Chiral Separations Using Dipeptide Polymerized Surfactants: Effect of

Kevin F. Morris, Bridget A. Becker, Bertha C. Valle, Isiah M. Warner, and Cynthia K. ... Eugene Billiot and Isiah M. Warner .... Gerald Gübitz , Mart...
0 downloads 0 Views 96KB Size
Download PDF
Anal. Chem. 1998, 70, 1375-1381

Chiral Separations Using Dipeptide Polymerized Surfactants: Effect of Amino Acid Order Eugene Billiot, Javier Macossay, Stefan Thibodeaux, Shahab A. Shamsi, and Isiah M. Warner*

Department of Chemistry, Louisiana State University, Baton Rouge, Louisiana 70803

Chiral separations using various polymerized dipeptide surfactants in electrokinetic capillary chromatography (EKC) are investigated. The two main dipeptide surfactants used in this study were sodium N-undecylenyl-Lvaline-L-leucine (L-SUVL), and sodium N-undecylenyl-Lleucine-L-valine (L-SULV). These studies were performed in order to determine if the order of amino acids in dipeptide surfactants is important in terms of chiral recognition and separations. Both the monomer and the polymer of these two surfactants were compared for the separation of two model atropisomers, (()-1,1′-bi-2naphthol (BOH) and (()-1,1′-bi-2-naphthyl-2,2′-diyl hydrogen phosphate (BNP). Some advantages and disadvantages of the polymer relative to the monomer are discussed. Four other surfactants, the polymers of sodium N-undecylenyl-L-leucine-L-leucine (L-SULL), sodium N-undecylenyl-L-valine-L-valine (L-SUVV), sodium N-undecylenyl-L-valine (L-SUV), and sodium N-undecylenyl-Lleucine (L-SUL), were also used in this study, and their performance was compared to that of poly(L-SULV). These data show conclusively that the order of amino acids in dipeptide surfactants has a dramatic effect on chiral recognition. Our investigations indicate that poly(L-SULV) provides the best enantioselectivity among the four dipeptide and two single amino acid surfactants for the separation of BNP and BOH. The advantages of poly(L-SULV) are demonstrated via the ultrafast separation of the enantiomers of BNP and BOH in less than 1 min. Separation of individual enantiomers in a racemic mixture is still one of the most challenging tasks facing separation scientists. This is true despite the fact that Louis Pasteur first successfully separated two optical isomers by use of a microscope and a pair of tweezers more than a century ago.1 Since then, several techniques have been developed for the separation of chiral compounds. One of the more recent methods employed for chiral separations is micellar electrokinetic capillary chromatography (MEKC). The MEKC technique was first introduced by Terabe et al. in 1984.2,3 They used MEKC to extend the range of achiral analytes to be separated with capillary electrophoresis to include neutral compounds. The first successful chiral separation using (1) Pasteur, L. C. R. Hebd. Seances Anad. Sci. (Paris) 1853, 37, 162. (2) Terabe, S.; Otsuka, K.; Ichikawa, K.; Ando, T. Anal. Chem. 1984, 56, 111113. (3) Terabe, S.; Otsuka, K.; Ando, T. Anal. Chem. 1985, 57, 834-841. S0003-2700(97)00956-6 CCC: $15.00 Published on Web 02/26/1998

© 1998 American Chemical Society

the MEKC method was performed by Cohen et al.,4 who used N,N-dodecyl-L-alanine and sodium dodecyl sulfate (SDS) along with a metal ion (Cu2+) to form a mixed-micelle chiral ligand. Several different types of chiral surfactants have been used for enantiomeric separations in MEKC. These include some naturally occurring surfactants such as bile salts,5-10 digitonin,11 glycyrrhizic acid,12 and β-escin12 as well as octyl maltopyranoside.13 Although there are a wide variety of synthetic surfactants available for use in separation science, few are chiral. One type of synthetic chiral surfactant contains a sugar derivative14-16 at the polar end, and another one of the most recently reported is derived from 6-aminopenicillanic acid.17 The most widely used artificial chiral surfactants contain amino acid derivatives (acyl amino acids11,18-26 and alkoxyacyl amino acids27-32) as polar headgroups. (4) Cohen, S. A.; Paulus, A.; Karger. B. L. Chromatographia 1987, 24, 15-24. (5) Cole, R. O.; Sepaniak, M. J.; Hinze, W. L.; Gorse, J. J. Chromatogr. A 1991, 557, 113-123. (6) Cole, R. O.; Sepaniak, M. J.; Hinze, W. L. J. High Resolut. Chromatogr. 1990, 13, 579-582. (7) Cole, R. O.; Sepaniak, M. J. LC-GC 1992, 10, 380-385. (8) Nishi, H.; Fukuyama, T.; Matsuo, M.; Terabe, S. J. Chromatogr. A 1990, 515, 233-243. (9) Nishi, H.; Fukuyama, T.; Matsuo, M.; Terabe, S. Anal. Chim. Acta 1990, 236, 281-286. (10) Nishi, H.; Fukuyama, T.; Matsuo, M.; Terabe, S. J. Microcolumn Sep. 1989, 1, 234-241. (11) Otsuka, K.; Terabe, S. J. Chromatogr. A 1990, 515, 221-226. (12) Ishihama, Y.; Terabe, S. J. Liq. Chromatogr. 1993, 16, 933-944. (13) Mechref, Y.; El Rassi, Z. Chirality 1996, 8, 518-524. (14) Tickle, D.; Okafu, G.; Camilleri, P.; Jones, R.; Kirby, A. Anal. Chem. 1994, 66, 4121-4126. (15) Mechref, Y.; El Rassi, Z. Electrophoresis 1997, 18, 912-918. (16) Mechref, Y.; El Rassi, Z. J. Chromatogr., A 1997, 757, 263-273. (17) Bouzige, M.; Okafo, G.; Dhanak, D.; Camilleri, P. Chem. Commun. 1996, 5, 671-672. (18) Dobashi, A.; Ono, T.; Hara, S.; Yamaguchi, J. Anal. Chem. 1989, 61, 19841986. (19) Dobashi, A.; Ono, T.; Hara, S.; Yamaguchi, J. J. Chromatogr. A 1989, 480, 413-420. (20) Otsuka, K.; Terabe, S. J. Chromatogr. A 1991, 559, 209-214. (21) Otsuka, K.; Terabe, S. Electrophoresis 1990, 11, 982-984. (22) Otsuka, K.; Kashihara M.; Kawaguchi, Y.; Koike, R.; Hisamitsu, T.; Terabe, S. J. Chromatogr. A 1993, 652, 253-257. (23) Wang, J.; Warner, I. M. Anal. Chem. 1994, 66, 3773-3776. (24) Wang, J.; Warner, I. M. J. Chromatogr. A 1995, 711, 297-304. (25) Dobashi, A.; Hamada, M.; Dobashi, Y. Anal. Chem. 1995, 67, 3011-3017. (26) Agnew-Heard, K. A.; Sanchez Pena, M.; Shamsi, S. A. Anal. Chem. 1997, 69, 958-964. (27) Mazzeo, J. R.; Grover, E. G.; Swartz, M. E.; Peterson, J. S. J. Chromatogr. A 1994, 680, 125-135. (28) Mazzeo, J. R.; Grover, E. G.; Swartz, M. E. Anal. Chem. 1995, 67, 29662973. (29) Swartz, M. E.; Mazzeo, J. R.; Grover, E. G.; Brown, P. R. J. Chromatogr. A 1996, 735, 303-310. (30) Hove, E. V.; Sandra, P. J. Liq. Chromatogr. B 1995, 18, 3675-3683.

Analytical Chemistry, Vol. 70, No. 7, April 1, 1998 1375

In 1994, Wang and Warner24 reported the use of a synthetic chiral polymerized surfactant for chiral separations. This polymer was synthesized from the amino acid L-valine and undecylenic acid. Such polymers have a zero critical micelle concentration (cmc) since the surfactants are polymerized. The separation mechanism in this technique is electrokinetic chromatography (EKC), which is similar to micellar electrokinetic chromatography (MEKC) except that the micelles are replaced with polymerized surfactants. According to Wang and Warner, polymerized surfactants have certain distinct advantages over conventional micelles in chiral separations. First, the elimination of the dynamic equilibrium between monomer and micelle may enhance chiral recognition, leading to better enantiomeric resolution. Second, the lack of a cmc means that the polymer can be used over a wider range of concentrations than the monomer. In addition, organic modifiers can also be used without disrupting the formation of the micelle. Finally, the structural rigidity and purification of the polymer should improve the mass transfer rate, thus reducing peak broadening. It is well-known that a chiral selector’s size and shape as well as the geometric arrangement of its functional groups help to determine its enantioselectivity.33 Many studies have varied the functional groups and even the positions of the functional groups on chiral selectors such as cyclodextrins. However, to the best of our knowledge, no one has outlined studies to date involving the effect of amino acid order on chiral separation. In fact, there have been no studies reported on the use of dipeptide surfactants in chiral separation before our initial studies in this area.34 This is interesting, since amino acid surfactants have been used for chiral separations for quite a few years. In this study, we report on the use of novel dipeptide surfactants for chiral separations. The main purpose of this study was to investigate the effect of the order of the amino acids in the dipeptide surfactants on chiral separation. The two main dipeptide surfactants under study were poly(sodium N-undecylenyl-L-valine-L-leucine) [poly(L-SUVL)] and poly(sodium N-undecylenyl-L-leucine-L-valine) [poly (L-SULV)]. Enantiomeric separations of two atropisomers, (()-1,1′-bi-2naphthol (BOH) and (()-1,1′-bi-2-naphthyl-2,2′-diyl hydrogen phosphate (BNP), were compared using the aforementioned polymerized surfactants. Additionally, their similarities with and differences from other related polymerized surfactants, i.e., sodium N-undecylenyl-L-leucine-L-leucine (L-SULL), sodium N-undecylenylL-valine-L-valine (L-SUVV), sodium N-undecylenyl-L-valine (L-SUV), and sodium N-undecylenyl-L-leucine (L-SUL), are discussed. EXPERIMENTAL SECTION Synthesis of Polymerized Surfactants. All surfactants in this study were synthesized using the procedure reported by Wang and Warner.23 Surfactant monomers were prepared by mixing the N-hydroxysuccinimide ester of undecylenic acid with the amino acid or dipeptide to form the corresponding N-undecylenyl chiral surfactant. The critical micelle concentration (cmc) of the surfactants was determined by use of surface tension measure(31) Peterson, A. G.; Ahuja, E. S.; Foley, J. P. J. Chromatogr. B 1996, 683, 1528. (32) Peterson, A. G.; Foley, J. P. J. Chromatogr. B 1997, 695, 131-146. (33) Gasper, M.; Berthod, A.; Nair, U.; Armstrong, D. Anal. Chem. 1996, 68, 2501-2514. (34) Shamsi, S. A.; Macossay, J.; Warner, I. M. Anal. Chem. 1997, 69, 29802987.

1376 Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

ments. Polymerization was achieved by 60Co γ-irradiation. Purification of the polymers was achieved by dialysis using a 2000 Da molecular mass cutoff cellulose membrane. The average number of monomer units per polymer of the surfactants used in this study was estimated to be 30-37. These numbers were calculated by use of the average molecular weights, which were determined by ultracentrifugation. All monomers and polymers used in this study were found to be 99% pure or better, as estimated from elemental analysis. Materials. (()-1,1′-Bi-2-naphthol (BOH), (-)-1,1′-bi-2-naphthol, (+)-1,1′-bi-2-naphthol, (()-1,1′-bi-2-naphthyl-2,2′-diyl hydrogen phosphate (BNP), (-)-1,1′-bi-2-naphthyl-2,2′-diyl hydrogen phosphate, and (+)-1,1′-bi-2-naphthyl-2,2′-diyl hydrogen phosphate were purchased from Aldrich (Milwaukee, WI). Tris(hydroxymethyl)aminomethane (TRIS) was purchased from Fisher Scientific Co. (Fair Lawn, NJ). N-Hydroxysuccinimide, undecylenic acid, valine, leucine, valine-valine, leucine-leucine, valine-leucine, and leucinevaline were purchased from Sigma (St. Louis, MO). All amino acids and dipeptides used in this study were in the L-form. These items were used as received. Choice and Preparation of Buffer. The background electrolyte (BGE) for all EKC experiments was 100 mM TRIS at pH 10.5. An appropriate percentage (w/v) of the polymerized surfactants was then added to the BGE, and the pH was readjusted with 1 N NaOH or 1 N HCl if necessary. The buffer TRIS was chosen because its low mobility would more closely match that of the analytes chosen, as compared to more conventional buffers such as borate and phosphate. The low mobility of TRIS also allows higher concentrations of buffer to be used without significantly increasing the current. In addition, the low current allows the use of higher voltages, thus yielding shorter retention times. The relatively high ionic strength of the buffer leads to sharper, more defined peaks. The pH of 10.5 was chosen because previous work23,34 in our laboratory had determined that binaphthyl derivatives are separated best at pH ∼10. It is important to note here that, although TRIS worked well in this system at pH, 10.5, TRIS is not normally used at this pH, since the pKa of TRIS is around 8; therefore, pH 10.5 is outside the normal range of buffering capacity. Capillary Electrophoresis Procedure. The EKC experiments were conducted on a Hewlett-Packard 3D CE model G1600AX. An untreated fused silica capillary (effective length 55 cm, 50 mm i.d.) was purchased from Polymicro Technologies (Phoenix, AZ). The surfactants were added to the buffer solution, and the solution was filtered through a 0.45-µm membrane filter. The analytes were prepared in a 50:50 methanol/water mixture at 0.1 mg/mL. The sample was pressure injected for 2 s with 25 mbar of pressure. Separations were performed at +30 kV, with UV detection at 215 nm. The temperature of the capillary was maintained at 25 °C by the instrument thermostating system, which consisted of a Peltier element for forced air cooling and temperature control. Prior to use, the new capillary was conditioned for 30 min with 1 N NaOH, followed by 30 min of 0.1 N NaOH. The capillary was then rinsed for 15 min with deionized water. Prior to each run, the buffer was pressure injected through the column for 2 min to condition and fill the capillary.

Figure 1. Comparison of poly(L-SULV) and poly(L-SUVL). (a, +) BNP with poly(L-SULV), (b, O) BOH with poly(L-SULV), (c, 0) BOH with poly(L-SUVL), and (d, ]) BNP with poly(L-SUVL).

RESULTS AND DISCUSSION Comparison of L-SUVL and L-SULV. (a) Separation of BNP and BOH with Poly(L-SULV) and Poly(L-SUVL). In our initial EKC study, the separations of 1,1′-bi-2-naphthyl-2,2′-diyl hydrogen phosphate (BNP) and 1,1′-bi-2-naphthol (BOH) were compared using two different dipeptide surfactants, poly(L-SULV) and poly(L-SUVL). In poly(L-SULV), valine is the C-terminal amino acid, while it is the first (N-terminal) amino acid in poly(L-SUVL). Figure 1 illustrates the variation in the enantiomeric separation of BNP and BOH as a function of poly(L-SULV) and poly(L-SUVL) concentrations. As can be seen in Figure 1, the difference in chiral recognition of BNP between poly(L-SULV) and poly (L-SUVL) was very dramatic. The maximum resolution achievable with poly(LSUVL) was less than 1, while poly(L-SULV) was able to resolve BNP with a resolution of almost 8 under the same conditions. While the difference in chiral selectivity observed for BOH was not as dramatic, Rs ≈ 2.5 for L-SUVL and Rs ≈ 6 for L-SULV, it is still very significant. In addition, it is interesting to note that the optimum concentration of the polymer appears to be analyte dependent, i.e., analyte resolution is dependent on polymer. In contrast, the optimum concentration of polymer for a given analyte appears to be independent of the polymer. For example, the optimum polymer concentration for BOH is approximately 0.6% (w/v), and the optimum concentration of surfactants for BNP is approximately 3% (w/v). (b) Comparison of the Monomers and Polymers of L-SULV and L-SUVL for the Separation of BNP and BOH. Several interesting differences are observed when one compares the separation performance of the polymers of L-SUVL and L-SULV to that of the monomers. One distinct advantage of the polymer over the monomer is illustrated in the separation of BOH. As can be seen from Figure 2, the monomers of L-SULV and L-SUVL show an expected drop in resolution as the concentration of monomer is decreased and the cmc is approached, the cmc being approximately 10 mM, which is slightly less than 1% (w/v). The polymers, on the other hand, show a significant increase in resolution at concentrations below the normal cmc of the monomers. The optimum resolutions of BOH achieved with the monomers of L-SULV and L-SUVL were respectively Rs < 2 and Rs < 1, while the optimum resolutions with the polymers were

Figure 2. Comparison of the polymers of L-SULV and L-SUVL to the monomers for the separation of BOH. (a, [) Poly(L-SULV), (b, 0) poly(L-SUVL), (c, 9) the monomer of L-SULV, and (d, 4) the monomer of L-SUVL.

Figure 3. Comparison of the polymers of L-SULV and L-SUVL to the monomers for the separation of BNP. (a, +) Poly(L-SULV), (b, 4) the monomer of L-SULV, (c, 3) the monomer of L-SUVL, and (d, ]) poly(L-SUVL).

respectively Rs ≈ 6 and Rs ≈ 3. The polymers were able to resolve BOH approximately 3 times better than the monomers. These data demonstrate a very important advantage of the polymerized surfactant over the monomeric micelles. Since the polymers do not have a cmc, they can be effective at concentrations below which the monomeric surfactants do not form micelles and are thus no longer capable of chiral separations. This is particularly important in this case, since the optimum concentration of polymeric surfactant is below the cmc of the monomer. In fact, many of our studies to date have shown that the optimum concentration of polymer is often below the normal cmc of the monomers. In a comparison of the polymers to the monomers for the separation of BNP, an interesting difference is observed for L-SUVL. As can be seen in Figure 3, both the monomer and the polymer of L-SULV result in approximately the same separation of BNP. The monomer for L-SUVL, however, provided better separation of BNP than did the polymer. Another batch of polymer was synthesized, and it performed in the same manner. The reason for this behavior is not clear at this time. Of the many chiral polymeric surfactants that we have studied, this is our first Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

1377

Figure 4. Comparison of poly(L-SUV) and poly(L-SUL) to poly(L-SULV). (A) For BNP: (a) poly(L-SULV), (b) poly(L-SUL), and (c) poly(L-SUV). (B) For BOH: (a) poly(L-SULV), (b) poly(L-SUL), and (c) poly(L-SUV).

observation of a monomeric surfactant system providing better chiral separation than the corresponding polymer. More studies are underway to try to understand the behavior of this particular polymer. These include high-resolution 1H NMR studies to get more detailed information concerning surfactant-analyte interactions. Additionally, the hydrophobicity, microviscosity, and polarity of the microenvironment of the core of the surfactants will be examined by use of fluorescent probes. Comparison of the Polymers of L-SULV to L-SUL and L-SUV. In an attempt to better understand why poly(L-SULV) provided better separation for BOH and BNP than poly(L-SUVL), poly(L-SUV) and poly(sodium N-undecylenyl-L-leucine) [poly(LSUL)] were studied. This was done to examine the hypothesis that possibly the valine or the leucine was responsible for the observed improvement in chiral resolution, depending on how far the analyte penetrates into the core of the polymerized surfactant. It is believed that, if either of these two surfactants showed separations comparable to that of poly(L-SULV), then the differences in chiral separations might be due to analyte interaction with one of the chiral centers rather than some type of synergism of the two chiral centers. As can be seen from Figure 4, poly(L-SULV) was able to separate BOH better than either poly(L-SUL) or poly(L-SUV). While the differences in resolution were not dramatic, Rs ≈ 6 for poly(L-SULV) and Rs ≈ 4 and 3.5 for poly(L-SUL) and poly(L-SUV), respectively, the differences were significant. However, the differences were not significant enough to allow us to draw any real conclusions about whether the observed improvement in chiral separation was due to interaction of the analyte with one of the chiral centers or, rather, some type of synergism of the two chiral centers on the dipeptide surfactant. A very significant difference, however, is observed in the enantioseparation of BNP with poly(L-SULV) as compared to those with poly(L-SUL) and poly(L-SUV). The maximum resolution for BNP was less than unity for poly(L-SUV) and poly(L-SUL), while poly(L-SULV) was able to resolve BNP with a resolution of approximately 8. The 1378 Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

Figure 5. Separation of BNP with 1% (w/v) of various polymerized surfactants. (a) Poly(L-SULV), (b) poly(L-SUL), (c) poly(L-SUVL), and (d) poly(L-SUV).

differences in resolving power of the various surfactants are clearly demonstrated in Figure 5, where the surfactant poly(L-SULV) was able to separate BNP in less than 6 min with a resolution of 5.2, with polymer concentration at 1% (w/v). In contrast, the other surfactants, poly(L-SUVL), poly(L-SUL), and poly(L-SUV), were unable to adequately separate BNP under the conditions used. From these data, it is reasonable to conclude that the observed improved chiral separation is due to some form of synergism between the two chiral centers or some type of steric effects of the dipeptide as compared to the single amino acid surfactant. Comparison of the Polymers of L-SULV to L-SUVV and L-SULL. To further investigate the hypothesis that the improvement in chiral selectivity was due to a synergistic effect of the two chiral centers or some type of steric factors of the dipeptide surfactants compared to the single amino acid surfactants, dipeptides with the same amino acids, i.e., poly(sodium N-undecylenylL-leucine-L-leucine) [poly(L-SULL)] and poly(L-SUVV), were studied. We desired to determine if a combination of either of those

Figure 6. Comparison of poly(L-SUVV) and poly(L-SULL) to poly(L-SULV). (A) For BOH: (a) poly(L-SULV), (b) poly(L-SUVV), and (c) poly(L-SULL). (B) For BNP: (a) poly(L-SULV), (b) poly(L-SULL), and (c) poly(L-SUVV).

two amino acids would show results comparable to those for poly( L-SULV). The polymer of L-SULV was again observed to perform better than either poly(L-SULL) or poly(L-SUVV) in the separation of BNP and BOH, (Figure 6). Poly(L-SULV) was able to resolve BOH with a resolution of about 6, while poly(L-SUVV) and poly(L-SULL) had resolutions of ∼3 and ∼2.2, respectively. The resolution achieved for BNP with the polymers of L-SULV, L-SUVV, and L-SULL were approximately 8, 2, and 4, respectively. In comparing the separations of BNP and BOH with the polymers of L-SULL and L-SUVV, it should be noted that, while poly(L-SUVV) was able to resolve BOH better than poly(L-SULL), the opposite trend is seen with BNP. Poly(L-SULL) separated BNP with a resolution of about 4, and poly L-SUVV was only able to provide a resolution of approximately 2. It is also interesting to note that, in a comparison of the polymers of the dipeptide surfactants L-SULL and L-SUVV to the polymers of the single amino acid surfactants L-SUL and L-SUV, the order of effectiveness of the surfactants in the separation of BNP and BOH seems to follow opposite trends. The bulkier surfactants poly(L-SULL) and poly(L-SUVV) separated BNP better than the less bulky, less sterically hindered, single amino acid surfactants. Poly(L-SUL) and poly(L-SUV), however, separated BOH better than the dipeptide surfactants poly(L-SULL) and poly(L-SUVV). The difference is not as great for BOH as it is for BNP, but there does seem to be a definite trend. It appears that the separation of BNP is favored by an increase in steric factors, while this same increase in steric factors decreases the resolution of BOH. Analysis of our data suggests that two different mechanisms are involved in the interaction of BNP and BOH with the chiral centers of these surfactants. At pH 10.5 (experimental conditions), BNP is completely anionic, while BOH is not completely ionized. The first pKa of BOH is about 9.5; thus, BOH is only partially ionized at pH 10.5. Experiments were performed at pH 12 (buffered with 50 mM CAPS, +30 kV applied voltage), where BOH would be completely ionized. Although the order of effectiveness of the surfactants for the separation of BNP and BOH remained the same, the optimum concentration of surfactant for BOH shifted

Figure 7. Comparison of optimum resolution for various surfactants.

to higher concentrations. The concentration versus resolution curve then became similar to that of BNP (figure not shown). It is not clear at this time what the major factors are that contribute to the two different mechanisms. Further studies are planned to investigate this observed behavior. Comparison of Optimum Resolutions and k′ at Optimum Resolution for the Various Polymeric Surfactants. As stated earlier, the order of effectiveness of the surfactants for the separation of these two compounds follows almost opposite trends. Although L-SULV is by far the best surfactant for the separation of BNP and BOH, the other surfactants used in this study show different results for these two compounds. In a comparison of all of the polymerized surfactants under study (Figure 7), the first two surfactants in the chart (L-SULV and L-SULL) provide better separation of BNP than BOH. The other surfactants, poly(LSUVV), poly(L-SUVL), poly(L-SUL), and poly(L-SUV), separated BOH better than BNP. Furthermore, as the resolution of BNP decreases, there is a relative increase in the resolution of BOH, with the exception of poly(L-SULV), which separates both compounds better than any of the other surfactants examined in this Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

1379

Figure 8. Comparison of k′ values at optimum resolution.

study. This trend can be better seen from the inset in Figure 7, where the ratios of the resolutions of BNP/BOH decrease in the following order: L-SULL > L-SUVV > L-SUVL > L-SUV > L-SUL. This trend suggests, as stated earlier, that two different mechanisms are involved in the interaction of BNP and BOH with the chiral centers of these surfactants. One possible reason for the improvement in selectivity of the chiral analytes could be differences in the binding of the analytes BNP and BOH to the various surfactants. The major factors involved in binding of analytes to the surfactants used in this study are hydrophobic-hydrophilic interactions, hydrogen bonding, and steric factors, which would either decrease or increase the binding of the analyte to the surfactant. The steric factors would include elements such as the size of the R-group attached to the chiral carbon of the amino acid and the configuration of the surfactant in solution. Configurational differences of the surfactants could serve to increase or decrease the flexibility of the surfactant core or increase or decrease the hydrogen-bonding ability of the analyte to the surfactant. To determine if a difference in binding is responsible for the differences in chiral selectivity observed, the “optimum” capacity factors (k′) were compared. The optimum k′ values are the k′ values at the concentration of surfactant which yielded optimum resolution. The optimum concentration of surfactant was approximately 3% (w/v) for BNP and about 0.6% (w/v) for BOH. The optimum k′ values, as seen in Figure 8, seem to be approximately the same for all the surfactants that gave adequate separation of analyte. The three surfactants that did not adequately resolve BNP were poly(L-SUVL), poly(L-SUL), and poly(L-SUV). As observed in Figure 8, the k′ values for these surfactants were significantly higher than the k′ values of the surfactants [i.e., poly(L-SULV), poly(L-SULL), and poly(L-SUVV)] that did adequately resolve BNP. The average optimum k′ values for those surfactants that yielded adequate separation was approximately the same, within experimental error, for both analytes, 1.3 ( 0.1 for BNP and 1.1 ( 0.2 for BOH. It should be noted that the k’s for BNP, for the surfactants poly(L-SUVL), poly(L-SUL), and poly(L-SUV), were not used to calculate the average optimum k′ since these surfactants did not adequately separate BNP. These surfactants gave a resolution of less than 1 for BNP. The increase in k′ values for those surfactants that did not adequately resolve BNP further supports the hypothesis that an increase in steric factors is responsible for the improvement in resolution of BNP. The larger k′ values of those surfactants that 1380 Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

Figure 9. Separation of BNP and BOH with 1% (w/v) poly(L-SULV) using short method. Same conditions as in Figure 6, except negative polarity (-30 kV) was used and injection was done at the detector end, making the effective capillary length 8.5 cm.

did not adequately resolve BNP suggest that BNP binds more strongly to these surfactants than to the other surfactants that did adequately resolve BNP. Since all the dipeptide surfactants in this study have approximately the same number of carbons, the hydrophobicity of the surfactant core can be assumed to be approximately the same for all the dipeptide surfactants in this study. Furthermore, since the numbers of heteroatoms available for hydrogen bonding are the same for all of the dipeptide surfactants, the major difference in binding is likely due to some type of steric factors that either would block the analyte from entering the core of the surfactant or would increase the binding or flexibility of the surfactant core. Since it is assumed that hydrophobic-hydrophilic interactions of all these surfactants are approximately the same, it is reasonable to conclude that the differences in binding must, then, be due to steric factors. At this time, it is unclear what these steric factors are. What is obvious, however, is that the separation of BNP is favored by the bulky dipeptide surfactants, while BOH is separated better by the less bulky, less sterically hindered, single amino acid surfactants. The one exception to this is the surfactant L-SULV, which separates both analytes better than any of the other surfactants examined in this study. Finally, due to the high selectivity of poly(L-SULV), it was possible to achieve baseline separation of BNP and BOH in less than 1 min (Figure 9). This was achieved by using 1% (w/v) polymer with reverse polarity and injecting the sample at the detector end, making the effective length of the capillary only 8.5 cm. These separations were done with enantiomeric excess of the R-form of BOH and BNP in order to determine the elution order of the enantiomers. The ultrafast separation achieved in this study is important because shorter analysis times mean higher sample throughput. This increase in sample throughput translates into increased laboratory efficiency. CONCLUSIONS This research shows conclusively that the order of amino acids in dipeptide surfactants can have a major effect on chiral recognition of analytes. The advantages of the polymer over the

monomer are clearly shown in the separation of BOH, where the optimum concentration of the surfactant is below the cmc of the monomer. One interesting exception to the superiority of the polymer over the monomer is seen with the surfactant L-SUVL, where the monomer was able to separate BNP better than the polymer. It was also shown that an increase in chiral recognition was not due to the interaction of the chiral center of the analyte with one of the chiral centers on the dipeptide surfactant, but rather some form of synergism of the dipeptide as compared to the single amino acid surfactant. Analysis of our data further suggests that two different mechanisms are involved in chiral recognition of BNP and BOH. An increase in steric factors favors the separation of BNP, while the resolution of BOH decreases with an increase in steric factors. The exception to this general trend is the surfactant L-SULV. Finally, the superiority of the polymer over the monomer is also demonstrated with baseline separation of BNP and BOH with 1% (w/v) poly(L-SULV) in less than 1 min. Such separations are not possible under these conditions with the monomer, since the cmc of these surfactants is about 1% (w/v), and the separation of BOH and BNP drops off rapidly as the cmc is approached.

Further studies are planned with a variety of other types of chiral analytes and various other dipeptide surfactants. A systematic study is presently underway in which all possible combinations of the L form of the three amino acids, alanine (A), valine (V), and leucine (L), in dipeptide surfactants are evaluated. Examination of the dipeptides from the nine possible combinations of these amino acids (A-A, A-V, A-L, V-A, V-V, V-L, L-A, L-V, and L-L) should aid in an understanding of the role of amino acid order in dipeptide surfactants for chiral recognition. Additionally, dipeptides with a single chiral center will be examined to determine what role the second chiral center plays in chiral recognition of the dipeptide surfactants. ACKNOWLEDGMENT This work was supported through a grant from the National Institute of Health (GM 39844). I.M.W. also acknowledges the Philip W. West endowment for partial support of this work. Received for review September 2, 1997. Accepted January 23, 1998. AC9709561

Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

1381