Characterization of primaquine imidazolidin-4-ones with antimalarial activity by electrospray ionization-ion trap mass spectrometry

A vailable online at www.sciencedirect.com International Journal of Mass Spectrometry 270 (2008) 81–93 Characterization of primaquine imidazolidin-4-ones with antimalarial activity by electrospray ionization-ion trap mass spectrometry Nuno Valea, Rui Moreirab, Paula Gomesa,∗ aCentro de Investiga¸c˜ao em Qu´ımica da Universidade do Porto, Departamento de Qu´ımica, Faculdade de Ciˆencias, Universidade do Porto, R. Campo Alegre 687, P-4169-007 Porto, Portugal bCentro de Estudos de Ciˆencias Farmacˆeuticas, Faculdade de Farm´acia, Universidade de Lisboa, Av. Prof. Gama Pinto, P-1649-019 Lisboa, Portugal Received 19 November 2007; accepted 5 December 2007 Available online 14 December 2007 Abstract The extensive characterization by electrospray ionization-ion trap mass spectrometry (ESI-MSn) of 20 imidazolidin-4-ones derived from the antimalarial primaquine was well obtained. These compounds are being under investigation as potential antimalarials, as they have been previously found to be active against rodent P. berghei malaria and to be highly stable under physiological conditions. Experiments by collision-induced dissociation (CID) in the nozzle-skimmer region or by tandem-MS have shown the title compounds to be remarkably stable. Mechanisms are proposed to explain the major fragmentations observed in ESI-MSnexperiments. Overall, this work represents an unprecedented contribution to a deeper insight into imidazolidin-4-one antimalarials based on a classic 8-aminoquinolinic scaffold. Data herein reported and discussed may be an useful guide for future studies on therapeutically relevant molecules possessing either the 8-aminoquinoline or the imidazolidin-4-one motifs. © 2007 Elsevier B.V. All rights reserved. Keywords: 8-Aminoquinoline; Antimalarial; ESI-MSn; Fragmentation; Imidazolidin-4-one 1. Introduction The diverse analytical methods focused on the antimalarial primaquine (PQ, 1,Scheme 1) or its analogues have been so far targeted at pharmacokinetics studies and metabolite analysis [1–9] or isomer separation/characterization [10,11]. The application of molecular modeling studies associated to voltammetric techniques showed to be an important way to understand the redox mechanism of electro-active drugs where aminoquinolines (AQs) such as PQ are included [7]. Two classes of AQs were investigated by infrared spectrometry to assess their geometry and coordination ability [6]. UV absorption and circular dichroism (CD) have also been used to investigate interactions between heparin and several quinolines [6]. Studies with HPLC are also currently used to characterize therapeutically relevant AQs, including PQ and its metabolites [1–3]. Other analytical tools to study PQ and its derivatives have included fast atom bombardment mass spectral analysis ∗Corresponding author. Tel.: +351 220402563; fax: +351 220402659. E-mail address: [email protected] (P. Gomes). of three new PQ oxidation products [4] or enantioselective capillary electrophoresis for the simultaneous determination of PQ and its main metabolite, carboxyprimaquine, in rat liver mitochondrial fraction [5]. Supercritical fluid chromatography–mass spectrometry (SFC–MS) was used to achieve shorter analysis times while having good resolution between PQ (free base), PQ diphosphate and the positional isomer quinocide, as well as to elucidate additional information about differences in their MS fragmentation patterns [8]. PQ and its isomers have also been identified by liquid chromatography–mass spectrometry (LC–MS) [10]. Electrospray ionization-mass spectrometry (ESI-MS) analysis, introduced by Yamashita and Fenn [12a,b], was described for the identification of NPC 1161, another antimalarial 8-AQ, and its metabolites in biological samples [9]. ESI-MS techniques are relevant in many research and development (R&D) areas [13–19], from food chemistry to the pharmaceutical industry [20–37]. Detailed structural information on drugs and related molecules can be obtained by resorting to cone voltage fragmentation with single MS instrumentation, to collisionally induced dissociation (CID) with triple quadrupole MS instruments, to MSntechniques using quadrupole ion-trap instrumentation or 1387-3806/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ijms.2007.12.006 82 N. Vale et al. / International Journal of Mass Spectrometry 270 (2008) 81–93 Scheme 1. Synthetic route to imidazolidin-4-one derivatives of PQ [40]: (i) BocAAOH + DCCI/HOBt or BocAAOSu (see text), DCM, 0 ◦C→r.t.; (ii) neat TFA, r.t.; 30% aq. Na2CO3until pH 10, extraction with CHCl3; (iii) R2(C O)R3, molecular sieves, refluxing MeOH. to time-of-flight mass spectrometry (ToF-MS) [31,32,35]. ESIMSnstudies of antimalarial compounds as bisthiazolium or quinolines can be found on reports by Nicolas et al. [38] or Smyth et al. and O’Donnell et al. [34,39], respectively. However, ESI-MS studies contributing to the development of antimalarials are scarce. Over the past few years, our research group has been working on imidazolidin-4-one derivatives of PQ as potential highly bioavailable antimalarials [40–42]. These imidazolidin-4-ones (4,Scheme 1) are stable at physiological pH and temperature, both in isotonic buffer and in human plasma, evidencing a weak conversion into their PQ ␣-aminoamide precursors (3, Scheme) with half-lives ranging from 9 to 30 days [41,42]. Moreover, these imidazolidin-4-one derivatives of PQ displayed gametocytocidal activity comparable or superior to that of the parent drug [41]. We now wish to report the characterization of 20 PQ-derived imidazolidin-4-ones 4(Table 1) by ESI-MSntechniques. This contributes to a deeper insight into the properties of these biologically relevant 8-AQs, through knowledge of their stability and fragmentation mechanisms under the gas-phase conditions of the ESI-MSnexperiments. 2. Experimental 2.1. Chemical synthesis The synthesis and spectral (NMR, HRMS) characterization of 14 out of the 20 imidazolidin-4-ones 4covered by the present study have been reported elsewhere [40]. The procedure for the initial condensation of PQ with a Boc-protected amino acid (BocAAOH) in dichloromethane (DCM) was slightly changed with respect to our previous report [40], as reaction times can be dramatically reduced through utilization of the succinimide ester of the amino acid, instead of using a dicyclohexylcarbodiimide (DCCI)/1-hydroxybenzotriazole (HOBt)-mediated condensation. The six new derivatives were obtained by the Table 1 PQ and derived imidazolidin-4-ones studied; average m/zvalues for the main species detected in the full-MS spectral analysis of compounds 1,3.2 and 4.1–4.20 are also given for comparison Compound R1R2R3MW (g mol−1)m/z[M+H] +m/z[M+ Na]+ 4.1 –H –CH3–CH3356.46 357.87 379.67 4.2 –CH3–CH3–CH3370.49 371.93 393.60 4.3 –CH(CH3)2–CH3–CH3398.45 399.93 422.00 4.4 –CH2Ph –CH3–CH3446.58 448.47 470.07 4.5 –CH2CH(CH3)2–CH3–CH3412.57 414.07 436.06 4.6 –CH(CH3)CH2CH3–CH3–CH3412.57 414.07 436.33 4.7 –CH(CH3)2–(CH2)4– 424.58 425.93 447.80 4.8 –CH(CH3)2–(CH2)5– 438.61 439.87 461.87 4.9 –CH(CH3)2–(CH2)6– 452.63 453.93 475.87 4.10 –CH(CH3)2–(CH2)2CH(CH3)(CH2)2– 452.63 453.93 476.33 4.11 –H –(CH2)2CH(CH3)(CH2)2– 410.55 412.33 433.80 4.12 –CH3–(CH2)2CH(CH3)(CH2)2– 424.58 425.80 448.07 4.13 –CH3–(CH2)6– 424.58 425.67 447.93 4.14 –CH2CH(CH3)2–(CH2)6– 466.66 467.87 490.93 4.15 –CH(CH3)CH2CH3–(CH2)6– 466.66 468.00 490.20 4.16 –H –(CH2)5– 396.53 397.93 419.87 4.17 –CH2CH(CH3)2–(CH2)5– 452.63 453.93 475.87 4.18 –CH(CH3)CH2CH3–(CH2)5– 452.63 453.93 476.07 4.19 –CH3–(CH2)4– 396.53 397.93 420.07 4.20 –CH2Ph –(CH2)4– 472.62 474.07 495.93 1– – – 259.35 260.80 –a 3.2 –CH3– – 330.42 331.23 354.00 aSodium adducts were never observed in the full-MS spectra of primaquine. N. Vale et al. / International Journal of Mass Spectrometry 270 (2008) 81–93 83 Table 2 Main species (relative abundance ≥20%) detected in the analyses in the nozzle-skimmer region Species VsCID Compounds on whose spectra the species was observed [M+H] +and/or [M+ Na]+70 4.2 4.3 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 1 80 4.2 4.3 4.4 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 1 90 4.2 4.3 4.4 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.20 1 100 4.2 4.3 4.4 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 1 [M2+H] +and/or [M2+ Na]+70 4.4 4.5 4.11 4.14 4.15 4.16 4.20 80 4.4 4.8 4.9 4.11 4.14 4.15 4.16 4.20 90 4.5 4.8 4.9 4.11 4.14 4.15 4.16 4.19 4.20 100 4.4 4.5 4.8 4.9 4.11 4.14 4.15 4.16 4.18 4.19 4.20 [M3+H] +70 80 90 4.19 100 4.19 [5+H] +, quasi-molecular ion of MAQ (m/z= 175) 70 4.1 4.2 4.3 4.4 4.5 4.8 4.9 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 1 80 4.1 4.2 4.3 4.4 4.5 4.8 4.9 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 1 90 4.2 4.3 4.4 4.5 4.8 4.9 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 1 100 4.2 4.3 4.4 4.8 4.9 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 1 [DHQ + H]+, quasi-molecular ion of 7,8-dihydroquinoline (m/z= 132) 70 4.1 4.2 4.3 4.4 4.5 4.16 4.19 1 80 4.1 4.2 4.3 4.4 4.5 4.8 4.11 4.12 4.13 4.16 4.19 1 90 4.1 4.2 4.3 4.5 4.7 4.8 4.9 4.11 4.12 4.13 4.15 4.16 4.19 1 100 4.1 4.2 4.3 4.5 4.7 4.8 4.9 4.11 4.12 4.13 4.15 4.16 4.19 1 84 N. Vale et al. / International Journal of Mass Spectrometry 270 (2008) 81–93 same procedure and were synthesized in goods yields from the corresponding ␣-aminoamides (3,Scheme 1) by refluxing with an excess of the appropriate ketone in methanol, in the presence of triethylamine and molecular sieves. The structures of the new compounds were confirmed by 1H and 13C NMR, and also by HRMS using the MALDI-TOF technique (cf. Supporting Information). 2.2. ESI-MSnanalysis ESI-MSnstudies on compounds 4.1–4.20 (Table 1) were achieved using a Finnigan Surveyor LCQ DECA XP MAX quadrupole ion trap mass spectrometer, utilizing electrospray ionization (ESI). Methanolic solutions of the compounds, containing 0.1% acetic acid, were infused into the ESI probe at a rate of 3 ␮L min−1. Inclusion of methanol to prepare samples was necessary to obtain good electrospray behavior [26].Inthe ESI source, nitrogen sheath gas flow was maintained at 20 psi, the capillary temperature was set to 275 ◦C and the spray voltage was set to 5 kV. Capillary voltage was of 15 V and tube lens voltage was of 30 V. All spectra were obtained in positive mode and data were collected and analyzed by using the Xcalibur software developed by ThermoFinnigan. To induce sample fragmentation, two processes were used: collision-induced dissociation (CID) in the nozzle-skimmer region and tandem-MS. In the first process, fragmentation is induced by colliding sample ions with the background gas in the intermediate-pressure region of the ESI interface. This region is also called “nozzleskimmer” and gas pressure is about 10−1mbar. Increasing the voltage between the nozzle and the skimmer (VsCID) increases the kinetic energy of the ions passing through this region. This raises the energy of collision between the ions and background gas, eventually causing fragmentation [43]. Tandem-MS or MS/MS occurs at the high vacuum region of the mass analyzer; the ion of interest is isolated within the ion trap by ejecting all other ions out of the trap. Then, this ion is accelerated, by applying a high-frequency AC voltage, and collides with a collision gas. The fragmentation ions generated are then detected by a mass scan [43]. The collision energy needed to achieve optimum fragmentation efficiency has been shown to follow a linear correlation with m/z. The normalized collision energy (NCE) principle automatically compensates for this mass dependency [44]. Three types of fragmentation experiments were conducted. The first was carried out on the nozzle-skimmer region where the VsCID parameter was varied between 70 and 100 V. The evaluation of the influence of the voltage in-source on the fragmentation patterns provides a means to distinguish stable adducts from unstable species. The second experiment was done by tandem-MS and was characterized by varying the NCE levels at 10, 20, 30 or 40%. This was done to establish the optimal NCE value that would allow, through MS2spectra, to analyze the structural stability of the sample, as this parameter has been used to predict drug metabolites at given NCE values as recommended by the MS manufacturer (25 or 30%) [32,44]. Scheme 2. Main MS/MS fragmentation of compounds 4at normalized collision energies of 30 and 40%. N. Vale et al. / International Journal of Mass Spectrometry 270 (2008) 81–93 85 The third experiment was also done by tandem-MS at a fixed NCE value of 30%, for subsequent MSnanalysis of the main fragment generated on the previous MSn−1fragmentation (n= 3, 4). MS2fragmentation was done on the quasi-molecular ion ([M+H] +) observed in the corresponding full-MS spectrum. This was not necessarily the base peak, as imidazolidin-4-ones 4were quite prone to form sodium adducts ([M+ Na]+species) that were frequently the base peaks in full-MS spectra. 3. Results and discussion 3.1. Experiments in the nozzle-skimmer region Spectra were obtained at four different voltages, 70, 80, 90 and 100 V. These can be regarded as high values (100 V is the maximum reached by the spectrometer) that were deliberately chosen to test compound resistance against bond disruption under highly fragmentative conditions. Results compiled in Table 2 demonstrate that compounds 4were mainly detected as [M+H] +and/or [M+ Na]+ions. The two major occurring fragmentations led to observation of peaks at m/z175 and m/z 132 that were respectively associated to the quasi-molecular ions ([M+H] +) of 6-methoxy-8-aminoquinoline (MAQ, 5on Scheme 2) and of, probably, 7,8-dihydroquinoline (DHQ, MW = 131 g mol−1) that can be generated from any of compounds 4. DHQ has also been previously found as the major product of MS2fragmentation of nicotine [34]. Some of the imidazolidin-4-ones, such as 4.1 and 4.5 gave rise to complex spectra, presenting a high number of peaks and not showing either the imidazolidin-4-one quasi-molecular ion [M+H] +or its sodium adduct [M+ Na]+at any of the four voltages employed, which was interpreted as an outcome of their higher instability at the analysis conditions employed. The above situation was exceptional, as the majority of compounds 4generated relatively clean spectra with no increase on fragmentative phenomena at higher voltages. An example is that of compound 4.20 whose spectrum is mainly composed by peaks due to the following ions: [M+H] +,[M+ Na]+, [M2+ Na]+and by a peak at m/z= 186 due to fragment cyclo[(CH2)4C+]–N CH–CH2Ph (see Section 3.3). Fig. 1 displays the spectra obtained for 4.20 at the four different voltages used, showing that increasing the potential provokes a decrease on peak intensity at m/z= 186, but favors the in-source formaFig. 1. Fragmentation spectra for compound 4.20 on the skimmer-CID region at 70, 80, 90 and 100 V. 86 N. Vale et al. / International Journal of Mass Spectrometry 270 (2008) 81–93 Table 3 MS2data obtained at varying normalized collision energies Original compound (MW/g mol−1) Base-peak observed (m/z) Relative abundance of the original species at NCE 30% Relative abundance of MAQ (5) at NCE 30%a Normalized collision energy (%) for MS/MS 10 20 30 40 4.1 (356.46) 357.19 357.19 183.15 183.15 23.86 7.10 4.2 (370.49) 371.20 371.20 197.13 197.13 11.83 10.52 4.3 (398.45) 399.20 399.20 225.13 225.13 15.03 11.31 4.4 (446.58) 447.12 447.18 273.15 273.15 24.22 3.83 4.5 (412.57) 413.27 413.33 239.20 239.20 30.23 6.34 4.6 (412.57) 413.27 413.27 239.20 239.20 27.93 7.44 4.7 (424.58) 425.29 425.29 251.18 251.18 27.89 4.31 4.8 (438.61) 439.27 439.27 265.16 265.16 11.63 7.48 4.9 (452.63) 453.30 453.30 279.20 279.20 18.00 12.28 4.10 (452.63) 453.20 453.20 279.27 279.13 31.80 8.12 4.11 (410.55) 411.20 411.20 237.33 237.33 23.28 7.12 4.12 (424.58) 425.22 425.22 251.25 251.25 18.36 8.62 4.13 (424.58) 425.22 425.22 251.25 251.18 28.98 10.80 4.14 (466.66) 467.27 467.27 293.27 293.27 23.80 14.49 4.15 (466.66) 467.33 467.33 293.33 293.20 44.06 9.26 4.16 (396.53) 398.20 398.20 224.27 224.27 23.50 13.75 4.17 (452.63) 453.24 453.24 279.27 279.20 33.06 6.28 4.18 (452.63) 453.27 453.33 279.27 279.20 22.74 8.31 4.19 (396.53) 397.20 397.20 223.20 223.13 19.54 4.91 4.20 (472.62) 473.20 473.20 299.27 299.20 18.27 1.63 1((259.35) 260.07 260.00 243.24 243.17 0.00 34.86 3.2 (330.42) 331.23 331.23 157.19 157.06 21.37 57.59 aFragment common to all compounds studied, with m/z([M+H] +) = 175. N. Vale et al. / International Journal of Mass Spectrometry 270 (2008) 81–93 87 tion of the [M2+ Na]+adduct. Peak intensity for ions [M+H] + and [M+ Na]+are similar, but there is a slight inversion on relative intensity when the potential is increased. The fragmentation behavior of compound 4.20 under these conditions suggests that its structure is quite stable. Interestingly, this compound has been previously characterized as highly stable in human plasma at physiological pH and temperature, where it remains intact after 72 h of incubation [41,42]. Another example of remarkable stability under the conditions of the ESI-MS analysis refers to compound 4.7, whose fragmentation behavior in-source is slightly different from that of 4.20. In the case of 4.7, voltage increase favors the formation of the [M+ Na]+ion, whereas the number and type of fragments remain unchanged (not shown). In fact, all imidazolidin-4-ones derived from the amino acid valine were seen to be stable towards in-source fragmentation, as further illustrated by the behavior of compound 4.10 that does not even originate the fragments at m/z= 175 and 132, currently observed for the generality of compounds 4(Table 2). Finally, it must be outlined that acetone-derived imidazolidin4-ones (4.1–4.6), especially smaller ones, were clearly more “fragmentable” than those derived from cyclic ketones. Again, it is interesting to notice that a similar stability pattern was observed for compounds 4under totally different conditions, namely, towards hydrolysis at physiological pH and temperature [41,42]. Both the parent drug (PQ, 1) and the linear ␣-aminoamide precursor of Ala-derived imidazolidin-4-ones 4.2,4.12,4.13, 4.19, compound 3.2 (Table 1), were subject of an identical study in the nozzle-skimmer region. We could observe that the insource behavior of both 1and 3.2 were quite different from those of compounds 4, in the sense that both compounds gave rise to much “dirtier” spectra and to significant signal carry-over between analyses (data not shown). This was interpreted, respectively, as due to higher compound instability in the course of the MS analysis and to the fact that both compounds have a terminal primary amino group that can strongly attach to the fused silica capillary. This latter fact turned comparisons between linear and cyclic compounds out of significance. 3.2. ESI-MS2analysis at varying NCE values This set of experiments was aimed at the determination of the influence of the normalized collision energy on the mechanism and extension of fragmentation of compounds 4.1–4.20. Compounds 1and 3.2 were included for comparison. Many reports in the literature have been devoted to characterization of different compounds by tandem-MS, with particular attention to the conditions of collision, i.e., NCE values [45–49]. In the majority of those reports, 25% was used as the standard value for NCE in drug fragmentation studies. In our case, we have raised the NCE up to 40% to fully characterize the behavScheme 3. Fragmentations of compounds 4in the MS3mode at 30% NCE: (A) main fragmentation pathway; (B) additional relevant fragmentations on structures derived from 4.1,4.2,4.13 and 4.16. 88 N. Vale et al. / International Journal of Mass Spectrometry 270 (2008) 81–93 ior of imidazolidin-4-ones 4. As shown by data in Table 3,at NCE values of 10 and 20%, the base-peak was still the quasimolecular ion [M+H] +, whereas at 30 and 40%, the base-peak occurs, for all cases, at m/z=m/z[M+H] +−174. This is due to loss of the 6-methoxy-8-aminoquinoline core (MAQ, 5), as depicted in Scheme 2. The fragmentation pathway proposed in Scheme 2, involving elimination of a primary amine (MAQ, 5), is corroborated by a previous LC/ESI-MS characterization of primaquine (1) and its major contaminant, quinocide, carried out by Brondz et al. [10]. Other examples where MS2fragmentation leads to cleavage of a C–N bond with amine release are mostly found for tertiary amines, as these have C–N bond dissociation energies (BDE) lower than those of their secondary amine counterparts [31]. For instance, the ESI-MS behavior of therapeutically relevant tertiary amines has been described by Smyth and co-workers to involve fragmentation to smaller amines [34,47], as nicely reviewed in [31]. Amine release is also common in MS2fragmentations of oligopeptides, though through different pathways, as for instance the y7fragmentation of the bioactive octapeptide Angiotensin II, corresponding to loss of the N-terminal Asp residue as 3-aminodihydrofuran-2,5-dione. The remaining heptapeptide is thus released, where Arg is the new N-terminal residue, having its ␣-amino group now free [49]. The behavior of compounds 4at the four different NCE values tested can be illustrated with data obtained for compound 4.4 as compared to that of the parent drug (PQ, 1).At NCE = 20%,some fragmentation is already observed for 1(MW, 260 g mol−1) with fragments at m/z243 and 175, respectively due to loss of the terminal primary amino group (as ammonia) and of the whole aliphatic chain linked to the 8-amino group yielding MAQ (5). Scheme 4. Hypothetical pathways leading to loss of NH3(A) or CH4(B) from species 7in MS4analyses [X=Y=–HorXandY=–(CH2)n– with n= 2, 3, 4 or X and Y = –(CH2)2–CH(CH3)–(CH2)2–]. N. Vale et al. / International Journal of Mass Spectrometry 270 (2008) 81–93 89 Fig. 2. MSnspectra for imidazolidin-4-one 4.17; main fragmentations are shown on the right.