Introduction

Nonproteinogenic amino acids such as β- and γ-amino acids play a significant role in many areas of chemistry. They occur as part of natural products many of which are of important therapeutical use. This has led to an increased demand in providing these compounds in enantiopure manner, resulting in a number of novel synthetic methods published recently [1–6].

β-Amino acids, for example, are known to possess antibiotic [7, 8], cytotoxic [9], antifungal [10], and other useful pharmacological properties [11, 12]. They serve as chiral building blocks for asymmetric catalysts [13] and for nonnatural modified peptides and also frequently occur incorporated in peptides, depsipeptides, and peptidomimetics [14].

The importance of (carbocyclic) γ-amino acids is mainly related to their potential to mimic the structure of γ-amino butyric acid (GABA) on specific receptor sites as well as to their occurrence in biologically active natural products, e.g., pepstatin [15]. GABA is the major inhibitory neurotransmitter in the mammalian central nervous system (CNS) [16]. In recent times, the application of conformationally restricted GABA mimics—typically cyclic compounds containing a rigid carbon backbone [17]—has contributed to a better understanding in GABA neuroreceptor research [18]. For example, 3-aminocyclopentanecarboxylic acid isomers were recognized to be particularly efficient stereomeric probes for GABA binding-site topography. 3-Aminocyclohexanecarboxylic acids, however, were found to act selectively as GABA uptake inhibitors [19]. Besides, analogues therefrom were investigated as therapeutic agents for a range of CNS system disorders [20].

Recently, highly enantioenriched β- and γ-amino acids and amides, respectively, have been made accessible in an enzymatic kinetic resolution step by biohydrolysis of the corresponding racemic β- and γ-amino nitriles, applying novel biocatalysts, such as nitrile hydratase [21, 22] or nitrilase [23, 24] (Scheme 1).

Scheme 1
scheme 1

Pathways of biotransformation of amino nitriles (n = 1,2)

Full size image

The determination of the enantiomeric purity of the products of an enzymatic/microbial nitrile biotransformations—amides and acids—is a challenging task. First, the strongly basic and polar amino group gives rise to serious troubles during all kinds of chromatographic analyses. Second, the isolation of this type of products from an aqueous reaction (so far the only medium for this type of biotransformation) has to take into account the high water solubility of the products—acids and amides—thus rendering an extraction into the organic phase always accompanied with high losses. Therefore and not at last by our long experience with these compounds and their pitfalls, we felt prompted to carry out our work with tolylsulfonyl protected amino groups (N-Ts) in all substrates [25].

In this paper, we wish to report on the enantioseparation of a range of N-protected β- and γ-amino acids, amides, and nitriles, respectively. The enantioseparation of β-amino acids by using chiral gas chromatography [26], chiral liquid chromatography (LC) [27–33], as well as by chiral derivatization LC [34–37] has been addressed recently. However, separations of γ-amino acids were reported less frequently [38], and a separation of N-Ts-protected acids presented in this work has not yet been reported. The same is true for the enantioseparation of the present amino amides and amino nitriles.

Experimental

Chemicals

n-Heptane (VWR), isopropanol (i-PrOH; Merck), ethanol (EtOH; Merck), methanol (MeOH; VWR), and acetonitrile (ACN; Merck) were high-performance liquid chromatography (HPLC) grade. n-Butanol (Roth), i-butanol (Roth), trifluoroacetic acid (Aldrich), triethylamine (TEA; Fluka), acetic acid (AcOH; Sigma-Aldrich), perchloric acid (PCA; 85%; Roth) and H3PO4 (85%; Roth) were all reagent grade. All aqueous mobile phases were filtered through 0.45 μm cellulose nitrate filter (Sartorius) prior to use.

Apparatus

The HPLC chromatographic system used for water compatible chiral columns was a Hewlett Packard Series 1100 HPLC equipped with a G1315A diode array detector. For nonwater-compatible chiral columns, an Agilent Series 1100 HPLC equipped with a multiwavelength UV-detector was used. All analytes bearing an aromatic ring were analyzed at 254 nm, all other analytes at 205 nm.

Chiral stationary phases

The columns used were a Chirobiotic R (250 × 4.6 mm) 5 μm particle size (Astec); a Chiral AGP 100.4 (100 × 4 mm) 5 μm particle size (ChromTech); a Crownpak CR(+) (150 × 4 mm) 5 μm particle size (Daicel); a Chiralpak AD-H (250 × 4.6 mm) 5 μm particle size (Daicel); a Chiralcel OD-H column (250 × 4.6 mm) 5 μm particle size (Daicel); and Chiralcel OJ column (250 × 4.6 mm) 5 μm particle size (Daicel). The hold-up time t 0 was determined by injection of air.

From time to time, each HPLC-chiral stationary phase (CSP) was checked for unchanged and best performance using test analytes. Thus, α1 acid glycoprotein (AGP) was found to be the only CSP which had to be regenerated several times by flushing with EtOH over night.

Substances

The syntheses as well as the associated physical and spectroscopic data for all compounds are given in [21–24] and in the literature cited therein.

Separation conditions

Chiralcel OD-H: 1a: heptane/i-PrOH 50/50 (0.38 mL min−1; 15 °C); 1b: heptane/i-PrOH 50/50 (0.38 mL min−1; 15 °C); 3a: heptane/i-PrOH 50/50 (0.38 mL min−1; 15 °C); 3b: heptane/i-PrOH 50/50 (0.38 mL min−1; 15 °C); 4a: heptane/i-PrOH 85/15 (0.45 mL min−1; 15 °C); 11a: heptane/i-PrOH 50/50 (0.38 mL min−1; 20 °C).

Chiralpak AD-H: 1a: heptane/i-PrOH 50/50 (0.5 mL min−1; 15 °C); 1b: heptane/i-PrOH 70/30 (0.5 mL min−1; 15 °C); heptane/i-PrOH 50/50 (0.5 mL min−1; 15 °C); 2a: heptane/i-PrOH 70/30 (0.5 mL min−1; 15 °C); 2b: heptane/i-PrOH 70/30 (0.5 mL min−1; 15 °C); 2c: heptane/i-PrOH 50/50 (0.5 mL min−1; 15 °C); 3a: heptane/i-PrOH 50/50 (0.5 mL min−1; 15 °C); 3c: heptane/n-butanol 70/30 (0.5 mL min−1; 15 °C); 4a: heptane/i-PrOH 70/30 (0.8 mL min−1; 15 °C); 4b: heptane/EtOH 70/30 (0.8 mL min−1; 15 °C); 6a: heptane/i-PrOH 50/50 (0.5 mL min−1; 20 °C);7a: heptane/EtOH 50/50 (0.5 mL min−1; 20 °C); 8a: heptane/EtOH 50/50 (0.5 mL min−1; 20 °C); 9a: heptane/EtOH 50/50 (0.8 mL min−1; 20 °C); 10a: heptane/i-PrOH 50/50 (0.5 mL min−1; 20 °C).

Chirobiotic R: 1c: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 23 °C); 2c: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 23 °C); 5a: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 30 °C); 5c: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 30 °C); 6a: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 23 °C); 6c: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 30 °C);

9c: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 30 °C); 10c: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 30 °C); 11c: MeOH/TEA/AcOH 1000/4/1 (0.8 mL min−1; 30 °C).

Chiral AGP: 2a: phosphate buffer 10 mM (pH 7.02)/ACN 90/10 (0.8 mL min−1; 15 °C); 2c: phosphate buffer 10 mM (pH 7.02)/ACN 90/10 (0.8 mL min−1; 15 °C); 4b: phosphate buffer 10 mM (pH 4.05)/ACN 95/5 (0.8 mL min−1; 15 °C);4c: phosphate buffer 10 mM (pH 7.02)/ ACN 95/5 (0.8 mL min−1; 15 °C); 7c: phosphate buffer 10 mM (pH 7.02) (0.8 mL min−1; 15 °C); Crownpak CR (+): 1a-c-4a-c: PCA (pH 1; 0.5 mL min−1; 15 °C).

Results and discussion

Generally, most of the existing work on enantioseparation of β- and γ-amino acids [27–37] was carried out using either pure amino acid standards or free racemic acids prepared by chemical methodology, thus reducing the problem to the chiral separation of a pair of pure enantiomers without other interfering compounds. Different from that, the analytes in our investigation are isolates from a biotransformation reaction, therefore a mixture of several enantiomeric compounds.

The amino nitriles (1a–11a)—substrates for the enzymatic transformation reaction shown in Scheme 1—as well as the expected products—amino acids (1c–11c) and amino amides (1b–11b)—are depicted in Fig. 1.

Fig. 1
figure 1

Structures of β- and γ-amino acid derivatives; a: R = CN (all nitriles are racemic); b: R = CONH2; c: R = COOH; (only one enantiomer is depicted)

Full size image

Initially, our ambitious goal was to separate all possible products of the biotransformation and determine the enantiomeric excess within a single chromatographic run. This goal, however, turned out to be impracticable. Besides, we also expected at the least the respective structurally related analytes (bearing identical functionalities, ring size as well as identical relative 1,2- and 1,3-substituent configuration) to share some chromatographic features to a certain extent, e.g., to be resolvable by the same CSP. In the end, however, it turned out that predictable similarities in separation behavior of even structurally closely related compounds could only be noticed for the chiral crown ether-based CSP .

Chiral crown ether-based CSP

Chiral crown ethers (Crownpak CR(+)) are known to be selectors for compounds bearing primary amine functionalities near a chiral center, capable of complex formation with the crown ether oxygen atoms. Hence, the separation of amides (e.g., sulfonamides or benzamides) is not possible on these CSPs. Unprotected β-amino acids (Fig. 2) have been successfully resolved on Crownpak CR(+) [39] (Table 1). The complex stability can be modulated by variation of the pH of the mobile phase.

Fig. 2
figure 2

N-unprotected β- and γ-amino acid derivatives; a: R = CN (all nitriles are racemic); b: R = CONH2; c: R = COOH; (only one enantiomer is depicted)

Full size image
Table 1 Separation of free β-amino acids 12c-15c and γ-amino acids 16c-19c
Full size table

The ring size of the alicyclic analytes 12a–c–15a–c was found not to affect the separation of the enantiomers substantially; however, due to different solubility of nitrile, amide and acid, the content of organic modifier had to be adapted in order to achieve suitable retention times (HClO4aq/ACN 90:10; pH = 1). Generally, we achieved better resolution for trans-isomers (13a–c and 15a–c) than for the cis-isomers (12a–c and 14a–c), which could only be partially separated, in accordance to previously reported results [35]. Increasing resolution efficiency was noticed in the order from β-amino amides and acids, respectively, to amino nitriles. As far as the free γ-amino acid derivatives 16a–c–19a–c were concerned, none of them could be separated using Crownpak CR(+). Instead, we had to resort to a traditionally successful method, the indirect separation via chiral derivatization by o-phthalic dialdehyde/N-acetyl cysteine using a RP-18 column.

Polysaccharide-based CSPs

The enantioseparation of amino nitriles 1a through 11a—the remaining enantiomers of the kinetic resolution—is best accomplished by Chiralpak AD-H, an amylose tris-(3,5-dimethylphenylcarbamate) CSP (Tables 2, 3, and 7). Interestingly, Chiralcel OD-H, a cellulose tris-(3,5-dimethylphenylcarbamate) CSP, completely failed in separating all γ-amino nitriles 5a–8a (Table 3) as well as heterocyclic nitriles 9a–11a (Table 7). Generally, the separation of trans-carbocyclic nitriles 2a, 4a, 6a, and 8a was superior to the cis-isomers, and Chiralpak AD-H was also found to be superior to Chiralcel OD-H in terms of resolution power. The results from the separation of β-amino amides are more diverse. With the exception of carbocyclic six-ring amides b and b, which could be separated complementary by Chiralcel OD-H and Chiralpak AD-H, all amides 1b–4b were resolved by these CSPs (Table 4). This was rather unexpected since we anticipated the high polarity and low solubility of the amides in the eluents (hexane/i-PrOH) to be an obstacle. The separation results of the respective carboxylic acids 1c through 11c are somewhat more diverse. Chiralpak-AD-H resolved all β-amino acids 1c–4c, but not so did Chiralcel OD-H (Table 5). Likewise, neither of the acids 5c–11c was separable on Chiralpak-AD-H or Chiralcel OD-H (Table 6). For synthetic reasons, none of the separations of γ-amino acid 8c could be worked out, since all attempts in preparing the racemic reference acid failed. This was due to extensive epimerization of the trans-amino nitrile 8a at carbon-1 to the respective all-equatorial 1,3-cis-amino acid 7c when subjected to strong alkaline hydrolysis to yield trans-acid 8c [23].

Table 2 Separation of β-amino nitriles 1a–4a
Full size table
Table 3 Separation of γ-amino nitriles 5a–8a
Full size table
Table 4 Separation of β-amino amides 1b–4b
Full size table
Table 5 Separation of β-amino acids 1c–4c
Full size table
Table 6 Separation of γ-amino acids 5c–8c
Full size table

The obscure separation behavior of Chiralcel OJ only enabled the separation of 3c, 4c, and 4b (Table 4), independent of their cis/trans-relative stereochemistry.

As mentioned in “Experimental,” the portion of alcohol in the eluent composition is generally high (15–50% i-PrOH in n-heptane when operating with Chiralcel OD-H and up to 30% EtOH in n-heptane, in certain cases 100% EtOH, when operating with Chiralpak AD-H). As a consequence of this unusually high alcohol portion in the mobile phase, the flow rates had to be adjusted according to maximum pressure requirements of the columns. In addition, to lower eluent viscosity, the temperature was raised to 40 °C when 100% EtOH was used. Irrespective of this, 3c could be only sufficiently separated when using n-BuOH in the eluent composition with n-heptane; all other alcohols failed.

Macrocyclic glycopeptide-based CSP

Ristocetin A derived CSPs (Chirobiotic R) are capable to offer multiple interaction sites, such as electrostatic, hydrogen bonding, dipole–dipole, π–π complexation as well as hydrophobic effects to diverse analytes. Armstrong and coworkers [40] have separated a plethora of free and N-protected α-amino acid derivatives and small peptides on Chirobiotic R. These authors knowingly commented on the benefits for detection and quantification of protected amino acids in “real world samples,” such as from a biotransformation reaction. However, a separation of N-protected β- and γ-amino acid derivatives of the present kind has not been reported with macrocyclic glycopeptidic CSPs up to now.

We investigated the separation capability of this CSP using reversed-phase eluents (MeOH/H2O or ACN/H2O mixtures) and the reportedly more successful polar organic mode elution conditions, utilizing MeOH/TEAA (MeOH/TEA/AcOH 100/0.4/0.1; v/v/v). Although Chirobiotic R represents the CSP of choice for (free and N-derivatized) amino acids, to our surprise, only cyclopentane carbocyclic acids 1c, 2c, 5c, and 6c could be sufficiently resolved but none of the respective six-membered acids, irrespective of any cis/trans stereoisomery as well as position of the two ring substituents. All heterocyclic amino acids (9c–11c, Table 7) were not nearly baseline separated.

Table 7 Separation of heterocyclic amino nitriles 9a–11a and amino acids 9c–11c
Full size table

Moreover, the complete failure of Chirobiotic R to separate a single amide among the amides 1b–11b came notably unexpected. The failure of resolution of amino nitriles 1a–11a was likely due to the Ristocetin macrocycle lacking the strong ionic interaction between its free amino group and the carboxylic functionality of the acids. However, in this case, the carbocyclic ring size also played an additional influential role on the separation behavior.

Protein-based CSP

Proteolytic analytes such as carboxylic acids, amines, etc. are the classical analytes for protein-based CSP. During our work, Chiral AGP, an α1 acid glycoprotein CSP, was found to be best applicable in resolving trans-β-amino carboxylic acids 2c and 4c (Table 5) as well as γ-amino acid 7c (Table 6). The separation of acid 8c (Table 6) could not be worked out due to the reasons mentioned before. The separation of amide 4b (Table 4) as well as nitrile 2a (Table 2) was achieved by slightly varying the organic modifier between 0% and 15% for optimal capacity ratio of the more lipophilic nitriles when compared to the acids. This is consistent with a certain reversed-phase behavior reported for this kind of CSP [41]. The complete loss of resolving power for five-membered amide trans-2b and six membered amide cis-3b is pointing out once more the unpredictable response of the CSP with respect to ring size, position of ring substituents, and their relative stereochemistry. In the instance of the amino acids not separable on this CSP, attempts to add ACN up to 5% resulted in a drop of retention and complete loss of separation. Therefore, the organic modifier was omitted giving rise to higher capacity ratios for the acids. Regardless of that, the resolution remained insufficient for a quantitative determination of the enantiomeric excess.

Conclusions

In view of the inconsistency in separation behavior of the present CSPs towards structurally closely related analytes (Fig. 1), the benchmark to separate a mixture of nitrile-, amide-, and acid-enantiomers in a single chromatographic run was initially set too high. However, this was not only due to the failure of the CSPs in separating them but also due to overlapping peaks even in the rare incident of concomitant enantioseparation of each of the acid derivatives.

Moreover, most of the resolutions required tedious optimization efforts.

In summary, the investigation of the separation properties of four different types of CSPs (six CSPs) towards N-protected β-amino acid and γ-amino acid derivatives has nevertheless revealed some trends: α1 acid glycoprotein CSP (Chiral AGP) and Chiralpak AD-H were found to be the most versatile CSPs for the compounds presented in this investigation, covering the majority of enantioseparations as depicted in Tables 2, 3, 4, 5, 6, and 7. Chiralcel OJ turned out to be the least suitable of all the CSP tested.

The overall performance of Chirobiotic R was somewhat disappointing, and good resolution could be achieved only for the cyclic five ring acids 1c, 2c, 5c, and 6c, regardless of the relative cis/trans-configuration of the ring substituents. At least, in this case, there is a certain consistency to notice in the separation behavior of the CSP.

The crown ether-based Crownpak CR(+) CSP efficiently separated all free β-amino acid derivatives and—evidently—any N-protection prevented from separation.