virtual screening (VS) was performed using a ligand based LDA (Linear Discriminant Analysis) model using FLAP software. LBVS allowed the rapid pre-screening of all compounds under analysis (
