Enantiomeric separations of terbutaline by CE with a suffated β- cyclodextrin chiral selector: A quantitative binding study

Article abstract of DOI:10.1021/ac980780i

Sulfated β-cyclodextrin, a negatively charged chiral selector, was used for the enantiomeric separation of racemic terbutaline by capillary electrophoresis. Chiral separation was found to increase with decreasing cyclodextrin concentration. Host-guest complex binding constants for this system were determined by UV difference spectroscopy (K(av) = 1490 M^-1) and by CE under conditions of minimal EOF and reversed polarity (K₁ = 1730 M^{- 1}, K₂ = 1590 M^-1, α = 1.09). The effect of organic modifiers, methanol, and acetonitrile was also studied over a wide range of modifier concentrations. Binding constants decreased while selectivity increased with increasing organic modifier concentration (10% MeOH: K₁ = 1590 M^-1, K₂ = 1130 M^-1, α = 1.41. 10% ACN: K₁ = 1320 M^-1, K₂ = 870 M^-1, α = 1.52). Experimental results are discussed in the context of existing separation models.

Full text of DOI:10.1021/ac980780i

Anal. Chem. 1998, 70, 5166-5171
En a n t io m e ric S e p a ra t io n s o f Te rb u t a lin e b y CE
w it h a S u lfa t e d â-Cyc lo d e x t rin Ch ira l S e le c t o r: A
Qu a n t it a t ive Bin d in g S t u d y
Sa m ue l R. Gra tz a nd Apryll M. Sta lc up*
Department of Chemistry, University of Cincinnati, ML 0172, Cincinnati, Ohio 45221
behavior and designing optimization strategies.5 Spectroscopic
methods have been used for the determination of host-guest
complex binding constants.6 More recently, however, CE has
been adopted for binding constant determinations based on the
migration equation that emerged from the work of Alberty and
King in 1951.7
Sulfated â-cyclodextrin, a negatively charged chiral selec-
tor, was used for the enantiomeric separation of racemic
terbutaline by capillary electrophoresis. Chiral separation
was found to increase with decreasing cyclodextrin con-
centration. Host-guest complex binding constants for
this system were determined by UV difference spectros-
-
1
copy (Kav ) 1 4 9 0 M ) and by CE under conditions of
Cyclodextrins (CDs) are among the most prevalent selectors
used in chiral CE. Currently, a wide range of native CDs (R, â,
γ) as well as charged and neutral derivatives are commercially
-
1
minimal EOF and reversed polarity (K
)
1
) 1 7 3 0 M , K
2
-
1
1 5 9 0 M , r ) 1 .0 9 ). The effect of organic modifiers,
8
-12
methanol, and acetonitrile was also studied over a wide
range of modifier concentrations. Binding constants
decreased while selectivity increased with increasing
available and have been employed as chiral selectors.
Al-
though neutral chiral selectors are the most frequently used,
several reports have demonstrated the advantages of negatively
charged selectors such as sulfobutylated CD,13 sulfated CD,14-16
dextran sulfate,17 and heparin.18 Tait et al. explained that the use
of a negatively charged chiral selector effectively increases the
“separation window”, as the maximum opportunity for separation
may exist when the analyte and chiral selector migrate in opposite
directions.13
Several applications of anionic CDs to chiral separations have
come with the applied voltage polarity reversed, under conditions
of suppressed electroosmotic flow (EOF).14,19 In this format,
analytes are injected at the cathodic end of the capillary and
detected at the anodic end, ensuring that the separated enanti-
omers will reach the detector only when complexed with the CD.
A recent review by Janini and Issaq concerning the suppression
of EOF pointed out that the use of coated capillaries can effectively
diminish EOF and may be superior to the use of uncoated
capillaries at low pH which suffer from instability and irreproduc-
ibility.20
organic modifier concentration (1 0 % MeOH: K
1
) 1 5 9 0
) 1 3 2 0
-
1
-1
M
M
, K
, K
2
) 1 1 3 0 M , r ) 1 .4 1 . 1 0 % ACN: K
) 8 7 0 M , r ) 1 .5 2 ). Experimental results are
1
-1
-1
2
discussed in the context of existing separation models.
The separation of enantiomers has received a great deal of
recent attention, specifically in the pharmaceutical industry.
Because chiral drugs are commonly administered as racemic
mixtures, and the two enantiomers often exhibit different phar-
macological effects, it is of significant interest to develop analytical
methods for chiral analysis. Still a relatively young technique,
capillary electrophoresis (CE) has quickly established itself in the
area of chiral separations.1 The separation of enantiomers by CE
is typically accomplished by exploiting stereospecific interactions
between chiral analytes and a chiral selector that has been added
to the background electrolyte. In brief, one enantiomer complexes
more favorably with the chiral selector, establishing a dynamic
equilibrium between the free analyte and the diastereomeric
complex.
(
(
5) Kord, A.; Strasters, J.; Khaledi, M. Anal. Chim. Acta 1 9 9 1 , 246, 131-137.
6) Benesi, H.; Hildebrand, J. J. Am. Chem. Soc. 1 9 4 9 , 71, 2703-2707.
The separation of two enantiomers can take place if there is a
difference in the binding or stability constants (K) between each
enantiomer and the chiral selector as well as a difference between
the mobility of the free analyte and the complex. Theoretical
models based on binding equilibria have been shown to give an
excellent account of the effects that changing the selector
concentration has on enantioresolution.2 Knowledge of binding
constants can provide a better understanding of separation
mechanisms and, consequently, can aid in predicting migration
(7) Alberty, R.; King, E. J. Am. Chem. Soc. 1 9 5 1 , 73, 517.
(
8) Fanali, S.; Cristalli, M.; Vespalec, R.; Bocek, P. Adv. Electrophor. 1 9 9 4 , 7,
-86.
9) Nishi, H.; Terabe, S. J. Chromatogr., A 1 9 9 5 , 694, 245-276.
1
(
(10) Terabe, S.; Miyashita, Y.; Ishihama, Y.; Shibata, O. J. Chromatogr. 1 9 8 5 ,
3
32, 211-217.
(
(
11) Schmitt, T.; Engelhardt, H. Chromatographia 1 9 9 3 , 37, 259-262.
12) Heuermann, M.; Blaschke, G. J. Chromatogr. 1 9 9 3 , 648, 267-274.
-4
(13) Tait, R.; Thompson, D.; Stella, V.; Stobaugh, J. Anal. Chem. 1 99 4 , 66, 4013-
018.
4
(
(
(
14) Stalcup, A.; Gahm, K. Anal. Chem. 1 9 9 6 , 68, 1360-1368.
15) Dette, C.; Ebel, S.; Terabe, S. Electrophoresis 1 9 9 4 , 15, 799-803.
16) Gahm, K.; Stalcup, A. Chirality 1 9 9 6 , 8, 316-324.
(
(
(
(
1) Ward, T. Anal. Chem. 1 9 9 4 , 66, 632A-640A.
2) Wren, S.; Rowe, R. J. Chromatogr. 1 9 9 2 , 603, 235.
3) Penn, S.; Goodall, D.; Loran, J. J. Chromatogr. 1 9 9 3 , 636, 149.
4) Ferguson, P.; Goodall, D.; Loran, J. J. Chromatogr., A 1 9 9 6 , 745, 25-35.
(17) Agyei, N.; Gahm, K.; Stalcup, A. Anal. Chim. Acta 1 9 9 5 , 307, 185-191.
(18) Stalcup, A.; Agyei, N. Anal. Chem. 1 9 9 4 , 66, 3054-3059.
(19) Janini, G.; Muschik, G.; Issaq, H. Electrophoresis 1 9 9 6 , 17, 1575-1583.
(20) Janini, G.; Issaq, H. J. Chromatogr., A 1 9 9 7 , 792, 125-141.
5166 Analytical Chemistry, Vol. 70, No. 24, December 15, 1998
10.1021/ac980780i CCC: $15.00 © 1998 American Chemical Society
Published on Web 10/24/1998

terbutaline was accomplished at the anodic end of the capillary.
All CE experiments were conducted using 0-40 mM sulfated
cyclodextrin and thermostated at 20 °C. Using the same cyclo-
dextrin concentration range, buffer viscosity measurements were
made using a Cannon-Ubbelohde No. 150 viscometer (Cannon
Inst., State College, PA) also thermostated at 20 °C.
UV difference experiments were performed using a Hewlett-
Packard 8452A diode array spectrophotometer and a 1-cm quartz
cell. Solutions prepared for analysis were 25.0 µM terbutaline in
pH 3 phosphate buffer. The range of sulfated-â-cyclodextrin
concentrations was from 0.0 to 0.8 mM. Identical conditions were
used in the presence of 10% (w/ w) methanol.
Spectroscopic sodium measurements were made using a
Plasma-Spec inductively coupled plasma atomic emission spec-
trometer with a photomultiplier detector (Leeman Labs Inc.,
Lowell, MA).
The sodium electrochemical measurements were made using
an Orion model 86-11 Ross sodium ion-selective electrode (Bev-
erly, MA). To each sodium standard and sample solution, equal
Fig u re 1 . Sulfated â-cyclodextrin.
In the present study, chiral separations of the antiasthmatic
drug terbutaline were performed using sulfated â-cyclodextrin
volumes of 4 M NH
ionic strength.
4 4
Cl/ NH OH were added to maintain constant
(
Figure 1) as the chiral selector in CE. Binding constants
determined spectroscopically through the use of Benesi-Hilde-
brand plots are presented and compared to those determined by
6
THEORY
capillary electrophoresis. The main objective of this work was to
explore the effects that the organic modifiers, methanol (MeOH)
and acetonitrile (ACN), have on the binding constants and
enantioselectivity. Unlike the results of several related studies
in which normal polarity was used,4,21-22 the addition of an organic
modifier to the run buffer was shown to enhance chiral selectivity
for terbutaline.
The interaction between sulfated cyclodextrin (SCD) and the
enantiomers of a chiral analyte is an equilibrium reaction which,
assuming 1:1 complexation, may be represented as
K
A + SCD 7
9
8 A-SCD
(1)
where A is the analyte, SCD is the sulfated CD, and A-SCD is the
complex with a binding constant of K. In most techniques used
to determine binding constants for enantiomers, the general
strategy is to measure some system response (absorbance,
mobility, retention time, etc.) to varying chiral selector concentra-
tion at constant analyte concentration. That response may then
be related to the relative concentrations of free and bound analyte
and subsequently to the selector-analyte binding constant.
In UV difference spectroscopy, the absorbance spectrum of
an unbound analyte is measured and subtracted from its spectra
at several CD concentrations. K is then obtained from a Benesi-
Hildebrand or double-reciprocal plot based on eq 2.6
EXPERIMENTAL SECTION
Materials. Sulfated cyclodextrin with ∼13 sulfates/ cyclodex-
23
trin (13 ds) was obtained from Aldrich Chemical Co. (Milwaukee,
WI). Racemic terbutaline was obtained from Sigma Chemical Co.
(
2 4 3 4
St. Louis, MO). The NaH PO , H PO , Na standards, and HPLC
grade methanol and acetonitrile were obtained from Fisher
Scientific (Pittsburgh, PA). CE buffers were prepared with
distilled, deionized water, and reagent grade NaH
2 4
PO and
adjusted to pH 3 with 85% H PO . For the experiments in which
3
4
organic modifiers were used, the sulfated cyclodextrin was added
to the hydroorganic mixture (w/ w).
Apparatus and Methods. All CE experiments were per-
formed using a Bio-Rad BioFocus 3000 automated capillary
electrophoresis system (Bio-Rad Laboratories, Inc., Hercules,CA)
interfaced with a Pentium personal computer (Gateway). Fused-
silica capillaries with a polyacrylamide-coated inner surface (25-
µm i.d., 24-cm total length, 19.6 cm to detector) were also obtained
from Bio-Rad. Prior to each experiment, the run buffers were
passed through a 0.2-µm filter and degassed. According to column
care guidelines provided by the vendor, capillaries were flushed
with water for 60 s and run buffer for 90 s before terbutaline was
injected hydrodynamically (3 psi s). The applied voltage for all
CE experiments was 10 kV. UV detection (λ ) 211 nm) of 1 mM
(∆Aobs)- ) (K∆Ac)-1[B]-1 + (∆Ac)
1
-1
(2)
where ∆Aobs is the observed absorbance difference, at some
particular λ, between the analyte-CD complex and the free analyte
at different CD concentrations; ∆A is the absorbance difference
at λ between the analyte-CD complex and the free analyte, [B]
is the molar concentration of the cyclodextrin, and K is the binding
c
constant.
In the case of CE, binding constants are determined using the
relationship between the CD concentration and the electrophoretic
mobility of the analyte. A variety of different equations and
graphical methods have been employed to determine binding
constants, depending on the format of the CE experiment (MEKC,
EKC, reversed-flow EKC, enantiospecific complexation). How-
(
21) Penn, S.; Bergstrom, E.; Knights, I.; Liu, G.; Ruddick, A.; Goodall, D. J. Phys.
Chem. 1 9 9 5 , 99, 3875-3880.
(
(
22) Shanks, P. C.; Franses, E. J. Phys. Chem. 1 9 9 2 , 96, 1794.
23) Information supplied by vendor.
Analytical Chemistry, Vol. 70, No. 24, December 15, 1998 5167

ever, according to a recent review by Rundlett and Armstrong,24
the migration equations for the various modes of CE are analogous
and originate from the 1951 publication of Alberty and King.7
The relationship used in this work, eq 3, was originally derived
by Kuhn et al.25 to calculate binding constants for lectin-sugar
systems and was later adapted by Tanaka et al.2 for enantiose-
6
1
1
1
1
)
+
(3)
µCor - µf (µCD - µ )K [CD] µCD - µf
f
lective complexation using anionic cyclodextrins where µCor is the
corrected solute mobility at a given cyclodextrin concentration
[
µ
CD], µCD is the electrophoretic mobility of the complexed solute,
is the mobility of the free analyte in the opposite direction of
Fig u re 2 . Effect of increasing SCD concentration on EOF mobility
nitromethane) in a 25-µm-i.d. coated capillary: applied voltage, 10
kV; buffer, 25 mM phosphate (pH 3); b, no organic modifier; 2, 5%
methanol; [, 10% methanol; f, 5% acetonitrile; half circle, 10%
acetonitrile.
f
(
the complexed analyte and is determined in the conventional
polarity mode in the absence of the chiral selector, and K is
1
determined by plotting (µ - µ
f
)- vs [CD]-1 and is equal to
(
intercept/ slope). This eliminates the need to measure µCD
.
RESULTS AND DISCUSSION
Mobility Corrections. Many of the methods used to deter-
mine binding constants by CE require correction or normalization
procedures to negate any changes in mobility caused by changes
in bulk buffer properties (e.g., viscosity).24 When neutral additives
are used, mobilities are typically corrected by multiplying each
value by the current ratio in the absence and presence of the
additive.2 However, in the case of polyvalent additives, measure-
ments of current do not provide an accurate assessment of the
additive’s contribution to the viscosity because the additive also
contributes to the conductivity of the background electrolyte
(
BGE). However, it is important to monitor current during
electrophoresis because charged additives may generate large
currents, possibly leading to Joule heating. This was not a source
of concern in this work as currents ranged from 10 µA in the
absence of SCD to 30 µA at the highest SCD concentration.
Mobilities were corrected for viscosity changes by multiplying
each mobility value by the ratio of the measured viscosity at zero
SCD concentration to the viscosity at the concentration of interest.
Fig u re 3 . (a) Electropherogram for terbutaline (0.5 mM) injected
into pristine capillary without sulfated cyclodextrin present in 25 mM
phosphate buffer (pH 3). Applied voltage, 10 kV. (b) Electropherogram
for terbutaline (0.5 mM) injected into that had previously been used
with sulfated cyclodextrin-containing buffers. In this run there was no
cyclodextrin in the 25 mM phosphate buffer (pH 3). Applied voltage,
2
Measured viscosities ranged from 1.13 mm / s for 1 mM SCD to
10 kV.
2
1
.37 mm / s for 40 mM SCD.
In this work, there was no measurable EOF in the absence of
SCD with a pristine column. However, charged additives such
as sulfated cyclodextrins may also adsorb onto the capillary’s inner
surface, thereby changing the electric double layer. Although the
use of coated capillaries and low-pH buffers is thought to minimize
this effect, it cannot be assumed that EOF is eliminated or even
negligible. As illustrated in Figure 2, there is a small but nonzero
EOF in the presence of SCD both in the presence and absence of
organic modifiers. The EOF mobility was found to increase with
increasing cyclodextrin concentration and is thought to arise from
adsorption of the sulfated cyclodextrin on the capillary wall. This
effect is illustrated in Figure 3, which shows two separate
injections of terbutaline without cyclodextrin in the background
electrolyte. Figure 3a, obtained using conventional polarity, shows
terbutaline injected into a pristine, coated capillary. As expected,
terbutaline migrates as a single peak with a migration time of 225
s. Figure 3b shows terbutaline injected into the same capillary
after it was used with SCD-containing buffers. In this case, not
only was terbutaline detected but the enantiomers were resolved,
indicative of interactions between the enantiomers and adsorbed
cyclodextrin. It should also be noted that terbutaline failed to
elute from this column under conventional polarity; the electro-
pherogram was obtained by reversing the polarity. Thus, the
dynamic cyclodextrin layer is the only mechanism of analyte
transport to the detector. Because this layer seems to be dynamic,
no correction for adsorption was incorporated in the binding
constant determination. The migration times for the separation
in Figure 3b were comparable to those observed with buffer
containing 5% SCD.
(
(
(
24) Rundlett, K.; Armstrong, D. J. Chromatogr., A 1 9 9 6 , 721, 173-186.
25) Kuhn, R.; Frei, R.; Christen, M. Anal. Biochem. 1 9 9 4 , 218, 131-135.
26) Tanaka, Y.; Terabe, S.; Yanagawa, M., J. High Resolut. Chromatogr. 1 9 9 6 ,
In this study, enantiomer mobilities were corrected for the
measured EOF at each sulfated cyclodextrin concentration. After
1
9, 421-433.
5168 Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

incorporating corrections for both viscosity and EOF, the µcor term
in eq 3 is defined as
+
a
T
a
b
l
e
1
.
N
a
M
e
a
s
u
r
e
m
e
n
t
(
p
p
m
)
i
n
S
C
C
D
S
a
m
p
l
e
s
sample
ICP-AES
ISE
4 ds SCD (50 ppm Na+)
48.6
51.5
45.7
26.9
µcor ) [(µobs) - (µeof)](η / η )
(4)
0
x
+
15 ds SCD (50 ppm Na )
a
See Experimental Section of text for conditions.
where µobs is the observed electrophoretic mobility for the analyte
at a given set of experimental conditions, µeof is the mobility of a
neutral marker under the same experimental conditions, η
0
is the
buffer viscosity in the absence of SCD, and η is the buffer viscosity
x
at a given SCD concentration. Note that because the migration
of the neutral marker is opposite that of the analytes, µeof is
effectively added to µanalyte
.
The use of charged additives such as SCD changes the ionic
strength of the background electrolyte, which can affect analyte
mobility in two different ways. First, increased ionic strength
decreases the effective field strength experienced by the analyte
thereby decreasing its electrophoretic mobility. Second, in
uncoated fused-silica capillaries, increased ionic strength causes
decreased EOF as a result of a decreased ú potential. However,
in this study, as seen in Figure 2, wall adsorption of the sulfated
cyclodextrin with the attendant EOF confounds this relationship.
It should be noted that polyvalent species with high charge
densities such as these sulfated cyclodextrins may not contribute
as much to ionic strength as their nominal degree of substitution
suggests. Counterion condensation theory predicts that if the
linear charge density of a polyionic covalent structure exceeds a
critical value, there will be a layer of condensed counterions along
the length of the polyion, effectively reducing its linear charge
density.27 In the case of micelles, it has been demonstrated that
only a fraction of the total charge is carried because counterions
are bound to the micelle to reduce electrostatic repulsion between
head groups. Specifically, SDS micelles bind approximately three-
Fig u re 4 . UV spectra illustrating SCD-terbutaline complexation and
the effect of increasing the SCD concentration while maintaining
-4
constant terbutaline concentration at 1.25 × 10 M. SCD concentra-
-5
-4
tion range, 1.25 × 10 -2.5 × 10 M. The spectrum with the highest
absorbance corresponds to the largest SCD concentration. The
solvent for all spectra was 25 mM phosphate buffer (pH 3), which
was subtracted as the blank.
mentally determined values than the more lightly substituted
cyclodextrin, which is consistent with the counterion condensation
theory.
The current CE models for binding constant determinations
contain no explicit correction for ionic strength although, to some
extent, compensation for the ionic strength is incorporated in the
EOF corrections. One measure of the validity of neglecting ionic
strength effects may be a comparison of binding constants
determined by an alternate technique.
quarters of their counterions in the absence of alcohol or
electrolyte.22 Ion-selective electrode (ISE) measurements of
sodium activity in SDS solutions show a significant decrease in
+
[
Na ] at the critical micelle concentration, indicative of counterion
binding upon micelle formation.28-29
The possibility of counterion condensation was explored by
measuring the total atomic sodium in an SCD solution using
inductively coupled plasma atomic emission spectroscopy (ICP-
AES) and comparing the value to the sodium concentration at an
ion-selective electrode. Solutions of sulfated â-cyclodextrin were
UV Difference Spectroscopy. Figure 4 shows the absor-
-
4
bance spectra of 1.25 × 10 M racemic terbutaline at SCD
-
4
concentrations ranging from 0.0 to 2.5 × 10 M. The spectrum
of racemic terbutaline in the absence of SCD exhibits a maximum
absorbance of 1.7 with λmax at 208 nm. As cyclodextrin is added,
the maximum absorbance value reaches 1.9 and λmax is shifted to
+
prepared to be 50 ppm in total Na , assuming the degree of
substitution provided by the manufacturer was correct. The
results of these measurements are shown in Table 1 and are
reasonably close to the expected values.
The results of the ISE analysis of the aqueous SCD solutions
containing 50 ppm total Na+ are also displayed in Table 1. As
2
12 nm.
Figures 5 and 6 are the difference spectra and the resulting
Benesi-Hildebrand plot from which the binding constant was
obtained. The experimentally determined value for K was 1490
(
(60) M-1. It is important to point out that this K is an average
can be seen from the table, the more highly substituted cyclo-
dextrin exhibits a greater discrepancy between the two experi-
binding constant because UV lacks the ability to discriminate
between the two different terbutaline enantiomer-cyclodextrin
complexes.
Mobility and Binding Constant. Figure 7 illustrates the
affect that varying SCD concentration has on the mobility and
chiral separation of terbutaline. As the selector concentration
increases, solute mobility increases continuously and levels off
as the mobility of the analyte-SCD complex is approached. The
mobility difference between enantiomers increases with decreas-
(
(
(
(
27) Manning, G. S. Acc. Chem. Res. 1 9 7 9 , 12, 443.
28) Lawrence, A.; Pearson, J. Trans. Faraday Soc. 1 9 6 7 , 63, 495.
29) Jain, A.; Singh, R. J. Colloid Interface Sci. 1 9 8 1 , 81, 536.
30) Atwood, J.; Davies, J.; Macnicol, D.; Vogtle, F. Comprehensive Supramolecular
Chemistry: Cyclodextrins; Elsevier Science Inc.: New York, 1996; Vol. 3, p
1
98.
(
(
31) Wren, S.; Rowe, R. J Chromatogr. 1 9 9 3 , 636, 57-62.
32) Allenmark, S. Chromatographic Enantioseparations: Methods and Applica-
tions, 2nd ed.; Ellis Horwood; Chichester, U.K., 1991.
Analytical Chemistry, Vol. 70, No. 24, December 15, 1998 5169

Fig u re 8 . Terbutaline-SCD binding constants calculated using eq
Fig u re 5 . UV difference spectra for SCD-terbutaline complexation.
Conditions shown in Figure 4. Each spectrum is the result of
subtracting the free terbutaline spectrum from that at each sulfated
cyclodextrin concentration.
3
and the data shown in Figure 7: coated capillary -10 kV; 20 °C;
, first enantiomer; 9, second enantiomer.
2
Organic Modifiers. Addition of an organic modifier to the
buffer in CE has become fairly routine. Consequently, several
groups have studied the effect that organic modifiers have on the
differential binding of enantiomers to cyclodextrins.4
,20-22
One
conclusion shared by most authors of such work is that the
addition of organic modifiers such as methanol or acetonitrile
decreases the magnitude of the CD-analyte binding constants.
The most common explanation for this is that the organic modifier
reduces the affinity of the analyte for the CD cavity and increases
it for the bulk buffer. Alternatively, the modifier is thought to
4
compete with the solute for the CD cavity. Goodall et al. reported
that selectivity remained constant from 0 to 15% MeOH or ACN
in the chiral separation of tioconazole using native â-CD. The
authors concluded that the constant selectivity observed supported
the “solvent effect” explanation and that a simple 1:1 competitive
binding scenario seemed unlikely for weakly binding species such
as methanol.21
Fig u re 6 . Benesi-Hildebrand plot for SCD-terbutaline complex-
ation resulting from the difference spectra in Figure 5. The binding
constant was calculated using eq 2 and found to be K ) 1450 ( 60
-
1
M
.
In contrast, Janini and Issaq20 proposed that the effect of the
organic modifier is different in the reversed-polarity MEKC format
where the effect of EOF is suppressed. In this case, the addition
of an organic modifier favors the partitioning of hydrophobic
solutes into the immobile aqueous phase, which increases solute
migration times as well as the resolution.
Figure 9 illustrates the effect that methanol has on the chiral
separation of terbutaline. As the methanol concentration increases
from 5 to 25%, with the SCD concentration remaining constant,
separation between enantiomers increases from 2.6 to 21 min. The
mobility difference between terbutaline enantiomers is plotted as
a function of organic concentration in Figure 10. As can be seen
in Figure 10, the mobility difference, ∆µ, increases with increasing
organic content and the effect of acetonitrile is greater than that
of methanol.
Fig u re 7 . Effect of increasing SCD concentration on the mobility
of terbutaline enantiomers. Buffer was 25 mM phosphate (pH 3)
containing 0-40 mM SCD: coated capillary: -10 kV; 20 °C; 2, first
enantiomer; b, second enantiomer.
ing SCD concentration but is limited by broad peaks and long
migration times at the lowest extreme.
Mobility data, corrected for viscosity and EOF, were plotted
against SCD concentration using a double-reciprocal format
Binding constants in the presence of several concentrations
of MeOH and ACN were determined as described above. As in
the aqueous buffer systems, mobilities used to calculate K values
were corrected for changes in both viscosity and EOF (see Figure
2) caused by the presence of the organic modifier. Table 2 reports
(
Figure 8) and related to K according to eq 3. Experimental values
for K were 1730 ( 50 and 1590 ( 38 M-1 for the two enantiomers.
These values were in fairly good agreement with the average value
determined by the spectroscopic method and are indicative of
relatively strong binding between terbutaline and SCD.
1 2
values of K , K , and R, at each modifier concentration. As
expected, the binding constants decrease with increasing organic
concentration and again acetonitrile has a greater influence than
methanol. It was also observed that the binding constant of the
5170 Analytical Chemistry, Vol. 70, No. 24, December 15, 1998

increases from 1.09 ( 0.04 with 0% organic to 1.41 ( 0.08 and
1
.52 ( 0.18 at 10% methanol and acetonitrile, respectively.
It is important to recognize that the binding constant is related
to the free energy associated with the transfer of the analyte from
the bulk buffer to the cyclodextrin. While the energy associated
with the inclusion of the analyte in the cavity may or may not
change with bulk buffer composition, the energy of the analyte
in free solution certainly does. Thus, it seems reasonable that
the second-eluting enantiomer, which has lower affinity for the
cyclodextrin, should exhibit greater differences in binding constant
with bulk buffer composition, and thus greater selectivity, even
though both enantiomers have the same energy in the bulk buffer.
This increase in selectivity with increasing organic content may
seem counterintuitive when compared to HPLC. However, CE
can exploit potential countercurrent migration of analyte and
additive while HPLC flow is unidirectional.
Fig u re 9 . Overlaid electropherograms of terbutaline chiral separa-
tion. SCD concentration was held constant at 2% in 25 mM phosphate
buffer (pH 3): Methanol concentrations 5, 10, and 15% (w/w); coated
capillary; -10 kV; 20 °C.
CONCLUSIONS
This study has shown that, for the chiral separation of racemic
terbutaline with sulfated â-CD, separation increases with decreas-
ing selector concentration but is limited by broad peaks and long
migration times at the lowest extreme. The addition of methanol
or acetonitrile to the CE run buffer increased enantioselectivity
for terbutaline under conditions of reversed polarity and minimal
EOF. It was also found that the presence of sulfated cyclodextrin
did induce a concentration-dependent EOF despite the use of a
coated capillary. Electrophoretic mobilities required adjustment
due to the changes in viscosity caused by the presence of the
chiral selector as well as changes in the electroosmotic flow
caused by adsorption of SCD in the capillary. AES and ISE results
indicate that counterion condensation occurs with these highly
charged additives. A binding study was performed to approximate
binding constants for this system and the effect that organic
modifiers have on binding constants. Values for K determined
spectroscopically compared well with those determined by CE.
Furthermore, CE and UV difference spectroscopy revealed a
substantial decrease in the values for K upon addition of organic
modifier to the buffer.
Fig u re 1 0 . Terbutaline enantiomer mobility difference plotted
against organic composition: 2% SCD; coated capillary; -10 kV, 20
°
C; 9, acetonitrile, b, methanol.
Ta b le 2 . Bin d in g Co n s t a n t s o f Te rb u t a lin e w it h
S u lfa t e d â-Cyc lo d e x t rin a
conditions
K1
K2
R
3
3
0
% modifier
% MeOH
0% MeOH
% ACN
1.73 × 10 ( 50
1.59 × 10 ( 38
1.09 ( 0.04
1.24 ( 0.07
1.41 ( 0.08
1.28 ( 0.07
1.52 ( 0.18
3
3
5
1
5
1
1.70 × 10 ( 90
1.37 × 10 ( 21
3
3
1.59 × 10 ( 60
1.13 × 10 ( 52
3
3
1.61 × 10 ( 90
1.26 × 10 ( 20
ACKNOWLEDGMENT
3
2
0% ACN
1.32 × 10 ( 80
8.70 × 10 ( 90
The authors greatly acknowledge the support of the National
Institutes of Health First Award (Grant 1R29 GM48180-04),
Novartis Pharmaceuticals Corp. (Summit, NJ) for the generous
donation of terbutaline, Cerestar, Inc. (Hammond, IN) for the
donation of sulfated cyclodextrin, Professor J. P. Caruso and Mr.
J. Waggoner of the University of Cincinnati for use and instruction
on the ICP instrument, and Professor G. Vigh of Texas A & M
University and Professor J. M. Davis at Southern Illinois University
for helpful discussions.
a
See Experimental Section of text for conditions.
second-eluting enantiomer was affected to a greater extent than
the first enantiomer. The effect of methanol on the binding
constant was also evaluated by UV difference spectroscopy. In
the presence of 10% methanol, K was 1320 ( 50 M-1, supporting
the CE results of a decreased binding constant in the presence
of organic modifiers. Unfortunately, a more exhaustive UV study
was not possible because of insufficient sulfated cyclodextrin.
Received for review July 15, 1998. Accepted September
21, 1998.
2 1
/ K
,28 was substantially
The selectivity, defined as R ) K
affected by both modifiers. As can be seen from the table, R
AC980780I
Analytical Chemistry, Vol. 70, No. 24, December 15, 1998 5171