High-performance liquid chromatographic enantioseparation of isopulegol-based ß -amino lactone and ß -amino amide analogs on polysaccharide-based chiral stationary phases focusing on the change of the enantiomer elution order

The enantioselective separation of newly prepared, pharmacologically signiﬁcant isopulegol-based ß - amino lactones and ß -amino amides has been studied by carrying out high-performance liquid chromatography on diverse amylose and cellulose tris -(phenylcarbamate)-based chiral stationary phases (CSPs) in n -hexane/alcohol/diethylamine or n -heptane/alcohol/ diethylamine mobile phase systems. For the elucidation of mechanistic details of the chiral recognition, seven polysaccharide-based CSPs were employed under normal-phase conditions. The effect of the nature of selector backbone (amylose or cellulose) and the position of substituents of the tris- (phenylcarbamate) moiety was evaluated. Due to the complex structure and solvation state of polysaccharide-based selectors and the resulting enantioselective interaction sites, the chromatographic conditions ( e.g ., the nature and content of alcohol modiﬁer) were found to exert a strong inﬂuence on the chiral recognition process, resulting in a particular elution order of the resolved enantiomers. Since no prediction can be made for the observed enantiomeric resolution, special attention has been paid to the identiﬁcation of the elution sequences. The comparison between the effectiveness of covalently immobilized and coated polysaccharide phases allows the conclusion that, in several cases, the application of coated phases can be more advantageous. However, in general, the immobilized phases may be preferred due to their increased robustness. Thermodynamic parameters derived from the temperature-dependence of the selectivity revealed enthalpically-driven separations in most cases, but unusual temperature behavior was also observed.


Introduction
β-Amino acid derivatives such as β-amino lactones and βamino amides have remarkable pharmacological importance. Lactones of natural β-amino acids, obtained from sesquiterpene-type α, β-unsaturated lactones, e.g ., alantolactone, isoalantolactone or ambrosin, possess significant biological activities, such as increasing the proportion of cells in the G2/M and S phase [1] . Their water-soluble derivatives, in turn, exhibit cytotoxic activity through * Corresponding author.
E-mail address: ilisz@pharm.u-szeged.hu (I. Ilisz). a prodrug mechanism for different human cancer cell lines [2] . In addition, ring opening of β-amino lactones with different amines results in β-amino amides, which are well-known subunits of biologically important compounds, such as α-hydroxy-β-amino amide bestatin, a potent aminopeptidase B. Its usefulness in the treatment of cancer through its ability to enhance the cytotoxic activity of known antitumor agents was described in the literature [3] . β-Amino amides exhibit other important biological activities as well. For example, pinane-based β-amino amides and similar bicyclic, norbornene-based amides with N -heteroaryl substituents possess tyrosine kinase inhibitor properties or even antibiotic activity [ 4 , 5 ]. Sitagliptin, a novel antidiabetic drug (Januvia®) bearing Fig. 1. Structure of isopulegol-based ß-amino lactones and ß-amino amides a β-amino amide moiety, is a lead antidiabetic agent [6] . Furthermore, some hydroxyl-substituted β-amino amides have remarkable HIV protease or renin inhibitor activities [7] . The determination of enantiomeric and diastereoisomeric purity of β-amino lactones and hydroxyl-substituted β-amino amides is of high significance, because these synthons are excellent starting materials for the synthesis of other families of bioactive building blocks, including aminodiols (by reduction of amino lactones), diamino alcohols (by reduction of hydroxyl-substituted β-amino amides), and their heterocyclic derivatives.
The main objective of the present paper is to reveal possible structure-separation relationships of the pharmacologically interesting ß-amino lactones and ß-amino amides. Our interest is based on the information that, to the best of our knowledge, no separation has been reported for ß-amino lactone enantiomers so far, and only a few cases were described for the enantiorecognition of ß-amino amides. Investigations were carried out on amyloseand cellulose-based tris -(phenylcarbamate)-type CSPs, due to their wide applicability and robust behavior described often in the literature. The study focused on exploring various effects observed with the variation of mobile phase composition, the nature and concentration of the alcohol modifier, the structure of chiral selectors and analytes, and the temperature on retention, selectivity, and resolution of stereoisomers. Elution sequences were determined in all cases.

Apparatus and chromatography
Liquid chromatographic measurements were performed with the use of two chromatographic systems. The Waters Breeze system consisted of a 1525 binary pump, a 2996 photodiode array detector, a 717 plus autosampler, and Empower 2 data manager software (Waters Corporation, Milford, MA, USA). A Lauda Alpha RA8 thermostat (Lauda Dr. R. Wobser Gmbh, Lauda-Königshofen, Germany) was used to maintain constant column temperature.
The 1100 Series HPLC system from Agilent Technologies (Waldbronn, Germany) contained a solvent degasser, a pump, an autosampler, a column thermostat, and a multiwavelength UV-Vis detector. Data acquisition and analysis were carried out with ChemStation chromatographic data software from Agilent Technologies.
All analytes were dissolved in 2-PrOH or EtOH in the concentration range 0.5-1.0 mg ml −1 and injected in a volume of 20 μL. The dead times of the columns were determined by injection of tri-t -butylbenzene.

Results and discussions
The ß-amino lactones and ß-amino amides as summarized in Fig. 1 are isopulegol-based analytes with benzyl, methylbenzyl or dibenzyl moieties attached to the N -atoms. Opening the ß-lactone ring (analyte 5, 6, and 7) modifies the structural characteristics of the molecules and may influence their interactions with chiral selectors.

The effect of mobile phase composition
Polysaccharide-based CSPs are most frequently employed in normal-phase mode (NPM), applying mixtures of a nonpolar hydrocarbon (typically n-hexane or n-heptane) and an alcohol of low molecular weight ( e.g ., EtOH, 1-PrOH, 2-PrOH, BuOH) as mobile phase [ 19 , 20 ]. The variation of the nature and concentration of alcohol serves most often for the modulation of the chromatographic behavior ( i.e ., retention and stereoselectivity) in NPM [33][34][35][36] .
To study the effect of the nature of alcohol modifier on chromatographic parameters, analytes 1, 2, 4, and 6 were selected as representatives of the complete set of analytes of this study. To avoid the generation of an unnecessary large data set among the nine polysaccharide-based CSPs, four of them were selected on the basis of structural similarities. These are amylose-and cellulosebased tris -(3,5-dimethylphenylcarbamate) (Chiralpak IA and IB) and tris -(3,5-dichlorophenylcarbamate) (Chiralpak IE and IC). For the purpose of a reliable comparison, the studied alcohols, namely EtOH, 1-PrOH, 2-PrOH, and BuOH, were used at the same molar concentration of 1.298 M. This corresponds to a different volume ratio of each alcohol in the mobile phase as follows: EtOH: 7.6 v%, 1-PrOH: 9.7 v%, 2-PrOH: 10.0 v%, and BuOH: 11.9 v%.
Data obtained with the change of the alcohol are presented in Supplementary Information (Table S1). Under normal phase conditions, increasing the apolar character of the alcohol usually results in enhanced analyte retention; however, opposite observations have also been described [ 35 , 36 ]. Under the applied conditions, no general trends can be observed in retention factors: k increased with alcohol apolarity unequivocally only for Chiralpak IE in the case of analyte 1 and 2. Interestingly, separation factors, in most cases, changed only slightly ( < 10%) with the variation of the nature of alcohol. From a practical point of view, it is important to note that unlike selectivity, resolution is much more dependent on the nature of the alcohol modifier. Depending on the structure of the analyte and the chiral selector, R S values were higher with EtOH or 2-PrOH, however, in some cases, the highest R S values were registered in the presence of BuOH. The changein enantioselectivity caused by changing the alcohol modifier was previously rationalized as a result of alteration of the steric environment of the chiral cavities within the chiral polymer material induced by different alcohol modifiers [ 17 , 18 ]. Taking into account all results obtained with respect to the effect of the nature of alcohol on chromatographic parameters in NPM, the use of 2-PrOH and, in some cases, EtOH was favored for this class of compounds. Consequently, these two solvents were chosen for further studies.
Besides studying how the nature of alcohol affects the chiral recognition ability, comparing n -hexane and n -heptane as the most frequently applied NP solvents is of scientific interest. (It is worth mentioning that n -heptane is less toxic compared to nhexane.) Previous works have shown improvements in selectivity with the use of n -heptane over n-hexane [37] . Applying Chiralpak IB with mobile phases of n -hexane/2-PrOH/DEA and n -heptane/2-PrOH/DEA and analytes 2 and 4, n -heptane showed no improvements over n -hexane: retention times, in most cases, were slightly shorter, but α and R S were significantly lower in mobile phases containing n -heptane. It should be noted here that this is only a limited data set (Fig. S3).
For the study of the effects of modifier concentration on chromatographic parameters, two pairs of isopulegol-based ß-amino lactone and ß-amino amide (analytes 1, 5 and 2, 6) were chosen. The mobile phase systems were n -hexane/2-PrOH/DEA and nhexane/EtOH/DEA containing 2-PrOH and EtOH at the same molar concentration (3.893, 2.596, 1.298, and 0.649 M), all containing 20 mM DEA, as the usual mobile phase additive used for the chromatography of basic analytes. Chiralpak IA and Chiralpak IE, as the best performing CPSs, were selected for this study. Regarding the retentive characteristics, a typical NP behavior was observed for both alcohol modifiers studied: increasing the apolar n -hexane to alcohol ratio resulted in an increased k 1 ( Fig. 2 ). Enantioselectivity exhibited only a small change with increasing n -hexane content. Most notably, R S , in most cases, increased significantly, in particular, for analyte 6 in mobile phase systems containing 2-PrOH. It is worth mentioning that the change in the chromatographic performance caused by the alcohol modifier depended on the structure of the chiral selector as well. Specifically, on Chiralpak IA, slightly higher k 1 , α, and R S were observed for analytes 1, 2, and 6 with the use of EtOH, while on Chiralpak IE, 2-PrOH had a similar effect for analytes 1, 5, and 6.
Not only the nature of the alcohol modifier, but also its concentration in a given mobile phase may affect the elution sequence as observed in several cases on polysaccharide-based CSPs [ 29 , 34 , 38 ]. In the present study, the reversal of elution order for analyte 5 on Chiralpak IA was registered by changing the composition of nhexane/2-PrOH/DEA mobile phase from 95/5/0.1 v/v/v to 60/40/0.1 ( Fig. 3 ), which probably due to the change in the solvation state of the chiral selector.

The effect of the structure of selectors
The amylose-and cellulose-based selectors are constructed of α or ß 1,4-linked glucopyranose units, respectively. The different linkage is responsible for a difference in the secondary structure of these polysaccharides and of their derivatives. Due to these differences, the interactions between analyte and selector may change and this results in different chromatographic behaviors. Table 1 summarizes chromatographic data for the seven ßamino lactones and ß-amino amides obtained on seven polysaccharide phases at the same mobile phase composition of n -hexane/2- The effect of the polysaccharide backbone can be evaluated by the comparison of the chromatographic data of amylose and cellulose tris -(3,5-dimethylphenylcarbamate) (Chiralpak IA vs. Chiralpak IB) and tris -(3,5-dichlorophenylcarbamate) (Chiralpak IE vs. Chiralpak IC), respectively. According to data in Table 1 , in most cases, k 1 , α, and R S were higher on amylose-than on cellulosebased CSPs. It appears that, with a few exceptions, the studied analytes fit better to the amylose-than to the cellulosebased polymeric CSP, especially in the case of ß-amino amides with the ß-lactone ring opened. The structural differences between amylose-and cellulose-based tris -(3,5-dimethylphenylcarbamate) or tris -(3,5-dichlorophenylcarbamate) were found to be reflected in the chiral recognition pattern toward some analytes. Reversal of elution order between amylose-and cellulose-based CSPs, containing the same substituents was registered for analytes 1, 4, and 6 on Chiralpak IA and IB, and for analytes 5 and 6 on Chiralpak IE and IC ( Table 1 and Fig. 4 A). Examples of reversed elution orders of analytes on amylose-or cellulose-based columns have been described previously [ 29 , 34 ].
The effect of the nature of the phenylcarbamate moiety can be estimated by comparing amylose tris -(3,5dimethylphenylcarbamate) (Chiralpak IA) and amylose tris -(3,5dichlorophenylcarbamate) (Chiralpak IE) or cellulose tris -(3,5dimethylphenylcarbamate) (Chiralpak IB) and cellulose tris -(3,5dichlorophenylcarbamate) (Chiralpak IC). Data in Table 1 reveal that much higher retentions were registered for all analytes on CSPs with tris-(3,5 -dichlorophenylcarbamate) moiety than on CSPs possessing the tris-(3,5-dimethylphenylcarbamate) moiety. Higher retentions were generally accompanied with higher α and R S values showing that dichloro rather than dimethyl substitution favored the enantioselective interactions, probably through enhanced π -π interactions. In a few cases lower α and R S were registered on Chiralpak IE than on Chiralpak IA, but these differences were not significant. In this study, the reversal of elution order was registered for analytes 1, 5, and 7 in the case of Chiralpak IA and IE and for analyte 4 in the case of Chiralpak IB and IC (related examples are depicted in Fig. 4 B). The reversal of elution sequence by the change of the chemical structure of substituents on the tris-(phenylcarbamate) moiety was also mentioned in earlier publications [ 29 , 34 , 39 , 40 ]. Table 2 Effect of mobile phase composition on k 1 , α, and R S of isopulegol-based β-amino lactones and β-amino amides The effect of the position of the methyl substituent in the phenylcarbamate moiety on the chromatographic performance was investigated by comparing chromatographic data obtained on amylose tris-(3-chloro-4-methylphenylcarbamate) (Chiralpak IF) and amylose tris-(3-chloro-5-methylphenylcarbamate) (Chiralpak IG). For all analytes, higher retentions were obtained on Chiralpak IG than on Chiralpak IF, but higher retention was accompanied with higher selectivity and resolution only for half of the studied analytes. It shows that the methyl substituent in position 5 offers stronger retentive interactions, but enantioselectivity may be reduced, probably for steric reasons.
The new generation of covalently immobilized polysaccharide phases are very robust and can be applied in different modalities with different bulk solvents [ 28 , 29 , 41 , 42 ]. A comparison of separation performances of covalently immobilized and coated polysaccharide CSPs were performed for analytes 1, 2, and 6 by applying immobilized and coated amylose tris -(3,5dimethylphenylcarbamate) (Chiralpak IA vs. Chiralpak AD-H) and cellulose tris -(3,5-dimethylphenylcarbamate) (Chiralpak IB vs. Chiralcel OD-H) with the same mobile phase composition of nhexane/2-PrOH/DEA (95/5/0.1 v/v/v ) and n -hexane/ethanol/DEA (95/5/0.1 v/v/v ) ( Table 2 ). Data in Table 2 revealed that in almost all cases higher k 1 , α, and R S values were registered on coated CSPs than on the immobilized CSPs. Interestingly, a reversal of elution sequence was registered for analyte 6 on Chiralpak IA vs. Chiralpak AD-H in the n -hexane/ethanol/DEA (95/5/0.1 v/v/v ) mobile phase system ( Fig. 5 A). A similar change was reported by Chankvetadze et al. [29] . Moreover, for analyte 6 on Chiralpak AD-H, the change of EtOH to 2-PrOH in n -hexane also resulted in a reversed elution sequence ( Fig. 5 B).
The strong dependence of the elution order of the individual enantiomers on the applied conditions calls particular attentions to the need of identification of each enantiomer in the case of polysaccharide-based CSPs. The complex structure of polysaccharide-based selectors and their applied conditions depending on solvation status do not allow to predict chiral recognition and elution order at these times.

The effect of the structure of analyte
Analytes 1-4 are ß-amino lactones, while 5-7, the ring-opened analogs of 1-3, are ß-amino amides. These structural differences may affect chromatographic behavior and chiral recognition. Analyte 4, compared to analyte 1, contains two benzyl moieties instead of a single benzyl group. According to chromatographic data ( Table 1 ), more bulky analyte 4 fits less well into the cavity of amylose or cellulose backbone resulting in a significantly shorter retention.Among the studied CSPs selectivity and resolutions were higher with Chiralpak IE, IF, and ID, probably due to enhanced π -π interactions of analyte 4. Analytes 2 and 3 possess an extra methyl moiety compared to analyte 1. This structural difference has marked influences on the chromatographic behavior. Analyte 2 and 3 are much less retained by each CSP, but in several cases, their enantiomers exhibited better resolution, possibly due to steric reasons. Analytes 5, 6, and 7, ring-opened analogs of analytes 1, 2, and 3, contain an extra hydroxyl and a secondary amino group capable of hydrogen bonding interactions with the carbamate moiety. Furthermore, the additional benzyl ring may be involved in π -π interactions. The presence of extra interaction sites, in most cases, led to enhanced enantioselectivity, while retention was generally smaller for the amino amide analogs, suggesting reduced nonselective interactions for these compounds.
It is interesting to examine how the structure of analyte affects the elution sequence. In case of analyte 1 the elution sequence depends strongly on the applied CSP, while no changes in elution order were observed for analytes 2 and 3 ( Table 1 ). This draws attention how a simple methyl substitution by creating a new chiral center can affect the chiral recognition. It is important to highlight that the methyl substitution in the same position in case of the amides (5 vs 6 and 5 vs 7) did not result in a consistent change in the elution sequences. On the basis of this limited data set no clear trend can be suggested how the structure of analytes affect the elution sequence.
For the quantitative characterization of the optimized methods, limits of both detection (LOD) and quantitation (LOQ) were determined for analytes 2 and 6 on Chiralpak IA and Chiralpak IE columns. Due to the better peak shapes sligthly lower LOD and LOQ values were obtained on Chiralpak IE, where LOD and LOQ values for analyte 2 were 6.9 pmol and 23.2 pmol, respectively, while these values for analyte 6 were 4.9 pmol and 16.3 pmol, respectively. Fig. 6 depicts the chromatograms obtained on Chiralpak IE for analytes 2 and 6 for the minor enantiomer in the presence of the major one.

Effect of temperature and thermodynamic parameters
By careful interpretations of the van't Hoff equation, the studies of temperature dependence of retention and enantioselectivity may offer valuable information on the chiral recognition process. For the enantiomeric pairs, the difference in the change in standard enthalpy ( H °) and entropy ( S °) can be obtained on the basis of the van't Hoff equation, not forgetting about the limitations of the simplified approach applied in this study ( i.e ., not differentiating between chiral and achiral contributions, which may vary in their magnitude) [43][44][45][46] .
In order to investigate the effects of temperature on the chromatographic parameters, a variable temperature study was carried out for analytes 1, 2, 5, and 6 on Chiralpak IA, Chiralpak AD-H, and  [46] . The corresponding experimental data are summarized in Table S2. Transfer of the analyte from the mobile phase to the stationary phase can commonly be described as an exothermic process. Because of this reason, retention decreases with increasing temperature. On the three studied columns with both mobile phase systems, k and α decreased with increasing temperature in most cases. However, for analyte 1 on Chiralpak IE and for analyte 6 on Chiralpak IA in n -hexane/ethanol/DEA (70/30/0.1 v/v/v ), k decreased, but α increased with increasing temperature (Table S2 and Fig. S4).
From the chromatographic data on the basis of Eq. 1 , where R is the universal gas constant, T is temperature in Kelvin, and α is the apparent selectivity factor, ln α vs. 1/T plots were constructed. As a general trend, linear plots were obtained as indicated Table 3 Thermodynamic parameters, ( H °), ( S °), Tx ( S °) 298K , ( G °) 298K , correlation coefficients, ( R 2 ), Q values, and T iso temperatures of isopulegol-based β-amino lactones and ß-amino amides on Chiralpak IA, Chiralpak AD-H, and Chiralpak IE columns.
Analyte by the correlation coefficients listed in Table 3 . In most cases, differences in the changes in standard enthalpy and entropy, -( H °) and -( S °), in both mobile phases were more negative on Chiralpak IA than on Chiralpak IE ( Table 3 ) indicating a stronger adsorption process. Interestingly, -( H °) and -( S °) values for Chiralpak IA and Chiralpak AD-H were very similar. The two CSPs possess the same selector in covalently bonded or coated form and, consequently, a retention mechanism independent of the immobilization of the selector can be suggested. According to the data of Table S2, retention decreases in every case, but selectivity increases with increasing temperature in two cases, as reported previously in chromatographic systems applying polysaccharide-type phases [ 28 , 29 , 34 , 38 , 47 ]. The T iso value (the temperature where the enantioselectivity cancels), in most cases, were above room temperature ( Table 3 ). To estimate the enthalpy/entropy contribution to the free energy, Q [ Q = ( H °)/[298 × ( S °)] values were calculated. According to data in Table 3 , Q values, in most cases, were higher than 1.0, indicating the relatively higher contribution of the enthalpy to the free energy. For the systems in which analytes possess negative T iso , Q < 1 suggests a predominantly entropic contribution to the free energy. That is, enantiodiscrimination was driven by entropy in these cases.

Conclusions
Enantioseparations of newly prepared ß-amino lactones and ßamino amides were carried out on amylose-and cellulose-based tris -(phenylcarbamate) stationary phases in n -hexane/alcohol/DEA and n -heptane/alcohol/DEA mobile phases. Regarding mobile phase composition, in case of the studied compounds, applications of 2propanol and ethanol in the mobile phase seem to be more advantageous, while changing between n -hexane and n -heptane leads to only slight differences in separation performances. The nature and content of alcohol modifier may have a significant influence on the elution sequence.
The nature of the chiral selector backbone (amylose or cellulose) together with the nature of substituents of the phenylcarbamate moiety influence not only the separation performance but also the elution sequence in several cases. In the applied chromatographic systems in general, much higher retentions were registered for all analytes on CSPs with tris-(3,5dichlorophenylcarbamate) moiety than on CSPs possessing tris-(3,5-dimethylphenylcarbamate) moiety, probably due to ππ acceptor type of interactions. The chemical structure of the substituent on the amylose or cellulose backbone may influence not only retention and selectivity but also the elution sequence.
The study of the effect of the position of the substituents of the phenylcarbamate moiety on the chromatographic performance in the case of amylose-based CSPs revealed that tris-(3-chloro-5methylphenylcarbamate) is more efficient regarding the chiral interaction between selector and the investigated analytes than that on tris-(3-chloro-4-methylphenylcarbamate).
The new generation of covalently immobilized polysaccharide phases are very robust. However, regarding separation performances for the analytes studied, higher k 1 , α, and R S were registered on coated CSPs than on the comparable immobilized ones. Rarely reported so far, but it is worth highlighting that the change between the two types of CSPs may result in a reversal of the elution sequence.
The structure of selector and analyte, the mobile phase composition (nature and content of bulk solvent and alcohol modifier), and temperature may affect the observed elution order. Consequently, the identification of enantiomers is mandatory for a valid interpretation of data.
Regarding the effect of the nature of analytes, it can be concluded that enantiodiscrimination of ß-amino amides were generally more pronounced, despite their shorter retention times.
The temperature-dependence study revealed enthalpically driven recognition in most cases, but entropy-controlled separation in n -hexane/ethanol mobile phase system was also observed under the chromatographic conditions employed in this study.

Declaration of Competing Interest
Authors declare no conflict of interest.