A practical route to β2,3-amino acids with alkyl side chains

Enantiopure N(Boc)-β3-amino nitriles, valuable synthetic intermediates in the multistep homologation of α-amino acids, were alkylated using n-BuLi as base. Alkylations afforded easily separable, almost equimolecular mixtures of diastereomeric N(Boc)-protected syn and anti β2,3-amino nitriles. Suitable manipulations of both cyano and amino groups eventually led to enantiopure N- and/or C-protected β2,3-amino acids. For example, methanolysis using conc. HCl gas in MeOH, provides C-protected β2,3 amino acids in excellent yields. This methodology is applied to the synthesis of a series N(Boc)-β2,3-dialkyl amino nitriles derived from l-phenylalanine, d-phenylalanine, l-valine and one C-protected β2,3 amino acid. We demonstrate an efficient procedure for the preparation of anti and syn β2,3-amino acids with alkyl side chains, from α-amino acids in reasonable yields.


Background
Beta amino acids have shown great potential for a wide range of applications in many fields of organic chemistry in recent years (Juaristi and Soloshonok 2005). The growing interest in this class of compounds, which are widely used in medicinal chemistry forming new secondary structures and as valuable synthetic building blocks, can be better appreciated by database searching (e.g. Sci-Finder). Almost all β 3 -amino acids with proteinogenic side chains are now commercially available, but are quite expensive. Several synthetic procedures have been reported for the preparation of β-amino acids, and the field has been extensively reviewed (Cole 1994;Liu and Sibi 2002;Weiner et al. 2010). β 2,3 -Amino acids have two substituent at the α (C2) and β (C3) positions ( Fig. 1). They are relatively rare in nature, although occur as substructures in several bioactive compounds and important metabolites (Juaristi and Soloshonok 2005). These β-amino acids bearing two side chains are of particular interest for the synthesis of β-peptides (oligomers of β-amino acids) in view of their conformation-inducing ability and thereby the ability to afford new foldamers (Seebach et al. 1999;Cheng et al. 2001;Martinek and Fülöp 2012). β 2,3 -Amino acids may be either homochiral (anti-or like-β 2,3 -amino acids) or heterochiral (syn-or unlike-β 2,3amino acids) and it is noteworthy that, when included as building blocks in peptides, the former (1) afford predominantly helical structures, with all substituents in lateral positions, whereas their syn diastereomers (2) adopt an extended conformation, with formation of pleated sheets (Seebach et al. 1999;Balamurugan and Muraleedharan 2015).
A number of synthetic procedure have been proposed for the preparation of β 2,3 -amino acids and many of them have been reported recently in an excellent review (Kiss et al. 2015), but very few have been originate from amino acid precursors (Burgess et al. 1993) which have the obvious advantage of transferring pre-existing structural information, such as the nature of side chain and the chirality, into the final products. Among these, the Arndt-Eistert homologation is the most commonly used procedure (Podlech and Seebach 1995). The process involves the conversion of N-protected α-amino acid mixed anhydrides into the corresponding α-diazoketones using diazomethane, followed by the Wolff rearrangement. It has seen a resurgence in popularity in recent years due to the work carried out at ETH in Zurich, which has also identified new secondary structures by inserting several β 3 residues in homologous peptide sequences (Seebach et al. 2004). However, this procedure is aimed at obtaining β 3 -amino acids and only involves intermediate α-diazoketones.
These intermediates have also been used in the preparation of α-methyl β 3 -residues via KHMDS/HMPA methylation, in modest yields (Yang et al. 2000). Antiβ 2,3 -amino acids can be prepared by the alkylation of β 3amino esters (Estermann and Seebach 1988;Cardillo et al. 1996;Capone et al. 2007). An alternative approach to the asymmetric synthesis of syn and anti β 2,3 -amino acids from non-amino acid precursors, exploits the conjugate addition of a chiral N-benzyl-N-(α-methylbenzyl) lithium amide to a specially prepared α,β-unsaturated ester, already bearing the side chains needed in the final product (Davies et al. 1994;Langenhan and Gellman 2003]. A remarkable, unified approach to both syn and anti β 2,3 -amino acids in enantiomerically pure form with natural or unnatural side chain, was reported (Yu et al. 2010). The strategy is based on asymmetric cycloaddition of a nitrone derived from d-gulose to Z or E acrylates, followed by a facile N-O cleaving fragmentation reaction.
The availability of convenient synthetic routes to monomers predisposed to form helical structures has accelerated the pace of discoveries involving β-peptide helices. In contrast, a general, scalable synthesis of syn-β 2,3 -amino acids from α-amino acids has not been reported so far. Our interests towards non-natural amino acids synthesis began some time ago, when we proposed a procedure for converting α-amino acids into their β 3 -homologues (Caputo et al. 1995). The methodology reported generated valuable synthetic intermediates, such as β-amino alcohols 3, β-amino iodides 4 and β-amino nitriles 5 (Scheme 1). This homologation procedure has since been enhanced, and the β 3 -amino acids, as well as all homologation intermediates, can be also prepared labeled with isotopic 2 H, using NaBD 4 in D 2 O in the reduction step [Caputo and Longobardo 2007].
The homologation intermediate N-protected β-amino iodides 4 are extremely useful starting materials for the preparation of new classes of unnatural amino acids (Bolognese et al. 2006;Sureshbabu et al. 2011;Longobardo et al. 2013]. Furthermore, N-protected β 3 -amino nitriles 5, with different amine Pgs, are substrates in biotransformation reactions to give the corresponding amides and/or amino acids catalyzed by nitrilases and nitrile hydratases (Liljeblad and Kanerva 2006; Veitía et al. 2009).
We therefore considered it timely to report a practical synthetic application of intermediates 5 in a novel approach to the simultaneous synthesis of syn and antidialkyl β 2,3 -amino acids (including deuterium labeled compounds) from α-amino acids. Scheme 1 Multistep homologation of α-amino acids to N-and/or C-protected β 3 -amino acids

Methods
Solvents, inorganic salts and organic reagents were purchased from commercial resources and used without further purification unless otherwise noted. All of the compounds for which analytical and spectroscopic data are quoted were homogeneous by TLC. TLC analyses were performed using silica gel plates (E. Merck silica gel 60 F-254) and components were visualized by the following methods: ultraviolet light absorbance, iodine adsorbed on silica gel, and ninhydrin spray. Melting points were measured with a Kofler apparatus and are uncorrected. Column chromatography was carried out on silica gel (E. Merck, 70-230 mesh). THF was dried over Na in presence of benzophenone under an Ar atmosphere. All the compounds were characterized by 1 H and 13 C NMR spectroscopy. NMR spectra were recorded using Varian Inova 500 and Bruker DRX-400 spectrometers: chemical shifts are in ppm and J coupling constants in Hz. High-resolution ES mass spectra were obtained with a Micromass Q-TOF UltimaTM API. Optical rotations were measured with a Jasco 1010 polarimeter (k = 589 nm). One suitable crystal was mounted at room temperature on a Bruker-Nonius Kappa-CCD diffractometer.

Results and discussion
LDA and metallated-HMDS which are commonly used to produce α-carbanions from N(Boc)-β 3 -amino esters, turned out to be completely ineffective toward N(Boc)protected nitriles 5. We found that chiral N(Boc)-β 3amino nitriles are smoothly alkylated at the α-position. The reaction is not at all diastereoselective, hence leads to essentially equimolecular mixtures of syn and anti disubstituted β-amino nitriles. The diastereomers can be easily separated and individually converted into the corresponding N-and/or C-protected β 2,3 -amino acids. The alkylations were carried out at −78 °C in anhydrous THF, using 2.2 equivalents of n-BuLi per mole of the starting N(Boc)-β 3 -amino nitrile 5. Under such conditions the resulting lithium dianion is rapidly formed and precipitates as a white solid from THF. These dianions are quite stable below −65 °C, and rapid quenching with alkyl halides afforded almost equimolecular mixtures of both diastereomeric anti and syn N(Boc)-α,β-dialkyl β-amino nitriles (e.g. 6 and 7 in Scheme 2). One example of moderate selectivity (from 60:40 to 95:5 anti:syn) observed in the alkylation of β 3 -amino nitriles has been reported, but the starting nitriles were fully protected with two bulky benzyl groups at the amino nitrogen atom in that case, therefore cannot be directly compared to our work (Reetz et al. 1994).
Various N(Boc)-β 2,3 -amino nitriles with simple alkyl side chains were prepared under the same conditions. The overall yields of the alkylated products were higher than 80 % in all cases. Details of the synthesis of a series N(Boc)-β 2,3 -dialkyl amino nitriles derived from l-Phe, d-Phe and l-Val are shown in Fig. 2 and Table 1.
Finally, a specially prepared C2-dideuterated N(Boc)β 3 -alkyl amino nitrile 5, derived from d-Phe, was alkylated with 4-iodochlorobutane yielding diastereomers 16 and 17 (containing one deuterium atom) in 84 % yield. These derivatives contain a side chain useful for further synthetic elaboration, for example in the preparation of β 2,3 -amino acids with a lysine side chain (Langenhan and Gellman 2003).
In all the alkylations studied, single enantiomers were obtained in very good yields after a simple chromatographic separation, as shown in Table 1. The calculation of the dipole moment for compounds 8-17, using the module Chem3D Ultra of ChemDraw software, suggested that for each pair of diastereomers, the anti-stereoisomer was the least polar, and we found that they were eluted first from the silica gel chromatographic columns (EtOAc-Hex 1:9). Two of the four N(Boc)-β 2,3 -dibenzylamino nitriles, namely 10 and 11 (prepared from d-Phe) could be readily converted into already known amino acids (Seki et al. 2000), which enabled the assignment of the absolute stereochemistry to the chiral centers of compounds 8-11. Nitrile 10 was converted by methanolysis (14 M HCl in MeOH, 0 °C to room temperature, 12 h) into the corresponding methyl ester hydrochloride 18 in 81 % yield (Scheme 3). The need for a high HCl concentration is necessary for both the transformation of the cyano group and the removal of the amino Pg. The remaining diastereomeric pair, N(Boc)-β 2,3 -amino nitriles 8 and 9 (prepared Scheme 2 Anti (6) and syn (7) N(Boc)-α,β-dialkyl β-amino nitriles from N(Boc)-β 3 -amino nitriles from l-Phe) were enantiomers of 10 and 11, respectively, and thus could be assigned the opposite configurations at the chiral center. All the stereochemical assignments were eventually confirmed by X-ray analysis of N(Boc)β 2,3 -dibenzylamino nitrile 8 (Fig. 3). The conversions above, along with our existing experience, emphasize the reliability of N(Boc)-β 2,3 -amino nitriles as intermediates to afford N-and/or C-protected β 2,3 -amino acids efficiently, preserving the integrity of the chiral centers.

Conclusions
The procedure we report here is suitable for the simultaneous preparation of both syn and anti β 2,3 -amino acids from α-amino acids. The straightforward preparation of the starting chiral N(Boc)-protected β 3 -amino nitriles (including deuterium-labeled derivatives) from commercial N(Boc)-protected proteinogenic α-amino acids is a noteworthy feature of the whole synthetic scheme. The easy preparation of β 2,3 -amino acids in enantiomeric pure form, with natural or unnatural side chains in position 2 and with the option to introduce other substituents than alkyl groups in this position, starting from low-cost materials, represents a valued synthetic methodology which may find a wide number of applications in the chemistry of amino acids and peptides. In light of the present example, the already mentioned homologation of α-amino acids via β-amino iodides appears to be an effective alternative to the classical Arndt-Eistert homologation procedure, due to the possibility of isolating valuable intermediates, which can be exploited in novel amino acid syntheses.
Authors' contributions LL design and performed the series of experiments and wrote the manuscript. MDG performed the NMR analysis, including 2D experiments. IdP performed the series of experiments, including purification and characterization of novel molecules. All authors read and approved the final manuscript.   N(Boc)-β 2,3 -dialkyl amino nitriles 8-17 (Fig. 2