A Family of Single-Isomer Chiral Resolving Agents for Capillary Electrophoresis. 2. Hepta-6-sulfato-β-cyclodextrin

Article abstract of DOI:10.1021/ac970418o

A new, hydrophilic, single-isomer charged cyclodextrin, the sodium salt of hepta-6-sulfato-β-cyclodextrin has been synthesized, characterized, and used for the capillary electrophoretic separation of the enantiomers of numerous noncharged, acidic, basic, and zwitterionic analytes. Hepta-6-sulfato-β-cyclodextrin proved to be a much stronger complexing agent for all the analytes tested, in both low-pH and high-pH background electrolytes, than the previously synthesized, moderately hydrophobic heptakis-(2,3-diacetyl-6-sulfato)-β-cyclodextrin. The separation selectivities of the two single-isomer, differently functionalized charged cyclodextrins often proved to be complementary. In agreement with the predictions of the charged resolving agent migration model, separation selectivity for the noncharged analytes decreased as the concentration of hepta-6-sulfato-β-cyclodextrin was increased. For acidic, basic, and zwitterionic analytes, selectivity could increase, decrease, or pass a maximum, depending on the binding strength of the enantiomers and ionic mobilities of both the complexed and noncomplexed forms of the enantiomers.

Full text of DOI:10.1021/ac970418o

Anal. Chem. 1997, 69, 4419-4428
A Family of Single-Isomer Chiral Resolving Agents
for Capillary Electrophoresis. 2.
Hepta-6-sulfato-â-cyclodextrin
J. Bryan Vincent, Dawn M. Kirby, Thanh V. Nguyen, and Gyula Vigh*
Department of Chemistry, Texas A&M University, College Station, Texas 77845-3255
agent isomers might change from batch to batch and might result
in irreproducible separations. Therefore, we proceeded to syn-
thesize another, well-characterized, single-isomer charged cyclo-
dextrin, hepta-6-sulfato-â-cyclodextrin that (i) offers intermolecular
interactions different from the previously described first member
of the family,1 (ii) has the same functional groups on the 2- and
3-positions of the glucose moieties as native â-cyclodextrin and,
(iii) is a maximally charged strong electrolyte that can be used at
any pH without compromising either separation selectivity or
efficiency.2,4,5 This paper describes the synthesis, characterization,
and use of hepta-6-sulfato-â-cyclodextrin (HS-âCD), the most
hydrophilic member of our new, single-isomer, charged cyclo-
dextrin family.
A new, hydrophilic, single-isomer charged cyclodextrin,
the sodium salt of hepta-6 -sulfato-â-cyclodextrin has been
synthesized, characterized, and used for the capillary
electrophoretic separation of the enantiomers of numer-
ous noncharged, acidic, basic, and zwitterionic analytes.
Hepta-6-sulfato-â-cyclodextrin proved to be a much stron-
ger complexing agent for all the analytes tested, in both
low-pH and high-pH background electrolytes, than the
previously synthesized, moderately hydrophobic heptakis-
(2 ,3 -diacetyl-6 -sulfato)-â-cyclodextrin. The separation
selectivities of the two single-isomer, differently function-
alized charged cyclodextrins often proved to be comple-
mentary. In agreement with the predictions of the charged
resolving agent migration model, separation selectivity for
the noncharged analytes decreased as the concentration
of hepta-6-sulfato-â-cyclodextrin was increased. For acidic,
basic, and zwitterionic analytes, selectivity could increase,
decrease, or pass a maximum, depending on the binding
strength of the enantiomers and ionic mobilities of both
the complexed and noncomplexed forms of the enanti-
omers.
EXPERIMENTAL SECTION
Synthesis of Hepta-6-sulfato-â-cyclodextrin. The chemicals
used in the synthesis and the CE application of hepta-6-sulfato-
â-cyclodextrin were purchased from Aldrich Chemical Co. (Mil-
waukee, WI), except for â-cyclodextrin, which was a gift from
Cerastar (Hammond, IN). The sodium salt of hepta-6-sulfato-â-
cyclodextrin was synthesized according to Figure 1 by combining
known derivatization and purification steps. The first intermediate,
heptakis-[6-(tert-butyldimethyl)silyl]-â-cyclodextrin was synthe-
sized according to ref 6 and purified by gradient elution preparative
column chromatography7 on silica gel using a simple eluent
system, n-hexane/ ethyl acetate/ ethanol.6 (The 50 mm i.d., 300
mm long preparative HPLC column packed with 30 nm pore size,
10 µm irregular silica (Merck, Darmstadt, Germany) was gener-
ously loaned by Dr. Y. Y. Rawjee of Smith-Kline Beecham, King
of Prussia, PA) The second intermediate was prepared by
acetylation with acetic anhydride6 and purification by gradient
elution preparative column chromatography on silica gel using
n-hexane/ ethyl acetate/ ethanol as eluent.6 In the third step, the
protecting silyl group was removed by boron trifluoride etherate6
and the product was repurified as in the first step. The pure third
intermediate was sulfated with SO3‚pyridine,8 freed of the sodium
sulfate byproduct,1 and, finally, deacetylated by dissolving the last
intermediate in water, raising the pH of the solution to 12 with
NaOH, adding 25% methanol, and stirring the mixture for 12 h.
Once indirect UV detection CE1,9 indicated that the reaction was
complete, the reaction mixture was poured into ethanol, and the
solid was collected by filtration, washed with ethanol, and dried
In part 1 of this series,1 we described the synthesis and
capillary electrophoretic (CE) use of a new, moderately hydro-
phobic, single isomer, fully charged anionic cyclodextrin, heptakis-
(2,3-diacetyl-6-sulfato)-â-cyclodextrin. The material was found to
obey the separation characteristics predicted by the charged
resolving agent migration model (CHARM model) of CE enanti-
omer separations2 and proved successful for the separation of
enantiomers belonging to very different compound classes.1
However, it is well-known from the CE use of noncharged
cyclodextrins that different functional groups connected to the 2-
and 3-positions of the glucose moieties of the cyclodextrins result
in different enantioselectivities.3 It is reasonable to expect that
the same holds true for the charged cyclodextrins as well.
Unfortunately, the commercially available charged cyclodextrins
are complicated mixtures of isomers which differ in their degree
of substitution and their substitution patterns. It was pointed out
in ref 1 that resolving agent mixtures are not as desirable as pure,
single-isomer resolving agents, because (i) the different isomers
may possess different chiral selectivities, (ii) they may possess
different, finite complexation rates, which lead to kinetic broaden-
ing, and (iii) the composition of a particular mixture of resolving
(4) Friedl, W.; Kenndler, E. Anal. Chem. 1 9 9 3 , 65, 2003.
(5) Rawjee, Y. Y.; Vigh, Gy. Anal. Chem. 1 9 9 4 , 66, 428.
(1) Vincent, B. J.; Sokolowski, A. D.; Nguyen, T. V.; Vigh, Gy. Anal. Chem. 1997,
69, 4226.
(2) Williams, B.ÅA.; Vigh, Gy. J. Chromatogr. 1 9 9 7 , 777, 43.
(3) Fanali, S. J. Chromatogr., A 1 9 9 6 , 735, 77.
(6) Takeo, K.; Hisayoshi, M.; Uemura, K. Carbohydr. Res. 1 9 8 9 , 187, 203.
(7) Vigh, Gy.; Quintero, G.; Farkas, Gy. J. Chromatogr. 1 9 8 9 , 484, 237.
(8) Bernstein, S.; Joseph, J.; Nair, U. U.S. Patent 4,020,160, 1997.
(9) Nardi, A.; Fanalli, S., Foret, F. Electrophoresis 1 9 9 0 , 11, 774.
S0003-2700(97)00418-6 CCC: $14.00 © 1997 American Chemical Society
Analytical Chemistry, Vol. 69, No. 21, November 1, 1997 4419

Figure 1. Synthesis scheme for hepta-6-sulfato-â-cyclodextrin.
in Figure 2. The 200 MHZ 1H NMR spectrum of the final product
is shown in Figure 3; the narrow, well-defined lines also attest to
the purity of the final product.
Electrophoretic Separations Using Hepta-6 -sulfato-â-cy-
clodextrin. All CE experiments were carried out with a P/ ACE
2200 CE system (Beckman Instruments, Fullerton, CA). The
wavelength of the UV detector was set at 214 nm, the separation
temperature was 20 °C. Untreated, 25 µm i.d. fused-silica
capillaries (Polymicro Technologies, Phoenix, AZ) were used with
a nominal total length of 45 cm and nominal injector-to-detector
length of a 39 cm. The samples were pressure injected for 1 s at
5 psi.10 The power dissipation was maintained between 500 and
700 mW/ m, brought about by 12-20 kV applied potentials.
As in part 1, both a low-pH and a high-pH buffer stock solution
was prepared by adding 0.0250 mol of concentrated phosphoric
acid (for the low-pH buffer) and 0.0250 mol ethanolamine (for the
high-pH buffer) to enough deionized water (Milli-Q, Millipore,
Milford, MA) to obtain solutions of ∼0.95 L. Using a combination
glass electrode and a precision pH meter (Corning Science
Products, Corning, NY), these solutions were titrated to pH 2.5
and pH 9.5, respectively, with aqueous LiOH and aqueous
methansulfonic acid to achieve solutions with similar ionic
strengths. After the solutions were quantitatively transferred to
1 L volumetric flasks, they were diluted to the mark with deionized
water and their pH was measured again.
Figure 2. Indirect UV detection electropherogram of a typical hepta-
6-sulfato-â-CD sample: BE, 20 mM p-toluenesulfonic acid, pH 8,
adjusted with Tris; wavelength, 214 nm; applied potential, 25 kV,
thermostat temperature, 25 °C; capillary, 25 µm i.d.; 39/45 cm
effective/total length; uncoated fused silica.
in a vacuum oven at 80 °C yielding the end product, pure hepta-
6-sulfato-â-CD. The purity of the final product was determined
by indirect UV detection CE using 20 mM p-toluenesulfonic acid
(PTSA) as background electrolyte (BE), whose pH was adjusted
to 8 with tris(hydroxymethyl)aminomethane (Tris).1 The elec-
tropherogram of a typical hepta-6-sulfato-â-CD sample is shown
Figure 3. 1H NMR spectrum of a typical hepta-6-sulfato-â-CD sample. Solvent, D2O.
4420 Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

U
Analytical Chemistry, Vol. 69, No. 21, November 1, 1997 4421

4422 Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

The low-pH and the high-pH BEs, which contained 10, 15, 30,
and 50 mM HS-âCD, were prepared from the buffer stock
solutions. As in part 1,1 the external EO flow marker method
described in ref 10 was used to determine the true effective
mobility (µeNfSf A-) of 2-naphthalenesulfonic acid (NSA-) in each
HS-âCD BE (Table 1). When NSA- is added to each analyte
sample, it can serve as an internal mobility marker for the
calculation of the effective mobilities of the analytes, because the
electroosmotic flow mobility, µEO, can be obtained as µEO
)
obs
µ
-
µeNfSf A-, and the effective mobilities of the enantiomers as
NSA-
µeffR ) µobsR - µEO. Separation selectivity, R, was calculated as R
) µeffR/ µeffS, where subscript S refers to the less mobile enanti-
omer. The dimensionless electroosmotic mobility values, â, were
5
calculated as â ) µEO/ µeff
.
S
The observed peak resolution values,
Rs, were calculated as usual, Rs ) 2 (tS - tR)/ (wS - wR), where
tS and tR are the observed migration times and wS and wR are the
observed peak widths of enantiomers.
RESULTS AND DISCUSSION
Effective Electrophoretic Mobilities of 2 -Naphthalene-
sulfonate in HS-âCD BEs. Since the knowledge of accurate
NSA- effective mobilities, µeffNSA-, is crucial for the calculation of
meaningful selectivities, the µeffNSA- determinations were repeated
five times in each BE. The results are shown in the first lines of
Tables 1 (low-pH BEs) and 2 (high-pH BEs). Just as with the
heptakis(2,3-diacetyl-6-sulfato)-â-cyclodextrin (HDAS-âCD),1 the
µeffNSA- values decrease as the HS-âCD concentration is increased
because the viscosity and the ionic strength of the BE, as well as
the extent of complexation with HS-âCD, increase. However, the
µeff
values are about 10-15% larger with HS-âCD than with
NSA-
HDAS-âCD,1 indicating that HS-âCD complexes more strongly
than the acetylated analog. In the high-pH BEs, µeffNSA- is again
consistently lower than in the low-pH BEs, indicating that the high-
pH BE constituents (ethanol amine/ ethanol ammonium and
methanesulfonate) compete more strongly with NSA- for binding
to HS-âCD than the low-pH BE constituents (phosphoric acid/
dihydrogen phosphate and lithium ion), just as they did with
HDAS-âCD.1
Separation of Enantiomers in HS-âCD-Containing BEs.
A series of neutral, strong base, weak base, weak acid, and
zwitterionic enantiomers were separated with both low-pH and
high-pH BEs. Many of these compounds are the same that we
used for the characterization of HDAS-âCD1 in order to facilitate
the comparison of the complexing properties and selectivities of
the two differently funtionalized CDs. The effective mobilities of
the less mobile enantiomers (µ), the separation selectivities (R),
the measured peak resolution values (Rs), the corresponding
dimensionless EO flow values (â), and the injector-to-detector
potential drop (U) are listed in Tables 1 and 2.
For neutral analytes, both at low-pH and at high-pH, the anionic
effective mobilities are about 5-10 times higher with HS-âCD than
with HDAS-âCD (Tables 1 and 2 in ref 1), indicating much
stronger complexation with HS-âCD. Some of the increased
mobility could be due to the higher charge-to-mass ratio of HS-
âCD over that of HDAS-âCD, but the decrease in molecular weight
(∼35%) alone is not enough to explain the large increase of
mobilities. At identical charged cyclodextrin concentrations, the
separation selectivities with HS-âCD are significantly lower than
with HDAS-âCD. For example, for methyl mandelate in the low-
(10) Williams, B. A.; Vigh, Gy. Anal. Chem. 1 9 9 7 , 69, 4410-4418.
Analytical Chemistry, Vol. 69, No. 21, November 1, 1997 4423

a
U
4424 Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

Analytical Chemistry, Vol. 69, No. 21, November 1, 1997 4425

Figure 4. Typical electropherograms of neutral analytes in low-pH
BEs. The numbers next to the structures indicate the HS-âCD
concentrations (mM): applied effective potential, 17 kV; capillary, 25
µm i.d., 39/45 cm effective/total length, uncoated fused silica. For
other conditions, see the Experimental Section.
pH BE at 10 mM charged CD concentration, R ) 1.31 with HDAS-
âCD,1 but only R ) 1.10 with HS-âCD, while the effective
mobilities are -0.5 vs -7.1 (in 10-5 cm2/ Vs units), confirming
the presence of strong, but parasitic (nonselective) complexation
with HS-âCD. Greater selectivity alone does not automatically
translate to greater resolution, because â and the effective charge
of the analyte also have important effects on peak resolution.2 For
example, for (R-methyl-R-phenylsuccinimide) in the low-pH BE
at 10 mM charged CD concentration, R ) 1.05 with HS-âCD and
R ) 1.50 with HDAS-âCD. Yet, Rs ) 1.3 with HS-âCD and Rs <
0.5 with HDAS-âCD, because â ) 0 with HS-âCD and â ) -57
with HDAS-âCD. Thus, the successful use of the two differently
functionalized charged CDs, HDAS-âCD and HS-âCD, requires
very different experimental conditions.
Due to solubility limitations, weak acids could not be analyzed
at low-pH with HDAS-âCD.1 However, since HS-âCD complexes
with all the analytes more strongly than HDAS-âCD, weak acids
became soluble in HS-âCD-containing BEs and could be separated
by CE with R values hovering around R ) 1.02 and Rs values
ranging between 0.5 and 1. Flurbiprofen and ibuprofen, which
could not be separated at low-pH with HS-âCD, were at least
partially separated in the high-pH BEs with HS-âCD, similarly to
what was observed with HDAS-âCD.1
For strong bases, the cationic effective mobility decreases (and
eventually becomes anionic) as the concentration of HS-âCD is
increased. Though very similar viscosities were measured in the
high-pH BEs and the low-pH BEs, the cationic mobilities of the
strong bases are much higher in the high-pH HS-âCD BEs than
in the low-pH BEs, once again confirming that the high-pH BE
constituents compete very effectively with the analyte for the HS-
âCD. In agreement with the predictions of the CHARM model
(Figure 6 in ref 3), separation selectivity increases with the
concentration of HS-âCD as long as the component migrates
cationically; then selectivity decreases as the strong base begins
to migrate anionically. In fact, selectivity becomes so high around
the changeover point that an Rs ) 4 can be obtained even when
4426 Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

Figure 7. Typical electropherograms of weak base analytes in low-
pH BEs. Conditions as in Figure 5.
Figure 5. Typical electropherograms of weak base analytes in low-
pH BEs. Conditions as in Figure 4, except applied effective potential,
12 kV.
Figure 8. Typical electropherograms of weak acid and zwitterionic
analytes in low-pH BEs. Conditions as in Figure 4.
Figure 6. Typical electropherograms of weak base analytes in low-
pH BEs. Conditions as in Figure 5.
zwitterionic analytes. As with HDAS-âCD,1 the peak resolution
values are quite large, allowing for the optimization of these
separations to meet diverse analytical objectives.
the separation is carried out from the short side of the capillary
(effective voltage 2.1 kV, Table 1).
In general, weak bases studied here migrate anionically both
in the low-pH and the high-pH BEs: their effective mobilities
decrease as the concentration of HS-âCD is increased. The only
exception is terbutaline, which is only slightly dissociated in the
high-pH BE and behaves more or less as a neutral species. In
low-pH BEs (and for the most part, in high-pH BEs as well),
separation selectivity is more or less constant, or decreases
slightly, as the concentration of HS-âCD is increased. The only
exception is oxyphencyclamine, which has a very high anionic
mobility at very low HS-âCD concentrations, and whose selectivity
increases with the concentration of HS-âCD. This again agrees
with the predictions of the CHARM model.3
A few typical separations are shown in Figures 4-8: the
numbers next to the electropherograms indicate the concentration
of HS-âCD (in mM). All BEs were made with the low-pH stock
buffer. Figure 4 shows the separation of neutral analytes, Figures
5-7 those of the basic analytes, and Figure 8 the weak acid and
CONCLUSIONS
A new, hydrophilic, single-isomer charged cyclodextrin, the
sodium salt of hepta-6-sulfato-â-cyclodextrin has been synthesized,
characterized, and used to separate the enantiomers of neutral,
weak acid, strong base, weak base, and zwitterionic analytes in
pH 2.5 and pH 9.5 BEs. As predicted by the CHARM model of
CE enantiomer separations3 and first verified in part 1 of this series
with heptakis(2,3-diacetyl-6-sulfato)-â-cyclodextrin, separation se-
lectivity for the neutral analytes decreases as the concentration
of HS-â-CD increases. For charged analytes, R can increase,
decrease, or pass a maximum, depending on the numeric values
of the complexation constants and ionic mobilities. The dimen-
sionless electroosmotic flow mobility, â,6 greatly influences the
magnitude of peak resolution obtained with HS-â-CD. HS-â-CD
complexed more strongly with most analytes than HDAS-â-CD
Analytical Chemistry, Vol. 69, No. 21, November 1, 1997 4427

and the two materials generally offered different separation
selectivities, indicating that both materials could play an important
role in the arsenal of separation scientists engaged in CE
enantiomer separations.
also indebted to Cerastar (Hammond, IN) for the â-cyclodextrin
sample used to synthesize HS-â-CD. The authors are grateful to
Professor D. Romo of Texas A&M University for valuable
discussions concerning the synthesis process.
ACKNOWLEDGMENT
Received for review April 21, 1997. Accepted August 6,
1997.X
Partial financial support of this project by the Texas Coordinat-
ing Board of Higher Education ARP program (Project 010366-
016), the NSF-REU program (Grant CHE9322109), and Beckman
Instruments (Fullerton, CA) is gratefully acknowledged. We are
AC970418O
X Abstract published in Advance ACS Abstracts, October 1, 1997.
4428 Analytical Chemistry, Vol. 69, No. 21, November 1, 1997