Circular dichroism, ultraviolet, and proton nuclear magnetic resonance

Pietro Bortolus, Giancarlo Marconi, and Sandra Monti , Bernd Mayer. The Journal of ... R. J. Tait , D. O. Thompson , V. J. Stella , and J. F. Stobaugh...
0 downloads 0 Views 989KB Size
Download PDF
Anal. Chem. 1992, 64, 1405-1412

1405

Circular Dichroism, Ultraviolet, and Proton Nuclear Magnetic Resonance Spectroscopic Studies of the Chiral Recognition Mechanism of P-Cyclodextrin Song Li and William C. Purdy' Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Quebec, H3A 2K6,Canada

The chlral recognltlon mechanism of 8-cyclodextrln was studled by UV-vlslble, circular dlchrolsm, and proton nuclear magnetic resonance spectroscoplc methods. The D and L enantiomers of DNP-vallne, DNP-leuclne, and DNP-methlonh e were used as model solutes. The results Indicatethat the dlnltrophenyl group, whlch forms stable lncluslon complexes wlth the &cyclodextrln cavlty and places the other functional groups around the chlral center In association wlth the hydroxyl groups at the edge of the cavlty, plays a very Important role In the chlral recognltlon. The alkyl groups of amino aclds, whlch form a secondary lncluslon complex wlth another 8cyclodextrln cavlty (In the case of DNP-L-aminoaclds) or are sterlcally repulsed by the hydroxyl groups at the edge of the cavity (In the case of DNP-mamlno aclds), are also major contrlbutorsto the chlral recognltlonprocess. The dlssoclatlon constants of the lncluslon complexes of 8-cyclodextrln wlth these model compounds were also obtained from the changes of UV absorbance, ellptlclty, and chemlcalshlfts, respectlvely. It was found that the DNP-L-amino aclds always have smaller dlssoclatlon constants than the menantlomers.

INTRODUCTION In recent years, cyclodextrins have received a great deal of attention as chiral mobile-phase additives1 or as chiral stationaryphases2Bfor direct enantiomeric separations. The tack of most investigators has been to expand their chiral separation spectrum as well as to improve the enantioselectivity by chemical modifications. Although a number of empirical and theoretical studies about the chiral recognition mechanism have been done, relatively little experimental effort has been focused on the explanation of the nature of the chiral discrimination interaction and the chirality forces responsible for the different retention of enantiomers. Enantioselective inclusion complex formation is considered to provide the essential discrimination in cyclodextrinstationary phases. However, very little direct experimental evidence can be found in the literature. As mentioned elsewhere, cyclodextrins are capable of forming complexes in aqueous solution with a variety of molecular species.4 The complexes are usually regarded as inclusion compounds in which hydrogen b ~ n d i n g ,and ~ , ~van ~~

* To whom correspondence should be addressed.

(1) Sybilska, D. In Ordered Media in Chemical Separation; Hinze, W. L., Armstrong, D. W., Eds.; American Chemical Society: Washington, DC, 1987; Chapter 12, p 219. (2) Hinze, W. L.; Riehl, T. E.; Armstrong, D. W.; Demond, W.; Alak, A.; Ward, T. Anal. Chem. 1985,57, 237-242. (3) Armstrong, D. W.; Yang, X.; Hand, S. M.; Menges, R. A. Anal. Chem. 1987,59, 2594-2596. (4) Bender, M. L.; Komiyama, M. Cyclodertrin Chemistry; Springer-Verlag: New York, 1978; p 10. (5) Cramer, F. Rev. Pure Appl. Chem. 1955, 5 , 143-164. (6) Cramer, F.; Dietsche, W. Chem. Ber. 1959, 92, 1739-1747. 0003-2700/92/0364-1405$03.00/0

der Waals' forces7,8are the main binding forces. Inclusion complex formation has been studied by a variety of methods, such as electron-spinresonance (ESR) spectroscopy?JOproton nuclear magnetic resonance (lH NMR) s p e ~ t r o s c o p y , ~ ~ - ~ ~ ultraviolet (UV) and visible spe~troscopy,~~-~6 circular dichroism (CD) spectroscopy or optical rotatory dispersion (ORD),l7-20 Raman spectroscopy,21fluorescence spectroscopy,22@ X-ray analy~is,24-2~ potentiometry,28 positron annihilation,29 and thermoanalytical methods.30 Among these methods, lH NMR spectroscopy, UV-visible spectroscopy, and the CD method are considered to be the most popular. Since lH NMR spectroscopy was first used for the study of adduct formation in aqueous solution, NMR studies of cyclodextrin complexes with aromatic compounds have been made by several re~earchers.~~331*32 In the structure of &cyclodextrin, only the 3' and 5' protons are located inside the cavity. The 3' protons form a ring near the larger opening of the cavity, while the 5'protons are near the smaller opening. The 6' protons are located on the upper surface and directed inward in the gg conformation. All other protons are located on the exterior of the cavity. It has been found that when an aromatic moiety of the guest molecule is included in the cavity of 0-cyclodextrin, protons (3' and 5' ) located within the cavity are susceptible to anisotropic shielding of the (7) VanEtten, R. L.; Sabastian, J. F.; Clowes, G. A,; Bender, M. L. J . Am. Chem. SOC. 1967,89, 3242-3253. (8) Bergeron, R.; Channing, M. A.; Gibeily, G. J.; Pillor, D. M. J. Am. Chem. SOC.1977, 99, 5146-5151. (9) Flohr. K.: Paton. R. M.: Kaiser. E. T. J. Am. Chem. SOC. 1975.97, 1209-1218. (10) Martinie, J.; Michon, J.; Rassat, A. J. Am. Chem. SOC.1975,97, 1818-1823. (11) Wood, D. J.; Hruska, F. E.; Saenger, W. J. Am. Chem. SOC.1977, 99,1735-1740. (12) MacNicol, D. D. Tetrahedron Lett. 1975,38, 3325-3326. (13) Inoue. Y.: Hoshi.. H.:. Sakurai.. M.:. Chuio. " . R. J . Am. Chem. SOC. 1~8;5,i07,2319-2323. (14) Bergeron, R. J. J. J. Chem. Educ. 1977,54, 204-207. (15) Ikeda, K.; Uekama, K.; Otagiri, M. Chem. Pharm. Bull. 1975,23, 201-208. (16) Hoffman, J. L.; Brock, R. M. Biochemistry 1970,9, 3542-3550. (17) Shimizu, H.; Kaito, A.; Hatano, M. Bull. Chem. SOC. Jpn. 1979, 52, 2678-2684. (18) Shimizu, H.; Kaito, A.; Hatano, M. Bull. Chem. SOC. Jpn. 1981, 54, 513-519. (19) Harata, K. Bioorg. Chem. 1981,10, 255-265. (20) Harata, K.; Uedaira, H. Bull. Chem. SOC.Jpn. 1975,48,375-378. (21) Sato, H.; Higuchi, S.; Teramae, N.; Tanaka, S. Chem. Lett. 1979 299-300. 1967,89, (22) Cramer, F.; Saenger, W.; Spatz, H. J . Am. Chem. SOC. 14-20. (23) Seliskar, C. J.; Brand, L. Science 1971, 171, 799-800. (24) Harata, K. Bull. Chem. SOC. Jpn. 1980, 53, 2782-2786. (25) Harata, K. Bull. Chem. SOC. Jpn. 1976, 49, 2066-2072. (26) Hingertz, B.; Saenger, W. Nature 1975,255, 396-397. (27) Noltemeger, M.; Saenger, W. J. Am. Chem. SOC.1980,102,27102722. (28) Gelb, R. I.; Schwartz, L. M.; Cardelino, B.; Fuhrman, H. S.;Johnson, R. F.; Laufer, D. A. J . Am. Chem. SOC.1981, 103, 1750-1757. (29) Hall, E. S.; Ache, H. J. J . Phys. Chem. 1979, 83, 1805-1807. (30) Sztatisz, J.; Gal, S.; Komives,J.; Stadler-Szoke, A,;Szejtli, J. Therm. Anal. 1980,2, 487-493. (31) Thakker, A. L.; Demarco, P. V. J. Pharm. Sci. 1971,60,652-653. (32) Bergeron, R.; Rowan, R. Bioorg. Chem. 1976, 5, 425-436. 0 1992 American Chemical Society

1406

ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992

aromatic moiety, while protons located on the exterior of the 8-cyclodextrin cavity, such as l', 2', and 4', are relatively unaffected. The 6' protons are also affected, though to a lesser extent. Alternatively, if association takes place at the exterior of the cavity, the protons on the exterior of the cavity will be significantly changed. It is now believed that 'HNMR spectra can give the most direct evidence for the inclusion complex formation between guest molecules and the cyclodextrin cavity in aqueous solution. It is known that the formation of inclusion complexes with cyclodextrins leads to the changes of UV-visible spectra of a variety of organic molecules ranging from organic dyes33 such as marine blue, methyl orange, and crystal violet to aromatic compounds such as phenol, naphthol, and substituted phenolic compounds. The stoichiometries of the inclusion complexes can be found from the UV-visible spectra in spectrophotometric titration, and the formation constants of inclusion complexes can be obtained from the absorption changes. Circular dichroic spectroscopy of active compounds is also a powerful tool for the study of the three-dimensional structures of organic molecules.34%5 This technique provides information on the absolute configuration, conformation, reaction mechanism, etc. It is shown that circular dichroic spectroscopy is also a powerful method for investigating the cyclodextrin-guest interaction, since the guest chromophore perturbed by the cyclodextrin molecule produces the induced circular dichroism spectrum. lB Both formation constants of inclusion complexes and the geometry of the host-guest interactions can be evaluated from the induced CD spectra. In this paper, we report our efforts to employ highresolution NMR, UV-visible, and circular dichroism spectroscopies as probes for the study of the host-guest interactions which lead to chiral recognition using DNP-amino acids as model compounds. The aim of this work is to show that these DNP-amino acids actually form enantioselective inclusion complexes with P-cyclodextrin in the aqueous solution and to characterize the chiral separation mechanism by UV-visible, CD, and 'H NMR spectrometric measurements. Furthermore, we wish to elucidate the mode of inclusion and the structure of the complex from the induced CD bands and NMR spectra. The induced CD bands, UV absorption changes, and NMR chemical shift changes were quantitatively investigated to obtain the dissociation constants. EXPERIMENTAL SECTION Materials. P-Cyclodextrin was purchased from Aldrich Chemical Co. and purified by recrystallization once from propanol followed by recrystallization once from water and drying at 80 OC with PzOsin vacuo for 12 h. The specificrotatory power of j3-cyclodextrinwas [a]% = 162.0& 0.5. Fluoro-2,4-dinitrobenzene (FDNB), DNP-L-valine,DNP-L-leucine,and other D- and L-amino acids were from Sigma Chemical Co. DNP-D-valine, DNP-D-leucine and DNP-D-and DNP-L-methionine were prepared by modifying the method described by Schroeder and Legette.36 Preparation of DNP-Amino Acids. A 20-mL aliquot of 1.0 M NaHC03solution is transferred to a 150-mLglass-stoppered flask followed by additions of a magnetic stirring bar, 0.01 mol of amino acid (solid), and 20 mL of FDNB-ethanol solution (containing 2.0 g of FDNB). The mixture is then stirred at room (33) Lautsch, W.; Broser, W.; Biedermann, W.; Gnichtel, H. J.Polym. Sci. 1955, 17, 479-510. (34) Velluz, L.; Legrand, M.; Grosjean, M. Optical Circular Dichroism,Principles,Measurements,and Applications;Verlag Chemie: Weinheim, 1965; p 79-200. (35) Snatzke, G. Optical Rotatory Dispersion and Circular Dichrioism in Organic Chemistry; Heyden & Son Ltd.: London, 1967; p 16. (36) Schroeder, W.A.;Legette, J. J. Am. Chem. Sac. 1953, 75,46124615.

temperature for 3 h. After evaporation of the ethanol, the residue is neutralized to pH 9.0 by 6 N HC1. This solution is then transferred to a 200-mL separation funnel with 20 mL of water. The solution is extracted six times with 30 mL of ether to remove the excess FDNB. After removalof the excess FDNB,the aqueous phase is adjusted to pH 1.5. DNP-amino acids are extracted with ether and dried for identification and application. For DNP-D-valine,the product is 1.5 g (yield 53%1. 'H NMR (inD20,pH = 11.0,0.05phosphoratebuffer): 8.96(3H,aromatic), 8.10 ( 5 H, aromatic), 6.78 (6 H, aromatic), 3.90 (a-CH-), 2.15 (P-CH-), 0.90 (CH3). UVmaxima (inO.O1Mphosphorate buffer, pH = 6.0): 264.4 and 361.8 nm. For DNP-D-leucine,the product is 1.3 g (yield 44 % ). 'H NMR: 8.97 (3 H, aromatic), 8.11 (5 H, aromatic), 6.75 (6 H, aromatic), 4.10 (a-CH-), 3.18 (P-CHz-), 1.70 (Y-CH-), 0.81 (4H3). UV maxima: 263.1 and 362.8 nm. For DNP-D-methionineand DNP-L-methionine,the products are 1.5 g (yield 47%) and 1.4 g (yield 44%),respectively. Both enantiomers have the same NMR, UV, and IR spectra at the measuring conditions. lH NMR (DzO): 8.89 (3 H, aromatic), 8.10 (5 H, aromatic), 6.76 (6 H, aromaticj), 4.18 (a-CH-), 2.08 (P-CHz-), 2.46 (SCHz-), and 1.92 (-CH3). UV maxima: 264.1 and 361.5 nm. CD,UV, and NMR Measurements. CD measurements were carried out in a 0.1 M sodium phosphate buffer, pH = 6.0, on a Jasco-500C spectropolarimeter(Japan SpectrosccopicCD,Japan) with a JASCO DP-J800/30 data processor. A 1-cm-thickquartz cell was used to hold the sample solution. The CD spectra in the range of 600-185 nm were measured at the following set of operating parameters: sampling wavelength, 1 nm; scan speed, 100 nmimin; time constant, 2 s; and temperature, 21 f 0.5 OC. The recording of each spectrum was repeated four times, and the averaged spectra were obtained on the data processor. NMR spectra were measured by a XL-200 spectrometer (Varian Canada Ltd., Montreal) at ambient probe temperature of 20 h 1"C. The sample was prepared in a 0.05 M phosphorate buffer solution. Since the solubilities of DNP-amino acids at pH = 6.0 are not high enough for NMR measurement, a buffer solution of pH = 11.0was used. The buffer solution was prepared with anhydrous Na3P04 in DzO and the pD was adjusted with deuterium chloride acid. The pD value was obtained by adding 0.4 to the pH meter reading,37using a glass electrode which had been calibrated with pH = 7.00 and pH = 10.00 buffers in water and the rinsed with D20. No internal 1H NMR reference was added since the possibility of reference binding to the 8-cyclodextrin could not be excluded. A solvent line (4.63 of Dz0) was used as the reference. The UV-visible spectra were measured on a DMS-300 UVvisible spectrophotometer (Varian Canada, Montreal) in a 0.1 M sodium phosphate buffer solution of pH = 6.0. The sample compartment contained a 1-cm-thick quartz cell with combined 0-cyclodextrinand DNP-amino acid solution while the reference compartment contained the same concentration of j3-cyclodextrin as the sample cell but with no DNP-amino acid. Determination of Dissociation Constants by UV-Visible Spectra. The dissociation constants, &, for j3-cyclodextrinDNP-amino acid complexes were measured according to the conventional Scott equation (1)3* where [Go] is the total

concentration of DNP-aminoacids, [Ho]is the total concentration of B-cyclodextrin, A6 is the difference of the molar absorptivities for free and complexed DNP- amino acids, and AAbs is the change in absorbance by the addition of 8-cyclodextrin. The data were treated by plotting [HoI[GoI/AA~~ vs ([Hol + [Go]),providing a slope of l/Ac and a intercept of &/A€. In order to meet the requirement for a linear plot, i.e. [Hol[Gol >> C2 (C is the concentration of host-guest complex),the concentration of DNPamino acid is held at least 10 times lower than the lowest j3cyclodextrin concentration. (37) Bates, R.G.Determination ofpH; Theory and Practice; Wiley: New York, 1964; p 220. (38) Scott, R. L.R e d . Trau. Chim. Pays-Bas 1956, 75, 787-789.

ANALYTICAL CHEMISTRY, VOL. 64, NO. 13,JULY 1, 1992 1S O

1

1.50

1407

4

1 .zo

0.90

0.60

0.30

0.00

0.00

200

250

300

350

400

450

500

550

600

200

WAVELENGTH (nm) Figure 1. UV-visible spectra of the DNP-amino acids in the solvent of 0.10 M phosphate buffer, pH = 0.0: (A) DNP-leucine, (B) DNP-

valine, and (C) DNP-methionine. The concentrationsof the DNP-amino M. acids are 1.0 X Determination of Dissociation Constants by CD Spectra. The changes of the ellipticities of DNP-amino acids were measured as a function of P-cyclodextrin concentrations. The data were also handled according to a modified version of Scott's equation 1 [HoI[GoI/A.B = Kd/A[6] + ([Hal + [GoI)/A.[61 (2) where A6 is the observed change of ellipticity of the guest molecule and A[@] is the difference of molecular ellipticity of guest before and after the inclusion complexation. Here, [Hol [GollAB was plotted against ([Ho] + [Go]). The Kd was obtained by dividing the intercept by the slope. In these measurements, the concentration of DNP-amino acids was held constant at 7.0 X 10-5 M while the concentrations of 8-cyclodextrin were changed from 0.001 to 0.014 M. Determination of Dissociation Constant by Nuclear Magnetic Resonance Spectra. lH Fourier transform NMR spectra (200 MHz) were measured on a LX-200 spectrometer. The chemical shifts of the 3' protons of P-cyclodextrin were measured as a function of the concentrationsof DNP-aminoacids. The concentrationof 6-cyclodextrin was held constant at 0.001 M, and the concentrations of DNP-amino acid were varied between 0.01 and 0.02 M. The data were treated according to Bergeron and Channing's equation (3)39 where [GO] is the (3)

concentrationof DNP-amino acids, [Ho] is the concentrationof P-cyclodextrin, A6 is the change of chemical shift, and Q = (6c - 6 ~ ) . [Go]/A6 wasplottedagainst([Go] + [&I). Thedissociation constants were obtained by dividing the intercept by the slope.

RESULTS AND D I S C U S S I O N UV-Visible Studies. Figure 1 shows the UV-visible spectra of all the DNP-amino acids in a 0.10 M sodium phosphate buffer solution in the absence of 8-cyclodextrin. The UV-visible spectra of all the DNP-amino acids investigated show two absorption maxima at about 361 and 263 nm, and both the L enantiomer and the D enantiomer of one DNP-amino acid have the same spectra in the absence of 8-cyclodextrin. The effect of 8-cyclodextrin on the UV-visible spectra of DNP-amino acids was investigated by holding the concentrations of DNP-amino acids constant at 1.0 X 10-4 M and varying the concentrations of P-cyclodextrin from 0.001 to 0.015 M. As a typical example of the observed spectral change with the addition of (3-cyclodextrin, Figure 2 shows the UVvisible spectra of DNP-L-valine at various /3-cyclodextrin concentrations. In the presence of Bcyclodextrin the absorption maximum at 361 nm was shifted to longer wave(39)Bergeron, R.;Channing, M.A. Bioorg. Chem. 1976,5, 417-449.

250

300

350

400

450

500

550

600

WAVELENGTH (nm) Figure 2. UV-visible spectra of DNP-L-valineat varying 8-cyciodextrin 5.0 X 1.0 X and 1.5 X concentrations: 0, 1.0 X M, read from A to 6. The concentration of DNP-L-valine Is 1.0 X lo4 M, and the measurements were carried out in a 0.10 M phosphate buffer (pH = 6.0).

0

0

2

4

6

6

10

12

14

16

(H + Ql x 10.'

Figure 3. Scott's plots for the interactions between &cyclodextrin and (A)DNP-L-valine,(A)DNP-o-valine,(0)DNP-L-leucine,(0)DNP-

c-ieuclne, (W) DNP-L-methionine,and (m)DNP+-methionine,

lengths (3-5 nm) along with some lowering of the absorption intensity. The well-defined isobestic points at about 330 and 385 nm indicate a 1:l complex. Similar UV-visible spectra were also observed for the DNP-D-valine and other DNP amino acids. However, the changes of absorption intensity at the absorption maximum on adding jhyclodextrin are different for L and D enantiomers. A plot of the data in the form of [H~l[Gol/AAbsvs ([Hal + [Go]) showed excellent straight-line fits (see Figure 3), indicating an H + G = HG equilibrium model. Table I summarizes the dissociation constants for all the L- and D-DNP-amino acids along with the capacity factors obtained on a 8-cyclodextrin bonded stationary phase. As is seen in Table I, all the L enantiomers have longer retention times on the 8-cyclodextrin bonded phase column and show larger formation constants than the corresponding D enantiomers. Circular Dichroism Studies. Comparison of t h e CD Spectra of DNP-Amino Acids with Those of Original Amino Acids. CD spectra of DNP-D-and DNP-L-valine in the absence of 8-cyclodextrin are shown in Figure 4. The DNP-L-valinegives a negative CD band centered at 225 nm, while one positive CD peak appears a t 410 nm. In contrast, the DNP-D-valine has a positive CD peak at 225 nm and a weaker negative peak a t 410 nm. It appears that the peak at 225 nm results from the n-Ir* Cotton effect of the carboxyl group while the CD peak at 410 nm is associated with the ir-ir* transition of the dinitrophenyl group. Similar CD spectra were also observed for other DNP-D- and DNP-Lamino acids. Compared with the CD spectra of the original amino acids, the observed molecular ellipticity of the DNP-

1408

ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992

Table I. Dissociation Constants for Inclusion Complexes of 8-Cyclodextrin-DNP-Amino Acids guest molecules

Kd

Kd (CD, W b

(UV,M)'

(1.78 f 0.10) x 10-3 (2.36 f 0.17) X (7.11 f 0.41) X (1.15 f 0.21) x 10-3 (1.38 A 0.19) x 10-3 (1.78 f 0.25) X

DNP-L-valine DNP-D-valine DNP-L-leucine DNP-D-leucine DNP-L-methionine DNP-D-methionine

Kd

k'

(NMR, M)'

2.85 x 3.38 x 1.52 x 2.43 x 2.47 x 2.96 x

(2.83 f 0.05) x (3.26 f 0.05) X (1.66 A 0.20) x 10-3 (2.15 f 0.25) X (1.32 f 0.07) X (2.16 f 0.41) X

10-3 10-3 10-3 10-3 10-3 10-3

6.93 5.87 11.73 8.53 7.60 6.70

Measurements were made in 0.10 M sodium phosphate solution, pH 6.0; temperature, 25 O C . Measurements were made in 0.10 M sodium phosphate solution, pH 6.0; temperature, 21 O C . Measurements were made in DzO containing 0.10 M sodium phosphate, pD 11.0;temperature, 20 O C . The capacity factors were obtained on a 250- X 4.6-mm j3-cyclodextrin bonded phase column with the mobile phase containing 20% methanol and 80% water. 1

50 I

'

-50 100

50 I

I

A

200

400

300

500

600

100

200

300

WAVELENGTH (nm)

-

Flgure 4. CD spectra of DNP-L-vailne (- -) and DNP-o-vallne(-) in the 0.10 M phosphate buffer (pH = 6.0). The concentrations of both M. enantiomers are 7.0 X

amino acids at 225 nm is about 3 times as large as that of the original amino acids. The most remarkable change is the sign of the CD band at about 225 nm. It has been reported that all L-a-aminoacids, except the cyclic amino acid proline, give positive Cotton effect curves with CD peaks in the wavelength range from 213 to 228 nma40 The negative CD peaks for DNP-L-amino acids at 225 nm indicate that the DNP derivatization of amino acids changes the sign of the Cotton effect associated with the carboxyl group. The changed sign of the Cotton effect for DNP-amino acids can be easily explained by considering the sector rule derived from the octant rule41942 by Klyne and Scopes.43 This discovery helped us to discover and correct the mistake we made in determining the elution order of DNP-DL-amino acids on a 8-cyclodextrincolumn.44 Now, it is found that the first eluted enantiomer possessing a positive CD peak at 225 nm is the DNP-D-amino acid, rather than the DNP-L-amino acid as earlier reported. CD Spectra of Amino Acids in the Presence of @-Cyclodextrin. Figure 5 shows the CD spectra of DNP-L-valine a t varying 8-cyclodextrin concentrations. In this case, the concentration of DNP-L-valine was held constant at 7.0 X 10-6 M, and the CD spectra intensity were measured a t 8cyclodextrin concentrations of 1.0 X 5.0 X and 9.0 X 10-3 M. An increase in 13-cyclodextrinconcentration lowers the intensity of the negative CD peak at 225 nm. In the presence of 8-cyclodextrin, a broadened CD band is observed in the region 270-430 nm. Compared to the CD spectra of (40) Jennings, J. P.; Klyne, W.; Scopes, P. M. J. Chem. SOC.1965, 294-296. (41) Moffitt, W.; Woodward, R. B.; Moscowitz,A.; Klyne, W.; Djerassi, C. J. Am. Chem. SOC.1961,83,4013-4018. (42) Moscowittz, A. Tetrahedron 1961, 13, 48-56. (43) Klyne, W.; Scopes, P. M. In Optical Rotatory Dispersion and Circular Dichroism in Organic Chemistry; Snatzke, G., Ed.; Heyden & Son: London, 1967; Chapter 9, p 139. (44) Li, S.;Purdy, W. C. J. Chromatogr. 1991, 543, 105-112.

400

500

600

WAVELENGTH (nm)

Figure 5. CD spectra of DNP-c-valine at varying 8-cyclodextrin 5.0 X and 9.0 X M, read concentrations: 0, 1.0 X from A to B. The concentratlon of DNP-L-valine Is 7.0 X M. The 0.10 M phosphate buffer soiutlon (pH = 6.0) was used as solvent.

1

A

-50

'

100

I 200

300

400

500

600

WAVELENGTH (nm)

spectra of DNP-o-valine at varylng 8-cyclodextrin 5.0X and 9.0 X concentrations: 0,5.0X lo-', 1.0 X M, read from A to B. The concentratlon of DNP-o-valineis 7.0 X M; solvent, 0.10 M phosphate buffer (pH = 6.0). Figure 6. CD

DNP-L-aminoacid in the absence of 8-cyclodextrin,we found that the broadened CD band consists of two compartments, the intrinsic part and the induced part, which is superimposed on the intrinsic part. CD spectra of the DNP-D-valine at different P-cyclodextrin concentrations are shown in Figure 6. The intensity of the CD peak at 225 nm becomes lower with increasing 8cyclodextrin concentrations. However, unlike the CD spectra of DNP-L-valine, only an independent induced CD band is observed in the range 290-370 nm. This induced positive band is centered at about 350 nm with an intensity higher than the induced CD peak of DNP-L-valine. The difference in the shape and intensity of the induced CD spectra of DNP-L-valineand DNP-D-valine in the presence of 8-cyclodextrinis mainly ascribed to the different orientation and/or disposition of the D and L enantiomers (or the DNP

ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992

i

-

1409

H

R

' .+COOH

Flgurr 7. (a, Left) coordinate system of the electric dlpole moment. (b, Right) most likely dlspositlon for j3-cyclodextrln-DNP-amino acM complexes.

groups) in the j3-cyclodextrin cavity. The intensity of the induced CD band by j3-cyclodextrin can be interpreted in terms of the Kirkwood-Tinoco coupled oscillator modek20~46

R

-

A ( 1 + 3 cos 2 6 ' ) ~ ~ where R is the rotation strength of CD intensity,A is a constant dependent only on the wavelength, and 6' is the angle made by the direction of the electric dipole moment ( F ) of the guest molecule and the symmetry axis of the 8-cyclodextrin. The rotation strength is independent of the rotation of the guest molecule around the symmetry axis of 8-cyclodextrin but sensitive to the angle 6'. The sign of the induced CD band is determined only by the relative orientation of the dipole moment in the j3-cyclodextrin cavity. The electric dipole moment parallel to the symmetry axis of j3-cyclodextrin gives a positive CD band, while the perpendicularly polarized moment produced the negative CD band. The observed positive CD bands induced by 8-cyclodextrin indicate that the electric dipole moment in parallel with the long axis of the 2,4-dinitrophenyl group is parallel to the symmetry axis of 8-cyclodextrin. It appears that axial inclusion complexes, as shown in Figure 7, are formed between D and L DNPamino acids and the j3-cyclodextrin cavity. The lower intensity of the induced CD band for DNP-L-valineindicates that DNPL-valine has a large tilt angle (6') against the symmetry axis of @-cyclodextrin,probably due to the formation of strong hydrogen bonding between the carboxyl group of DNP-Lvaline and the hydroxyl group on the edge of j3-cyclodextrin. The higher intensity of the induced positive CD band for DNP-D-valinemeans a smaller tilt angle (6') between the long axes of the dinitrophenyl group of DNP-D-valine and the symmetric axis of j3-cyclodextrin. The smaller tilt angle may be caused by the steric repulsion between the alkyl group (R) and the hydrophilic surface of 8-cyclodextrin. The CD spectra shown in Figures 5 and 6 were also observed for the L and D enantiomers of DNP-leucine and DNP-methionine, respectively. In all the cases, it was found that the intensity of the induced CD peaks at 350 nm increased with the increasing j3-cyclodextrin concentrations. However, attempted fit of the ellipticity data for these induced CD peaks to the modified version of Scott's equation failed. No straight line can be found for the plots of [H0l[GollA6'versus ([Hal [Go]) for all the DNP-amino acids studied. The intensity of the intrinsic CD peaks at 225 nm decreased with increasing j3-cyclodextrin concentrations. The ellipticity changes when plotted accordingto the modified Scott equation gave straight lines with correlation coefficients larger than 0.9950 for all the D- and L enantiomers of DNP-amino acids. The dissociation constants obtained are also listed in Table I. These dissociation constants reflect, to some extent, the interactions between the carboxyl group of the DNP-amino acids and the exterior surface of j3-cyclodextrin. NMR Studies. The 1H NMR spectrum of free j3-cyclodextrin in D2O is shown in Figure 8A. The spectrum of free

+

r .

I""I'

~

4.0

~

3.8

~

.

)

3.8

~

~

~

3.4

~

I

)

3.2

)

PA(

~

~

Flgurr 8. 'H NMR (200 Hz) spectra of j3-cyclodextrln at dlfferent molar ratios of DNP-L-vallne to flcyclodextrin: (A) 0, (B) 0.5, (C) 1.O, and (D) 2.0. The j3-cyclodextrln concentration is constant at 0.005 M.

0-cyclodextrin was initially assigned by Demarco and Thakker.46 The assignments were confirmed in the present work by decoupling and COSY experiments. The free j3-cyclodextrin resonance positions a t 20 f 1 "C relative to solvent line (DzO, 4.63 ppm) are 1' at 4.925 (doublet), 2' at 3.477 (doublet of doublets), 3'at 6 3.824 (triplet),4'at 3.428 (triplet), and 5' at 3.664 (doublet of triplets), and both 6'protons nearly overlap at 3.728. In DzO solution only resonances from the nonchanging hydrogens attached to the carbons are detected. The resonances for the active 2', 3', and 6' hydroxyl protons were not observed. (We use i' to assign the j3-cyclodextrin protons and use iH to assign the aromatic protons of DNPamino acids). Effect of DNP-Amino Acid on the lH NMR Spectra of B-Cyclodextrin. The effects of DNP-amino acids on the spectrum of j3-cyclodextrin were investigated by holding the concentration of j3-cyclodextrin constant and changing the molar ratios of j3-cyclodextrin to DNP-amino acid from 0 to 20. Parts B-D of Figure 8 show the effect of DNP-L-valine on the lH NMR spectrum of 8-cyclodextrin. As expected, the lower field triplet assigned to the 3' proton resonance is progressively shifted to higher field with increasing concentration of DNP-L-valine in the solution. An upfield shift is also observed for the 5' proton resonances which originally superimposed on the 6' proton signals. No significant chemical shift or line broadening is observed for the 2', 4', and 6' protons. Figure 9 shows the plots of chemical shift values for the protons of j3-cyclodextrin as a function of R, the molar ratio of DNP-L-valine to P-cyclodextrin. The chemical shift of the 1' proton is used as the reference for the calculation of A6 values. As can be seen in Figure 9, the A6 values for the 2') 4', and 6' protons are constant to within 2 Hz over the range of R values from 0 to 4.0. This means that the relative shift of the exterior protons are unaffected by the addition of DNP-L-valine. The A6 value for the resonance of

~

(45) Harata, K. Bull. Chem. SOC.Jpn. 1979,52, 1807-1812.

,

3.0

(46) Demarco, P. V.; Thakker, A. L. Chem. Commun. 1970, 2-4.

~

~

~

~

1410

ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992

280

0

1

3

2

4

5

R

Flgure 11. Plots of the chemical changes for 3’ protons of @-cycledextrln in the presence of (A)DNP-L-valine, (0)DNP-Pvailne, (V) DNP-L-leucine,(+) DNP-deucine, (m) DNP-L-methlonlne, and (0)DNPo-methionine. R is the molar ratio of DNP-amino acid to j3-cyclodextrin.

310

280

-

(

I

-

270

r -

-

:

:

- -

~

:

A

2’

I

3‘

-

0

4’

Y

210

’

0

1 1

3

2

4

5

R

Figure 10. Plots of the chemlcal shift changes for the protons of 8-cyclodextrln against the molar ratio of DNP-Pvailne to 8-cyclodextrin.

the 3’ protons increased 38.42 Hz with increasing molar ratio (R)from 0 to 2.0. The effect of DNP-L-valineon the resonance of the 5’ protons is less substantial. The A6 value for the resonance of the 5’ protons only increased 19.33 Hz within the range of molar ratio from 0 to 2.0. The effect of DNP-D-valine on the lH NMR spectra of j3-cyclodextrin was also examined. In Figure 10, we plotted the chemical shift (A&) values for the protons of P-cyclodextrin against the molar ratio of DNP-D-valine to 8-cyclodextrin. Again, the A6 values of the 2’, 4‘, and 6’ protons are constant over the investigated R range, and upfield shifts were observed for the resonances of the 3’ and 5’ protons. However, the effects of DNP-D-valine on the resonances of the 3’ and 5’ protons are significantly different from the effect of DNP-L-valine. The effect of DNP-Pvaline on the resonance of the 3’ protons is less substantial. Within the range of R from 0 to 2.0, the A6 value for the 3‘ protons increased 33.62 Hz, about 5 Hz lower than the A6 value observed in the presence of DNP-L-valine. In contrast, the effect of DNPD-valine on the resonance of the 5‘ protons is more significant than the effect of DNP-L-valine. The A6 value for the 5’ protons increased 21.43 HZ with increasing molar ratio from 0 to 2.0, about 2 H z higher than the A6 value produced by the effect of DNP-L-valine. The strong molar ratio dependence of the A6 values for the 3‘ and 5‘ protons is the direct evidence for the inclusion complex formation between the cavity of 8-cyclodextrin and the dinitrophenyl group of DNP-amino acid, since only when the dinitrophenyl group of the DNP-amino acid includes into

the j3-cyclodextrin cavity can the strong anistropic shielding of the aromatic ring become accessible to the 3’and 5’protons. The A6 value for the 3’ protons which form a ring near the larger opening of the cavity is related to the stability of the inclusion complex,ll while the A6 value for the 5’ protons which form a ring near the smaller opening of the cavity can be taken as an indicator for the penetration depth of the aromatic group of the DNP-amino acid. The larger A6 value for the 3’ protons observed in the presence of DNP-L-valine suggests that DNP-L-valine forms a more stable inclusion complex with P-cyclodextrin than ib D enantiomer. The larger A6 value observed for the 5’ protons in the presence of DNPD-valine indicates that the dinitrophenyl group of DNP-Dvaline penetrates into the cavity more deeply than the dinitrophenyl group of DNP-L-valine. The effects of L and D enantiomers of DNP-leucine and DNP-methionine on the spectra of P-cyclodextrin demonstrate similar results to DNP-L-valine and DNP-D-valine, respectively. Figure 11 shows the plots of A6 values for the resonance of the 3’ protons versus the molar ratio ( R )for all the DNP-amino acids studied. In all the cases studied, the effect of the L enantiomer on the resonance of the 3’ protons is more pronounced than that of the D enantiomer. The chemical shift changes of the 3‘ protons with the addition of DNP-amino acids are used to calculate the dissociation constants of the inclusion complex according to Bergeron and Channing’s equation. The results are also listed in Table

I. Effect of B-Cyclodextrin on ‘HNMR Spectra of DNPDL-Amino Acids. The effect of 8-cyclodextrin on the lH NMR spectra of DNP-amino acids was studied by setting the concentration of DNP-amino acid at 0.005 M and varying the molar ratio of j3-cyclodextrin to DNP-amino acid from 0 to 3.0. It is found that in the presence of j3-cyclodextrin, the resonance of the aromatic 6 H is shifted to lower field, and no significant difference can be observed between D and L enantiomers. Downfield proton resonance shifts of a molecule can usually be observed when this molecule binds to another by interactions of dipole-dipole, dipole-induced dipole, and London, which physical chemists have referred to as van der Waals’, forces,47by steric perturbation,@or by diamagnetic anisotropy of particular bonds or regions of the host mole~ u l e .In~ this ~ case, the downfield of aromatic 6 H may be (47) Howard, B. B.; Linder, B.; Emerson, M. T. J. Chem. Phys. 1962, 36, 485-490. (48)Cheney, B. V.; Grant, D . M. J. Am. Chem. SOC.1967,89, 53195327.

ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992

mainly caused by the van der Waals’ interactions with the hydrophobic cavity of P-cyclodextrin. In contrast, for all the DNP-amino acids, the chemical shift of aromatic 3H,which is located between the two nitro substituents, is moved upfield with inclusion complexation with P-cyclodextrin. No significant chemical shift change was observed for the aromatic 5H.This fact further confirms the inclusion complex structure predicted by the CD studies. As shown in Figure 7, the DNP group of the guest molecules is included into the /3-cyclodextrin cavity in such a way that the 6H is closer to the wall of the cavity than the 5H. It is known that van der Waals’ forces are extremely short-range, being proportional to l/1.6,50where r is the distance between the two interaction groups. Therefore, the interactions between the cavity wall and 5H are very weak, and so no substantial chemical shift change is expected for this proton upon the inclusion complex formation with 0-cyclodextrin. In the presence of P-cyclodextrin, substantialchemical shift, changes were also observed for alkyl protons of both D and L DNP-amino acids. However, the shift changes are more significant for the alkyl protons of the L enantiomers. For example, the chemical shifts of the methyl protons moved 6.62 Hz to lower field for both DNP-L-leucine and DNP-Lvaline after inculsion complex formation with P-cyclodextrin (R = l),while the changes are only 1.1and 1.64 Hz for DNPD-leucine and DNP-D-valine, respectively. This kind of behavior may be indicative of a secondary inclusion complex formation between the smaller side of the 0-cyclodextrin cavity and the methyl group of the DNP-L-amino acids. The line width for the resonances of the alkyl protons also increased significantly with increasing P-cyclodextrin concentration. Although perfectly quantitative line width comparison is impossible due to the lack of an internal line width standard, it is clear that the resonance lines assigned to the protons in the vicinity of the chiral center are broadened more significantly than those of the protons far from the chiral center. The shift changes and line broadening may be caused mainly by the steric perturbations upon inclusion complex formation with 0-cyclodextrin. I t should be pointed out that the host-guest system is in the NMR chemical-shift fast-exchange limit.8 In this case, the measured resonance positions for the protons of DNPamino acid appear as the average of the chemical shift of free DNP-amino acid and the chemical shift of DNP-amino acid bound in each possible orientation to P-cyclodextrin, weighted by the fractional population of DNP-amino acid molecules in each environment. For P-cyclodextrin protons, the measured resonance positions are the average of the chemical shift of “empty” P-cyclodextrin molecules and the chemical shift of P-cyclodextrin molecules which have guests, weighted by the fractional population in each state. Dissociation Constants of @-Cyclodextrin-DNP-Amino Acid Inclusion Complexes. All the dissociation constants determined by UV-visible, CD, and NMR methods are listed in Table I, along with the retention data obtained on the P-cyclodextrin bonded phase column. The Kd values listed in Table I show large deviations between different determination methods. As can be seen, the& values measured by NMRspectroscopy are about twice as large as that determined by the UV method. This difference is mainly attributed to the different media used for the measurements. UV measurements, as mentioned in the Experimental Section, were carried out in a pH = 6.0 (49) Apsimon, J. W.; Graig, W. G.; Demarco, V. P.; Mathieson, D. W.; Saunder, L.; Whalley, W. B. Tetrahedron 1967, 23, 2339-2355. (50) Chang, R. Physical Chemistry with Applications to Biological Chemistry; Macmillan Publishing: New York, 1977; p 134.

1411

phosphate buffer, while NMR measurements were done in a pH = 11.0 phosphate buffer. The results demonstrated that the stability of the P-cyclodextrin-DNP-amino acid inclusion complexes decreased with increasing pH. The Kd values obtained by the CD method, which mostly reflect the interaction extent between the carboxyl group of the amino acid and the 8-cyclodextrin,also show significant differences from the UV results. Although large Kd deviations exist between the determination methods, it is clear that P-cyclodextrin did form enantioselective inclusion complexes with DNP-amino acids. The Kd values determined by all of these three methods clearly indicate that the inclusion complex formed between @-cyclodextrin and DNP-L-amino acid is more stable than the corresponding 0-cyclodextrin-DNP-D-amino acid inclusion complex. This fact coincides with the chromatographic data which show that the DNP-L-amino acid always has longer retention times than its D enantiomer on a 0-cyclodextrin bonded phase column. Structure of the Inclusion Complexes and Chiral Recognition Mechanism. In principle, DNP-amino acids can penetrate the 8-cyclodextrin cavity in only two orientations, either the amino acid part first or the p-nitro group of the DNP substituent first. The orientation of inserting the ortho or meta position of the DNP substituent is impossible because very little of the substituent would actually fit onto the cavity. The CD and lH NMR studies clearly suggest that the penetrations for both D and L DNP-amino acids are by the p-nitro group head-on into the cavity from the wide 2,3-dihydroxyl side,with the amino acid part sticking out, as shown in Figure 7. It is observed that the depth and the tilt of the aromatic ring in the cavity are significantly different for the D and L enantiomers. The lower intensity of the induced bands for DNP-L-amino acids indicate that DNP-L-amino acids have a tilt angle in the cavity larger than their D enantiomers. The large angle may be caused by the formation of strong hydrogen bonding between the carboxyl group of the DNP-L-amino acid and the hydroxyl group on the edge of the P-cyclodextrin cavity. The larger chemical shift change observed for the 5’ protons of P-cyclodextrin in the presence of DNP-D-aminoacid suggests that the insertion is deeper for a DNP-D-amino acid than for a DNP-L-amino acid because the large tilt of the DNP group in the latter case hinders it from penetrating further. The shallower insertion for the DNP-L-amino acids leaves enough space in the 0cyclodextrin cavity to host the alkyl group from another DNPL-amino acid. In the presence of P-cyclodextrin, the large chemical shift changes observed for the alkyl protons of DNPL-aminoacids are direct evidence for the existence of secondary inclusion complex formation between the P-cyclodextrin and the alkyl groups. On the basis of observations and the above discussions, the proposed structure of 6-cyclodextrin complex with DNP-Lamino acids is in Figure E A , where two or even more inclusion complexes of 1:lstoichiometryare associated. In the primary 1:lcomplex, the dinitrophenyl portion of DNP-L-amino acid is included within the cavity and the carboxylic group interacts with the hydroxyl group a t the edge of the cavity through hydrogen bonding, leaving the alkyl group to insert into another cyclodextrin cavity from its smaller 6-hydroxyl side. Figure 12Bis the proposed structure of P-cyclodextrin with DNP-D-amino acids. In this case, the repulsion between the alkyl group of a DNP-D-amino acid and the hydroxyl groups a t the edge of the @-cyclodextrincavity results in a smaller tilt angle and deeper penetration of the DNP group in the cavity, and leaves no space to hold the alkyl group into the cavity. On the other hand, since the alkyl groups of DNPD-amino acids are held tightly against the edge of the @-cy-

1412

ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992 R

.H

major contributors to the chiral recognition. CONCLUSION

p-CD : DNP-L-amino acid p-CD : DNP-D-amino acid Flgure 12. Structures of inclusion complexes of 8-cyclodextrin with

DNP-o-aminoacid and with DNP-L-amino acid.

clodextrincavity, even if there were enoughspace in the cavity, it is difficult to form a secondary inclusion complex due to the steric hindrance. Therefore, no secondary inclusion complex can be formed between 8-cyclodextrin and the alkyl groups of DNP-D-amino acids. These proposed structures for DNP-L-amino acids and DNP-D-amino acids show the precise nature of the chiral discrimination interaction and chirality forces responsible for the chiral resolutions. It appears that the DNP group, which forms a stable inclusion complex with the 8-cyclodextrin cavity and places the other functional groups around the chiral center in association with the polar hydroxyl groups at the edge of the cavity, plays a very important role in the chiral recognition. The alkyl groups of amino acids, which form a secondary inclusion complex with another 8-cyclodextrin cavity (in the case of DNP-L-amino acids) or play a role of steric repulsion with the hydroxyl groups at the edge of the cavity (in the case of DNP-D-amino acids), are also

UV, CD, and lH NMR studies give detailed information about the stability and structures of the inclusion complexes of 8-cyclodextrinwith DNP-L-aminoacids and DNP-D-amino acids. Such direct spectroscopic observation of the “soluble models of the chromatographic system” substantiate the presumed chiral recognition mechanism and add further details as to conformational preference during inclusion complex formation. The results demonstrate that the formation of an inclusion complex is an essential requirement for the chiral separation, but it is clear that only inclusion complexation is not sufficient for chiral recognition. Sufficient chiral recognition also requires the interactions of other functional groups around the chiral center with the mouth of the fi-cyclodextrin cavity and/or with another cyclodextrin cavity to form a secondary inclusion complex. These spectroscopic studies account for the contribution of mobile-phase composition but still neglect the intricate effect of the silica surface. These spectroscopic studies, together with the chromatographic data in a previous report,U the theoretical studies,51and the computer-modeling method,52 give a more clear picture for the chiral recognition process and will aid the design of new chiral stationary phases to meet specific chiral separation problems. ACKNOWLEDGMENT We are indebted to the Natural Sciences and Engineering Research Councilof Canada for financial support of this work. RECEIVED for review December 17, 1991. Accepted March 23, 1992. (51) Boehm, R. E.; Martire, D. E.; Armstrong, D. W. Anal. Chem. 1988,60, 522-528.

(52) Armstrong, R. D. In Ordered Media in Chemical Separations; Hinze. W. L.. Armstrona. D. W.. Eds.: American Chemical Society: Washington, DC, 1987; p-272.