SPECTRAL ANALYSIS OF 3-( ADAMANTAN-1-YL )-4-ETHYL-1-[ ( 4-PHENYLPIPERAZIN-1-YL ) METHYL ]-1 H-1 , 2 , 4-TRIAZOLE-5 ( 4 H )-THIONE

Vibrational IR (3200–650 cm–1) and Raman spectra (3200–150 cm–1) of adamantane-containing 3-(adamantan1-yl)-4-ethyl-1-[(4-phenylpiperazin-1-yl)methyl]-1H-1,2,4-triazole-5(4H)-thione, which is promising for drug design, were examined. The UV/Vis spectrum (450–200 nm) of the compound in EtOH was measured. Full geometry optimization using density functional theory (DFT) in the B3LYP/cc-pVDZ approximation allowed the equilibrium confi guration of the molecule to be determined and IR and Raman spectra to be calculated. Based on these, the experimental vibrational IR and Raman spectra were interpreted and the biological activity indices were predicted. The UV/Vis spectrum of the title compound was simulated at the time-dependent DFT/CAM-B3LYP/cc-pVDZ level with and without solvent effects and at the ab initio multi-reference perturbation theory XMCQDPT2 level. The UV/Vis spectrum that was simulated using the multi-reference XMCQDPT2 approximation agreed very successfully with the experimental data, in contrast to the single-reference DFT method. This was probably a consequence of intramolecular charge transfer.

resolution). Samples were prepared by grinding crystals into a powder that was spread into a thin fi lm on the refl ecting surface of the substrate (Al foil). The obtained signal was both scattered radiation from the sample and radiation that passed through the sample and was refl ected from the substrate surface. Raman spectra (3200-150 cm -1 ) were recorded using a continuous DPSS laser (λ = 532 nm, power 7 mW) as the excitation source. The angle between the excitation and recorded signal directions was 180 o . The recording system consisted of a Solar TII S3901 spectrograph equipped with a diffraction grating (1200 lines/mm) and CCD array (Princeton Instruments) cooled by liquid N 2 . The signal accumulation time was 60 s. Electronic absorption spectra (450-200 nm) of the compound in EtOH were recorded using a Cary 500 spectrophotometer. The solvent effect was minimized by placing a cuvette with EtOH in the reference channel.
Structures and vibrational spectra of the compound were calculated using the GAMESS-US quantum-chemical suite [23]. Results were illustrated using MacMolPlt [24] and ORTEP [25] programs. Density functional theory (DFT) with a Dunning correlation-consistent, polarized valence, double-zeta basis set [26] and hybrid exchange-correlated functional B3LYP [27][28][29] was used to optimize the equilibrium structure and to calculate the Hessian (force fi eld), eigenfrequencies of modes in the harmonic approximation, and IR and Raman intensities.
Calculations in the fi rst stage searched for a stable confi guration using gradient optimization of all molecular geometric parameters. When the optimization was completed, the eigenfrequencies of modes in the harmonic approximation and intensities in IR and Raman vibrational spectra were calculated. The force fi eld was computed considering two shifts (in positive and negative directions) of each atom relative to its equilibrium position along each of three Cartesian axes. This approach improved by 20-50 cm -1 (or ~1%) the accuracy of the calculated vibrational frequencies. In several instances, imaginary frequencies for low-frequency bending vibrations were avoided. The aforementioned improvement in the accuracy turned out to be very important because the harmonic frequencies of stretching modes of terminal groups that are calculated using the B3LYP/cc-pVDZ approximation are usually 4-5% greater than the experimental ones. The lack of imaginary frequencies in the calculated spectrum confi rmed that the obtained structure was located at a minimum on the potential-energy surface. Vibrational bands and lines were assigned by calculating the potential-energy distribution over internal coordinates.
Time-dependent density functional theory (TDDFT) [30] is currently widely used to simulate electronic spectra of polyatomic molecular systems. Quantum-chemical calculations using the TDDFT formalism with the local spin density approximated by gradient corrections (i.e., using standard exchange-correlation functionals, e.g., hybrid functional B3LYP) are known to give deviations up to 0.4 eV for the energies of the lower excited states from the experimental ones [31]. This is especially typical of molecular systems with charge transfer. Phenyl, piperazine, and triazole groups in the studied molecule allowed it to be classifi ed as a molecular system that could exhibit intramolecular charge transfer (ICT). The Coulomb-attenuating method (CAM) approximation (or refi ned LRC method) that was developed for such instances [32] could sometimes improve the qualitative and quantitative calculations. Thus, the spectral and energy characteristics of the excited singlet states of the studied compound were calculated using the TDDFT method with cc-pVDZ basis set [26] and hybrid functional CAM-B3LYP [32]. The solvent (EtOH) effect was accounted for in the solvation model density (SMD) approximation [33]. The quantum-chemical GAMESS-US suite was also used for the calculations.
Multi-reference approximations (e.g., the ab initio multi-reference perturbation theory method) are alternatives to the TDDFT method. Our recent work [21] demonstrated that ab initio calculations in the multi-reference approximation CASSCF/XMCQDPT2 [34] explained very successfully the electronic absorption spectrum of N′-(adamantan-2-ylidene) benzohydrazide considering the presence of four conformers in solution [21].
The electronic absorption spectrum of the studied molecule was calculated in the CASSCF/XMCQDPT2 approximation using the Firefl y quantum-chemical suite [35] and standard basis set cc-pVDZ [26]. The fi rst stage included CASSCF calculations with the density-matrix averaged over seven states (four singlets and three triplets) for two active electrons involving four active orbitals. Then, calculations used the XMCQDPT2 approximation [34]. The electronic correlation energy for all 117 doubly occupied orbitals in the ground confi guration was taken into account. The energy denominator shift (EDS) was 0.01.
The structure determination of the studied molecule enabled calculation of its biological activity indices (probability to be active/inactive, Pa/Pi), i.e., the probability that biological activity of a particular type was present (absent). The biological activities were obtained using the PASS database [36], the online version of which [37] predicted greater than 4,000 different biological activities based only on the compound structural data.
The calculated structural parameters of the isolated molecule gave deviations of ≤1% for most calculated bond lengths from the experimental ones. Only bonds involving the N atoms (i.e., C-N and N-N of the piperazine and triazole rings) had errors of ~2% for the calculated bond lengths. A similar situation was observed previously for N′-(adamantan-2-ylidene)benzohydrazide [21], 4,4′-methylenediphenyldiisocyanate [38], and benzohydrazide [39] and was most probably characteristic of the used combination of basis set and functional. An analogous tendency was found for the in-plane angles. Angles between C-N bonds typically had the greatest errors (up to 4%). In other instances, the deviations of the calculated values from the experimental ones were usually ≤1%. It is also noteworthy that the C-H bond lengths that were fi xed in the x-ray structure analysis [19] at 0.97 Å (A, P, M), 0.96 (Et), and 0.93 (Ph) differed noticeably from the experimental values for adamantane (1.112 ± 0.004 Å [40]), phenol (1.080-1.086, average 1.083 Å [41]), piperazine (1.133 Å [42]), and methylene (1.085 Å [43]). Figure 2 shows the experimental and calculated IR and Raman spectra of the studied compound. Table 1 presents experimental vibrational frequencies and those calculated at the B3LYP/cc-pVDZ level of theory and the corresponding intensities in the IR and Raman spectra. In most instances, the vibrations were localized on one of the functional groups (adamantyl, phenyl, ethyl, piperazine, methylene, or triazole).
Vibrations of the adamantyl fragment. Frequencies of CH-and CH 2 -stretching vibrations in monosubstituted adamantyl lay in the range 2952-2850 cm -1 [44]. According to the literature [44] and our calculations [20,22], bands at 2943, 2934, 2911, 2907, 2888, and 2953 cm -1 in the IR spectrum and lines at 2941, 2917, 2909, 2889, 2855, and 2847 cm -1 in the Raman spectrum belonged to adamantyl CH and CH 2 stretching vibrations. Adamantyl CH 2 scissoring vibrations were located at 1475-1440 cm -1 [44]. Monosubstitution lifted the degeneracy of CH 2 wagging vibrations (~1350 cm -1 ) and led to the appearance of a specifi c sequence of bands in the range 1400-1300 cm -1 [45]. Thus, bands at ~1472 and 1441 cm -1 were assigned to scissoring vibrations of adamantane CH 2 groups; bands and lines at 1362, 1321, and 1311 cm -1 , to wagging vibrations. Furthermore, adamantyl vibrations appeared near 1200 (CH 2 twisting vibrations), 1100 (CH 2 rocking vibrations), 800 (CC stretching vibrations) and in the ranges 780-740 and 460-330 cm -1 (adamantyl CCC bending vibrations) [44,45]. Thus, bands at 1201 and 1190 cm -1 corresponded to CH 2 twisting vibrations; bands and lines at 865 (IR), 868 (Raman), a hypsochromic shift of 9 nm if the solvent was considered. It was supposed that errors in the TDDFT calculations were responsible for the ICT in the studied compound. Figure 4 shows the upper "frozen" (doubly occupied) and four active molecular orbitals (MOs) from the total active space and also their occupancy factors (second, third, and fourth active MOs had the same occupancy factor). Figure 4 shows that the electron density shifted upon excitation partially from the piperazine and phenyl groups to the triazole ring although it was mostly localized on the S atom. Thus, the phenyl and piperazine rings in the studied compound acted as electron-density donors; the triazole, as an acceptor.
The wavelength of the S 1 ← S 0 -transition that was calculated in the CASSCF(2,4)/XMCQDPT2 approximation [34] was 254 nm. The transition had a high intensity (Table 2; Fig. 3, curve 4). Thus, the electronic absorption spectrum of the studied compound that was calculated in terms of the multi-reference perturbation theory method agreed fully with the experimental data.   Table 3 presents the biological activity indices of the studied compound (i.e., the probability of activity Pa or inactivity Pi of the determined type) that were found using the PASS Online database [37]. According to the literature [36], if Pa > 0.7, then the examined compound will most probably demonstrate this activity experimentally. If 0.5 < Pa < 0.7, then the compound may demonstrate experimental activity but the probability of this is less. If Pa < 0.5, the likelihood of this compound demonstrating this activity is low. According to Table 3, the studied compound probably acted as a proteasome ATP-ase inhibitor.
Conclusions. Spectral characteristics of 3-(adamantan-1-yl)-4-ethyl-1-[(4-phenylpiperazin-1-yl)methyl]-1H-1,2,4-triazole-5(4H)-thione, an adamantane-containing compound that is promising for drug development, were analyzed systematically. Vibrational IR and Raman spectra of the crystalline compound were interpreted based on quantum-chemical simulation in the B3LYP/cc-pVDZ approximation. Biological activity indices of the molecule were predicted based on the molecular structure and established that the studied compound was likely to act as a proteasome ATP-ase inhibitor. Simulation of the UV/Vis absorption spectrum using TDDFT and ab initio perturbation theory demonstrated that the fi rst approximation was inadequate for describing the experimental spectrum of the studied compound. This was most likely a consequent of ICT. The calculations in the second of the aforementioned approximations agreed with the experimental data. The results could be used in medicinal chemistry and for analytical purposes.