Possible interactions between fused pyrazole derivative and magnesium ions - NMR experiments and theoretical calculations

The paper verifies our recently described NMR approach for the determination of interaction sites between a heterocyclic ligand and magnesium ions. The method is based on the comparison of chemical shifts of condensed pyrazole derivative before and after the complexation with magnesium salt and is supported by DFT (B3LYP/6-31G(d,p) level of theory, CPCM solvation model, GIAO method) and classical molecular dynamics theoretical calculations.


Introduction
Pyrazoles and their benzanalogues, indazoles, are bioisosters of pyrroles and indoles.The activity spectrum of pyrazole derivatives is quite wide.This class of heterocyclic compounds includes drugs of anti-inflammatory analgesic, antipyretic, antiestrogenic, antibacterial, antiviral, and anticancer activity.2][3][4] Among the anticancer drugs, pyrazole or indazole derivatives like Celecoxib, anthrapyrazoles (piroxantrone, losoxantrone), and Axitinib have been found to be effective antineoplastic agents useful in the treatment of breast and prostate cancers, as well as acute leukemia. 5,6 gnesium ions, Mg 2+ , participate in many biochemical processes including these involved in cancerogenesis like proliferation, angiogenesis and apoptosis. 7,8 he role of magnesium in cancerogenesis is, however, not clearly understood.Magnesium deficiency, hypomagnesaemia, was observed during the treatment of cancer patients with cisplatin or cetuximab alone or with combination with other chemotherapeutics.It was shown that hypomagnesaemia had a beneficial effect in laboratory animals due to inhibition of tumour cell proliferation and neoangiogenesis.On the other hand, this effect was counterbalanced or even outweighted by accelerated cancer invasion and metastatisation. 9s mentioned above, hypomagnesaemia was detected in patients with colorectal cancer treated with cisplatin or cetuximab.Cisplatin induced hypomagnesaemia is due to the nephrotoxicity of this drug but the mechanism of cetuximab magnesium deficiency has not been clarified. 9ur earlier studies evidenced that some condensed pyrazole derivatives had cytostatic and proapoptotic activity against the HT29 human colorectal cancer cells. 10We suggested therein that the mechanism of cytotoxic activity of these compounds could be related to their ability to adhere to DNA. 11In addition, we observed DNA cleavage. 10,12 t is also possible that they can interfere with the cellular Mg 2+ ions owing to the presence of pyridinic nitrogen in the pyrazole skeleton.Although the structural difference between the small-molecular pyrazole derivative and proteinaceous antibody cetuximab is huge, the molecular mechanism of hypomagnesaemia induced by both compounds might be similar and may involve complexation of the metal.
4][15][16][17][18][19][20][21][22] Investigations on these interactions have been carried out by a variety of analytical methods, including 1 H NMR spectroscopy.Noteworthy is the fact that NMR spectroscopy has found its use for the determination of interactions of some ligands with different chemical targets and allows to assess quantitatively these interactions if the equilibrium constants and changes in chemical shifts are taken into account. 23,24  our previous paper, we discussed the interactions between an indazole derivative and magnesium ions. 25The analysis of chemical shifts before and after the complexation with Mg 2+ led to the conclusion that the most susceptible indazole region for the interaction with the metal ion is the NH functionality and position 5 of the ring.In a continuation of our attempts to clarify the anticancer mechanism of the condensed pyrazoles, 26,27 we present herein our theoretical studies concerning interactions of another fused pyrazole derivative with magnesium cations.We also focused on calculations of the proton chemical shifts of this azole by applying various basis sets and methods in gaseous phase.To give careful consideration to the solvation sphere, we used the conductor-like polarizable continuum (CPCM) solvation model.Moreover, the goal of this work is to improve our protocol for the estimation of interactions between metal ions and small to medium size molecules.Similarly to the previous studies, this approach is very efficient with respect to experimental and measuring times and repeatability of its results under different conditions. 24,25 sults and Discussion
In order to analyse the interactions between compound 7 and the metal ion, compound 7 was dissolved in DMSO-d6 and separately in DMSO-d6 with the admixture of Mg(NO3)2 × 6H2O.The NMR analysis was based on our previous conclusions drawn from the spectral examination of a complex of dimeric pyrazole with magnesium ions. 25The 1 H NMR technique, compared to other methods, is fast and cheap, and enables to follow changes in chemical shifts with no need for a time-consuming alternative NMR approach.Herein it must be added that the use of 15 N or 17 O NMR techniques would result in serious errors and could not be such informative as there are only few nitrogen and oxygen atoms in the structure of 7. Very low natural abundance and a relatively large quadrupole moment renders 17 O NMR method difficult for routine NMR measurements.The Ts Ts first drawback is relevant to 15 N NMR as well.31] The NMR measurements were carried out at three different temperatures, namely at 25, 40 and 60 °C.The shifts in the signal positions refer to the experiments involving separately compound 7 and the same molecule in the presence of magnesium hydrate.In the preceding paper 25 , we concluded that the analysis of the drifts in all proton signals of the molecule indicated that the reference line was not locked during the temperature increase.Therefore, we used the method based on the analysis of relationship between the analyte signals.In order to identify the signals with the highest sensibility to temperature, we focused on the interval changes between the signals.Obviously, the signals that have similar tendencies in the chemical shifts changes will have the same interval between the signals.If one signal drifts downfield and the second stays at the same position or moves less than the first one, then the distance between signals will increase. 25n that paper, we also revealed that the change in pH caused by the addition of magnesium hydrate was insignificant due to the high concentration of DMSO in relation to the analyte.This observation has been confirmed in the present paper as well.The water signal (3.31 ppm) is omitted from the tables due to relatively large drifting that equals to -0.070 and -0.157 ppm at 40 and 60 °C, respectively, or to 0.390, 0.365, and 0.369 ppm at 25, 40 and 60 °C, respectively, after the addition of magnesium salt (the values referenced to the water signal in DMSO).The tendencies in the signal shifts upon temperature change appear to be identical, which means that with the temperature increase, the remaining signals are shifted upfield (Table 1).In order to determine the magnesium salt impact on the change in chemical shifts as a temperature function (matrix MATG after addition of magnesium ions), the following equations were used: a The results are gathered in a graphical form (Figs. 2; Fig. S17, supplementary material).Both graphics demonstrate that the chemical shift of protons B and D changes the most after the addition of magnesium salt.It means that magnesium is chelated in the close proximity to these protons (Scheme 1).Moreover, small changes of ca 0.002 ppm of the chemical shifts of the tosyl C protons (Fig. S17, supplementary material), together with the above finding concerning D protons, may indicate that the neighbouring sulfonyl group is also engaged in the complexation.Interestingly, there are no apparent changes in the shifts of protons A (Tos methyl group) and E (N−methyl group).The latter observation implies that both nitrogens of the N−substituted pyrazole part are not the complexation site.
We also measured the change in pH after the addition of magnesium salt: pH raised from 10.01 to 10.12.That was a small change and therefore it was neglected.Usually, such small changes are omitted in this type of publications. 24,25,30,31 Th pH alterations, even small, should be considered if the molecule contains easily ionisable groups or for compounds of specific physicochemical properties (amino acids, proteins).They are unimportant for small molecules that, like compound 7, contain practically one protonation site.Additionally, we prove that such compounds do not tautomerise. 11,25 oreover, the matrices constructed in our paper cancel the effects of ionic strength and pH, and show that the results are reproducible.It is interesting that the same tendency that determine the changes in chemical shifts was also observed when concentration of compound 7 was lowered (Table S7, supplementary material).The decrease in concentration allowed to eliminate potential influence of the pH alterations on the NMR analysis.

Theoretical investigations
As was suggested before, the investigations and understanding of the relationship between molecular structure and NMR parameters can sometimes be quite difficult, and therefore are often supported by theoretical calculations.The information gathered from quantum-chemical calculations, especially on solvation model and ion solvation sphere, is very valuable and allows for a rapid visualization of the results.This especially relates to the processes proceeding in solvents.The theoretical calculations of the chemical shifts for compound 7 were executed for all energy minima found during the alteration of C3-C2-S1-C1 dihedral angle effected by rotation in 10° increments of the fused pyrazole moiety and tosyl fragment, respectively (optimization at the commonly used B3LYP/6-31G(d,p) level of theory); the four most stable rotamers I−IV are shown in Fig. 3.The calculations demonstrate that the energies of different tautomers of the same rotamer are essentially the same, the energy difference between tautomers is about 0.005 Hartree (14.95 kJ mol -1 ).Therefore tautomerism involving protonation of the nitrogen atoms can be neglected in the above case as possible tautomeric variants of rotamers I−IV of compound 7 have almost identical energy.The calculated proton chemical shifts for the optimized rotamers (the GIAO 32,33 method, CPCM solvation model and water as a solvent) are generally accurate and consistent with the experimental values (Table 2).It is noteworthy that the dielectric constant values of water and DMSO are on a comparable level (ε = 78.3 and 46.8, respectively). 26,27,34 Mreover, the results of the NMR calculations carried out using DMSO as a solvent in comparison with the corresponding data for water are very similar.
Table 2. Calculated and experimental data for 1 H NMR spectra of compound 7, rotamers I-IV; the following parameters were determined for the proton groups of the rotamers: experimental (Exp.) and calculated values of the chemical shifts (rotamers I-IV), percentage relative errors (E1-E4), mean values of the absolute and relative errors (∆δ and E, respectively; average); calculated H ref = 31.7397ppm (GIAO method), mean absolute deviation parameter MAD = 0.92 The relative error for the NH group is large due to the NH proton exchangeability and dynamic dissociation equilibrium in the solvation sphere.This is supported by a simulation of NH proton interactions with a water molecule.The formation of a N-H … H2O hydrogen bond results in a significant drop in the relative error down to 0.6%.Furthermore, we have shown that the relative error for the methyl protons A and E is fairly small (Table 2) and equals to 2% for protons A, and 3% for protons E. The smaller error for the methyl protons A results from their deshielding and the lack of steric hindrance in contrast to the steric obstacle connected with the rotating tosyl and azole systems (protons E).In the latter case, the mean value for the distance between the N−methyl protons E and the tosyl residue is 2.4 Å at most.The theoretical NMR analysis shows also that the mean relative error for the phenyl protons D is 4%.Thus it is notably smaller than for the phenyl protons C located closer to the sulfonyl group.The steric hindrance as well as electron withdrawing properties of the sulfonyl group results in an increase of the relative error of these protons shift to 8%.
One limitation concerning the changes in chemical shifts in the theoretical spectrum of the compound 7-magnesium ion complex is the difficulty to combine conformational analysis with the correct estimation of both various interaction points and solvation sphere for magnesium ions.With this in mind, we carried out optimization of the compound 7-magnesium ion complex using the following models of rotamers (Fig. 4): 7−Mg(H2O)6 (V), 7−Mg(OH) + (VI), 7−Mg(OH)2 (VII).Moreover, we opted for the use of a cheaper, less time consuming GIAO method for the calculations involving 7−Mg complexes (Table 3).The analysis of rotamers V-VII geometries shows that the NH and CH (B) bond lengths undergo changes after optimization (ΔNH=0.06,0.05 and 0.03 Å, respectively; ΔB = 0.05, 0.0002 and 0.0001 Å, respectively).The results of the analysis of the chemical shifts for protons A-E and NH are given in Table 3.It should be added that the use of computation method is tantamount with making assumptions concerning the ligand structure.The above findings seem to indicate that the rotamers in which the magnesium ion is close to the heterocyclic ring are the most appropriate.Moreover, to check accuracy of the computation model, the energy of the interaction N-H ... O-Mg for rotamers V−VII (Fig. 4) was determined using the basis set superposition error (BSSE). 35This energy amounted to ca. 25.1 kJ mol -1 and is located within the energy limits for weak hydrogen bonds.This correlates well with the solvation of magnesium ions observed in our experiment.On the other hand, possibility of the direct contact of magnesium ions with the N1 nitrogen atom was ruled out due to the small value of the calculated coupling constant JN-Mg that was below the experimental error (0.02 Hz, 0.03 Hz, 0.04 Hz calculated at B3LYP/6-31G(d,p), CAM-B3LYP/6-31G(d,p), B3LYP/6-311+G(d,p) level of theory, respectively).Table 3. Calculated and experimental data for 1 H NMR spectra of complex 7−Mg, rotamers V-VII; the following parameters were determined for the proton groups of the rotamers: experimental (Exp.) and calculated values of the chemical shifts (rotamers V-VII), percentage relative errors (E5-E7), mean values of the absolute and relative errors (∆δ and E, respectively; average); calculated H ref = 31.7397ppm (GIAO method), MAD = 0.63.The final stage of the DFT calculations was devoted to investigation of two molecules of 7 that coordinate one hydrated magnesium cation.This led to additional two rotamers IX and X in which an important feature of the realistic system was included.Probably at the experimental basic pH, the hydration shell of the magnesium cation includes at least one hydroxyl ion due to the interplay between basicity of Mg 2+ and nitrogen atoms of compound 7.The starting points for the two optimizations assumed different numbers of the OH -moieties in the complex, but finally the two complexes turned out to be rotamers of 7−[Mg(H2O)5(OH)]−7.This, however, has a little effect on the signals of C-H protons A−E (Table 5) with notable exception of the proton B that moves upfield from ca 7.5 ppm in complexes IV-VIII to ca 6.8 ppm in complexes IX and X (Fig. 6).This is due to a change of theoretically possible protonation site in the azole system.However, such a strong shift was not observed experimentally.Therefore, proton B is indeed among the most perturbed atoms upon addition of the magnesium salt (Table 1).This points out to a conclusion that, regardless whether hydroxyl ions are present or not in the coordination sphere of the magnesium cation, the protonated species of compound 7 and the presence of the hydroxyl anion are insignificant in the solution under these experimental conditions.The DFT calculations reveal that multiple positions are attainable for the interactions between 7 and Mg 2+ .An approximate insight into their relative importance can be gained by molecular dynamics simulation with classical force field (GAFF). 36The model employed in our study is a single molecule of 7 together with one Mg 2+ cation solvated in a 30 Å cubic box of TIP3P water 37 .The resulting radial pair distance distribution functions for the contacts between magnesium cation and selected atoms of 7 are presented in Fig. 7. Evidently, and in agreement with the strong negative electrostatic potential of the sulphonyl oxygen atoms, they are the primary contacts for the Mg 2+ .The cation visits also the vicinity of N1, and also the NH moiety.However, its contacts with the C protons of the phenyl ring are much closer (4−5.5 Å) than with the proton B (7−8 Å).Thus, the experimental observations of slight, but rather similar changes of the 1 H NMR signals of B and C upon addition of the Mg salt suggests that the interaction of the cation with the phenyl ring changes its properties by a "through-space" mechanism.The change in chemical shifts of the CH moiety and the associated proton B is rather propagated through the molecular skeleton, or induced by changes in the solvation sphere.
Prediction of the accurate solvation sphere, conformation and site of the azole-ligand interaction seem to be a complicated process, and we intend to improve the procedure in future.Nevertheless, the depicted herein theoretical calculations have provided an interesting visualization of compounds complex from the quantum chemistry standpoint.

Conclusions
The above discussed method is fast and cheap and enables to follow changes in chemical shifts with no need for time-consuming use of reference compounds and various solvents.We have proved that reliable results can be obtained from the comparative analysis of the changes in chemical shifts in complexes at different temperatures.A potential application of the method has been checked on a model NMR analysis at 25, 40 and 60 °C for the condensed pyrazole derivative 7 and its complex with magnesium ions.The interactions have also been investigated by DFT calculations performed for several rotamers of compound 7 before and after complexation; a stoichiometry of 7−Mg−7 was also considered.Basing on the analysis of changes in chemical shifts and matrix operations we have identified the sites susceptible to interactions with magnesium ions, i.e. the pyrazole nitrogens of the N−unsubstituted heterocyclic part of the condensed pyrazole derivative 7. We have also evidenced that the adduct 7−Mg can additionally be stabilised by the sulfonyl group of the tosyl substituent.The above approach can be used for the analysis of interactions ligand-protein, between a couple of organic compounds, biofluids with metal salts and biofluids with metabolites based on further statistical tests.

Experimental Section
General.The reactions progress and purity of the products were monitored by TLC using silica gel plates Kieselgel 60F254 (Merck).Melting points were determined on a Boetius apparatus and are uncorrected.The 1 H NMR and 13 C NMR spectra were recorded in 5-mm NMR tubes Norrell XR-55 on a Bruker Avance III 300 MHz NMR spectrometer equipped with a variable temperature unit VTU 3200 (300 and 75 MHz) using TMS as internal standard.The temperature of the probe head was calibrated with perdeuterated methanol according to Berger with perdeuterated methanol assuming a linear correlation of temperature and shift differences. 38For temperature-dependent 1 H NMR measurements stock solutions of the compounds with ca. 1 mg of material were dissolved in the appropriate amount of DMSO-d6.After mixing the solutions by vortexing the NMR sample tubes and turning them several times upside down, the NMR samples were thermostatted for at least 10 min at the appropriate temperature (298, 313, and 333 K) in the probe head and, after determination of the water resonance a 1D NOESY spectrum with gradient purge pulses and with a delay time of 10 ms was recorded (manufacturer's pulse program noesy1dpr).For the recording of the 1 H NMR spectra, four dummy scans were used and 64 scans were accumulated with an interpulse delay of 3.5 s.Thereby 32768 data points were collected with a sweep-width of 16.0 ppm, which were for resolution enhancement zero-filled to 131072 data points before fouriertransformation.For the recording of the 13 C NMR spectra with proton decoupling at 298 K, fivefold concentrated solutions were employed and a pulse program was used with an acquisition time of about 0.4 s, 10000 scans and an interpulse delay of 2 s (manufacturer's pulse program udeft, with a flip back pulse at the end of acquisition to reduce the interpulse delay 39 ).Linear prediction to about 262000 data points was made.The 13 C spectra were calibrated with either an external solution of TMS in DMSO-d6 or TMS in DMSO-d6 with a large excess of magnesium nitrate hexahydrate salt.The characterization of the compound 7 was performed with higher concentrated solutions (2.50 mg in 450 µl of DMSO-d6), that is a standard for H,H-COSY, H,C-HSQC, H,C-HMBC and H,H-NOESY experiments.In order to discriminate protons, they were marked according to Scheme 1 (values in ppm): positions: A 2.38, s; B 7.65, s; C 7.43-7.46,m AA'XX'; D 7.85-7.88,m AA'XX'; E 4.09, s; NH 13.67, bs.Low resolution MS spectra were taken on an AMD 402 instrument.IR spectra were recorded on a Specord 75-IR apparatus.Elemental analyses were performed on a Vario EL III instrument.Starting materials 3, 4, and 5 were obtained according to known methods. 28,40,41 Mreover, values of chemical shifts were directly taken from FID files (ASCII files recorded in TopSpin), and all experiments were correctly calibrated.It is worth mentioning that the changes in chemical shifts were caused not by a potential field drift, but by the interactions of the analyte with environment.This conclusion is proved by the matrix operations described in the present paper that shows that the significant changes in chemical shifts correspond only to protons NH and B-D of azole 7.

Theoretical calculations
The theoretical DFT calculations were executed using Gaussian G09 D.01 suite code. 34The structures of the different rotamers of compound 7 were optimized (no negative frequencies were observed in the generated vibrational spectrum of the optimized rotamers) using the density functional theory (DFT) 42 with the commonly used B3LYP 43 functional and standard basis 6-31G(d,p). 44The solvent effect on the geometry of compound 7 and complex 7−Mg was determined using a conductor-like polarized continuum (CPCM) model (water as a solvent). 45Vibrational frequencies and thermodynamic properties were calculated applying the ideal gas, rigid rotor, and harmonic oscillator approximations. 46Energy minimum was confirmed by the frequency calculation for all rotamers (no negative frequencies were detected in the generated vibrational spectrum of the analyzed rotamers).The initial structures of magnesium complex 7−Mg were designed by a change in the C3-C2-S1-C1 torsion angle and rotation of the tosyl group.The NMR parameters, were calculated using the coupled perturbed density functional theory (CP-DFT) 47 method with B3LYP functional.This functional has been found to be in good agreement with experimental data. 48The calculations were performed using the NMR dedicated GIAO 32,33 method.The same solvent models as employed for the geometry optimization were used in the calculation of scalar couplings. 44The proton chemical shifts were referenced to the central signal of TMS.The proton shift equalled 31.7397ppm (for B3LYP/GIAO//B3LYP/6-31G(d,p) approach).The shifts of compound 7 and reference TMS were calculated using the same method whereas the reference TMS was used to obtain the chemical shifts of 1 according to the following equation: δi = σref − σi, where δi was the chemical shift of i-nuclei of 1 and σref and σi were the calculated isotropic magnetic shielding tensor for the TMS and 1, respectively. 26,27,50 Te calculated chemical shifts for the homotopic protons were averaged.The Chemcraft 1.7 software was utilized for visualization of all optimized rotamers. 51he classical molecular dynamics calculations with the GROMACS 5.0.5 suite 52 employed the GAFF general organic force field 36 and TIP3P water model 37 .The molecule of 7 was immersed in a 30 Å cubic box of water and a randomly placed magnesium cation was added.To remove bad contacts, 100 steps of minimization was carried out, then a two-stage equilibration was performed.First, an NVT (constant number, volume, and temperature conditions) run at 300 K was carried out for 0.3 ns, and then an NPT (constant number, pressure, and temperature conditions) run at 300 K and 1 atm was performed for 0.2 ns.Finally the NPT production run lasted for 5 ns.The timestep was 1 fs, and PME electrostatics with 10 Å real space nonbonded cutoff was employed.The trajectory was analysed using the VMD 1.9.1 program. 53 , and Fig. 1; Tables S1−S6 and Fig S13−S20, supplementary material).

Figure. 1 .
Figure. 1. Enhanced 1 H NMR spectra of aromatic region of compound 7 with (d-f) and without (a-c) magnesium salt taken at 25 (a or d), 40 (b or e) and 60 °C (c or f).The next step involved construction of a 5 × 5 matrix (MAT) for the intervals between signals.The results of the analysis of NMR spectra registered at 40 and 60 °C were elaborated identically (Tables S1−S6, supplementary material).Then, the matrices were subtracted for the visualisation of the interval changes in chemical shifts influenced by temperature.As a result a MAT40-MAT25 matrix and a MAT60-MAT40 matrix (without Mg 2+ ) were obtained for temperatures 40 and 25 °C as well as 60 and 40 °C, respectively.For the analysis of the obtained results a matrix was proposed in which colours were used instead of numbers (Fig.2; Fig. S13−S17, supplementary material).The matrix reveals clearly that the largest temperature-dependent changes in chemical shifts for azole 7 refer to proton 3-H (B, Fig. S14, supplementary material).On the other hand, the presence of Mg 2+ ions affects insignificantly the proton D signal of the tosyl substituent (Fig. 2; Fig. S15−S17, supplementary material).

Figure. 3 .
Figure. 3. Visualisation of the selected optimized rotamers of compound 7.The structures I−IV were optimized at the B3LYP/6-31G(d,p) level of theory with CPCM solvent model.

Figure. 4 .
Figure. 4. Binding modes of Mg 2+ cations to compound 7.The structures V−VII were optimized at the B3LYP/6-31G(d,p) level of theory with CPCM solvent model.Basing on NMR experiments, we claim that chemical shifts of protons B and D change the most upon interaction.However, the data in Tables 1, S1−S6 (supplementary material) indicate that the changes in C and D protons chemical shifts are slight.This is probably due to the migration of magnesium cations around the phenyl moiety during optimization caused by the close proximity of the sulphonyl group.Additionally, the results of calculations given in Table 4 and Fig. 5 (rotamer VIII) proved the possible influence of the presence of magnesium ions on changes in chemical shifts of proton A−E and NH of azole 7, especially A−D.Noteworthy is good agreement of the calculated chemical shifts with the experimental values.

Figure. 6 .
Figure. 6. Binding modes of complexes of compound 7.The structures of IX-X were optimized at the B3LYP/6-31G(d,p) level of theory with CPCM solvent model.

Figure 7 .
Figure 7. Radial distribution function of distances between the magnesium cation and the indicated atoms of compound 7 resulted from the classical molecular dynamics simulation.

Table 4 .
Calculated and experimental data for 1 H NMR spectra of complex 7−Mg(H2O) + , rotamer VIII; the following parameters were determined for the proton groups of the rotamers: experimental (Exp.) and calculated values of the chemical shifts (rotamer VIII), mean values of the absolute and percentage relative errors (∆δ8 and E, respectively); calculated H ref = 31.7397ppm (GIAO method), MAD = 0.49.

Table 5 .
Calculated and experimental data for the 1 H NMR spectra of complex 7−[Mg(H2O)5(OH)]−7 (rotamers IX-X); the following parameters were determined for the proton groups of the rotamers: experimental (Exp.) and calculated values of the chemical shifts (rotamers