Methylone and pentylone: structural analysis of new psychoactive substances

Racemic standards of methylone hydrochloride and pentylone hydrochloride with purity exceeding 98% were obtained from Alfarma (Černošice, Czech Republic). Deuterium oxide (D₂O; 99.9% D) for the dissolution of methylone and pentylone was purchased from ISOSAR GmbH, Germany. HPLC grade solvents (hexane, heptane, propan-2-ol IPA) were obtained from Labicom (Olomouc, Czech Republic) and ethanol was purchased from Merck (Darmstadt, Germany). Diethylamine, used as a basic additive, was obtained from Sigma-Aldrich (St. Louis, MO, USA).

Both the analytical and the preparative chiral separation of the drugs were performed using the Waters AutoPurification System (Milford, MA, USA) equipped with a UV–Vis detector (detection wavelength of 254 nm). The analytical separation was performed on a chiral polysaccharide column ChiralArt Amylose-SA (250 × 4.6 mm i.d., 5 μm, YMC Europe GmbH, Dinslaken, Germany) with an analyte concentration of 1 mg ml⁻¹ and a flow rate of 1 ml min⁻¹ at ambient temperature. For the preparative chromatography, an analogous polysaccharide column available in our laboratory—Chiralpak IA (250 × 20 mm i.d., 5 μm, Chiral Technologies Europe, Illkirch, France)—was used. For methylone, the sample concentration was 17.5 mg ml⁻¹, while in case of pentylone, for which a different mobile phase was required (vide infra), the sample concentration was 4 mg ml⁻¹. In both cases, at flow rate of 15 ml min⁻¹ and ambient temperature conditions were employed. For analytical chromatography screening, commercially available drugs were dissolved in the mobile phase containing 0.1% of diethylamine. The sample for preparative chromatography was dissolved in a mixture of 1% diethylamine in propan-2-ol (0.45 ml) and a total volume of 1 ml was adjusted by the addition of heptane. In both cases, the addition of diethylamine ensured the liberation of a free base of the drug from the corresponding hydrochloric acid salt.

The experimental ECD spectra were acquired using a J-815 spectrometer (JASCO Corporation, Tokyo, Japan) purged with nitrogen gas (purity 99.99%, Siad, Prague, Czech Republic) during the measurement. To measure the ECD spectra, solutions of individual enantiomers of a concentration of 15 mg l⁻¹ in demineralized water (UCT Prague, Czech Republic) were prepared. These sample solutions (3 ml) were pipetted into a quartz cuvette (Hellma, Müllheim, Germany) with a pathlength of 1 cm and measured in a spectral range of 185–380 nm at ambient temperature with a 20 nm min⁻¹ scanning speed, 8-s response time, 3 accumulations, and 1 nm band width. The baseline was corrected by subtracting the spectra of demineralized water obtained under identical experimental conditions. The corresponding UV absorption was calculated from the detector HT voltage using the Spectra Analysis module of the Spectra Manager software (JASCO Corporation, Tokyo, Japan).

The VCD and IR absorption spectra were acquired using the FTIR IFS 66/S spectrometer equipped with a VCD/IRRAS PMA 37 module (Bruker, Bremen, Germany), a ZnSe beam splitter (Hinds Instrument, Hillsboro, OR, USA), a BaF₂ polarizer, and an MCT detector (InfraRed Associates, Stuart, FL, USA) [32]. The solutions of individual enantiomers of methylone hydrochloride and pentylone hydrochloride in D₂O (100 mg ml⁻¹) were placed individually into a BioCell cuvette with CaF₂ windows (BioTools, Inc., Jupiter, FL, USA) and an optical pathlength of 27.3 μm. The spectra were measured at ambient temperature in a spectral range of 1750 to 1250 cm⁻¹ with a resolution of 8 cm⁻¹. The VCD spectra were averaged from 9 to 12 blocks, each measured for 20 min and containing 3680 interferometric records. The baseline of each enantiomer was corrected by the subtraction of the solvent (D₂O) spectra measured under identical experimental conditions. The VCD noise spectra were offset from zero for clarity.

The starting geometries of (R)-methylone and (R)-pentylone hydrochlorides were optimized using the DFT method at the B3LYP/6-31G(d) level. Reoptimization of the final lowest energy conformers, considering their Gibbs free energies, was performed by the DFT method at several higher levels of theory (B3LYP/6-311++G(d,p), B3PW91/aug-cc-pVDZ, B3PW91/6-311++G(d,p), CAM-B3LYP/6-311++G(d,p), CAM-B3LYP/aug-cc-pVDZ, wB97XD/6-311++G (d,p), wB97XD/aug-cc-pVDZ, and wB97XD/TZVP, Electronic Supplementary Material, Table S1 and S2) using the Gaussian 09 program package [33]. In this work, only the results providing the best level of agreement with the experimental spectra are presented in detail (B3LYP/6-311++G(d,p), B3PW91/6-311++G(d,p)). All the calculations were carried out on a supercomputer Altix UV 2000 (UCT Prague, Czech Republic). In general, 12 processors with total memory of 64 GB were used for individual computation tasks, which took 40–48 h.

The D₂O solvation effect was considered via the conductor-like polarizable continuum model (CPCM). The DFT calculations of VCD spectra included the effect of deuteration of the amine hydrogens. The population weighted average VCD, ECD, IR and UV spectra were calculated using the Boltzmann distribution based on Gibbs free energy at a temperature of 298 K. The similarity overlaps of the experimental and calculated spectra were undertaken using the CDSpecTech program, based on the similarity of the dissymmetry factor to determine quantitatively the agreement between both spectra and to choose the optimal scaling factor [34, 35]. A scaling factor of 1.0 marks the maximum similarity.

For visualization of the simulated VCD and IR spectra, a Lorentz profile function with a 10 cm⁻¹ bandwidth was assumed. For ECD and UV absorption, the 30 lowest singlet states of each conformer were calculated and Gaussian band profiles with a 12 nm bandwidth were assumed.

Since methylone and pentylone are in the cathinone family, they can be enantiomerically resolved by several different methods including, for example, capillary electrophoresis, supercritical fluid chromatography, and HPLC [36‐38]. However, only some methods from this list can provide pure enantiomers in preparative amounts of the milligram to gram scale. The most prominent method for the separation of chiral substances nowadays is preparative HPLC. We have already employed this method in our previous study focused on the determination of the absolute configuration of butylone [31], thus HPLC was selected as the method of choice in this study.

Starting from the previously optimized conditions, we have found that although they are in the same family of NPS, methylone, butylone and pentylone exhibit significantly different chromatographic behaviours. Therefore, different methods of chiral separation for methylone and pentylone have been developed; more detailed information on the screening of the mobile phase conditions can be found in the Electronic Supplementary Material. Methylone, as the more polar substance, was resolved using a mobile phase composed of a mixture of hexane/EtOH (4/1, v/v) with diethylamine (0.1%) (Fig. 2), whereas the more lipophilic pentylone was successfully resolved using a mixture of heptane/IPA (95/5, v/v) with diethylamine (0.1%) (Fig. 3).

In both cases, automatic fraction collection was set to ensure high purity of the collected enantiomers. In order to assure the stability of the separated enantiomers and to avoid a cyclization reaction occurring for the cathinone free base [39], the separated substances were transformed to hydrochlorides by adding an etheric solution of hydrogen chloride. Both enantiomers (methylone 1 and methylone 2) were found to be of comparable optical purity with the enantiomeric excess of > 96.3%. This value was still acceptable for the subsequent spectroscopic measurements. For pentylone, even better values of > 99.4% and > 98.9% for the first and the second eluting enantiomers, respectively, were found.

The individual (R)-enantiomers of methylone and pentylone hydrochloride were selected for the conformational analysis. Energetically preferred geometries of these molecules were determined by three or five dihedral angles α₁, α₂, α₃, and optionally α₄ and α₅ (Fig. 1) for methylone and pentylone, respectively. The dihedral angle α₁ was systematically changed by 180° rotation about a single bond and the remaining angles α₂, α₃ and possibly α₄ and α₅ by 120° using the MCM software [40]. As a result, 18 and 162 starting geometries for methylone and pentylone, respectively, were obtained. Those geometries were subsequently optimized by the DFT method at the B3LYP/6-31G(d) level of theory. Since most starting structures were converging to the same geometry or their population was too low, conformational analysis revealed all the relevant stable conformers (relative energy < 2 kcal mol⁻¹): 5 and 9 for methylone and pentylone, respectively. These stable conformers were reoptimized at a more refined level of theory: B3LYP/6-311++G(d,p) or B3PW91/6-311 ++G(d,p) to provide the final structures (Fig. 4) and their relative abundances based on the Boltzmann distribution (Table 1).

All the conformers were stabilized by the hydrogen bonding between the oxygen atom of the carbonyl group and the hydrogen atom of the amino group. From all the conformers of methylone, this interaction most stabilized the structure of the conformer M I, probably resulting in its high relative abundance (85%). The distance of hydrogen and oxygen in the case of the conformer M I was ~ 2.1 Å, for other conformers of methylone this distance was 2.2–2.5 Å. As for the individual dihedral angles (Table 1), the angle α₁ occupied predominantly two positions that differed by ~ 180°. The total number of stable conformers of pentylone was higher than methylone due to the elongation of the alkyl chain and the associated increase in the number of dihedral angles in the molecule. All 9 conformers of pentylone were also stabilized by the hydrogen bonding between the oxygen of the carbonyl group and the hydrogen of the amino group, with the distance of these atoms being in the range of 2.0–2.2 Å for all the stable conformers. Relatively small differences in the length of this stabilizing interaction may be the reason for the more uniform relative abundances of conformers (33, 26 and 19%) than in the case of methylone.

Since the experimental ECD spectra of the first and the second eluting compounds (Fig. 5) were mirror images, the enantiomeric character of the separated products was confirmed. The comparison of the experimental and simulated spectra determined the absolute configuration of the individual enantiomers: methylone 1 and 2 were determined as (R)- and (S)-enantiomer, respectively. On the contrary, pentylone 1 and 2 were determined as (S)- and (R)-enantiomer, respectively. These results could be confirmed further using the similarity overlap plots (Electronic Supplementary Material, Figs. S3 and S4). For the correct assignment of the bands in the experimental and simulated spectra, the simulated spectra of methylone were scaled by a factor of 0.96 in order to reach the maximum value of the similarity index. After scaling, a similarity index of 0.85 was achieved in this case. For pentylone, a scaling factor of 0.97 was applied and the corresponding index of similarity reached 0.75. Both results were more than sufficient to reliably determine the absolute configuration of individual enantiomers, which was subsequently confirmed by a detailed VCD spectra analysis.

In the experimental ECD spectra of methylone and pentylone hydrochloride (Fig. 5), six very similar bands reflecting the combination of electron transitions π → π *, n → π * and n → σ * were observed. The bands at 330 nm and partially at 278 nm were mainly an expression of an electron transition from the highest occupied to the lowest unoccupied molecular orbitals (HOMO → LUMO, Fig. 6). Three positive and three negative bands in the spectra of (R)-enantiomers were also present in the simulated weighted average spectra of (R)-enantiomers. The spectra of individual conformers of the (R)-enantiomers differed considerably from one another (Fig. 5, top), indicating the exceptional sensitivity of ECD to the 3D structure of the molecules compared to non-polarized methods (UV, IR). However, the experimental ECD spectra did not allow reliable differentiation between the two studied compounds.

The simulated UV spectra of (R)-methylone and (R)-pentylone hydrochloride correctly predicted four significant bands of the experimental spectra (Fig. 7). The simulated spectra were scaled by a factor of 0.97 and 0.98 for methylone and pentylone, respectively, and the final similarity overlap reached 0.82 and 0.78 for methylone and pentylone, respectively (Electronic Supplementary Material, Figs. S3 and S4). However, this method does not seem to be suitable either for the differentiation of substances differing in the number of –CH₂ groups (since the experimental spectra of both studied compounds did not significantly differ) nor for the studies of the 3D structural changes of the individual conformers. The shapes of the partial spectra (Fig. 7, top) were very similar, with slight differences up to 10 nm observed in the change of the position of the band maximum at 330 nm. Differences between the intensities of band at ~ 234 nm in the simulated spectra of (R)-pentylone were caused mainly by the change of dihedral angle α₁: if its value was positive, the intensity of the band was lower. In contrast, the intensity of the band at 270 nm was higher for positive values of α₁.

The shape of the experimental IR absorption spectra of methylone and pentylone in the spectral range 1750–1250 cm⁻¹ was very similar with a total number of 11 and 9 bands for methylone and pentylone, respectively (Fig. 8). For the correct assignment of the individual bands and their respective vibrational modes (Tables 2, 3), the spectra were scaled by a factor of 0.99 and 0.98 and resulted in a very convincing index of spectra similarity: 0.85 and 0.91 for methylone and pentylone, respectively (Electronic Supplementary Material, Figs. S3 and S4). Looking at the individual spectra of the conformers (Fig. 6, top), it is worth noting the large influence of the dihedral angle α₁ on the spectral shape. Band 1 was shifted to the lower wavenumbers in the spectra of all the conformers with a positive value of angle α₁ (~ 5°) compared with the spectra of the other conformers. Band 2 was apparent only in the spectra of the conformers with the positive value of angle α₁ (~ 5°), hence the intensity of this band was lower in the averaged spectra and appeared only as a shoulder. In contrast, bands 9 in the spectra of methylone and 8 in the spectra of pentylone were not present in the spectra of those conformers. The bands in this area reflected the bending vibration of the C–H group, including the umbrella bending vibration of –CH₃. Band 7 in the spectra of methylone and band 6 in the spectra of pentylone were more intense in the case of the conformers with positive value of α₁. Band 10 in the spectra of methylone reflected the bending vibration of the C–H group in CH–CH₃ and was correctly predicted by the DFT calculations. This group was not present in the structure of pentylone, therefore we did not expect a band reflecting the bending vibration of the C–H group in CH–CH₃ to occur in its IR spectra. This was confirmed by the experimental spectra as well. Furthermore, slight differences in the spectral shapes were observed between the experimental IR absorption spectra of methylone and pentylone hydrochloride (Fig. 8, bottom), particularly in the spectral range of 1420–1280 cm⁻¹. Therefore, this method can be used to distinguish structurally similar substances, but, unfortunately, with low sensitivity.

The experimental VCD spectra (Fig. 9, bottom) were presented in the spectral range of 1750–1263 cm⁻¹ because of the high level of VCD noise caused by overabsorption at lower wavenumbers. The simulated VCD spectra were scaled by a factor of 0.98 and 0.99 for methylone and pentylone, respectively, for the correct assignment of the individual bands (Tables 4, 5). The resulting index of spectra overlap reached convincing values of 0.86 and 0.70 for methylone and pentylone, respectively (Electronic Supplementary Material, Figs. S3 and S4). In the experimental VCD spectra of (R)-methylone and (R)-pentylone hydrochloride (Fig. 9, bottom), there were several areas in which individual substances could be reliably distinguished, mainly in a spectral range of 1400–1263 cm⁻¹, where the molecules not only had a different number of bands but also different spectral shapes. Moreover, that was predicted correctly by the quantum chemical simulations (Fig. 9, middle). In addition, in the spectral range of 1700–1580 cm⁻¹, there were different intensities of bands 1 and 2, which was also correctly predicted by the simulations. Our results demonstrating a higher propensity of distinguishing between two structurally similar compounds by VCD than ECD are in very good agreement with the literature [22]. The conformers of methylone and pentylone with a positive value of dihedral angle α₁ again had a distinct spectral shape different from the rest of the conformers, e.g. band 3 in the VCD spectra of both (R)-methylone and (R)-pentylone as well as bands 5, 6 and 7.