|Home | About | Journals | Submit | Contact Us | Français|
In the present work we performed a combined experimental and computational study on the interaction of the natural antimalarial endoperoxide plakortin and its synthetic analogue 4a with heme. Obtained results indicate that the studied compounds produce reactive carbon radical species after being reductively activated by heme. In particular, similarly to artemisinin, the formation of radicals prone to inter-molecular reactions should represent the key event responsible for Plasmodium death. To our knowledge this is the first experimental investigation on the reductive activation of simple antimalarial endoperoxides (1,2-dioxanes) by heme and results were compared to the ones previously obtained from the reaction with FeCl2. The obtained experimental data and the calculated molecular interaction models represent crucial tools for the rational optimization of our promising class of low-cost synthetic antimalarial endoperoxides.
Malaria, an infectious disease caused by protozoans belonging to the genus Plasmodium, continues to affect a large part of the world population, with a special incidence in the sub-Saharan Africa. Indeed, approximately 88% of malaria cases and 90% of malaria deaths occurred in the WHO African Region, with children aged under 5 years and pregnant women most severely affected1. Despite the introduction of artemisinin(1)-based combination therapies (ACTs) and the mass use of insecticide treated mosquito net (ITN), the pace of decrease in estimated malaria mortality rates slowed down1. Indeed, millions of people at risk of malaria still do not have access to interventions such as an ITN, indoor residual spraying (IRS), diagnostic testing, and ACTs. In addition, the emergence of artemisinin-resistant Plasmodium falciparum (Pf) strains is raising severe concerns1 and the rate at which resistance is growing outpaces the development of new effective and low-cost antimalarials2. In the course of a search for new antimalarial lead compounds, we reported that plakortin (2) and dihydroplakortin (3) (Fig. 1), simple endoperoxide-containing polyketides isolated from the Caribbean sponge Plakortis simplex3,4,5, show significant in vitro activity against chloroquine (CQ)-resistant strains of Pf (i.e., IC50 0.4μM), promising in vivo activity (P. berghei infected mice), and no observed toxicity3,4,6.
The preparation of several plakortin semi-synthetic derivatives7 allowed the development of a first set of structure-activity relationships (SARs), which strongly suggested that the antimalarial mechanism of action of plakortins, similarly to that of 18,9,10,11, could include a FeII-induced reductive activation which generates toxic carbon radicals thus leading to the death of the parasite. By using an integrated computational and experimental (FeCl2) model, we therefore investigated the FeII-induced reactivity of plakortins12. Results confirmed that 2 and 3 react with FeII undergoing a dissociative electron transfer (DET) of the endoperoxide bond and forming an oxygen-centered radical at O1, which is simultaneously transferred to C10 carbon of 2 or to C13 carbon of 3, by means of an intra-molecular radical shift mechanism. The resulting carbon radicals were supposed to represent the toxic species able to alkylate biological molecules and to unbalance the redox Plasmodium environment. Further SAR studies performed on additional plakortin analogues13,14,15 supported the above mechanism of action. Thus, we set up a low cost MnIII-mediated synthesis of a new series of affordable antimalarial endoperoxides, designed on the basis of the plakortin pharmacophore model and characterized by the 3-methoxy-1,2-dioxane scaffold (4; Fig. 1)16. The new synthetic derivatives showed in vitro antimalarial activity on Pf CQ-resistant strains comparable to that of the natural leads. Computational and SAR studies performed on the new series of synthetic 1,2-dioxane derivatives indicated that, as in the case of plakortins, the antimalarial activity is related to their ability to react with FeII generating carbon centered radicals (in this case at C3 and/or at C6 alkyl chain)16,17,18,19. Nevertheless, the reaction of 4a with FeCl2 did not evidence any product whose formation could be related to the antimalarial activity16.
A recent analysis of the mechanism of action of 120 showed that free heme-FeII, rather than free FeII, is predominantly responsible for its reductive activation, in agreement with the observed heme-FeII-mediated reactivity of 1 in in vitro models21,22,23,24,25,26,27,28,29,30. By contrast, neither plakortins nor their synthetic analogues, have been ever investigated for their heme-FeII-mediated reactivity. In order to fulfill this gap and put the base for a rational optimization of this new and promising class of low-cost antimalarials, in the present work we investigated for the first time the interaction of 2 and one of its active synthetic analog (4a; Fig. 1) with heme-FeII chloride dimethyl ester (hereafter heme-FeII-Cl), by NMR spectroscopy and MS analyses. Obtained data are discussed in the frame of our previous results12,13,14,15,16,17,18,19 and a molecular interaction model of heme-FeII in complex with the studied ligands was generated.
NMR was selected as method of choice for the initial characterization of the interaction between 2 and heme. In particular, as shown in Fig. 2, 2 (3 equiv.) was incubated with FeIII protoporphyrin IX dimethyl ester chloride (hemin-FeIII-Cl, 1 equiv., 5) in presence of a large excess of the reducing agent 2,3-dimethylhydroquinone (DMHQ, 10 equiv., 6), according to the procedure described by Meunier and coworkers for 122.
The progress of the reaction was closely monitored in real-time in the NMR tube by acquiring a series of mono-dimensional (1D) 1H spectra. As can be seen in Supplementary Figure S1, the resonance at 6.53 ppm, assigned to the aromatic protons of 6, decreased in intensity, while that at 6.72 ppm, the corresponding protons of its oxidized quinone form, increased in intensity. The oxidation of 6, not observed in the presence of the sole 2 (data not shown), indicates, as expected, the concomitant reduction of hemin-FeIII to heme-FeII. Interestingly, the analysis of the spectra revealed also that the signals of 2 changed over time. In particular, upon the addition of the reducing agent, some signals decreased in intensity and vanished almost completely in ~15h, whereas other ones exhibited significant broadening (Fig. 2b). These changes were not observed for 2 in the presence of the sole hemin-FeIIICl or reducing agent (data not shown). In particular, the double bond proton signals at 5.38 (H10) and 5.10 (H9) ppm, the endoperoxide oxymethine signal (H3 δ=4.49 ppm), and signals of other protons surrounding the endoperoxide ring (H15 δ=1.37, H18 δ=3.05) (Fig. 2b) virtually disappeared, suggesting a substantial modification (likely a ring opening) of the endoperoxide skeleton. The broadening of 2 signals, i.e. the methyl groups at around 0.9ppm (starred H12, H14 and H17 peaks in Fig. 2b), could be due to the formation of an adduct between 2 and hemin-FeIII, as a consequence of the presence of paramagnetic FeIII.
To gain insight on the reaction products, 2D 1H-13C HSQC NMR spectra were also acquired. From the spectra analysis, one of the products identified was the chlorinated compound 9 (Fig. 3).
This product, confirmed by MS spectrum, as indicated by the presence of an ion peak at m/z 349.24, was obtained also when 2 was allowed to react with FeCl212. We can then hypothesize that 9 is formed by interaction of 2 with heme-FeII-Cl following the reaction mechanism depicted in Fig. 3.
The one-electron reduction of the peroxide function yields an oxygen radical which is promptly transferred to C10 by an intramolecular rearrangement (1,5-radical shift; 8, Fig. 3). Once this radical is formed, it reacts with a chlorine atom, possibly linked to the hemin-FeIII-Cl, leading to compound 9. However, while the reaction with FeCl2 yielded a mixture of two C10 epimers, in the case of heme only the 10S stereoisomer was formed. Indeed, 2D NMR spectra revealed the presence of a single set of characteristic proton and carbon signals, e.g. those at positions 10 (δH=3.98, δC=65.2) and 9 (δH=3.89, δC=79.2), assigned by comparison with literature data12,31 (Supplementary Figure S2).
Additional reaction products of 2 with heme were observed by NMR and MS spectra, which could not be fully characterized. Among them we could identify: (i) a reduced open form of 2 containing two –OH groups at C3 and C6 (Supplementary Figure S2); (ii) an adduct formed by a modified 2 and hemin, as indicated by the presence in the MS spectrum of an ion at m/z 957.59, corresponding to the sum of the masses of hemin and 2 plus one mass unit. Finally, hemin signals, both in the mass (m/z 644.25) and in the NMR spectra (Supplementary Figure S3), indicate that most of hemin remains unmodified upon reaction. Conversely, no MS signals attributable to intact 2 are observed, evidencing that all 2 reacted.
In order to test the generality of the mechanism of action of 3-methoxy-1,2- dioxane scaffold16,17,18,19, we selected the active compound 4a (Fig. 1), a recently developed synthetic analogue of 2, to reproduce the experiment of interaction with heme-FeII-Cl.
Similarly to 2, 4a (3 equiv.) was allowed to react with 5 (1 equiv.) in the presence of the reducing agent 6 (10 equiv.). The progress of the reaction was closely monitored in real-time in the NMR tube by acquiring a series of 1D 1H NMR spectra, following peculiar methyl protons of the reagent (H7, H8, H14 and H18, Fig. 4).
The NMR analysis indicated that: (i) a fast reaction occurs immediately after the addition of the reducing agent, as shown by the reduced NMR signal intensities already after 10min; (ii) NMR signals of 4a (e.g. methyl resonances) decreased at about 50% after 30min and disappeared almost completely after 6h, indicating the completeness of the reaction at this time; (iii) simultaneously to the disappearance of the signals of 4a, several new signals, ascribable to reaction products, appeared in the NMR spectrum (Fig. 5).
A detailed 2D NMR analysis, performed by 1H-1H TOCSY, 1H-13C HSQC, 1H-13C HMBC spectra registered on the reaction mixture, allowed us to identify the structure of several reaction products (Fig. 6; Supplementary Figures S4–S9).
Among them, 13, 14, 15, and 17 were also identified by MS analysis, whereas 10, 11, 12, and 16 could not be detected in the conditions used because of their volatility and low molecular weight. In addition, as for 2, a signal corresponding to the molecular ion of hemin-FeIII was observed. This finding is in agreement with the NMR analysis, since the hemin signals remain mainly unmodified at the end of the reaction (data not shown).
The formation of the products reported in Fig. 6 can be rationalized by assuming the reductive activation of 4a by heme-FeII, generated in situ by the reducing agent 6, the homolytic cleavage of the peroxide bond and the formation of the two different alkoxy radicals, either on O1 (pathway a) or on O2 (pathway b) (Fig. 7). The subsequent homolytic cleavage of an adjacent C-C bond (beta scission) can give rise to two different alkyl carbon centered radicals, namely 18 from pathway a, and 19 from pathway b. At this step of the reaction, products 12 and 10 can be formed in the pathway a and b, respectively. In pathway a, the primary C5 centered radical 18 undergoes a classic carbon-carbon bond cleavage with oxidation of O2 and reduction of FeIII to FeII. These intra-molecular rearrangements can lead to the formation of the products 10 and 11. In pathway b, the α-carbonyl radical 19 can provide the covalent adduct 14 with heme. NMR assignment of 14 signals revealed the presence of the enol form of the carbonyl group (Supplementary Figure S7) as a consequence of its participation in heme conjugation.
Finally, products 13, 15, 16, and 17 are likely formed as a result of different evolutions of the O2 centered radical (pathway b). In particular, the formation of compounds 13, 15, and 16 can be rationalized through the following mechanism: a butyl radical is ejected via α-cleavage forming first a methyl ester intermediate bound to the heme by the O1, which is then protonated to 13 with release of heme. Compound 13 then undergoes spontaneously lactonization to 15. Finally, we hypothesize that the butyl radical undergoes aerobic oxidation to butanal (16) under our experimental conditions32.
Remarkably, we also identified the chlorinated product 17. A rationale for the formation of 17 could involve an intra-molecular 1,4-H shift reaction between the endoperoxide oxygen O2 and the C3 butyl chain that leads to the formation of a carbon centered radical on the alkyl chain eventually quenched by a chlorine atom, thus following a reaction pathway similar to that reported for the formation of 9.
Docking studies on 4a in complex with heme-FeII were performed by using a Monte Carlo/Simulated Annealing (SA) based protocol12,33. In order to enhance the variance of the resulting complexes, two conformers of 4a were used as starting structures, one presenting intra-molecular distances suitable for the 1,4-H shift and the other presenting intra-molecular distances suitable for the 1,5-H shift (Supplementary Figure S10). Then, following the principles of statistical mechanics, the Monte Carlo-minimization approach randomly generated a total of 19 complexes which were then subjected to a molecular dynamics SA protocol. This included a dynamics run divided in multiple stages, during which the temperature of the system is linearly decreased from 500K to 300K. To ensure that the results obtained are independent from the starting geometries, during all calculations the conformational space of the ligand and the target are explored and all rotatable bonds are left free to move. The resulting complexes are subjected to a final optimization without any restraint (see the Methods Section for further details on the computational procedure). All obtained complexes (Supplementary Figure S11) were ranked by their conformational energy (Supplementary Table S1), the lowest energy complex was selected and then subjected to unconstrained DFT full optimization using B3LYP hybrid exchange-correlation functional34,35 and LanL2DZ basis set36,37,38. A DFT full optimization was also performed on the previously reported12 2-heme-FeII complex, obtained applying the same docking procedure herein reported for 4a. The resulting 2- and 4a-heme-FeII docked complexes are reported in Fig. 8.
It is noteworthy that, comparing the structure of the two starting complexes with those obtained after DFT full optimization (Supplementary Figure S12), a significant change in ligand-heme orientation and C3 (2) or C4 (4a) chain conformation can be observed, while overall the conformation of C6 (2) or C3 (4a) alkyl chain is preserved. Interestingly, the relative orientation of the C6 (2) or C3 (4a) substituent with respect to the endoperoxide function resulted suitable for a 1,5- or 1,4-radical shift from O1 (2) or O2 (4a), respectively, although in the case of 4a the docking procedure started from the conformer presenting distances suitable for 1,5-radical shift.
As showed in Fig. 8a, the DFT optimized 2-heme-FeII complex evidenced the ability of 2 to interact with FeII by O1 and O2 (FeII -O1=2.40Å and FeII -O2=2.98Å) and, at the same time, to place the double bond close to O1, with C9 correctly positioned for the putative 1,5-radical shift from O1 (O1-C9=3.38Å) (Fig. 3). On the other hand, in the resulting 4a-heme-FeII complex (Fig. 8b) the methoxy oxygen atom (O7) drives the interaction with the metal (FeII-O7 distance=2.28Å), by consequence, 4a presents a different peroxide-FeII coordination mode - including O7, O1, and O2 - from that resulting in the 2-heme-FeII complex (Fig. 8b vs 8a8a).
According to our previous DFT calculations16, the O7/O1/O2 coordination allows 4a to form the oxygen radical on both O1 and O2 (FeII -O1=3.67Å and FeII-O2=3.41Å), although only the O2 radical could be then transferred to a C3 alkyl chain carbon. In the DFT optimized complex the positioning of the C3 butyl chain is indeed optimal for an intra-molecular 1,4-H shift from O2 (O2-HC10=2.69Å), and this is true not only for the lowest energy docked complex (Fig. 8b), but also considering all the low energy complexes generated by the docking procedure (Supplementary Figure S11 and Table S1). The stability of the obtained 4a-heme-FeII complex was further tested by performing a 50ns Molecular Dynamics (MD) calculation at 298K (equilibration phase=20ns, production phase=30ns; see the Methods Section for further details on the computational procedure) (Supplementary Figure S13; Supplementary Table S2). All complexes obtained during the MD production phase presented a distance suitable for the intra-molecular 1,4 H-shift with calculated non-bond interaction energies ranging between -28kcal/mol and -51kcal/mol.
In summary, the resulting bioactive conformations of 2 and 4a can account for the experimentally derived intra-molecular radical shifts described in Figs 3 and and7,7, as well as, for the previously observed SARs of this class of this simple antimalarial endoperoxides (1,2-dioxanes)13,14,15,16,17,18,19. Accordingly, the calculated complexes represent reliable molecular models which can be used, together with the new experimentally obtained data, for future structure optimization.
The results of the present investigation supply new evidences to the antimalarial mechanism of action we previously proposed for plakortins and their synthetic analogues12,13,14,15,16,17,18,19. Upon interaction with heme-FeII-Cl, 2 and 4a undergo the reductive cleavage of the endoperoxide bond through a DET mechanism, with the formation of an oxygen-centered radical which quickly evolves to carbon-centered toxic radicals, the putative “bioactive species” that kill the parasite (Figs 3 and and7).7). In particular, after the reaction of 2 with heme-FeII-Cl, the chlorinated compound 9 and a covalent 2-heme adduct were detected as main products, demonstrating the propensity of the formed carbon radicals to promote inter-molecular reactions. These reactions represent a crucial issue for antimalarial activity; strong evidences have been, indeed, recently provided20 that heme-FeII-activated 1 kills the malaria parasite through a promiscuous targeting mechanism including the covalent binding to 124 protein targets. The observation of 2-heme adduct signals after a few days witnesses it is not a short-lived species in these reaction conditions, the same was observed for the 4a-heme adduct discussed below. The formation of 9, also observed upon FeCl2 reaction12, results by an inter-molecular reaction of the C10-centered radical species, obtained, in turn, through an intra-molecular radical shift from O1 to the C9-C10 double bond (Figs 3 and and8a).8a). The precise configuration obtained at C10, not observed in the reaction with FeCl2, suggests that complexation of 2 with heme makes the two diastereotopic faces of the C10 centered radical differently accessible to the approaching chlorine atom.
This hypothesis is supported by the calculated 2-FeII-heme interaction model, showing that just the Si face of the C9-C10 bond is accessible to inter-molecular reactions (Fig. 8a). If we hypothesize that the chlorine atom approaches C10 by using this face only the 10S stereoisomer would be formed, as indeed found in 9. A higher steric accessibility of both faces is likely to characterize, on the contrary, the C10-centered radical when FeCl2 is used12.
In the case of 4a, a faster reaction with heme was observed (Figs 2b vs. vs.4b).4b). However, the intra-molecular 1,4-radical shift reaction, involving the endoperoxide function and the C3 butyl chain (Figs 7 and and8b),8b), competes with several carbon-carbon bond cleavage reactions. This could explain the lower activity of 4a with respect to 2 despite its higher reactivity. In particular, as predicted by our calculations, the FeII coordination mode of 4a allows the formation of the oxygen radicals on both O1 and O2 producing not only the chlorinated product 17, resulting from the intra-molecular 1,4-hydrogen shift reaction, but also compounds 10–16, obtained by the carbon-carbon bond cleavage reactions (Fig. 7). Nevertheless, similarly to what already observed by Posner et al. for 110,11, the lack of any SAR indicating that the observed carbon-carbon cleavage reactions are responsible for the antimalarial activity, suggests that these reactions could compete with the formation of the toxic radical species by producing carbon radicals which promptly evolve into neutral species through intra-molecular rearrangements (Fig. 7). An important exception is represented by the radical species 19 (generated by the radical shift from O2 to C4), which is responsible for the formation of the heme-adduct 14 (Fig. 7). Interestingly, a pharmacodynamic role of the C4 substituent was recently observed by us when replacing the ester group of 4a with an amino group, and a possible involvement in the putative toxic radical formation/propagation has been evoked19. The C4 substitution determined, indeed, a new and more potent series of analogues showing peculiar SARs compared to the previous derivatives, being able to increase the antimalarial activity independently from the presence and position of the C3/C6 butyl chain(s)19. Thus, the newly obtained results provide a possible explanation for such behavior putting the rational bases for the development of new analogues.
The comparison of the present results with those previously obtained by testing the heme-alkylating ability of 1 and synthetic antimalarial trioxanes and trioxolanes21,22,23,24,25,26,27,28,29,30 supports the hypothesis that the trend to form covalent adducts with heme is related to the antimalarial potency. However, the formation of chlorinated products 9 (Fig. 3) and 17 (Fig. 7) starting from 2 and 4a, respectively, reveals that further alkylating radicals are in action, resulting from the shift of the oxygen radical to a carbon of the alkyl chain. We already reported12,13,14,15,16,17,18,19 that the ability to form such carbon radicals can account for the observed antimalarial potency of 2 and its natural and synthetic derivatives.
It has to be underlined that the putative “bioactive species” of 4a were determined only when 4a was allowed to react with heme, as in the present study, and not when we performed the reaction with FeCl216. Indeed, in the latter case, we identified as major product only the carbon-carbon cleavage product 15 (Fig. 7). This strongly suggests, as reported for 120, that heme, rather than free ferrous iron, is predominantly responsible for antimalarial endoperoxide “bio-activation”, showing that the investigation of the interaction of natural or synthetic compounds with heme by using NMR spectroscopy and MS analyses, could represent an efficient screening method for the evaluation of their antimalarial activity.
During the erythrocyte stage of the Plasmodium the (bio)activation of the endoperoxide function likely occurs by the heme-FeII derived from hemoglobin digestion. The reduction of the endoperoxide bond by agents, such as free heme-FeII, which are not present in normal host cells, represents a key issue for the selective toxicity of these compounds, and, hence, for their potential development as antimalarial drugs. Remarkably, both 23,4 and 4a16 showed no toxicity against human cells and a recent biochemical investigation demonstrated that 2 induces ROS production and cause cellular damages only in infected human erythrocytes39.
Altogether, our results indicate that 2, 4a, and their analogues, similarly to 120, do not interact with a specific protein target but, rather, produce toxic radical species after being reductively activated by heme-FeII. The formation of radicals prone to intermolecular reactions, should represent the key event responsible for a reaction cascade eventually leading to Plasmodium death. Such varied and unspecific mechanism of action is more difficult to be bypassed by parasite resistance than a mechanism of action based on the interaction with a specific protein target.
We have herein presented the first experimental investigation on the interaction of simple antimalarial endoperoxides (1,2- dioxanes) with heme. Obtained results provided new information with respect to the previously reported reaction with FeCl2. The knowledge of further details on the molecular mechanism of action of plakortins and their synthetic analogues constitutes an issue of key importance to continue the investigation on antimalarial endoperoxides. Indeed, only this knowledge can lead to a more detailed comprehension of the structural features responsible for antimalarial activity and their rational optimization. For example, more efficient substituents patterns on the 1,2-dioxane ring could be devised, in order to exploit all possible reactions for the generation of various toxic radical species.
NMR spectra were recorded at 25°C using Inova 400MHz and Inova 600MHz, equipped with a cryogenic probe optimized for 1H detection, spectrometers (Varian Inc., Palo Alto, CA, USA). Chemical shifts were referenced to internal Tetramethylsilane (TMS). Reactions were monitored by mono-dimensional (1D) 1H NMR spectra. Products were characterized by 1D and bi-dimensional (2D) homo- and hetero-nuclear spectra. Homo-nulclear 1H-1H scalar correlations were determined by TOCSY (Total Correlation Spectroscopy) experiments with mixing time of 70ms. One-bond heteronuclear 1H-13C correlations were determined by HSQC (Heteronuclear Single Quantum Coherence) experiments. Two and three-bond heteronuclear 1H-13C correlations were determined by HMBC (Heteronuclear Multiple Bond Correlation) experiments with gradients of refs 2 and 3 J from 5 to 15Hz. Low- and high-resolution ESI-MS spectra were performed on a LTQ Orbitrap XL (ThermoScientific) mass spectrometer.
Ferriprotoporphyrin IX dimethyl ester chloride [hemin-FeIII-Cl, 5] and 2,3-dimethylhydroquinone (DMHQ, 6) are from commercial sources (Livchem Logistics GmbH and Carlo Erba Reagents, respectively). Plakortin (2) was obtained as previously reported40. 3-methoxy-1,2-dioxane derivative (4a) was synthesized, purified and characterized as already reported16.
Plakortin (2, 2.8mg, 9.0μmol) was mixed with the hemin-FeIII-Cl (5, 2.0mg, 2.9μmol) in deuterated chloroform (500μL) in a NMR tube. A 1D 1H NMR spectrum of the obtained solution was acquired before the reductant was added. Large spectral width of 31974.42Hz and short relaxation delay 0.4s were used to visualize the very low-field and broad signals of high-spin ferric hemin. Then, 2,3-dimethylhydroquinone (6, 4.1mg, 29.7μmol) was added as solid. The oxygen at the top of the reaction mixture was replaced with nitrogen gas. The reaction was monitored acquiring 1D 1H NMR spectra every hour for 15h. The following product was identified. Compound 9: 1H and 13C NMR: see data reported in ref. 12 MS: m/z: 349.24 [M]+.
Compound 4a (12.5mg, 48μmol) was mixed with the hemin-FeIII-Cl (5, 10.9mg, 16μmol) in deuterated chloroform (500μL) in a NMR tube. Then, 2,3-dimethylhydroquinone (6, 22.1mg, 160μmol) was added as solid. The oxygen at the top of the reaction mixture was replaced with nitrogen gas. The reaction was monitored for 15h, acquiring 1D 1H NMR spectra after 10, 30 and 60min and after every hour. The following products were identified. Compound 10. 1H NMR (600MHz, CDCl3): δ=3.66 (s, 3H; COOMe); 2.31 (t, J=7.3Hz, 2H; H-2); 1.60–1.62 (m, J=7.3Hz, 2H; H-3); 1.35-1.33 (m, J=7.3Hz, 2H; H-4); 0.91 (t, J=7.3Hz, 3H; H-5). 13C NMR (600MHz, CDCl3): δ=174.45 (C1); 51.43 (COOMe); 33.89 (C2); 27.09 (C3); 22.27 (C4); 13.81 (C5). In agreement with chemical shifts reported in Spectral Database for Organic Compounds (SDBS) No 2890. Compound 11. 1H NMR (600MHz, CDCl3): δ=6.41 (d, J=18.6Hz, 1H; H-3b); 6.15-6.10 (dd, J=18.6, 10.8, H-2); 5.82 (d, J=10.8, 1H; H-3a); 3.76 (s, 3H; COOMe). 13C NMR (600MHz, CDCl3): δ=166.69 (COOMe-1); 130.61 (C3); 128.41 (C2); 51.65 (COOMe). In agreement with chemical shifts reported in SDBS No 1220. Compound 13. 1H NMR (600MHz, CDCl3): δ=3.74 (s, 6H; COOMe); 3.63 (overlapped, 1H; H-2); 2.16 (overlapped, 2H; H-3); 1.24 (s, 6H; H-5 and H-6). 13C NMR (600MHz, CDCl3): δ=170.69 (COOMe-1 and -7); 70.11 (C4); 52.65 (COOMe); 47.72 (C2); 41.77 (C3); 29.38 (C5 and C6). MS: m/z: 227.09 [M+Na]+ Compound 14 (Enol form). 1H NMR (600MHz, CDCl3): δ=2.77 (s, 2H; H-3); 1.60 (s, 6H; H-5 and H-6). 13C NMR (600MHz, CDCl3): δ=194.20, (C1); δ=81.20 (C4); 46.70 (C3); 29.40 (C5 and C6). MS: m/z: 804.28 [M+16]+ Compound 15: 1H and 13C NMR: see data reported in ref. 14a. MS: m/z: 195.06 [M+Na]+ Compound 16. 1H NMR (600MHz, CDCl3): δ=9.77 (s, 1H; H-1); 2.40 (t, 2H; H-2); 1.68 partially overlapped (m, 2H; H-3); 0.97 (t, J=7.3Hz, 3H; H-4). 13C NMR (600MHz, CDCl3): δ=203.09 (C1); 45.96 (C2); 15.75 (C3); 13.80 (C4). Compound 17. 1H NMR (600MHz, CDCl3): δ=3.70 (s, 3H; COOMe-12); 3.65 (1H; H-3); 3.30 (s, 3H; OMe-13); 3.05 (1H; H-6); 1.56 (m, 2H; H-2); 1.32 (s, 6H; H-9 and H-10); 0.92 (d, 3H; H-1). All proton signals was overlapped in the 1H NMR spectrum and were assigned with 2D COSY, HSQC and HMBC spectra. 13C NMR (600MHz, CDCl3): δ=171.44 (COOMe-11); 102.00 (C5); 70.15 (C8); 51.99 (COOMe-12); 48.53 (OMe-13); 41.77 (C3); 34.85 (C2); 29.24 (C9 and C10). MS: m/z: 319.12 and 321.12 (100:33) [M+Na]+. When J values are not indicated, they could not be measured for overlapping.
Molecular modeling calculations were performed on SGI Origin 200 8XR12000 and E4 Server Twin 2X Dual Xeon-5520, equipped with two nodes. Each node: 2X Intel® Xeon® QuadCore E5520-2.26Ghz, 36 GB RAM.
Docking studies were carried out on 4a in complex with heme-FeII, using a Monte Carlo-metropolis/SA docking methodology which considers all the systems flexible (Affinity, SA_Docking; Insight 2005, Accelrys, San Diego, CA)33. Heme apparent pKa values were calculated by using the ACD/Percepta software41. Atomic potentials were assigned by the Heme29.frc42 force field while atomic partial charges were assigned by PM7 calculations43. Two different starting confomers of 4a were selected among those previously calculated16: the one resulting from DFT calculations in complex with FeII and the PM7 global minimum conformer. All atoms in the complex were left free to move during the entire docking procedure with the exception of heme pyrrolic carbons, which were tethered to their original position. Non-bond interactions were calculated using the Cell_Multipole method44. A Monte Carlo/minimization approach for the random generation of a maximum of 20 4a-heme-FeII complexes was applied. During the first step, starting from the previously obtained roughly docked structures, the ligand was moved by a random combination of translation, rotation, and torsional changes (Flexible_Ligand option, considering all rotatable bonds) to sample both the conformational space of the ligand and its orientation with respect to the heme-FeII (MxRChange=1Å; MxAngChange=180°). During this step, van der Waals (vdW) and Coulombic terms were scaled to a factor of 0.1 to avoid very severe divergences in the Coulombic and vdW energies. If the energy of a complex structure resulting from random moves of the ligand was higher by the energy tolerance parameter than the energy of the last accepted structure, it was not accepted for minimization. To ensure a wide variance of the input structures to be successively minimized, an energy tolerance value of 106kcalmol−1 from the previous structure has been used. After the energy minimization step (conjugate gradient, 2500 iterations, ε=80*r), the Metropolis test, at a temperature of 310K, and a structure similarity check (rms tolerance of 0.3kcalÅ−1) were applied to select the 20 acceptable structures. Each subsequent structure was generated from the last accepted structure. All the accepted complexes resulting from the Monte Carlo/minimization approach were subjected to a molecular dynamics simulated annealing protocol, including 5ps of a dynamic run divided in 50 stages (100fs each) during which the temperature of the system was linearly decreased from 500 to 300K (Verlet velocity integrator; time step of 1.0fs). Molecular dynamics calculations were performed using a constant temperature and constant volume (NVT) statistical ensemble, and the direct velocity scaling as temperature control method (temp window, 10K). In the first stage, initial velocities were randomly generated from the Boltzmann distribution according to the desired temperature, while during the subsequent stages initial velocities were generated from dynamics restart data. A temperature of 500K was applied to surmount torsional barriers, thus allowing an unconstrained rearrangement of 4a and heme-FeII (initial vdW and Coulombic scale factors of 0.1). Successively temperature was linearly reduced to 300K in 5ps, and concurrently the scale factors were similarly increased from their initial values (0.1) to their final values (1.0). Distance restraints were applied between FeII and the interacting oxygens in order to avoid unrealistic results during the SA procedure (100kcal/mol/Å). A final round of 105 unconstrained minimization steps (conjugate gradient, ε=80*r) followed the last dynamics steps, and the minimized structures were saved in a trajectory file. After this procedure, the resulting docked structures were ranked by their conformational energy and classified according to the conformational properties of the ligand (i.e., torsional angles). The complex with the best conformational energy was selected as the complex representing the most probable 4a-heme-FeII binding mode.
The selected 4a-heme-FeII complex and the previously calculated12 2-heme-FeII complex were subjected to Density Functional Theory (DFT) optimization. DFT calculations were performed with Gaussian 09 program suite45 by using B3LYP hybrid exchange correlation functional34,35 and LanL2DZ basis set36,37,38. FeII was considered in the quintet spin multiplicity46. Complex optimization was carried out without imposing geometric constraints and vibrational analysis was performed at the same level of theory.
DFT optimized structure of 4a in complex with FeII-heme was subjected to a Molecular Dynamics (MD) calculation using Discover_3 module of Insight 2005 (Accelrys Software Inc., San Diego, CA). MD simulation was carried out in two phases: an equilibration phase of 20ns and a production phase of 30ns. All MD simulation was performed using a constant temperature and constant volume (NVT) statistical ensemble and mimicking the polarity of a water media (ε=80*r). Non-bond interactions were calculated using the Group Based method (CUT_OFF: 100). During the entire MD procedure all atoms in the complex were left free to move whereas a tethering restraint was applied on heme pyrrolic carbons, which were tethered to their original position, and distance restraints (2.5Å–4.0Å) were applied between FeII and the interacting oxygens (10kcal/mol/Å). During the equilibration phase, the 4a-heme-FeII complex was subjected to a 20ns of MD at 298K using a time step of 1fs (Verlet velocity integrator). Initial velocities were randomly generated from the Boltzmann distribution according to the initial temperature (T=10K) and the direct velocity scaling method was used to control the temperature of the system with the Temperature Difference parameter set to 10K. After the equilibration period, the production phase was run for 30ns, using a time step of 0.5fs with a constant temperature of 298K (Verlet velocity integrator). Initial velocities were generated from dynamics restart data and the Nosé–Hoover method was used to control the temperature of the system (Q_ratio=1.0). The coordinates of the system were saved in a trajectory file every 1ns and the non-bond interaction energies of the corresponding complexes were calculated (vdW and electrostatic energy contribution; CUT_OFF=100; Insight 2005 Docking module (Accelrys Software Inc., San Diego, CA)).
How to cite this article: Persico, M. et al. The interaction of heme with plakortin and a synthetic endoperoxide analogue: new insights into the heme-activated antimalarial mechanism. Sci. Rep. 7, 45485; doi: 10.1038/srep45485 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by the following grants: MIUR - PRIN 2008 20084MMXNM, MIUR - PRIN 2015 20154JRJPP_003, POR Campania FESR 2007-2013 FARMABIONET (B25C1300023007), MIUR - FIRB 2012 RBFR12WB3W, and EU Project Bluegenics (Grant 311848). We thank Leopoldo Zona for NMR technical assistance. Dedicated to the beloved memory of prof. Ernesto Fattorusso.
The authors declare no competing financial interests.
Author Contributions B.F. and C.F. designed experimental and computational studies, B.F., M.P., and C.F. wrote the main manuscript text, C.T., M.L., and A.Q. synthesised the compounds, B.F. and I.d.P. prepared the reaction mixtures, B.F. acquired the NMR spectra, B.F., L.Z. E.F. and R.F analyzed the NMR results, G.C. isolated plakortin and run MS spectra, I.d.P., B.F., and O.T.S. analysed the MS spectra, M.P. performed docking studies and SAR analysis., F.R. performed DFT calculations, C.F. and M.P. analysed computational results,. All authors reviewed the manuscript.