Background
Despite encouraging levels of progress in international control efforts, as many as 3.3 billion people in the world are at continued risk for malaria infection [
1]. While
Plasmodium falciparum exacts a greater burden in mortality and morbidity, the impact of
Plasmodium vivax is also significant. It has been estimated that 40% of the world’s population are at risk of vivax malaria and, on the whole, more people are at risk of vivax than falciparum malaria [
2,
3].
A key component of continued control and eradication efforts is the development of effective drugs for treatment and prophylaxis. Only the 8-aminoquinoline (8-AQ) class of compounds have demonstrated the ability to target the key survival stages of the parasite — the sleeping liver stages, or hypnozoites, of
P. vivax and
Plasmodium ovale, and Stage V gametocytes of
P. falciparum, with primaquine (PQ) being the only drug from this class in clinical use. Unfortunately, the use of 8-AQs is limited by their tendency to cause haemolytic anaemia in individuals with a genetic deficiency in glucose-6-phosphate dehydrogenase (G6PD), an enzyme implicated in the body’s defence against oxidative stress [
4].
Although in use for several decades, PQ’s mechanisms of efficacy and toxicity are not well understood and its metabolic profile has not been fully elucidated. These mechanisms of efficacy/toxicity are believed to involve the formation of reactive oxygen species or interference by PQ and/or its metabolite(s) with electron transport in the parasite [
5]. Further, it is generally believed that PQ’s haemolytic toxicity is due to one or more metabolites and not the parent compound [
6]. For example, Link
et al. show direct methaemoglobin (MHb) formation in canine hemolysates and purified human oxyhaemoglobin (Hb) upon exposure to the putative PQ metabolite 5-hydroxyprimaquine (5-HPQ) [
7,
8]. Further, Ganesan
et al. have recently demonstrated, in a human erythrocyte-based model of PQ toxicity, the ability of several CYPs, most notably 3A4, 2D6, and 2B6, to form reactive oxygen species resulting in generation of methaemoglobin [
9]. MAOs have also been implicated in PQ metabolism and may play a role in carboxyprimaquine (CPQ) formation [
10].
Elucidation of the fundamentals of PQ metabolism, including the number, type, and relative contribution of involved metabolic enzymes, as well as the metabolites produced in each pathway, will help to determine if the drug’s efficacy is inextricably linked to its toxicity in G6PD-deficient individuals, or if future drug design efforts could overcome toxicity, and potentially enhance efficacy, by directing metabolism. To this end, a phenotyping study using the relative activity factor (RAF) method of Crespi
et al. was completed to determine the number and type of mixed function oxidases involved in PQ metabolism and the relative contribution of each [
11]. Overlap, if any, between the CYPs implicated in toxicity by Ganesan
et al. and any CYPs shown to metabolize PQ to a great extent could help direct future metabolism studies toward ultimately resolving the question of which enzyme and which metabolite(s) is ultimately responsible for efficacy and observed haemolytic effects [
9].
Methods
Chemicals
Chemicals used were: primaquine diphosphate (Sigma, # 160393), nicotinamide adenine dinucleotide phosphate, oxidized form (NADP) (Sigma, # 077K7000), acetonitrile (Fisher Scientific, # 972970), glucose-6-phosphate (G6P) (Sigma, # 046K3779), glucose-6-phosphate dehydrogenase (G6PD) (Sigma, # 068K3795), and magnesium chloride (MgCl2) (Sigma, # 102K0154). The mobile phase for LC separations consisted of acetonitrile, distilled water and formic acid; all were of HPLC-grade.
Isoenzymes
Baculovirus-insect cell expressed recombinant Cytochrome-P450 Supersomes® containing CYP-1A2 (BD Biosciences, Billerica, MA, USA, # 18682), 2D6 (BD, # 16902), 3A4 (BD, # 24175), 2C9 (BD, # 25315), 2C19 (BD, # 15466), MAO-A (BD, # 31000), MAO-B (BD, # 36152), were stored at −80°C until required and were rapidly thawed by submersion in a 37°C water bath before use.
Enzyme activity screening
Cofactor concentrations were as follows in all experiments: 1 U/ml G6PD, NADP 1.3 mM, G6P and MgCl2 3.3 mM, and 0.5 mg/mL CYP, MAO-A, MAO-B in 0.1 mM pH 7.4 phosphate buffered saline (PBS). Final reaction volume was 1 mL. Reaction mixtures containing all cofactors and enzymes were pre-warmed for 10 min at 37°C, and reactions were started with the addition of primaquine. Aliquots of 250 μL were quenched after 2 h with an equal volume of cold acetonitrile. The resulting samples were centrifuged at 13,000 rpm for 10 min and the supernatant collected for analysis. Each experiment was conducted with an n of four to eight. Error was represented as the standard deviation.
Primaquine analysis
LC-MS analysis for phenotyping studies and enzyme activity were performed using a Flux Rheos 2000 pump (Flux Instruments AG, Switzerland) coupled with a CTC PAL autosampler (LEAP Technologies, Carrboro, NC, USA) and a Micromass ZQ (now Waters Corp., Milford, MA, USA) mass spectrometer. Initial gradient conditions were 98% water and 2% acetonitrile, each with 0.1% formic acid. Organic content was raised from 2% to 40% over 3 min before returning to the initial conditions for equilibration for subsequent injections. The MS method detected primaquine (parent ion) by positive ion electrospray ionization at a m/z of 259.88. Analytical separations were achieved using a Waters X-Terra RP 5 cm x 2 mm, 3 μm C18 column, with a flow rate of 300 ml/min.
RAF determination
Sample stocks of testosterone (3A4) (Sigma, # 160393), bufuralol (2D6) (Sigma, # UC168), S-mephenytoin (2 C19) (Sigma, # UC175), and serotonin (MAO-A) (Sigma, # H9523) at 10 mM in DMSO were diluted to a final concentration of 1 μM into a mixture containing 0.5 mg/mL of pre-incubated pooled human liver microsomes (BD Biosciences, Billerica, MA, USA) or recombinant enzyme of interest, 1.3 mM NADP, 3.3 mM MgCl2, and 0.1 M pH 7.4 PBS using a TECAN Genesis RSP 150 (Tecan, Durham, NC, USA) robotic liquid handler. The reaction was started with the addition of 1 U/mL G6PD. The mixture was incubated on a shaking platform at 37°C, and aliquots were taken and quenched with the addition of an equal volume of cold acetonitrile at 0, 10, 20, 30 and 60 min. Samples were centrifuged at 3,700 rpm for 10 min at 20°C to remove debris. Sample quantification was carried out by LC/MS. RAF was calculated using the equation:
[
11].
Kinetic studies
Incubations containing cofactors, primaquine (ranging from 0 μM to 40 μM), and enzyme (CYP, MAO-A, MAO-B) were treated as above, but quenched after 30 min (to remain within linear range of activity) with an equal volume of cold acetonitrile. Each experiment was conducted with an n of four to eight. Experimental data were fit to the Michaelis-Menten equation using non-linear least squares approximation. Error was represented as the standard deviation.
Hepatocyte incubations
Pooled human hepatocytes and culture medium were purchased from Celsis/IVT (Chicago, IL). Cultures in suspension were initiated from cryopreserved vials using InVitro HI incubation buffer. Cell viability was verified under a microscope by trypan blue exclusion. Approximately 5.0x105/ml hepatocytes were incubated with PQ for 2 h. Reactions quenched using two volumes cold ACN.
All isoenzyme incubations were performed as mentioned in the enzyme activity screening section above, except PQ was fixed at 50 μM to increase the chances of detecting low-level metabolites, and incubations were quenched after 1 hr. Hepatocyte incubations were performed as outlined in the hepatocytes incubations section. Metabolites in the accurate mass data were found using AB Sciex MetabolitePilot™ software. The data were searched using an algorithm that used all of the following criteria for finding potential metabolites: predicted metabolites, mass defects, and isotope pattern (all three for typical phase I and phase II and combinations with dealkylations), and also common fragment ions and neutral losses found in the primaquine product ion spectrum. The data were scored by four criteria with user-selectable weighting: mass defect (20% of total) with an all-or-nothing score; isotope pattern (10% of total); MS/MS similarity and quality (30% of total); and mass accuracy (40% of total). The last three scoring criteria were given weighted values based on how close the measure property matched theory. Most metabolites had a score of 70% or better. The software allows for up to five controls. The data were processed using three controls (a time 0 sample, and two buffer blanks).
MetabolitePilot™ software also was used to help match probable structures to the MS/MS spectra of potential metabolites using its interpretation module. Using the accurate mass (typically within ca. 1 mDa except for very low m/z fragments) of the fragment ions and a program which allows for bond breakages, ring closures, and re-arrangements while also using chemical logic, the program presents possible structures for the fragments, highlighting them within the proposed structure.
Discussion
Isoenzyme activity screening and steady-state kinetic data suggest that CYPs 3A4, 2D6, 2C19, and also MAO-A all play a role in PQ metabolism. However, using the RAF-weighted, steady-state kinetics approach to analyse the data [
11], MAO-A appears to be the predominant enzyme involved in Phase I PQ metabolism, followed by 2C19, 3A4, and 2D6, respectively. Brossi
et al proposed a role for MAOs in the formation of the carboxy (CPQ) metabolite, and demonstrated differential K
i values for (+)-PQ and (−)-PQ [
12]. Further, they directly demonstrated CPQ formation in liver fractions containing only MAOs. These observations appear to support the findings of the present study, indicating a major role for MAO-A, in particular, in PQ metabolism.
While 3A4 and 2C19 are commonly involved in the metabolism of anti-malarial compounds [
13], neither MAO-A nor 2D6 are generally considered to be significant contributors to anti-malarial drug metabolism. However, in this study, metabolites directly observed in MAO-A and 2D6 incubations accounted for more than 93% of the total metabolite peak area observed after incubation with hepatocytes. It is also interesting to note that several anti-malarials are known inhibitors of CYP 2D6, including chloroquine (CQ), quinine, and quinidine [
13,
14].
Many of the metabolites identified here have been observed during the decades of research into the metabolism of PQ [
15‐
18]. Key new findings of the present study are the attribution of many of these metabolites to production from specific CYP or MAO mediated pathways. For example, while 3A4 and 2C19 cannot be definitively eliminated from consideration as the source of PQ’s haemolytic and/or therapeutic metabolites, this study clearly demonstrates that 2D6 has a much greater intrinsic affinity for the metabolism of PQ and, at equal concentrations, produces a significantly higher amount of phenolic metabolites than either of those isoforms. The nature of the metabolites formed, including phenolic metabolites of the type first identified by Idowu
et al., by CYP 2D6 is highly suggestive that 2D6 may account for the well documented haemolytic effects of the drug and its efficacy, as several of the phenolic metabolites observed have been shown to re-dox cycle in the presence of NADPH and/or oxidoreductase [
5,
16]. This is believed to be the first time that these metabolites have been directly attributed to one single CYP pathway. If this is the case, then it is unlikely that haemolytic toxicity could be separated from efficacy through medicinal chemistry efforts aimed at avoiding CYP 2D6 metabolism, as the metabolite(s) responsible for each effect are generated in the same pathway. This has important implications for the efficacy of PQ as a treatment against relapsing malaria in 2D6 polymorphs of the poor or intermediate metabolizer phenotype, as prevalence of the poor metabolizer phenotypes are noted to be as high as 21% in some populations (German Caucasians) and the intermediate metabolizer phenotypes are estimated as high as 50% in others (Asians) [
19]. Further, haemolytic toxicity could be exacerbated in extensive metabolizers. The formation of the alcohol and ring-closed form of the aldehyde by MAO-A, suggests that this enzyme acts to catalyze the first step in the pathway leading to the formation of CPQ, the major metabolite of PQ found in plasma, and ultimately drug clearance via the acyl glucoronide. It should be noted however that 2D6 produced the alcohol to a much lower level. This provides evidence that carboxy primaquine production may be mediated by both MAO-A and to a lesser extent by CYP 2D6. Further investigation in this area is needed to determine the effects of common CYP 2D6 and MAO-A inhibitors and inducers on PQ’s efficacy and toxicity.
Acknowledgements
This research and development project is supported by a grant to the University of Mississippi, and was awarded and administered by the US Army Medical Research & Materiel Command (USAMRMC) and the Telemedicine & Advanced Technology Research Center (TATRC), at Fort Detrick, MD, under award number: W81-XWH-07-2-0095. The views, opinions and/or findings contained in this presentation are those of the author(s) and do not necessarily reflect the views of the Department of Defense and should not be construed as an official DoD/Army position, policy or decision unless so designated by other documentation. No official endorsement should be made.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
BSP and JCS: manuscript preparation, experimental design, data analysis and interpretation; JF, REC and CV: mass spectrometry; RB: enzyme kinetics; RJS and GAR: data interpretation, figure preparation; XJ: experimental design, data analysis and interpretation; MPK: funding, experimental design, manuscript preparation; LAW, CO and VM: funding, experimental design. Each author read and approved the final version of this manuscript.