Characterization of Amide Bond Conformers for a Novel Heterocyclic Template of N-acylhydrazone Derivatives

Herein we describe NMR experiments and structural modifications of 4-methyl-2-phenylpyrimidine-N-acylhydrazone compounds (aryl-NAH) in order to discover if duplication of some signals in their 1H- and 13C-NMR spectra was related to a mixture of imine double bond stereoisomers (E/Z) or CO-NH bond conformers (syn and anti-periplanar). NMR data from NOEdiff, 2D-NOESY and 1H-NMR spectra at different temperatures, and also the synthesis of isopropylidene hydrazone revealed the nature of duplicated signals of a 4-methyl-2-phenylpyrimidine-N-acylhydrazone derivative as a mixture of two conformers in solution. Further we investigated the stereoelectronic influence of substituents at the ortho position on the pyrimidine ring with respect to the carbonyl group, as well as the electronic effects of pyrimidine by changing it to phenyl. The conformer equilibrium was attributed to the decoplanarization of the aromatic ring and carbonyl group (generated by an ortho-alkyl group) and/or the electron withdrawing character of the pyrimidine ring. Both effects increased the rotational barrier of the C-N amide bond, as verified by the ΔG≠ values calculated from dynamic NMR. As far as we know, it is the first description of aryl-NAH compounds presenting two CO-NH bond- related conformations.


Introduction
The bioactive N-acylhydrazone (NAH) moiety has been identified in a great number of lead compounds that act on different types of molecular targets [1][2][3][4]. Because of the assemblage of amide and imine functions, NAH compounds may exist as C=N double bond stereoisomers (E/Z) (Scheme 1) and as syn/antiperiplanar conformers about the amide CO-NH bond (Scheme 1) [5].

Scheme 1. General structure and stereochemistry of NAH.
A research program to develop a series of 2-phenyl-4-methylpyrimidine-N-acylhydrazone compounds with antinociceptive and anti-inflammatory activities led to the discovery of LASSBio-1083 (1, Figure 1), a promising lead compound that shows an ED 50 of 27 µmol/Kg in the acetic acid-induced mouse writhing test [6]. Regarding full structural characterization, the 1 H-NMR spectra revealed the duplication of some signals, indicating the presence of a mixture of stereoisomers or conformers.
Using 1 H-and 13 C-NMR data, Palla and coworkers posited that NAH compounds derived from the condensation of hydrazides and aromatic aldehydes are present in solution as the E geometric isomer, which is less sterically hindered compared to the Z form [7]. However, starting from the pyridine-2-carboxaldehyde, the Z isomer can be detected in less polar solvents due to its stabilization with intramolecular H-bonds [5]. On the other hand, the duplication of NMR signals has been attributed to the presence of anti-and synperiplanar conformers in benzyl and alkyl NAHs [8][9][10], but there is no description of this phenomenon in aryl NAHs in the literature. Because the complete knowledge of structure, including stereochemistry, is essential for lead optimization in drug discovery, we herein describe our studies concerning the elucidation of the signal duplication in the 1 H-NMR of 1 and the modifications of its structure leading to compounds 2a-d and 3a to better understand the stereoelectronic properties that promote this effect ( Figure 1).

Synthesis of N-acylhydrazone Derivatives 1-3
The key precursors for the synthesis of NAH 1 and 2a-d, the phenylpyrimidine and biphenyl hydrazides 11a-e, were obtained in 79% to 95% yield by nucleophilic acyl substitution reaction of 4 with 100% hydrazine hydrate in ethanol at 55-80 °C (Scheme 5) [16]. The desired compounds 1 and 2a-d were synthesized using the classic acid-catalyzed condensation of hydrazides 11a-d with 4-dimethylaminobenzaldehyde in ethanol at room temperature in 85%-99% yield (Scheme 5 and Table 1). Analysis of the 1 H-NMR in DMSO-d 6 showed the formation of NAH 1 and 2a-d by the presence of signals attributed to the N=CH and CONH protons, but for some NAH (1, 2a-c), these signals appear to be duplicated (Table 1). Finally, the N-methyl-NAH derivatives 3a were obtained from a selective N-methylation of 1 using methyl iodide in a basic medium of K 2 CO 3 at 43 °C in 77% yield (Scheme 5 and Table 1) [17]. The presence of signals referring only to the N-CH 3 protons indicates the selective N-methylation of NAH. Additionally, no signal duplication was observed for these N-methyl-NAH derivatives. The 1 H-NMR spectrum of the analgesic lead compound LASSBio-1083 (1) is presented in Figure 2A and shows the duplication of the hydrogen signals. Assuming that these duplications could be attributed to the presence of the two possible isomers (E/Z) of the imine double bond, we decided to proceed with differential Nuclear Overhauser Effect (differential NOE) experiments to assess the spatial proximity of 1 H-1 H. The hydrogen atom selected for irradiation was the NH of the amide. Due to the presence of two singlets related to the NH hydrogen, we chose the one that shifted less and that was more prevalent at 11.67 ppm ( Figure 2B).
Although only one amide signal at 11.67 ppm was selected for irradiation in the NOE experiments, the amide hydrogen at 11.79 ppm presented the same irradiated signal phase. Furthermore, both signals for H 6 and N=CH showed increased intensities, presenting a positive NOE of 11% and 15%, respectively ( Figure 2B).
Because of the angle and distance, a positive NOE on N=CH was not expected from NH irradiation of the Z isomer. Calculations using Mspin 1.03 software [18] were performed to simulate the NOE effect at a radius of 5 Å around the irradiated hydrogen. For this simulation, more stable E (A/B) and Z (C) stereoisomers, obtained through a process of geometric optimization by molecular mechanics followed by conformational analysis using the semi-empirical method AM1 in the PC Spartan Pro 1.0.5 software [19], were selected (Table 2). Although the B conformer was not as stable as A, the choice was based on observations from other NAH compounds in X-ray diffraction studies (Table 2) [17]. The theoretical results for amide NH irradiation showed that this effect should be observed mainly in H 6 and N=CH for both E (A and B) conformers and be more pronounced at N=CH in the B conformer ( Table 2). The same analysis for the Z stereoisomer showed that the greatest positive NOE should be observed at H 2 " followed by H 6 , but no significant positive NOE would be expected for N=CH (Table 2).
To clarify the reason for the negative phase of the minor NH signal and exclude the possibility of a technical artifact due to the proximity of signals, we decided to irradiate the signal of the less prevalent H 2 "/H 6 " at 7.20 ppm. This location was relatively distant from the multiplet and doublet neighbor signals, whereas the signal corresponding to H 2 ′′/H 6 ′′ in greater proportion was located inside a multiplet at 7.54 to 7.57 ppm ( Figure 3). The irradiation led again to two negative signals corresponding to both signals of H 2 ′′/H 6 ′′ (7.20 and 7.55 ppm) and a positive one for both signals of N=CH and H 3 ′′/H 5 ′′, presenting a positive NOE of 9% and 8%, respectively ( Figure 3). Curiously, only the H 2 ′′/H 6 ′′ signal shifted at 7.55 ppm inside the multiplet, and no other hydrogen presented a negative signal , showing that the inversion of phase was not an artifact of the technique due to the proximity of signals ( Figure 3).
Additionally, we obtained a 2D-NOESY spectrum and verified that each expected diagonal signal from a conformer presented a cross-correlation in-phase with the correspondent signal of the other conformer (supplementary material). This effect was dependent on the experimental mixing time (D8 of 40, 80 and 300 ms), with longer times resulting in more intense cross-correlations.  Normally this type of cross-correlation can be attributed to an artifact because of a chemical exchange with solvent. However, this is not the case, as we observed a cross-correlation for all protons including those that were not exchangeable. The cross-correlation is most plausibly due to an exchange among multiple conformers in which a 1 H has a distinct chemical shift in each conformer. The time required for one conformer to become another explains the increase in intensity with longer mixing times.
These NOE experiment results led to the hypothesis that each signal could be related to the presence of two conformers. Moreover, we could exclude the Z stereoisomer because all NOE results are in agreement with the more stable E stereoisomer [20], as theoretically predicted. Thus, the hydrogen of NH could be synperiplanar (sp) or antiperiplanar (ap) in relation to the carbonyl oxygen atom of the E isomer ( Figure 3). This hypothesis was corroborated by HPLC studies in which only one compound was detected as well as by the change in the ratio between the pair of duplicated signals when CDCl 3 was used in place of DMSO-d 6 for the 1 H-NMR analysis, as previously observed by Palla and coworkers for alkyl-NAH [7].

NAH Derivative 1 Conformers Determination
Based on the work of Quattropani and colleagues [10], who described the presence of conformers for alkyl NAH compounds (Figure 4), we carried out a 1 H-NMR experiment in DMSO-d 6 at 80 °C to determine whether the coalescence of duplicated signals would be observed ( Figure 5). A complete coalescence was observed for the NH, H 6 , H 3 ′′/H 5 ′′, N(CH 3 ) 2 and CH 3 signals, while partial coalescence was observed for N=CH and H 2 ′′/H 6 ′′. The reversibility of these changes was verified when the experimental temperature was returned to 20 °C. This result corroborated the hypothesis of the conformers: the energy required to faster overcome the rotational barrier is reached upon an increase in temperature, leading to a rapid conversion between the conformers. Traditionally, the amide group is represented by the canonical forms 1 and 3.

A B
The resonance effect between the lone pair on the nitrogen sp 2 atom and the carbonyl π bond contributes to the double bond character of the amides in the planar form. In this context, the resonance structure 3 is considered more significant for the double bond character and rotational barrier [21].
Converting a conformer into another requires rotation around the C-N bond, which consequently changes the nitrogen sp 2 hybridization to a pyramidal arrangement. In the pyramidal geometry, the lone pair of nitrogen is placed in an orbital with high s character for stabilization. The disruption of favorable interactions between N and C raises the rotational barrier. In the case of formamide, the barrier is 18 kcal mol −1 [22][23][24][25].
Wiberg et al. conducted theoretical studies concerning the amide bond rotation. Their calculations showed that the C-N bond lengthens by 0.08 Ǻ upon rotation from the planar form to the transition state, whereas the C-O bond shortens slightly (by 0.01 Ǻ). Moreover, an insignificant difference in charge density at the oxygen was observed in the planar form and the transition state. Hence, the oxygen is assumed to simply polarize the C-O bond, resulting in a large δ + on the carbon [22].
As the earlier model with two canonical structures did not explain these results, one that incorporated a carbonyl dipolar canonical form was suggested (2). Therefore, the rotational barrier is attributed mainly to the resonance structures 2/3 and 5/6 [22,23].
N-Methylation of NAH 1, which results in N-methyl-NAH 3a, causes conformational restriction of the amide bond, thus preventing its rotation, as previously described by our group [17,26]. In fact, this modification abolished the duplication of the signals in the NMR spectra. Another approach used to evaluate the presence of two conformers was the synthesis of isopropylidene hydrazones. The latter was based on the works of Wyrzykiewicz and Prukala [27] and Palla [5], to eliminate the possibility of forming E/Z stereoisomers. The compound (isopropylidene)-2-phenyl-4-methyl-pyrimidine-5-hydrazide (12) (Figure 6) was synthesized by condensation of 2-phenyl-4-methyl-pyrimidine-5-hydrazide (11a) with acetone. Its 1 H-NMR spectrum presented two singlets for each NH, H 6 and CH 3 a and another three for CH 3 b and CH 3 c. Regarding CH 3 a, the signal present to a lesser extent showed the same chemical shift of the solvent DMSO-d 6 (2.48 ppm). Likewise, the signal for the smaller proportion of CH 3 c and that for the greater proportion of CH 3 b also coincided (both were at 1.9 ppm). This result confirmed the presence of conformers. Considering the proportion of the conformers, we believe that the more deshielded signal from N=CH matches the conformer ap, based on the work by Palla in which the more deshielded signal was attributed to imine hydrogen of conformer ap in DMSO-d 6 and in CDCl 3 [7]. Furthermore, in the ap conformation, the two dipolar moments (C δ+ =O δ− and N δ− -H δ+ ) are aligned, which does not occur in the sp conformer [21].
According to literature reports, the presence of conformers for the NAH compounds is typically observed in those with a spacer between the carbonyl group and the aromatic ring or in alkyl-N-acylhydrazones. However, NAH compound 1 does not structurally match the compounds described in previous reports. Perhaps an ortho steric effect, exerted by the methyl group on the carbonyl, and/or an electronic effect induced by the pyrimidine ring could be related to the observed phenomenon [28]. Thus, we decided to study the stereoelectronic requirements for the existence of conformers.

Stereoelectronic Effects on NAH Derivative (1) ap-sp Amide Bond Rotamers
To evaluate the ortho steric effect, the methyl group was replaced by an H atom (2a) and an ethyl group (2b). The first modification represents the absence of any steric effect while the second one increases the steric effect. Both compounds showed duplicated signals in the 1 H-NMR spectra, as shown in Table 3. Additionally, the pyrimidine ring was replaced with a phenyl ring to verify the influence of the electronic effect on the amide bond, generating compounds 2c and 2d. Thus, we could study the influence of the heterocyclic ring both in the presence (2c) and absence (2d) of an ortho steric effect. Only 2d did not show duplicated signals. Therefore, we found that the steric effect and/or the pyrimidine ring induced the formation of conformers in this class of NAH. Thus, the decoplanarization between the carbonyl and the aromatic ring (induced by the ortho-alkyl group) as well as the electron withdrawing effect exerted by the pyrimidine ring on the carbonyl group increase polarization. Both effects are disadvantageous in regard to the transition state between one conformer and the other. The transition state between amide rotamers is when the C-N bond is turned 90° and the nitrogen is not conjugated with the carbonyl (resulting in a pyramidalized N atom). As stated by Olsen and colleagues for nicotinamide and picolinamide, in the transition state, the π electrons from the aromatic ring can stabilize the resonance forms B, lowering the transition state energy and facilitating amide bond rotation ( Figure 7). As shown, 2d and 2a are capable of coplanarizing their aromatic π electrons in the B canonical form with C + -O − . However, while this feature stabilizes the canonical form B1 in compound 2d, the recognized electron withdrawing effect of pyrimidine reduces stabilization (form B2 in compound 2a), leading to increases in the transition state energy and rotational barrier (Figure 7). This type of effect is supported by the observed 5.4 kcal/mol difference in the amide bond rotational barrier between nicotinamide and picolinamide [25]. Additionally, due to the more difficult coplanarization generated by ortho-alkyl groups, compounds 1 and 2c present a more energetic transition state B3 and thus a higher rotational barrier.
The 13 C-NMR spectra support these assumptions, as the carbon from the carbonyl bond is deshielded by approximately 2 and 3 ppm after the introduction of an ortho-alkyl group or the replacement of the pyrimidine ring with a phenyl, respectively (Table 4).  The higher sp 2 amide nitrogen character attributed to compounds 1 and 2a-d was evaluated using HMBC, which allowed us to distinguish between the aromatic carbons directly attached to a carbonyl that is cis and those attached to a carbonyl that is trans to the amide N-1 H three bonds away. Due to the local bond constraints and planarity, two vicinal bond angles (0° and 180°) are present in the stable conformers. The HMBC correlations show that for compounds 1 and 2a-c, the signals of the amide N-1 H of each conformer are coupled differently to the aromatic 13 C ( 13 C-CO-N-1 H). Only the minor conformer, which has a trans correlation between these atoms, shows a 3 J CH coupling (Table 5). These results agree with the literature showing a higher coupling for the trans coupling compared to cis [29,30]. On the other hand, compound 2d did not show a 3 J CH coupling in the same experiments, possibly due to the minor sp 2 amide nitrogen character and the longer length of the C-N bond, which results in a decrease in the rotational barrier compared to the other compounds (1, 2a-c). Because of this small barrier, 2d does not possess the fixed trans conformation necessary for 3 J CH coupling in this series. In addition, 3 J CH coupling of the amide N-1 H and imine 13 C=N was present in all compounds, indicating the similar chemical environments with no great differences in conformation for this moiety of NAH compounds. The Gibbs free energy of activation (∆G ≠ ) related to the transition state among one conformer and another can be calculated using Equation 1. Through a dynamic NMR process, we could estimate the coalescence temperature (T c ), which provides, in association with the maximum peak separation (Δv in Hz) at low-temperature (i.e., slow exchange between conformers), the energy of the rotational barrier (∆G ≠ ) [31]: The ∆G ≠ values showed that the steric effect exerted by the methyl group on the ring decoplanarization and the electron withdrawing effect of pyrimidine have similar effects on the rotational barrier because compounds 2a and 2c have almost the same ∆G ≠ (Table 6). A small additional increase in the barrier was observed when both effects are present, as observed for 1 and 2b, while no change was found when replacing the methyl group with an ethyl group. Elemental analyses were carried out on a Thermo Scientific Flash EA 1112 Series CHN-Analyzer. Thin layer chromatography was performed on Merck Kieselgel 60 HF 254 plates and detection took place using UV (254 and 365 nm). All reagents and solvents were purchased from commercial suppliers and used without further purification.

2-Phenylpyrimidine-5-carbohydrazide (11c)
To a solution of ethyl 2-phenylpyrimidine-5-carboxylate (4c) (0.080 g, 0.350 mmol) in ethanol (2 mL) was added hydrazine monohydrate 100% (0.351 g, 7.01 mmol). The reaction mixture was stirred at room temperature for 2 h. Ice was added to the reaction flask and the solid, filtered and washed with cold water. The title compound was obtained as a white amorphous solid in 85% yield.

General Procedure to Synthesize the N-Acylhydrazones 1, 2a-d
To a solution of an aromatic cabohydrazide (1.10 mmol) in ethanol (5 mL) were added HCl catalytic (2 drops) and the appropriate aromatic aldehyde. The reaction mixture was stirring at room temperature for ca 30 min, and then poured into ice. The precipitate was filtered, washed with cold water, petroleum ether and recrystallized in EtOH.

Molecular Modeling
The sketching, geometry optimization and conformational search of compounds were performed in PC Spartan Pro 1.0.5 software. They were subjected to structural minimization by the use of molecular mechanics, using the base MMFF. Subsequently, the conformational analysis was performed, using the semi-empirical AM1.
The NOE prediction study was performed by selecting the minimum conformation of each stereoisomer (E/Z), and additionally the election of a second conformer (B) for the E isomer which ΔHf = 114.68 kcal/mol. Conformers that have the carbonyl group synperiplanar related to NH hydrogen were obtained by a dihedral angle search in 12 steps of 30° on the amide bond CO-NH, leaving the other angles free. Then we selected the lowest energy conformer, generated by the conformational analysis with the semi-empirical AM1 method, which presented the carbonyl synperiplanar related to the NH. The NOE calculations were carried out in the MSpin 1.03 software from the selected structures from the conformational analysis.

Conclusions
In this work, we revealed that duplicated signals observed in NMR spectra of 4-methyl-2phenylpyrimidine-N-acylhydrazone correspond to the presence of two CO-NH bond-related syn and antiperiplanar conformers. To our knowledge, this was the first description of conformers in aryl-NAH compounds. Our assumptions were based on NMR data from NOEdiff spectra of NH and phenyl H 2'' /H 6'' irradiations, 2D-NOESY and dynamic 1 H NMR and also by synthesis of isopropylidene hydrazone. The possibility of conformer's observation rises from the increase of rotational barrier (∆G ≠ ) among them, which results from both the decoplanarization of the aromatic ring and carbonyl group (induced by an ortho-alkyl group) and the electron withdrawing nature of the pyrimidine ring.