Article pubs.acs.org/IC
Solution and Solid State Correlations of Antimalarial Drug Actions: NMR and Crystallographic Studies of Drug Interactions with a Heme Model Erin L. Dodd, Dagobert Tazoo, and D. Scott Bohle* Department of Chemistry, McGill University, Montreal, H3A OB8, Quebec, Canada S Supporting Information *
ABSTRACT: Solution NMR has been used in tandem with a diamagnetic non-iron heme model compound as a simple and effective tool to rapidly probe the structures of the bound complexes formed between the metalloporphyrin and antimalarial drugs from the 4-aminoquinoline, 4-methylenehydroxylquinoline, and 8-aminoquinoline subfamilies. The ability of gallium(III) protoporphyrin IX to mimic heme chemistry is exploited. The 4-aminoquinolines quinacrine and amodiaquine and two novel 3-halo chloroquine analogues are found to bind to the metalloporphyrin through hydrogen-bonding and stacking interactions, while halofantrine and the 4-methylenehydroxylquinolines, quinine and mefloquine bind through the alcohol group of the drug. In each case, detailed structural information is available from the NMR assessment. The mefloquine model is confirmed crystallographically. The 8-aminoquinoline primaquine does not interact strongly. These tools show promise for future applications in assessing antimalarials in preclinical development for heme-binding drug targets.
■
INTRODUCTION Despite past advances in the development of antimalarial therapies, malaria continues to be a global health problem due to the evolution of resistance by the parasite to each new drug in turn. To date, multidrug resistance has been documented in three of the five malaria species known to affect humans in nature: Plasmodium falciparum,1−3 P. vivax, and P. malaria. P. ovale and P. knowlesi also pose a health risk. All of these species produce hemozoin in the blood stage of human infection. Quinoline-based antimalarials have not only been the most successful class of compounds to treat malaria to date, but most are inexpensive, relatively easy to synthesize, and have acceptable levels of toxicity.4 These drugs hold tantalizing targets for novel therapies. In vitro and analytical studies show that most of these compounds interfere with the formation of hemozoin.5−13 The specific mechanism of this interference has been the subject of vigorous debate for many years. Recent years have seen leaps and bounds in our understanding of the underlying chemistry of quinoline-type antimalarials, with two differing modes of action suggested through crystallographic structural studies.14,15 Iron−proton distances determined through solid state paramagnetic NMR of the drug−heme complex precipitated from aqueous solution highlight the lengths to which researchers have gone to try to obtain sensible structural information on this interaction in physiologically relevant conditions16,17 Studies quantifying the strength of interaction between quinoline-based antimalarial drugs and heme species in solution have mostly used spectrophotometric methods to follow the reaction.18−20 Isothermal titration calorimetry has also been used.21 None of these methods allow observation of the © 2017 American Chemical Society
structural details of binding, or offer any means to differentiate between modes of binding of the different quinoline antimalarials. Here we demonstrate the use of NMR and a non-iron model heme compound to probe the structures of the bound complexes formed between the metalloporphyrin and each member of a small library of antimalarial drugs representing three quinoline-based subfamilies, including the 4-methylenehydroxylquinoline group and the non-quinoline halofantrine, whose structures bound to Fe(III) heme have been structurally characterized,14,22,23 and the 4-aminoquinoline group, for which the structural nature of the interaction remains ambiguous. Gallium(III) protoporphyrin IX (Ga(PPIX)) is uniquely suited as a Fe(III) protoporphyrin IX model, being nearidentical in radius, valence, and charge. The more commonly used Zn(II)(PPIX) is also diamagnetic, but its valence and charge lower its affinity for axial coordination to relevant anionic ligands such as propionate and hydroxy. Ga(III) has a full rather than half-full 3d subshell, and the resulting diamagnetism, permits NMR to be used to investigate complex formation in solution.24 An insoluble Ga(PPIX) μ-carboxylatobridged dimer analogous to hematin anhydride has also been isolated.24 The strength of the gallium model has been demonstrated in observing structural details of solution-phase interactions between Ga(PPIX) and the drug chloroquine.15 A solid-state structure of the complex formed between chloroquine and μ-carboxylato-bridged dimeric Ga(PPIX) was determined crystallographically, which may represent at least Received: March 2, 2017 Published: June 26, 2017 7803
DOI: 10.1021/acs.inorgchem.7b00526 Inorg. Chem. 2017, 56, 7803−7810
Article
Inorganic Chemistry
alkoxide.14 This report was followed by reports of the mefloquine,22 quinine, and quinidine23 heme-bound structures showing the same motif. Solid-state paramagnetic 13C NMR of the complex precipitated from aqueous solution supported the biological relevance of this structure, with Fe−O distances consistent with an iron−alkoxide bond.16 In each of the crystalline structures there is a Fe(III)−OR bond in a high spin five coordinate complex with iron 0.5 Å out of the porphyrin plane. Diffraction clarified the significance of the alcohol group in this class of compounds, separating them as a class from the compounds for which the quinoline ring itself is of key structural importance. Great similarity in the structural interactions between drug and heme and with those observed with the gallium model strengthens the argument for the use of the model to describe interactions with the 4-aminoquinoline family, and indeed to extend its use for the study of novel antimalarials targeting hemozoin formation in the parasite. Halofantrine. The structural information afforded by NMR is sufficient to confirm that in solution binding through the alcohol is comparable to that found in the solid state by de Villiers et al.20 In the crystal binding through the hydroxy group is augmented by extensive intermolecular H-bond formation connecting the heme propionates to both the propionate groups of neighboring porphyrin/halofantrine complexes and to the amino group of the halofantrine side chain of the neighboring molecule, as well as to solvate molecules. The gallium protoporphyrin IX−halofantrine coordination/exchange is fast on the NMR time scale, allowing us to discern structural information on the bound complex based on the location of each averaged 1H peak. One key observation was the sharpening of the porphyrin propionate methylene proton signals. These signals are usually observed as broad singlets because of the exchange dynamics of the dimerization/oligimerization which occurs spontaneously in methanol solution for gallium(III) protoporphyrin IX species.26 Upon addition to halofantrine, however, these signals remain as distinct triplets, and the signals from each separate propionate can be distinguished. This strongly supports a 5-coordinate, monomeric bound complex, Ga(PPIX) (halofantrine). These results show quite clearly that the protons that experience the most upfield shift, and therefore must be located closest to the ring current field above the porphyrin plane, are the protons closest to the alcohol group in the halofantrine molecule (Figure 1). This result strongly supports binding through the alcohol in a manner analogous to that seen by de Villiers et al. That our gallium model binds halofantrine in a mode which matches a known structure is encouraging, because it confirms the validity of our model as a probe for investigating the binding of antimalarial drugs to heme and its dimers. Mefloquine. Mefloquine (MQ) forms a 5-coordinate complex with gallium(III) protporphyrin IX via its hydroxyl group. X-ray quality crystals of the Ga(PPIX)-mefloquine complex grow slowly from methanol solution by slow evaporation. Analysis of the determined structure shows binding of the mefloquine molecule directly to the gallium(III) center through the alcohol group of the drug (Figure 2). The structure is centrosymmetric, including both enantiomers of the complex, which are the only isomers seen. The gallium is 0.391 Å out of the plane of the porphyrin, and the quinoline ring system plane is only slightly oblique to the porphyrin plane at an angle of 10.21 deg at an average plane separation of 3.2 Å, suggesting π-stacking interactions of
one possible bound structure of the drug−heme complex in the parasitic digestive vacuole.15 1 H NMR studies of drug−porphyrin interactions are here used to reveal detailed structural information about the change in chemical environments around the drugs upon binding in addition to binding affinity. These tools have proven to be both simple and effective in the observation of the clear emergence of two distinct types of bonding across this library. The differences in binding mode appear consistent across subfamilies of antimalarials.
■
MATERIALS AND METHODS
Protoporphyrin IX dimethyl ester was purchased from Frontier Scientific, Inc. Gallium trichloride was purchased from STREM chemicals. Chloroquine diphosphate (racemate), quinocrine dihydrochloride (racemate), primaquine bisphosphate (racemate), and quinine were purchased from Sigma-Aldrich and prepared as specified below. Quinine was purchased as free base and used without further preparation. Amodiaquine dihydrochloride dihydrate was purchased from Sigma-Aldrich and used without further preparation. Mefloquine hydrochloride (racemate) was kindly provided by the Institute of Parasitology of McGill University. All other reagents were purchased from Sigma-Aldrich and used without further purification. HPLCgrade methanol, HPLC-grade dichloromethane, and double-distilled 2,6-lutidine were purchased from Sigma-Aldrich and used without further purification. NMR-grade methanol-d4 was purchased from Cambridge Isotopes and used without further purification. All volume measurements were performed using Hamilton gastight syringes for accuracy. All single 1H, NOESY, and 1H titration NMR experiments were performed on a 500 MHz Varian Mercury NMR spectrometer. Variable temperature experiments were run on a 500 MHz Varian Mercury NMR spectrometer. Infrared spectroscopy was performed on an ABB Bomem MB series IR spectrometer. NMR spectra were analyzed using MestreNOVA software. Equilibrium constants were determined using WinEQNMR2.25 Gallium(III) protoporphyrin IX hydroxide and gallium(III) octaethylporphyrin chloride (Ga(OEP)Cl) were synthesized by literature methods.24 Preparation of Free Base Antimalarial Drugs. A quantity of the commercially available salt of the drug (10 mg to 1 g) was dissolved in water (200 mL) in a separatory funnel. Sodium hydroxide solution (1 M, 200 mL) was added until all drug precipitated. The suspension was shaken with dichloromethane (200 μL) to extract the free base drug, and the organic layer was separated and dried over anhydrous magnesium sulfate. Extraction with dichloromethane was repeated two more times. The drying agent was filtered and the solvent removed in vacuo. The drug residue was dried at room temperature under high vacuum for 24−48 h in the presence of desiccant (P2O5). NMR Titration of Ga(PPIX) (OH) against Free-Base Drug. A solution of Ga(PPIX)(OH) (0.02 M) was prepared in methanol-d4 (500 μL). Separately, free-base drug (6 mmol) was dissolved in methanol-d4 (500 μL) in an NMR tube. Dichloromethane (2 μL, HPLC-grade) was added as an internal standard. Aliquots (5 μL or appropriate) of metalloporphyrin solution were added to the sample in the NMR tube over the course of the titration, with 1H NMR spectra taken after 20 inversions to obtain homogeneity initially and again upon each addition. The Ga(PPIX)(OH) sample was freshly made, kept dark, prepared immediately before use, and used quickly, as some aggregation occurs over the first few hours at this concentration.
■
RESULTS AND DISCUSSION The 4-Methylenehydroxylquinoline Family − Halofantrine, Mefloquine, and Quinine. The first strong evidence of a large difference between the binding of aminoquinoline drugs and 4-methylenehydroxylquinoline drugs is found in the report of a crystal structure of halofantrine, a non-quinoline-based phenanthrene analogue of quinine, bound to iron(III) protoporphyrin IX through the 7804
DOI: 10.1021/acs.inorgchem.7b00526 Inorg. Chem. 2017, 56, 7803−7810
Article
Inorganic Chemistry
Figure 1. Top: stacked spectra (increasing [Ga(PPIX)(OH)] toward the top), bottom: plot of Δδ of halofantrine phenanthrene ring 1H NMR peaks with increasing Ga(PPIX) concentration. The legend refers to the protons highlighted on the labeled structure.
significant strength. The propionate side chains of the porphyrin are curled upward to hydrogen bond with the piperdinyl amino group of the drug at distances of N(6)−O(4) 2.662 Å and N(6)−O(3) 2.922 Å as part of an extended Hbonding network that connects pairs of adjacent units of the Ga(PPIX)−drug complex and solvated methanol molecules through the crystalline lattice. Evidence of formation of salt bridges between the alkyl side chains of drug and porphyrin have been observed in other structural studies.22,26 The formation of hydrogen bonds within the crystalline structure suggests that the salt bridges formed in the bound structure may interact with protic solvent in solution as well. The C−O bond lengths are within normal range for deprotonated carboxylates with delocalized π-systems, and their orientation is consistent with interaction with two protons of a fully protonated secondary amino group N(6), as is expected from the relative predicted pKa values. The Ga(PPIX)(MQ) structure is nearly identical to the reported structure of Fe(PPIX)(MQ) reported by De Villiers et al.22 Like that structure, the complex crystallizes in a centrosymmetric space group, with both enantiomers present in the crystalline lattice and related by inversion symmetry. The difference between the two structures is seen in the crystalline packing, with the Ga(PPIX)(MQ) structure being much less densely packed, and with an extended network of hydrogen bonding interactions between neighboring molecules as well as within each complex (Figures S3 and S4). The methanol solvent of crystallization takes part in this network as well, with one methanol molecule per Ga(PPIX)(MQ) unit. The difference in the crystallization packing preferences is almost
Figure 2. ORTEP diagram of Ga(PPIX)(MQ), with 40% thermal ellipsoids. Carbon-bound hydrogens are omitted for clarity. Key metric parameters (Å) include Ga−O(5) 1.863(6); O(1)−C(23) 1.254(15); O(2)−C(34) 1.396(16); O(3)−C(34) 1.225(15); O(4)−C(23) 1.280(15).
certainly determined by the solvent from which the complexes were crystallized. The coordination/exchange binding of mefloquine to Ga(PPIX)(OH) in the solution phase is slow on the NMR time scale at 20 °C, leading to broad signals. A slight upfield shift of the quinoline ring protons is evident, and broadening is much more pronounced for the ring protons nearest the alcohol group. Binding through the alcohol is again supported by the shift and extreme broadening of the signals of the protons nearest the alcohol functional group, indicating that the solid state structure is likely present in solution, consistent with the solution-phase studies reported for the heme-MQ complex.22 In particular, the proton geminal to the alcohol group experiences a downfield shift from its initial point (overlapped with the internal standard) and considerable broadening. Porphyrin protons also show broadening, and the β-protons of one of the porphyrin vinyl groups (but not both!) are shifted and broadened, indicating dynamic exchange for one but not the other. This suggests a bound structure with a preferred binding diastereomer. Given that both enantiomers of the drug are present, a pair of enantiomeric diastereomers are present. Hydrogen bonding between the porphyrin propionic acid 7805
DOI: 10.1021/acs.inorgchem.7b00526 Inorg. Chem. 2017, 56, 7803−7810
Article
Inorganic Chemistry
Figure 3. NOESY of a 1:1 mixture of Ga(PPIX)(OH) and quinine shows negative exchange peaks for the methine protons of two distinct states of Ga(PPIX) in exchange. Other cross peaks indicate through-space NOE between porphyrin methine and vinyl groups (in blue circle), likely from the same porphyrin unit.
group and the amino group on the drug’s piperidinyl group is a probable cause. The nearer α-protons of the vinyl groups are both too broadened to observe. Quinine. The heme-bound structure of quinine has also been reported, alongside spectrophotometric determination of the bound complex in solution.23 Here we find support for quinine binding to Ga(PPIX) through the alcohol group by 1H NMR, with complex formation slow on the NMR time scale and clear resolution of a separate set of signals for the bound complex. The shift of unbound quinoline signals indicates that protonation of quinine occurs, and its coordination is in the fast exchange regime of NMR. The observation of a new set of porphyrin methine protons far upfield of their normal positions is indicative of these protons being influenced by the π-cloud of an aromatic system in close physical proximity. This pattern is analogous to that observed in the formation of μ-oxo27 and μ-fluoro24 gallium protoporphyrin IX dimers. Each of these signals give rise to exchange crosspeaks in a NOESY experiment of a mixture of Ga(PPIX)(OH) and quinine at near 1:1 molar ratio (Figure 3). NOESY spectra also indicate that both these species of Ga(PPIX) give NOE crosspeaks with what appear to be the “bound” quinoline 1H signals (Figures 4 and S7), observable as NOESY exchange peaks between bound and unbound quinine molecules. These are most clearly observable through the signals for the protons from the alkyl quinuclidine ring, seen at well below 0 ppman upfield shift of over 4 ppm. The magnitude of peak displacement is proportional to the position of the proton within the porphyrin ring current, supporting a structure for the bound complex which is in good agreement with that reported by de Villiers et al.23 The quinuclidine ring of bound quinine is positioned with its vinyl group at a maximum distance from the porphyrin, minimizing steric interactions. Protons 14a and 11a experience the most negative displacement in the bound state, suggesting they are the closest
Figure 4. NOESY exchange peaks between the bound and unbound quinine protons from the quinuclidine portion of the spectrum allow for structural analysis of the preferred orientation of the aliphatic ring above the porphyrin in the bound state. Protons experiencing the greatest upfield shift are considered nearest to the porphyrin plane, consistent with the expected effect of porphyrin extended aromatic ring current.
to the center of the porphyrin ring current. Protons 10, 13a, and 13b are much less displaced and are probably closer to the center of the porphyrin ring, where the ring current experiences 7806
DOI: 10.1021/acs.inorgchem.7b00526 Inorg. Chem. 2017, 56, 7803−7810
Article
Inorganic Chemistry a node. A matching pattern is observed for the quinoline ring portion of the molecule (Figure S7), with H(3) displaced upfield by over 6 ppm upon binding. We have depicted the bound structure as hydrogen-bonded to a propionate group of the porphyrin because this is consistent with the positioning of the quinuclidine in a single bound state, which would require a structural feature favoring one single position of the bound quinine over others. A hydrogen bond is predicted under these conditions and would definitely provide such a structural feature. Repetition of this experiment with the simpler Ga(OEP)Cl was inconclusive. Slight broadening of 1H signals of both species was observed, but no NOE signal was detected (Figure S8). The absence of propionate side chains on the porphyrin means no orienting preference of the drug to the porphyrin face could predominate, and also binding strength could not be augmented by intramolecular H-bonding. It is possible that binding occurred between these species to a small degree in methanol solution. This result highlights the importance of the salt-bridge formation in the formation of these complexes. The same pattern was observed in the titration of halofantrine against Ga(OEP)Cl (Figure S9). The 4-Methylenehydroxylquinoline Antimalarials Conclusions. It becomes strikingly obvious that the rate of exchange for the formation of the drug−Ga(PPIX) complex increases in the order quinine < mefloquine < halofantrine, with halofantrine being fastest, i.e., the most labile. This observation clarifies the difficulty in interpreting the results for the other two drugs; in the cases of quinine and mefloquine, the observed peaks are neither fully averaged nor fully resolved. Rather, each equilibrium exists near the cusp of the definition of fast or slow exchange based upon the NMR time scale. For mefloquine, this results in very broad peaks, while for quinine the observed peaks are not sharp, and assignment of the bound complex peaks is difficult. It is well understood at this point that this class of drugs binds to the metal center of heme through the alcohol group, and our work indicates that these interactions are maintained in methanol solution and suggests that they would be in aqueous environments as well. The gallium model emphasizes the importance of understanding the salt bridges and intramolecular hydrogen bonding between drug and porphyrin side chains in predicting antimalarial activity. 4-Aminoquinoline Family. We have previously described the binding of chloroquine to propionate-linked dimeric gallium(III) protoporphyrin IX,15 and we have repeated our solution-phase studies to explore the interactions of gallium(III) protoporphyrin IX (Ga(PPIX)(OH)) in methanol solution with chloroquine’s closest structural homologues. Two 4-aminoquinoline antimalarial drugs structurally similar to chloroquine, quinacrine (QC) and the 4-aminoquinoline amodiaquine (AQ), were both titrated against Ga(PPIX)(OH) to observe characteristics of the structure of the metalloporphyrin-bound complexes. Quinacrine. Quinacrine reacts with Ga(PPIX)(OH) in a dynamic equilibrium which is fast on the NMR time scale. In the region of the acridine ring, we see the 1H NMR signal of the ring protons nearest the ring nitrogen exhibits a large upfield shift as the amount of metalloporphyrin in solution is increased, while the other ring protons remain minimally perturbed (Figure 5) in a pattern exactly analogous to that seen with chloroquine, indicative of close proximity between these protons and the strong ring current of the aromatic porphyrin ring. Side chain interactions,
Figure 5. Top: stacked spectra of quinacrine (acridine ring) Ga(PPIX) titration. Note dramatic change of chemical shifts. Bottom: plot of Δδ with increasing Ga(PPIX) mole fraction.
as well, are analogous to those seen in chloroquine, with the largest 1H signal displacements seen for protons of the terminal ethyl groups of the drug, and a corresponding upfield shift of over 0.35 ppm for the porphyrin methine H(20) proton. Together these results strongly suggest a structure of the QCGa(PPIX) bound complex that closely resembles that of the CQ-Ga(PPIX) complex. The degree of dimerization, if present, is ambiguous. Amodiaquine. Binding was observed with the acidic amodiaquine dihydrochloride dihydrate salt, which is soluble in methanol while the free base was not. The overall displacement of chemical shifts of the quinoline ring region of amodiaquine was small compared to that of free-base chloroquine, and the binding was found to be much weaker, as in this case the solution was acidic overall. The pattern of the 1 H NMR chemical shift displacements mirrored that of chloroquine, with the proton signals for H(2) and H(8) shifting upfield and broadening more than the other signals. The side chains of the drug and the propionic acid groups of the porphyrin also exhibit behavior similar to that observed in chloroquine. The proton signals for each of these regions broaden significantly, but of course lack the observed shift associated with proton transfer. As with chloroquine and quinacrine, this pattern is indicative of protons which are brought into a chemical environment which is closer to the ring current of the porphyrin, and suggests strongly that 7807
DOI: 10.1021/acs.inorgchem.7b00526 Inorg. Chem. 2017, 56, 7803−7810
Article
Inorganic Chemistry
that these drugs all induce dimerization in Ga(PPIX) species in the way that chloroquine does. In Table 1 the combined results of the binding constants of 4-aminoquinoline based drug to gallium porphyrin are listed with the pKa’s of the drugs. One trend is immediately observable, which is that the binding constant of drug to metalloporphyrin increases with increasing basicity of the quinoline ring nitrogen. The Kassociation values obtained for binding to gallium(III) protoporphyrin IX in methanol are seen to be consistently weaker by an order of magnitude than those observed for iron species in various aqueous media.11,20,30 All are relatively weak binding interactions, and yet these weak interactions are enough to prevent crystallization of hematin anhydride and/or to destabilize aggregates. In a dynamic equilibrium of aggregates and solvated compound, transient interactions which increase the solubility of each heme molecule momentarily are expected to disturb the entire equilibrium even without permanently binding to the metalloporphyrin. To date, no single crystal structures have been determined for any complex of a 4-aminoquinoline antimalarial with Fe(III) heme in monomeric, μ-oxo dimeric, or propionate-linked dimeric form. The axial lability and accessibility of soluble dimer in methanol solutions of Ga(PPIX)24 are a key aspects of this model system. Previous studies based on paramagnetic relaxation measurements of 1H NMR shifts in complexes formed between chloroquine and the soluble μ-oxo dimer of Fe(III)(PPIX)31,32 lacked this flexibility, and conclusions reached were limited to the conformations possible as modulated by a μ-oxo scaffold. Nonetheless these studies provide powerful support of a plane−plane orientation of quinoline ring to porphyrin ring. The involvement of the ring N of chloroquine covalent interaction at a 2-bond distance from the Fe is consistent with contact shifts observed in solid state 13 C and 15 N NMR work previously reported. 17 The diamagnetic gallium model proposes a structure that is consistent with the above observations and has emphasized the effect of the formation of salt bridges between drug and porphyrin side chains on enhancing the solubility of the propionate-linked dimer of Ga(PPIX). The 8-Aminoquinoline family. Primaquine. Primaquine is structurally unable to form the same orientation as chloroquine when binding to metalloprotoporphyrin and lacks the basicity of chloroquine’s quinoline ring nitrogen. Primaquine is able to target the parasite in its exoerythrocytic liver stage, known as the hypnozoite stage,33 which other drugs are not capable of doing, thereby killing all latent stages of the parasite and preventing relapse of disease.34 Inclusion of primaquine as a negative control was warranted since primaquine is not efficacious in the blood stage of the parasite’s cycle. From the 1H NMR data obtained for primaquine free base, we see protonation of both chain nitrogens upon addition of Ga(PPIX)(OH), but no actual drug binding−primaquine proton peak shifts were very small and characteristic of protonation only.
amodiaquine has a similar mechanism of binding metalloporphyrins to that of chloroquine. What is not seen in the case of amodiaquine is the upfield shift of porphyrin methine proton H(20), which is the proton between the two propionic acid groups. Rather, in this case, methine H(20) is seen to give a signal which is displaced by a mere 0.02 ppm, but significantly broadened. This may indicate that the binding takes place in the absence of porphyrin deprotonation in the acidic solution. Here we have a completely acidic system in which we see binding of drug in the absence of the base that is presumed to catalyze the heme dimerization process.15 Biologically, the digestive vacuole of the malaria parasite is thought to be acidic, even when infused with antimalarial drug up to millimolar concentrations.28 A small amount of dimerized [Ga(PPIX)]2 exists in methanol solution in acidic conditions,24 and that increased dimerization is observed in the presence of a large excess of pyridine.29 It is evident that even a very weakly binding drug in acidic conditions is enough to solubilize the porphyrin as a drug−porphyrin complex, which makes a strong case for biological relevance of this model. Two Novel 4-Aminoquinoline Potential Antimalarials. The novel chloroquine analogues 3-bromochloroquine and 3iodochloroquine were found to be active in vitro against nonchloroquine-resistant strains of P. falciparum, although chloroquine resistance conferred resistance to these compounds as well.32 Our gallium(III) protoporphyrin IX system efficiently predicted the mechanism of binding of these molecules to hematin anhydride, showing the promise of this methodology when extended to new antimalarials in development. Despite overall similarity to the chloroquine structure, the large halogen in the (3) position on the quinoline ring was expected to interfere with the free rotation of the drug side chain due to sterics. While the crystal structure of the chloroquine−[Ga(PPIX)]2 dimer complex shows a drug side chain oriented away from this position, the steric effects of the added halogen atom could still have an effect on the solutionphase binding. This was not seen. The binding of 3iodochloroquine and 3-bromo-chloroquine to Ga(PPIX)(OH) were explored using 1H NMR titrations and found to show binding behavior exactly analogous to that of chloroquine; however, the overall binding was an order of magnitude weaker (Table 1). Table 1. Binding Constants of 4-Aminoquinoline Drugs to Gallium(III) Protoporphyrin IX Hydroxide log Kassociation to Ga(PPIX)(OH) chloroquine quinacrine amodiaquine HCl 3-bromochloroquine 3-iodochloroquine
4.17 ± 0.02 3.09 ± 0.06 2.3 ± 0.7 2.48 ± 0.13 2.41 ± 0.05
■
4-Aminoquinoline Family Conclusions. The observation of the binding of each of these drugs to Ga(PPIX) species in solution by 1H NMR suggests a bound structure very similar to that formed by chloroquine and Ga(PPIX) reciprocal dimer. The binding consistently involves stacking of the quinoline ring over the porphyrin with the ring nitrogen directed closest to the metal center, and interactions between the side chains of both molecules. It is highly likely based on these observations
CONCLUSION To summarize, we have compiled a small library of quinolinebased antimalarial drugs and researched structural changes observable upon binding for each case. We have used solution NMR and fluorescence15 and a noniron model heme compound to probe the mechanisms of action of a small library of known the quinoline-based and 7808
DOI: 10.1021/acs.inorgchem.7b00526 Inorg. Chem. 2017, 56, 7803−7810
Inorganic Chemistry
■
related antimalarial drugs. These tools have proven to be both simple and effective in the observation of the clear emergence of two distinct types of bonding across this library. The 4aminoquinolines appear to promote dimerization of the gallium(III) protoporphyrin IX in methanol solution, and bind to the metalloporphyrin through hydrogen-bonding and stacking interactions. The strength of these binding interactions is comparable to those observed by other means with ironbased heme compounds. The 4-methylenehydroxylquinolines and halofantrine bind to gallium(III) protoporphyrin IX through the alcohol group of the drug as they do with iron(III) protoporphyrin IX. In light of the strong evidence for metal binding through the hydroxy group found by de Villiers et al.,14 the usefulness of our gallium model against ferriprotoporphyrin IX itself was supported by verification of similar modes of binding of gallium protoporphyrin IX to 4-methylenehydroxylquinoline compounds. The changes in the mefloquine and quinine 1H NMR spectra upon addition of Ga(PPIX)(OH) correspond well with bonding through the 4-methylenehydroxyl alkyl group. These distinct patterns of heme binding effectively separate the quinoline antimalarials into two subgroups, the first being similar to chloroquine15 and the later being similar to halofantrine as first proposed elsewhere.14 Each of these subsets contains key structural features now clearly identified, presenting informed targets for future drug development. In particular, these results suggest that, following upon the emergence of chloroquine as an anticancer therapeutic35 and evidence that the binding motif of chloroquine extends to nonheme systems,36 the other 4-aminoquinoline antimalarials may also find new life as new tools against cancer. The successand easeof the use of gallium(III) protoporphyrin IX NMR binding studies in predicting the heme-binding patterns of known antimalarials makes us optimistic for the future extension of this methodology toward the preclinical testing of emerging antimalarials. It is encouraging to note that, in fixating on the quinoline ring in the development of the range of drugs described in this work, researchers unwittingly opened doors to the discovery of many varied mechanisms to disrupt hemozoin aggregation and crystallization. This suggests that the formation of hemozoin, a process unique to the parasite, is a fragile process by nature, and is readily interrupted by the introduction of what could become, in the future, a very wide range of heme-binding intercalators.
■
Article
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
D. Scott Bohle: 0000-0001-5833-5696 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We gratefully acknowledge NSERC and the CRC for their generous support of this research.
■
REFERENCES
(1) Wellems, T. E.; Plowe, C. V. Chloroquine-Resistant Malaria. J. Infect. Dis. 2001, 184 (6), 770−776. (2) Martin, R. E.; Marchetti, R. V.; Cowan, A. I.; Howitt, S. M.; Bröer, S.; Kirk, K. Chloroquine Transport via the Malaria Parasite’s Chloroquine Resistance Transporter. Science 2009, 325 (5948), 1680− 1682. (3) Foote, S. J.; Kyle, D. E.; Martin, R. K.; Oduola, A. M.; Forsyth, K.; Kemp, D. J.; Cowman, A. F. Several alleles of the multidrug-resistance gene are closely linked to chloroquine resistance in Plasmodium falciparum. Nature 1990, 345, 255−258. (4) O’Neill, P. M.; Bray, P. G.; Hawley, S. R.; Ward, S. A.; Park, B. K. 4-Aminoquinolines-past, present, and future: a chemical perspective. Pharmacol. Ther. 1998, 77, 29−58, DOI: 10.1016/s0163-7258(97) 00084-3. (5) Loria, P.; Miller, S.; Foley, M.; Tilley, L. Inhibition of the peroxidative degradation of haem as the basis of action of chloroquine and other quinoline antimalarials. Biochem. J. 1999, 339 (2), 363−370. (6) Sullivan, D. J.; Matile, H.; Ridley, R. G.; Goldberg, D. E. A common mechanism for blockade of heme polymerization by antimalarial quinolines. J. Biol. Chem. 1998, 273, 31103−31107. (7) Sullivan, D. J.; Gluzman, I. Y.; Russell, D. G.; Goldberg, D. E. On the molecular mechanism of chloroquine’s antimalarial action. Proc. Natl. Acad. Sci. U. S. A. 1996, 93 (21), 11865−11870. (8) Solomonov, I.; Osipova, M.; Feldman, Y.; Baehtz, C.; Kjaer, K.; Robinson, I. K.; Webster, G. T.; McNaughton, D.; Wood, B. R.; Weissbuch, I.; Leiserowitz, L. Crystal Nucleation, Growth, and Morphology of the Synthetic Malaria Pigment β-Hematin and the Effect Thereon by Quinoline Additives: The Malaria Pigment as a Target of Various Antimalarial Drugs. J. Am. Chem. Soc. 2007, 129, 2615−2627, DOI: 10.1021/ja0674183. (9) Cohen, S. N.; Phifer, K. O.; Yielding, K. L. Complex formation between chloroquinone and ferrihemic acid in vitro, and its effect on the antimalarial action of chloroquinone. Nature (London, U. K.) 1964, 202, 805−6, DOI: 10.1038/202805a0. (10) Chai, A.; Chevli, R.; Fitch, C. Ferriprotoporphyrin IX Fulfills the Criteria for Identification as the Chloroquine Receptor of Malaria Parasites. Biochemistry 1980, 19 (8), 1543−1549. (11) Dorn, A.; Vippagunta, S. R.; Matile, H.; Jaquet, C.; Vennerstrom, J. L.; Ridley, R. G. An Assessment of Drug-Haematin Binding as a Mechanism for Inhibition of Haematin Polymerisation by Quinoline Antimalarials. Biochem. Pharmacol. 1998, 55 (6), 727−736. (12) Egan, T. J. Haemozoin (malaria pigment): a unique crystalline drug target. Targets 2003, 2, 115−124. (13) Egan, T. J.; Ncokazi, K. K. Quinoline antimalarials decrease the rate of [beta]-hematin formation. J. Inorg. Biochem. 2005, 99 (7), 1532−1539. (14) de Villiers, K. A.; Marques, H. M.; Egan, T. J. The crystal structure of halofantrine−ferriprotoporphyrin IX and the mechanism of action of arylmethanol antimalarials. J. Inorg. Biochem. 2008, 102 (8), 1660−1667. (15) Dodd, E. L.; Bohle, D. S. Orienting the heterocyclic periphery: a structural model for chloroquine’s antimalarial activity. Chem. Commun. 2014, 50 (89), 13765−13768.
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.7b00526. Additional figures, crystallographic data, and tables are supplied in the supplementary data (PDF) Accession Codes
CCDC 1545184 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_
[email protected], or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033. 7809
DOI: 10.1021/acs.inorgchem.7b00526 Inorg. Chem. 2017, 56, 7803−7810
Article
Inorganic Chemistry (16) Alumasa, J. N.; Gorka, A. P.; Casabianca, L. B.; Comstock, E.; de, D. A. C.; Roepe, P. D. The hydroxyl functionality and a rigid proximal N are required for forming a novel non-covalent quinineheme complex. J. Inorg. Biochem 2011, 105, 467−475. (17) de Dios, A. C.; Tycko, R.; Ursos, L. M. B.; Roepe, P. D. NMR Studies of Chloroquine−Ferriprotoporphyrin IX Complex. J. Phys. Chem. A 2003, 107 (30), 5821−5825. (18) Bohle, D. S.; Dodd, E. L.; Kosar, A. J.; Sharma, L.; Stephens, P. W.; Suárez, L.; Tazoo, D. Soluble Synthetic Analogues of Malaria Pigment: Structure of Mesohematin Anhydride and its Interaction with Chloroquine in Solution. Angew. Chem., Int. Ed. 2011, 50 (27), 6151−6154. (19) Egan, T. J.; Mavuso, W. W.; Ross, D. C.; Marques, H. M. Thermodynamic factors controlling the interaction of quinoline antimalarial drugs with ferriprotoporphyrin IX. J. Inorg. Biochem. 1997, 68, 137−145. (20) Egan, T. J.; Hempelmann, E.; Mavuso, W. W. Characterization of synthetic β-haematin and effects of the antimalarial drugs quinidine, halofantrine, desbutylhalofantrine and mefloquine on its formation. J. Inorg. Biochem. 1999, 73, 101−107. (21) Bachhawat, K.; Thomas, C. J.; Surolia, N.; Surolia, A. Interaction of chloroquine and its analogues with heme: An isothermal titration calorimetric study. Biochem. Biophys. Res. Commun. 2000, 276, 1075− 1079. (22) Gildenhuys, J.; Sammy, C. J.; Muller, R.; Streltsov, V. A.; le Roex, T.; Kuter, D.; de Villiers, K. A. Alkoxide coordination of iron(iii) protoporphyrin IX by antimalarial quinoline methanols: a key interaction observed in the solid-state and solution. Dalton Transactions 2015, 44 (38), 16767−16777. (23) de Villiers, K. A.; Gildenhuys, J.; le Roex, T. Iron(III) Protoporphyrin IX Complexes of the Antimalarial Cinchona Alkaloids Quinine and Quinidine. ACS Chem. Biol. 2012, 7 (4), 666−671. (24) Bohle, D. S.; Dodd, E. L.; Pinter, T. B. J.; Stillman, M. J. Soluble Diamagnetic Model for Malaria Pigment: Coordination Chemistry of Gallium(III)protoporphyrin-IX. Inorg. Chem. 2012, 51, 10747. (25) Hynes, M. J. EQNMR: a computer program for the calculation of stability constants from nuclear magnetic resonance chemical shift data. J. Chem. Soc., Dalton Trans. 1993, 2, 311−312. (26) Gorka, A. P.; de Dios, A.; Roepe, P. D. Quinoline Drug−Heme Interactions and Implications for Antimalarial Cytostatic versus Cytocidal Activities. J. Med. Chem. 2013, 56 (13), 5231−5246. (27) Parzuchowski, P. G.; Kampf, J. W.; Rozniecka, E.; Kondratenko, Y.; Malinowska, E.; Meyerhoff, M. E. Gallium(III) and indium(III) octaethylporphyrin dimeric complexes with a single [mu]-hydroxo bridge: synthesis, structure and stability in anion-containing organic media. Inorg. Chim. Acta 2003, 355, 302−313. (28) Geary, T. G.; Jensen, J. B.; Ginsburg, H. Uptake of [3H]chloroquine by drug-sensitive and -resistant strains of the human malaria parasite Plasmodium falciparum. Biochem. Pharmacol. 1986, 35, 3805−3812, DOI: 10.1016/0006-2952(86)90668-4. (29) Bohle, D. S.; Dodd, E. L. [Gallium(III) protoporphyrin IX]2: A Soluble Diamagnetic Model for Malaria Pigment. Inorg. Chem. 2012, 51 (8), 4411−4413. (30) Egan, T. J.; Mavuso, W. W.; Ross, D. C.; Marques, H. M. Thermodynamic factors controlling the interaction of quinoline antimalarial drugs with ferriprotoporphyrin IX. J. Inorg. Biochem. 1997, 68 (2), 137−145. (31) Leed, A.; DuBay, K.; Ursos, L. M. B.; Sears, D.; de Dios, A. C.; Roepe, P. D. Solution Structures of Antimalarial Drug−Heme Complexes†. Biochemistry 2002, 41 (32), 10245−10255. (32) Schwedhelm, K. F.; Horstmann, M.; Faber, J. H.; Reichert, Y.; Büchner, M.; Bringmann, G.; Faber, C. Spin State of ChloroquineHeme Complexes: Formation of a Hemin Tetramer Adduct. Open Spectrosc. J. 2008, 2, 10−18. (33) Markus, M. The hypnozoite concept, with particular reference to malaria. Parasitol. Res. 2011, 108 (1), 247−252. (34) Baird, J. K.; Rieckmann, K. H. Can primaquine therapy for vivax malaria be improved? Trends Parasitol. 2003, 19 (3), 115−120.
(35) Solomon, V. R.; Lee, H. Chloroquine and its analogs: A new promise of an old drug for effective and safe cancer therapies. Eur. J. Pharmacol. 2009, 625 (1−3), 220−233. (36) Leung, K. K. K.; Shilton, B. H. Chloroquine Binding Reveals Flavin Redox Switch Function of Quinone Reductase 2. J. Biol. Chem. 2013, 288 (16), 11242−11251.
7810
DOI: 10.1021/acs.inorgchem.7b00526 Inorg. Chem. 2017, 56, 7803−7810
