22 CHIRAL SEP ARATION Nelu Grinberg, Thomas Burakowski, and Apryll M. Stalcup 22.1 INTRODUCTION Chirality plays a major role in biological processes, and the enantiomers of a bioactive molecule often possess different biological effects. For example, all pharmacological activity may reside in one enantiomer of a molecule, or enan- tiomers may have identical qualitative and quantitative pharmacological activ- ity.In some cases, enantiomers may have qualitatively similar pharmacological activity, but different quantitative potencies. Since drugs that are produced by chemical synthesis are usually a mixture of enantiomers, there is a need to quantify the level of the isomeric impurity in the active pharmaceutical ingre- dient. Accurate assessment of the enantiomeric purity of substances is critical because isomeric impurities may have unwanted toxicological, pharmacologi- cal, or other effects. Such impurities may be carried through a synthesis and preferentially react at one or more steps and yield an undesirable level of another impurity. The determination of a trace enantiomeric impurity in a sample of a single enantiomer drug substance in the presence of a range of other structurally related impurities and a large excess of the major enan- tiomer remains challenging. The history of enantiomeric separation starts with the work of Pasteur. In 1848 he discovered that the spontaneous resolution of racemic ammonium sodium tartrate yielded two enantiomorphic crystals. Individual solutions of these enantiomorphic crystals led to a levo and dextro rotation of the polar- ized light. Because the difference of the optical rotation was observed in solu- tion, Pasteur suggested that like the two sets of crystals, the molecules are 987 HPLC for Pharmaceutical Scientists, Edited by Yuri Kazakevich and Rosario LoBrutto Copyright © 2007 by John Wiley & Sons, Inc. mirror images of each other and the phenomenon is due to the molecular asymmetry [1]. W hile Pasteur made the historical discovery, subsequent advances in the resolution of enantiomers by crystallization were based on empirical results. Several attempts to separate enantiomers using paper chromatography were met with unsystematic results. In 1952 Dalgliesh postulated that three points of simultaneous interaction between the enantiomeric analyte and the sta- tionary phase are required for the separation of enantiomers [2]. Developments in the field of life sciences and in the pharmaceutical indus- try brought enantiomeric separation to a new level. In the late 1950s/early 1960s, many of the drugs were synthesized and used in a racemic form. An example with tragic consequences was the use of thalidomide, a sedative and a sleeping drug used in the early 1960s which produced severe malformations in newborn babies of women who took it in the early stage of pregnancy. Later it was demonstrated that only the (S)-enantiomer possesses teratogenic properties [3]. Introduction of gas chromatography gave a burst to the field of enan- tiomeric separation. In 1966 a group from the Weizmann Institute of Science in Israel reported the first successful separation of enantiomers using gas chromatography. In a letter addressed to Emanuel Gil-Av after the publication of the first separation of enantiomers on a chiral separation of enantiomers on a chiral gas chromatography (GC) stationary phase [4], A. J. P. Martin wrote: “As you no doubt know, I had not expected such attempts to lead to much success, believing that the substrate-solvent association would normally be too loose to distinguish between the enantiomers.” At the time there were just several reports on the separation of enantiomers using chromatographic methods. Later developments in HPLC gave an additional boost to the field. Today, there are over 60 types of rugged, well-characterized columns capable of separating enantiomers. Unfortunately, there is a great deal of trial and error in choosing a particular column for a chiral separation. Therefore this chapter will summarize a rationale for choosing a stationary phase that is based on the relationship that exists between the analytes and the chiral stationary phases. 22.1.1 Enantiomers, Diastereomers, Racemates Chirality is due to the fact that the stereogenic center, also called the chiral center, has four different substitutions. These molecules are called asymme- trical and have a C 1 symmetry. When a chiral compound is synthesized in an achiral environment, the compound is generated as a 50:50 equimolar mixture of the two enantiomers and is called racemic mixture. This is because, in an achiral environment, enantiomers are energetically degenerate and interact in an identical way with the environment. In a similar way, enantiomers can be differentiated from each other only in a chiral environment provided under 988 CHIRAL SEPARATION the conditions offered by a chiral stationary/mobile phase [5]. The separation of enantiomers using chiral stationary/mobile phases involves the formation of transient diastereomeric complexes between the enantiomeric analytes and the chiral moiety present in the chromatographic column. Thus, diastereomers are chiral molecules containing two or more chiral centers with the same chemical composition and connectivity. They differ in stereochemistry about one or more chiral centers. If two stereoisomers are not enantiomers of one another, they can in principle be separated in an achiral environment—that is, using a nonchiral stationary phase [5]. 22.2 SEPARATION OF ENANTIOMERS THROUGH THE FORMATION OF DIASTEREOMERS Formation of diastereomers for chromatographic purposes can be generated in two ways: transient diastereomers, which occur between the enantiomers and the chiral stationary phase (CSP) during the chromatographic process. Such a process is also called direct separation. The second way is to generate long-lived diastereomers that are formed by chemical reaction between the enantiomer and a chiral derivatizing reagent prior the chromatography. Such a process is called indirect separation. Indirect separation of enantiomers is usually a good technique when everything in direct separation fails. However, it requires suitable functionality in the enantiomers for reaction with a chiral derivatizing agent. The effectiveness of this approach may also depend on a variety of other conditions such as structural rigidity and the spatial relation- ship between the stereogenic centers of the enantiomers and the chiral center introduced through derivatization. When two chiral compounds,racemic A and racemic B, react to form a cova- lent bond between them without affecting the asymmetric center, the stereo- chemical course of the reaction can be as follows [6]: [(±) − A] + [(±) − B] → [+A + B] + [+A − B] + [−A + B] + [−A − B] where the first and the last products constitute an enantiomeric pair and the second and the third products constitute a second enantiomeric pair. In con- trast, the first and the third products and the second and fourth products are diastereomeric pairs. In a chiral environment, one should be able to separate all of these four products.However, because diastereomers possess slightly dif- ferent physicochemical properties, achiral chromatography of this mixture should lead to two peaks (corresponding to the two diastereomers). Indirect approaches such as chiral derivatization with chiral derivatiz- ing reagents (CDR) offers a variety of advantages. For instance, CDRs are cheaper than chiral columns. Separation of the product diastereomers is gen- erally more flexible than the corresponding enantiomeric separation because achiral columns can be used in conjunction with various mobile-phase SEPARATION OF ENANTIOMERS THROUGH THE FORMATION 989 compositions. Depending on the functional groups on the enantiomers, there is a variety of CDRs on the market (chiral anhydrides , acid chlorides, chloro- formates, isocianates, isothiocianates, etc.) which can be applied, which in turn can change the selectivity of a chromatographic system. There are also disadvantages to the chiral derivatization approach includ- ing extra validation. For instance, the derivatizing reagent has to be optically pure, or the analysis can generate false-positive results.In addition, special care needs to be taken that the chiral center of the enantiomers or derivatizing agent is not racemized during the derivatization reaction. Furthermore, unequal detector response of the diastereomers must be corrected via stan- dard procedures [7]. Often, the derivatization requires a long reaction time, which adds to the analysis time. 22.2.1 Mechanism of Separation The separation of diastereomeric pair is due to the effect of their nonequiva- lent shape, size, polarity, and so on, on their relative solvation and sorption energies [8]. Their interaction with a particular stationary phase is dependent upon their molecular structure and availability of functional groups able to interact with the stationary phase. For instance, unsaturated bicyclic alcohols, which are capable of internal hydrogen bonding, show shorter retention than epimers or dihydro derivatives, which cannot undergo such types of interac- tions [9] (Figure 22-1). The compounds of Figure 22-1 were separated by gas chromatography on a 12-ft × 1 / 4 -in. column packed with 23% by weight of Ucon No. 50HB 2000 available from Union Carbide on Celite. As the number of double bonds increases in the molecules, the possibility of intramolecular hydrogen bonds between the hydroxyl groups and the double bond increases. Simultaneously, the potential for hydrogen bond formation between the com- pounds and the stationary phase decreases. As a consequence, the retention time of each isomer decreases as the number of double bonds in the mole- cules increases [10–13]. 990 CHIRAL SEPARATION Figure 22-1. Retention time of bic yclic alcohols. The numbers under each structure represent the retention time in minutes. (Reprinted from reference 9, with permission.) There are few differences between the separation in gas chromatography [14–16] and the separation in liquid chromatography (LC), because it is assumed that the differential solvation of the diastereomeric compounds during the LC separation does not play a very important role [17]. Helmchen et al. [18] explained the separation of diastereomeric amides using LC with a silica gel stationary phase under normal-phase conditions. In order to explain their separation, the authors made some assumptions: 1. Secondary amides adopt essentially the same conformation in polar solu- tions and in the adsorbed state (on silica gel). 2. In the adsorbed state, a parallel alignment of the planar amide group and the surface of silica gel is preferred. 3. Apolar groups (i.e., alkyl, aryl) outside the amide plane cause a distur- bance of this preferred arrangement in proportion to their steric bulk in a direction perpendicular to the amide plane. Such groups are classified as large and small by indices L and S, respectively. 4. That member of a diastereomeric pair in which both faces of the amide plane are more shielded than the least shielded face in the other member is eluted first. 5. There is an attractive interaction between small polar groups and the silica gel, particularly if they are hydrogen bond donors not internally bonded to the amide group. Formally, such groups are assigned to the S (small) class. The actual magnitude of the interaction of a given substituent with the adsorbent depends on the adsorbent, other substituents present, and the type and rigidity of the backbone of the diastereomeric analytes. Although no serious attempts at quantification have been made, repulsive interactions toward silica and alumina can be ranked roughly as H < methyl < phenyl = ethyl < tert-butyl < trifluoromethyl <α-naphthyl < 9-anthryl = pentafluoroethyl < heptafluoroethyl. Size and hydrophobicity are both relevant; incorporation of polar functionality (hydroxyl, carbalkoxy, cyano) leads to attractive rather than repulsive interactions with silica. 22.2.2 General Concepts for Derivatization of Functional Groups As noted previously (Section 22.2), derivatization with a chiral derivatizing reagent (CDR) requires the presence of suitable functionality (e.g., —OH, Ar—OH, —SH,—COOH, —CO—, —NH 2 , —NRH) within the chiral analyte to serve as a reactive site. Before addressing specific issues with regard to CDR and analyte classes, it may be helpful to review general considerations for achiral derivatization in chromatographic assays. Desirable achiral derivatization reaction properties include fast, unidirec- tional reactions with no or minimal side reactions.In addition, both the reagent SEPARATION OF ENANTIOMERS THROUGH THE FORMATION 991 and the product should be stable. Most derivatization methods use an excess of reagent which can present as an interfering chromatographic peak. Of course, incorporating a derivatization step in an assay requires additional materials, time, and effort as well as additional method validation. In the case of chiral derivatization, there are some unique considerations in addition to the ones noted above for achiral derivatization. Extra valida- tion is required to establish the optical purity of the derivatizing agent. In addi- tion, nonracemization of either the analyte or the derivatizing reagent during the derivatization must be confirmed. Excess reagent must be used to elimi- nate any potential chiral discrimination in the derivatization reaction. The presence of more than one type of reactive group (e.g., amine and alcohol) must be considered if the selected reagent has different reaction potentials for each moiety. In some cases, chiral derivatization may be coupled with achiral derivatization. If more than one reactive functional group is present in the analyte, usually the derivative in which the two stereogenic centers are in closest proximity yields the most favorable diastereomeric pair for separation by achiral chromatography. Also, derivatives that incorporate the most struc- tural rigidity (e.g., amides versus esters) tend to be the most amenable to sep- arations by achiral chromatography. 22.3 MOLECULAR INTERACTIONS Generally speaking, there are three properties involved in an intermolecular interaction: the probability of the interaction occurring, the strength of the interaction, and the type of interaction. These properties will be discussed in the following sections. 22.3.1 The Probability of Molecular Interactions Achieving enantiomeric discrimination requires understanding the interac- tions between the selector and the selectand. In his Ph.D. thesis [19], Feibush postulated that attaining an enantiomeric separation on a chromatographic chiral system required that certain conditions should exist: A necessary condition for having a difference in the standard free energy of the two enantiomers in solution is that the solvent is chiral. The fact that the solvent is chiral is in itself not sufficient to sustain such difference. A certain solute–solvent correlation should exist to cause the difference in the behavior of the enantiomers. There should be strong (solute–solvent) interactions, such as p-complexation, coordinative bonds, [and] hydrogen bonds, to form associates between the asym- metric solvent/solute molecules. Such association can be regarded as short-living diastereomers. When the bonds that form these associates are in immediate prox- imity of their asymmetric carbons, a difference in the behavior of the enantiomers in the active phase is possible.We search for active phases and enantiomeric solutes that can form associates through (preferably) more than one hydrogen bond, and 992 CHIRAL SEPARATION where these bonds are formed in the immediate proximity of the asymmetric carbons . In an associate formed through a single H-bond, free rotation of the bonded molecules still exists, on the other hand, more bonds prevent this possi- bility to a large extent, and a solute–solvent associate with a preferred conforma- tion is formed. In addition, having more H-bonds between the asymmetric solute and solvent increases the interaction between these neighboring molecules and increases the population of (the selective) associates where asymmetric carbon are in close proximity. With the increase of the relative population of these particular associates from all the possible associates, an increase in the gap of the free solvation energy of the enantiomers is expected, which enables their GC separation. This model can also be extended to enantiomeric separation using liquid chromatography. Yet enantioerecognition is still a matter of debate [20–22]. More recently, Sundaresan and Abrol [23] proposed a novel stereocenter recognition (SR) model for describing the stereoselectivity of biological and other macromole- cules toward substrates that have multiple stereocenters, based on the topol- ogy of substrate stereocenters. The SR model provides the minimum number of substrate locations interacting with receptor sites that need to be consid- ered for understanding stereoselectivity characteristics. According to this model, the substrate locations and receptor sites can have binding, nonbind- ing, or repulsive interactions that may occur in a many-to-one or one-to-many fashion. The interactions between the two chiral entities must involve a minimum number of locations in the correct geometry.The model predicts that stereoselectivity toward a substrate with N stereocenters in a linear structure involves N + 2 substrate locations distributed over all stereocenters in the sub- strate, such that at least three locations per stereocenter effectively interact with one or more receptor sites. In building models of possible enantioselective associates, conformational searching during docking of the selectands (enantiomeric solutes) with the selector (chiral solvent or ligand) is necessary. Usually it is not known which conformation of a ligand interacts more favorably with a particular receptor, and the flexibility of the ligand plays a major role in such computational approaches [24]. Associations where each of the pairing partners is not in its preferred conformation play only a minor role in the overall interaction between the selectand and the selector, and their contribution to the enan- tioselectivity is minimal. In Figure 22-2, the diastereomeric associates between the selectand/selec- tor are formed through one, two, or three substituents of the asymmetric carbon. The chirality of the selector or the selectand can arise from an asym- metric carbon, the molecular asymmetry, or the helicity of a polymer.Also, the bonds between substituents of the selectand and the selector can involve a single bond, but could also involve multiple bonds or surfaces. Such bonds rep- resent the leading interactions between selectand and selector. Only when the leading interactions take place and the asymmetry of the two bodies are MOLECULAR INTERACTIONS 993 brought in close proximity do the secondary interactions (e.g., van der Waals, steric hindrance , dipole–dipole) become effectively involved. The secondary interactions can affect the conformation and the formation energy of the diastereomeric associates. If the interaction between the selectand and the selector takes place through one leading interaction (Figure 22-2A), then the enantioselectivity of the system is governed by the position of unbounded substituents B, C, and D of the selectand relative to the sub- stituents F, G, and H of the selector. One particular enantiomer will interact more strongly with a particular selector if the contour and polarity of the two molecules are better complements of each other. When the interaction between the selectand and the selector occurs through two leading interac- tions (Figure 22-2B), the enantioselectivity of the system is determined by the effective size of the groups that do not participate in interactions. If, for example, G of the selector is an alkyl and H a hydrogen substituent, and C of the selectand is an alkyl group and D a hydrogen, then one enantiomer has 994 CHIRAL SEPARATION Figure 22-2. Schematic representation of selectand/selector associations. Dashed lines represent the leading interactions between the two chiral entities. (Reprinted from ref- erence 25, with permission.) the larger G and C groups in syn arrangement and the other in anti arrange- ment. In a variety of cases involving interactions through hydrogen bonding or ligand metal complexes, the enantiomer whose larger nonbonded groups are positioned syn to the corresponding larger group of the selector will elute last from a chromatographic column, as compared to the opposite isomer that forms the anti arrangement [25]. The solvation energy of one enantiomer in the active chiral phase can be described as the contribution of all possible forms of solvent/solute associates. These associates are in equilibrium with fast interconversion rates. Each form contributes to the total free energy according to its particular formation energy and its particular molar fraction [25, 26].These complexes between the selector and selectand should also be as mutually exclusive as possible, to prevent a given interaction from occurring at multiple sites in the diastere- omeric complexes [5]. 22.3.2 The Types of Molecular Interactions Chiral separations generally rely on the formation of transient diastereomeric complexes with differing stabilities. Complexes are defined as two or more compounds bound to one another in a definite structural relationship by forces such as hydrogen bonding, ion pairing, metal-ion-to-ligand attraction, π-acid/ π-base interactions, van der Waals attractions, and entropic component desol- vation. In the following sections, the most important types of molecular interactions in chiral separations are discussed. 22.3.3 Chiral Separation Through Hydrogen Bonding Hydrogen bonding is a donor–acceptor interaction specifically involving hydrogen atoms [27]. When a covalently bonded hydrogen atom forms a second bond to another atom, the second bond is referred to as a hydrogen bond. A hydrogen bond is formed by interaction between the partners R—X— H and :Y—R′ according to R—X—H + :Y—R′↔R—X—H···Y—R′ where R—X—H is the proton donor and :Y—R′ makes an electron pair avail- able for the bridging bond. Hydrogen bonding can be regarded as a prelimi- nary step in a Brønsted acid–base reaction, which would lead to a dipolar reaction product R—X − ···H—Y + —R′. According to their bonding energy, hydrogen bonds can be subdivided into three categories: strong, moderate, and weak hydrogen bonds. Strong hydro- gen bonds are formed by groups in which there is a deficiency of electron density in the donor group, (i.e., —O + —H, >N + —H) or an excess of electron MOLECULAR INTERACTIONS 995 density in the donor group (i.e., F − ,O − —H, O − —C, O − —P, N − <). They are referred to as forced strong H-bonds [27]. Moderate hydrogen bonds are generally formed by neutral donor and acceptor groups, such as —O—H, ¨N—H, or —N(H)—H and O¨,O¨C, or N¨, in which the donor X is electronegative relative to hydrogen and the Y atom (the acceptor) has a lone pair of unshared electrons. These are the most common hydrogen bonds and are essential contributors to the structure and function of biopolymers. Weak hydrogen bonds are formed when the hydrogen atom is covalently bonded to a slightly more electrically neutral atom relative to hydrogen (e.g., C—H, Si—H) or when the acceptor group has no lone pair but has π elec- trons, (e.g., C¨C or an aromatic ring). Although F is a very electronegative atom, F—C or F—S groups are only weak acceptors. These interactions have energies and geometries similar to those of van de Waals complexes, and they are distinguished from them by evidence of a directional involvement of the X—H bond. The H-bond is generally assumed to be linear with θ between 175–180°. The geometrical requirement can, in certain cases, lead to arrangements in which a covalently bonded H-atom is located close to more than one poten- tial acceptor atom,leading to a bifurcated hydrogen bond [28]. Such complexes have lower stability than those with a single hydrogen bond. An example of a bifurcated hydrogen bond between two drug enantiomers and amylose car- bamate stationary phase is presented in Figure 22-3.The right-hand side enan- tiomer undergoes a bifurcated hydrogen bond with the amylose phase, forming a complex less stable than that from the left-hand side. As a conse- quence, the enantiomer forming the bifurcated hydrogen bond eluted earlier from the chromatographic column [29]. 996 CHIRAL SEPARATION Figure 22-3. Interaction of two drug enantiomers with amylose carbamate stationary phase . (Reprinted from reference 29, with permission.) [...]... conformation They may be considered as rigid building blocks giving fairly limited conformational freedom of the macrocycle in rotation of the C6–O6 groups and limited rotational movements about the glucosidic link C1(n)-O4(n − 1)-C4 1004 CHIRAL SEPARATION Figure 22-9 Chemical structure of CA7 (β-CD) where the numbering of glucose unit (1–7) is performed counterclockwise (left) Atom numbering scheme for. .. section, we will present several chiral phases employed either in GC or in normal-phase HPLC for which the hydrogen-bonding interactions discussed above governs the interactions between the selectand and the selector It should be noted that the interactions occurring in GC are similar to those occurring in normal-phase HPLC The first successful chiral phases used under GC conditions were Ntrifluoro-acetyl... entity was incorporated in a polysiloxane backbone for higher thermal stability Some of the compounds separated on Chirasil-Val® contained only groups, such as N-TFA-proline esters, that are able to accept hydrogen bonding To undergo such an interaction, the diamide phase has to have a conformation where both NH groups point toward the selectand in a conformation similar to the α-helix structure of proteins... either trans or cis configuration, forms that are in equilibrium: Figure 22-4 The geometry of the peptide backbone, with the trans peptide bond, showing all the atoms between two Cα atoms of adjacent residues (Reprinted from reference 31, with permission.) 998 CHIRAL SEPARATION The trans form is energetically favored, due to less repulsion between nonbonded atoms [31] For an amide group to hydrogen bond... triethylamine, can also be effective for the separation of enantiomers mediated by the CDs Under these conditions, the interior of the CD cavity is occupied by acetonitrile The overwhelming concentration of acetonitrile renders its displacement by the enantiomeric analytes basically impossible Acetonitrile is a polar aprotic solvent, with limited capacity for hydrogen bond formation As a consequence, under... presence of aromatic substitution provides possibilities for π–π interaction with the aromatic substituents of the enantiomeric analytes [61, 62] For example, in (R)-(−)-, or (S)(+)-1-(1-naphthyl)ethyl carbamate of β-CD, the naphthyl ethyl moiety has some π donor character Incorporation of 3,5-dinitrophenyl substituents on chiral analytes promotes formation of a π–π complex At the same time, the carbamate... (Figure 22-20) In this conformation, where the (R)-(−)-TAPA and M-(−)helicene molecules are parallel to each other, the substituents of the asymmetric carbon sterically hinder the π–π overlap and impair the interactions of the M-(−)-isomer In other complex conformations where the selectand–selector molecules are antiparallel to each other, both M-(−)- and P-(+)-helicenes can form readily π–π overlapping... configurationally known solutes support this model Controlling the conformational mobility of the chiral selector on the CSP can enhance chiral recognition For instance, chiral phases incorporating l-proline were designed to separate the enantiomers of N-(3,5dinitrobenzoyl)amino acid esters and related analytes Separation factors as high as eight were obtained for N-(3,5-dinitrobenzoyl)leucine amides [95] The structure... role in the design of chiral selectors In principle, if a single molecule of a chiral selector has different affinities for the enantiomers of another substance, then a single enantiomer of the latter will have different affinities for the enantiomers of the initial selector In an effort to design a chiral stationary phase capable of separating naproxen, Pirkle et al [97] first designed two stationary... groups through esters or carbamate formation greatly enhanced the enantioselective properties of these polysaccharides In 1973, Hesse and Hagel reported [109] for the first time the complete separation of Tröger’s base on a column filled with microcrystalline cellulose triacetate The triacetate cellulose is believed to preserve a structure closely related to native cellulose (form I) (CTA-1) The sorption of . crystals, the molecules are 987 HPLC for Pharmaceutical Scientists, Edited by Yuri Kazakevich and Rosario LoBrutto Copyright © 2007 by John Wiley & Sons,. error in choosing a particular column for a chiral separation. Therefore this chapter will summarize a rationale for choosing a stationary phase that is based