Introduction

Since their introduction in 1982 [1], chiral Ni(II) complexes-based synthons of α-amino acids gained noticeable popularity as a very simple, robust and cheap tool for the preparation of enantiomerically pure non-coded amino acids and diagnostics of brain tumours and malfunction of dopaminergic neurons by positron emission tomography (PET) [2–12].

An important improvement occurred when the first generation synthons based on substituted 2-aminobenzaldehyde (as biomimetic analogues of pyridoxal-5-phosphate) [13] were overperformed in asymmetric alkylation reactions by derivatives of 2-aminoacetophenone followed by even more efficient derivatives of 2-aminobenzophenone [14]. The second generation synthons delivered a higher selectivity of C19 monomethylation and much higher ratio of diastereomers of alkylated products under thermodynamically controlled conditions when C19 is epimerised in a basic reaction mixture or in a separate epimerisation step carried out in NaOMe/MeOH [15, 16] (Y. N. Belokon, personal communication). For example, equilibrium ratio of diastereomers of the alanine complex increased from 15% d.e. for the first generation to 82% d.e. for the second generation [15]. New synthons enabled the preparation of enantiomerically pure α-methyl amino acids by sequential monomethylation of the glycine complex followed by the introduction of a bulkier second C19 substituent (Scheme 1) [17].

Scheme 1
scheme 1

Selective monomethylation of the second-generation synthon. For the 2-aminobenzophenone-derived complex (GK) a bulkier electrofile can be used in the second alkylation step for the preparation of enantiomerically pure α-methyl amino acids

Full size image

This high selectivity was explained by the destabilisation of a carbanion generated from the monomethylated product due to intramolecular steric repulsion, when compared to the carbanion generated from the starting non-substituted complex (Scheme 2) [18]. Such a distortion leads to a non-planar destabilised carbanion (Scheme 2). In this article we present a complementary explanation based on the disclosure of a non-trivial bond strain in the least sterically hindered Ni(II) complex of the Schiff base of (S)-N-(2-benzoylphenyl)-1-benzylpyrrolidine-2-carboxamide and glycine (GK) when compared with its α-carbon (C19) monoalkylated (HomoSK) and dimethylated (Me2GK) derivatives (see Scheme 3 and Fig. 1).

Scheme 2
scheme 2

Steric repulsion leading to the distortion of the sterically hindered monomethylated carbanion

Full size image
Scheme 3
scheme 3

Non-, mono- and di-substituted complexes derived from 2-aminobenzophenone

Full size image
Fig. 1
figure 1

Numbering scheme for GK (From ref. [26])

Full size image

Results and discussion

Recently achieved breakthrough in the reduction of the environmental impact of a high-scale production of α-amino acids via Ni(II) complexes [19, 20] and a substantial improvement of the diastereoselectivity of the alkylation reaction [16, 21–25] made the complexes a tool of choice for the preparation of carbon-11 or fluorine-18 enantiomerically pure α-amino acids for PET. In order to achieve simple work-up of the reaction mixture after the alkylation reaction, and easy, cheap and environment-friendly purification of the desired diastereomer of the alkylated complex by the crystallization or simple purification of the reaction mixture in the remote-controlled preparation of labelled α-amino acids for PET, it is important to understand and use the factors determining the high ratio of GK alkylation rates k 1/k 2 (Scheme 1). During the course of our development of the approaches to carbon-11 labelled α-methyl amino acids and to a strategy of fluorine-18 labelling of peptides, we prepared a number of single crystals of nickel (II) complexes and performed ab initio MP2 quantum-chemical modelling of these structures. In this work we describe the quantum theory of atoms-in-molecules (QTAIM) analysis of the atoms and bonds of interest based on MP2 single-point calculations of the non-, mono- and di-substituted complexes, X-ray structures of which we have published previously (Scheme 3) [26–28]. QTAIM analysis based on MP2 calculations using double-ζ basis sets is considered to be an adequate approach for modelling the large systems where π–π and σ–π interactions play an important role [29]. Initially we expected that the introduction of substituents into the α-position (C-19) of the complexes should subsequently increase the steric hindrance in these complexes. In reality, the atomic volume of C-19 in the non-substituted GK is the highest one among the three complexes (Table 1). As a rule, atomic volumes increase with higher electron density at the corresponding atoms (more negative or less positive atomic charges) in analogous structures [30, 31] otherwise an increasing reactivity at these atoms is indicated. In GK, O2 atom exhibits a lower atomic volume and a more negative charge whereas C19 atom exhibits a higher volume and a more positive charge in comparison with HomoSK and Me2GK complexes (Table 1). This unusual behaviour is accompanied by the increased BCP ellipticities of Ni–O2, O3–C20 and C19–N3 bonds of GK in comparison with the remaining complexes (Table 2) which may be explained by the mechanical strain. The increased mechanical strain at O2 and C19 atoms indicates the higher reactivity at these sites of GK [30, 31] in comparison with HomoSK and Me2GK complexes. Significantly stronger O2–C20 bond and weaker C19–N3 bond in GK in comparison with the remaining compounds (Table 2) confirm the observed differences in their electronic structure as well. The atomic volumes of the iminic N3 bonded to C19 follow the similar pattern—they are the biggest in the least substituted GK (Table 1). We hypothesised that such an unexpected behaviour originates in the more compact core of GK where the shorter distances in the internal coordination sphere result in the higher strain of these bonds. Monoalkylated (e.g. monomethylated) complexes derived from 2-aminobenzophenone, where the distances between the nickel atom and the heteroatoms are longer, keep a lower strain and thus they are more stable. In the complexes derived from 2-aminobenzaldehyde, the differences are probably not so profound due to a lower steric hindrance, and so there is a little or no energetic benefit of the monosubstitution in such complexes.

Table 1 Atomic charges and volumes of selected atoms in the systems under study (see Scheme 3 and Fig. 1)
Full size table
Table 2 BCP electron densities, ρBCP, and ellipticities, εBCP, of selected bonds in the systems under study (see Scheme 3 and Fig. 1)
Full size table

Experimental

The electronic structure of the complexes under study is evaluated at MP2 level of theory using triple-ζ valence (TZV) basis for Ni [32] and double-ζ valence (DZV) basis sets for the remaining atoms [33]. Firefly software package is used for single-point calculations of experimental geometries [34].

The electronic structure of the species under study is investigated using QTAIM topological analysis of electron density [33, 34]. The results are evaluated in terms of atomic volumes V and atomic charges q obtained using the electron density integrated over atomic basins (up to 0.001 e/Bohr3 level). Bond characteristics are evaluated in terms of electron density ρ (proportional to bond strength) and bond ellipticity ε

$$ \varepsilon= \lambda_{1}/{\lambda_{2}}-1$$
(1)

at bond critical points (BCP) where λ1 < λ2 < 0 < λ3 are the eigenvalues of the Hessian of the BCP electron density. AIM2000 [35] software package is utilized for QTAIM analysis (http://www.aim2000.de/).

Conclusions

We have disclosed an important factor giving rise to the high selectivity of the monoalkylation of nickel (II) complex of the Schiff base of BPB and glycine—the higher strain of the bonds of this non-substituted complex when compared to its mono- and di-substituted analogues. Further MP2/QTAIM analyses of similar less sterically hindered complexes are desirable to verify our findings.