Anal. Chem. 1996, 68, 1360-1368
Application of Sulfated Cyclodextrins to Chiral Separations by Capillary Zone Electrophoresis Apryll M. Stalcup* and Kyung H. Gahm
Department of Chemistry, 2545 The Mall, University of HawaiisManoa, Honolulu, Hawaii 96822
Mixtures of randomly substituted sulfated cyclodextrins (degree of substitution, ∼7-10) were successfully used as chiral additives for the enantioseparation of 56 compounds of pharmaceutical interest, including anesthetics, antiarrhythmics, antidepressants, anticonvulsants, antihistamines, antihypertensives, antimalarials, relaxants, and bronchodilators. The separations were accomplished at pH 3.8, with the anode at the detector end of the column. Under these conditions, in which electroosmotic flow is directed toward the injection end of the column and the electrophoretic mobility of the negatively charged cyclodextrin is toward the detector, none of the analytes reached the detector in the absence of the sulfated cyclodextrin. Most (40) of the successfully resolved enantiomers contained basic functionality and a stereogenic carbon. However, the versatility of this sulfated cyclodextrin additive was also demonstrated by the fact that three atropisomers, 1,1′-binaphthyl-2,2′-diyl hydrogen phosphate, 1,1′-binaphthyl-2,2′-diol, and Troger’s base, and several neutral analytes were also successfully enantioresolved under these conditions. The separation mechanism seems to involve inclusion complexation. Cyclodextrins (CDs) have demonstrated tremendous utility as chiral additives (CAs) for liquid chromatography and for capillary zone electrophoresis (CZE).1-3 The low solubility of the native β-CD, the most widely applicable CD in liquid chromatography, has engendered the widespread use of functionalized CD,4 such as methylated5,6 and hydroxypropylated CD,7 in CZE. Modification of the net electrophoretic migration of the chiral analyte through enantiospecific complexation between the CD additive and the chiral analyte is thought to drive the enantioseparation. One limitation in this approach is that neutral analytes in the presence of neutral cyclodextrins are not enantioresolvable unless the system is further modified, through the addition of either surfactants8 or ionic cyclodextrins.9 (1) Ward, T. J. Anal. Chem. 1994, 66, 632A-640A. (2) Terabe, S.; Otsuka, K.; Nishi, H. J. Chromatogr. 1994, 666, 295-319. (3) Nishi, H.; Terabe, S. J. Chromatogr. 1995, 694, 245-276. (4) Bechet, I.; Paques, P.; Fillet, M.; Hubert, P.; Crommen, J. Electrophoresis 1994, 15, 818-823. (5) Sepaniak, M. J.; Cole, R. O.; Clark, B. K. J. Liq. Chromatogr. 1992, 15, 1023-1040. (6) Soini, H.; Riekkola, M. L.; Novotny, M. V. J. Chromatogr. 1992, 608, 265274. (7) Rawjee, Y. Y.; Vigh, G. Anal. Chem. 1994, 66, 619-627. (8) Nishi, H.; Fukuyama, T.; Terabe, S. J. Chromatogr. 1991, 553, 503-516. (9) Sepaniak, M. J.; Copper, C. L.; Whitaker, K. W.; Anigbogu, V. C. Anal. Chem. 1995, 67, 2037-2041.
1360 Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
It has been pointed out,10 however, that the maximum opportunity for chiral separation may exist when the electrophoretic mobility of the CA is opposite that of the analyte. The general utility of this approach has been demonstrated using polysulfated carbohydrates, such as heparin11-13 and dextran sulfate,13 for the chiral separation of basic drugs. The promise of sulfonated cyclodextrins, first introduced by Stobaugh,10 has been further highlighted by Lurie et al.14 and Terabe et al.15 Blaschke et al.16 noted the advantage of the “countercurrent” flow of the negatively charged additive with respect to the electroosmotic flow. Other anionic cyclodextrins used in this manner include sulfated17,18 and carboxymethyl-β-cyclodextrin.19-21 In each of these cases, the electrophoretic mobility of the CA opposed the electroosmotic flow, which was directed toward the detector end of the column. Thus, in all of these studies, the chiral selectivity may have been limited by the leveling effect of the electroosmotic flow. As a result, there has been some work done to minimize the electroosmotic flow.18,22 An extension of using anionic cyclodextrins for chiral separations by CZE might be to reverse or suppress the electroosmotic flow altogether and reverse the applied voltage polarity.23 In this way, with the electrophoretic mobility of cationic species directed toward the injector and the absence of other mechanisms for transporting cationic or neutral species toward the detector, the only possibility for analyte migration toward the detector is through complexation, possibly stereospecific, with a negatively charged CA. The potential for this approach became apparent during studies designed to probe the effect of low pH on the net migration of heparin.24 The reversal of the polarity of the applied voltage allowed confirmation that the electrophoretic mobility of heparin was, indeed, able to overcome the electroosmotic flow of (10) Tait, R. J.; Thompson, D. O.; Stobaugh, J. F. Anal. Chem. 1994, 66, 40134018. (11) Stalcup, A. M.; Agyei, N. M. Anal. Chem. 1994, 66, 3054-3059. (12) Nishi, H.; Nakamura, K.; Nakai, H.; Sato, T. Anal. Chem. 1995, 67, 23342341. (13) Agyei, N. M.; Gahm, K. H.; Stalcup, A. M. Anal. Chim. Acta 1995, 307, 185-191. (14) Lurie, I. S.; Klein, R. F. X.; Dal Cason, T. A.; LeBelle, M. J.; Brenneisen, R.; Weinberger, R. E. Anal. Chem. 1994, 66, 4019-4026. (15) Dette, C.; Ebel, S.; Terabe, S. Electrophoresis 1994, 15, 799-803. (16) Chankvetadze, B.; Endresz, G.; Blaschke, G. Electrophoresis 1994, 15, 804807. (17) Wu, W.; Stalcup, A. M. J. Liq. Chromatogr. 1995, 18, 1289-1315. (18) Mayer, S.; Schurig, V. Electrophoresis 1994, 15, 835-841. (19) Terabe, S.; Ozaki, H.; Otsuka, K.; Ando, T. J. Chromatogr. 1985, 332, 211217. (20) Anigbogu, V. C.; Copper, C. L.; Sepaniak, M. J. J. Chromatogr. A 1995, 705, 343-349. (21) Roussel, C.; Favrou, A. J. Chromatogr. A 1995, 704, 67-74. (22) Belder, D.; Schomburg, G. J. Chromatogr. A 1994, 666, 351-365. (23) Schmitt, T.; Engelhardt, H. Chromatographia 1993, 37, 475-481. (24) Gahm, K. H.; Agyei, N. M.; Stalcup, A. M., unpublished results. 0003-2700/96/0368-1360$12.00/0
© 1996 American Chemical Society
the system at pH < 4. Although we have previously reported17 the application of sulfated cyclodextrins (mixtures of randomly substituted sulfated cyclodextrins; degree of substitution (ds), ∼7-10) as CA, here we demonstrate the wider versatility of these sulfated cyclodextrins to a large number of neutral as well as cationic compounds of pharmaceutical interest. EXPERIMENTAL SECTION Materials. The sulfated cyclodextrin (ds 7-10) was obtained from Aldrich Chemical Co. (Milwaukee, WI). The analytes were obtained from either Sigma Chemical Co. (St. Louis, MO) or the U.S. Pharmacopoeia. NaH2PO4 and Na2HPO4 were obtained from Fisher Scientific. The water used to prepare the buffer solutions was distilled and doubly deionized. Buffers of the desired pH were prepared by mixing equimolar solutions of NaH2PO4 and Na2HPO4. Apparatus. The CZE experiments were carried out using a Waters Quanta 4000 capillary zone electrophoresis system equipped with a UV detector (214 nm), interfaced to a Shimadzu Chromatopac CR-501 data station. The CZE system was operated with the cathode at the injector end of the capillary. The fused silica capillary column was 75 µm i.d., with a column length of 60 cm (52.4 cm to detector window). The capillary was cleaned after each run by flushing for 2-3 min with 0.5 M KOH, followed by distilled water. Samples were dissolved in buffer and introduced by hydrostatic injection. RESULTS AND DISCUSSION The results obtained are tabulated in Tables 1 and 2. In our previous reports on the chiral selectivity of polysulfated carbohydrates, stronger analyte/selector interactions were indicated by longer migration times. However, in this study, the situation is reversed. That is, with the reduction or complete lack of electroosmotic flow at low pH and the reversed polarity of our system relative to the earlier study, the positively charged analytes migrate away from the detector, and the neutral analytes are essentially stationary. Thus, analytes should not migrate toward the detector unless they are complexed with the negatively charged CD. Indeed, in the absence of the sulfated cyclodextrin additive, no peaks were observed for any of the analytes used in this study. Hence, the shortest migrations times are generally indicative of the strongest interactions with the polysulfated CD. However, as noted previously,25 strength of complexation does not necessarily guarantee chiral recognition. Analyte or chiral additive structural features which inhibit complexation may actually increase chiral recognition, in this method, by increasing analyte column residency time, thus effectively widening the “chiral window”. Neutral Analytes. Discerning the interactions responsible for chiral recognition may be facilitated by first examining the data for the neutral analytes successfully enantioresolved in this study. As implied previously for these analytes, the principal mechanism for transport is through complexation with the sulfated CD. Thus, the migration times for these analytes are a direct indication of the strength of the analyte/CD association. The data for the neutral analytes are tabulated in Table 1. In general, the migration times for the neutral analytes were longer (tav ) 25 ( 11 min; n ) 16) than those observed for the basic compounds used in this study (tav ) 13 ( 7 min; n ) 38). The longer migration times for (25) Gahm, K. H.; Stalcup, A. M. Anal. Chem. 1995, 67, 19-25.
Figure 1. Plot of resolution vs migration times for the first eluting enantiomers of the neutral analytes: 5-cyclobutyl-5-phenylhydantoin (1); hydrobenzoin (2); R-cyclopropylbenzyl alcohol (3); 5-(4-methylphenyl)-5-phenylhydantoin (4); 5-(4-hydroxyphenyl)-5-phenylhydantoin (5); warfarin (6); 1,1′-binaphthyl-2,2′-diol (7); 2,2-dimethyl-1phenyl-1-propanol (8); 1,1′-binaphthyl-2,2′-diyl hydrogen phosphate (9); 1,2,3,4-tetrahydro-1-naphthol (10); 1,2-diphenyl-2-propanol (11); benzoin (12); 2-phenylcyclohexanone (13); 1-(3-chlorophenyl)ethanol (14); 9-methyl-∆5,10-octalin-1,6-dione (15); and phensuximide (16). CZE conditions: 2% sulfated cyclodextrin; 10 mM phosphate buffer, pH 3.8; 15 kV.
the neutral analytes seem reasonable because, presumably, hydrophobically driven complexation between the analytes and the sulfated CD would lead to weaker interactions than ion-pairing interactions. As can be seen from Figure 1,26 the neutral analytes seemed to be classifiable into three groups. In the case of the compounds which had the lowest overall resolution, although the analytes all contained hydrophobic moieties which fostered inclusion complexation with the CD, most seemed to lack additional functional groups which could further assist in immobilizing or orienting the analytes within the cavity. In addition, these analytes tended to have smaller or fewer hydrophobic moieties, possibly leading to less “tight” inclusion, than the analytes which had intermediate or high resolution. For these neutral analytes, in general, as the migration times increased, enantioresolution also increased. However, several compounds (e.g., 5-cyclobutyl-5-phenylhydantoin, hydrobenzoin, and cyclopropylbenzyl alcohol) had exceptionally high enantioresolution. Comparisons of the results for structurally related compounds reveals some interesting trends. For instance, phensuximide had the longest migration time (t ) 51.12/60.65 min), implying the weakest interactions with the CD of all the analytes used in this study. Comparison of the migration times and enantioresolution obtained for phensuximide with those of other neutral anticonvulsants and glutethimide in Figure 2 demonstrates the ability of the sulfated CD to discriminate not only enantioselectively but also between structurally similar analytes. The single phenyl ring of phensuximide, with no para hydrogen bonding (e.g., 5-(4hydroxyphenyl)-5-phenylhydantoin; t ) 14.47/15.63 min) or hydrophobic moiety (e.g., 5-(4-methylphenyl)-5-phenylhydantoin; t ) 12.88/14.04 min) to assist in immobilizing the analyte within the CD cavity, may partially account for its relatively weak binding. It should also be noted that the structurally related hydantoins (26) The lines are not intended to imply any empirical correlation but are merely included to guide the eye.
Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
1361
Table 1. Electrophoretic Data for Neutral Analytes Using Sulfated Cyclodextrin (2%) as a Chiral Additive (10 mM Phosphate Buffer, pH 3.8; 15 kV) compound
structure
migration time (min)
Rs
CH3
12.88/14.04
5.4
OH
14.47/15.62
5.6
21.30/32.33
26.0
51.12/60.65
8.3
19.55/21.16
2.8
30.31/57.95
26.0
42.72/48.79
6.5
16.33/16.92
2.4
35.41/50.24
11.1
16.66/17.22
2.2
29.49/31.91
4.9
34.24/47.01
15.7
12.50/13.32
2.6
18.42/19.18
2.3
24.15/25.34
2.7
25.20/30.89
8.6
O HN
NH
5-(4-methylphenyl)-5-phenyl-hydantoin O
O HN
NH
5-(4-hydroxyphenyl)-5-phenyl-hydantoin O
O HN
NH
5-cyclobutyl-5-phenylhydantoin O
CH3
phensuximide
O
N
O
O
C6H5 OH
C
C
benzoin
H OH OH
hydrobenzoin
C
C
H
H CH3
O
9-methyl-∆5,10-octalin-1,6-dione O
1,1′-binaphthyl-2,2′-diylhydrogen phosphate
O O
1,1′-binaphthyl-2,2′-diol
O P OH
OH OH
1,2,3,4-tetrahydro-1-naphthol OH Cl
1-(3-chlorophenyl)ethanol
CHCH3 OH CH
R-cyclopropylbenzyl alcohol
OH OH
2,2-dimethyl-1-phenyl-1-propanol
CHC(CH3)3 CH3
1,2-diphenyl-2-propanol
CH2C OH
2-phenylcyclohexanone O O
O
warfarin CHCH2COCH3 OH
1362 Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
C6H5
Table 2. Electrophoretic Data for Cationic Analytes Using Sulfated Cyclodextrin (2%) as a Chiral Additive (10 mM Phosphate Buffer, pH 3.8; 15 kV) compound
structure
migration time (min)
Rs
13.08/13.62
3.0
19.79/21.43
4.0
24.43/25.22
2.6
8.60/8.68
1.0
8.26/8.44
1.9
8.55/8.80
2.3
8.63/8.96
3.2
8.88/8.97
0.6
8.87/9.32
4.1
10.45/11.02
5.0
12.40/13.08
3.6
20.83/22.64
7.4
8.25/8.37
1.3
8.19/8.26
0.8
9.74/9.98
1.7
Anesthetics O
ketamine
NHCH3
Cl CH3
CH3
mepivacaine
N
CONH CH3 CH3
CH2(CH2)2CH3
bupivacaine
N
CONH CH3
Antihistamines CH2CH2N(CH3)2
pheniramine
N
brompheniramine
N
CH CH2CH2N(CH3)2 CH
Br
CH2CH2N(CH3)2 N
chlorpheniramine
CH
Cl
OCH2CH2N(CH3)2 N
carbinoxamine
Cl
CH
OCH2CH2N(CH3)2 N
doxylamine
CH CH3 (CH3)2NCH2CH2 N CH
dimethindene
CH3
Antiarrhythmics CH3O
verapamil CH3O
OCH3
CN C(CH2)3NCH2CH2
OCH3
CH(CH3)2 CH3
mexiletine
CH3
OCH2CHNH2 CH3 CH2CH2N(CH(CH3)2)2
disopyramide
N C CONH2
Anticholinergics OH
trihexyphenidyl
CCH2CH2
N
OH
oxyphencyclimine
N CCOOCH2 N CH3 CH3
mepenzolate
OH
N+
CH3
CCOO
Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
1363
Table 2 (Continued) compound
structure
migration time (min)
Rs
β-Blockers O
piperoxan
N
CH2
8.15/8.61
6.1
10.87/11.12
2.3
11.62/12.12
4.0
26.92/30.98
8.4
30.47/31.23
1.9
O OH
pindolol
OCH2C HCH2NHCH(CH3)2 HN OH OCH2CHCH2NHCH(CH3)2
alprenolol
CH2CH
CH2
OH OCH2CHCH2NHCH(CH3)2
oxprenolol
OCH2CH
CH2
COCH3
acebutolol CH3CH2CH2CONH
OH
OCH2CHCH2NHCH(CH3)2 OH
metoprolol
CH3O(CH2)2
OCH2CHCH2NHCH(CH3)2
9.69
OH
propranolol
OCH2CHCH2NHCH(CH3)2
OH
O
atenolol
10.43
H2NCCH2
OCH2CHCH2NHCH(CH3)2
13.46
Antimalarials N
Cl
chloroquine
10.27/10.58
2.6
10.38/10.75
3.8
NHCH(CH2)3N(C2H5)2 CH3 N
Cl
hydroxychloroquine CH2CH3 NHCH(CH2)3N
CH2CH2OH
CH3 CH3 NHCH(CH2)3NH2
primaquine
N
8.60
slight
8.34/8.56
1.6
8.55/8.69
1.1
10.27/10.64
2.4
10.75/10.91
1.1
CH3O
Antidepressants Cl
CH3
bupropion
COCHNHC(CH3)3
trimipramine N CH2CHCH2N(CH3)2 CH3 NH2
tranylcypromine C6H5 C6H5 O
nefopam N CH3
1364 Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
Table 2 (Continued) compound
structure
migration time (min)
Rs
8.75/9.48
5.8
8.32/8.71
3.4
8.83/9.15
2.2
10.24/10.37
1.0
OCH3
33.97/37.87
8.1
OH
10.58/12.56
5.6
17.16/22.81
12.6
14.00/16.53
4.6
8.01/8.20
1.5
8.37/8.75
2.8
11.35/11.98
3.8
9.26/9.82
3.4
10.00/13.06
16.1
Miscellaneous O
NH
O C2H5
aminoglutethimide
NH2 O O
canadine N
CH3O CH3O
N
idazoxan
NH
O O OH
isoxsuprine
HO
CH3
CHCHNHCHCH2O CH3 OCH3
laudanosine
CH3O CH3O N
CH3 OH
HO
tetrahydropapaveroline HO N OCH3
methoxyphenamine
H
CH3
CH2CHNHCH3 OCH3 OH
midodrine
O
CHCH2NHCCH2NH2 CH3O CH3 C6H5
orphenadine
CHOCH2CH2N(CH3)3 HO
terbutaline
OH CHCH2NHC(CH3)3
HO
tetramisole
N
S N
tolperisone
CH3
O
H
C
C
CH2
N
CH3 N
CH3
Troger’s base CH3
N
all contain two bulky substituents on the stereogenic center, which is incorporated in a ring, whereas phensuximide contains only one substituent. Phensuximide also lacks an amide nitrogen adjacent to the stereogenic carbon to participate in additional hydrogen bonding interactions with the CD. Although aminoglutethimide also lacks an R amide nitrogen, the presence of the primary amine, protonated under the experimental conditions, allows ion-pairing interactions with the CD, as evidenced by its short migration times (t ) 8.75/9.48 min). In the case of
5-cyclobutyl-5-phenylhydantoin, decreased binding to the CD, as indicated by longer migration times, resulted in dramatically enhanced resolution (Rs ) 26.0) relative to 5-(4-hydroxyphenyl)5-phenylhydantoin (Rs ) 5.6) or 5-(4-methylphenyl)-5-phenylhydantoin (Rs ) 5.4). The compound with the second longest migration times of all the analytes studied is 9-methyl-∆5,10-octalin-1,6-dione. Energy minimization calculations using Hyperchem for this compound reveal that the two rings lie in approximately the same plane, with Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
1365
Figure 3. Resolution vs migration times of the first eluting enantiomer in which the stereogenic center is incorporated in a ring. CZE conditions are given in the text.
Figure 2. Electropherogram showing the enantioseparations of aminoglutethimide (1,1′), 5-(4-methylphenyl)-5-phenylhydantoin (2,2′), 5-(4-hydroxyphenyl)-5-phenylhydantoin (3,3′), 5-cyclobutyl-5-phenylhydantoin (4,4′), and phensuximide (5,5′). CZE conditions are given in the text.
the methyl group oriented above the plane. The presence of this methyl group may sterically inhibit inclusion into the CD cavity, and the lack of protons for hydrogen bonding interactions may further weaken the attraction for the sulfated CD; hence, the long migration times. However, as can be seen from the table, the lack of strong interactions does not preclude enantioselective interactions. In the case of benzoin, the presence of the carbonyl decreases the congestion near the stereogenic center and confers additional rigidity on the analyte, thereby affording stronger complexation than is observed with hydrobenzoin. However, hydrobenzoin offers an additional hydroxyl group for hydrogen bonding interactions with the CD, producing much greater enantioresolution than that observed for benzoin. Another interesting comparison can be made between the two atropisomers, 1,1′-binaphthyl-2,2′-diyl hydrogen phosphate and 1,1′-binaphthyl-2,2′-diol. Energy minimization calculations using Hyperchem for these two compounds reveal that in the case of the binaphthol compound, the two aromatic ring systems are almost perpendicular to each other, whereas in the case of 1,1′binaphthyl-2,2′-diyl hydrogen phosphate, the alignment of the aromatic moieties is restricted by the presence of the phosphate bridge. Evidently, the conformational rigidity imposed by the phosphate group and near alignment of the aromatic rings allows for better inclusion into the CD cavity, resulting in shorter migration times but less resolution than that obtained for the binaphthol compound. Interestingly, warfarin was reported to be unenantioresolvable using a CD-based chiral stationary phase (CSP).27 In this previous work, the authors postulated, on the basis of energy minimization calculations, that under optimal complexation, the hydrogen bonding moieties of the warfarin were too far removed from the (27) Armstrong, D. W.; Ward, T. J.; Armstrong, R. D.; Beesley, T. E. Science 1986, 232, 1132-1135.
1366 Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
stereogenic secondary alcohols lining the mouth of the CD cavity for effective enantioselective complexation of warfarin with the native CD. Thus, one possible role of the CD sulfate substituents may be to provide an extension of the chiral cavity. Cationic Compounds. The data for the cationic compounds used in this study are tabulated in Table 2. Presumably, interactions between the sulfated CD and the cationic compounds may arise through a combination of ion-pairing/inclusion complexation. The importance of ion-pairing may be ascertained from the fact that these compounds, in general, had much shorter migration times than the neutral analytes. This is particularly important because these analytes have their electrophoretic mobility vector directed away from the detector. Hence, the binding between the cationic compounds and the sulfated CD is stronger than that with the neutral analytes. The importance of ion-pairing for some compounds may also be indicated in Figure 3, which illustrates the relationship between the resolution and migration times for a series of analytes in which the chiral center is incorporated in a ring. As can be seen from the figure, the resolution decreases and the migration time increases as the bulkiness on the amine functionality increases. Of course, when comparing migration times and resolution for the cationic compounds, it should be noted that the measured net migration times of the analytes are a function of not only the binding constant and the fraction of the analyte that is complexed at any given moment but also the intrinsic electrophoretic mobility of these cationic analytes. It has also been noted that there is an optimum CA concentration for specific analytes.28 Hence, caution must be used when comparing enantioresolution obtained for a series of compounds at a single CA concentration. However, because chiral separations were obtained for such a large number of structurally diverse compounds, no attempt was made in this study to optimize the CA concentration. (i) Anesthetics. The electrophoretic data for the three anesthetics used in this study are tabulated in Table 2. All three anesthetics contain an aromatic and a saturated six-membered ring which includes the chiral center. In all three compounds, the chiral center is juxtaposed between an amine and a carbonyl functionality. The structural analogs, mepivacaine and bupivacaine, contain a tertiary amine functionality which is part of the saturated ring system, while ketamine contains a secondary amine (28) Wren, S. A. C.; Rowe, R. C. J. Chromatogr. 1992, 603, 235-241.
that is external to the ring system. Also, the aromatic rings of bupivacaine and mepivacaine are considerably bulkier, with 2,6dimethyl substituents, than the aromatic ring of ketamine. The migration order of these analytes (ketamine < mepivacaine < bupivacaine) would tend to support an inclusion complexation as an important contribution to solute migration. (ii) Antihistamines. In the case of the antihistamines studied, the chiral center is positioned between two aromatic rings. The migration times (Table 2) for these analytes are fairly similar and indicate fairly strong analyte-CD interactions. Ion-pairing between these analytes and the sulfated CD may occur either through the alkyl amine or through the pyridinium nitrogen. Brompheniramine seems to have the strongest interaction with the sulfated CD. However, as noted previously, care must be taken when trying to correlate migration times with strengths of interaction for charged analytes. Presumably, part of the interaction involves inclusion complexation of the nonheterocyclic aromatic ring into the CD cavity. Armstrong et al.29 reported a larger k′ for bromide ion than for chloride ion on a β-CD column. Thus, the halide substituent may contribute to the formation of an inclusion complex between the analyte and the CD cavity. The lack of a halogen substituent to assist in immobilizing the analyte in the CD cavity may account for the longer migration times of the pheniramine and doxylamine enantiomers, as well as their decreased enantioresolution. (iii) Antiarrythmics. It is interesting to compare the migration time of disopyramide with those of doxylamine and pheniramine, two antihistamines (Table 2). The increased congestion about the chiral center introduced by the amide substituent and the bulkiness resulting from the two isopropyl substituents on the alkyl amine lead to greatly increased migration times but also improved enantioselectivity for disopyramide (t ) 20.83/22.64 min; Rs ) 7.4) relative to pheniramine (t ) 8.60/8.68 min; Rs ) 1.0) and doxylamine (t ) 8.88/8.97 min; Rs ) 0.6). The short migration times and good enantioresolution obtained for verapamil are somewhat surprising, considering the bulkiness of the substituents on both aromatic rings as well as the stereogenic center and the presence of only a single nitrogen moiety, which is somewhat distant from the stereogenic center. (iv) Anticholinergic Agents. In this class of compounds, the stereogenic centers for three of the compounds are substituted with a hydroxyl group, an aromatic ring, and a nonaromatic ring. The slight structural difference between biperiden and trihexyphenidyl (e.g., the nonaromatic ring substituent) does not seem to affect the relative binding of the two compounds with the sulfated CD (as evidenced by their nearly equivalent migration times). However, the reduced enantioselectivity obtained for biperiden relative to trihexyphenidyl is somewhat surprising, given that increased rigidity of the stereogenic center or substituents on the stereogenic center have often been cited as leading to enhanced chiral recognition. Oxphencyclimine has an additional nitrogen, which leads to higher affinity for the CD but with loss of enantioresolution. In the case of mepenzolate, in contrast to the other analytes in this group, the stereogenic center is encompassed in the heterocyclic ring. The congestion around the quartenary amine sterically inhibits the CD/analyte association, as evidenced by the longer migration times, thus leading to (29) Armstrong, D. W.; Alak, A.; Bui, K.; Demond, W.; Ward, T. J. J. Inclusion Phenom. 1984, 2, 533-545.
enhanced chiral recognition relative to the other analytes in this group. (v) β-Blockers. In the case of the β-blockers, with the exception of piperoxan, the stereogenic center consists of a secondary alcohol γ to an aromatic ring and β to an N-isopropyl group. Although both the aromatic and the isopropyl groups may participate in inclusion complexation, the N-isopropyl portion is identical in all of these analytes. Hence, differences in migration and enantioselectivity must be ascribed to other features of the analytes. Armstrong et al.30 noted that many β-blockers are unresolvable by HPLC using a β-CD column in the reversed-phase mode, where interactions are presumably dominated by inclusion complexation. In HPLC, using CD-based phases, the proximity of the stereogenic center to hydrophobic moieties and hydrogen bonding moieties seem to play important roles in chiral selectivity. Interestingly, metoprolol, propranolol, and atenolol were not enantioresolved under the conditions used in this study. Other workers have reported difficulties with chirally resolving propranolol with the native CD.31,32 In this work, metoprolol and propranolol seemed to be the most tightly complexed with the CD, as evidenced by the shortest migration times (Table 2). No doubt the presence of the hydrophobic substituent on the aromatic ring para to the substituent containing the stereogenic carbon enhances complexation for metoprolol. The presence of the ortho substituent seems to inhibit complexation for acebutolol relative to metoprolol, as evidenced by the longer migration times (t ) 30.47/31.23 vs 9.69 min). In the case of atenolol, the relatively hydrophilic amide functionality may reduce the complexation, leading to increased migration times relative to metoprolol, with its relatively hydrophobic alkoxy substituent, but the lack of ortho substitution in both cases leads to no enantioselectivity. In an effort to determine if inhibition of complexation for these three analytes would improve chiral recognition, a small amount of the sulfated CD was methylated, and this new material was successfully able to enantioresolve propranolol, metoprolol, and atenolol. A more complete study with this material is currently being conducted and will be reported elsewhere. It is also interesting to compare the migration times and enantioselectivities obtained for alprenolol and oxprenolol. The addition of a single oxygen between the aromatic ring and the nonstereogenic alkyl ortho substituent on the phenyl ring in the case of oxprenolol results in dramatically increased migration times relative to alprenolol (t ) 26.92/30.98 vs 11.62/12.12 min) and enantioresolution (Rs ) 8.4 vs 4.0). (vi) Antimalarials. In the case of the three antimalarials used in this study, hydroxychloroquine and chloroquine differ only by the presence of a hydroxyl group on the alkyl substituent. Thus, not surprisingly, their migration times are very similar. However, slightly better resolution was obtained for hydroxychloroquine. Virtually no separation was obtained for primaquine. The much shorter migration time obtained for primaquine relative to the other two antimalarials suggests stronger complexation, which may be facilitated by the proximity of the amines on the same side of the aromatic ring system. (vii) Miscellaneous. Of the cationic compounds studied, Troger’s base and methoxyphenamine had the highest enantioresolution (Table 2). A comparison of the migration times and (30) Armstrong, D. W.; Chen, S.; Chang, C.; Chang, S. J. Liq. Chromatogr. 1992, 15, 545-556. (31) Fanali, S. J. Chromatogr. 1991, 545, 437-444. (32) St. Pierre, L. A.; Sentell, K. B. J. Chromatogr. 1994, 657, 291-300.
Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
1367
enantioresolution obtained for laudosine and tetrahydropapaveroline, its exhaustively methylated analog, again highlights the role of analyte bulkiness inhibiting complexation but enhancing chiral recognition by effectively widening the chiral window. However, comparison of piperoxan and idazoxan (Table 2) reveals that piperoxan exhibited higher enantioresolution (Rs ) 6.1) than idazoxan (Rs ) 2.2), despite shorter migration times (t ) 8.15/ 8.61 vs 8.83/9.15 min).
seen from the data, a large number of structurallydiverse analytes, neutral as well as cationic, were successfully enantioresolved.
CONCLUSIONS The combination of sulfated cyclodextrins, limited electroosmotic flow, and operation of the CZE instrument with the anode at the detector end of the column provides a very versatile method for the enantioresolution of a large variety of analytes. As can be
Received for review August 1, 1995. Accepted November 27, 1995.X
1368
Analytical Chemistry, Vol. 68, No. 8, April 15, 1996
ACKNOWLEDGMENT The authors gratefully acknowledge the generous support of the National Institutes of Health (1R29 GM48180-03) and helpful discussions with Professor K. B. Sentell (University of Vermont), Professor J. P. Foley (Villanova University), and Professor T. J. Ward (Millsaps College).
AC950764A X
Abstract published in Advance ACS Abstracts, March 1, 1996.
