Tautomerism troubles: proton transfer modifies the stereochemical assignments in diastereoisomeric structures of spirocyclic 5-methyl-2H-imidazol-4-amine dimers

Single-crystal structure analysis was required to correctly identify the molecular structure and stereochemistry assignment of unexpected impurities in the preparation of a novel pharmaceutical spirocyclic imidazole-amine compound.


Chemical context
During the racemization of an enantiopure spirocyclic 5-methyl-2H-imidazol-4-amine, two impurities were observed by reverse phase-HPLC, which were subsequently rationalized as a combination of the homochiral compounds (S),(S)-(R),(R)-, and heterochiral compounds (S),(R)-and (R),(S)-(see Fig. 1 for the proposed 2D structures). Solution-state NMR and mass spectrometry analysis revealed that these impurities were dimers of the 5-methyl-2H-imidazol-4-amine compound e (see Fig. 2); no diagnostic signals were observed in the solution-state NMR and therefore single-crystal structure determination was required to allow assignment of the absolute configuration of the impurities observed.
The chemical shifts from the solution-state NMR are given in Section 6 below. As related enantiomers are indistinguishable by solution-state NMR, single crystal X-ray diffraction analysis was sought to enable an unambiguous assignment, revealing structures 1 and 2. This analysis not only enabled the identification of the correct absolute structure, but also revealed that there was, in fact, a subtle variation to the proposed structures. Crystal structures were obtained for both the impurities observed, which revealed that the homochiral and heterochiral structures differed from those proposed (a, b, c and d) due to hydrogen migration from the bridging nitrogen centre to the closest imidazole group.

Structural commentary
Both structures solved and refined satisfactorily in the centrosymmetric space group P2 1 /n. Therefore, both possible diastereoisomers, RR/SS (in structure 1) and RS/SR (in structure 2), are present in equal amounts in their respective crystal. The structures along with their atomic numbering schemes are illustrated in Fig. 3. Fig. 3 shows that in both cases an unexpected proton transfer from the bridging amine centre (N1 in both structures) to the spiro-imidazole nitrogen (N2 in both structures) had occurred. Examination of residual electron density maps (see Refinement section and Fig. 6) of both structures, supported by interpreting the bond lengths around these nitrogen centres, confirmed the location of the hydrogen atom and therefore the fact that this migration has occurred. It was, however, necessary to restrain the N2-H2 bond in structure 1, otherwise it refined to a value slightly shorter than expected. This transfer results in a perturbation of the bonding pattern within the imidazole rings for both structures. The bond conjugation between these rings is extended via the bridging nitrogen (N1), which makes the formal nature of the double and single bonds in these ring systems less clear, as depicted in Fig. 4 and Table 1.
This perturbation of bonding means particularly close attention must be paid to the formal chirality assignment of the stereocentres C4 and C22 in both structures 1 and 2, as it is dependent on analysis of the surrounding imidazole bond lengths. The definitive Cahn-Ingold-Prelog assignment (Cahn et al., 1966) of these stereocentres required a manual approach as the algorithms in both PLATON (Spek, 2020) and Mercury (Macrae et al., 2020)  The molecular structures with atomic numbering schemes (non-carbon and hydrogen atoms only for clarity) for 1 and 2 respectively. The chiral centres are marked with an asterisk and for clarity only the enantiomer solved in the asymmetric unit of each structure is shown.

Figure 4
Labelled bonds in the N-bridged bis-imidazole core, with associated bond lengths for 1 and 2 denoted in Table 1. Table 1 Bond lengths (Å ) in the N-bridged bis-imidazole core for structures 1 and 2, with bonds denoted as in Fig. 4  the ambiguity of bond order altering the priority of the bonds connected to the stereocentres. In both structures, the distances between atoms C1 and N2 (1) and C19 and N4 (2) have more double-bond character than single. This results in a formal designation of RR (and SS) and RS (and SR) in the refined structures of 1 and 2, respectively. The hydrogen-atom location also results in the formation of a strong intramolecular hydrogen bond between both imidazole rings, mediated by a N2-H2Á Á ÁN4 interaction in each structure (Tables 2 and 3).

Supramolecular features
The packing arrangement for both structures 1 and 2 are shown in Fig. 5. In both compounds there are no hydrogenbonding interactions present within the structure, other than that of the N2-H2Á Á ÁN4 intramolecular hydrogen bond. This intramolecular interaction is an additional factor influencing the delocalization of bonding in these ring systems.
The packing is likely to be dominated by dispersive interactions and the differences between the two motifs will be small. These molecules have a ridge-tile shape and the structure of 1 involves insertion of a sidewall of one molecule into the cleft of another; this motif contains some small voids with a volume of approximately 30 Å 2 calculated using Mercury (Macrae et al., 2020). However, molecules in the structure of 2 assemble in a side-by-side manner into a strand, which allows complementary head-to-tail stacking of strands and is more packing efficient (using the same settings in Mercury no voids are calculated).

Database survey
These molecular structures are relatively unique in solid-state chemistry. A search of the CSD (CSD version 5.42, updates of Feb 2021; Groom et al., 2016) yielded no results for structures with a similarity to the overall molecule or to the motif of the methyl-imidazole-amine bridged dimer. There were also no structures found for a spiro group with similar substituents.
2H-Imidazoles are well known in organic chemistry and a name search for these in the CSD revealed 677 structures. A combination of this search with that of the substructure of the imidazole core of this system, where the bonds are considered to be 'double' produces 50 hits, while there are no results if these bonds are defined as 'delocalized'. Analysis of the hit lists does not reveal any structures similar to those reported herein and therefore does not provide any insight as to how the bonding should be assigned.

Synthesis and crystallization
Solid samples of 1 and 2 were isolated from a reaction to form a spirocyclic 5-methyl-2H-imidazol-4-amine, during which they were formed as an impurity product and were subsequently isolated. Single crystals of compounds 1 and 2 were grown by slow evaporation at room temperature from individual solutions of ethyl acetate (200 mL g À1 ). Each mixture was allowed to evaporate to dryness over the period of a week. Both compounds formed colourless block-shaped crystals.

Characterization by spectroscopic techniques
The following NMR and mass spectrometry data were collected.

Refinement
The crystal data, data collection and refinement details for structures 1 and 2 are summarized in Table 4   The disorder modelling in structure 2, with displacement ellipsoids drawn at the 50% probability level and the minor component highlighted in orange (10.8%).

Figure 6
The 3D residual electron difference maps from the refinements of 1 and 2. The green wireframe is drawn at a threshold of >0.4 electrons /Å À3 . obtained by following a previously published approach (Coles & Gale, 2012). Further details of these experiments are given at the end of this section. The structure refinements of both 1 and 2 demonstrated that the hydrogen atom anticipated to be bound to the bridging nitrogen (labelled N1 in both structures) was in fact bound to one of the adjacent imidazole nitrogen atoms (labelled N2 in both structures). This was confirmed by inspection of residual electron difference maps. Fig. 6 depicts 3D representations of the residual electron difference map around the bis-imidazole cores of 1 and 2, the green wireframes are drawn at a threshold of >0.4 electrons/ Å À3 and highlight the location of the hydrogen atoms. The structures of 1 and 2 solved in the space group P2 1 /n (# 14) using dual methods in the SHELXT (Sheldrick, 2015a) structure-solution program and refined by full-matrix leastsquares minimization on F 2 using SHELXL2018/3 (Sheldrick, 2015b). All non-hydrogen atoms were refined anisotropically. The position of the N-H atom H2 was located from the difference map and refined with its thermal parameter linked to that of its parent atom, N2. The positions of the remaining C-H atoms were calculated geometrically and refined using a riding model.
The disordered atoms of 2 (Br1A/Br1B, O1A/O1B and C5A/C5B > C18A/C18B), have been modelled over two positions using geometric parameter restraints. In addition, the geometry of the minor benzene ring (C5B > C10B) was constrained to be a regular hexagon with bond lengths of 1.39 Å . All minor atomic positions were modelled isotropically with the thermal parameters of atoms Br1A and Br1B restrained and those of atoms O1B and C5B > C18B, constrained to be the same. Applying the above to the refinement conserved realistic chemical geometries and lowered the R 1 value from 2.74% to 2.20%. Fig. 7 depicts the disorder modelling in structure 2, with displacement ellipsoids drawn at the 50% probability level and the minor component highlighted in orange (10.8%). ; program(s) used to refine structure: SHELXL2018/3 (Sheldrick, 2015b); molecular graphics: OLEX2 (Dolomanov et al., 2009); software used to prepare material for publication: OLEX2 (Dolomanov et al., 2009). where P = (F o 2 + 2F c 2 )/3 (Δ/σ) max = 0.001 Δρ max = 0.33 e Å −3 Δρ min = −0.39 e Å −3 Extinction correction: SHELXL-2018/3 (Sheldrick 2015b), Fc * =kFc[1+0.001xFc 2 λ 3 /sin(2θ)] -1/4 Extinction coefficient: 0.00013 (2) Special details Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes. Refinement. This diastereoisomer has crystallised in the centrosymmetric space group P2 1 /n; meaning that both RS and SR forms of the API must be present in equal amounts within the crystal. The atoms of both methoxy groups (O1, C18 and O2, C36), have been modelled as single sites with large thermal ellipsoids. This was found to be the most appropriate model; however, these large ellipsoids result in two checkCIF C-alerts. The N2-H2 bond length has been restrained, otherwise it refined to an unrealistic value.