Conformational and Stereodynamic Behavior of Five- to Seven-Membered 1-Aryl-2-iminoazacycloalkanes

The stereodynamic behavior of 1-arylpyrrolidin-2-imines, having a Caryl–N stereogenic axis, has been studied by means of dynamic nuclear magnetic resonance and density functional theory calculations, evaluating the steric effect of ortho-aryl substituents. The rotational barrier due to E/Z isomerism about the −C=N–H bond was also determined. The dynamic stereochemistry of homologous six- and seven-membered iminoazacycloalkane rings and their oxo-analogues was also comparatively investigated, evidencing a ring size effect. It was found that the seven-membered heterocycle shows additional dynamic features because of ring inversion.


■ INTRODUCTION
Many drugs containing small-ring nitrogen heterocycles exhibit a wide range of biological activities, 1 and it is widely reported that the potency and selectivity are strongly sensitive to the conformational constraints, as the ring size enlarges from five to seven members. The combination of ring size and the appropriate spatial and steric properties of the substituents bonded to the ring lead to the preparation of potent and selective drugs of great interest in medicinal chemistry. In particular, 2-iminoazaheterocycles are potent inhibitors of human nitric oxide synthase (iNOS), 2 and a recent synthesis of 1-aryl-2-iminoazacycloalkanes from ω-halonitriles using polyphosphoric acid esters such as ethyl polyphosphate (PPE) or trimethylsilyl polyphosphate (PPSE) 3 gave the possibility to study the conformational behavior about the N-aryl stereogenic axis in five to seven membered heterocyclic rings. These compounds can be easily converted into their oxo-analogues by exposure to air (Scheme 1). Examples of dynamic conformational analysis due to a C sp 2 −N stereogenic axis have been reported: 4 barbiturates are the most studied class of biologically active compounds, 5 and lactams, 6 pyrroles, 7 indoles, 8 imides, 9 azalidine-4-ones, 10 thiazolidine-2-thiones, 11 and xanthines 12 are found to exhibit stereodynamic properties.
1-Aryl-2-iminoazacycloalkanes are expected to have the plane of the ortho-substituted phenyl ring significantly twisted with respect to the time-averaged plane of the five-, six-, and seven-membered heterocyclic ring. The existence of such an Ar−N stereogenic axis would originate, in principle, a pair of stereolabile enantiomeric forms when the rotation rate is rendered sufficiently slow in the nuclear magnetic resonance (NMR) time scale. The aim of this paper is a thorough investigation of the stereodynamic behavior of a new class of compounds having a C aryl −N stereogenic axis in a fivemembered saturated ring, 13 (1-arylpyrrolidin-2-imines 1−9 and their oxo-analogues 18−20 Scheme 1), and the evaluation of the steric effect of the ortho-aryl substituents. Moreover, the six-and seven-membered imino homologues (10−13 and 14− 17, respectively) and their oxo-analogues 21−22 have been prepared to evaluate the effect of the ring size and of the exocyclic heteroatom on the N-aryl rotational barrier.

■ RESULTS AND DISCUSSION
When an ortho-substituted aryl ring is bonded to pyrrolidin-2imines (Scheme 1), the steric hindrance caused by the orthosubstituent forces the aryl ring to adopt a skewed conformation with respect to the almost planar heterocyclic ring, thus generating a stereogenic axis and a pair of conformational enantiomers when the rotational barrier is frozen; depending on the value of the rotational barrier, stereolabile enantiomers or atropisomers can be generated. In addition to the stereogenic axis, a second source of stereoisomerism could be anticipated because of the presence of the imine moiety that can have E or Z configuration. 14 This conformational feature must be carefully considered because of the different steric hindrance exerted by the NH moiety in the E or Z stereoisomer on the N-aryl ring rotation. To evaluate the energy barriers for the N-aryl rotation, density functional theory (DFT) calculations were performed using the M06-2X functional and the 6-311+G(d,p) basis set. All the optimized structures were verified to be true ground states by frequency analysis, showing the absence of imaginary frequencies. For each E/Z stereoisomer, the optimized structures showed that the aryl ring was twisted with respect to the heterocyclic ring with C 2′ −C q −N−C 2 dihedral angles (ϕ) of about 60°and 120°, yielding a total of four ground states (for compound 1, see Figure 1, top). The interconversion of the two conformational enantiomers can occur by the rotation of the aryl ring through two different transition states where the heterocyclic and the aromatic ring are almost coplanar, with dihedral angles ϕ close to 0°or 180°(TS1 and TS2 in Figure 1, bottom). In the case of compound 1, the calculated energy barriers were 9.5 and 14.3 kcal mol −1 for the E stereoisomer, whereas they were calculated as 15.5 and 17.5 kcal mol −1 in the Zstereoisomer (all the DFT-calculated energies and the computational details for compounds 1−22 are reported in the Supporting Information). The higher values calculated of the Z-stereoisomer are due to the larger steric hindrance caused by the imine hydrogen in the transition state geometries.
The lowest energy transition state is that of the E stereoisomer with a dihedral angle ϕ close to 180°; this geometry avoids the steric clash between the ortho-substituent of the aryl ring and the imine NH and simultaneously minimizes the steric interaction between the ortho-H and the NH. The calculated energy barriers range from 2.9 kcal mol −1 for compound 4 (o-fluoro) up to 21.3 kcal mol −1 for compound 3, bearing a tert-butyl substituent in the ortho position. These energy barriers can be conveniently observed by the variable-temperature NMR technique (D-NMR), 15 thanks to the presence of enantiotopic groups such as the two hydrogens of CH 2 , which turn into anisochronous signals when the aryl rotation becomes slow in the NMR time scale. An example of the dynamic NMR experiment is shown in Figure 2 for compound 1.
At ambient temperature, the enantiotopic CH 2 in position 5 shows a triplet signal because of the coupling with CH 2 in position 4. On lowering the temperature, the signal broadens and reaches the coalescence point at −64°C, yielding a diastereotopic ABX 2 system below −91°C. From the rate constants obtained by line shape simulations at different temperatures, the activation energy value of 10.2 ± 0.15 kcal mol −1 was derived using the Eyring equation. As usually happens in conformational processes, the activation entropy was found to be negligible. 15 The same approach was used for compounds 2−22, and the experimental and calculated values are reported in Table 1 (variable-temperature NMR spectra and line shape simulations are reported in the Supporting Information). 16 In the case of compound 3, the calculated energy barrier (21.3 kcal mol −1 ) is much higher than the other compounds and indeed the two diastereotopic 5-CH 2 signals do not show any sign of dynamic exchange up to +120°C, where the product begins to decompose. Therefore, a kinetic study was performed using one-dimensional exchange spectroscopy NMR (1D-EXSY NMR) 17 at +105°C and +110°C in dimethyl sulfoxide (DMSO-d 6 ). The saturation transfer due to aryl rotation increases on raising the mixing time. First-order kinetic analysis allowed to determine the rate constants, yielding a ΔG ⧧ value of 22.0 kcal mol −1 , in very good agreement with the DFT-calculated data (Figures 3 and S21 in the Supporting Information).
On the other side, compound 4 has the lowest calculated energy barrier (2.9 kcal mol −1 ). In this case, the spectra had to be recorded below −100°C using CDFCl 2 as the solvent ( Figure 4). The dynamic phenomenon begins at −149°C,

ACS Omega
Article when the CH 2 signal broadens, reaches the coalescence at −156°C, and eventually splits showing four signals at −165°C . This occurrence implies that two diastereomeric conformations are present in a 73:27 ratio, both showing diastereotopic hydrogens. An additional dynamic process, ascribable to the E/Z conversion, has thus to be taken into account. The line shape simulations at different temperatures should be performed using three rate constants (N-aryl rotation in the E stereoisomer, N-aryl rotation in the Z stereoisomer, and E/Z interconversion), but good simulations were obtained by using only two kinetic constants. The first one accounts for the N-aryl rotation through the more stable transition state where the imine is the E configuration ( Figure  4, AB peaks exchange). The second rate constant considers the exchange of E/Z stereoisomers ( Figure 4, AC and BD peaks exchange), obtaining two energy barrier values of 5.3 ± 0.3 and 6.5 ± 0.3 kcal mol −1 , respectively. It must be underlined that the rotational rate constant for the aryl ring in the imine Z configuration is negligible because of the larger steric hindrance caused by the hydrogen in the planar TS. This occurrence implies that the rotational barrier in the stereoisomer is higher than the E/Z barrier, so the aryl rotation occurs only when the imine is in the E geometry. 18 To confirm the assignment of this second dynamic process (i.e., E/Z interconversion of imine), we synthesized compound 9 with a p-methylphenyl substituent. This compound cannot have enantiomeric conformations because of local symmetry of the aryl ring, and any dynamic phenomenon has to be related to the E/Z conversion, which yields two diastereomeric conformations. Figure 5 shows the NMR spectra of compound 9. At −128°C, the CH 3 signal broadens and eventually coalesces at −133°C. At −149°C, two distinct diastereomeric singlet signals in a 70:30 ratio are visible. An energy barrier of 6.9 ± 0.2 kcal mol −1 was derived from line shape simulations at different temperatures, a value identical to the higher barrier detected for compound 4.
Using the B-value scale reported by Lunazzi et al. 19 as an indicator to estimate the steric hindrance of various orthosubstituents, the very good correlation between the B-values and the experimental ΔG ⧧ values of 1, 3−8 (black rhombus in Figure 6, the table of values is reported in the Supporting Information) suggests that the magnitude of the enantiomerization barriers is mainly due to steric effects, as also forecasted by DFT calculations. It should be noted that the rotational barrier of the o-chloro compound is slightly smaller than that of the o-methyl, while the B-values series has opposite trend. However, this result is in agreement with a recent evaluation of the steric size of common substituents 11b and with the steric size scale proposed by Sternhell. 20 Six-and Seven-Membered Amidines. Switching from five-membered rings to the six-membered analogues, the energy barriers increase by about 5 kcal mol −1 (see Tables 1 and S6 and Figure S22 in the Supporting Information). This effect was previously observed for the rotational barriers of Naryl fiveand six-membered heterocyclic amidines (imidazolines and tetrahydropyrimidines, respectively). 21 For 2iminoazaheterocycles, the higher experimental value of sixmembered derivatives is most probably the result of two opposite factors. On the one side, the higher steric constraints imposed in the TS by the different ring size should raise the rotational barrier because of the higher steric hindrance of the imino group in the TS. This effect is due to the geometry of the six-membered ring that moves the NH moiety toward the aryl ring in the TS geometry. On the other side, the higher flexibility of the six-membered ring allows the N-aryl group displacement in a pseudo-equatorial position, where the steric hindrance of the exocyclic imino moiety is reduced. This preference is well reproduced by DFT calculations on compound 11 ( Figure S23 of the Supporting Information). The global minimum (GS4-Z in Figure S23) has the aryl ring in a pseudo-equatorial disposition, while the threshold transition state corresponds to the rotation of the ofluorophenyl in a pseudo-axial disposition and with the fluorine on the CNH side (TS3-Z in Figure S23 of the Supporting Information).
In the case of the seven-membered ring, the increased flexibility of the ring compensates the higher steric hindrance on the TS and the energy barriers are similar to the sixmembered rings. However, the larger size of the ring combined with the planarity of the imino portion makes the system to behave similar to a cyclohexane, with the N−C 1 moiety replacing a single carbon of cyclohexane. 22 The N-aryl ring can arrange itself onto a pseudo-equatorial or pseudo-axial position, and both conformations have similar energies (GS5-in and GS7-out, Figure 7).
In the case of compound 15 ( Figure S11 in the Supporting Information), the triplet of the CH 2 signal splits into two diastereotopic signals below −62°C and splits again into four signals on further lowering the temperature below −75°C. This behavior can be rationalized by considering the simultaneous freezing of the aryl rotation and the ring inversion, yielding two conformational diastereoisomers in a ≈50:50 ratio, each one showing diastereotopic hydrogens for CH 2 (the E/Z interconversion takes place at much lower temperatures and cannot be responsible for the observed

ACS Omega
Article spectra). The two dynamic processes have similar energies (9.5 ± 0.2 and 11.1 ± 0.2 kcal mol −1 , respectively). Compound 16 (aryl = o-chloro) was prepared to obtain information on which of the two barriers is due to the rotation of the aryl and which one is related to the ring inversion. In compound 16, the aryl rotational barrier is much higher, and the ambient temperature spectrum shows diastereotopic signals for CH 2 ( Figure S12 in the Supporting Information). On lowering the temperature, each diastereotopic signal broadens again and splits into two sets of diastereotopic signals below −70°C, with a diastereomeric ratio of 61:39 ( Figure S19 in the Supporting Information). This dynamic process must be assigned to the ring inversion, and the different diastereotopic ratio is a consequence of the higher steric hindrance exerted by chlorine with respect to fluorine. The experimental energy barrier was 9.5 ± 0.2 kcal mol −1 , in agreement to the smallest barrier observed for compound 15. The higher barrier observed for 15

ACS Omega
Article (11.1 kcal mol −1 ) has therefore to be assigned to the ofluorophenyl rotation.
Carbonyl Compounds. As a last structural change, we prepared some carbonyl analogues (compounds 18−22) to investigate the effects of the exocyclic heteroatom on the rotational barrier. From a theoretical point of view, the steric size of the carbonyl should be very similar to that of the Estereoisomer of the imine. In the case of the five-membered rings (compounds 18−20), the aryl rotational barriers are indeed very close to the corresponding imino compounds (R = Me, Cl, and NO 2 ), with differences lying just above the experimental error (±0.2 kcal/mol). On the contrary, in the case of the six-and seven-membered rings (compounds 21 and 22, respectively), the rotational energy difference is about 2 kcal mol −1 higher than the imino compounds.
DFT calculations suggest that the geometry of the threshold TS ( Figure S24 middle in the Supporting Information) in compound 21 is higher in energy and different in geometry with respect to the TS of compound 12. In compound 21, the o-chlorine atom is located over CH 2 in position 6 of the ring, and the two rings are almost coplanar. In compound 12, the aryl ring is in a pseudo-axial position and the chlorine is on the side of the CNH moiety. The difference is due to the lower double-bond character of the amidine moiety with respect to the amide. This difference is confirmed by the observation of a lower C−N rotational barrier in N,N-dimethylacetamidine with respect to N,N-dimethylacetamide. 23 The same effect can be invoked to explain the larger barrier of 22 with respect to 16 ( Figure S24 in the Supporting Information). On the contrary, the smaller five-membered ring (e.g., compound 5 vs 19) is more rigid and does not allow the aryl ring to bend out of the ring plane, thus the steric effect of the imino is similar to that of the carbonyl. DFT calculations confirmed that the geometries of both TS have the chlorine close to CH 2 in position 5 ( Figure S24 in the Supporting Information). As observed for compound 16, the low-temperature NMR spectra of compound 22 showed the presence of two diastereomeric conformations because of the ring inversion, with an energy barrier at the coalescence point of 9.7 ± 0.2 kcal mol −1 , very similar to that observed for compound 16.

■ CONCLUSIONS
We investigated the conformational dynamics of 1-aryl-2iminoazacycloalkanes on increasing the ring size from five to seven members. The N-aryl rotational barrier was found to be driven by the steric size of the ortho substituent, and the barrier increases by about 5 kcal mol −1 when changing the ring size from five to six and seven members. The E/Z interconversion barrier of the imino moiety was measured as 6.9 kcal mol −1 in compound 4. Two conformational diastereoisomers, generated by ring inversion, were observed in the case of the sevenmembered rings, with an energy barrier of 9.5 kcal mol −1 . A comparison with some carbonyl analogues showed very similar

ACS Omega
Article N-aryl rotational barriers for the five-membered rings and slightly higher barriers for the larger rings.

■ EXPERIMENTAL SECTION
Spectroscopic Data. NMR spectra were recorded using a spectrometer operating at a field of 14.4 T (600 MHz for 1 H, 150.8 for 13 C). Chemical shifts are reported in ppm (δ) relative to tetramethylsilane as an internal standard. D 2 O was employed to confirm exchangeable protons (ex). Splitting multiplicities are reported as singlet (s), broad signal (bs), doublet (d), double doublet (dd), triplet (t), quartet (q), heptet (h), and multiplet (m). The assignment of the 13 C signals was obtained by means of DEPT, gs-HSQC, and gs-HMBC spectra. The 150.8 MHz 13 C spectra were acquired under proton decoupling conditions with a 36 000 Hz spectral width, 5.5 μs (60°tip angle) pulse width, 1 s acquisition time, and 9 s delay time. The long relaxation time was needed to observe some quaternary carbons of the heterocycle. A line broadening function of 1−2 Hz was applied before the Fourier transformation. 1D-EXSY spectra were obtained at 600 MHz using the DPFGSE sequence 24 and a 50 Hz wide selective pulse with a R−SNOB shape. 25 High-resolution mass spectrometry (HRMS) was performed with an ESI-QTOF spectrometer. Reagents, solvents, and starting materials were purchased from standard sources and purified according to literature procedures. Melting points were determined with a Buchi capillary apparatus and are uncorrected. Variabletemperature NMR spectra were recorded as previously reported. 26 The rate constants were derived from line shape simulations (QCPE DNMR6 program), 27 and the free energies of activation (ΔG ⧧ ) were obtained by means of the Eyring equation. Within the experimental uncertainty, the latter values were found essentially invariant in the examined temperature range, thus implying an almost negligible activation entropy ΔS ⧧ . 15 Calculations. The ground-state and transition-state geometries were obtained using the Gaussian 09 rev D.01 series of programs, 28 with standard optimization parameters. The calculations employed the M06-2X hybrid-DFT functional 29 and the 6-311+G(d,p) basis set. Vibrational analysis was performed to validate the ground states (zero imaginary frequencies) and the transition states (one imaginary frequency). Visual inspection of the corresponding normal mode 30 confirmed the identification of the correct transition states.
General Procedure for the Synthesis of 1-N-Aryl-2iminopyrrolidines 1−9. 3 A mixture of the corresponding 4arylaminobutyronitriles (0.5 mmol) and a chloroform solution of PPE (4 mL) was reacted in a microwave reactor (Monowave 300, Anton Paar) at 100°C for 5 min for compounds 1 and 4−9 or 30 min for compounds 2 and 3. After reaching room temperature, the resulting solution was extracted with water (5 × 8 mL). The aqueous phases were pooled, filtered, and made alkaline in an ice bath, and the mixture was extracted with dichloromethane (3 × 40 mL). The organic layer was washed with water (5 mL), dried over sodium sulfate, and filtered. The solvent was removed in vacuo. The crude products were purified by column chromatography [Silica gel 60, dichloromethane (DCM)/methanol 30:1, DCM/isopropylamine 30:1].

ACS Omega
Article