Analytical- and preparative-scale isoelectric focusing separation of enantiomers

Article abstract of DOI:10.1021/ac9902749

Isoelectric focusing has been used to achieve the analytical- and preparative-scale separation of the enantiomers of amphoteric analytes. By considering the simultaneous multiple equilibria involved in the chiral recognition process, a model has been developed to describe the magnitude of the ΔpI value that develops between the enantiomers in the presence of a noncharged chiral resolving agent, such as a noncharged cyclodextrin. Theoretical analysis of the model indicates that three kinds of IEF enantiomer separations are possible: aniono-selective and cationo-selective, when only the identically charged forms of the enantiomers bind selectively to the resolving agent, and duo-selective, when the differently charged forms of the enantiomers bind selectively to the resolving agent. The model predicts that the ΔpI vs cyclodextrin concentration curves approach limiting ΔpI values which can be as large as 0.1, even when the binding constants of the enantiomers differ only by 10%. The parameters of the model can be readily determined by free solution capillary electrophoretic or pressure- mediated capillary electrophoretic experiments. The validity of the proposed model has been tested with hydroxypropyl β-cyclodextrin as resolving agent and dansyl phenylalanine as probe. Capillary IEF enantiomer separations have been achieved using both ampholytes and binary propionic acid-serine buffers (Bier's buffers). Preparative-scale IEF enantiomer separations with production rates as high as 1.3 mg/h have been achieved in an Octopus continuous free-flow electrophoretic system.

Full text of DOI:10.1021/ac9902749

Anal. Chem. 1999, 71, 3814-3820
Analytical- and Preparative-Scale Isoelectric
Focusing Separation of Enantiomers
Pavel Glukhovskiy and Gyula Vigh*
Chemistry Department, Texas A&M University, College Station, Texas 77842-3012
capillary gel electrophoresis.19 Isoelectric focusing (IEF) tech-
Isoelectric focusing has been used to achieve the analyti-
cal- and preparative-scale separation of the enantiomers
of amphoteric analytes. By considering the simultaneous
multiple equilibria involved in the chiral recognition
process, a model has been developed to describe the
magnitude of the ∆pI value that develops between the
enantiomers in the presence of a noncharged chiral
resolving agent, such as a noncharged cyclodextrin.
Theoretical analysis of the model indicates that three
kinds of IEF enantiomer separations are possible: aniono-
selective and cationo-selective, when only the identically
charged forms of the enantiomers bind selectively to the
resolving agent, and duo-selective, when the differently
charged forms of the enantiomers bind selectively to the
resolving agent. The model predicts that the ∆pI vs
cyclodextrin concentration curves approach limiting ∆pI
values which can be as large as 0 .1 , even when the
binding constants of the enantiomers differ only by 1 0 %.
The parameters of the model can be readily determined
by free solution capillary electrophoretic or pressure-
mediated capillary electrophoretic experiments. The va-
lidity of the proposed model has been tested with hydroxy-
propyl â-cyclodextrin as resolving agent and dansyl phe-
nylalanine as probe. Capillary IEF enantiomer separations
have been achieved using both ampholytes and binary
propionic acid-serine buffers (Bier’s buffers). P repara-
tive-scale IEF enantiomer separations with production
rates as high as 1 .3 mg/ h have been achieved in an
Octopus continuous free-flow electrophoretic system.
niques,20 including, more recently, capillary isoelectric focusing
(cIEF),21 have been used extensively to separate amphoteric
compounds on the basis of their different isoelectric points (pI
values). However, to the best of our knowledge, there is only one
published report, by Righetti’s group, on the successful use of
IEF, in the slab gel format, for the separation of the enantiomers
of dansyl amino acids.22 By adding 40 mM â-cyclodextrin (CD), 4
M urea, and 10% methanol to an immobilized pH-gradient gel,
they were able to separate, using 6-h-long focusing times, the
enantiomers of dansyl isoleucine (DNS-Ile), phenylalanine (DNS-
Phe), and tryptophan (DNS-Trp). The pI values of the two CD-
complexed DNS amino acid enantiomers were found to differ by
as much as 0.1. They argued that the enantiomer separation was
feasible because CD altered the base strength (pKb value) of the
dimethylamino group of the DNS moiety more for
L-DNS-Phe,
than for -DNS-Phe, but no equation was presented to describe
D
the CD concentration dependence of the complexation-generated
∆pI value. Though they speculated that IEF enantiomer separa-
tions should be feasible in multicompartmental electrolyzers
equipped with Immobiline membranes,24 no such follow-up report
appeared in the literature.
The objectives of this paper are to (i) present an explicit
relationship that predicts the magnitude of the ∆pI value that can
be generated between the enantiomers upon the addition of a
(6) St. Claire, R. L. Anal. Chem. 1 9 9 6 , 68, 569R-586R.
(7) Camilleri, P. Chem. Commun. 1 9 9 6 , 16, 1851-1858.
(8) Bressolle, F.; Audran, M.; Pham, T. N.; Vallon, J. J. J. Chromatogr., B 1 9 9 6 ,
687, 303-336.
(9) Vespalec, R.; Bocek, P. Electrophoresis 1 9 9 7 , 18, 843-852.
(10) Fanali, S. J. Chromatogr., A 1 9 9 7 , 792, 227-267.
(11) Vigh, Gy.; Sokolowski, A. D. Electrophoresis 1 9 9 7 , 18, 2305-2310.
(12) Landers, J. P. Handbook of Capillary Electrophoresis, CRC Press: Boca Raton,
1997.
As indicated by numerous recent reviews and monographs,1-15
capillary electrophoretic (CE) enantiomer separations underwent
spectacular development over the past few years. CE is a logical
choice for the separation of water-soluble weak electrolyte enan-
tiomers because, due to the high innate separation efficiency of
CE, the enantiomers can be resolved even when the separation
selectivity is as low as R ) 1.005. Successful enantiomer separa-
tions have been achieved early on in all major electrophoretic
operation modes: in micellar electrokinetic chromatography,16
isotachophoresis,17 free solution capillary electrophoresis,18 and
(13) Chankvetadze, B. Capillary Electrophoresis in Chiral Analysis; J. Wiley: New
York, 1997.
(14) Beale, S. C. Anal. Chem. 1 9 9 8 , 70, 279R-300R.
(15) Khaledi, M., Ed. High Performance Capillary Electrophoresis: Theory,
Techniques and Applications; J. Wiley: New York, 1998.
(16) Terabe, S.; Ozaki, H.; Ando, T. J. Chromatogr. 1 9 8 5 , 348, 39-51.
(17) Jelinek, I.; Snopek, J.; Smolkova-Keulemansova, E. J. Chromatogr. 1 9 8 7 ,
405, 379-384.
(18) Fanali, S. J. Chromatogr. 1 9 8 9 , 474, 441-446.
(19) Guttman, A.; Paulus, A.; Cohen, A. S.; Grinberg, N.; Karger, B. L. J.
Chromatogr. 1 9 8 8 , 448, 41-53.
(1) Chankvetadze, B.; Endresz, G.; Blaschke, G. Chem. Soc. Rev. 1 9 9 6 , 25,
141-152.
(20) Righetti, P. G. Isoelectric Focusing: Theory, Methodology and Applications;
(2) Nishi, H.; Terabe, S. J. Chromatogr., A 1 9 9 6 , 735, 3-27.
(3) Nishi, H. J. Chromatogr., A 1 9 9 6 , 735, 57-76.
(4) Fanali, S. J. Chromatogr., A 1 9 9 6 , 735, 77-121.
(5) Nguyen, N. T.; Siegler, R. W. J. Chromatogr., A 1 9 9 6 , 735, 123-150.
Elsevier: Amsterdam, The Netherlands, 1983.
(21) Hjerten, S.; Zhu, M. D. J. Chromatogr. 1 9 8 5 , 346, 265-270.
(22) Righetti, P. G.; Ettori, C.; Chafey, P.; Wahrmann, J. P. Electrophoresis 1 9 9 0 ,
11, 1-4.
3814 Analytical Chemistry, Vol. 71, No. 17, September 1, 1999
10.1021/ac9902749 CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/24/1999

NR- + D a NR-D
(6)
(7)
noncharged resolving agent, (ii) analyze this relationship to
determine the advantageous regimes of IEF enantiomer separa-
tions, (iii) demonstrate the feasibility of analytical-scale cIEF
enantiomer separations using both ampholyte and binary Bier
buffers,25 and (iv) demonstrate the feasibility of preparative-scale
IEF enantiomer separations using the Octopus32-34 continuous
free-flow electrophoretic unit. The history and recent advances
of continuous free-flow electrophoresis were extensively reviewed
recently.33 The design and characteristics of the Octopus unit were
also discussed in detail,32 and its long-term stability has been
demonstrated.34
H+NRH + D a H+NRHD
The equilibrium expressions for reactions 1-7 are
Ka,R ) [NR-][H3O+]/ [H+NR-]
(8)
(9)
Kb,R ) [H+NRH][OH-]/ [H+NR-]
Kw ) [H3O+][OH-]
(10)
(11)
(12)
(13)
THEORY
KH
) [H+NR-D]/ [H+NR-][D]
+
-
NR D
1 . Complexation-Induced Shifts in the Effective pI Values
of Enantiomers. To account for the effects of the noncharged
cyclodextrin (D) concentration of the background electrolyte (BE)
on the effective pI values of the individual enantiomers, both
protonation equilibria and complexation equilibria must be con-
sidered simultaneously. Let the BE contain a buffer system to
control the pH and a noncharged cyclodextrin, D, to create an
anisotropic environment. Let the enantiomers of an amphoteric
analyte be NRH and NSH. Cyclodextrin will interact with both
the buffer components and the analytes. When the buffer
concentration is much higher than that of D, and the concentration
of D is much higher than that of the analytes, the concentration
and original species distribution of D will remain practically the
same whether analytes are present or absent. Therefore, in a first
approximation, the equilibria between the buffer components and
D can be omitted from consideration.
KNR D ) [NR-D]/ [NR-][D]
-
KH
) [H+NRHD]/ [H+NRH][D]
+
NRHD
The mass balance equations for the neutral, negatively charged,
and positively charged R-related species are
cH - ) [H+NR-] + [H+NR-D]
(14)
(15)
(16)
+
NR
cNR- ) [NR-] + [NR-D]
cH
) [H+NRH] + [H+NRHD]
+
NRH
The respective mole fractions of the neutral, negatively charged
and positively charged noncomplexed species are
Once in solution, both solute enantiomers, NRH and NSH,
participate in protic equilibria according to eqs 1-4. (Only the
equations pertaining to the R enantiomer are shown, but identical
expressions exist for the S enantiomer as well.)
RH - ) [H+NR-]/ c
(17)
(18)
(19)
+
+
-
NR
NR
H
RNR- ) [NR-]/ c
-
NR
NRH a H+NR-
(1)
(2)
RH
) [H+NRH]/ cH
NRH
+
+
NRH
H+NR- + H2O a H3O+ + NR-
Analytical expressions can be obtained from eqs 8-19 for
[H+NR-], [NR-], and [H+NRH] yielding
H+NR- + H2O a OH- + H+NRH
2H2O a H3O+ + OH-
(3)
(4)
RH - ) 1/ (1 + K NR D[D])
(20)
(21)
(22)
+
+
H
-
NR
RNR- ) 1/ (1 + KNR D[D])
-
Cyclodextrin will complex with all three forms of the enantiomers:
H+NR- + D a H+NR-D
(5)
RH
) 1/ (1 + KH NRHD[D])
+
+
NRH
In the absence of a noncharged complexing agent, the ratio of
the acid and base dissociation constants of the R enantiomer
becomes
(23) Righetti, P. G.; Wenisch, E.; Faupel, M. J. Chromaotgr. 1 9 8 9 , 475, 293-
309.
(24) Rawjee, Y. Y. Thesis, Texas A&M University, College Station, 1994.
(25) Rawjee, Y. Y.; Staerk, D. U.; Vigh, Gy. J. Chromatogr. 1 9 9 3 , 635, 291-306.
(26) Rawjee, Y. Y.; Williams, R. L.; Vigh, Gy. J. Chromatogr. 1 9 9 3 , 652, 233-
245.
(27) Rawjee, Y. Y.; Vigh, Gy. Anal. Chem. 1 9 9 4 , 66, 619-627.
(28) Bier, M.; Ostrem, J.; Marques, R. B. Electrophoresis 1 9 9 3 , 14, 1011-1018.
(29) Cai, H.; Vigh, Gy. Anal. Chem. 1 9 9 8 , 70, 4640-4643.
(30) Glukhovskiy, P.; Vigh, Gy. Electrophoresis 1 9 9 8 , 19, 3166-3170.
(31) Righetti, P. G.; Gelfi, C.; Perego, M.; Stoyanov, A. V.; Bossi, A. Electrophoresis
1 9 9 7 , 18, 2145-2153.
(32) Weber, G.; Bocek, P. Electrophoresis 1 9 9 6 , 17, 1896-1910.
(33) Krivankova, L.; Bocek, P. Electrophoresis 1 9 9 8 , 19, 1064-1074.
(34) Weber, G.; Bauer, J. Electrophoresis 1 9 9 8 , 19, 1104-1109.
Ka,R [NR-][H3O+]2
)
(23)
[H+NRH]Kw
Kb,R
In the isoelectric point, [H+NRH] ) [NR-] and pH ) pI, which
leads to
pIR ) (1/ 2)(pKa,R + (14 - pKb,R))
(24)
Analytical Chemistry, Vol. 71, No. 17, September 1, 1999 3815

In the presence of a noncharged complexing agent, the ratio of
the acid and base dissociation constants of the R enantiomer
becomes
1 + KNS D[D]
-
1
2
∆pI′
)
log
(32)
R,S
(
)
1 + KNR D[D]
-
Ka,R [NR-][H3O+]2
In a cationo-selective IEF enantiomer separation KNR-D
KNS-D; i.e., the negatively charged enantiomers complex identically
)
)
(25)
[H+NRH]Kw
Kb,R
with the resolving agent. This also simplifies eq 31 to
Equation 25 can be rewritten with eqs 17-22 as
Ka,R RNR-c -[H O+]2
1 + KH NRHD[D]
+
1
2
∆pI′
)
log
(33)
R,S
(
)
1 + KH NSHD[D]
+
NR
3
)
(26)
Kb,R RH NRHcH NRHKw
+
+
The functions described by eqs 32 and 33 are similar: as the
noncharged resolving agent concentration is increased toward
infinity, the absolute value of ∆pI′R,S increases from zero to the
-
+
If the pH at which cNR ) cH
is called pI′R, the effective pI for
NRH
enantiomer R in the presence of a noncharged complexing agent,
then eq 26 becomes
limiting value, ∆pI′R,Smax. When KH+NRHD[D] . 1, ∆pI′R,S
)
∆pI′R,Smax ) 1/ 2 log(KH+NRHD/ KH+NSHD). Figure 1 shows the ∆pI′R,S
vs [D] function calculated by eq 33; the top curve was obtained
with KH+NRHD ) 114 and KH+NSHD ) 150 (estimated complexation
RNR
-
1
2
pI′R ) pIR + log
(27)
RH
NRH
+
constant values for a DNS-Phe type compound24), yielding
max
∆pI′R,S
) -0.06.
max
The rate with which ∆pI′R,S approaches ∆pI′R,S depends on
the magnitude of KH+NRHD and KH+NSHD. This effect can be studied
by introducing a new variable, FR,S+, the ratio of the complexation
constants:
Similarly, pI′S, the effective pI for enantiomer S in the presence
of a noncharged complexing agent becomes
RNS
-
1
2
pI′S ) pIS + log
(28)
RH
NSH
+
FR,S+ ) K NRHD/ KH
(34)
+
+
NSHD
H
Thus, the complexation-induced effective pI difference between
the enantiomers, ∆pI′R,S, becomes
With this definition, eq 33 can be rewritten as
RNR RH
1 + FR,S+K NSHD[D]
-
+
+
+
1
2
NSH
1
2
H
∆pIR′ ,S ) pI′ - pI′ ) pIR -pIS + log
(29)
(30)
(31)
∆pI′
)
log
(35)
R
S
R,S
(
)
(
)
RNS RH
-
1 + KH NSHD[D]
NRH
+
Since pIR ) pIS , ∆pI′R,S becomes
In Figure 1, the ∆pI′R,S vs [D] function is plotted at a constant
FR,S+ value (FR,S+ ) 0.76) for increasing KH+NSHD values (150 to
750 to 1500). ∆pI′R,Smax remains the same, but the rate with which
this limiting value is approached increases as the complexation
constant is increased 5-fold and then 10-fold. A cyclodextrin
concentration as low as 15 mM can generate ∼90% of the
theoretically possible complexation-induced ∆pI′ value, even for
a moderately strong binding constant (KH+NSHD ) 750).
It is instructive to see how ∆pI′R,Smax changes as FR,S+ is varied,
i.e., what kind of enantiomer binding strength difference leads to
(Figure 2). Though the relationship is
slightly nonlinear, the larger the FR,S+, the greater the ∆pI′R,S
RNR RH
-
+
+
1
2
NSH
∆pIR′ ,S
)
log
(
)
RNS RH
-
NRH
By substituting eqs 20-22 into eq 30, we obtain
1 + KNS D[D] 1 + KH NRHD[D]
-
+
1
2
pIR′ ,S
)
log
(
)
1 + KNR D[D] 1 + KH NSHD[D]
-
+
max
what kind of ∆pI′R,S
max
This means that, in chiral IEF separations, the resolving agent-
induced ∆pI′R,S depends on both the material characteristics of
the solutes (the complexation constants of the negatively and
positively charged enantiomers) and the operating variable: the
resolving agent concentration.
.
In the 0.9 < FR,S+ and 1.1 < FR,S+ ranges, which cover the realistic,
modest enantiomer binding strength differences of ∼10% or
max
greater, ∆pI′R,S
is 0.03 or larger.
In duo-selective IEF enantiomer separations, both the anionic
and the cationic forms of the amphoteric enantiomers complex
differently with cyclodextrin, i.e., KNR-D * KNS-D and KH+NRHD
2 . Discussion of the ∆pI′R,S Model. It can be seen from the
analysis of eq 31 that, depending on the relative magnitudes of
*
the complexation constants, KNR-D, KNS-D, KH+NRHD, and KH+NSHD
there are three major types of chiral IEF separations: aniono-
selective, cationo-selective, and duo-selective separations.
,
K
H+NSHD. Therefore, eq 31 cannot be simplified. As in the case of
duo-selective CE enantiomer separations,25-27 the first and the
second multiplier terms in the argument of the logarithm function
in eq 31 have opposing effects. Therefore, depending on the
numeric values of KNR-D, KNS-D, KH+NRHD, and KH+NSHD, the ∆pI′R,S
vs [D] function will have two distinct shapes.
In an aniono-selective IEF enantiomer separation KH+NRHD
)
KH+NSHD; i.e., the positively charged enantiomers complex identi-
cally with the resolving agent. This simplifies eq 31 to
3816 Analytical Chemistry, Vol. 71, No. 17, September 1, 1999

Figure 3. Effective pI difference for the enantiomers as a function
of the cyclodextrin concentration in an antagonistic duo-selective IEF
enantiomer separation. Dash-dot line: KNS-D ) 186, FR,S- ) 0.76,
KHNSH+D ) 150, and FR,S+ ) 0.736. Dotted line: KNS-D ) 930, FR,S-
) 0.76, KHNSH+D ) 750, and FR,S+ ) 0.736. Solid line: KNS-D ) 1860,
FR,S- ) 0.76, KHNSH+D ) 1500, and FR,S+ ) 0.736.
Figure 1. Effective pI difference as a function of the cyclodextrin
concentration in a cationo-selective IEF enantiomer separation. For
all curves, FR,S+ ) 0.76. Dash-dot line, KH+NSD ) 150. Dotted line,
KH+NSD ) 750. Solid line, KH+NSD ) 1500.
where
KNS
KH
KNR D
-
+
-
D
NRHD
+
)
+
1 + KNS D[D] 1 + KH NRHD[D] 1 + KNR D[D]
-
+
-
KH
+
NSHD
(36)
1 + KH NSHD[D]
+
i.e., where the positive root(s) of the quadratic equation obtained
from the rearrangement of eq 36 are located:
[D]2[KH NRHDKNR D(KNS - KH NSHD) + KH NSHDKNS
×
+
-
-
+
+
-
D
D
(KH
- KNR D)] + 2[D](KH NRHDKNS
-
-
D
+
-
+
NRHD
Figure 2. Relationship between the ∆pI′max value and the com-
plexation constant ratio (FR,S+) for a cationo-selective IEF enantiomer
separation.
K
NR DKH NSHD) + (KH
- KNR D) -
NRHD
-
+
+
-
(KH
- KNS D) ) 0 (37)
+
-
NSHD
max
When FR,S+ > 1 and FR,S- > 1, or FR,S+ < 1 and FR,S- < 1, the
first and second terms act antagonistically (they counteract each
∆pI′R,S
depends only on the value of the FR,S+/ FR,S- ratio: the
max
more this ratio differs from unity, the larger the ∆pI′R,S value.
When FR,S+ > 1 and FR,S- < 1 or FR,S+ < 1 and FR,S- > 1, the
first and second terms in the argument of the logarithm in eq 31
act cooperatively: the ∆pI′R,S values will be much higher at any
cyclodextrin concentration than in a corresponding aniono-
selective or cationo-selective case. Obviously, this is the most
desirable scenario. The shape of the ∆pI′R,S vs [D] function is
other’s influence): there is a finite cyclodextrin concentration at
max
which ∆pI′R,S is at its maximum. This ∆pI′R,S
is smaller than
the value one would observe in an aniono-selective or cationo-
selective case with the same FR,S- or FR,S+ value. The shape of the
∆pI′R,S vs [D] function is shown in Figure 3 for a realistic set of
K
NR-D, KNS-D and KH+NRHD, KH+NSHD values. (The complex forma-
tion constants for the anionic species are measured values24 for
DNS-Phe with hydroxypropyl â-cyclodextrin, KNR-D ) 137 and
KNS-D ) 186; the complex formation constants for the cationic
species are estimated values, KH+NRHD ) 114 and KH+NSHD ) 150.)
For the three curves, the numeric values of the formation
constants increase in the 1:5:10 ratio, but the FR,S values are the
same: FR,S+ ) 0.736 and FR,S- ) 0.76. Obviously, ∆pI′R,Smax remains
the same in all three cases, but the optimum CD concentration
becomes lower as the complex formation constants increase, just
as in a duo-selective free solution CE enantiomer separation.25-27
The optimum concentration is at the point where the first
derivative of the ∆pI′R,S (D) function is equal to zero. This occurs
shown in Figure 4 for FR,S- ) 0.76, FR,S+ ) 1.359 and KNS-D )
186, 930, and 1860, KHNSH+D ) 150, 750, and 1500 (the same
numeric values as in Figure 3, except that the R and S subscripts
are interchanged). The shape of the curves is similar to that of
an aniono-selective or cationo-selective ∆pI′R,S vs [D] curve. As
max
the K values increase, ∆pI′R,S
∆pI′R,S
is approached more rapidly:
max
depends only on the value of the multiple of FR,S+ and
F
R,S-. The more this multiple differs from unity, the larger the
max
∆pI′R,S
value. In such a system, ∆pI′R,S values as large as 0.1
can be realized easily.
Obviously, there is yet another, theoretically possible case. This
occurs when both the anionic and the cationic forms of the two
Analytical Chemistry, Vol. 71, No. 17, September 1, 1999 3817

Figure 5. Simplified schematic of the Octopus free flow electro-
phoretic system.
Figure 4. Effective pI differences for the enantiomers as a function
of the cyclodextrin concentration in a cooperative duo-selective IEF
enantiomer separation.. Dash-dot line: KNS-D ) 186, FR,S- ) 0.76,
KHNSH+D ) 150, and FR,S+ ) 1.359. Dotted line: KNS-D ) 930, FR,S-
) 0.76, KHNSH+D ) 750, and FR,S+ ) 1.359. Solid line: KNS-D ) 1860,
FR,S- ) 0.76, KHNSH+D ) 1500, and FR,S+ ) 1.359.
acid binary Bier buffers28 was determined using the modified five-
band PreMCE technique.29,30 The serine-propionic acid BEs used
in the preparative-scale IEF separations contained 0.1% (w/ w)
hydroxypropylmethyl cellulose, HPMC (Aldrich, Milwaukee, WI),
and 30 mM HPâCD. The enantiomeric purity of the fractions
collected in the preparative IEF runs was determined in a 200
mM iminodiacetic acid buffer,31 which contained 200 mM HPâCD.
Equipment. All analytical-scale CE separations were com-
pleted on a P/ ACE 5510 CE unit (Beckman Instruments, Fullerton,
CA). Its UV detector was operated at 340 nm for the ampholyte
pH 3-5 cIEF measurements and at 214 nm for all other BEs. All
ampholyte-based separations and IDA buffer-based separations
were carried out in Ld ) 19 cm, Lt ) 26 cm, 25 µm i.d., 150 µm
o.d. uncoated fused-silica capillaries (Polymicro Technologies,
Phoenix, AZ). All cIEF separations with Bier’s buffers were carried
out in Ld ) 37 cm, Lt ) 44 cm, 25 µm i.d., 150 µm o.d. uncoated
fused-silica capillaries. The capillaries were thermostated at 25
°C.
All preparative-scale IEF separations were completed with an
Octopus continuous free-flow electrophoretic unit32-34 (Dr. Weber
GmbH, Kirchheim-Heimstetten, Germany). The schematic of the
unit is shown in Figure 5. The unit is equipped with a pair of
anolyte recirculating ports, a pair of catholyte recirculating ports,
seven independently fed BE inlets, a central sample inlet, and a
counterflow inlet. All these ports are fed by a variable rate,
multichannel peristaltic pump. There are 96 sample collection
ports at the exit end of the separation chamber that offer a lateral
resolution of 1.04 mm/ collection port. The 400-µm-thick IEF
separation chamber spacer was used in each experiment. The
chamber coolant was thermostated at 10 °C. The residence time
in the chamber was adjusted to 12.5 min; the sample feed rate
was set at 3 mL/ h. Fraction collection was continued until 2-mL
aliquots were obtained at each port; these aliquots were analyzed
for enantiomeric purity as described in Procedures.
enantiomers complex identically; i.e., KNR-D ) KNS-D and KH+NRHD
) KH+NSHD . Clearly, ∆pI′R,S ) 0 and there can be no IEF
enantiomer separation.
Often, the pKa and pKb (or pI) values are known for the
enantiomers, but the model parameters are not. However, reason-
able parameter estimates can be obtained from free solution CE
experiments (or PreMCE experiments) as discussed in refs 25-
27. Once the complex formation constants are known, the type
of the separation can be established (aniono-selective, cationo-
selective, or duo-selective), the best cyclodextrin concentration
max
can be determined and the ∆pI′R,S
value can be calculated.
Naturally, the model developed here applies not only for un-
charged cyclodextrins as resolving agents, but for any other
noncharged resolving agent (such as maltodextrins, chiral crown
ethers, etc.), as long as the interactions can be described by
stoichiometric equilibrium reactions. The model covers the
complexation-assisted IEF separation of nonenantiomeric analytes
as well, with the added complexity that in such cases it is often
true that the original pI values of the two analytes are not exactly
identical: pIR * pIS .
Since protein cIEF separations have been achieved for ∆pI
values smaller than 0.03-0.1, the ∆pI′R,S values predicted here,21
cIEF enantiomer separationssand even preparative IEF enanti-
omer separationssshould be feasible.
EXPERIMENTAL SECTION
Chemicals. Both polydisperse ampholytes (ampholyte pH
3-5, Sigma, St. Louis, MO) and Bier’s serine-propionic acid
binary buffers28 (OptiFocus, Protein Technologies, Tucson, AZ)
were used as IEF background electrolyte components. Their pH
range covered the pI value of the uncomplexed enantiomers.
Dansylglycine (DNS-Gly), racemic and pure enantiomer dan-
sylphenylalanine (DNS-Phe) were obtained from Sigma. Hydroxy-
propyl â-cyclodextrin (HPâCD) was obtained from Cerastar
(Hammond, IN). All solutions were freshly prepared using
deionized water from a Millipore Q unit (Millipore, Milford, MA).
The pI value of uncomplexed DNS-Phe in the serine-propionic
P rocedures. The pI value of the uncomplexed DNS-Phe in
the serine-propionic acid Bier buffers28 was determined using
the five-band PreMCE method30 and was found to be pI ) 3.65.
The cIEF separations with 2.0% ampholyte pH 3-5, and 15, 30,
60, or 120 mM HPâCD, were obtained at 18 kV in electroosmotic
flow mobilization mode, with UV detection at 340 nm.
The cIEF separations with the 0.1% HPMC, 30 mM HPâCD,
serine-propionic acid binary Bier buffers were obtained by first
filling the capillary with a 0.1% HPMC, 30 mM HPâCD, 90 mM
3818 Analytical Chemistry, Vol. 71, No. 17, September 1, 1999

serine-10 mM propionic acid mixture, pH 3.9, using the injection
pressure of the P/ ACE. Next, a 1-s-wide band of nitromethane
marker, dissolved in the pH 3.9 BE was injected. Then, this band
was moved into the capillary by pushing it with a 0.1% HPMC, 30
mM HPâCD, 40 mM serine-60 mM propionic acid mixture, pH
3.4, using the injection function of the P/ ACE. The band was
stopped 9.25 cm away from the detector window. Next, a 1-s-wide
band of the DNS-Phe sample, dissolved in the pH 3.4 BE was
injected. Then, it was pushed into the capillary with a 0.1% HPMC,
30 mM HPâCD, 35 mM serine-65 mM propionic acid mixture,
pH 3.3, using the injection function of the P/ ACE and was stopped
27.75 cm away from the detector window. The detector-side vial
(cathode vial) was filled with the pH 3.9 BE; the injector side vial
(anode vial) was filled with the pH 3.3 BE. The separation potential
(25 kV) was turned on for 48 min, followed by the injection of a
1-s-wide band of nitromethane, dissolved in the pH 3.3 BE. Finally,
the content of the entire capillary was pushed out, using the pH
3.3 BE and the injection pressure, and the UV detector trace was
recorded. Since the push velocity is constant during the last step,
the time variable can be directly translated into position in the
capillary.
Figure 6. EOF-mobilized cIEF separation of the enantiomers of
DNS-Phe in 2.0% ampholyte pH 3-5, 30 mM HPâCD background
electrolyte. For other conditions, see text.
The preparative, continuous free-flow IEF enantiomer separa-
tions were completed by first filling the Octopus separation
chamber with deionized water and removing all air. Then, the
anode compartment was filled with 40 mM serine-360 mM
propionic acid and the cathode compartment was filled with 360
mM serine-40 mM propionic acid. The inlet ports were pumped
at 13.5 mL/ h with 0.1% HPMC, 30 mM HPâCD, x mM serine/ y
mM propionic acid/ pH z BEs as follows: first port, 50/ 150/ 3.37;
second port, 60/ 140/ 3.42; third port, 30/ 130/ 3.49; fourth port, 80/
120/ 3.55; fifth port, 90/ 110/ 3.59; sixth port, 100/ 100/ 3.63; seventh
port, 110/ 90/ 3.70. The sample, 2 mM DNS-Phe, dissolved in the
pH 3.55 BE, was injected through the central sample port, at a
rate of 3 mL/ h. The counterflow, 0.1% HPMC in deionized water,
was pumped at 55.0 mL/ h. Once the flows were established,
1700-V potential was applied across the 10-cm-wide separation
channel and the current was monitored; it became steady in ∼0.5
h. Fractions of 2 mL were then collected in 96-well deep titer
plates, and the pH of each fraction was determined using a
combined glass electrode and model 05669-20 precision pH meter
(Cole-Parmer, Vernon Hills, IL).
Then, 200 µL of 8.4 mM DNS-Gly was added to each 2-mL
aliquot taken from the collected fractions and passed through
methanol-activated, C18 solid-phase extraction cartridges (Adsor-
bex, E. Merck, Darmstadt, Germany). The cartridges were washed
with 1.5 mL of deionized water and then eluted with 1 mL of
methanol. The methanol solutions were evaporated to dryness in
an AES1010 SpeedVac drier (Savant Industries, Holbrook, NY),
reconstituted in 300 µL of 200 mM IDA, 200 mM HPâCD BE,
and analyzed for enantiomeric purity on the P/ ACE.
Figure 7. cIEF separation of the enantiomers of DNS-Phe in Bier’s
serine-propionic acid, 0.1% HPMC, 30 mM HPâCD background
electrolyte. Separation time: 48 min, 25 kV, pressure mobilization.
For other conditions, see text.
3.6, and 2.9 cm, respectively, in the 10, 30, 60, and 120 mM HPâCD
BEs. The electropherogram obtained in the 30 mM HPâCD, 2.0%
pH 3-5 ampholyte BE with electroosmotic mobilization is shown
in Figure 6. Though the pH 3-5 ampholyte BE permits the
analytical-scale separation, the separated enantiomer bands also
contain the ampholytic components. This is undesirable in a
preparative application.
Therefore, the cIEF separation was repeated using 30 mM
HPâCD serine-propionic acid binary Bier buffers, as shown in
Figure 7. To follow more easily what happened during the
separation, the time axis is replaced by the distance (from the
detector window) axis. The neutral marker, which was placed 9.25
cm away from the detector window and marked the boundary
between the pH 3.9 and pH 3.4 buffers, is now 4.8 cm away from
the window. This 4.4 cm movement was caused by the residual
electroosmotic flow (EOF) during the 48-min-long separation. The
RESULTS AND DISCUSSION
The cIEF separations of the enantiomers of DNS-Phe were
obtained in 15, 30, 60, and 120 mM HPâCD, 2.0% ampholyte pH
3-5 BEs. As predicted by eq 31 for an antagonistic duo-selective
case, the two DNS-Phe enantiomer bands were well separated
from each other and there was not much change in the ∆pI′ values
as the HPâCD concentration was varied from 15 to 120 mM. The
enantiomer bands were separated from each other by 3.6, 3.4,
D
- and
L-DNS-Phe enantiomers were originally 18.5 cm away from
the nitromethane neutral marker. For the
D
-enantiomer, this
distance decreased to ∼6 cm; the high-pH side of the peak is
sharper, the low-pH side is broader, reflecting focusing. For the
Analytical Chemistry, Vol. 71, No. 17, September 1, 1999 3819

mixture can be collected in channels 48-52 (over a distance of
∼0.4 cm). With the electric field on, both bands are only five
channels wide (covering a distance of ∼0.5 cm), indicating that
the detrimental electrohydrodynamic band-broadening processes35
are well controlled in the system. The accuracy of the analytical
system is good: the collected fractions were found to contain 1.988
mg of L-DNS-Phe and 2.012 mg of D-DNS-Phe, vis-a`-vis the added
2.0-mg quantities. This represents a production rate of 1.33 mg/
h, with an enantiomeric excess of better than 99.99% in the
combined fractions.
CONCLUSIONS
A rigorous theoretical model has been developed to describe
chiral IEF separations that rely on complexation-generated pI′
differences between the enantiomers. Though all the model
parameters can be determined by straightforward PreMCE or
conventional CE experiments, only the pI of the uncomplexed
analyte need to be known to set up successful IEF separations.
Since the predicted ∆pI′ values are about 0.05-0.1, narrow,
nonnatural pH gradients generated by simple binary buffers, such
as Bier’s buffers, can be used to resolve the cyclodextrin-
complexed enantiomers. This simplifies the downstream process-
ing of the collected fractions prior to their intended end use. With
the Octopus continuous free-flow electrophoretic unit, production
rates of 1.33 mg/ h yielding an enantiomeric excess of better than
99.99% have been demonstrated.
Figure 8. Preparative continuous free-flow IEF separation of the
enantiomers of DNS-Phe using Bier’s serine-propionic acid, 0.1%
HPMC, 30 mM HPâCD background electrolyte in the Octopus unit.
Residence time: 12.5 min, 1700 V. For other conditions, see text.
L
-enantiomer, the distance decreased to 19 cm; the low-pH side
of the peak is sharper, once again reflecting focusing. In 48 min,
the - and -enantiomers became separated by 13 cm. When the
D
L
focusing time was almost doubled to 90 min, the neutral marker
was swept past the detector by the residual EOF, but the distance
between the
D
- and
L
-enantiomers increased only by ∼1 cm, to a
total of 14 cm, indicating that the final migration position in the
pH gradient was almost completely achieved. When the focusing
time was further increased to 120 min, even the
swept past the detector by the EOF.
D-enantiomer was
ACKNOWLEDGMENT
The authors are indebted to Dr. E. Pungor and Berlex
Figure 8 shows the enantiomeric analysis results of the fraction
streams collected during the continuous free-flow preparative-scale
IEF separation in the Octopus unit. The pH profile is superim-
Biosciences (Richmond, CA) for partial financial support of this
project, to Beckman Instruments (Fullerton, CA) for the loan of
the P/ ACE unit, CeraStar Inc. for the HPâCD sample, and Dr. G.
Weber for valuable discussions concerning the Octopus unit. Initial
cIEF experiments with carrier ampholytes (not shown here) by
former graduate students E.T. Horton, A.D. Sokolowski, and B.A.
Williams are acknowledged.
posed on the graph (right axis). It shows that the
is located at around pH 3.54 where pH varies evenly from fraction
to fraction. The -DNS-Phe band is located in an eight-fraction-
wide constant pH region, at pH 3.59. The band shape of -DNS-
L-DNS-Phe band
D
L
Phe indicates that there is good focusing on the low-pH side. With
the 12.5-min-long residence time in the separation chamber, there
Received for review March 15, 1999. Accepted June 15,
1999.
is not enough time for
pH zone and approach its final focusing position in channels
somewhere above 60. Nevertheless, the -DNS-Phe enantiomer
is completely resolved from the -DNS-Phe enantiomer. When
there is no electric field applied, the racemic - and -DNS-Phe
D-DNS-Phe to migrate out of the constant
AC9902749
D
L
(35) Rhodes, P. H.; Snyder, R. S.; Roberts, G. O.; Baygents, J. C. Appl. Theor.
Electrophor. 1 9 9 1 , 2, 87-98.
D
L
3820 Analytical Chemistry, Vol. 71, No. 17, September 1, 1999