Anal. Chem. 1999, 71, 1252-1256
Articles
Amino Acid Order in Polymeric Dipeptide Surfactants: Effect on Physical Properties and Enantioselectivity Eugene Billiot, Rezik A. Agbaria, Stefan Thibodeaux, Shahab Shamsi, and Isiah M. Warner*
Department of Chemistry, Louisiana State University, Baton Rouge, Louisiana 70803
The effect of amino acid order on chiral selectivity in polymeric dipeptide surfactants, as well as the physical properties of the surfactants, is investigated. An understanding of enantioselectivity of such dipeptide surfactants is crucial to the design of more efficient polymeric surfactants and has implications in other areas of research such as enantioselective interactions of amino acid based compounds (i.e., enzymes, hemoglobin, antibodies, etc.). It should be noted that such polymeric surfactants are not easily crystallized. Therefore, in a manner similar to the study of proteins, fluorescence spectroscopy is a powerful tool used to study the structure-function relationship of these polymeric surfactants. The microenvironments inside the core of 18 polymeric surfactants were characterized using the environmentally sensitive probes pyrene and 6-propionyl-2-(dimethylamino)naphthalene (Prodan). The surfactants examined in this study include all possible dipeptide combinations of the L-form of alanine, valine, and leucine and the achiral amino acid glycine (except glycine-glycine) as well as the single amino acid surfactants of alanine, valine, and leucine. The results of the fluorescent probe studies led to a proposed structure of the polymeric dipeptide surfactants in solution. The implications of the proposed structure for chiral selectivity were tested with two model atropisomers, (()1,1′-bi-2naphthol and (()1,1′-bi-2-naphthyl-2,2′-diyl hydrogen phosphate, using capillary electrokinetic chromatography. In 1994, Wang and Warner1 reported the use of a synthetic polymeric chiral surfactant, poly(sodium undecyl-L-valine) (poly L-SUV), for chiral separations in capillary electrokinetic chromatography (EKC). Polymeric chiral surfactants have several potential advantages over normal micelles for chiral separations in EKC. One such advantage is that polymeric surfactants do not have a critical micelle concentration (cmc). The lack of a cmc means that the micelle polymer can be used as a pseudostationary phase over a wider range of concentrations than the unpolymerized form. This is particularly important when the optimum concentration of the (1) Wang, J.; Warner, I. M. Anal. Chem. 1994, 66, 3773-3776.
1252 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999
surfactant is below the cmc of the micelle.2 In addition, higher concentrations of organic modifiers can be used without disrupting the micelle.3 Finally, the structural rigidity and purification of the polymer is believed to improve the mass-transfer rate, thus reducing peak broadening. Since 1994, several other papers have been published exploring the potential of polymeric chiral surfactants as pseudostationary phases for EKC. In a subsequent study by Wang and Warner,4 a synergistic effect was observed for the chiral separation of four enantiomeric pairs with poly D-SUV in combination with γ-cyclodextrin. Additional research by our group5 and by Dobashi et al.6 extended the range of analytes separated with poly L-SUV. Another study compared the polymeric single amino acid surfactant poly L-SUV to the polymeric dipeptide surfactant poly(sodium undecyl L-valine-valine) (poly L-SUVV).7 The polymeric dipeptide surfactant poly L-SUVV showed a significant improvement in chiral selectivity for three out of the four analytes examined, as compared to the single amino acid surfactant poly L-SUV. Recently, the effect of amino acid order in polymeric dipeptide surfactants on chiral separation was examined.2 The primary surfactants used in that investigation were poly(sodium undecyl L-valine-leucine) (poly L-SUVL) and poly(sodium undecyl L-leucine-valine) (poly L-SULV). The results of that research showed that the amino acid order can have a dramatic effect on chiral selectivity. In the present paper, we report further advances toward our goal of understanding the effects of amino acid order in polymeric dipeptide surfactants through the examination of a larger group of homologue amino acid surfactants. All possible dipeptide combinations of the L-form of alanine, valine, and leucine and the achiral amino acid glycine (except glycine-glycine) were synthesized, and their functionality for chiral recognition and chiral separations was investigated. In addition to the dipeptide surfactants, the single amino acid surfactants of alanine, valine, and leucine were also examined. (2) Billiot, E.; Macossay, J.; Thibodeaux, S.; Shamsi, S.; Warner I. M. Anal. Chem. 1998, 70, 1375-1381. (3) Shamsi, S. A.; Akbay, C.; Warner, I. M., submitted to Anal. Chem. (4) Wang, J.; Warner, I. M. J. Chromatogr., A 1995, 711, 297-304. (5) Agnew-Heard, K.; Sanchez Pena, M.; Shamsi, S.; Warner I. M. Anal. Chem. 1997, 69, 958-964. (6) Dobashi, A.; Hamada, M.; Dobashi, Y. Anal. Chem. 1995, 67, 3011-3017. (7) Shamsi, S.; Macossay, J.; Warner, I. M. Anal. Chem. 1997, 69, 2980-2987. 10.1021/ac980461r CCC: $18.00
© 1999 American Chemical Society Published on Web 02/24/1999
Fluorescent probe studies were conducted in order to better understand the interactions of the analytes with these new polymeric surfactants and to better correlate structure with function. The fluorescent probes 6-propionyl-2-(dimethylamino)naphthalene (Prodan) and pyrene were used to characterize the hydrophobicity of the microenvironment of the surfactant core. The characterization of these polymeric surfactants by fluorescence spectroscopy led to a proposed structure of the dipeptide surfactants in solution. There are two major implications of the proposed structure as far as chiral recognition is concerned for large bulky hydrophobic analytes. First, if the larger of the two amino acids in the dipeptide surfactant is in the first (N-terminal) position, then the second (C-terminal) amino acid can act as a “finger” to help hold the analyte, restricting its movement. Second, if the larger amino acid is in the second position, it could block access to the first chiral center, resulting in a significant decrease in the chiral selectivity of the surfactant. To test the above structural theory, EKC experiments were performed. It was postulated that if we examined a large group of homologue dipeptide surfactants, then the best arrangement of the amino acids for the enantiomeric separation of large bulky analytes should be with the larger of the two amino acids in the first (N-terminal) position. To examine this hypothesis, the enantiomers of (()1,1′-bi-2-naphthol (BOH) and (()1,1′-bi-2naphthyl-2,2′-diyl hydrogen phosphate (BNP) were selected for separation using equivalent monomer concentrations (EMCs) of the polymeric surfactants. The results from the EKC experiments support the proposed structural theory. EXPERIMENTAL SECTION Synthesis of Polymeric Surfactants. All surfactants in this study were synthesized using the procedure reported by Wang and Warner.1 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. Polymerization was achieved by 60Co γ-irradiation. Purification of the polymers was achieved by dialysis using a 2000 molecular weight cutoff cellulose membrane. All monomers and polymers used in this study were found to be 99% pure or better as estimated from elemental analysis. Materials. The pyrene and various enantiomers of BOH and BNP were purchased from Aldrich (Milwaukee, WI). The tris(hydroxymethyl)aminomethane (TRIS) was ordered from Fisher Scientific Co. (Fair Lawn, NJ). N-Hydroxysuccinimide, undecylenic acid, and the various amino acids and dipeptides were acquired from Sigma (St. Louis, MO). The Prodan was purchased from Molecular Probes (Eugene, OR). All compounds were used as received, except pyrene which was recrystallized twice from ethanol. Capillary Electrophoresis Procedure. The EKC experiments were conducted on a Hewlett-Packard3D CE model G1600AX instrument. An untreated fused-silica capillary (effective length 55 cm, 50 µm i.d.) was purchased from Polymicro Technologies (Phoenix, AZ). The background electrolyte (BGE) for all EKC experiments was 100 mM TRIS and 10 mM borate at pH 10.0. An appropriate percentage (w/v) of the polymeric surfactants was then added to the BGE and the pH readjusted with 1 N NaOH or 1 N HCl if necessary. After adjusting the pH, the solution was
filtered through a 0.45-µm membrane filter. 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. All analytes were prepared in 50:50 methanol/water at 0.1 mg/mL. 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 triply distilled deionized water. Prior to each run, the buffer was pressure injected through the column for 2 min to condition and fill the capillary. Fluorescence Procedure. Steady-state fluorescence measurements were acquired on a Spex model P2T 211 spectrofluorometer. Samples were measured in 1-cm2 quartz cells. All measurements were performed at ambient room temperature. The Prodan samples were excited at 390 nm and emission spectra measured from 400 to 600 nm. The pyrene samples were excited at 335 nm and emission spectra recorded from 360 to 450 nm. Excitation and emission slit widths of 8.6 and 1.7 nm were respectively employed. Preparation of Samples. Stock solutions of 1.0 × 10-3 M of pyrene and 1.0 × 10-3 M Prodan were each prepared using spectroquality grade cyclohexane. Surfactant stock solutions were prepared by adding 90 mg of surfactant to 15 mL of triply distilled deionized water. The samples were prepared by adding 100 µL of the stock probe solutions to 10-mL bottles. The cyclohexane was evaporated under dry nitrogen. A 5-mL aliquot of the stock surfactant solution was transferred to the bottle containing the probe residue. The samples were allowed to equilibrate for at least 24 h before analysis. RESULTS AND DISCUSSION Proposed Structure of Dipeptide Surfactants. The results of our studies suggest that the order of the amino acids in dipeptide surfactants has a pronounced effect on the physical characteristics of the surfactant. For example, the amino acid order significantly affects the hydrophobicity of the surfactant core. More importantly, the amino acid order dramatically affects the chiral recognition ability of the polymeric dipeptide surfactant. It is proposed here that the lowest energy configuration of dipeptide surfactants in solution is when the larger of the R-groups, i.e., the most hydrophobic group, is inside facing the core of the polymeric surfactant. The proposed structure of the dipeptide surfactants is shown in Figure 1. The backbones of these dipeptide surfactants are the same. The only difference between the surfactants is the size of the R-groups. The final conformation of the dipeptide in the polymeric surfactant is governed by two major effects which are involved in determining the minimum energy configuration. First, one would expect the two hydrophobic groups, R1 and R2, to face the inner core of the polymeric surfactant structure rather than be exposed to the bulk aqueous phase. However, because of the packed configuration of the dipeptide surfactant, this preferred conformation is unlikely to occur due to steric hindrance. Therefore, the smallest R-group would be forced to twist toward the aqueous phase. It is not possible at this stage to determine the extent of such a twist of the small R-group. However, such a twist toward the water phase is expected to bring the adjacent carbonyl group closer to the inner core. Consequently, the inner core becomes more polar as Analytical Chemistry, Vol. 71, No. 7, April 1, 1999
1253
Table 1. Hydrophobicity Data of Various Polymeric Surfactants Prodanb pyrene I/IIIa Prodan λmax, nm surfactant pyrene I/III (normalized data) (normalized data) alanine valine leucine Gly-Ala Gly-Val Gly-Leu Ala-Gly Ala-Ala Ala-Val Ala-Leu Val-Gly Val-Ala Val-Val Val-Leu Leu-Gly Leu-Ala Leu-Val Leu-Leu
Figure 1. Proposed structure of dipeptide surfactants.
Figure 2. Proposed interaction between of BOH with poly L-SUAL and poly L-SULA.
a result of closer proximity of the carbonyl group. Second, another structural implication occurs if the larger amino acid is in the second position. In such a case, the large bulky R-group could limit access of a large bulky analyte to interact with the first chiral center attached to R1. The implications of the proposed structure can be better illustrated by a comparison of the proposed structures of the dipeptide surfactants of alanine-leucine [poly (L,L)-SUAL] and leucine-alanine [poly (L,L)-SULA] and their possible interactions with BOH (Figure 2). When leucine (the larger of the two amino acids) is in the first position (Figure 2b), it is believed that the R-group of alanine is directed away from the hydrophobic core and more toward the aqueous phase. In this configuration, BOH can interact more with the heteroatoms on the alanine, thus restricting the movement of BOH and thereby enhancing the chiral selectivity of the surfactant. Conversely, if the larger amino acid (leucine) is in the second position (Figure 2a), it could block access to the chiral center attached to alanine, thereby reducing the chiral selectivity of the dipeptide surfactant. 1254 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999
1.098 ( 0.008 1.033 ( 0.006 1.013 ( 0.002 1.029 ( 0.001 0.997 ( 0.002 0.958 ( 0.003 1.059 ( 0.002 1.037 ( 0.001 1.041 ( 0.002 1.012 ( 0.004 0.959 ( 0.003 1.051 ( 0.003 0.986 ( 0.002 0.975 ( 0.002 0.959 ( 0.005 0.949 ( 0.002 0.945 ( 0.005 0.962 ( 0.002
0.861 ( 0.006 0.915 ( 0.005 0.933 ( 0.002 0.918 ( 0.001 0.948 ( 0.002 0.986 ( 0.003 0.892 ( 0.002 0.911 ( 0.001 0.908 ( 0.002 0.934 ( 0.004 0.985 ( 0.003 0.899 ( 0.003 0.958 ( 0.002 0.969 ( 0.002 0.985 ( 0.005 0.996 ( 0.002 1.000 ( 0.005 0.982 ( 0.002
493.0 ( 1.0 490.3 ( 0.6 490.0 ( 0.0 489.0 ( 1.0 484.6 ( 0.6 482.3 ( 0.6 490.6 ( 0.6 488.0 ( 0.0 489.3 ( 0.6 487.6 ( 0.6 485.3 ( 0.6 489.6 ( 0.6 485.6 ( 0.6 485.6 ( 0.6 486.3 ( 0.6 486.6 ( 0.6 482.0 ( 0.0 481.6 ( 0.6
0.809 ( 0.002 0.849 ( 0.001 0.853 ( 0.000 0.868 ( 0.002 0.943 ( 0.001 0.988 ( 0.001 0.844 ( 0.001 0.884 ( 0.000 0.864 ( 0.001 0.891 ( 0.001 0.930 ( 0.001 0.859 ( 0.001 0.925 ( 0.001 0.925 ( 0.001 0.913 ( 0.001 0.908 ( 0.001 0.994 ( 0.00 1.000 ( 0.001
a I/III of polymer divided by I/III of Leu-Val. b 1 + [(λ max of polymer - λmax of Leu-Leu)/λmax of Leu-Leu].
Fluorescent Probe Study. To better understand the role of the R-groups and the carbonyls in the dipeptide surfactants, fluorescent probe studies were conducted. The fluorescent probes used in this study were Prodan and pyrene. Prodan was first introduced by Weber and Farris8 in 1979. Prodan has both an electron-donor (amino) group and an electron-withdrawing (carbonyl) group on opposite sides of the naphthalene moiety. A large excited-state dipole moment and extensive solvent polaritydependent fluorescence shift occurs due to localization of charges on opposite sides of the naphthalene moiety upon excitation. The increase in solvent polarity results in shifts of the emission spectrum to longer wavelengths. For example, the emission maximum of Prodan shifts from 392 nm in cyclohexane to 523 nm in water. The fluorescence emission spectrum of pyrene, the other fluorescent probe used in this study, has five major vibrational bands. The ratio of the I to III vibronic bands at the respective wavelengths of 372 and 383 nm has been shown to depend on solvent polarity.9 For example, the III/I ratio of pyrene in water and cyclohexane is reported to be 0.63 and 1.68, respectively.10 In this paper, pyrene and prodan were used to compare the polarity/hydrophobicity of the microenvironment of the polymeric surfactant core. Table 1 lists the I/III ratios of pyrene and the λmax observed for Prodan. The table also lists the normalized data. These data were normalized in order to be able to compare all data on the same scale. A comparison of the hydrophobicity trends for various surfactants is shown in Figure 3. As observed in Figure 3a, the core of poly L-SUA is the least hydrophobic of the three single amino acid surfactants, followed by poly L-SUV and then poly L-SUL. This is the trend that would be expected in going from a less hydrophobic amino acid such as alanine to the more hydrophobic amino acids valine and leucine. The same trend is observed when glycine is (8) Weber, G.; Farris, F. J. Biochemistry 1979, 18, 3075-3078. (9) Nakajima, A. Bull. Chem. Soc. Jpn. 1971, 44, 3272-3277. (10) Kalyanasundaram, K.; Thomas, J. K. J. Am. Chem. Soc. 1997, 99, 2039.
Figure 3. Comparing the hydrophobicity of various polymeric dipeptide surfactants.
held constant in the first (N-terminal) position of the dipeptide surfactant and the size of the amino acid in the second position is increased (Figure 3b). Note that when valine or leucine is kept constant in the second position of the dipeptide surfactant, and the amino acid in the first position is increased in size from left to right (i.e., alanine to valine to leucine), the hydrophobicity again follows expected trends Figure 3c,d). Interestingly, if glycine is present in the first position of the dipeptide surfactant, an increase in hydrophobicity is observed as compared to the intuitive decrease that one would expect in going from a less hydrophobic compound, poly L-SUGV, to a more hydrophobic compound, poly (L,L)-SUAV (Figure 3c). The same unexpected behavior is observed in Figure 3d. The hydrophobicity of the microenvironment of the core of poly L-SUGL is greater than poly (L,L)-SUAL and poly (L,L)-SUVL and about the same as poly (L,L)-SULL. The reason for this apparent anomaly is that the probe experiences only the microenvironment inside the core of the polymeric surfactant. It is important to note that, because of the relatively packed interior of the covalently linked polymeric surfactant, these analytes cannot penetrate as deeply into the core as compared to noncovalently linked micelles. Therefore, due to steric effects, these analytes will partition closer to the bulk aqueous phase with polymeric surfactants as compared to regular micelles. This distinction is important in understanding the trends observed. The probes are believed to partition near polar headgroups rather than deep within the core, away from the chiral centers. Thus, the environment that the probe experiences is the same environment that the chiral analytes examined in this study are likely to experience. Indeed, this is true because it has been well established in the literature that pyrene partitions near the polar headgroup of normal micelles rather than penetrating deep into the inner core. Also, since the chiral analytes examined in this study can be enantiomerically separated with these surfactants, then they too must partition near the polar headgroups; otherwise no chiral separation would be observed. As discussed earlier and shown in Figure 1, the proposed conformation of these dipeptide surfactants is such that the larger of the two R-groups associated with the two amino acids in these dipeptide surfactants would be facing the inside of the polymeric surfactant core while the smaller, less hydrophobic R-group would be facing the outside water surface. The R-group facing the water
Figure 4. Effect of amino acid order in dipeptide surfactants on the chiral resolution of BOH. EKC conditions: applied voltage +30 kV, buffer solution prepared with 100 mM TRIS and 10 mM sodium borate at pH 10.0, column temperature 25 °C, and 5 mM each of the polymeric surfactants at equivalent monomer concentrations.
layer would then have minimal effect on the hydrophobicity of the polymeric surfactant core. In the case of poly L-SUGV and poly L-SUAV, the larger of the two R-groups would be the R-group on valine and it would be facing the core of the polymeric surfactant. With poly L-SUGV, there is no competition between the two R-groups since the R-group on glycine is a hydrogen. Therefore, the carbonyl adjacent to R1 is free to rotate and face the aqueous phase without the steric hindrance associated with the competing R-group (R1). When the carbonyl adjacent to R1 faces the aqueous phase, the polarity of the surfactant core decreases, thus increasing the hydrophobicity of the core. Capillary Electrophoresis Study. As stated earlier, it is proposed that if the more hydrophobic bulky R-group of the polymeric dipeptide surfactants is in the second position, it could limit access of a bulky analyte to the first chiral center attached to R1. It was speculated that this orientation of the surfactant in solution would cause a decrease in chiral selectivity of the surfactant to that analyte. Two large bulky chiral compounds (BNP, BOH) were selected to test this hypothesis. In this study, the resolution of the BNP and BOH were compared at EMC of all the surfactants, (i.e., 15 mM). The EMCs were used because the molecular weights of these polymers are not known at this time. Furthermore, at EMCs it is assumed that the average number of repeat units of the polymeric surfactants is approximately the same as the aggregation number of the monomers, or at least close enough in value to make the comparison by EMCs valid. Preliminary molecular weight studies using analytical ultracentrifugation support this assumption.11 The average number of repeat units for various polymeric surfactants ranged from approximately 32 ( 2 to 37 ( 2 for the dipeptide surfactants and 31.5 ( 2 for the single amino acid surfactants. A comparison of the enantiomeric resolution of BOH with various polymeric dipeptide surfactants is shown in Figure 4. These surfactants are grouped together as pairs. The first polymeric dipeptide surfactant in each pair always has the smaller of the two amino acids in the first position and the second surfactant has the larger of the two amino acids in the first (11) Yarabe, H.; Billiot, E.; Cush, R.; Russo, P.; Shamsi, S.; Warner, I. M., manuscript in preparation.
Analytical Chemistry, Vol. 71, No. 7, April 1, 1999
1255
for two of the polymeric dipeptide surfactants (i.e., poly L-SUGV and poly L-SUGL). However, with the larger amino acid in the first position, nearly twice the value of baseline separation was achieved for poly L-SUVG and more than twice the value of baseline separation was achieved in the other cases. The only exception to this trend was found with the first pair of surfactants, poly L-SUGA and poly L-SUAG (Figure 5a). Both poly L-SUGA and poly L-SUAG were not able to separate the enantiomers of BNP.
Figure 5. Effect of amino acid order in dipeptide surfactants on the chiral resolution of BNP. (EKC conditions same as Figure 4).
position. It was speculated that the preferred order of the amino acids in the dipeptide for chiral recognition of large bulky analytes would be with the larger of the two amino acids in the first position (i.e., the second dipeptide surfactant in the paired grouping). As can be seen in this figure, the prediction holds true for the separation of the enantiomers of BOH. In fact, baseline resolution (defined as resolution of at least 1.5) was not achieved for any of the surfactants when the smaller of the two amino acids is in the first position, except poly (L,L)-SUVL. In contrast, when the larger of the two amino acids is in the first position, resolution values of 4 or better are achieved. Similar trends are also observed in comparing the resolution of the enantiomers of BNP (Figure 5). When the smaller amino acid is in the first position, baseline separation was only obtained
1256 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999
CONCLUSIONS This research shows conclusively that the order of amino acids in polymeric dipeptide surfactants does have an effect on the physical characteristics of the surfactant, e.g., hydrophobicity of the surfactant core. More importantly, the amino acid order has a major effect on the chiral selectivity of the surfactant. Such considerations of the dipeptide conformations have proven useful in explaining all spectroscopic and EKC data and provide a fundamental understanding of the molecular recognition principles of these polymeric dipeptide surfactants. Further studies using high-resolution NMR are planned in order to further evaluate the proposed structure. ACKNOWLEDGMENT This work was supported by a grant from the National Institute of Health (GM39844). I. M.W. also acknowledges the Philip W. West endowment for partial support of this research. Received for review April 28, 1998. Accepted December 2, 1998. AC980461R
