Upload
others
View
12
Download
0
Embed Size (px)
Citation preview
Chiral separations using capillary electrophoresis
Citation for published version (APA):Ingelse, B. A. (1997). Chiral separations using capillary electrophoresis Eindhoven: Technische UniversiteitEindhoven DOI: 10.6100/IR492451
DOI:10.6100/IR492451
Document status and date:Published: 01/01/1997
Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication
General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:
www.tue.nl/taverne
Take down policyIf you believe that this document breaches copyright please contact us at:
providing details and we will investigate your claim.
Download date: 09. Sep. 2019
CHIRAL SEPARATIONS USING
CAPILLARY ELECTROPHORESIS
Ingelse, Benno A
Chiral separations using capillary electrophoresis I by Benno A.
Ingelse. - Eindhoven: Technische Universiteit Eindhoven, 1997.
Proefschrift. -
ISBN 90-386-0958-2
NUGI 813
Trefw.: capillaire elektroforese I stereoselectieve scheidingsmethoden
Subject headings: capillary electrophoresis I enantiomers
CHIRAL SEPARATIONS USING
CAPILLARY ELECTROPHORESIS
PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de
Technische Universiteit Eindhoven, op gezag van
de Rector Magnificus, prof.dr. M. Rem, voor
een commissie aangewezen door het College
van Dekanen in het openbaar te verdedigen op
donderdag 19 juni 1997 om 16.00 uur
door
BENNO ALLARD INGELSE
geboren te Breda
Dit proefschrift is goedgekeurd door de promotoren:
prof.dr.ir. F.M. Everaerts
en
prof.dr.ir. C.A.M.G. Cramers
co-promotor:
dr.ir. J.C. Reijenga
TABLE OF CONTENTS
TABLE OF CONTENTS
INTRODUCTION~~~~~~~~~~~~~~~~~~~
1. PRINCIPLES OF CAPILLARY ELECTROPHORESIS 7
1.1 Introduction to electrophoresis 7
1.2 Electrophoretic mobility 8
1.3 Factors influencing resolution 9
1.4 Electroosmosis 12
1.5 Different modes of electrophoresis 13
1.6 Instrumentation 14
2.THE CHIRAL SEPARATION MECHANISM IN 19
2.1 Introduction 19
2.2 Chiral selectors in CE 20
2.2. l Rules for chiral recognition 20
2.2.2 Cyclodextrins 21
2.2.3 Crown-ethers 24
2.2.4 Macrocyclic antibiotics 25
2.2.5 Proteins 25
2.2.6 Micelles 25
2.2. 7 Other chiral selectors 26
2.3 Determination of equilibrium constants of complex formation 26
3.ENANTIOMERIC SEPARATION OF DRUGS BY CE USING A SOLUBLE
NEUTRAL ~-CYCLODEXTRIN POLYMER 35
3.1 lntroduction 35
3.2 Experimental 36
3.2.l Chemicals 36
3.2.2 Apparatus 37
3.2.3 Methods 37
TABLE OF CONTENTS
3.3 Results and 01s1cuss101fl __________________ 37
3.3.1 Effect of the concentration of EP-~-CD on the mobility of the analytes __ 39
3.3.2 Effect of the concentration of EP-~-CD on chiral recognition 42
3.3.3 Effect of buffer concentration on the resolution of the enantiomers 46
3.3.4 The effect of organic solvent added to the BGE 48
3.3.5 The effect of tèmperature 51
3.4 Conclusions 52
4.ERGOT ALKALOIDS AS CHIRAL SELECTORS IN CAPILLARY ELECTROPHORESIS ________ ~ _________ 55
4.1 Introduction _____________________ 55
4.2 Experimental --------------------- 58 4.2.1 Equipment 58
4.2.2 Chemicals and sample preparation 58
4.2.3 Methods and electrophoretic systems 59
4.3 Results and discussion 60
4.3.1 Characterization of the ergot alkaloids 60
4.3.2 Comparison of stereoselectivity of different ergot alkaloids towards some
racemic organic acids. 63
4.3.3 Determination of the mobility of the analyte interacting with allyl-TER __ 64
4.3.4 Influence of the pH on stereoselectivity 68
4.3.5 Direct determination of formation constants 69
4.3.6 Reversed determination of the formation constants of mandelic acid 72
4.3.7 The influence ofMeOH on chiral separation 74
4.4 Conclusions 77
5. THE INFLUENCE OF THE NATURE OF THE BUFFER ON CHIRAL
SEPARA TION IN CE 79
5.1 Introduction 79
5.2 Experimental 80
5.2.1 Chemicals 80
5.2.2 Apparatus 81
5.2.3 Methods 81
ü
TABLE OF CONTENTS
5.3 Results and discmisi<J1n ___________________ 83
5.3. l DIME-P 84
5.3.2 TRIME-P 88
5.3.3 Neutral a- and P-cyclodextrin polymer 90
5.3.4 HP-P-CD 96
5.4 Coraclusioi:as __________________ 96
6.COMPUTER SIMULATION AND BASIC THERMODYNAMICS OF
CHIRAL SEPARATIONS IN CE -------------- 99
6.1 Introduction ______________________ 99
6.2 Experimental --------------------- 101 6.2.l 101
6.2.2 Chemicals 101
6.2.3 Methods 101
6.3 Simulations of chiral separations _______________ 103
6.3.1 Chiral sub-menu 103
6.3.2 Calculation of migration behavior 105
6.3.3 Visualization of results 105
6.3.4 Selectivity vs. resolution 106
6.3.5 Comparison with literature results 107
6.3.6 Comparison with experimental results 108
6.3.6. l Determination of the pK values and mobility of mandelic acid and
terbutaline ---------------------- 108 6.3.6.2 Determination of formation constants 110
6.4 Temperature etTects and basic thermodynamics of the chiral separation of
ibuprofen enantiomers 118
6.4.l Effect oftemperature on BGE conductivity and pK 118
6.4.2 Effect of temperature on EOF 118
6.4.3 Effect of temperature on the mobility and pK. of ibuprofen 119
6.4.4 Determination of K2 of ibuprofen-~-CD at different temperatures 120
6.4.5 Determination of the average Ki of ibuprofen-P-CD at different temperatures _______________________ 122
6.4.6 The effect of temperature on selectivity 123
6.4.7 Thermodynamic model for K1 and 124
iii
TABLE OF CONTENTS
6.5 Conclusions, ____________________ 129
7. APPLICABILITY OF CE IN CHIRAL SEPARATIONS ______ 131
7.1 Introduction ____________________ 131
7.2 Herbicides _____________________ 131
7 .2.1 Introduction 131
7.2.2 Experimenta1 132
7 .2.3 Results and discussion 133
7.3 Quality control of fenfluramine enantiomers ---------- 136 7 .3.1 lntroduction 136
7 .3.2 Experimental 136
7.3.3 Results and discussion 137
7.4 Determination of thiopental enantiomers in plasma ------- 140
7.4.1 lntroduction 140
7.4.2 Experimental 140
7.4.3 Results and discussion 141
7.5 Conclusions _______ ~------------ 144
ABSTRACT ____________________ l47
SAMENV A TIING 149
SYMBOLS AND ABBREVIA TIONS 153
DANKWOORD 157
CURRICULUM 159
BIBLIOGRAPHY 161
iv
INTRODUCTION
INTRODUCTION
What is chirality?
The word chiral originates from the Greek word "xetp", rneaning hand. The
left and right human hands are non-superimposable forms, that can be represented as
mirror images. An object is "handed" if it has an identical mirror image counterpart,
which cannot be superimposed onto itself. Analogously, a molecule is chiral if its
îdentical mîrror image counterpart cannot be superimposed onto itself.
Figure 1 shows two forms of mandelic acid. The two forms are not
superimposable. They are known as enantiomers or optica[ isomers. A 1: 1 mixture of
the two enantiomers is called a racemic mixture.
yOOH
Ho-9-H
© yOOH
H-Ç-OH
mirror
Enantiomers have identical physîcal and chemica]
properties, in an isotropic environment. lts
chirality is only observed when the molecule is
subjected to a chiral influence. A well known
example is the optica! rotation of polarîzed light.
Polarized light is rotated when passing through
solutions containing chiral molecules (but not
Figure 1 D(.) and L(+)-mandelic when passing through racemic mixtures). Optica! acid isorners rotate the light in an equal degree but in
opposite direction. If the enantiorner rotates the
light to the right, it will be indicated as dextrorotatory (Latin: dexter), "d" or "(+)".
Optica! isorners that rotate light to the left, on the other hand will be indicated as
levorotatory (Latin: laevus), "l" or"(-)". According to the Fischer convention [1] the
absolute configuration around a chiral center can be noted as D or L. This notation,
which is nowadays mainly used for amino acids and carbohydrates, correlates the
configuration of a chiral center to the configuration of D and L glyceraldehyde. The
RIS notation (from Latin; rectus (right) and sinister (left)), which has largely replaced
the DIL notation, is related to the Cahn-Ingold-Prelog convention [2], and can also be
used for molecules containing more than one chiral center.
Optica] activity was first discovered in 1815 by Jean-Baptiste Biot [3]. Louis
Pasteur was the first to separate enantiomers [4]. In 1848, he used a pair of tweezers
and a microscope to isolate the crystals of the optical isomers of tartaric acid.
Stereochemistry was further stimulated by Van't Hoff [5] and Le Bel [6]. They
INTRODUCTION
proposed that the four valences of the carbon atom are directed towards the vertices of
a tetrahedron with the carbon atom at its center.
The best known and easiest recognizable form of chirality is obtained when a
carbon atom possesses four different substituents, like mandelic acid (Figure 1 ). This
carbon is called a stereocenter or chiral center. This kind of chirality is not only
restricted to carbon chemistry. A1so nitrogen, sulphur, phosphorus, and boron can
produce stable chiral centers. In this thesis, chirality caused by a chiral carbon center is
the main issue of attention. Formation of chiral axes (e.g. dialkenes) or planes (e.g.
binaphtol) can also cause chirality.
Consequences of chirality Probably the best example of chiral influence is given by nature. Most
recognition systerns in nature (e.g. enzymes, receptors) distinguish between
enantiomers. It is well known that the enantiomers of chiral pharmaceuticals can
behave very different in the human body. The (-)-enantiomer of the ~-blocker
propranolol is about 100 times more active than the (+)-form. Another example is
given by the thalidomide (softenon) tragedy in the early 1960s. Thalidomide was
administered as a racemate. However, only the (R)-(+)-enantiomer possessed the sleep
inducing action. The S-(-)-enantiomer possessed teratogenic action, responsible for
serious malformation in newborn babies of women who took the drug during
pregnancy. Recently, it bas been reported that among 523 natura! and semi-synthetic
drugs, 98.8% are chiral and 98.4% of them are sold as a single isomer, while in case of
the 1327 synthetic drugs, chiral compounds represent 39.8% and only 11.6% of them
are sold as a single enantiomer [7]. Ariëns [8] emphasized the need for administration
of the optically pure active drug (the eutomer), calling the "inactive" drug enantiomer
(the distomer) "isomerie ballast".
A sirnilar problem is encountered in the agrochemical industry. A report in
1981 [9] showed that of the 550 pesticides, 98% were synthetic products of which
17% were shown to be chiral. Only 8% of these chiral synthetic pesticides were
marketed as single isomers.
The properties of products from the food and drink industry can also be highly
dependent on enantiomeric composition. For example, the S-isomer of asparagine bas a
bitter taste while the R-isomer tastes sweet. Asparagine is a precursor to the sweetener
aspartame.
2
INTRODUCTION
Separation of optical isomers
These examples underline the need for chiral separation methods, preparative
as well as analytical. Enantiomers can be separated either by the direct or the indirect
separation rnethod. The indirect separation rnethod is based on the formation of a
covalent bond between the optical antipodes on the one hand and a pure chiral
compound, called the chiral selector, on the other hand. This chemica] reaction will
result in a product consisting of two isomerie compounds which are not mirror images
anymore. They are known as diastereoisomers and they can, in principle, be separated
by any analytica! method using an achiral separation mechanism. This method is, first
of all, time consuming since sample pretreatment involving a chemica! reaction is
necessary. Secondly, the chiral selector has to be very pure, since optica! impurity will
result in two more diastereomeric products.
In the direct separation mode, the separation of the optica! isomers is based
upon complex formation between the enantiomers and a chiral selector, resulting in the
formation of labile diastereoisomers. Separation can be accomplished if the complexes
possess different stability constants. The afore mentioned disadvantages do not apply
for the direct separation mode. The chiral purity of the selector only influences the
resolution. It has been shown that relatively good results can be obtained using a chiral
selector containing up to 10% of its antipode [ 10).
Analytica! methods used so far for the enantiomeric separation include high
performance liquid chromatography (HPLC) [l l-13], thin-layer chromatography
(TLC) [14], gas chromatography (GC) [15], supercritical fluid chromatography (SFC)
[16], and capillary electrophoresis (CE) [17-30]. The application of gas
chromatography is mainly restricted to more volatile compounds. Therefore, until now,
the method of choice for the separation of more polar compounds, as are most drugs,
is HPLC. The main drawback of CE compared to HPLC is that until now, CE bas not
shown to be useful as a preparative separation tool. Another advantage of HPLC over
CE is the low detection limit, due to the much Jonger path length of the detection cell
and the much higher injection volume. However, the very high efficiencies usually
obtained in CE, and the ease of method development, make it a very good alternative
for analytica! separation of enantiomers. Other advantages of CE over HPLC are the
low consumption of both analyte and chiral selector and the short analysis times.
Moreover, CE has no need for expensive chiral stationary phases, since the chiral
selector is simply added to the buffer. Alternatively, CE might be very useful for the
rapid screening of novel chiral selectors, thus avoiding the waist of the laborious
synthesis of new chiral HPLC stationary pha<>es.
3
INTRODUCTION
Despite the many studies already published on the subject of chiral separations
using CE, little is known about the fundamental aspects and hardly any data is available
on complex formation equilibria, resulting in chiral separation. The primary aim of this
thesis was to obtain a better insight into the chiral separation mechanism and to
characterize and quantify the above mentioned equilibria. This knowledge should result
in a better understanding of the capillary electrophoretic separation of optica\ isomers.
Chapter 1 describes the principles of capillary electrophoretic separations. Factors
intluencing mobility, such as the pH and complexing agents, are discussed. In the work
presented in this thesis, complexing agents are required for the separation of
enantiomers.
Chapter 2 describes the chiral separation mechanism. A survey is presented of the
most important chiral selectors, applied in CE. A genera! model is presented to
determine formation constants between neutra! chiral selectors and optica] isomers.
Chapter 3 describes the use of a neutral 1)-cyclodextrin polymer as chiral selector in
CE. Factors intluencing mobility and resolution, such as concentration of the chiral
selector, organic modifier content and ionic strength of the background electrolyte
(BGE), and temperature are investigated.
Chapter 4 introduces ergot alkaloids as chiral selectors in CE. Different ergot
alkaloids are studied and compared. with respect to their enantioselectivity towards
some chiral organic acids. The effects of pH, and the addition of MeOH to the BGE
were investigated. The formation constants of the dissociated and the non-dissociated
chiral acids were detennined.
Chapter 5 discusses the intluence of the nature of the buffer on chiral separations.
Different (modified) cyclodextrins (CD's) are applied for the separation of some
sulfonamide enantiomers. Formation constants are determined in different electrolyte
systems. The intluence of the co-migrating buffer anion on the formation constant and
enantioselectivity is discussed.
In Chapter 6, the use of a steady-state simulation program for CE, extended with a
sub-menu for chiral interaction is demonstrated. From mobility determinations at
different temperatures and CD-concentrations, it was possible to calculate some basic
therrnodynamic parameters concemed with complex fonnation.
Chapter 7 illustrates the potential of CE for chiral analyses of various cornpounds in
different matrices. The detennination of drug enantiomers is shown in pharmaceutical
formulations and in human plasma.
4
INTRODUCTION
References
N.V. Alllinger, M.P. Cava, D.C. De Jongh, C.R. Johnson, N.A. Lebel, C.L.
Stevens, Eds., Organic Chemistry, Worth Publishers, New York, 1976, 98-101
2 R.S. Cahn, C.K. lnghold, V. Prelog, The specification of asymmetrie
configuration in organic chemistry. Experientia, 12 (1956) 81
3 D.E. Drayer, Drug stereochemistry: analytica] methods and pharmacology,
l.W. Wainer and D.E. Drayer Eds., Marcel Dekker, New York, 1988, 3-29
4 L. Pasteur, Comptes Rendus del' Academie des Sciences, 6 ( 1848) 535
5 J.H. Van't Hoff, Arch Netherland Sci Extracts et Naturelles, 9 (1874) 445
6 J.A. Le Bel, Bull Sci Chimique de France, 22 (1874) 337
7 J.S. Millership and A. Fitzpatrick, Chirality, 5 (1993) 573
8 E.J. Ariëns, Stereochemistry, a basis for sophisticated nonsense in
pharmacokinetics and clinical pharmacology, Eur. J. Clin. Pharmacol., 26
(1984) 663
9 E.Y. Spencer, Guide to chemicals used in erop protection. Ottawa, Canadian
Govt. Publ. Centre, 7th ed., 1981
10 S. Fanali, M. Cristalli, R. Vespalec and P. Bocek, in A. Chrambach, M.J.
Dunn and B.J. Radola (Editors), Advances in Electrophoresis, VCH,
Weinheim, 1994, 3
11 J. Debowski, D. Sybilska and J. Jurczak, J. Chromatogr., 282 (1983) 83
12 T..J. Ward and D.W. Armstrong, J. Liq. Chromatogr., 9 (1986) 407
13 G. Blaschke, J. Liq. Chromatogr" 9 (1986) 341
14 D.W. Armstrong, Faulkner, Jr., and S.M. Han, J. Chromatogr., 452 (1988) 323
15 C.P. Granville, B. Gebreke, W.A. Konig and I.W. Wainer, J. Chromatogr., 622
(1993)21
16 S. Hara, A. Dobashi, K. Kinoshita, T. Hondo, M. Saito, and M. Senda,
J. Chromatogr., 1986 (371) 153
17 T. Schmitt and H. Engelhardt, Chromatographia, 37 (1993) 247
18 A. Nardi, A. Eliseev, P. Bocek and S. Fanali, J. Chromatogr., 638 (1993) 247
19 R. Vespalec, V. Sustacek and P. Bocek, J. Chromatogr., 638 (1993) 255
20 L. Valtcheva, J. Mohammed, G. Pettersson and S. Hjerten, J. Chromatogr.,
638 (1993) 263
21 R. Kuhn, F. Stoekklin and F. Emi, Chromatographia, 33 (1992) 32
22 R. Kuhn and S. Hoffstetter-Kuhn, Chromatographia, 34 (1992) 505
23 E. Hohne, G.J. Krauss and G. Gubitz, J. High Resolut. Chromatogr" 15 (1992)
698
24 S. Busch, J.C. Kraak and H. Poppe, J. Chromatogr" 635 (1993) 119
25 H. Nishi, T. Fukuyama, M. Matsuo and S. Terabe, J. Microcol. Sep"1 (1989)
234
5
INTRODUCTION
26 S. Terabe, H. Shibata and Y. Miyashita, J. Chromatogr., 480 (1989) 403
27 P. Gozel, E. Gasmann, H. Michelsen and R.N. Zare, Anal. Chem., 59 (1987)
44
28 S. Fanali, L. Ossicini, F. Foret and P. Bocek, J. Microcol. Sep., l ( 1989) 190
29 A. Guttman, A. Paulus, A.S. Cohen, N. Grinberg and B.L. Karger, J.
Chromatogr., 448 (1988) 41
30 S. Fanali, J. Chromatogr" 474 (1989) 441
6
PRINCIPLES OF CAPILLARY ELECTROPHORESIS
1. PRINCIPLES OF CAPILLARY ELECTROPHORESIS
Abstract
This chapter describes the principles of capillary electrophoresis. The basic concept
of electrophoretic mobility is explained. Furthennore, factors influencing the mobility
and basic parameters, such as resolution and efficiency, are described.
1.1 Introduction to electrophoresis
Electrophoresis is the separation principle in which charged particles or
molecules are separated under the influence of an external electric field. Already at the
beginning of the last century, von Reuss performed the first electrophoretic
experiments [l]. Today exactly 100 years ago, Kohlrausch developed his regulating
functions [2], which made it possible to theoretically describe all electrophoretic
methods. Electrophoresis has, since then, been mainly applied for the separation of
large biomolecules like DNA and proteins, using stabilizing and sieving media such as
gels. The introduction of narrow bore tubes as an anti-convective medium made it
possible to use free solutions in stead of these gels. Hjertén describes the use of a
rotating glass tube of 3 mm inner diameter (l.D.) [3]. Smaller I.D. capillaries were
successfully applied by Everaerts [4] and Virtanen [5]. The reduction of the I.D.
allowed the use of higher electric field strengths, resulting in higher efficiencies and
shorter times of analysis. Mikkers et al. [6,7) showed that the high efficiencies,
theoretically described by Giddings [8], could be achieved. Jorgenson [9] used 75 µm
LD. glass capillaries, in which longitudinal diffusion was shown to be practically the
only source of band broadening. Capillary electrophoresis (CE) has since then proven
to be a highly efficient, analytica! separation tool, not only for the separation of
macromolecules but also for smaller molecules. Fundamental studies as well as
numerous applications have been reported in the last decade. A survey is presented in
reviews [ 10, 11) and textbooks [ 12-15).
7
CHAPTER 1
1.2 Electrophoretic mobility
The velocity of solute molecules (v, [m.s-1] ) will be proportional to the applied
electric field (E, [V.m"1]) and the electrophoretic mobility (µ, [m2.V-1.s·1
] ):
v=µ E (1-1)
The electrophoretic mobility is dependent on both the charge (q, [C] ) and the radius
(r, [m] ) of the particle. The electric force (F,1 , [N] ) equals:
F,1 =qE (1-2)
In steady state, the electric force is counter balanced by the friction force (F1) which
equals:
(1-3)
Eq. (1-3) is known as Stokes' law, in which 1J [N.m"2.s] is the dynarnic viscosity of the
surrounding medium. Assuming constant velocity:
(1-4)
and thus follows for the mobility at infinite dilution (µ0):
0 v q µ =-=-
E 61r17r (1-5)
lt should be noted that Stokes' law is only valid for rigid spherical particles, having a
homologous charge distribution. Most molecules do not meet these conditions. In
genera!, it can be concluded that the mobility of a molecule is related to its charge to
mass ratio.
At finite dilution, other forces should be taken into account, originating from
the relaxation and retardation effects. A quantitative description of these effects is
given by Debye, Hückel and Onsager [16]. The mobility at finite dilution (µ) is shown
to decrease, with increasing concentration of the BGE.
8
PRINCIPLES OF CAPILLARY ELECTROPHORESIS
Another parameter which influences the mobility is the temperature.
Electrophoretic mobilities increase at a rate of approximately 2% °C1 • Obviously,
temperature not only influences the mobility, but also equilibrium constants, such as
pK-values and complex formation constants, which can have a major influence on
electrophoretic separations of ions.
The mobility of any molecule can be altered by changing its charge to mass
ratio. This ratio can be changed due to e.g. acid dissociation or complex formation.
Obviously, the mobility of a dissociated acid will differ from that of the non-dissociated
form. Analogously, the mobility of cations such as potassium or amino compounds will
decrease upon complex formation with (neutra!) crown ethers. The ejfective mobility
(µett) can be expressed as a summation of the products of the concentration of the
different subspecies relative to the analytica! concentration (a) and the mobilities of
these subspecies [17]:
( 1-6)
This equation is important in order to understand the chiral separation mechanism, as
further explained in the next chapter.
1.3 Factors influencing resolution
The quality of a separation of two eomponents (1 and 2) is described by its
resolution (Rs), defined as:
(1-7)
in which tm [s] is the migration time and er [s] is the standard deviation of the peak (<f
is the peak variance).
The value of the numerator can be increased by increasing the effective mobility
difference, which is a measure for the selectivity. In the previous section, it is discussed
briefly how to influence the effective mobility. Obviously, separation of weak acids or
bases can be optimized through pH-optimization. Analogously, separation of (optica!)
isomers can be optimized by the addition of the optimum amount of a suitable
complexing agent. The Jatter will be discussed extensively in chapter 2.
9
CHAP1ER 1
In order to optimize resolution, also the denominator of eq. ( 1-7} should be
minimized. The peak variance or band broadening can be characterized by the
efficiency. The efficiency is quantified by the number of theoretica! plates (N):
(1-8)
where Id is the length [m] of the separation tube until the point of detection and <J is the
standard deviation [m]. Assuming that variances of independent sources of band
broadening are additive, the total peak variance will be:
2 2 2 2 2 2 2 2 C1tot =(Jinj +C1det +O', +O'ther +O'dif +<Jcnnc +O'EOF
a1!i = variance due to injection
a;er = variance due to the width of the detector slit,
a; = variance due to the time constant of the detector
a,!,, = variance due to thermal dispersion
aJif = variance due to longitudinal diffusion
a;onc = variance due to electromigration dispersion
aioF = variance due to electroosmosis
(1-9)
The dispersion due to injection is, in the ideal case, proportional to the square of the
length of a (stacked) sample plug. Analogously, dispersion due to detector slit width is
proportional to the square of the slit width. Also the square of the time constant of the
detector is proportional to the peak variance.
Limited heat dissipation is the main problem encountered in fused silica
capillaries at elevated current densities. Since heat is only dissipated at the inner
surface wall of the capillary, while heat is produced throughout the separation medium,
a parabolic heat profile across the capillary will exist. Consequently, sample
constituents which are migrating in the warmer center of the capillary, will have higher
velocities than identical ions, migrating in the same zone, but close to the capillary
wall. This will cause unwanted zone broadening, referred to as thermal dispersion.
Reduction of the 1.D. will result in reduced thermal dispersion, according to Ref. [5].
Longitudinal diffusion is related to the diffusion coefficient (D [m2 s·1n by the
Einstein relation:
10
PRINCIPLES OF CAPILLARY ELEC1ROPHORESIS
(l-10)
In diluted solutions, the diffusion coefficient is related to the mobility at infinite dilution
by the Nemst-Einstein relation:
µ 0 RT D:::--
zF (1-11)
in which R is the gas constant [J K 1 mor1], T is temperature [K], z is the charge
number and Fis Faradays constant [C mor1]. If, in the ideal case, Jongitudinal diffusion
would be the only source of band broadening, and assuming IJ "" l1, a simp Je re lat ion for
the efficiency can be derived [9]:
N = 2D
Combining eq. (1-11) and (1-12), and assuming µ = µ0:
zVF N=
2RT
(1-12)
(1-13)
Eq. (1-13) shows that the efficiency is proportional to the applied voltage (V, [V]). In
the above mentioned ideal case, increasing voltage will result in an increased efficiency.
Dispersion due to electroosmosis is considered negligible in open
electrophoretic systems, assuming an ideal plug flow (see section 1.4). However,
electroosmosis might cause band broadening if the plug flow is not constant
throughout the separation system.
Dispersion due to electromigration is caused by a mobility difference between
sample ions and co-ions in the BOE. This will result in triangularly shaped peaks [6].
Electrodispersion can seriously decrease resolution. As an example, it has been shown
that mobility matching between optica! isorners on the one hand and the co-migrating
ion on the other hand can greatly improve chiral resolution [18]. Obviously the
resolution of other ( difficult) separations can be improved similarly. All above
mentioned sources of band broadening are discussed in more detail in Ref. [ 19).
The last section gave a brief overview of several factors influencing the
efficiency. The most obvious way to improve the efficiency is, according to eq. (l-13),
increasing the voltage. Unfortunately, voltage cannot be endlessly increased. lncreasing
11
CHAPlER 1
voltages lead to an increased heat generation and consequently to increased thermal
dispersion.
Besides this effect, deficient heat dissipation can even cause the separation
medium to boil. Heat formation is enforced by the increased mobilities, leading again
to higher currents, and thus to more heat formation, etc .. Obviously, boiling of the
separation medium will result in a total break-down of the system.
The most effective solution to combine high voltages with limited heat
dissipation is to reduce the tubes inner diameter (LD.). First of all, the ratio between
tube surface area and tube volume will increase, resulting in a more efficient heat
dissipation. Reducing the l.D. will also have a very beneficia! effect on thermal
dispersion [5). Consequently, the use of capillaries allows the use of high voltages,
which will simultaneously result in high plate numbers and short analysis times.
1.4 Electroosmosis
Electroosmosis is known as the flow of an electrolyte solution caused by an
electric field across a capillary. If a fused silica capillary wal! is in contact with a
solution, the wall will be negatively charged, due to dissociation of the silanol groups.
Consequently, the wall will attract cations from the solution, resulting in the formation
of an electric double layer adjacent to the capillary surface. If an electric field is applied
across the capillary length, the mobile part of this double layer will start rnigrating
towards the cathode. Tuis results in a flow of the electrolyte solution, called the
electroosmotic flow or EOF. The most pronounced property of the EOF is the flat
flow profile, compared with the parabolic flow profile of hydrodynarnic flows,
common in chromatography. The flat flow profile will result in a decreased band
broadening compared to e.g. open tubular liquid chromatography (OTLC). The
mobility of the EOF (µEoF) is found from the Helmholtz-Smoluchowski equation [20]:
(l-14)
where Ç is the potential of the capillary surface [V] and e is the dielectric constant of
the solvent [C2.J1.m-1).
The mobility of the EOF will be superimposed on the electrophoretic mobility
of the solute molecules, resulting in the apparent mobility of the analyte {µaj,p):
12
PRINCIPLES OF CAPILLARY ELECTROPHORESIS
( 1-15)
If the magnitude of the EOF is sufficiently high, it is possible to give both cations and
anions a net velocity in the direction of the cathode. Consequently, it is possible to
detect both anions and cations in one single electrophoretic run.
In some cases it is desirable to suppress the EOF. Efficient ways to suppress
the EOF are the reduction of (and increasing the viscosity near the wall. Modification
of the capillary surface by covalent linking of the silanol groups with polyacrylamide
has proven to be a useful method to simultaneously obtain both desired effects [21].
1.5 Different modes of electrophoresis
Four main electrophoretic techniques can be distinguished. The most popular
and most widely applied mode is zone electrophoresis (ZE). In ZE, a minute amount of
sample is introduced in the separation system, which is filled with one electrolyte. This
electrolyte is called the background electrolyte (BGE). After applying an electric field
across the separation system, all analytes will start migrating in distinct zones, with
different velocities. A high value of the EOF allows the separation of both anions and
cation in one single run, using ZE.
In rnoving boundary electrophoresis, the separation system is filled with a so
called leading electrolyte. The sample is introduced at the beginning of the separation
compartrnent. The leading electrolyte consists of a co-ion having a higher mobility than
the separands. After applying the voltage, the most mobile analyte will form a pure
zone, followed by a mixed zone consisting of the most mobile and the second most
mobile analyte. According to the Kohlrausch regulation function, the concentration of
the components in the zones are adapted to the concentration of the leading
electrolyte. The boundaries between the zones have self-correcting properties, due to
differences in electric field strength between the zones. The boundary between the
leading zone and the first pure zone can be considered as an isotachophoretic
boundary. In a single run, moving boundary electrophoresis is only applicable to either
anions or cations.
In isotachophoresis (ITP) a discrete amount of sample is sandwîched in
between a leading electrolyte and a terrninating electrolyte. Generally, similar to
rnoving boundary electrophoresis, only anîons or cations can be separated. The leading
13
CHAPTER 1
electrolyte consists of a so called leading-ion, having a higher mobility than any of the
separands. The terminating ion, on the other hand, has the Iowest rnobility. In steady
state, all components will rnigrate with equal velocities in consecutive zones, in order
of their mobility. The Jeading ion will rnigrate in front. As in rnoving boundary
electrophoresis, the concentration of the components in the zones are adapted to the
concentration of the leading electrolyte. Likewise, the boundaries between the distinct
zones have self-correcting properties.
In isoelectric focusing, the BGE consists of ampholytes which will create a pH
gradient along the separation tube. The amphiprotic sample components will rnigrate
through the pH-gradient until the place where the pH equals their isoprotic point,
which approaches their isoelectric point (pi). At the isoprotic point, the effective
velocity of the analytes will be zero, and consequently they will stop rnigrating.
Isoelectric focusing allows separation of e.g. proteins, due to differences in pl-values.
As mentioned before, most people using capillary electrophoretic techniques
apply ZE. However, it should be noted that in the initia! phase of almost any
electrophoretic separation, the moving boundary principle is present.
Micellar electrokinetic chromatography (MEKC), or rnicellar electrokinetic
capillary chromatography (MECC) allows the separation of neutral molecules [22, 23].
A surfactant is added to the BGE in a concentration higher than its critica] micelle
concentration (CMC). These rnicelles will act as a pseudo-stationary phase.
Components will be separated due to a distribution equilibrium between the rnicellar
phase and the aqueous phase, both of which move at different velocities.
1.6 Instrumentation
The most important modes of CE, as discussed in the last section, can be
executed in the same equipment. Basically, the equipment consists of a high voltage
power supply, a separation tube, an injection module, a detector and a data collection
system. Figure 1.1 schematically shows an electrophoretic separation system.
Generally, for the electrophoretic experiments, presented in this thesis, a
P/ACE 2200 (Beckman, Palo Alto, CA) was used. The Beckman instrument is fully
automated, and consists of, besides the earlier mentioned features common for all
electrophoretic separation systerns, an auto sampler and a Iiquid cooled capillary
cartridge. The Jatter aJlows temperature control between 20°C and 50°C.
14
PRINCIPLES OF CAPILLARY ELECTROPHORESIS
Fused silica capillaries, with an inner diameter of 50 or 75 µm were
used as separation tubes. The required minimum length was about 27 cm. Prior to use,
new capillaries were rinsed for approximately 30 minutes with a 1 M KOH solution.
For some applications, the capillary inner wall was coated with linear polyacryl amide,
according to the procedure described elsewhere [24]. The coating procedure could be
fully automated, using the PI ACE 2200.
power supply-=-
-- - - - -- -- -- ... __ - -- --- - -1
detector ---------
~electrolyte vessels--->
1 1 1
data collection
Figure 1.1 Schematic representation of a capillary electrophoretic separation system
The high voltage power supply is capable of delivering voltages ranging from 1
kV up to 30 kV, and currents up to 250 µA. In most experiments, electric field
strengths were about 500 Vim, resulting in currents of 5-50 µA. The P/ACE control
software allows separations to be performed at either constant voltage, constant
current or constant power. Generally, the constant voltage mode was selected.
Samples can be introduced either hydrodynamically or electrokinetically. The
former method results in more reproducible data, and was therefore selected for most
applications.
The Beckman instrument is standard equipped with a selectable wavelength
UV absorbance detector. On line detection is made possible by removing a short
15
CHAP1ER 1
section of the protective non-transparent polyimide coating from the capillary.
Generally applied wavelengths were 200 nm, 214 nm, 230 nm, 254 nm, and 280 nm.
The separation method is fully controlled by the P/ ACE control software. Data
collected with this software were generally analyzed by Caesar for Windows. The
algorithm of this software was developed by Wanders [25].
References
l F. von Reuss, Comment. Soc. Phys. Univ. Mosquensem, 1 (1808) 141
2 F. Kohlrausch, Ann. Phys. (Leipzig), 62 (1897) 209
3 S. Hjertén, Chromatogr. Rev" 9 (1967) 122
4 F.M. Everaerts and W.M.L. Hoving-Keulemans, Sci. Tools, 17 (1970) 25
5 R. Virtanen, Acta Polytech. Scand" 123 (1974) 1
6 F.E.P. Mikk.ers, F.M. Everaerts and Th.P.E.M. Verheggen, J. Chromatogr.,
169 (1979) 1
7 F.E.P. Mikk.ers, F.M. Everaerts and Th.P.E.M. Verheggen, J. Chromatogr"
169 (1979) 11
8 J.C. Giddings, Separ. Sci., 4 (1969) 181
9 J.W. Jorgenson and K.D. Lukacs, Anal. Chem" 53 (1981) 1298
10 W.G. Kuhr and C.A. Monnig, Anal. Chem., 64 (1992) 389R
11 C.A. Monnig and R.T. Kennedy, Anal. Chem., 66 (1994) 280R
12 S.F.Y. Li (Ed.), Capillary Electrophoresis, J. Chromatogr. Libr" vol. 52,
Elsevier, Amsterdam, 1992
13 P.D.Grossman and J.C. Colbum (Eds.), Capillary Electrophoresis, Academie
Press, San Diego, 1992
14 N.A. Guzman (Ed.), Capillary Electrophoretic Technology, Chromatogr. Sci.
series, vol 64, Marcel Dekker, New York, 1993
15 R. Weinherger, Practical Capillary Electrophoresis, Academie Press,
San Diego, 1992
16. H. Falkenhagen, Elektrolyte, Verlag von S. Hirzel, Leipzig, 1932
17 Tiselius, Nova Acta Reg. Soc. Sve. Sci., Upsala, 4, 7 no 4 (1930)
18 Y.Y. Rawjee, R.L. Williams and G. Vigh, Anal. Chem" 66 (1994) 3777
19 J.C. Reijenga, E. Kenndler, J. Chromatogr. A, 659 (1994) 403
20 S. Hjertén, Chromatogr. Rev" 9 (1967) 122
21 S. Hjertén, J. Chromatogr., 347 (1985) 191
16
PRINCIPLES OF CAPILLARY ELECTROPHORESIS
22 S. Terabe, K. Otsuka, K. Ichikawa, A. Tsuchiya and T. Ando, Anal. Chem.,
56 ( 1984) 111
23 P. Muijselaar, thesis, Eindhoven University ofTechnology, Eindhoven, 1996
24 M.J. van der Schans, J.L. Beckers, M.C. Molling and F.M. Everaerts,
J. Chromatogr. A, 717 (1995) 139
25 B.J. Wanders, thesis, Eindhoven University ofTechnology, Eindhoven, 1993
17
CHAP'IER 1
18
CHIRAL SEPARA TION MECHANISM
2. THE CHIRAL SEPARATION MECHANISM IN CE
Abstract
In this chapter, the basic principles of the chiral separation process in CE are
described. Chiral separation is based upon the f onnation of diastereomeric
complexes between the optica/ isomer and a chiral selector. Chiral separation can be
obtained only if these complexes have different equilibrium constants of complex
fonnation. A genera/ model is presented to detennine these fonnation constants.
Furthennore, an overview of the most commonly used chiral selectors in CE is
presented.
2.1 Introduction
As mentioned in the introductory chapter, separation of racemic mixtures can
be accomplished either by the direct or by the indirect separation mechanism. The
indirect separation mode in CE has been mainly applied for the enantiomeric separation
of amino acids [1-3]. However, in these cases, electrokinetic chromatography (EKC)
had to be employed, either by the addition of sodium dodecyl (SDS) micelles [l] or by
the addition of a chiral polymer; polyvinylpyrrolidone [2,3]. The separation of D- and
L-carnitinine, on the other hand, bas been performed after derivatization with (-)-[ 1-
fluorenyl)ethyl]chloroformate (FLEC) in a 50 mM phosphate buffer at pH 2.6 without
any additives. However, also in this case, resolution could be improved by the addition
of 20 mM of the surfactant tetrabutylammonium bromide [4]. Generally, it can be
concluded that most electrophoretic separations of diastereorneric compounds are
performed using MEKC, in order to optimize selectivity. Since the actual separation
rnechanism of the indirect separation method is achiral, this thesis only deals with the
direct separation method.
Successful application of the direct separation method involves interaction
between the optica! isomers on the one hand and a chiral selector on the other hand. In
most cases, the chiral selector is simply added to the BGE [5-8]. Chiral selectors can
also be incorporated or bound to a gel matrix [9,10], or bound to the capillary wal]
[l l]. Interaction between analytes and the chiral sclector will depend on the stability of
the diastereomeric complex formed. In case the chiral selector is either bound to the
capillary surface or incorporated or bound to a gel matrix, the net velocity of the
19
CHAP1ER2
complex will be zero. In case the chiral selector is added to the BGE, the net velocity
of the complex will not necessarily be zero, but (in most cases) differ from the velocity
of the free analyte. Therefore, complex formation will result in an average velocity of
the analyte, which is different from the velocity of the free analyte. As a consequence,
a difference in complex stability between two optical antipodes, will result in a
difference in the average velocity of these compounds. In order to maximize
enantioselectivity, one should obviously maximize this difference in average velocity
between the two optica! antipodes.
The resolution attainable in any separation system is a function of both
selectivity and efficiency. As mentioned earlier, efficiencies obtained usually in CE are
very high and exceed the values usually obtained in HPLC. Some factors influencing
efficiency have been discussed in more detail in section 1.3. Selectivity is influenced by
chemica] and physical parameters. Parameters influencing enantioselectivity include the
pH of the BGE, the nature and concentration of chiral selector present in the BGE and
the capillary temperature. Obviously. the structure of the chiral selector will have a
decisive influence on the separation. The next section will give a brief survey of the
existing chiral selectors and their application in CE. In section 2.3, the chiral separation
mechanism will be discussed in detail using a mathematica! model which describes
mobility differences as a function of the pH and the concentration of the chiral selector.
2.2 Chiral selectors in CE
2.2.J Rules /or chiral recognition
In order to separate optica) isomers it is necessary to introduce a chiral element
into the separation process. For CE, this chiral element or chiralselector will, in most
cases, be added to the BGE. The addition however, of a chiral selector to a
electrophoretic system does not guarantee the successful separation of all optica]
isomers. The most important rule for chiral recognition is that the chiral selector must
be compatible in size and structure to the racemate; a minimum of three molecular
interactions bas to occur. These interactions can be both attractive or repulsive.
Possible modes of interaction include:
• Ion-ion bonds;
• Dipole-dipole bonds like hydrogen bonds;
• Van der Waals forces;
• Ion-dipole bonds.
20
CHIRAL SEPARA TION MECHANISM
Furthermore, only one of the two enantiomers needs to interact with the chiral se lector
via the three-point minimum mode. Not all interactions between the chiral selector and
the solute will meet this criterion; also achiral înteractions will occur. In these cases,
separation optimization should be accomplished by maximizing the 3-point 'chiral
interactions' at the expense of the non-chiral interactions.
2.2.2 Cyclodextrins
Cyclodextrins (CD's) are by far the most popular chiral selectors used in CE
and wil! therefore be discussed in more detail than the other chiral selectors mentioned
in this section. CD's are torus-
O CH,OH
HOH 2c/2o;~~
Uoy v" HOH~O OH CH20H
HOH2C 0 HO
OH !!HO o
0 CH,OH HO
HO OH OH
HOH,c o~o 0
O CH20H
Figure 2.1 Structure of [3-cyclodextrin
hydroxyl groups.
shaped cyclic D-gluco
oligosaecharides produced form
starch by enzymatic degradation.
Although CD's containing
between 6 to 12 D( + )-glucopyranose units have been
isolated, only those containing 6
(a-CD), 7 (~-CD) or 8 (y-CD)
residues are currently used. The
interior of the CD cavity is
relatively hydrophobie, while the
outside rim is more hydrophilic.
The rim on the wider side of the
CD cavity contains the chiral
secondary hydroxyl groups, while
the opposite smaller opening is
occupied by achiral primary
Figure 2.1 shows the structure of ~-CD, while the dimensions are sehematically
shown in T ABLE 2.1. The size of the hydrophobic cavity is sueh that, in genera!, the
a-CD can aeeommodate a single phenyl ring, while ~-CD and y-CD can aceommodate
substituted single- and multiple ring systems. This inelusion alone is not enough for
chiral recognitîon: interaction between substituents on the asymmetrie center of the
analyte and the hydroxyl groups on the CD-rim are responsible for chiral recognition.
The mechanism of inclusion complexation in CE is schematically shown in Figure 2.2.
21
CHAPTER2
TABLE 2.1 SOME PHYSICO-CHEMICAL PROPERTIES OF CYCLODEXTRINS
b
CD Dimensions Cavity Molecular Specific Solubility
volume mass optical in water
Á A3 (g/mol) rotation at 25°C
[a]~ (g /100 ml)
a b c
a. 5.7 13.7 7.8 174 972 +150 14.5
~ 7.8 15.3 7.8 262 1135 +162 l.85
y 9.5 16.9 7.8 427 1297 +177 23.2
Inclusion complex fonnation and the size of the analyte's binding constant to
the cyclodextrin are determined by several different factors. The most important
factors are the 'hydrophobic effect', which induces the apolar portion of the molecule
to preferentially reside in the cyclodextrin cavity, and hydrogen bonding between
appropriate polar segments of the guest molecule and the secondary hydroxyl groups
at the mouth of the cyclodextrin cavity. Other factors which can influence complex
formation are Van der Waals interactions, release of high energy water from the CD
cavity and a change in ring strain upon complexation.
22
CHIRAL SEPARA TION MECHANISM
Cation
> EOF
8
Figure 2.2 Schematic representation of the mechanism of inclusion complexation
with neutra! cyclodextrins in CE
Besides the already mentioned native cyclodextrins, a wide range of modified
cyclodextrins is comrnercially available. Actually, the first attempt to perform chiral
separation using CE was by the group of Smolkova-Keulemansova [ 12] using modified
cyclodextrins. They used ITP with the addition of dimethylated (DIME-8) or
trimethylated (TRIME-8) ~-cyclodextrin to the leading electrolyte. For DIME-8 and
TRIME-8, the hydroxyl groups on the CD-rim are (partially in case of DIME-B)
replaced by methoxy groups. Another modification comprises hydroxypropylated CD's
(HP-a-CD, HP-~-CD or HP-y-CD) [13], in which the various hydroxyl groups are
substituted by (O-C3H10H) moieties. The above mentioned modifications of the native
cyclodextrins obviously lead to a different stereoselectivity, but also to an improved
solubility. As mentioned earlier, the solubility of native ~-CD in water is not more than
16 mM, whereas e.g. the solubility of DIME-8 is as much as 200 mM. Depending on
the magnitude of the formation constants (see section 2.3), solubility can have a
limiting effect on chiral resolution.
Hydroxyl groups on the cyclodextrin rim can also be substituted by charged or
chargeable groups [14,15]. Firstly, the introduction of chargeable groups will result in
an increased solubility. Secondly, the use of these modified CD's allows the separation
23
CHAPTER2
of uncharged species. Furthermore, the separation mechanism is altered by the
introduction of electrostatic interactions. Finally, the use of chiral selectors carrying a
charge opposite to that of the analytes can greatly improve the mobility difference
between the two optical antipodes.This is explained in section 2.3.
The following chapter describes the use of uncharged ~-CD polymer for chiral
drug separation by CE. A negatively charged ~-CD polymer has been applied
successfully by Aturki and Fanali [16] for the separation of basic compounds of
pharmaceutical interest. Recent reviews give an excellent survey of the use of CD's in
CE including hundreds of references [ 17, 18]
2.2.3 Crown-ethers
Crown-ethers are macrocyclic polyethers capable of forming host-guest
complexes with especially inorganic and organic cations. Modification of the crown
ether by the introduction of four carboxylic groups makes it possible to use this class
of compounds as chiral selectors in CE.
The crown-ether can incorporate protonated primary amino compounds by
formation of ion-dipole bonds with the oxygen atoms of the chiral selector (Figure
2.3). The chiral crown-ether (18-crown-6-ether tetracarboxylic acid) can be used for
the chiral separation of several basic compounds [19].
Figure 2.3 Structure of the complex formed between 18-crown-6 and a protonated
primary amino compound
24
CHIRAL SEPARA TION MECHANISM
2.2.4 Macrocyclic antihiotics
A new, very promising class of chiral selectors are the macrocyclic antibiotics.
Vancomycin, rifamycin B and ristocetin A have proven to be highly selective towards
the enantiomers of a broad class of compounds [20-22]. These antibiotics are
amphilytic, and are strong UV-absorbers. However, in most cases, detection of the
analytes is not disturbed by the high background absorption of the chiral selector since
only very low concentrations of the antibiotics are needed. The chiral recognition is
obtained mainly by charge-charge interactions, hydrogen bonding, hydrophobic
inclusion and 1t-1t interactions. These interactions can be either attractive or repulsive.
2.2.5 Proteins
Also proteins have been applied successfully as chiral selectors in CE. One of
the characteristics of proteins is the isoprotic and the isoelectric point, pl. The protein
will be positively charged if pH < pl, and negatively charged if pH > pl. This indicates
that the pH will be a very important operating parameter for the optimization of chiral
selectivity. Similar to the charged CD-derivatives, it is possible to separate both
charged and uncharged species using this chiral selector. Among the many proteins
used as chiral selector in CE, bovine serum albumin (BSA) is most widely applied
[23,24]. The mechanism, involved in chiral recognition is comparable with that of
macrocyclic antibiotics.
2.2.6 Micelles
Micellar electrokinetic chromatography (MEKC) was first introduced by
Terabe et al.[25] for the separation of non-charged compounds. Terabe's group was
also the first to use chiral surfactants as micellar phase for the separation of optica]
isomers [26]. Both natura! surfactants such as bile salts, as wel! as optically active
amino-acid derived synthetic surfactants have been used as chiral selector in CE. New
chiral surfactants often have a low critica] micelle concentration, are highly soluble and
can be synthesized in both the L- and D-form [27]. The last feature makes it possible
to easily change the migration order of the optica] isomers. For the determination of
the optica! purity of e.g. drugs, it is highly favorable that the minor component
migrates in front of the major component.
25
CHAPTER2
2.2. 7 Other chiral selectors
Many other classes of compounds have been used as chiral selectors in CE. The
most important group, not mentioned so far, is probably the oligosaccharides
consisting of rnaltodextrin, heparin and dextran sulphate. As mentioned before, some
recent comprehensive reviews give an excellent survey on the state of the art of chiral
separations in CE [17,18,28]. Chapter 4 of this thesis will introduce ergot alk:aloids as
a novel class of chiral selectors in CE.
2.3 Determination of equilibrium constants of complex fonnation
In several early studies on chiral separations using CE, it was shown that the
concentration of the chiral selector influences the mobility of the optical antipodes
[6,29,30]. Moreover, it was shown that optical resolution varied with the
concentration of the chiral selector, and that resolution could reach a maximum value
at a certain optimum concentration of the chiral buffer additive.
A recent study of Rundlett and Armstrong [31] presents a survey of possible
approaches to determine formation constants. Wren and Rowe were the first .to apply
CE for the determination of formation constants between chiral analytes and
cyclodextrins [32]. Rawjee et al. [33] extended this model by including the pH as a
separation parameter. In both models, it is assumed that the concentration of the buffer
is much higher than the concentration of the CD, and that the concentratîon of the CD
is much higher than that of the analyte. The consequence of thîs assumption is that
practically all the CD will exist as CD-buffer complex, or in other words, the analytica)
concentration of CD is practically the same as the concentration of the CD-buffer
complex and wiU remain more or less constant, whether the analyte is present or not.
For this reason, the term CD and [CD] will be used in stead of the more proper terms
CD-buffer and its concentration [CD-buffer]. Chapter 5 will discuss the influence of
the buffer composition on complex formation in CE in more detail.
The model will focus on the complex formation between a racemic acid (HS
and HR) and CD. lt should be noted however that this model is not only applicable to
CD's, hut also to other chiral selectors, if the above mentioned condition, concerning
the concentration of the selector-buffer complex, is valid. The following equations
26
CHIRAL SEPARA TION MECHANISM
show the acid dissociation and complex formation equilibria with the corresponding
equilibrium constants:
KHR [W][Hp+]
[HR] (2-1)
KHS [S-][Hp+]
[HS] (2-2)
HR+CDHHRCD K _ [HRCD]
HRCD - [HR][CD] (2-3)
K :::: _[_RC_D_-_] RCD- [W ][CD]
(2-4)
HS+CDHHSCD K _ _ [_HS_C_D_]
HSCD - [HS][CD] (2-5)
s- +CD+-+ SCD- (2-6)
The acid dissociation constants of both optica! isomers are assumed identical and will
from now on be expressed as Ka. The analytica! concentration of HR and HS (C8 R and
CHs) can be expressed as:
CHs = [HSJ + [SJ + [HSCDJ + [SCDJ (2-7)
CHR =[HR]+ [RJ + [HRCDJ + [RCDJ (2-8)
The mole fractions of the negatively charged species are:
as· = [S] 1 CHs (2-9)
(2-10)
(2-11)
asco- = [SCD-} 1 CHs (2-12)
27
CHAP1ER2
By substituting eq. (2-1)-(2-6) in eq. (2-7)-(2-12), it is possible to eliminate the molar
concentrations of the acid related species:
( 2-13)
( 2-14)
(2-15)
(2-16) asco- = [H30+] 1 + KSCD_[CD]+-K-(1 + KHSCD[CD])
a
The effective mobilities of the optical isomers can be expressed as the mole-fraction
weighed ionic mobilities of the respective (charged) species (see eq. (1-6)),
(µR_ ,µ5_ ,µRm- ,µsco- ). The effective mobilities of the fully dissociated acids s- and K
( µ5
_ ,µR_) are equal, and from here on referred to asµ_.
(2-17)
(2-18)
These equations, combined with the expressions for the mole fractions of the
negatively charged species (eq. (2-13)-(2-16)), yield:
28
µRCD-1 +--KRCD [CD)
µ~ff = µ_ --------------[H,0+]
1+ KRCD_[CDJ+~(l + KHRCD[CD]) a
µ;tr = µ_ ------'--------[H,0+]
1 + KSCD_ [CD] +-·--(1 + KHSCD[CD]) Ka
CHIRAL SEPARA TION MECHANISM
(2-19)
(2-20)
Several definitions can be used to express (chiral) selectivity in CE. In this thesis we
will use:
• S, defined as the ratio of the effective mobility difference and the mean effective
mobility;
• SF, the separation factor, defined as the ratio of the equilibrium constants of
complex formation;
• ARts, defined as the ratio of the effective mobilities of both optical isomers.
For the sake of convenience, ARts, was chosen to express chiral selectivity. If we
introduce the relative complex mobility ( M' ):
M' R
µRCDµ_
M' = µSCD s µ_
we obtain:
(2-21)
(2-22)
29
CHAPTER2
From eq. (2-23) it can be seen that chiral selectivity is a function of the relative
complex mobility, the cyclodextrin concentration, the degree of dissociation of the
chiral acid, and the equilibrium constants of complex formation of both the dissociated
and the non-dissociated acid. Similar expressions can be derived for basic compounds
[34]. The mobility of the basic optica! isomer R (or HR") can be described as:
µf µ+ [on-1 1 + KHRCD'[CD] +--(1 + KRcv[CDJ)
K"
( 2-24)
in which µ+ is the mobility of the fully protonated basic enantiomer, in absence
of a chiral selector, and Kb is the base dissociation constant. In genera!, the mobilities
of the fully charged ions are referred to as Jlo.
Three different types of chiral interaction can be discriminated [33,35,36]:
• Type 0 or non-selective interactions; neither the charged nor the non-charged
enantiomers interact selectively with the chiral selector;
• Type l or desionoselective interactions; only the non-dissociated optica! antipodes
interact selectively with the chiral selector;
• Type II or ionoselective interaction; only the charged forms (the dissociated
solutes) interact selectively with the chiral selector;
• Type lil or duoselective; both the charged and the non-charged species interact
selectively with the chiral selector.
In genera!, K1 is used to quantify interaction between a chiral selector and a
non-dissociated acid or a non-protonated base, whereas Ki is used to describe the
interaction between the chiral selector and the dissociated acid or protonated base. Kc
can also be used to quantify complex formation.
Figure 2.4 shows the chiral selectivity plot of a ionoselective compound
interacting with a cyclodextrin with realistic values for the chiral parameters. From this
plot it can be seen that the migration order of the optica) isomers can be reversed by
changing the concentration of CD or the pH of the BGE. The Jatter will influence the
degree of dissociation a. This is only observed for ionoselective and duoselective
interactions. Reversal of the migration order of the optica! isomers by changing either
the pH or the CD-concentration is not possible for non-ionoselective interactions.
The equation derived by Wren and Rowe [32], which describes the mobility of
the optica! isomers interacting with a chiral selector, is found if full dissociation or
30
CHIRAL SEPARATION MECHANISM
protonation of the acidic respectively basic compounds is assumed, i.e. [H30+]/K. resp.
[Off]/Kb « 1 (see eq. (2-19), (2-20) and (2-24)). For acids it follows:
( 2-25)
1.10 1.10
1.05
1.05
1.00
1.00 ~
< 0.95
~ < 0.95
0.90
0.85
l.00
Figure 2.4 Three-dimensional chiral selectivity suiface for an ionoselective chiral
compound. KHRCD =KHscD =500, M;_ M;_ = 0.25, KRcD- = 100, KsCD 120.
And, with the commonly valid assumption thatµRCD- µsCD-, it follows that
the mobility difference between the optica] antipodes can be expressed as:
31
CHAPTER2
( 2-26)
where JJ-c is the mobility of complex. Tuis equation is valid, both for anions as well as
for cations. A closer look to this equation learns that the mobility difference of the two
enantiomers is linearly proportional to the mobility difference between the free (µ., µ+
or µo) and complexed (JJ-c) species. This explains the high selectivities usually obtained
by employing CD's, carrying a charge opposite to that of the analyte. The optimum
CD-concentration, resulting in a maximum mobility difference of both optica! isorners
is found when:
i)Aµelf iJ[CD) =O
and thus follows:
(2-27)
( 2-28)
For the exarnple given in Figure 2.4, this results in [CD]op1= 11 mM (assuming full
dissociation of the acidic analyte ).
References
1. H. Nishi, T. Fukuyama and M. Matsuo, J. Microcol. Separ., 2 ( 1990) 234
2. W. Schutzner, S. Fanali, A. Rizzi and E. Kenndler, J. Chrornatogr" 639 (1993)
375
3. W. Schutzner, G. Caponecchi, S. Fanali, A. Rizzi and E. Kenndler,
Electrophoresis, 15 (1994) 769
4. C. Vogt, A. Georgi and G. Werner, Chromatographia, 40 (1995) 287
5. S. Terabe, Trends Anal. Chem" 8 (1989) 129
6. S. Fanali, J. Chrornatogr" 474 (1989) 441
7. K.D. Altria, D.M. Goodall and M.M. Rogan, Chrornatographia, 37 (1993) 475
8. R. Kuhn and S. Hoffstetter-Kuhn, Chromatographia, 34 (1992) 505
32
CHIRAL SEPARA TION MECHANISM
9. A. Guttman, A. Paulus, A.S. Cohen, N. Grinberg and B.L. Karger, J.
Chromatogr" 488 (1988) 41
10. I.D. Cruzado and G. Vigh, J. Chromatogr" 608 (1992) 421
ll. S. Mayer and V. Schurig, J. High Resol. Chromatogr" 15 (1992) 129
12. J. Snopek, 1. Jelinek and E. Smolkova-Keulemansova, J. Chromatogr.
438(1988)211
13. B. Chankvetadze, G. Endresz and G. Blaschke, J. Chromatogr. A, 700 (1995)
43
14. C. Dette, S. Ebel and S. Terabe, Electrophoresis, 15 (1994) 799
15. C. Desiderio and S. Fanali, J. Chrornatogr. A, 716 (1995) 183
16. A. Aturki and S. Fanali, J. Chrornatogr. A, 680 (1994) 137
17. H. Nishi and S. Terabe, J. Chromatogr. A, 694 (1995) 245
18. S. Fanali, J. Chromatogr. A, 735 (1996) 77
19. R. Kuhn, F. Emi, T. Bereuter and J. Hausler, Anal. Chem" 64 (1992) 2815
20. D.W. Armstrong, Y.B. Tang, S.S. Chen Y.W. Zhou, C. Bagwill and J.R. Chen,
Anal. Chem., 66 (1994) 1473
21. D.W. Armstrong, K.L. Rundlett and J.R. Chen, Chirality, 6 (1994) 496
22. D.W. Armstrong, K.L. Rundlett and G.L. Reid, Anal. Chem., 66 (1994) 1690
23. S. Busch, Kraak and H. Poppe, J. Chromatogr., 635 (1993) 119
24. Y. Tanaka and S. Terabe, J. Chromatogr. A, 694 (1995) 277
25. S. Terabe, K. Otsuka, K. Ichikawa, A. Tsuchiya and T. Ando, Anal. Chem" 56
(1984) 111
26. S. Terabe, H. Shibata and Y. Miyashita, J. Chromatogr., 480 (1989) 403
27. D.C. Tickle, G.N. Okafo, P. Camilleri, R.F.D. Jones and A.J. Kirbu, Anal.
Chem., 66 (1994) 4121
28. H. Nishi, J. Chromatogr. A, 735 (1996) 57
29. S. Fanali, J. Chromatogr., 545 (1991) 437
30. H. Nishi, T. Fukuyama ans D. Terabe, J. Chromatogr" 553 (1991) 503
31 K.L. Rundlett and D.W. Armstrong, J. Chromatogr. A, 721 (1996) 173
32. S.A.C. Wren and R.C. Rowe, J. Chromatogr" 603 (1992) 235
33. Y.Y. Rawjee, D.U. Staerk and G. Vigh, J. Chromatogr., 635 (1993) 291
34. Y.Y. Rawjee, R.L. Williams and G. Vigh, J. Chromatogr. A, 652 (1993) 233
35. Y.Y. Rawjee, R.L. Williams, L.A. Buckingham and G. Vigh, J. Chromatogr. A,
688 (1994) 273.
36. J.C. Reijenga, B.A. Ingelse and F.M. Everaerts, J. Chromatogr. A, in press
33
CHAPTER2
34
POLYMER
3. ENANTIOMERIC SEPARATION OF DRUGS BY CE
USING A SOLUBLE NEUTRAL ~-CYCLODEXTRIN
POLYMER
Abstract
The separation of the enantiomers of several basic compounds of phannaceutical
interest and three tryptophan derivatives was investigated by capillary electrophoresis
employing a soluble neutral {3-cyclodextrin polymer. The eff ects of the composition of
the background electrolyte on the effective mobility and the resolution and selectivity
of the optica! isomers were examined. An increased concentration of the chiral
polymer led to a decreased mobility of the analytes. Both selectivity and resolution
were influenced by the concentration of the {3-cyclodextrin polymer. Also, it was
shmvn that increasing the ionic strength of the background electrolyte could lead to
increased resolution. The addition of different organic additives to the background
electrolyte generally resulted in a decrease of resolution. However, in some cases,
e.g. ephedrine, the organic solvent proved to be essential to achieve enantiomeric
separation. Furthennore, the influence of the capillary temperature on the resolution
of the racemic analytes was investigated. lncrease of temperature had a deleterious
effect on the resolution of the enantiomers. For ephedrine however, relatively high
temperatures proved to be advantageousfor the resolution of the optica[ isomers.
The study presented in this chapter has resulted in the following publications:
B.A. lngelse, F.M. Everaerts, C. Desiderio and S. Fanali, J. Chromatogr. A, 709 (1995) 89-98
B.A. Ingelse F.M. Everaerts, J. Sevcik, Z Stransky and S. Fanali. J. High Res. Chromalogr" 18 (1995)
348-352
J. Sevcik, Z. Stransky, B.A. Ingelse and K. Lemr, Journal of Pharmaceutical and Biomedical Analysis, 14
(1996) 1089-1094
3.1 Introduction
The development of new chiral substances, especially in the pharmaceutical
field, places increasing demands on analytica! methods, for the separation of these
35
CHAPTER3
kinds of isomers for, e.g., the chiral purity control of drugs and pharmacokinetic
studies.
When cyclodextrins (CD's) or their derivatives are used for the separation of
optica! isomers, the chiral resolution is based on selective inclusion complexation with
analytes. Hydrophobic interactions between analytes and the CD cavity and hydrogen
bonding between analytes and the hydroxy (or modified) groups on the CD rim can
lead to the formation of labile diastereomeric cömplexes with different stability
constants. The optica! isomer that forms the most stable complex with the neutra] CD
wil! migrate with the lowest effective mobility.
Cyclodextrins have been widely applied as chiral selectors in CE for many
applications [1-4]. Derivatization ofCD's can change both selectivity and solubility. A
chargeable ~-cyclodextrin polymer was used for the enantiomeric resolution of several
basic compounds [5]. Nishi et al. studied the enantiomeric resolution of trimetoquinol
and related substances using uncharged ~-cyclodextrin polymer (EP-~-CD). This
polymer consists of ~-cyclodextrin cross-linked with epichlorohydrin. A similar
polymer has been used earlier in HPLC for the chiral recognition of warfarin and
dansylated amino acids [6].
CE has proven to be a powerful technique for the separation of the optica!
isomers of drugs, as extensively reviewed by Nishi and Terabe [7,8] and Fanali [9]. In
this study, the use of chiral ~-CD polymer for the separation of enantiomers of several
basic compounds of pharmaceutical interest, was investigated by CE. The effect of the
polymer concentration on the effective mobility, the resolution and the selectivity was
studied. Furthermore the role of the organic solvent added to the BGE, the ionic
strength and composition of the BGE, and the capillary temperature were examined.
3.2 Experimental
3.2.1 Chemicals
Soluble ~-cyclodextrin polymer (EP-~-CD) was purchased from Cyclolab
(Budapest, Hungary). The characteristics of the polymer were: molecular weight m = 3000-5000; cyclodextrin content: 58.2 %; solubility in water: 40-50%; cross linking
agent: epichlorohydrin. All standards were of analytica! grade and were obtained from
Sigma (St. Louis, MO, USA). The structures of the analytes are shown in Figure 3.1.
36
POLYMER
3.2.2 Apparatus
A P/ACE 2200 capillary electrophoresis system (Beckman, Fullerton, CA) was
used for all the experiments in the temperature range from 20°C up to 50°C. The
Beckman instrument used an untreated fused-silica capillary, 370 mm x 50 µm LD.,
with an effective length of 300 mm. The UV-detector was operated at 214 nm.
3.2.3 Methods
All solutions were prepared in demineralized water. For most of the
electrophoretic experiments a 50 mM phosphate buffer, pH 2.5, was used. The buffer
was prepared by titrating a 50 mM phosphoric acid solution with NaOH. The
background electrolyte (BGE) containing EP-P-CD was filtered before use with a 0.45
µm pore size filter (Lida, Kenasha, WI). The concentration of the analyzed standards
was 5. l0.5M. The applied voltage was 15 kV. Before every electrophoretic run, the
capillary was rinsed with 10 mM KOH for two minutes and with phosphate buffer
(without EP-P-CD) for two minutes. Before applying the sample, the capillary was
rinsed with BGE containing a specific EP-P-CD concentration for 20 seconds. No
polymer was present in the electrode vials during separation. In this setup, only a few
µl of BGE containing the chiral polymer were needed per analysis. The absolute
consumption of EP-P-CD per analysis was less than 1 mg. All experiments were
performed at 20°C, except for the experiments to study the influence of temperature
on chiral separation.
3.3 Results and discussion
Different basic cornpounds of pharmaceutical interest, namely a-adrenergic
agonîsts (ephedrine, epinephrine and norepinephrine), P-adrenergic agonists
{isoproterenol, terbutaline and clenbuterol), P-adrenergic blockers (atenolol,
metoprolol, oxprenolol and propranolol), anaesthetics (ketamine and bupivacaine),
anorexie (norephedrine and methamphetamine) and tryptophan methyl, ethyl and butyl
esters were selected for the electrophoretic experiments. Deprenyl, methamphetamine
and ephedrine are known as drugs of abuse. These compounds represent particular
steps of the synthesis of R-(-) deprenyl (selegiline), where ephedrine is the starting
compound and methamphetamine an intermediate.
37
CHAPTER3
Clenbuterol
Buph·aeaine
OH CH3 CH,.-CH-CH2-NH-CH
ó CH3
Ç,-a<,.o-CH, Metoprolol
0
Q)5 Cl
Ketamine
OHh_OH
~-J:H·CH,-NH-C(CH3)3
OH Terbutaline
cq ~Hl CH2-CH-COOCH3
Tryptophan methyl ester
~CH,-CH-CH, ~-,-
NHCH,
Methamphetamine
Atenolol
lsoproterenol
OH
o-cH,-tH-CH;rNH-CH(CH3)2
©-O-CH2-CH.CH,
Oxprenolol
Propranolol
r--.Nis. ~. NH2
CH,-CH-COOC2'i5
Tryptophan ethyl ester
Figure 3.1 Structures of the studied compounds.
38
Deprenyl
CH, OH-CH-CH:_NH-CH, © .
Ephedrine
NH 2
OH-CH-CH-CH, ©. Norephedrine
Epinephrine
Norepinephrine
f""r-N\ ~. NH2
CH,-CH-COOC4H9
Tryptophan bntyl ester
P-CYCLODEXTRIN POL YMER
3.3.1 Effect of the concentration of EP-fj-CD on the mobility of the analytes
The racerrric mixtures were run in a phosphate buffer at pH 2.5 in the absence
of chiral additive. Owing to the protonation of the basic nitrogen atom, the analytes
moved towards the cathode. (The pKa values of most of these analytes vary in between
8 and 10, and therefore full protonation can be assumed). As expected, no
enantiomeric separation was obtained.
Under the experimental operating conditions (pH 2.5), the electroosmotic flow
(EOF) was relatively low compared to the effective mobility of the analytes. In order
to separate the optica) isomers, the BGE was supported with different amounts of EP
P-CD. Upon addition of EP-P-CD, the EOF decreased even more. This is partly
explained by the increased viscosity of the BGE upon addition of the polymer. Another
explanation rrright be the absorption of EP-~-CD to the capillary surface. The effect of
the polyrner concentration on the velocity of the electroosmotic flow is shown in
Figure 3.2.
0 50 100 150 200
conc. EP·P·CD [mg/ml]
Figure 3.2 Effect of the concentration of EP-/3-CD on the electroosmotic flow. BGE:
50 mM phosphate, pH 2.5. Capillary 37 cm x 50 µm, effective length 30 cm.
Separation voltage: 15 kV. Capillary temperature: 20°C.
39
CHAPTER3
25
20
5
0 50 100
coru:. EP·~·CD [mg/ml]
.._Selegiline
--Methamphetamine ....-Clenbuterol
150 200
Figure 3.3 Effect of the concentration of EP-fJ.CD on the effective mobility of
methamphetamine, deprenyl and clenbuterol. Experimental conditions as in Figure
3.2.
25
20
';;
"~ 15 e
" . ... ~ lO
J 5
0
0 20 40 60
oom:. EP·J\..CD [mg/ml]
.._Oxprenolol
-tl- Atenolol
....-Propranolol
--Metoprolol
80 100
Figure 3.4 Effect of the EP·f3·CD polymer concentration on the effective mobility of
f3·adrenergic blockers. Experimental conditions as in Figure 3.2.
40
~-CYCLODEXTRIN POL YMER
Figure 3.3 shows the effect of the concentration of the chiral polymer added to the
BGE, on the effective mobility of racemic deprenyl, methamphetamine and clenbuterol.
The effective mobilities were not corrected for changes in electrolyte viscosity, since,
in this study, complexation is not expressed in a quantitative way. An increase in the
concentration of EP-~-CD leads to a genera) decrease of the mobility of the analytes.
The analytes form labile diastereomeric complexes with the cyclodextrin polymer
during the electrophoretic run, which causes a decrease of the electrophoretic mobility.
This decrease depends mainly on the concentration of the chiral selector and on the
stability constant of the complex formed. Also for all the other analytes, the mobility
was determined as a function of the EP-~-CD concentration.
According to the results presented in Figure 3.4 the formation constants of EP
~-CD towards the ~-adrenergic blockers decreases in the order propranolol ""
metoprolol > atenolol "' oxprenolol. (In this figure, no distinction is made between the
mobility of the (+) and (-)-isomers). Wren and Rowe (10] proposed to relate the
interaction between ~-blockers and cyclodextrins to the hydrophobicity of the analytes.
This means that the most hydrophobic compound has the strongest interaction with the
hydrophobic cavity of the CD, and hence will have the largest value of formation
constant. The influence of the hydrophobicity of the . analytes on the inclusion
complexation is best illustrated by the results obtained with the methyl, ethyl, and butyl
ester of tryptophan.
Figure 3.5 shows the effect of the polymer concentration on the effective
mobility of these tryptophan derivatives. The strongest complexation was observed for
the most hydrophobic ester (TBE), whereas the most hydrophilic ester (TME) showed
the weakest interaction with EP-~-CD. This behavior is in accordance with that
described in Ref. [ t O]. Also in this case, an increase in the concentration of the
complexing additive caused a genera) decrease in the effective mobility of all three
esters. A similar effect was obtained for the two anesthetic cornpounds studied, where
bupivacaine was more retarded than ketamine.
41
CHAP1ER3
25
20
'iii' "'t:. 15 E
" . ... - IO ~
5
0
0 20 40 60 80 100
com:. EP·f}-CD [mg/mlJ
Figure 3.5 Dependence of the effective mobility of three tryptophan esters on the
concentration of EP-/3-CD. Different mobilities between the optica/ isomers of TBE are
indicated with the solid and dotted lines. Experimental conditions as in Figure 3.2.
Symbols as in TABLE 3.1.
3.3.2 Effect of the concentration of EP-{3-CD on chiral recognition
To study the chiral recognition of the polymer, the analyses were performed
using a BGE at pH 2.5 supported with different amounts of EP-1}-CD in the range of
0-200 mg/ml. The amount of l}-cyclodextrin units per milligram of polymer could be
calculated according to the specification of the manufacturer (CD content 58.2 %
wlw). Consequently, taking into account the molecular weight of P-CD (m = 1135
g/mol), 200 mg/ml EP-P-CD should be compared with 100 mM (or 100 meq/ml) PCD. In aqueous BGE's, !}-CD is soluble up to 20 mM.
Resolution was obtained for all the analytes, except for ephedrine and
metoprolol. Propranolol, norephedrine and oxprenolol were only partly resolved
(resolution up to 0.8) whereas the maximum resolution for the optica! isomers of
atenolol was smaller than 0.5.
Figure 3.6 shows the effect of the concentration of EP-1}-CD on the selectivity
(S) as defined by the ratio of the mobility difference and the effective mobllity of the
butyl, ethyl, and methyl ester of tryptophan, in the concentration range of 0- l 00 mg
EP-1}-CD I ml. An increase in the concentration of P-cyclodextrin polymer added to
the BGE at pH 2.5 led to a genera] increase in resolution and selectivity for the methyl
42
~-CYCLODEXTRIN POL YMER
and ethyl ester of tryptophan. The butyl ester shows a clear maximum in selectivity.
The data presented in this figure confirm the previous findings [ 11] and the theoretica!
model discussed by Wren and Rowe [12] concerning the existence of a maximum in
selectivity at a certain concentration of chiral selector.
This maximum in selectivity is only observed for TBE and propranolol in the
examined range of concentration of chiral selector, as can be found in T ABLE 3.1.
Consequently, according to Ref. [10], it can be concluded that complex formation with
these compounds ·is stronger than with all the other analytes which show optica]
resolution.
TABLE 3.1 SELECTIVITY OF THE INVESTIGATED ANALYTES AS A
FUNCTION OF EP-P-CD CONCENTRATION AND AMOUNT OF MEOH (% IN
BGE).
no organic solvent 10% methanol 30% methanol
EP-~-CD 10 20 50 100 200 20 50 100 200 100 150 300
(mg/ml)
Deprenyl 0.006 0.016 0.027 0.036 0.048 0.010 0.018 0.028 0.034 0.020 0.029 -
Amphetamine 0 0.013 0.023 0.032 0.044 0.008 0.017 0.024 0.038 0.017 0.019 -
Clenbuterol 0.010 0.019 0.037 0.047 0.064 0.014 0.021 0.045 0.053 0.040 - -
Propranolol 0.011 0.016 0.015 0.010 0 0.014 0.016 0.015 0 0.025 - -
Terbutaline 0.034 0.056 0.087 0.099 - - 0.054 0.085 - 0.066 - -
TBE' 0.056 0.062 0.062 0.041 - - 0.045 0.046 - 0.042 - -
TEEb 0.009 0.015 0.023 0.031 - - 0.019 0.028 - 0.027 - -
TME' 0.004 0.010 0.012 0.021 - - 0.014 0.021 - 0.020 - -
i-Proterenol 0.004 0.010 0.023 0.031 - - 0.016 0.025 - 0.024 0.025 -
Ephedrine 0 0 0 0 0 0 0 0 0 0.006 0.010 0.019
Norephedrine 0 0 0.005 0.010 0.013 - 0.004 0 - 0.009 0.015 0.026
Epinephrine 0 0.009 0.013 0.021 0.034 0 0.009 0.016 0.031 0.009 0.017 0.020
Norepinephrine 0 0.003 0.011 0.017 - - 0.006 0.011 - 0.004 0.009 0.023
Ketamine 0 0 0.007 0.014 0.020 - 0 0.010 - 0.006 0.009 -
Atenolol 0 0 0 0.006 - - 0 0.004 - 0.004 0.007 -
Bupivacaine 0 0 0 0.011 0.019 - 0 0.008 - 0 0.006 -
Metoprolol 0 0 0 0 - - 0 0 - 0.005 0.007 0
Oxprenolol 0 0 0.008 0.011 0.013 - 0 0.009 - 0.008 0.009 -
a) tryptophan butyl ester
b) tryptophan ethyl ester
c) tryptophan methyl ester
Metoprolol and ephedrine might still form stronger complexes with the polymer,
without forming enantioselective complexes.
43
50%
100
-
-
-
-
-
-
0.014
0.011
0
0.007
0.007
-
0
0
0
0
0.005
0
CHAP1ER3
0.07
0.06
0.05
~ 0.04 Ie ... 'S! 0.03
~ ;;J 0.02
0.01
0
0 20 40 60
oonc. EP·P·CD [mg/ml]
80
..,._TBE --TEE ....... TME
100
Figure 3.6 Selectivity of tryptophan esters versus concentration EP-/3-CD.
Experimental conditions and symbols as in Figure 3.2 and TABLE 3.1.
Figure 3.7 shows chiral separation of the tryptophan esters in presence of 20
mg/ml EP-P-CD. Another example of the separating power of EP-(3-CD is shown in
Figure 3.8. This electropherogram shows the enantiomeric separation of deprenyl,
amphetamine and clenbuterol, using a BGE containing 200 mg/ml EP-P-CD.
The migration order of the enantiomers was verified for deprenyl,
methamphetamine, propranolol, epinephrine, norepinephrine, isoproterenol and the
methyl and ethyl ester of tryptophan by spiking the racemic mixtures with optically
pure isomers (commercially available) and performing electrophoretic experiments.
The L-isomers of the tryptophan esters moved faster than the D-forms. For the other
analytes, the (-)-enantiomers rnoved faster than the ( + )-enantiomers, indicating that the
( + )-antipodes formed complexes with higher stability constants than their optica)
antipodes.
The enantiomeric separation power of EP-P-CD was compared with that of
native P-CD. The resolution of propranolol, terbutaline, tryptophan butyl ester and
epinephrine obtained at 2.5, 5, 10, and 20 mM P-CD was compared with the resolution
obtained at 5, 10, 20 and 50 mg/ml EP-P-CD (50 mg/ml of EP-P-CD should be
compared with 25 mM P-CD but aqueous solutions of > 20 mM P-CD cannot be
prepared).
44
J3-CYCLODEXTRIN POL YMER
0.0085
TME
s 0.0045 ~
TEE " " TBE c " -= " j
0.0005 <
-0.0035 +------+-----1-------+-------l 0 10 15 20
time[min]
Figure 3. 7 Electropherograms for the enantiomeric separation of racemic mixtures of
tryptophan esters. BGE: 50 mM phosphate buffer, pH 2.5 containing 20 mglml EP-{3-
CD. Other experimental conditions as in Figure 3.2. Symbols as in TABLE 3.1.
0.0025
s 0.0005 ~ " "
i < -0.0015
CLE
DEP MAT
-0.0035 -+------+------+------+------+------!
0 5 10 15 20 25
time[min]
Figure 3.8 Electropherogram of the enantiomeric separation of methamphetamine
(MAT), deprenyl (DEP) and clenbuterol (CLE). BGE: 50 mM phosphate buffer
containing 200 mglml EP-{3-CD. Other experimental conditions as in Figure 3.2.
Propranolol and the butyl ester of tryptophan were not resolved at any
concentration of P-CD, as shown earlier for propranolol by Fanali [11], whereas
epinephrine showed R, = 0.3 at 20 mM. These three analytes could be separated into
45
CHAPTER3
their enantiomers with the 13-CD-polymer (for propranolol R, = 0.8 at 20 mg/ml EP-13-
CD, for TBE R, = 2.7 at 50 mg/ml EP-13-CD and for epinephrine R" = 2.6 at 200
mg/ml EP-13-CD). Only for terbutaline, the resolving power of the native 13-CD was
higher than that of the polymer, at comparable concentrations of the chiral selector; at
2.5 mM 13-CD, Rs was 1.26, whereas 5 mg/ml EP-13-CD gave R, = 0.85. Fanali [11]
also successfully separated the optica! isomers of terbutaline. He obtained R, = 2, at a
13-CD concentration of 15 mM. The resolving power of EP-13~CD towards the
enantiomers of terbutaline exceeds that of native 13-CD at higher concentrations of the
chiral polymer (R, = 7.5 at 100 mg/ml EP-13-CD).
Another study [4] shows the enantiomeric separation of norephedrine,
ephedrine, norepinephrine, epinephrine, and isoproterenol using modified
cyclodextrins. According to this study, the native 13-CD was not able to resolve these
drugs into their enantiomers. Only very poor resolution was obtained for the optica]
isomers of ephedrine and isoproterenol at relatively high concentrations of 13-CD (20
mM). Using EP-13-CD, these enantiomers could be easily baseline resolved, except for
racemic ephedrine. The higher resolving power can also be supported by previous
results obtained for the methyl ester of tryptophan, which could not be resolved with
13-CD [13].
The high resolution capacity of EP-13-CD, compared to native 13-CD, towards
the compounds studied could be interpreted by considering the structure of the chiral
selector. In fact, the polymerization changes the properties of the 13-CD units. As
stated earlier, the CD content in the polymer is Jess than 60% w/w. According to the
manufacturer, about 40 % of the product consists of epichlorohydrin branches. Also,
the cooperation of two CD rnoieties of the polymer for inclusion complexation with
analytes possessing more than one guest part in their structure, must be considered
[14].
3.3.3 Effect of buffer concentration on the resolution of the enantiomers
The experiments for the study on the effect of the buffer concentration on
enantiomeric resolution were carried out on bupivacaine enantiomers. As expected, an
increase in buffer ionic strength caused a general increase in migration time.
46
f)-CYCLODEXTRIN POL YMER
2
1.6
-,... 1.2
~
j !
0.8
0.4
0
0 50 100 150 200
ronc. phospate in BGE [mMJ
Figure 3.9 Dependence of the resolution, Rs, of bupivacaine enantiomers on the
concentration of the BGE containing 200 mg/ml EP-{3-CD. Experimental conditions
as in Figure 3.2.
0.0005
200mM
~ j -0.0015
j
25mMN
-0.0035 +----------1---------+------------< 10 15 20 25
time[min)
Figure 3.10 Electropherograms for racemic bupivacaine at two different
concentrations of phosphate buffer containing 200 mg!ml EP-{3-CD. All other
experimental conditions as in Figure 3.2 ..
47
CHAPTER3
At the same time, resolution increases with increasing concentration phosphate in the
BOE. This is shown in Figure 3.9. Figure 3.10 shows the electropherograms of the
enantiomeric separation ofbupivacaine at 25 and 200 mM phosphate buffer, containing
200 mg/ml ~-CD polymer. The concentration of the sample is adapted to the
concentration of the surrounding BGE, according to Kohlrausch law [ 15]. Obviously,
the efficiency of the separation increases, whereas the ionic strength is not expected to
have a major intluence on the enantioselectivity ( èxpressed in terms of separation
factor (SF) which is defined as the ratio of the formation constants of both optica]
isomers). This concentrating effect of high ionic strength BGE's is counterbalanced by
an increased Joule-heating.
3.3.4 The effect of organic solvent added to the BGE
As outlined by Wren and Rowe [16], the addition of methanol reduces the
affinity of the analyte for the cyclodextrin cavity, due to the decreased polarity of the
solvent, i.e. the formation of an inclusion complex is less favored by the high energy
release of water out of the cyclodextrin cavity with an increase of the concentration of
methanol.
Comparing the electrophoretic data obtained by running the analytes with
BGE's containing 100 mg/ml of EP-~-CD in absence and in presence of organic
solvents (MeOH, isopropylalcohol (IPA) and acetonitril (ACN)}, a genera! increase in
migration time was obtained. The addition of 10% (vlv) tetrahydrofuran (THF), on the
other hand, led to a decrease of migration times for all the analytes investigated. The
increase in migration time may be explained by the increase of the viscosity. The
decrease in migration times upon adding 10% (vlv) ofTHF to the BGE containing the
neutra! polymer may be explained by a reduction of the complex formation constant
between the analytes and the cyclodextrin cavity.
T ABLE 3.1 shows the calculated values of the selectivity S as a function of the
concentration of MeOH added to the BGE and the amount of chiral selector added to
the BGE. A genera! decrease of both enantiomeric selectivity and resolution was
observed upon adding the organic solvent to the BGE, however S was enhanced for
some of them. For example, propranolol showed a maximum selectivity of 0.025 when
the BGE contained 30% (vlv) MeOH and 100 mg/ml of EP-~-CD, whereas the
maximum selectivity without the addition of MeOH was 0.016 at a chiral selector
concentration of 20 mg/ml.
48
0.002 a
0
J ! -0.002
10
0 b
-0.001
0%
15 20 25 30
thne [min]
35 40
S -0.002 +--------------------'A s Ij ; -0.003
1-0.004 r------------------' -0.005 t_ ........ ______________ j
6 8 to 12 14
time [min]
16 18
POLYMER
30%
45 50
0%
20 22
Figure 3.11 Electropherograms of the enantiomeric separation of a) propranolol and
b) ephedrine using different concentrations of methanol as organic modifier in the
BGE. Concentration of EP-/3-CD in the experiments with propranolol and ephedrine
is 100 mglml (a) and 200 mglml (b) respectively. Other experimental conditions as in
Figure 3.2.
49
CHAPTER3
The most pronounced effect was observed for the enantiomers of metoprolol and
ephedrine. In fact, these compounds did not show resolution at any concentration of
EP-~-CD without the presence of MeOH in the BGE. Still, no baseline separation for
the optica! isomers of these two compounds could be obtained by the addition of the
organic modifier.
Figure 3. l l shows the electropherograms of the enantiomeric separations of
propranolol and ephedrine using BGE's containing different amounts of methanol. It is
noteworthy to pay attention to the change in migration times upon the addition of
MeOH. The peaks of propranolol are shifted to much longer migration times, whereas
the migration times of ephedrine enantiomers are not markedly influenced by the
addition of the organic modifier. The addition of MeOH influences the mobility of an
analyte in several ways. Firstly, the viscosity of the BGE will increase, thus causing a
decreased mobility. On the other hand, as mentioned earlier, the decreased affinity of
the analyte towards the cyclodextrin cavity will mostly result in an increased effective
mobility. These effects are countereffective and the net effect will depend on the nature
of the analyte. However, in genera) a slightly decreased mobility was observed for
most analytes, with the increase of the amount of MeOH added to the BGE.
TABLE 3.2 shows the effect of the presence of ACN, IPA, MeOH and THF in
the BGE at pH 2.5, containing 100 mg/mJ of EP-~-CD, on the enantioselectivity of the
studied compounds. Cornparing these results with those obtained in the absence of
organic modifier, a general decrease of selectivity with IPA and THF is observed,
excluding the results obtained for propranolol. The maximum selectivity of the
propranolol enantiomers was obtained in a BGE containing 10% IPA, supported with
100 mg/mJ EP-~-CD. Acetonitril enhances the selectivity of the chiral polymer towards
the optica! isomers of epinephrine.
Methanol seems to be the most appropriate organic modifier, with respect to
the enhancement of chiral selectivity for the compounds studied. In most cases
however, it appears to be inadvisable to add any of the above mentioned organic
modifiers to the BGE.
50
~-CYCLODEXTRIN POL YMER
TABLE 3.2 SELECTIVITY OF THE INVESTIGATED ANAL YTES IN A BGE
CONTAINING 100 MG/ML EP-P-CD AND 10% (VN) OF DIFFERENT
ORGANIC ADDITIVES.
l0%IPA l0%ACN 10%THF 10%MeOH 100% H20
(mg/ml) 100 100 100
Selegiline 0.011 0.023 0 0.028 0.036
Amphetamine 0.012 0.022 0 0.024 0.032
Clenbuterol 0.021 0.035 O.Ol l 0.045 0.047
Propranolol 0.017 0.012 0.013 0.015 0.010
Terbutaline 0.042 0.069 0.018 0.085 0.099
TBE 0.047 0.032 0.031 0.046 0.041
TEE 0.013 0.021 0 0.028 0.031
TME 0.009 0.015 0 0.021 0.021
i-Proterenol 0.005 0.019 0 0.025 0.031
Ephedrine 0 0 0 0 0
Norephedrine 0 0.008 0 0 0.010
Epinephrine 0.006 0.030 0.008 0.016 0.021
Norepinephrine 0.005 0.010 0.006 0.011 0.017
Ketamine 0.006 0.008 0.008 0.010 0.014
Atenolol 0 0 0 0.004 0.006
Bupivacaine 0 0.008 0 0.008 0.011
Metoprolol 0 0 0 0 0
Oxprenolol 0 0.008 0 0.009 0.011
3.3.5 The effect oftemperature
In order to study the effect of the capillary temperature on the optica!
resolution of the analytes using EP-P-CD, experiments were carried out varying the
temperature between 20°C and 50°C. As expected, the migration time of all analytes
decreased by încreasing the temperature of the capillary [3]. Most obvîously, înclusion
complexation is influenced by temperature [ 17]; the increased temperature will
decrease the stability of the labile diastereomerîc complexes formed between the
analytes and the chîral selector. The resolutîon R, decreased by încreasing the
temperature for all analytes, except for ephedrîne that showed no noticeable change of
R, at the highest and the lowest temperature (0.8 and 0.7 at 50°C and 20°C,
respectively, at 300 mg/ml EP-P-CD and 20% MeOH (v/v) added to the BGE).
Moreover, the analysis time was decreased from about 35 minutes to less than 15
minutes. The slight increase in enantioresolution for ephedrine might be explained by
the increase in efficiency. The electropherograms are shown in Fîgure 3.12. The most
51
CHAPTER3
dramatic decrease in resolution was recorded for terbutalline, R, = 7.5 and 2.2. at 20°C
and 50°C, respectively, using 100 mg/ml EP-~-CD.
0
-0.0005
S' ~ ·0.001
j -0.0Hl
<
-0.002
10
50°C
15. 20 25
time[min]
20°C
30 35
Figure 3.12 Electropherograms of the enantiomeric separation of ephedrine
enantiomers at 20°C and 50°C. BGE: 50 mM phosphate buffer (pH 2.5), 20% MeOH,
supported with 300 mg!ml EP-f:J-CD. Other experimental conditions as in Figure 3.2.
3.4 Conclusions
The use of uncharged ~-cyclodextrin polymer as a chiral selector in CE allows
the enantioseparation of several classes of compounds (~-blockers, cx-adrenergic
agonists, ~-adrenergic agonists, tryptophan esters and anesthetics).
The complexation, resolution and selectivity are all influenced by the
concentration of the CD-polymer added to the BGE. Generally, the mobility of the
analytes decreases, with an increasing concentration of EP-~-CD.
The solubility of the chiral polymer in water is high in comparison with that of
the native ~-CD. This allows the separation of a wider number of compounds. The
method is cheap, since only a few microliters of the chiral polymer solution are
consumed per analysis (the chiral additive was only present in the capillary).
The addition of organic modifiers to the BGE has a deleterious effect on the
resolution for most of the analytes. However, for propranolol, metoprolol and
ephedrine, the resolution was enhanced upon adding MeOH. The increase of the
52
~-CYCLODEXTRIN POL YMER
capillary temperature had a negative effect on the resolution, except for ephedrine. The
resolution of this compound slightly increased at increased temperatures.
Simultaneously, the analysis time decreased, which makes the use of elevated
temperatures highly favorable for the separation of the optica) isomers of this
compound.
References
1. T. Schmitt and H. Engelhardt, Chromatographia, 37 ( 1993) 475
2. A. Nardi, A. Eliseev, P. Bocek and S. Fanali, J. Chromatogr., 638 (1993) 247
3. A. Guttman, A. Paulus, A.S. Cohen, N. Grindberg and B.L. Karger, J.
Chromatogr., 448 (1988) 41
4. S. Fanali, J. Chromatogr" 474 (1989) 441
5. Z. Aturki and S. Fanali, J. Chromatogr" 680 (1994) 137
6. N.Thuaud, B.Sebille, A.Deratani and G.Lelievre, J. Chromatogr.,
555 (1991) 53
7 H. Nishi and S. Terabe, J. Chromatogr. A, 694 (1995) 245
8 H. Nishi, J. Chromatogr. A, 735 (1996) 57
9 S. Fanali, J. Chromatogr. A, 735 (1996) 77
10. S.A.C. Wren and R.C. Rowe, J. Chromatogr" 635 (1992) 113
11. S. Fanali, J. Chromatogr., 545 (1991) 437
12. S.A.C. Wren and R.C. Rowe, J. Chromatogr" 603 (1992) 235
13. A. Nardi, L. Ossicini and S. Fanali, Chirality, 4 (1992) 56
14. A. Harada, M. Furue and S. Nozakura, Polymer J., 13 (1981) 777
15. F. Kohlrausch, Ann. Phys. (Leipzig), 62 (1897) 209
16 S.A.C. Wren and R.C. Rowe, J. Chromatogr., 609 (1992) 363
17. S.M. Han and N. Purdie, Anal. Chem., 56 (1984) 2825
53
CHAP'IER3
54
ERGOT ALKALOIDS
4. ERGOT ALKALOIDS AS CHIRAL SELECTORS IN
CAPILLARY ELECTROPHORESIS
Abstract
This chapter introduces ergot alkaloids as novel chiral selectors in CE. In this study,
stereoselectivities of several ergot alkaloids towards a number of racemic acidic
compounds were compared. The 1-allyl derivative of terguride ( allyl-TER) proved to
be the best chiral selector f or these analytes. The eff ects of pH, and the addition of
MeOH to the background electrolyte (BGE) were investigated. Low pH proved to
have an adverse effect on enantioseparation, indicating ionoselective complex
formation. These observations we re confirmed by determining the formation constants
of the dissociated and the non-dissociated acid. Good separation for the enantiomers
of some a-hydroxy acids was obtained at pH 4.2, at a 25 mM allyl-TER
concentration. The addition of 50% MeOH to the BGE altered stereoselectivity and
increased the solubility of the chiral selector. Using a BGE containing 50% MeOH,
and 62.5 mM allyl-TER at pH' 5.5, the optica/ isomers of all the test compounds
could be baseline resolved.
The study presented in this chapter has resulted in the following publications:
B.A. Ingelse, J.C. Reijenga, H.A. Claessens and F.M. Everaerts, J. High Res. Chromatogr., 19 (1996) 225-
226
B.A. Ingelse, M.Flieger, H.A. Claessens, F.M. Everaerts, J. ofChromatogr. A, 755 (1996) 251-259
4.1 Introduction
As recently reviewed, capillary electrophoresis has proven to be a suitable
method for chiral analysis [ l ]. Several different chiral se lectors have been applied in
CE, such as proteins, crown ethers, metal-chelator complexes, macrocyclic antibiotics,
chiral micelles, and native and modified cyclodextrins. In this study, a novel group of
chiral selectors for use in CE, namely ergot alkaloids, is introduced.
55
CHAP1ER4
Figure 4.1 Chemical structure of
Ergot alkaloids are a large group of
natura! compounds that are derivatives of
(5R)-lysergic acid (2). The chemica! structure
of lysergic acid is shown in Figure 4.1.
Previously, the optica! purity of some semi
synthetic ergot pharmaceutical preparations,
like (5R, BS, /OR) and (5S, BR, JOS)-lisuride
and (5R, BS, JOR) and (5S, BR, /OS)-terguride,
was determined by CE using cyclodextrins as
chiral selector [3]. Flieger et al. [4] prepared a
chiral stationary phase, using the 1-(3-
( jR)-lysergic acid. aminopropyl) derivative of (+)-(5R, BS, JOR)-
terguride (AMP-TER) to resolve optically active isomers of sorne semisynthetic ergot
alkaloids by HPLC. This alkaline stationary phase was also used to separate the optical
isomers of some carboxylic acids and dansyl derivatives of arnino acids [5,6]. A recent
study shows the successful enantiomeric separation of some halogen-substituted 2-
arylpropionic acids using this terguride-based chiral stationary phase [7]. The capillary
electrophoretic separation of the optica] isomers of these compounds is shown in
chapter 7.
Castellani et al. [5] studied the retention of some 2-arylpropionic acids as a
function of the pH of the mobile phase. They observed a maximum capacity at pH 4
for all pairs of enantiomers. This could he explained by considering the acid-base
equilibria of both the analytes and the chiral selector since the strongest electrostatic
interaction can be expected if the analytes are fully dissociated and if the basic nitrogen
at position (N(6)) of the ergot alkaloid is fully protonated. An NMR study on the
diastereoisomer complexes formed between AMP-TER and naproxen was performed
to study the mechanism of chiral recognition [8]. From the results of this study it could
be concluded that electrostatic and stabilizing 1Mt stacking interactions are the most
significant bonding interactions resulting in the formation of diastereomeric complexes.
A new procedure for the preparation of an ergot alkaloid based chiral
stationary phase is based on the bonding of a 1-allyl derivative of (+)-(5R, BS, JOR)
terg~ride (allyl-TER) to mercaptopropyl silanized silic.a gel [9]. This packing exhibited
a higher concentration of chiral selector, higher stability, reproducibility and
enantioselectivity towards some arnino acid. derivatives compared to the AMP-TER
statfonary phase. As expected, the influence of the pH is similar as described above.
In this chapter, it is shown that ergot alkaloid derivatives, specifically allyl
TER, have great potential for use as chiral selectors in CE. In order to study the
56
ERGOT ALKALOIDS
separation mechanism, the migration behavior of the acidic analytes was studied at
different pH values. Moreover, the influence of the addition of MeOH on the chiral
separation was investigated.
In order to obtain more insight in the separation of enantiomers, the separation
mechanism is discussed in several studies [ 10-15]. These studies focus on the
determination of equilibrium constants of complex formation between analytes and
cyclodextrins as chiral selector. For this purpose, the chiral selector is dissolved in the
background electrolyte (BGE) and the mobility of the optica! isomers is determined as
a function of the cyclodextrin concentration. Ina similar way, an attempt was made to
determine the formation constants between some racemic organic acids and allyl-TER.
In some cases, this method has drawbacks. Above all, if the chiral selector is
very high-priced, the determination of the formation constant can become very
expensive, since not only the capillary but (in most cases) also the in- or outlet vial (or
both) have to be filled with BGE containing chiral selector. Only in case coated
capillaries (or low pH BGE's) and neutra! chiral selector's are used, it is sufficient to
fill only the capillary with the chiral selector. Another obstacle occurs if the chiral
selector is a strong UV-absorbing compound. In case detection of the analytes is not
possible on top of a high background absorption of the chiral selector, it is necessary to
fill only part of the capillary with BGE containing the chiral selector [ 16, 17]. In all
these cases, the determination of formation constants can be troublesome, or even
impossible.
Allyl-TER is a strong UV-absorbing cation, and therefore (as outlîned earlier)
it was hard to determine the formation constants, without making some assumptions.
Initially, it was assumed that only the dissociated acid interacted with the ergot
alkaloid.
Recently, Lee and Lin [18] determined formation constants between
cyclodextrins and some non-chiral compounds (salicylic acid and benzylamine) by
injecting the cyclodextrins in a BGE containing different concentrations of these
compounds. A similar approach was used in the second part of the current study. The
formation constants were determined by measuring the mobility of the ergot alkaloid,
injected as analyte, in BGE's consisting of varying concentrations of either L-( +) or D
(-)-mandelic acid at different pH-values. In this way, from here on indicated as
reversed determination, it was possible to determine both selectivity and formation
constants for complexes between protonated allyl-TER and both the dissociated and
the non-dissociated chiral acids [12, 13].
57
CHAP1ER4
4.2 Experlmental
4.2.1 Equipment
All analyses were performed on a P/ ACE 2200 capillary electrophoresis system
(Beckman, Fullerton, CA). The Beckman instrument used polyacrylamide coated
capillaries [19], with a total length of 37 cm and an effective length of 30 cm. The
intemal diameter was 50 µm. The capillary temperature was kept constant at 22°C.
Before every electrophoretic run, the capillary was first flushed with BGE and then
flushed with BGE containing allyl-TER with concentrations varying between 0-62.5
mM. In this way, allyl-TER was only present in the capillary and not in the in- or outlet
vial, as described earlier [ 16, 17]. Samples were introduced by pressure injection (10
seconds at 3.3. 103 Pa). The detection wavelength was 200 nm in BGE's containing no
MeOH, and 214 nm in BGE's containing 50% MeOH. The applied voltage was 20 kV
or 25 kV.
4.2.2 Chemicals and sample preparation
The 1-allyl derivatives of ergot alkaloids (lisuride [18016-80-3], terguride
[37686-84-3], luol [35121-60-9], lysergol [602-85-7], festuclavine [569-26-6],
ergotarnine [113-15-5), and dihydroergotamine [511-12-6]) were synthesized by the
method published earlier [9]. Dihydroergocristine [17479-19-5] and optically pure(+)
and (-)-terguride were a gift of Mirko Flieger of the Academy of Sciences of the Czech
Republic (Prague, Czech Republic). The chemica] structure of the alkaloids applied as
chiral selector is shown in Figure 4.2. Racemates of mandelic acid, p-hydroxymandelic
acid, 3,4,-dihydroxymandelic acid, vanilmandelic acid, and tropie acid were purchased
form Sigma (St. Louis, MO). Racemates of N-acetylphenylalanine, N-formyl
phenylalanine, 2-phenylglycine, and a-methoxyphenylacetic acid were a kind gift of
DSM Research (Geleen, The Netherlands). L-(+) And D-(-) mandelic acid were
obtained from Fluka (Buchs, Switzerland). The above samples were dissolved in
demineralized water. The concentration of the analytes was 104 M. Samples of ergot
alkaloids were dissolved in glacial acetic acid and diluted with 50 parts of water to a
final concentration of 104 M.
58
o H3Lo,?Hn
~,"""J-T~ :'~H30 H ÁO 0
UN 1-allyl-dihydroergotamme (
~
ERGOT ALKALOIDS
" :.: 3
N j 1-allyl-lysergol
0 H3CECYF?H~H 0 ;
~,"""". N0N N H 0 H o·".",, 0 'CH H 3
dihydroergocristine
0 H:r-1 Yf?H 0 ;
~"".""" ~0N
N H 0 H c··"", 0 'CH H 3 1: 1-allyl-ergotamine
1:
Figure 4.2 Chemica/ structures of the applied ergot alkaloids.
4.2.3 Methods and electrophoretic systems
The pK.-values of the ergot alkaloids were determined by measuring mobilities
from pH 3 up to pH 9. For this purpose, background electrolytes (BGE's) were
prepared with constant ionic strength i (! 0 mM). The composition of these BGE's is
listed in T ABLE 4.1. The pK.-values of some of the acidic analytes were determined,
by measuring the (apparent) pH of a solution containing 10 mM NaOH and 20 mM of
the acid.
The BGE for the non-chiral analysis of the racemic organic acids was prepared
by adjusting a 200 mM ~-alanine solution with acetic acid up to pH 4.2. The ionic
strength of the BGE was 50 mM. For the preparation of the chiral BGE, ergot
alkaloids were dissolved in glacial acetic acid and diluted with demineralized water.
59
CHAPTER4
Subsequently, the pH was adjusted with a ~-alanine solution up to pH 4.2. The ionic
strength of the BGE containing chiral selector was also 50 mM. A similar approach
was used to prepare BGE's at pH 3.2 using formate instead of acetate as background
anion.
Ergot alkaloids were also soluble in MeOH. Solutions of allyl-TER in MeOH
were diluted with NaAc-solutions up to pH* 5.3 and i = 50 mM, or with a 200 mM ~
alanine/acetate-solution up to pH* 5.5, both with a final MeOH concentration of 50%
(vlv).
Experiments for the reversed determination of the formation constants were
performed at pH 4.9 and pH 2.2. BGE's at pH 4.9 consisted of 200 mM creatinine and
100 mM L-(+) or D-(-)-mandelic acid or acetic acid. BGE's at pH 2.2 consisted of
5.00 mM aniline and 100 mM L-(+) or D-(-)-mandelic acid or formic acid. The sample,
consisting of 5.104 M allyl-TER and i.10·3 M tetrabutylammonium bromide, was
injected hydrodynamically for 2 seconds (3.103 Pa).
TABLE 4.1 BGE's USED FOR THE pKa-DETERMINATION OF ERGOT
ALKALOIDS
pH
2.0
3.0
3.8 5.0
6.0
7.0
8.0
9.0
CATION
lOmMH+
lOmMNa+
lOmMNa+
JOmMNa+
lOmMNa+
lOmMNa+
TRIS+
lOmMNa+
MES 2-(N-morpholino)ethanesulfonate, MOPS
Tris(hydroxymethyl)aminomethane
4.3 Results and discussion
4.3.1 Characteriza,tion of the ergot alkaloids
ANION
phosphate
forma te
forma te
acetate
MES
MOPS
10 mM chloride
borate
morpholinopropanesulfonate, TRIS
Figure 4.2 shows the chemica! structures of the ergot alkaloids, applied in this
study. These natura! compounds possess two or three asymmetrie carbons, a 7t
interaction site represented by the indole ring, and a basic nitrogen on the N(6)
60
ERGOT ALKALOIDS
position as electrostatîc interaction site. This electrostatic interaction between the
acidic function of a sample molecule, as well as stabilizing 1Mt stacking interactions are
the most significant bonding interactions concurring to the formation of diastereomeric
complexes [8). Furthermore, these compounds have strong UV-absorbance between
200 and 300 nm. (Àmax forterguride: 292, 281, 224 nm, loge 3.72, 3.81, 4.42 [20]).
In order to estimate the charge of the chiral selectors, the pKa values were
determined. For this purpose, effective mobilities were measured as a function of the
pH of the BGE. The ionic strength of these BGE's with varying pH was kept constant
at 10 mM (T ABLE 4.1 ). The degree of dissociation ( a) was calculated from the ratio
of the measured mobility and the mobility of the ergot alkaloid at pH 2. At this pH, the
basic alkaloids were assumed to be fully protonated. These mobilities at pH 2 are
tabulated in T ABLE 4.2.
TABLE 4.2 MOBILITIES AND pKa-VALUES OF THE APPLIED ERGOT
ALKALOIDS AT 22°C.
Ergot alkaloid
1-allyl terguride
1-allyl festuclavine
1-allyl luol
1-allyl lysergol
dihydroergocristine
1-ally l Iisuride
1-allyl dihydroergotamine
1-allyl ergotamine
mobility*
18.J
22.0
19.9
21.0
14.2
17.9
13.9
14.2
pKa
7.1
8.4
8.9
7.6
6.7
7.0
6.4
6.1
*mobility x[l0'9 m2N.s] at i = 0 mM Measured at i 10 mM and corrected by multiplying with 0.93
according to eq. (7) in Ref. [21]
The pKa-values could be determined graphically, according to the equation of
Henderson-Hasselbalch:
pK. = pH - log[_!!_] 1-a
( 4-1)
This is shown in Figure 4.3 for allyl-TER and 1-allyl-ergotamine. The pK.-value is
determined by the intersection of the linear curve with the pH-axis. As listed in
TABLE 4.2, the pKa-values varied between 6.1 and 8.9. Therefore, the assumption of
61
CHAPTER4
full protonation at pH 2 was justified. Standard deviations (s.d.) of the pKa values were
within 0.1 units, whereas the s.d. for the mobilities was within 3%.
Figure 4.3 Detennination of pK,,-values of 1-allylterguride (.li..) and 1-allylergotamine
(•).
The mobilities of the ergot alkaJoids were plotted versus their molecular
weights on a double logarithmic scale (Figure 4.4). The mass versus charge relation of
these alkaJoids indicates monovalent cations. According to Figure 4.4, their mobility is
proportional to (molecular mass)°5. The components behave like a homologous series.
Ergot alkaloids are sparingly soluble in water. In fact, at pH > 4.5, it was not
possible to dissolve allyl-TER in 100% water. For this reason, experiments were
perforrned using methanol as organic buffer modifier. Allyl-TER is freely soluble in
MeOH.
62
ERGOT ALKALOIDS
10'--~~~~-'-~~~--'~~~'--~-'-~---'-~--'-~-'---'
200 500
molecular weight [g/mol]
1000
Figure 4.4 Double logarithmic plot of ergot mobilities (listed in TABLE 4.2) versus
molecular weight.
4.3.2 Comparison of stereoselectivity of different ergot alkaloids towards some
racemic organic acids.
To determine the quality of a chiral separation, two different definitions were
used. The enantioseparation factor (SF) is defined as the ratio of the equilibrium
constants of complex formation of the two optica! isomers. Selectivity (S) is defined as
the ratio of the mobility difference and the mean effective mobility of the two optica!
isomers. Mandelic acid, mandelic acid derivatives (p-hydroxymandelic acid, 3,4-
dihydroxymandelic acid, vanilmandelic acid), and tropie acid were chosen as test
compounds.
63
CHAPTER4
0.03
O.QI
j-0.0l j -0.03
<
-0.05
2 3
AT
5
DH
-0.07 +----f----.........,1------+----+----+------l
4 5 6 7
time [mln]
8 9 10
Figure 4.5 Comparison of the stereoselectivity of different ergot alkaloids towards some
racemic organic acids. FC = 1-allylfestuclavine, AT = 1-allylterguride, DH =
dihydroergocristine. 1 = mandelic acid, 2 = p-hydroxymandelic acid, 3 = 3,4-
dihydroxymandelic acid, 4 = vanilmandelic acid, 5 = tropie acid. BGE: /3-alanine-
. acetate + 25 mM ergot alkaloid, pH 4.2, i = 50 mM. Capillary: 300-370 mm, 50 µm
/.D" Separation voltage 20 kV.
These experiments were performed at pH 4.2. At their highest possible concentration
in water, it was found that the 1-allyl derivatives ofluol, lisuride, and lysergol showed
no selectivity towards the optical isorners of any of the selected compounds. Mandelic
acid and mandelic acid derivatives were partly resolved usiilg 90 mM 1-allyl
festuclavine as chiral selector (S < 0.008). Slightly higher selectivities were obtained at
40 mM dihydroergocristine. This ergot alkaloid showed also selectivity towards the
tropie acid enantiorners (S = 0.005), but no selectivity towards the optica! isomers of
mandelic acid. The best results were obtained with aJJyl-TER as shown in Figure 4.5.
Already at 25 mM, this chiral selector showed the highest selectivities towards all the
selected racernic compounds (S = 0.01 for mandelic acid up to 0.02 for vanilmandelic
acid enantiomers), except tropie acid (S = 0.002). For this reason, the study was
focused on allyl-TER.
4.3.3 Determination of the mobUity of the analyte interacting with allyl-TER.
The racernic analytes were injected at the cathode side of the capillary (phase 1 in
Figure 4.6). All analytes were moving as anions in the direction of the anode. To
64
ERGOT ALKALOIDS
suppress EOF, coated capillaries were used. The measured EOF was negligible
(µEOJ< L 10-9 m2 N .s). Interaction with the chiral se lector occurred only in the first part
of the capillary, due to the experimental setup (phase I and II in Figure 4.6).
I
___ JGE
II +
~--....... ~~~~~_,GE
III +
Figure 4.6 Separation scheme. Enantiomers in black, BGE in white, BGE supported
with ergot alkaloid in dark gray. Further explanation see text.
Due to this interaction, the mobility of the analytes in this part of the capillary (µ;) will
be lower than the mobility of the free analyte (µ1).
Since all the experiments were performed at pH 3.2 and pH 4.2 (in BGE's
consisting of 100 % water), the chiral selectors were fully charged. The mobility of the
ergot alkaloids is higher than the mobility of the other counter ion, ~-alanine.
Therefore a self-sharpening rear boundary between BGE containing ergot alkaloid
(gray in Figure 4.6) and pure BGE (white in Figure 4.6) will be formed. The rear
boundary will migrate in the direction of the inlet via!, opposite to the migration of the
acidic analytes. Because of the strong UV-absorbance of the ergot alkaloids, this rear
boundary should pass the detection window before the analytes. Detection of the
analytes would not have been possible on top of this high background absorption.
After the analytes have pa<;sed the rear boundary (phase III, Figure 4.6), they wil!
migrate in a BGE with an adjusted concentration of ~-alanine and acetate, following
the Kohlrausch regulation function.
65
CHAPTER4
A dynamic simulation [22] of the separation is depicted in Figure 4.7. The simulation
conditions were: -35 µA (for ~-alanine) or -47 µA (for sodium) in a 75 µm I.D.
capillary with suppressed electroosmosis. The software could only simulate the
'constant current' mode. The above mentiöned values for the current, correspond with
an average voltage of 25 kV across a capillary of 370 mm. The situation shown in
Figure 4.7 is approximately 60 s after current switch-on. Coordinate 0 in Figure 4.7
refers to the capillary outlet. The self-correcting sharp boundary at -70 mm, between
BGE's with and without allyl-TER, migrates to the left in the direction of the inlet.
The sharp boundary is only obtained if the mobility of the counter-ion is lower than the
mobility of allyl-TER. The upper figure clearly shows the difference between the use of
~-alanine and the fast cation sodium as a counter-ion. In the Jatter case, no sharp
boundary is obtained, since sodium has a higher mobility than allyl-TER.
According to the Kohlrausch regulation function, the pH will decrease from 4.2
to 4.07, thus slightly decreasing the degree of dissociation and therefore the effective
mobility of the analytes. The electric field strength (E) in this zone however, was
calculated to be approximately 6 % higher. Consequently, the effects of the pH and of
E on the effective mobility of the analytes are counterproductive. Therefore, for the
determination of µ; both pH and E are assumed to be homogeneous throughout the
capillary. The dynarnic simulation [22] of the pH and the electric field strength (E),
versus the distance from the capillary outlet is shown in the lower part of Figure 4. 7.
Initially, after about 1 minute, the rear boundary passes the detection window
(see Figure 4.8). This time (to) could be accurately determined due to the sharp drop in
absorbance. lt is interesting to notice that to in Figure 4.8 corresponds well with the
simulated situation in Figure 4. 7.
From to, the velocity of the rear boundary (VRs) is calculated, according to the
following equation:
( 4-2)
In this equation, ld is the effective capillary length, and 11 is the total capillary length.
(In all these equations, the velocities have positive values.)
At t = t;, the analyte will pass the rear boundary, which is migrating in the
opposite direction. This time is dependent on the analyte velocity (v;) and will therefore
be different for each sample.
66
ERGOT ALKALOIDS
25 BGE: 8-alanine/Ac·
î 20 ! llC BGE: Na./Ac· Ol
i 15
~ 10
0 +-~~-+~~~-+-~~--+~~-->--+-~-"'--Il--~~-+-~~--+
-0.14 -0.12 -0.1 -0.08 -0.06 -0.04 -0.02 0
capillary coordinate [m]
70 E I \
60 4.8
50
4.6
4.4
pH 20
4,2
10
-0.14 -0.12 -0.1 -0.08 -0.06 -0.04 -0.02 0
caplllary eoordinate [m]
Figure 4.7 Dynamic simulation of the rear boundary after 60 seconds. Upper:
concentration allyl-TER versus distance from capillary outlet (x=O) in BGE's
containing /)-alanine or sodium as counter-ion. Lower: local field strength and pH
versus distance from the capillary outlet, in BGE containing {3-alanine.
( 4-3)
During this time the analyte will cover a distance equaling µ;.E.t;. The remaining part
of the capillary up to the detection window should equal µ1-E.(tm-t;) and thus follows:
( 4-4)
67
CHAP1ER4
0.025
5' 0.015 ~
1 ." < -0.005
4
2 6
rear boundary (to)
-0.015 +------t--+---+------------+-------1 0 2 3
time [mini
4 5 6
Figure 4.8 Chiral separation of some racemic organic acids. 6 = a-methylphenyl
acetic acid. Separation voltage 25 kV. Other symbols and experimental condinons as
in Figure 4.5.
Rearrangement of eq. (4-2) - (4-4) gives:
( 4-5)
The interaction time of the analytes with the chiral selector measured 60-75% of the
analysis time, depending on µ;. As an example, the time of interaction of the mandelic
acid enantiomers in Figure 4.8 was 155 seconds (72 % of the analysis time), and 180
seconds (64%) for vanilmandelic acid enantiomers.
4.3.4 Influence of the pH on stereoselectivity
According to Rawjee et al. [23] the pH can have a major influence on chiral
selectivity in CE. In order to study the influence of the pH on the separation of the
optica! isomers of the selected organic acids, complex formation and enantioselectivity
at pH 3.2 and pH 4.2 were studied. Since the pKa-values of the analytes applied in
these experiments were about 3.5 for the mandelic acid derivatives and o:methoxyphenylacetic acid, the pH was expected to have a major intluence on the
degree of dissociation, and consequently on the stereoselectivity.
68
ERGOT ALKALOIDS
As illustrated in Figure 4.9, the analytes have a Jower migration velocity at pH
3.2 than at pH 4.2. This is due to the difference in the degree of dissociation of the
weak acids. The selectivity of allyl-TER towards the analytes was very Iimited at pH
3.2. At pH 4.2 good enantioselectivities were obtained towards the optica] isomers of
the acidic compounds. Besides the compounds shown in Figure 4.5 and Figure 4.9,
baseline resolution was also obtained for the enantiomers of 2-phenylglycine.
The observed pH-dependency is in good agreement with HPLC experiments
using terguride as a stationary phase [9]. From these results it was concluded that only
the dissociated acids interact stereoselective with the ergot alkaloid.
4 0.02
6 2 pH4.2
~ 0.01 4 ~ 2 " 5 IJ = "' .&>. 0 " Q 6 "' .&>. <
-0.01 pH3.2
-0.02
3 4 5 6 7
time [ minutes]
Figure 4.9 lnfluence of pH on chiral separation of racemic organic acids. BGE of
bottom electropherogram: 25 mM allyl-TER + /3-alanine-formate, pH 3.2, i = 50
mM. BGE of top electropherogram: 25 mM allyl-TER + {3-alanine-acetate, pH 4.2, i
= 50 mM. Symbols and other experimental conditions as in Figure 4.8.
4.3.5 Direct determination of formation constants
To study the interaction between analyte and chiral selector, the BGE in the
capillary was supported with 0, 5, JO ,15, 20 and 25 mM allyl-TER. The theoretica!
model of Wren & Rowe [JO, 11] was used to determine the equilibrium constants (Kc).
According to this model, the electrophoretic mobility of the analytes (µ;) will be a
function of the mobility of the free enantiomer (µf), the mobility of the analyte/chiral
selector complex (Jlc), the concentration of the chiral selector ( C), and the equilibrium
constant (see Chapter 2):
69
CHAPTER4
( 4-6)
As a first approximation, only interactions between the dissociated acid and the chiral
selector were tak.en into account. Since the dissociated acid bas a single negative
charge, and the chiral selector has a single positive charge, the complex will be
uncharged. Consequently, the complex rnobility (µ,;) will be negligible, and eq. (4-6)
can be simplified:
µ;=l+KC c
( 4-7)
This rearranges to:
( 4-8)
Consequently, a plot of (µ/µ 1) versus C, should result in a linear relation, with a slope
equaling Kc. The Kc-value thus determined should be interpreted as a pH dependent
formation constant. Figure 4.10 shows the determination of the Kc-values of the
vanilmandelic acid enantiomers, using allyl-TER as chiral selector. The two solid lines
show the best linear fit for these two optica! isomers. The upper solid line represents
the isomer which has the strongest interaction with the chiral selector (steepest slope).
T ABLE 4.3 lists Kc-values of all the compounds studied, in BGE's without, and
with 50% MeOH as organic buffer modifier. Also the SF is tabulated. Vanilrnandelic
acid and tropie acid have the highest affinity towards the chiral selector.
Enantioselectivity towards the tropie acid enantiomers however, is very limited, under
these conditions (0% MeOH, pH 4.2). The pK.-value of tropie acid is relatively high,
resulting in a relatively low degree of dissociation, at pH 4.2 (TABLE 4.5). This could
be a possible explanatîon of the limited enantioselectivity. A structural difference
between tropie acid and the other analytes is that tropie acid is ~-substituted, while the
other analytes are a-substituted, counting both from the carboxylic acid functional
group, and from the aromatic group. This could also be a possible explanation for the
low selectivity obtained for this analyte.
70
ERGOT ALKALOIDS
1.30
~ 1.20
1.10
5 10 15 20 25
concentration l·allyl terguride [m.VI]
Figure 4.10 Detennination of Kc-values of the vanilmandelic acid enantiomers. Trend
lines represent best linear fits /or the stronger (e) and weaker (0) interacting isomer.
Experimental conditions as in Figure 4.8.
TABLE 4.3 Kc-VALUES AND ENANTIOSEPARATION FACTORS (SF) OF
ALL YL-TER TOW ARDS SOME RACEMIC ORGANIC ANAL YTES. COMPONENT
NUMBERS AS IN FIGURE 4.5 AND FIGURE 4.8.
100% water, pH 4.2 50%MeOH, 5.3
compound * K1sr K2nd SF K1s1 Kznt1 SF c c c c
1 5.5 6.3 1.15 11 13.2 1.18
2 7.8 8.9 1.14 12. l 13.2 1.09
4 12.0 14.9 1.24 13.5 14.9 1.10
5 14.1 14.3 1.01 6.7 7.2 1.07
6 7.3 8.7 1.19 10.6 13.2 1.25
to the first and second migrating enantiomer
71
CHAP1ER4
The migration order was determined for the optica! isomers of mandelic acid by
spiking the racemic sample with L-(+)-mandelic acid. L-(+)-mandelate appeared to
have the strongest interaction with allyl-TER.
4.3.6 Reversed determination of the formation constants of mande/ic acid
The mobility of allyl-TER was determined as a function of the concentration of
mandelate present in the BGE. Different concentrations of mandelate in the BGE were
obtained by mixing BGE's containing mandelate with the BGE's containing acetate (pH
4.9) or formate (pH 2.2). In this way, unlike the rnethod presented in Ref. [18], the
ionic strength of the BGE's was independent on the mandelate concentration. This was
very important in these experiments since very small differences in mobility had to be
determined as accurately as possible, and it is well known that the ionic strength
considerably influences the mobility. Since, especially at low pH-values, the magnitude
of the EOF can be difficult to determine accurately, tetrabutylammonium (TBA) was
chosen as a mobility reference. This relatively slow cation is migrating closely to allyl
TER, and has no interaction with rnandelate. The mobility of TBA was 14.2 10·9
m2N.s in the BGE containing creatinine (pH 4.9), and 14.5 10·9 m2N.s in the BGE
containing aniline (pH 2.2). This difference in mobility is due to the higher ionic
strength of the creatinine BGE. The mobility of allyl-TER is slightly lower than the
mobility of TBA in these experiments and ranges in between 10 and 13.10·9 m2N.s,
depending on the composition of the BGE. The mobility of allyl-TER in absence of
mandelate (µo) is 13.0 10·9 m2N.s in the BGE containing creatinine and 12.0 10"9
m2N.s in the BGE containing aniline.
At pH 4.9, mandelic acid has a degree of dissociation a > 0.95, since its pK,.
values 3.4. For the sake of simplicity, a =1 was assurned. At pH 2.2, the degree of
dissociation is about 0.06, and as a rough estimation a = 0 was assumed. The simple
model ofWren and Rowe [10] was used to determine the formation constants between
allyl-TER and L-(+) and D-(-)-mandelate. The results are presented in TABLE 4.4.
According to these results, Ki (referring to interaction with the non-dissociated acid) is
higher than Ki (referring to interaction with the dissociated acid). Consequently, the
assumption that interaction between the non-dissociated acid and allyl-TER could be
neglected, was erroneous. However, it is true that only the dissociated acid interacts
stereoselective with the ergot alkaloid. The mobility of the complex between the
dissociated acid and the chiral selector does not significantly differ from zero, justifying
the earlier a.<>sumption. From the results obtained by the indirect determination, the Kc
value at pH 4.2 was recalculated applying eq. (4-6). As shown in TABLE 4.4, Kc was
overestimated in the earlier experiments.
72
ERGOT ALKALOIDS
TABLE 4.4 COMPARISON OF CHIRAL PARAMETERS OF MANDELIC ACID
OBT AINED BY THE REVERSED AND THE DIRECT DETERMINA TION.
Average Difference ((+)vs.(-))
reversed direct reversed direct
4.6 (0.5)' 0% 0%
2.05 (0.17) 21%
Kc(a= 0.86) 2.4 (0.2) 6.0 (0.3) 13% 17%
10 (0.3) 0%
0.3 (1.9) 0 0% 0%
Standard deviation on parenthesis; Kc (L-(+)-mandelate) > Kc(D-(-)-mandelate)
These result stress the importance of applying high pH BGE's. At high pH, the non
selective interactions are minimized, and similarly the time of analysis is decreased.
In an earlier study, Fanali et al. successfully separated ergot alkaloids using y
cyclodextrins as chiral selector [3]. Present study proves that these alkaloids might also
be separated using relatively inexpensive optically pure mandelic acid as counter-ion.
Figure 4.11 shows the optimized separation of a mixture of (-)-TER : (+)-TER = 1
1.5.
0.004
0.003 (-)-TER (+)-TER
S' 0.002 ~ " 0.001 "' = "' i: 0 0
~ -0.001
-0.002
-0.003
0 2 4 6 8 10 12 14 16
time [min]
Figure 4.11 Chiral separation of a mixture of terguride (TER) enantiomers; (-)-TER:
(+)-TER= 1 : 1.5. BGE: 200 mM e-aminocapronic acid, 100 mM L-(+)-mandelic
acid, pH 4.5. Capillary 400-470 mm, l.D. 50 µm, polyacrylamide coated. Detection:
280 nm. Separation voltage: 20 kV. Temperature: 17°C.
73
CHAPTER4
This separation was achieved by applying a BGE containing 100 mM L-( + )-mandelic
acid at pH 4.5. At this pH, the degree of dissociation of mandelic acid a > 0.9. (+)
TER Is the slowest migrating alkaloid, and therefore it can be concluded that (+)-TER
has the strongest interaction with L-( + )-mandelate, confirming the earlier results.
Reversal of the migration order was easily accomplished by using D-(-)-mandelate in
stead of L-( + )-mandelate as counter-ion.
4.3.7 The influence of MeOH on chiral separation
Also, the effect of the addition of MeOH to the BGE on the interaction
between the chiral selector and the racemic analytes was investigated. The chiral
selector is well soluble in MeOH, which makes it very advantageous to use this organic
buffer modifier. Upon addition of 50% MeOH to the BOE at pH 4.2 (200 mM ~
alanine/acetate) the apparent pH shifted up to pH* 5.3. Since the pK.-value of ~
alanine was not known under these conditions, the strong ion sodium was used as a
cation, in order to make a proper calculatîon of the ionic strength of the BGE. Sodium
however, migrates faster than allyl-TER. For this reason, the rear boundary between
BOE with and without chiral selector, will not have self sharpening properties (see
Figure 4.7 and upper electropherogram in Figure 4.12).
The addition of MeOH to the BOE caused a shift in the apparent pH of the
buffer. At the sarne time the pK.-value of the analytes was shifted in the same
direction. The change in the degree of dissociation ( a) was calculated for mandelic
acid and for tropie acid by the determination of the pK.-values in BGE's with and
without MeOH. As shown in T ABLE 4.5, no significant change of the a-value of these
two analytes was observed between BOE's with and without MeOH.
TABLE 4.5 INFLUENCE OF MeOH ON THE pK.-V ALUES AND DEGREE OF
DISSOCIATION (a) FOR MANDELIC AND TROPIC ACID.
Mandelic acid Tropie acid
BGE pK. a pK. a 0% MeOH, pH 4.2 3.3 0.89 4.0 0.60
50% MeOH, pH* 5.3 4.2 0.90 5.1 0.62
Figure 4.12 shows the influence of MeOH on the chiral separation of the
racemic analytes. First, as expected, a Jonger migration time was observed in BGE's
containing MeOH. More interesting, an increase in resolution is observed for some of
74
ERGOT ALKALOIDS
the racemic analytes. The increased resolution is very pronounced for the tropie acid
enantiomers. As mentioned earlier, this effect could not be attributed to a change in a
value of this analyte. Therefore, it can be concluded that MeOH can have a positive
influence on the enantioselectivity of allyl-TER towards some racemic organic acids.
This is also shown in T ABLE 4.3. The enantioseparation factor (SF) increased for
mandelic acid, a-methylphenylacetic acid, and for tropie acid. On the other hand, a
decrease of SF is observed for p-hydroxymandelic acid and for vanilmandelic acid. At
the same time however, the addition of MeOH to the BOE caused an increase in the
Kc-value for all analytes, except for tropie acid. According to the model of Wren &
Rowe [10,11] an increased Kc-value results in a lower optimum concentration of the
chiral selector. This is of course beneficia] for the separation of the racemic samples,
since solubility of the chiral selector is a limiting factor under these circumstances.
Another main advantage of MeOH is the increased solubility of the ergot
alkaloid. Figure 4.13 shows a high resolution separation of some racemic organic acids
using a BOE supported with 62.5 mM allyl-TER at pH* 5.5. This was the highest pH
at which the chiral selector was still soluble, under these conditions. Using pure water
as solvent, it was not possible to dissolve such a high concentration of this ergot
alkaloid.
75
CHAP1ER4
0
-0.01
~ -0.02
e i -0.03
~ .! < -0.04
-0.05
6
2
50% MeOH
0% MeOH
5
2
6
5
-0.06 +----1------+---------+-----1------1
3 4 5 6
time [min]
7 8 9
Figure 4.12 lnfluence of MeOH on the chiral separation of some racemic organic acids.
BGE bottom electropherogram: 25 mM allyl-TER + {3-alanine!acetate, pH 4.2, i = 50
mM. BGE top electropherogram: 25 mM 1-allyl terguride + sodiumlacetate, pH* 5.3, i
= 50 mM. Separation voltage 25 kV. Symbols and other experimental conditions as in
Figure 4.5 and Figure 4.8.
0.007 5
5' 0.005 ~ 0.003 B ä 0.001
,/;; s -0.001 .! < -0.003
-0.005 +--+----lir-----+--+---+---t--+--t-----l
0 2 3 4 5 6 7 8 9
time [min]
Figure 4.13 Chiral separation of racemic Njormylphenylalanine (7), N
acetylphenylalanine (8), tropie acid (5) and 2-phenylglycine (9). BGE 100 mM {3-
alanine/acetate, 62.5 mM 1-allyl terguride, 50% MeOH, pH* 5.5. PVA coated capillary:
20-27 cm, 50 µm l.D .. Separation voltage 25 kV.
76
ERGOT ALKALOIDS
4.4 Conclusions
lt is shown that ergot alkaloids have great potential as chiral selectors in CE.
Baseline resolutions are obtained for all the racemic acidic test compounds used in this
study. The analysis times are short and the method is cheap due to the low
consumption of the chiral selector. Results have shown that a high degree of
dissociation of the analytes is beneficia] for enantioselectivity. This was confirmed by
determination of the formation constants in a reversed setup, applying the chiral
selector as analyte, and the analyte as buffer component. In this way, the determination
of even very low formation constants was made possible.
Solubility of allyl-TER in 100% water is limited, but can be increased by the
addition of 50% MeOH to the BGE. The addition of MeOH also alters the
stereoselectivity and the degree of complex formation.
References
1 S. Fanali, J. Chromatogr. A, 735 (1996) 77
2 B. Berde and H.O. Schild, Handbook ofExperimental Pharmacology, Vol. 49,
Berlin, Springer-Verlag (1978)
3 S. Fanali, M. Flieger, N. Steinerova and A. Nardi, Electrophoresis, 13 (1992)
39
4 M. Flieger, M. Sinibaldi, L. Cvak, and L. Castellani, Chirality, 6 (1994) 549
5 L. Castellani, M. Flieger, and M. Sinibaldi, J. Liq. Chromatogr., 17 (1994)
3695
6 M. Sinibaldi, M. Flieger, L. Cvak, A. Messina, and A. Pichini,
J. Chromatogr. A, 666 (1994) 471
7 P. Padiglioni, C.M. Polcaro, S. Marchese, M. Sinibaldi and M. Flieger,
J. Chromatogr. A, 756 (1996) 119
8 L. Castellani, M. Flieger, L. Mannini, P. Sedmera, A. L. Segre, and
M. Sinibaldi, Chirality, 6 (1994) 543
9 A. Messina, A.M. Girelli, M. Flieger, M. Sinibaldi, P. Sedmera and L. Cvak,
Anal. Che m., 68 ( 1996) 1191
10 S.A.C. Wren and R.C. Rowe, J. Chromatogr., 603 (1992) 235
11 S.A.C. Wren, Electrophoresis, 16 (1995) 2127
12 Y.Y. Rawjee, D.U. Staerk and G. Vigh, J. Chromatogr" 635 (1993) 291
13 Y.Y. Rawjee, R.L. Williams and G. Vigh, J. Chromatogr., 652 (1993) 233
14 Y.Y. Rawjee and G. Vigh, Anal. Chem., 66 (1994) 619
77
CHAP1ER4
15 K.L. Rundlett and D.W. Armstrong, J. Chromatogr. A, 721(1996)173
16 L. Valtcheva, J. Mohammad. G. Petterson and S. Hjerten, J. Chromatogr" 638
(1993) 263
17 Y. Tanaka and S. Terabe, J. Chromatogr. A, 694 (1995) 277
18 Y. Lee, T. Lin, Electrophoresis, 17 (1996) 333
19 M.J. van der Schans, M.C. Molling, J.L. Beckers and F.M. Everaerts,
J. Chromatogr. A, 717 (1995) 139
20 S. Budavari, M.J. O'Neil, A. Smith, P.E. Heckelman (&is), The Merck Index,
l lth ed" 1989
21 W. Friedl, J.C. Reijenga, and E. Kenndler, J. Chromatogr. A, 709 (1995) 163
22 J.H.P.A. Martens, J.C. Reijenga, J.H.M. ten Thije Boonkkamp,
R.M.M. Mattheij and F.M. Everaerts, J. Chromatogr. A, in press
23 Y.Y. Rawjee, D.U. Staerk and G. Vigh, J. Chromatogr., 635 (1993) 291
78
THE INFLUENCE OF THE NATURE OF THE BUFFER
5. THE INFLUENCE OF THE NATURE OF THE
BUFFER ON CHIRAL SEPARATION IN CE
Abstract
The capillary electrophoretic separation of the enantiomers of a number of anionic
sulf onamides was studied. Enantioselectivity of a range of native and modified
cyclodextrins was examined. Also, from the experimental results, equilibrium
constants and complex mobilities in different electrolyte systems were detennined. In
the present study it was found that the nature of the co-migrating buffer anion may
significantly influence the magnitude of equilibrium constants, depending on the type
of modification of a specific cyclodextrin. Consequently, this may also strongly
influence the optimum cyclodextrin concentration f or a particular separation. The
results of the separation of sulfonamide enantiomers with cyclodextrins do not agree
with the theoretica! model suggested by Wren and Rowe, concerning the existence of a
maximum in the mobility difference between two optica[ isomers.
The study presented in this chapter has resulted in the following publication:
B.A. Ingelse, H.A. Claessens, Sj. van der Wal, A.L.L. Duchateau, F.M. Everaerts, J. Chromatogr. A,
745 (1996) 61-71
5.1 Introduction
Until now, cyclodextrins (CD's) have been the most widely used chiral
se lectors in CE. CD' s have the ability to form enantioselective inclusion-complexes
with many kinds of solutes involving the secondary hydroxyl groups on the Jarger rim.
lnclusion complexation is controlled by charge, size and shape of both the solute and
the cyclodextrin [ 1,2]. Besides that, the composition of the buffer may also play a role
in enantioselectivity. Optimization of selectivity must therefore be accomplished by
screening of native (a-,~-,y-CD) and derivatized cyclodextrins, combined with
different buffer systems.
Recently, an increasing number of studies has been published conceming the
retention and selectivity mechanism in chiral separation. Wren [3,4] described the
difference in mobility of two enantiorners as a function of the cyclodextrin
79
CHAP1ER5
concentration, predicting an optimum cyclodextrin concentration as a function of
equilibrium constants. Rawjee et al. [5] extended this model studying the chiral
selectivity as a function of pH and CD-concentration. Other parameters influencing the
resolution of enantiomers that were studied include temperature [6,7], concentration of
organic modifier and buffer concentration [8].
The models presented in Ref. [3-5] have been discussed in detail in chapter 2. It
is important to notice that these models use the assumption that each CD molecule
contains buffer molecule(s). As a result, there will be a competition for inclusion
complexation between the analyte and buffer and solvent molecules. Since, under
normal conditions, the buffer concentration is much higher than the CD concentration,
the CD-buffer complex is omitted from further considerations. In genera!, this
assumption is valid (see section 2.3). However, it should be noted that complex
formation between CD and solute demands the release of buffer molecules from the
CD cavity. This was also recognized by Wren & Rowe [9] when they investigated the
role of organic solvent in chiral separations in CE. They proposed that the addition of
MeOH to the BGE reduces the affinity of the analyte for the CD cavity, and will
consequently reduce the size of the equilibrium constant. lt might be expected that the
buffer ions can influence the complex fonnation constant in a similar way
In this chapter, the chiral separation of some sulfonamide enantiomers was
investigated, using native and modified cyclodextrins as chiral selectors. The main
objective of this study was to investigate the influence of the co-migrating buffer anion
on complex formation and enantioselectivity. Another airn of this study was to
investigate both complex formation and stereoselectivity for these solutes as a function
of the properties of different native and modified cyclodextrins.
5.2 Experimental
5.2.1 Chemicals
Native a- and ~-cyclodextrins, as well as heptakis(2,6-di-0-methyl)-~
cyclodextrin (DIME-~), heptakis(2,3,6-tri-O-methyl)-~-cyclodextrin (TRIME-P), (2-
hydroxy)-propylated ~-cyclodextrin; degree of substitution 4.6 (HP-P-CD), and
soluble ex- and ~-cyclodextrin polymers were from Cyclolab (Budapest, Hungary). All
CD's and the sodium salts of the racemic sulfonamides, 2-rnethylbutylsulfonamide
(MBS), ~-methyl-phenethylsulfonamide (MPS), and 2-butylsulfonarnide (BS) were
kind gifts from DSM Research (Geleen, Netherlands). The enantiomers (R)- and (S)-
80
THE INFLUENCE OF THE NATURE OF THE BUFFER
BS, and (S)-MBS were synthesized in the laboratory of Organic Chemistry of the
Eindhoven University of Technology. The structural formulas of these substances are
presented in Figure 5.1.
5.2.2 Apparatus
A P/ACE 2200 capillary electrophoresis system (Beckman, Fullerton, CA) was
used for all electrophoretic experiments. Polyacrylamide coated and uncoated fused
silica capillaries (50 µm LD.) of different lengths were applied. Capillaries were coated
according to the procedure of van der Schans et al. [ 10]. The capillary cartridge was
thermostated at 20°C. When coated capillaries were used, only the capillary was filled
with BGE containing a specific cyclodextrin concentration. In case of uncoated
capillaries, the capillary and the inlet via! were filled with BGE containing cyclodextrin.
In none of the experiments cyclodextrins were present in the outlet via! during
separation. The UV-detector was operated at 214 nm, except for the indirect UV
experiments using TRIS/benzoate buffer (230 nm) and TRIS/chromate buffer (280
nm).
5.2.3 Methods
All solutions were prepared in demineralized water. The BGE's were prepared
by adjusting a 100 mM TRIS (Tris(hydroxymethyl)-aminomethane, Merck, Darmstadt,
Germany) solution with maleic acid, fumaric acid, hydrochloric acid or chromic acid to
pH 8.2, or with benzoic acid to pH 8.5.
TABLE 5.1 COMPOSITION OF THE APPLIED BUFFERS
Abbr. CATION ANION i (mmo!/!)
BGEAl lOOmMTRIS chromate 64.6
BGEA2 40mMTRIS chromate 25.8
BGEB lOOmMTRIS benzoate 27.6
BGECl 100 mM TRIS chloride 43.1
BGEC2 60mMTRIS chloride 25.9
BGED lOOmMTRIS maleate 64.6
BGEE 100 mM TRIS fumarate 64.6
81
CHAP1ER5
Chromic acid was prepared by percolating a potassium chromate (Merck, Darmstadt,
Germany) solution over a strongly acidic cation exchanger (Type I, Merck, Darmstadt,
Germany). All BGE's applied in this study, with their respective ionic strengths (i), are
listed in TABLE 5.1.
To measure relative viscosities of BGE's containing cyclodextrin, the capillary
was first filled with a BGE containing cyclodextrin in a specific concentration, after
which a small amount of water was injected, and finally the BGE under study was
pumped through the capillary at constant low pressure (3.3. I03 Pa), until the water
plug was detected. The same experiment was also carried out with the corresponding
BGE without cyclodextrin. The relative viscosity of BGE containing a specific
cyclodextrin was calculated from the ratio of the detection time of the water dips
obtained from two corresponding BGE solutions, with and without a specific
cyclodextrin.
Samples were injected hydrodynamically for 10 s (3.3.103 Pa). The
concentration of injected samples was 104 M. The applied fieldstrength was in the
range 30-50 kV/m using the constant voltage mode.
82
THE INFLUENCE OF THE NATURE OF THE BUFFER
5.3 Results and discussion
Electrophoretic experiments were performed in different BGE's at pH 8.2 or
8.5. At this pH, the analytes were moving as anions, opposite to the direction of the
MPS
BS
~NHS03-Na+ MBS
EOF. The influence of the acid dissociation
equilibria can be neglected, since at this pH the
sulfonamides can be considered strong acids. Under
these conditions, using positive polarity, both the
electroosmotic mobility (water dip) and the mobility
of the analytes could be determined in one run,
providing an accurate calculation of effective
mobilities. Native cyclodextrins ( a- and ~-CD),
modified CD's (DIME-~, TRIME-~, HP-~-CD) and
neutra! a- and ~-cyclodextrin polymers were
dissolved in several concentrations in the BGE's
under study. One of the objectives of this research
was to study the influence of the co-migrating buffer
anion on the inclusion-complex formation between
cyclodextrin and analytes. Therefore, different
electrolytes were prepared (see TABLE 5.1).
Wren and Rowe [3] suggested a model for
Figure 5.1 Chemica[ structure the separation of optica! isomers in capillary
of the sulfonamide optica/ electrophoresis describing the change in effective
isomers. For abbreviations see mobility (corrected for the electroosmotic mobility)
section 5.2. 1.
µ_ + µcKclCDJ 1 + Kc[CD]
of enantiomers (µtorµ;!!, in genera! µ,!! ) as a
function of the equilibrium constant for complex
formation between analytes and cyclodextrin (KRcv
or Kscv , from here on referred to as Kc-
( 5-1)
[CD] = concentration of cyclodextrin [mol/l]
µ. = effective mobility of fully dissociated free enantiomer [m2/V .s]
µc = effective mobility of cyclodextrin-enantiomer complex [m2 IV .s]
Kc = equilibrium constant for complex forrnation [-]
83
CHAPTER5
This equation can be transformed to:
( 5-2)
Consequently, according to this model, if complex-formation occurs, a graphical
representation of (µ.-µ,.tr)l[CDJ versus µ•ff must show a linear relationship, with a slope
equaling Kc and the µ,raxis being intercepted at µeff = µc- In these experiments,
standard errors in Kc and µc calculations were smaller than 10%, depending on the
magnitude of the formation constant. Day to day reproducibilities were also within
10%.
The relative viscosities ofBGE's containing cyclodextrins were determined as a
function of the concentration of cyclodextrins added to the BGE and all measured
mobilities were corrected for the influence of viscosity by multiplying with the relative
viscosity.
5.3.J DIME-/3
First the influence of the co-migrating buffer ion on complex formation was
studied for the modified cyclodextrin DIME-~. As an example in Figure 5.2, effective
mobilities of MPS, BS, and MBS enantiomers are plotted as a function of the
concentration of DIME-~ in the BGE. Obviously, increasing concentrations of DIME
~ result in a decrease of the apparent mobilities of MPS and MBS, with MPS showing
a higher affinity towards the cyclodextrin as compared to MBS. However, BS
apparently shows very little interaction with the cyclodextrin, since hardly any decrease
in mobility is observed.
In Figure 5.3 the relationships of (µ.-µ,ff)/[CD] versus the apparent mobility µ,ff
of MPS and MBS for the different BGE's are plotted. As can be seen from the slopes
of the curves, the mean equilibrium constant Kc is higher for MPS than for MBS. This
confirms the conclusions drawn from Figure 5.2.
In TABLE 5.2 all Kc-values for the different chiral selectors and BGE's are
summarized. The higher Kc-values for MPS in case of DIME-~ could be attributed to
the aromatic structure of MPS, which is more likely to fit in the hydrophobic cavity of
the cyclodextrin than the aliphatic chain of MBS. Furthermore, for DIME-~ as the
chiral selector, the Kc-values of MPS and MBS appeared to be independent on the
applied BGE-anion.
84
THE INFLUENCE OF THE NATURE OF THE BUFFER
30
• • .~,;---..-----.-----·----------------------" 25 \
" \
~ 20
l.,.& ,~" ,, " ............ 5
10
• \ \ \ •,
.... ... ,_
---~-----·-----------------------· "
.... . "" • "'-1 -----·-----! _______________________ .
5'--~~~-'-~~~~-'-~~~~'--~~~--'
0 50 100 150 200
conc. DIME-8 [mM]
--•--· BS
__ .._ __ MBS
__ .._ __ MPS
Figure 5.2 Effective mobilities (correctedfor viscosity effects) of MPS, BS and MBS
enantiomers versus concentration DIME-/3. BGE: 100 mM TRIS!benzoate, pH 8.5.
Capillary: 401-468 mm, 50 µm l.D" Separation voltage I 5 kV.
Since benzoic acid is likely to enter the CD-cavity and as a consequence could
decrease the interaction between analyte and CD it was expected that the equilibrium
constant in buffers containing benzoic acid would be lower compared to BGE's not
containing aromatic compounds. However, both for MPS and MBS, K,-values
appeared to be independent of the buffer-anion under these conditions. lt also appears
from Figure 5.3 that the magnitude of the complex mobility (µc) is approximately the
same for the MPS- and the MBS-DIME-P complexes, in case the same BGE was
applied. Furthermore, it should be noted that the formation constant seems
independent on the concentration of the BGE's (Al versus A2).
85
CHAP1ER5
BGEAl BGEA2 BGE B BGE Al BGE A2 BGE B _._ .. 0- --·- -- .. -6· . ---A---MPS MPS MPS MBS MBS MBS
0.12
0.10
..... ~ 0.08
.! ë
..,. 0.06
t:::' s }
0.04 " ~ 0.02
0.00 =::......::..:;,,;,_~ ....... :.::.=------'--------'---------J 5 10 20 25
Figure 5.3 Graphical determination of Kc values and mobilities of selector
enantiomer complexes (µc) for complexes between DIME-/3 and MBS and MPS in
different BGE's. Experimental conditions as in Figure 5.2. Composition of BGE's is
listed in TABLE 5.1.
Opposite to that, µc values for both MPS and MBS appeared to be dependent
on the BGE nature and composition. When benzoate (BGE B) was used as the co
migrating anion, the µc-values for both MBS and MPS were about 50% Jarger as
compared to chromate (BGE Al). Besides the nature of the buffer this observation
might also be attributed to the higher ionic strength of BGE Al compared to the BGE
B ( see T ABLE 5.1). In order to allow a fair comparison at equal ionic strengths,
experiments were carried out applying BGE A2. As illustrated in Figure 5.3, µc does
increase upon decreasing the ionic strength, although the complex mobility still doesn't
match the value obtained in BGE B. Furthermore, µc-values for MPS and MBS were
calculated and compared for BGE B, BGE Cl and BGE C2. The mobilities of the
MPS and MBS-complexes were -9.0.10-9 in BGE B, -7.5.10-9 in BGE Cl, and -8.2.10-9
m2N.s in BGE C2 respectively. These results show that µc-values decrease upon
increasing the ionic strength of the BGE, as expected.
86
THE INFLUENCE OF THE NATURE OF THE BUFFER
TABLE 5.2 EQUILIBRIUM CONSTANTS OF COMPLEX FORMATION (Kc)
FOR DIFFERENT MODIFIED CYCLODEXTRINS AND DIFFERENT BUFFER
SYSTEMS. AVERAGE VALUES ARE TABULATED FOR EACH PAIR OF
ENANTIOMERS. COMPOSITION OF THE BGE's IS LISTED IN TABLE 5.1.
BGE DIME-P a-CD- P-CD- HP-P- TRIME-p
polymer polymer CD
MPS MBS MPS MBS MPS MPS MPS MBS
Al 117 22 52 59 14 20
B 120 22 30 28 80 70
Cl 120 48 155 130 16
D 25 48
E 108 25 51
According to the model suggested by Wren & Rowe [3], optimum difference in
effective mobility (as defined by eq. (5-1)) between two optica] isomers is obtained at:
{CD ]opt = 11 ~ K Rcv- K scv- ( 5-3)
This is in good agreement with the results of these experiments, showing optimum
mobility differences between the optica! isomers of MPS, at a DIME-P concentration
of IOmM.
As discussed earlier Kc-values appeared to be independent on the nature of the
BGE. The same might be true for the selectivity (S) which is given by the ratio of the
mobility difference of the two enantiomers and the mean mobility value of both optica]
isomers. The selectivity of DIME-P towards the optica! isomers of MPS seemed to be
independent on the choice of the co-migrating anion. For MBS however, no resolution
was obtained using BGE B whereas the optica! isomers were partly resolved (Rs=0.5)
using either maleate, chromate or fumarate as co-migrating buffer ion. This maximum
in resolution was obtained at about 50 mM DIME-P, which corresponds with the
optimum concentration predicted by eg. (5-3), using the results from T ABLE 5.2. For
the enantiomers of BS a partial resolution at 200 mM DIME-P was obtained, which
was shown to be independent of the nature of the corresponding BGE. Evidently,
interaction occurred between this analyte and DIME-p. Spiking of the racemic sample
with (S)-BS demonstrated that the (R)-enantiomer had a stronger interaction with
87
CHAPTER5
DIME-j} than the (S)-enantiomer. None of the chiral analytes could be baseline
resolved using DIME-j} as the chiral selector.
5.3.Z TRIME-{3
In Figure 5.4 the apparent mobilities of MPS and MBS enantiomers, after
correction for the BGE viscosity, were plotted against the TRIME-j} concentration for
two different BGE's. Compared to the results for DIME-j} (Figure 5.2), the decrease
in mobilities of the MBS and MPS solutes with increasing concentrations of the chiral
selector is less steep. This might be attributed to steric hindrance by the methoxy
group on the 3-position located on the wider entrance of the cyclodextrin rirn, while
DIME-j} is only methoxy substituted on the 2- and 6-position. On the other hand,
methylation of all the hydroxy groups makes the j}-CD more flexible. This might also
results in a less "tight fit", and consequently less stable complexes.
Kc-Values for TRIME-~ were experimentally determined for the BGE Al and
Cl, while no reliable data could be obtained for BGE B. This latter result might be
explained by the weaker interaction and retardation of the analytes with TRIME-~ in
BGE B. Consequently in such cases, standard errors in Kc-calculations become
relatively high. However, it can be safely assumed that Kc-values in BGE B are
significantly smaller than in any of the other buffers. The results are summarized in
T ABLE 5.2. Due to the lower ionic strength of BGE B, it might be expected that
complex-mobilities (µc) are higher than in other electrolyte systems.
In Figure 5.4 it is shown that under the experimental conditions MPS is more
retarded than MBS. This is not in agreement with the calculated equilibrium constants
(Kc) for the optica] isomers of MBS (MBS J and MBS2) of 17 respectively 22, in BGE
Al whereas Kc = 14 for MPS in the same BGE. The calculated complex-mobility (µc)
however for MBS is high (-19.5.10·9 m2N.s) compared to MPSffRIME-j} cornplexes
(-9.2.lff9 m2N.s), thus explaining the higher retardation for MPS. A possible
explanation of the high value of Pc for TRIME-j}/MBS cornplexes could be that
complex formation differs from the regular "single analyte - single CD-molecule"
mechanism [11].
88
THE INFLUENCE OF THE NATURE OF THE BUFFER
~ 5
"i = ::::.
" ::s.
25
20
"·
·······6·"'''""'"'"''''''
······•········•···········
"· ., .. ..
15~~~~~~~~~~~~~~~~~~
0 20 ~ ~ w conc. TRIME-R [mM]
- MPS: BGEB
." .•. " MPS: BGEAI
___.,...._ MBSI: BGE B
-tr- MBS2: BGE B
"" .... ". MBSl:BGEAI
""Ó"" MBS2:BGEAI
Figure 5.4 Effective mobilities f or MPS and MBS enantiomers ( corrected for
viscosity effects) versus the concentration TRIME-/3. MBSI is the strongest
interacting enantiomer. BGE's as in TABLE 5.1. Experimental conditions as in
Figure 5.2.
Furthermore, from the data it can be seen that the analytes are more retarded in
BGE Al than in BGE B. This can be explained either by a stronger complex-formation
or by a lower complex mobility in BGE Al. As an example in Figure 5.5 the baseline
resolution of the MBS enantiomers using TRIME-~ as the chiral selector in BGE Al
and BGE B is presented.
Resolutions (R,) were calculated according to eq. (5-4):
( 5-4)
with t1 and t1 being the migration times of the first and second enantiomers, and wh the
peak width at the baseline. Resolution of the MBS-enantiomers in BGE B is slightly
better (1.2) than in BGE Al (1.0). The higher resolution has to be explained by a
89
CHAPTER5
higher efficiency since the selectivity in BOE Al is higher than in BOE B (see Figure
5.6). The higher efficiency might be clarified by a much better mobility-matching
between MBS (fl.u= -23. 10·9 m2N.s) and benzoate (Jl,u= -30. 10·9 m2N.s), compared
to MBS and chromate {µeg= -65. 10·9 m2N.s) [ 12]. Spiking of racemic MBS with (S)
MBS proved that (S)-MBS interacts stronger with TRIME-P than (R)-MBS, under
these conditions.
Also selectivities of the MBS enantiomers for different BOE' s were calculated
according to the definition mentioned earlier. From the results in Figure 5.6 it can be
seen that selectivities of the MBS enantiomers are higher in BOE Al than in BOE B.
These observations are in good agreement with the results obtained with DIME-P for
the separation of MBS enantiomers. According to the data in T ABLE 5.2 and to eq.
(5-3), a maximum mobility difference between MBSI and MBS2 was expected at 50
mM TRIME-P using a 100 mM TRIS/chromate BOE. From these results however, no
clear maximum in mobility differences could be distinguished. Tuis could be explained
by a different mobility (pc) between MBS l/TRIME-P and MBS2/TRIME-P
complexes. According to these data this mobility difference valued 7%. Wren recently
modified the mathematica! model [3] to deal with such differences [13], although in his
case (atenolol in modified p-cyclodextrin) the relative differences were smaller.
5.3.3 Neutra/ a;. and f>cyclodextrin polymer
Neutra! ex- and P-cyclodextrin polymers were also included in this study as
chiral additives to the BOE. P-Cyclodextrin polymer did not show any
enantioselectivity towards BS and MBS. However for the MPS enantiomers, good
selectivity was obtained. Equilibrium constants between the 1)-cyclodextrin polymer
and MPS were determined for BOE B and BOE Cl. Kc· Values for the MPS
enantiomers were 78 and 83 for benzoate and 151 and 160 for chloride as the co
migrating anion. In order to calculate these values, the polymer concentration was
expressed as eq.r1• The amount of cyclodextrin units per mg of polymer was calculated
according to the specification of the manufacturer (!)-CD polymer sample: !)-CD
content 58.2% w/w, ex-CD polymer sample: ex-CD content 54% w/w) as in Ref. [10].
Consequently 100 mg P-CD-polymer/ml equals 50 meq P-CD/I, while 100 mg/ml ex
CD-polymer is equal to 55 meq cx-CD/l. Opposite to DIME-~, for the P-cyclodextrin
polymer, the Kc-values for MPS were strongly influenced by the nature of the co
migrating anion. Presumably, complex formation may be suppressed by benzoate,
competing with the analyte for inclusion in the cyclodextrin cavity, without however
affecting selectivity (expressed as SF).
90
THE INFLUENCE OF THE NATURE OF THE BUFFER
-0.002
-0.003
-0.005
-0.006 +-----~-----+-----~-------<
6 7 8
time [min]
Figure 5.5 Chiral separation of MBS enantiomers applying BGE Al and BGE B,
supported with 80 mM TRIME-/3. Composition of the BGE's as in TABLE 5.1. Other
experimental conditions as in Figure 5.2.
91
CHAPTER5
0.025
0.020
::::: ,..... f!t, 0.015 ;., ;: ~ ~
il 0.010 'il
(it.)
0.005
0.000 0
"" ..... .à:"···
."" ... "."··ll····
... ".~ ... "" ... ·····
... " .............. "." .... " .. .....
20 40 60
conc. TRIME·8 [mMJ
"." " ... " " ..
80
Figure 5.6 Selectivity of MBS enantiomers versus the concentration of TRIME-{J
applying BGE Al (à dotted curve) and BGE B ( t::.solid curve). Other experimental
conditions as in Figure 5.2. Curve fit according to a power function.
From the Kc-values shown in TABLE 5.2, similar trends can be observed for
the MPS and MBS racemates with the a-CD polymer. Kc-values increased twofold if
chromate was used instead ofbenzoate as the co-migrating anion. According to eq. (5-
3), this will influence the optimum cyclodextrin concentration in the BGE. Since the
experimentally measured Kc·values are nearly the same for MPS and MBS, [CD]opr
was expected to be about the same for both pairs of enantiomers, equaling 34 meq/l
(63 mg/ml) for BGE B, and 18 meq/l (32 mg/ml) for BGE Al. In Figure 5.7 and
Figure 5.8 selectivities of MPS and MBS enantiomers respectively, are plotted as a
function of the a-CD polymer concentration, applying different BGE' s.
From these results two observations can be made. First, as was also the case
for TRIME-~ with MBS (Figure 5.6), no clear maximum in either selectivity or
mobility-difference could be distinguished. As was the case with TRIME-~, this might
be explained by a different mobility (µc) of the complexes of a-CD polymer and the
two optica! isomers. According to these data this mobility-difference valued 4% for
MPS complexes. For MBS complexes this difference valued 3%. Also from the data it
92
THE INFLUENCE OF THE NATURE OF THE BUFFER
is obvious that the selectivity in BGE B is again substantially lower than for the other
BGE's. E.g., to obtain a selectivity for MBS enantiomers (Figure 5.8) of 0.013, a
concentration of 100 mg/ml o.-CD-polymer is needed in a BGE B, while 20 mg/ml
polymer is sufficient in the other BGE's. Thus, from this data it is obvious that
benzoate has a negative effect on the enantioselectivity (S) for both these sulfonamides.
In this case this effect was even more pronounced than with TRIME-~ as the chiral
selector (Figure 5.6).
:c @ .....
;<:::
~ 1 ~
0.025
0.020
0.015
0.010
0.005
. .. """." "".""."."." .... , ... " ..
0.000 Jf""''-----+------1----+----~,f-----+ 0 20 ~ w w 100
conc. a-CD-polymer [mg/ml]
Figure 5.7 Selectivity of MPS enantiomers versus the concentration of the a
cyclodextrin polymer applying BGE Al (à and dotted curve), BGE B ( D. and solid
curve) and BGE Cl (Tand "dash-dot-dash" curve). Composition of the BGE's as in
TABLE 5.1. Other experimental conditions as in Figure 5.2. Curve fit according to
polynomial functions.
93
CHAPTER5
0.025
0.020
::::: @: 0.015 .~ l>-
'-C ál
0.010 jj
0.005
0.000 0
... "."."" .. """" ····•··· ·······~::~::~"-------··J .".·.:-.:.:~:-"-·--.. "·.,.,......-
."-~,,,.,." ·""'" ·V
"~~ ,.;•;;;-' .~
./ ..,/" ...
/ 20 40 60 80 100
conc. a-CD-polymer [mg/ml]
Figure 5.8 Selectivity of MBS enantiomers versus the concentration of the a
cyclodextrin polymer applying BGE Al (.à and dotted curve), BGE B ( b. and solid
curve) and BGE E (Y and "dash-dot-dot-dash" curve). Composition of the BGE's as
in TAB LE 5. 1. Other experimental conditions as in Figure 5.2. Curve fit according to
polynomial functions.
As an example Figure 5.9 shows a full resolution electropherogram of MPS
and MBS enantiomers using a-cyclodextrin polymer as the chiral selector. Spiking of
racemic MBS with (S)-MBS demonstrated that (S)-MBS has the strongest interaction
with the polymer under these conditions, and was the second peak to be detected, as
was also the case for TRIME-P.
94
10 11
THE INFLUENCE OF THE NATURE OF THE BUFFER
12
time [min]
13
Figure 5.9 Separation of MPS and MBS enantiomers applying BGE Al (see TABLE
5.1) containing 100 mg/ml a-cyclodextrin polymer. Coated capillary: 300-370 mm,
50 µm l.D .. Separation voltage 15 kV.
95
CHAPTERS
5.3.4 HP-{J-CD
HP-13-CD (degree of substitution 4.5) was also included in this study for its potential
as a chiral selector. The results in T ABLE 5.2 show that complex formation is
influenced by the nature of the co-migrating anion, similarly as for a- and 13-CD
polymer. HP-13-CD showed enantioselectivity for the MPS enantiomers (separation
factor SF = 1.05) but no selectivity was observed for the optica! isomers of BS and
MBS.
S.4 Conclusions
Baseline resolution was obtained for the enantiomers of MPS and MBS for
several modified cyclodextrins. Only partial resolution was obtained for the optica!
.isomers of BS. Complex formation was markedly influenced by the nature of the co
migrating buffer anion, in the case of HP-13-CD and a- and 13-CD polymers being
applied as chiral selectors, with an exception of DIME-13. The formation constants of
MPS and MBS with HP-13-CD, and a- and 13-CD polymers were approximately
halved, when applying a BGE containing benzoate instead of any of the other BGE's.
MBS enantiomers were partly separated using DIME-13 in maleate, fumarate,
and chromate containing BGE's, but no separation was obtained in a TRIS/benzoate
BGE. Also for TRIME-13 and the a-CD polymer it is shown that benzoate has a
negative effect on the enantioselectivity. However in these cases resolution is not much
influenced by the nature of the BGE, probably since benzoate has a better mobility
matching with the analytes than the other buffer anions such as chromate or chloride.
The use of benzoate in the BGE generally has an adverse effect on chiral interaction
due to competitive inclusion in the CD cavity.
96
TIIE INFLUENCE OF TIIE NATURE OF THE BUFFER
Reference
L S. Fanali, J. Chromatogr" 474 (1989) 441
2. B. Chankvetadze, G. Endresz and G. Blaschke, J. Chromatogr. A.,
700 (1995) 43
3. S.A.C. Wren and R.C. Rowe, J. Chromatogr., 603 (1992) 235
4. S.A.C. Wren, J. Chromatogr., 636 (1993) 57
5. Y.Y. Rawjee, D.U. Staerk and G. Vigh, J. Chromatogr" 635 (1993) 291
6. A. Guttman, A. Paulus, A.S. Cohen, N. Grinberg, and B.L. Karger,
J. Chromatogr., 448 (1988) 41
7. S.M. Han and N. Purdie, Anal. Chem., 56 (1984) 2825
8. B.A. Ingelse, F.M. Everaerts, C. Desiderio and S. Fanali, J. Chromatogr. A"
709 (1995) 89
9 S.A.C. Wren and R.C. Rowe, J. Chromatogr" 609 (1992) 363
10. M.J. van der Schans, J.L. Beckers, M.C. Molling and F.M. Everaerts,
J. Chromatogr. A, 717 (1995) 139
11 D.W. Armstrong, F. Nome, F. Spino and T.D. Golden, J.A.C.S, 108 (1986)
1418
12. Y.Y.Rawjee, R.L. Williams, G.Vigh, Anal. Chem., 66 (1994) 3777
13. S.A.C.Wren, Electrophoresis, 16 (1995) 2127
97
CHAPTER5
98
SIMULA TIONS AND THERMODYNAMICS
6. COMPUTER SIMULATION AND BASIC THERMO-
DYNAMICS OF CHIRAL SEPARATIONS IN CE
Abstract
A previously published steady-state simulation program for CE was extended with a
sub-menu for chiral interaction. The interaction was modeled with a hypothetical
(neutra/) selector with properties similar to cyclodextrins (CD's). A three-type chiral
interaction model was implemented in such a way that it was valid for both anionic
and cationic analytes. The pe1formance of the program is illustrated with simulations,
using chiral parameters obtained from literature (homatropine) and from own
experiments and similar experiments peifonned in another laboratory (terbutaline
and mandelic acid). Simulation results were compared with experimental
electropherograms. Furthermore, the effect of temperature on the electrophoretic
chiral separation of ibuprofen with [3-CD is investigated. From mobility
determinations at different temperatures and CD-concentrations, it was possible to
calculate some basic thennodynamic parameters concemed with complex fonnation.
Por the chosen example, it is shown that temperature increase always results in the
need of an increased concentration of /3-CD, in order to achieve equal selectivity.
However, it is shown that temperature optimization can be used to minimize
separation time.
The study presented in this chapter has resulted in the following publications:
J.C. Reijenga, B.A. Ingelse and F.M. Everaerts, J. Chromatogr. A, in press
B.A. Ingelse, K. Sarmini, J.C. Reijenga, E. Kenndler and F.M. Everaerts, Electrophoresis, in press
J.C. Reijenga, B.A. Ingelse and F.M. Everaerts, Electrophoresis, in press
6.1 Introduction
For training purposes, fast simulation programs based on steady-state rnodels
were found a useful addition to real laboratory experiments [!]. This was already
obvious for techniques such as gas chromatography [2] and rnicellar electrokinetic
chrornatography [3], where elementary equilibrium thermodynarnics can be readily
99
CHAP1ER6
used to model the temperature dependent partition between mobile and (pseudo)
stationary phase.
In capillary electrophoresis however, the fundamental basis for separation
behavior is the concept of mobility, which together with basic acid-base equilibria, can
be used to model the separation pattem in detail. A steady-state simulator based on
such a model is now available for training purposes [4,5]. Although in that simulator,
temperature and ionic · strength effects are modeled, using empirica! relations,
complications such as complex formation and stereoselectivity were not taken into
account. The use of electrophoretic techniques for separation of optica) isomers has by
now gained widespread attention. Recent reviews on the subject [6,7], with hundreds
of references, list many applications of the use of cyclodextrins and other chiral
selectors in CZE, MEKC and EKC. In spite of the large number of references on chiral
applications now available, quantitative data on modeling the chiral interaction are not
easily obtained. A three-type interaction model for chiral separation of acidic [8] and
basic compounds [9] was clearly outlined by Rawjee et al.. Both ionic and non-ionic
interaction of the two optical isomers were quantified with corresponding complex
formation constants. If, in addition, the effective rnobility of the free analyte and of the
analyte-selector complex are known, a 6 parameter model is obtained that was
confirmed to agree well with experimental results [8,9].
The previously published HPCESIM [3,4] simulation program was extended
with a sub-menu for chiral interaction, based on the above mentioned chiral
parameters. Although this software is mainly intended for training purposes, in cases
where sufficient data is available about the analyte-selector equilibria, the program is
also suitable for method development of chiral separations in CE. In current study,
simulations of chiral separations were compared with literature. Furthermore, an acidic
and basic compound were selected and all parameters which are relevant for complex
forrnation were determined; pKa values, mobilities of the solutes and their formation
constants with the chiral selector. Two cyclodextrins were chosen in order to
distinguish the three different mechanisms of selective complex formation, as proposed
by Rawjee et al. [8,9]. Experimental results were compared with computer simulations
using the experirnentally determined chiral parameters listed above. In addition, the
results were compared with the results obtained in another laboratory.
As a next step, the influence of temperature on chiral separation was studied
quantitatively. The effect of temperature on electrophoretic separations is well known
because it influences, in principle, rnany of the parameters, variables and constants
involved in the separation, such as mobilities and pK values of both analyte and buffer
ions [10-12]. In chiral separations in CE, several additional formation constants
100
SIMULATIONS AND THERMODYNAMICS
between analyte and chiral selector, and their temperature dependence are involved as
well. Detailed knowledge of the magnitude of these effects will lead to a better
understanding. As a result, temperature may in some instances also be used as a tool
for fine-tuning resolution, provided that the separation compartment can be sufficiently
thermostated in order to ensure a homogeneous temperature throughout the analysis
time. The second part of this chapter focuses on the chiral separation of ibuprofen,
using ~-cyclodextrin as chiral selector. Equilibrium constants of complex formation
were determined at different temperatures. The results allowed the determination of
the corresponding thermodynamic properties: AH and LiS.
6.2 Experimental
6.2.1 Simulation
The simulation software was programmed and compiled in PowerBasic version 3.0
(PowerBasic, Carmel, CA). For running the software, a 386 type PC with 33 MHz
clock speed is a minimum requirement in order to keep calculation times within a
fraction of a second. This enables "instant" response necessary for use of short-cut
keys.
6.2.2 Chemicals
BGE's were prepared with analytica! grade chemicals from the usual sources
(Merck, Darmstadt, Germany or Sigma, St. Louis, MO). ~-Cyclodextrin (~-CD),
racemic mandelic acid, and the pure optica! isomers of mandelic acid were obtained
from Fluka (Buchs, Switzerland). Heptakis-(2,3,6-tri-O-methyl)-~-cyclodextrin
(TRIME-~) and hydroxypropylated ~-CD (HP-B-CD) were obtained from Cyclolab
(Budapest, Hungary). The degree of hydroxypropyl substitution was 6.5. Terbutaline
was obtained from Sigma (St. Louis, MO).
6.2.3 Methods
Experirnents were performed using P/ACE 2100, P/ACE 2200 and P/ACE
2500 equipment (Beckman, Fullerton, CA) with fused silica capillaries of different
lenghts. Detection wavelength was 214 nm. Samples were injected for 5 seconds. The
101
CHAPTER6
injection pressure was 3.3.103 Pa. All experiments were performed at 25°C. The
ibuprofen experiments were performed at 25, 32, 40 and 50°C.
The chiral parameters of mandelic acid and terbutaline were determined in the
Laboratory of lnstrumental Analysis at the Eindhoven University of Technology
(Eindhoven, The Netherlands) and in the Institute for Analytical Chemistry at the
University of Vienna (Vienna. Austria). For the experiments performed in Eindhoven,
410-480 mm uncoated fused silica capillaries, I.D. = 50 µm were used. The dimensions
of the capillary used in Vienna were 300-370 mm and 100 µm I.D" A constant voltage
varying between 10 and 30 kV was applied. The resulting current was always lower
than 50 µA. For the determination of the pKa of mandelic acid, BGE's were used with
a pH in between 2.54 and 3.22. These BGE's were prepared by adjusting the pH of a
10 mM NaOH solution with formic acid. Similarly, BGE's for the determination of the
pK,. of terbutaline were prepared by adjusting the pH of a 10 mM HCJ solution with
tris(hydroxymethylamino)methane (TRIS) up to pH values in between 8.22 and 8.96.
To determine K1 and K2, 10 mM Na+/formate, pH 2.98 and 10 mM Cr/TRIS, pH 8.70
were used. ~-Cyclodextrin was dissolved in these BGE's at concentrations of 2.5, 5,
10 and 15 mM. HP-~-CD was used at concentrations of 5, 10, 20 and 40 mM. Prior to
use, these solutions were filtered with disposable 0.45 µm pore size filters.
Experiments for the determination of the thermodynamic properties, MI and
L1S, of complex formation were performed using ibuprofen as test compound. The
mobilities and pKa values of ibuprofen were determined at +20 kV in a 400-476 mm,
75 µm l.D. uncoated capillary with BGE's consisting of 10 mM NaOH, adjusted to pH
values in the range 3.0-5.0 with acetic or formic acid. The driving current was in the
range 20-40 µA, depending on pH and temperature. Mobilities were measured with
reference to the EOF dip. K1 and K2 were determined by measuring the effective
mobility of ibuprofen as a function of the CD concentration in the range 0-15 mmol/1 at
pH 4.20 (in 10 mM Na+/Ac') and pH 6.55 (in 10 mM Na+/MES) respectively. These
experiments were performed at +20 kV in a 400-470 mm, 50 µm LD. uncoated
capillary. Chiral resolution and L1K1 values were obtained from experiments at -25 kV
in a 300-370 mm, 50 µm LD. polyacrylamide coated capillary, with a BGE of 10 mM
sodium/acetate of pH 4.47. The pressure injection time was 2 seconds.
102
SIMULA TIONS AND THERMODYNAMICS
6.3 Simulations of chiral separations
6.3.1 Chiral sub-menu
The following user-defined variables were provided as model input: the
concentration of the driral selector, the complex formation constants of both non-ionic
(K1) and ionic analyte species (K2), the relative K differences (L1K1 and L1K2) between
the two optica! isomers (also for both non-ionic and ionic species) and the mobility of
the analyte-selector complex, relative to that of the fully dissociated free analy1e (M').
Chiral components in the database are simulated as nrixtures of optica! isomers with
variable enantiomeric concentration ratios. When adding chiral selectors, such as
cyclodextrins, to the BGE, the bulk viscosity often increases and as a result, the Ohnric
resistance of the capillary adjusts according to the decreased mobilities. These effects
can be of the order of 10%. The bulk viscosity was already included in the program as
an independent variable. Experiments in chapter 5 have shown that for cyclodextrins,
the viscosity increases with the cyclodextrin concentration according to a quadratic
relation:
1JIT/o 1 + aJ.[CD] + ai.[CD]2 (6-1)
in which T/o is bulk viscosity without chiral selector and [CD] the analytica!
concentration of the chiral selecter in either [moVI] or [equivalents/I] (see section
3.3.2. and 5.3.3.). The coefficients ai and a2 are now included as program parameters.
T ABLE 6.1 shows the values of these coefficients for some common modified
cyclodextrins. These values were obtained according to the method presented in
section 5.2.3.
The mobility-viscosity relation in non-sieving media was estimated by
monitoring the current at constant voltage as a function of the BGE viscosity. It was
observed that the BGE mobilities decrease linearly with increasing viscosity according
to the relation:
(6-2)
in which J1o * is the mobility of the fully charged ion in water and a3 is a constant that
surprisingly was found to be 0.78 for the ~-CD polymer mentioned. In other cases,
103
CHAPTER6
values of a1 close to zero were observed [13]. For this reason, the constant a.1 was also
included as a parameter.
TABLE 6.1 VISCOSITY COEFFICIENTS FOR SOME COMMON MODIFIED
CYCLODEXTRINS INWATER AT 20°C.
cyclodextrin conc. range coefficients
a1 a2
DIME-~ 0-200mM 2.42xlff3 2.79x10·5
TRIME-~ 0-80mM 2.89xlff3 4.00xl0-5
HP-~-CD (D.S.=4.6) 0-80mM 3.09xlff3 4.79xlff5
a-CD-polymer 0-100 mg/ml 6.57xl0-3 l.25x104
~-CD-polymer 0-100 mg/ml l.l 8xlff2 2.12x104
T ABLE 6.2 summarizes the additional chiral and non-chiral parameters with
their allowed range of values. The interaction type listed in T ABLE 6.2 refers to the
definitions introduced by Rawjee et al. [8,9].
TABLE 6.2 CHIRAL PARAMETERS IN THE HPCESIM SIMULATION
PROGRAM.
variable dimension range from to
Interaction Type 0 (none-chiral) m (ionic and non-ionic)
Selector Concentration mmol/I 0 200
K1 (non-ionic interaction) 0 32000
&1 (+vs.-) % -40 +40
K2 (ionic interaction) 0 32000
&2 (+VS.-) % -40 +40
M' (relative mobility) % 0 100
Purity (%+) % 0.01 99.99
a1 in eq. (6-1) 0 5
a2 in eq. (6-1) 0 100
a1 in eq. (6-2) 0 2
104
SIMULA TIONS AND THERMODYNAMICS
These definitions are explained in section 2.3. All these chiral parameters can be loaded
and saved as buffer files, together with the other buffer parameters (pH, composition
and ionic strength).
6.3.2 Calculation of migration behavior
Migration behavior in chiral systems is rather more complicated than in non
chiral systems. For example, an acidic analyte can be present in either of four species:
as an charged or as a non-charged acid (equilibrium subject to K"), as a non-charged
acid complexated with the chiral selector (subject to K1) and as a charged acid
complexated with the chiral selector (subject to K1). Only two of these species migrate,
in case a non-charged chiral selector is applied.
The effective mobility of chiral components in the presence of a chiral selector
(µeJJ) is calculated from the mobility of the fully charged ion (µo) according to eq. (2-
19) and eq. (2-20) for anions and eq. (2-24) for cations. For multivalent ions, the
denominators of these equations are accordingly adjusted.
In the software, two kinds of sample components are distinguished: those with
potential chiral properties (the names are stored in an auxiliary ASCII-type file) and
those without such properties. The former are all treated alike, in that they are
supposed to interact with the hypothetical chiral selector in exactly the same manner
(all chiral complexation constants are the same for all chiral analytes).
The non-chiral analytes also interact with the selector, again with the same K1
and K1 values, but the '1.K1 and '1K2 values are set to zero. Although not set as a
separate goal for the current version of the program, in this way some form of (non
chiral) complexation can be modeled as well, with the limitation of a non-charged
complex forming agent.
As long as the chiral selector concentration is zero, program performance
remains exactly what it used to be. Only with non-zero selector concentrations,
additional information is available: selectivity and resolution now refer to the
separation of the respective optica] isomers, resulting in zero values for non-chiral
analytes. Other tabulated information on migration and dispersion always refers to the
slowest of the two optica! isomers.
6.3.3 Visualization of results
As an example, the separation of a number of cations with potential chiral
properties was simulated with both K1 and K2 set to 100. The '1.K1 value is zero, '1.K2 is
set to 5%, the relative mobility to 10% and the optica] purity to 40%. This is an
105
CHAPTER6
ionoselective interaction where only the ionic form of the analytes shows chiral
discrirnination by the se lector. In Figure 6.1 a the analysis using non-chiral conditions is
shown. The separation is optirnized applying a BGE with a pH value in the range of
the pK values of the analytes. Consequently, they were separated according to their pK
value. Figure 6.1 b simulates a chiral selector concentration of 5 mmol/I.
a
~---L...--
b
~ ~ } ........
3 4 5 6 7 8
migration time [mini
Figure 6.1 Simulation of electrophoretic separation of procaine, /3-eucaine, cocaïne,
diocaine, psicaineleu, tetracaine and bupivacaine respectively without (a) and with 5
mM chiral selector (b). The buffer is a JO mM TRJS/borate at pH 8.0. Further
conditions as described in the text
6.3.4 Selectivity vs. resolution
As can be seen from Figure 6.1, chiral resolution is complete for the first few
analytes. The last rnigrating components show less resolution because of two reasons.
The degree of ionization of these components is less, so that with an ionoselective
interaction less selectivity can be expected under these conditions. In addition to this
difference in selectivity, resolution is also Jess because at increased rnigration times,
increased dispersion due to diffusion is expected. From th~ point of selectivity the
model proposed in the literature [8,9] appears to form a complete description of
interaction phenomena. The resulting equations for selectivity are independent of all µ0
106
SIMULA TIONS AND THERMODYNAMICS
values, under the assumption that the charged analyte-selector complex mobility has no
chiral distinction. Whether or not racemic mixtures can be separated, also depends on
additional peak-broadening contributions from injection volume, concentration
overload, diffusion, radial temperature gradient, detector cell length, and time constant.
These additional effects are included in the present sirnulations as well.
6.3.5 Comparison with literature results
Unfortunately, most publications on chiral separations Jack quantitative
information on the complex formation constants that can be used to model the analyte
solute interactions. In their publications describing the previously mentioned types I-III
interaction mechanîsms, Rawjee et al. [8,9] give a number of examples for anionic and
cationic sample components. Results of the optimization strategy were presented as
selectivities vs. pH and selector concentration plots. In addîtion, it wîll be shown that
the present software is capable of calculating not only these selectivities, but also the
resolution. As long as analysis time is considered as well, this is often a more practical
optimization criterion
As an example, the chiral separation of the cation homatropine [9], was
simulated. The chiral interaction is of type III, meaning that both the ionîc and non
ionic form of the analyte exhibit stereoselective interaction with the ~-CD, although to
a different extend (duoselective interaction). In this example, Rawjee et al. [9] found
for K1 1328, IJ.Ki 3.4%, K1 95.9, and !J.K2 8.4%. In other words, the non-ionic form
shows stronger but less stereoselective interaction, the ionic form shows weaker but
more stereoseleetive interaction. Although in the practical concentration range of ~
CD (0-15 mM) the migration sequence remains the same and a distinct selectivity
optimum is observed, at higher hypothetical concentrations a reversal of the migration
order is found. Figure 6.2 illustrates simulated analyses at three different chiral selector
concentrations. Obviously, if possible, benefit should be taken from the high value of
IJ.Ki which results in shorter migration times.
107
CHAPTER6
b
a
c
' ' ' 0 10 20 30 40
migration time [min]
Figure 6.2 Simulation of electrophoretic separation of homatropine at 5, 25 and 50
mM hypothetical chiral selector with model parameters and experimental conditions
as in Ref [9 ]. BGE: 10 mM phosphoric acid adjusted to pH 9.0 using LiOH.
6.3.6 Comparison with experimental results
An acidic (mandelic acid) and a basic (terbutaline) compound were selected and
all parameters, relevant for chiral separation, were experimentally determined in two
laboratories: Eindhoven and Vienna. Two cyclodextrins were chosen in order to
distinguish the three different mechanisms of selective complex formation, as proposed
by Rawjee et al. [8,9]. Experimental results were compared with computer simulations
using the experimentally determined chiral parameters.
6.3.6. J Detennination of the pK values and mobility of mandelic acid and terbutaline
In order to determine the pKa of mandelic acid and terbutaline, the effective
mobilities, Jleft, of these analytes were determined as a function of the pH of the BGE.
In Eindhoven, mobilities were determined by the dual-marker method by Williams et
al. [14], whereas mobilities in Vienna were determined from the total migration time of
108
SIMULA TIONS AND THERMODYNAMICS
the analyte and that of a neutra! EOF marker. The pKa could be determined graphically
by plotting the reciprocal value of the absolute value of the effective mobility versus
the hydrogen concentration ([H+]), according to the linearized equation of Henderson
Hasselbalch:
(6-3)
For the determination of the pK.i, of basic compounds, the [Off] concentration IS
plotted instead of the [H•] concentration.
0.16 0.16
b 0.14 a 0.14
0.12 0.12
"à O.l ~ 0.1
"" " ·==. 0.08 ~~ 0.08
" Jo.06
,; 1006
O.o4 0.04
0.02 0.02
0 0 0 0 2 4 6 10
IH'l [10'3 mol/IJ [OH') [xlO" mol/!]
Figure 6.3 Graphical detennination of the acid (A) and base (B) dissociation
constant/or mande/ic acid and terbutaline (datafrom Eindhoven).
The resulting graphs and forthcoming data, obtained in Eindhoven, are presented in
Figure 6.3 and T ABLE 6.3. Data obtained in Vienna in comparison with data obtained
in Eindhoven are presented in T ABLE 6.4. The differences in the p.K. values, obtained
in the two laboratories are smaller than 1 %. The difference between the experimentally
obtained data and literature data is substantial, especially for mandelic acid. (2.93 vs.
3.34). Literature values, however, are extrapolated to an ionic strength of i = 0 mM,
whereas the pKa values, presented in this study, were determined at i = 10 mM.
Obviously, the use of the literature value of the pKu of mandelic acid would have
109
CHAPTER6
introduced a major error in the determination of the formation constant of non-ionic
mandelic. acid with cyclodextrins. This stresses the importance of determining the
pK.'s, using identical experimental conditions (temperature, ionic strength), as those
applied for the determination of the formation constants.
TABLE 6.3 RELEVANT CHIRAL PARAMETERS OF MANDELIC ACID AND
TERBUTALINE INTERACTING WITH j)-CD.
parameter analyte
mandelic acid terbutaline
µo 10·9 m2N.s -28.7 (0.1) 20.5 (0.1)
pK. 2.94 (0.02) 8.71 (0.02)
M' 0.62 (0.15) 0.46 (0.03)
Ki 198 (5) 33.3 (1.5)
LlK1 % 0 0
K2 55.8 (7.8) 74.7 (2.2)
&2 % 0 18.2 (2)
Standard deviations are given in parentheses
6.3.6.2 Detennination of fonnation constants
Formation constants for terbutaline and mandelic acid were determined using
native j)-CD, as well as HP-j)-CD. Initia! experiments pointed out that only the
rnodified cyclodextrin was able to resolve the optica! isomers of both racemic analytes
whereas native j)-CD showed only enantioselectivity towards the terbutaline
enantiomers. For this reason, only HP-j)-CD was used in both laboratories, while the
determination of formation constants using j)-CD was only performed in Eindhoven
(results presented in TABLE 6.3).
In order to determine the formation constants of non-ionic interaction (K1), and
ionic interaction (K2), the effective mobilities of terbutaline and mandelic acid
enantiomers were determined at different concentration levels of cyclodextrin, using a
high and a low pH BGE. For the determination of Ki of terbutaline, a 10 mM
sodium/formate buffer of pH 2.98 was applied. For the determination of K1 of
mandelic acid, a 10 mM hydrochloric acid!fRIS buffer of pH 8. 70 was used. Under
these experirnental conditions, mandelic acid and terbutaline could be regarded as a
strong acid respectively base. After switching, the sarne BGE's could be used for the
determination of the K1 values of both racemic analytes, since the pH values of these
110
SIMULATIONS AND THERJ\.10DYNAMICS
buffers are close to the pK. values ofmandelic acid (pK. = 2.93) and terbutaline (pKa = 8.68). Consequently, at these pH values, both ionic and non-ionic interaction occur.
The K2 values could be determined by plotting the ratio of the mobility difference and
the concentration of chiral selector versus the effective mobility according to eq. (5-2),
presented in the former chapter.
µ_ (5-2)
This equation is only valid for strong acids and bases. Using the specified
experimental conditions (mandelic acid at pH 8.70 and terbutaline at pH 2.98) the
separands are treated as such.
Figure 6.4 shows the resulting graphs for mandelic acid and terbutaline
respectively, using the two different cyclodextrins: native ~-CD and HP-~-CD. The K2
values of both analytes could be determined from the slope of the linear regression line.
The resulting data is presented in T ABLE 6.3 and T ABLE 6.4. T ABLE 6.3 presents
the data relevant for the determination of the formation constants using native ~-CD,
whereas TABLE 6.4 presents these data for HP-~-CD. K1 and Ki, listed in these tables
are average values for both enantiomers.
TABLE 6.4 RELEVANT CHIRAL PARAMETERS OF MANDELIC ACID AND
TERBUT ALINE INTERACTING WITH HP-~-CD. A COMP ARISON BETWEEN
DATA FROM EINDHOVEN AND VIENNA.
parameter mandelic acid terbutaline
EINDHOVEN VIENNA EINDHOVEN VIENNA
/1<J 10·9 m2N.s -28.7 (0.1) -29.8 (0.1) 20.5 (0.1) 21.2 (0.1)
pKa 2.94 (0.02) 2.92 (0.04) 8.71 (0.02) 8.63 (0.04)
M' 0.76 (0.12) 0.79 (0.09) 0.38 (0.07) 0.42 (0.05)
K1 127 (4) 143 (3) 20.9 (0.5) 38.5 (1)
t1K1 % 6 (1) 13 (2) 22 (2) 46 (3)
K1 24.6 (2.3) 25.9 (2.4) 47 (5) 41 (4)
!JK2 % 0 0 44 (9) 36 (11)
Standard deviations are gîven in parentheses
111
CHAP1ER6
0.60
! :L 0.30 <;::;'
0.0018
0.60
9 ~ ,;. ~
:L 0.30 <;::;'
a
20
b
8 JO
, ___ ......... -·-22 24
11.,. uo·• m 2/V.s)
12 14
11.., (10 ·• m 2/V.s)
-a· ,A.···"· ••••• 11.r··
26 28 30
18 20
Figure 6.4 Determination of K2 for mandelic acid (a) and terbutaline (b)
enantiomers. The data/or /3-CD isfitted with solid lines andfilled symbols (Á in (a);
• and T in (b)), no discrimination could be made between the enantiomers of
mandelic acid; data/or HP-/3-CD isfitted with dotted lines and open symbols ( /::,. in
(a);o and 'Vin (b)). (•or o) = (+)-terbutaline, (Tor 'V) = (-)-terbutaline For y
axis, see text. (Data/rom Eindhoven, all experiments in duplicate).
112
SIMULA TIONS AND THERMODYNAMICS
In order to determine the K1 values of both analytes, the data was fitted
according to the theoretica! model of Rawjee et al. [8,9]. In this model it is assumed
that the pK values of the two isorners in the analyte-selector complexes are identical,
which is not obvious for all cases [15]. The obtained relations (see eq. (2-19), (2-20)
and (2-24)) can be linearized as follows:
For weak acids:
l+M'K2 [CD]µ0 [H~O+] --;z--K2 [CDJ-l
a
(6-4)
And for weak bases:
(6-5)
Figure 6.5a shows the left-hand side of eq. (6-4) plotted versus [CD], so that the slope
equals K1.[H30+]1Ka, resulting in the non-ionic complex formation constants for
mandelic acid enantiorners using the two different cyclodextrins. Analogously, K1
values for terbutaline could be determined by plotting the left-hand side of (6-5) versus
the cyclodextrin concentration. This is shown in Figure 6.5b for both cyclodextrins. All
forthcoming results are listed in T ABLE 6.3 and T ABLE 6.4.
All parameters determined experimentally were obtained from linearized
graphs. Error analysis of the scatter of these graphs readily yields standard deviations
of both slope and intercept, resulting in standard deviations of mobilities and K values.
These are tabulated along with their average values. The number of determinations
used in the regression analyses were 8-12 for the pKa determination and 6-8 for the
determination of the fonnation constants.
Seve~al observations regarding lab-to-lab differences can be made. In view of
calibrating the pH meter with pH-references, an error of 0.02 is a fair estimate. This
corresponds to the error in the pK. or pKb thus determined. In the determination of the
K1 values using equation (5-2), the error in [CD) is neglected. Small differences in
effoctive mobility are measured, accounting for the large standard deviation of the K1
values (around 10%).
113
CHAPTER6
4
~ 3
=-~ 2 "i I::'
0 10 20 30 40
conc. CD [mmol/I]
0.40
~ ~
! =: l;:;'
0.00
0.80 .-----------------....,,--_-__ -_-__ ....,_ ........
b --
-------------- .----- ...... -__ , ,,--' ,•
•' ,, ,, ... --- ---_9------- _ ...... ----- ... --,- ... -. .. __ Er... ---
,,-- v---,.--··
-0.40 0 10 20 30 40
conc. CD [mmol/I]
Figure 6.5 Determination of K1 for mandelic acid (a) and terbutaline (b)
enantiomers. L-( +)-mande late = (+ ), D-(-)-mandelate = ( L ). Other explanation of
symbols and styles of trend lines as in Figure 6.4. For y-axis, see text. (Data from
Eindhoven, all experiments at least in duplicate ).
114
SIMULA TIONS AND THERMODYNAMICS
If only regression statistics are taken into account for the detennination of the
K1 values, using equation (6-4) or (6-5), scatter alone yields a lower relative standard
deviation than obtained for Ki. This doesn't seem entirely correct, as the values for Ka,
Kb, [H+], [Off] and Ki are included in these equations as well. Having used averaged
values for these, their effect on the standard deviation is lost, so the standard deviation
of K1 is underestimated. For example, consider the right-hand side of equation (6-4) or
(6-5). The error in the Ka or Kb is 4.5% (except where the error in pH is 0.04, then it is
9%). The regression error is in the range 2-3%. The error in [W] or [Off] is around
5%. As a result amore realistic relative standard deviation of K1 would be around 5-
10%.
Native P-CD <lid not show any selectivity towards the enantiomers of rnandelic
acid. The enantiomers of terbutaline could be separated at low pH, where the basic
compound is fully protonated. No resolution was obtained at high pH (pH pK,). This
kind of separation is referred to as a Type II separation [8,9] or ionoselective
separation [ 16].
HP-P-CD was the more successful selector. It allowed the separation of the
enantiomers of both mandelic acid and terbutaline. The enantiomers of mandelic acid
could be separated at low pH, but not at high pH, indicating that the non-dissociated
acid interacts stereoselectively with the modified cyclodextrin. This kind of separation
is referred to as a Type I separation [8,9] or desionoselective separation [16]. Chiral
interaction, resulting in optica] resolution, between the enantiomers of terbutaline and
HP-~-CD was observed at both low and high pH. Both the non-ionic as wel! as the
protonated terbutaline internet stereoselective with the chiral selector. This kind of
separation is referred to as a Type III separation [8,9] or duoselective separation [16].
The migration order of the enantiomers was deterrnined by spiking the racernic
sample with pure standards. The L-(+)-isomer of mandelic acid rnigrated slower than
the 0-(-)-isomer, and therefore Ku>KDJ. This confirms earlier experiments ofNardi et
al. [17). Fanali [18) showed that (+)-terbutaline has the highest affinity towards the
cyclodextrin cavity. This was verified for native ~-CD, heptakis(2,6-di-O-rnethyl)-~
CD and heptakis(2,3,6-tri-O-methyl)-P-CD using a BGE at low pH (pH 2.5).
Using the experimentally obtained parameters presented in T ABLE 6.4, the
chiral separation of mandelic acid and terbutaline enantiomers was simulated, using
HP-P-CD at a concentration of 40 mM, using the low pH BGE. The sirnulation, using
the parameters obtained in Eindhoven, is shown in Figure 6.6a, the sirnulation using
parameters obtained in Vienna is shown in Figure 6.6b whereas the actual separations
(performed in Eindhoven) are shown in Figure 6.6c.
115
CHAPIER6
Reasonable coincidence is found between both simulated electropherograms
and the experimentally obtained electropherograms. The selectivities (S) calculated
using the "Vienna" simulation match the selectivity of terbutaline enantiomers in Figure
6.6c (0.077 vs. 0.073), whereas the selectivity for mandelic acid enantiomers is
overestimated (0.082 vs. 0.039). The "Eindhoven" simulatîon matches exactly the
experimentally obtained selectivity for mandelic acid enantiomers (0.039). This
simulation ho wever overestimated the selectivity for terbutaline enantiomers (0.104 vs.
0.073). The simulated efficiencies are overestimated in all cases resulting in much
higher resolutions than those calculated from Figure 6.6c. A possible explanation for
these high efficiencies can be an incorrect modeling of the "stacking" process by the
simulation algorithm, occurring after the injection of the sample.
In conclusion, it can be observed that the accuracy and precisîon of the
determination of chiral complex equilibrium constants in CE are limited. This is
especially so in case of the relatively small K values in the exarnples chosen, because
calculations are based on deterrninations of very small mobîlity dîfferences. In addîtîon,
small errors in pK., [H+], and M" values propagate in the calculatîon of the final K
values. Considerîng these relatively large standard deviations, reasonable agreement
was found between the values of the chiral parameters determined in both laboratories.
Agreement between simulations based on the two different sets of parameters and
between simulation and experiment were satisfactory.
116
SIMULA TIONS AND THERMODYNAMICS
100 ~
TB 8()
MA
s ~ 6()
"' ~ " € 40 ~ ~
20
0
0 2 4 6 10 12
time [min)
100
b 80 TB
s ~ 60
MA
~ " 1: j
40
<( 20
0
0 4 6 10 12
time[min]
0.010
0.008 c TB
s 0.006 . ~
~ 0.004
" M €
~ 0.002
0.000
-0.002
0 4 6 10 12
time [min]
Figure 6.6 Simulated electropherograms using chiral parameters from TABLE 6.4
determined in Eindhoven (a) and Vienna (b) versus the experimentally obtained
electropherograms in Eindhoven (c). TB = terbutaline enantiomers, MA = mande/ic
acid enantiomers. BGE: pH 2.98 supported with 40 mM HP-j3-CD. Separation
voltage 30 kV, T = 25°C.
117
CHAPTER6
6.4 Temperature eff ects and basic thermodynamics of the chiral
separation of ibuprofen enantiomers
In section 3.3.5, it is shown that increased temperatures generally have an
adverse effect on chiral selectivity. In this section, the effects of temperature on chiral
separations are studied in more detail, yielding some basic thermodynamic parameters.
Using these parameters, it is possible to predict the effect of temperature on chiral
selectivity. Consequently, temperature control can be utilized to fine-tune chiral
separations.
6.4.1 Effect of temperature on BGE conductivity and pK
Mobilities generally have a temperature coefficient of ca 2% K 1, that amounts
to a factor 1.64 between 25°C and 50°C. Experimentally, a factor of 1.52 was found
for the BGE consisting of 10 mM Na+/Ac· pH 4.98, a factor 2.0 for the BGE
consisting of 10 mM Na+/formate pH 3.06, as indicated by the driving current.
The choice of BGE buffering co-ion was determined by the availability of data
concerning the temperature dependence of their pK value. For formic and acetic acid,
~KIST values found from literature [12] were less than 0.0005 units per degree at
25°C. For this reason, it is assumed that the pH of the BGE is independent of
temperature under the experimental conditions used in the lower pH range. For the
experiments performed at pH 6.55, MES was used as a counter-ion. The temperature
dependence of the p.K,. of MES is considerable (-0.02 K 1), but this BGE was only used
for experiments to determine Ki. For these experiments, the pH is irrelevant as long as
pH> p.K,.+2 (for anions).
6.4.2 Effect of tempera/ure on EOF
Electroosmosis is caused by a negative Ç-potential of the capillary wall,
according to eq. (l-15). There is no reason to expect that the di-electric constant (e) or
the Ç-potential depend on temperature. This would mean that the temperature
dependence of µEOF can be modeled with the temperature dependence of viscosity,
which amounts to 2% K 1• This is confirmed by the experimental results. For each of
the BGE's, the temperature coefficient of µEoF was determined. The average value was
1.9% K 1 with a standard deviation of 0.3% K 1• As expected, the EOF strongly
118
SIMULATIONS AND THERMODYNAMICS
increases with the pH of the BGE. Values of µEoF at pH 4.98 are also tabulated in
TABLE 6.5. In the pH and temperature range mentioned, the absolute values for the
electroosmotic mobility were successfully fitted to the following two-parameter model:
µroF = -10.29 + 5.562.pH 0.5455.T + 0.2503.T.pH (6-6)
where µEoF is in 10'9m2Ns and T in °C. This relation can be quite useful but not
necessarily valid at higher pH values or in other buffer systems, especially at different
ionic strengths.
TABLE 6.5 THE EFFECT OF TEMPERATURE ON THE ELECTROPHORETIC
PARAMETERS OF EOF (AT pH 4.98) AND IBUPROFEN AT IONIC STRENGTH
10.0 mM. MOBILITIES ARE GIVEN IN 10'9 M2NS.
Temperature µEOF -µo s.d. pKa s.d.
25°C 35.35 20.91 0.7 4.37 0.02
32"C 40.23 23.10 1.6 4.36 0.02
40°C 45.85 27.18 l.9 4.39 0.02
50°C 51.63 31.95 2.3 4.40 0.02
%/oC 1.85 2.0 0
6.4.3 Effect of temperature on the mobility and pK" of ibuprofen
Inverse absolute values of the effective mobilities of ibuprofen were plotted vs.
the [H+] concentration, as shown in Figure 6.7. The intercept corresponds to 11µ0,
whereas the slope of this curve equals lOPKfµo. The highest [H+] concentration yielded
outlier values, which were discarded. The remaining 4 points, each determined in
duplicate, yielded good correlations. The results are summarized in T ABLE 6.5. The
pK. of ibuprofen has an average value of 4.38 (s.d. 0.02) which was independent of
temperature in the measured range. The value was compared with the following
literature values: 4.48 at 100 mM ionic strength and 37°C [8], or 5.10 at zero ionic
strength and 25°C [19]. No temperature dependence of the pKa value of ibuprofen was
found in the literature.
119
CHAPTER6
0.40
0.30 ii ,e-
! 0.20
0.10
0.00 ...._ ___ ....._ ___ __.._ ___ __,, ____ ..__ ___ _,
0.00 0.10 0.20 0.30 0.40 0.50
1000x [11•]
Figure 6. 7 Inverse absolute value of the effective mobility of ibuprofen as a function of the [F] concentration atfour different temperatures. lntercept and slope yield µo and pK0 values respectively. 't' =25°C, 'il =32°C, •=40°C, 0=50°C. All experiments in duplicate.
The mobility µ0 is -20.91.109 m2Ns at 25°C and has a temperature coefficient of2.0 %
K 1• A literature value of -21.32.10·9 m2Ns was found at 100 mM ionic strength and
37°C [8]. When applying correction for temperature and ionic strength according to an
empirical correction model [20], the difference between these mobility values is well
within standard deviation. The standard deviation of µ0 thus measured was high, due to
the fact that is was obtained through extrapolation of the reciprocal value.
6.4.4 Determination of K2 of ibuprofen-f:J-CD at different temperatures
Firstly, the K2 was deterrnined at pH 6.55, where only interaction between the
fully charged ibuprofen and ~-CD takes place. According to the literature [8), the
interaction involved is desionoselective, meaning that no resolution is expected under
these conditions. In Figure 6.8, the effective mobility of ibuprofen µ,g is plotted vs. the
~-CD concentration [CD). Mobility decreases strongly at low CD concentrations and
120
SIMULA TIONS AND THERMODYNAMICS
saturation is visible at l 0 mM, indicating a high K1 value. The values were fitted
according to eq. (2-25):
µ0 + µcK2 [CD] µ,ff = l+K
2[CD]
where µc is the mobility of the ibuprofen ~-CD complex.
30
.} 20
15
(2-25)
10~~~~~~~~~~~~~~~~~~~~~~----'
0 2 4 6 8 10
conc. 8-CD [mM]
Figure 6.8 Effective mobility of ibuprofen as a function of {3-CD and temperature at
pH 6.55 and ionic strength JO mmolll. Symbols as in Figure 6.7. For curve fitting of
K1 see text. All experiments in duplicate.
Coefficients of correlation ranged between 0.9990 and 0.9998. Values of µo, µc and K2
obtained from the fit are listed in TABLE 6.6. As can be seen, µo values obtained here
are systematically different from those measured without CD in the pH range 3-5. The
latter values were obtained through extrapolation of the data in Figure 6.7, the former
were directly measured at [CD] = 0. For this reason it was decided to use the µo values
in T ABLE 6.6 for subsequent calculations. The mobilities of the ibuprofen-~-CD
complex range around half of the value of the mobility of the free analyte, and increase
slightly at elevated temperatures. The K2 values, listed in T ABLE 6.6 are somewhat
121
CHAP1ER6
higher than previously published values for ibuprofen at 37"C and ionic strength l 00
mM [8].
TABLE 6.6 THE EFFECT OF TEMPERATURE ON THE CHIRAL
INTERACTION PARAMETERS BETWEEN IBUPROFEN AND P-CD.
Temp. -µo s.d. -µc s.d Ki s.d K1 s.d M1 s.d.
25°C 23.30 0.1 11 0.1 5256 740 10124 142 339 5
32°C 25.83 0.1 12.48 0.1 3550 253 6089 58 213 22
4û°C 28.72 0.1 14.48 0.1 2139 176 3692 33 112 3
50°C 32.30 0.1 18.07 0.1 1675 145 3011 21 78 2
%/K 1.5 2.0
i = 10 mM, mobilities (fto andµ,,) in 10·9 m2N.s
6.4.5 Determination of the average Ki of ibuprofen-{J-CD at different temperatures
The interaction between uncharged ibuprofen and P-CD was determined in a
BGE of 0.01 M sodium/acetate at pH 4.20, where both charged and uncharged forms
of the analyte are present. Once again, effective mobilities were measured at different
P-CD concentrations and temperatures applying an uncoated capillary with
mesityloxide as a neutral EOF marker for reference. Under these conditions, no
resolution was observed, in spite of the fact that the interaction is desionoselective [8].
The formation constant (K1) was determined sirnilar as in section 6.3.6.2,
applying eq. (6-4). Results obtained in this way are shown in Figure 6.9. Coefficients
of correlation were in all cases at least 0.999. Average K1 values and their standard.
deviations are also listed in TABLE 6.6. The values for K1, found in this study, are
higher than the Iiterature values at 37°C [8]. This was also the case with K2.
122
SIMULA TIONS AND THERMODYNAMICS
250
200
150
9 :e,
i:7< 100 " ~
50
0
-50 0.000 0.005 0.010 0.015 0.020 0.025
Figure 6.9 Determination of Kif or ibuprofen. Symbols as in Figure 6. 7. Explanation
see text. All experiments in duplicate.
6.4.6 The effect of temperature on selectivity
The formation constants between the non-charged ibuprofen and P-CD are
high and unequal for both optica] isomers [8], indicating desionoselective interaction.
In order to obtain chiral resolution, electroosmosis was suppressed by applying coated
capillary with the cathode on the inlet side [8]. In that case however, no EOF marker
for use as mobility reference could be observed. Therefore, the residual electroosmotic
mobility was calculated from the average experimental migration time and the average
effective mobility of ibuprofen, calculated from the data in T ABLE 6.6. Using this
information, the individual values of KJ for both optica] îsorners could be calculated
frorn theîr experimental migration times, assuming equal values of Ki for both isomers.
The difference between the individual values of K1 for both optica! isomers is listed as
'1K1 in T ABLE 6.6. The values of '1K1 are well outside the standard deviation of the
average KJ values, deterrnined previously. It can be seen that not only K1 and K1 but
also '1K1 decreases monotonously with increasing temperature.
123
CHAPTER6
6.4.7 Thermodynamic model/or K1 and Kz
Temperature dependence of equilibrium constants is usually modeled using a free
energy (.1G) relationship of the fonn [21]:
~ = exp(-LlG/Rî) (6-7)
with R the gas constant (8.314 J/mol.K) and T the absolute temperature. Using basic
thermodynamics, this can be rewritten using enthalpy (Ml) and entropy (LlS) changes
associated with the formation of the analyte-selector complex:
~ = exp(-MljRT+LlSJR) (6-8)
Experimentally, both Ml and LlS can be obtained from a so-called Van 't Hoff plot: the
logarithmic of ~ vs. l/T. This is only valid under the assumption that both Ml and LlS
are independent of temperature. Although there seem to be indications that this is not
always the case for ~-CD interactions [21], the data was processed under this
assumption. The results are shown in Figure 6.10 and TABLE 6.7. The negative sign
of Ml indicates an exothermic process, due to the release of high energy water out of
the cyclodextrin cavity. The negative sign of LlS indicates a decrease of entropy, due to
complex formation, which consequently results in a decrease of the degree of freedom
of the components involved in the interaction. As expected, the dominant force for
analyte binding arises from enthalpy changes (ILiHJ "'2x IT LlSI). The same was
concluded in Ref. [24]. From the results presented in this study, it was not possible to
assign enantioselectivity to either .1Ml or .1LlS since the error in both Ml and LlS is
higher than .1Ml or .1LlS, respectively.
Probably the main source of systematic errors arises from the temperature
difference between the untherrnostated part (first 4 cm) and the therrnostated part of
the capillary. This is clearly observed for the data points at 50°C, but a systematic error
in the Kc-determination at Jower temperatures cannot be excluded. However, this
systematic error will be largest at 50°C and almost absent at 25°C. The leftmost point
in both graphs (corresponding to 50°C) is considered an outlier.
The random error of Ml and LlS depends, among others, on the number of data
points used, i.e. the number of different temperatures applied for the deterrnination of
the formation constants. Since the Kc-determination at 50°C is considered an outlier,
only 3 data points are left. For obvious reasons, these data points are chosen in a
relatively small temperature range: 298K-323K. Therefore, random errors in Ml and
124
SIMULATIONS AND THERMODYNAMICS
L1S are relatively high, especially for L1S since this parameter is obtained through
extrapolation. Overall, accuracy and precision can be improved by increasing the
number of temperatures and the number of experiments, and by insuring that the
rnobilities are measured exclusively in the thermostated part of the capillary. The Jatter
can be achieved by the pressure mobilization method presented by Williams et al. [22] .
...!..
fl
l ~
1000
100
20'--~~~~"'--~~~~"'--~~~~"'--~~~--'
3.00 3.10 3.20
1ooorr [1/KJ
3.30 3.40
Figure 6.10 Van 't Hojf plots of the {3-CD formation constants K1 (and L1K1) and Kz with uncharged and charged ibuprofen respectively. Outliers at 50"C (the leftmost, solid points) were not included in calculating L1H and Af.
Values of L1H and L!S were calculated, together with their standard deviations
and tabulated in TABLE 6.7. No literature data on ibuprofen were available. The
values obtained in this study were somewhat higher thall literature values for other
analytes, possibly due to the fact that ibuprofen has very high stability constants with
P-CD.
Using circular dichroism spectropolarimetry, Han et al. [23] determined L1H
and L1S for interaction between ~-CD and 8 barbitals and found values around -20
kJ/mol and -10 J/molK respectively. In a liquid chromatography study with a P-CD
chiral stationary phase, Lipkowitz et al. (24] studied the enantioseparation of methyl
125
CHAPTER6
mandelate. Their values for .&:l were around -30 kJ/mol, but LWS values were 4
J/molK, 10 times higher than the values for ibuprofen shown in T ABLE 6. 7.
TABLE 6.7 THERMODYNAMIC PARAMETERS .&:l AND .1S FOR CHIRAL
INTERACTION OF IBUPROFEN AND ~-CD.
Value at 25°C .&:l s.d . .1S s.d.
[-] [kJ/mol] [J/molK]
Ku 9955 -52.08 1.6 -98.22 5.3
K1.2 10294 -52.25 1.4 -98.49 4.7
Ki 5256 -46.58 2.3 -84.89 7.5
Figure 6.11 Left ( a): contour plot for selectivity of enantioseparation of ibuprofen as
afunction oftemperature and {3-CD concentration at pH 4.47. Mesh size is 2.5°C and
0.001 M respectively. Right (b): concentration of {3-CD and analysis time requiredfor
a selectivity of 1.01, as a function of temperature, at pH 4.47. See text for further
details.
Effects of temperature on chiral resolution in CE were also measured
qualitatively by Guttrnan et al. [25], who observed a decrease in both resolution and
analysis time when increasing the temperature.
126
SIMULA TIONS AND THERMODYNAMICS
Using all data gathered, mobilities and selectivities can be calculated for any
combination of parameters. As an example Figure 6.11 a shows a contour plot of
selectivity vs. temperature and ~-CD concentration at pH 4.47. As expected,
selectivity increases with increasing ~-CD concentration and decreasing temperature.
When constructing the same contour plot for a higher pH value, for example pH 5.00,
the three dimensional surface is shifted down as far as 0.01 selectivity units, making
enantioseparation virtually impossible.
So far there seems to be no reason to increase the operating temperature above
25°C, unless analysis time is taken into account as well. Consider for example a fixed
selectivity of 1.01. In order to visualize a constant selectivity, the information
contained in the three dimensional plot of Figure 6.11 a is reduced to 2 dimensions (T
and [CD]) shaped as a horizontal cross section of the three dimensional figure. Such a
cross-section is shown as a dotted line in Figure 6.11 b. As expected, the CD
concentration, necessary to obtain a certain selectivity, increases strongly with
increased temperatures. Next, the effective mobility was calculated (eq. (2-19)) of the
slowest migrating isomer. The solid line in Figure 6.11 b shows the migration time
required to obtain a fixed selectivity of 1.01, applying a 300/370 mm coated capillary
at -25 kV, assuming µEoF = 0. Now it is visualized that, although temperature increase
has an adverse effect on the amount of ~-CD required, it might favor analysis time.
Temperature optimization can lead to a gain in the speed of analyses, which might be
favorable if the costs of the chiral selector are low (e.g., P-CD). The optimum
temperature is very much dependent on the required selectivity. Increasing the required
selectivity will result in a decrease of the optimum temperature.
Summarizing, when finally choosing a set of separation parameters, through
method development at room temperature, it seems certainly worthwhile to
subsequently try different temperatures, as illustrated in the experimental
electropherograms of Figure 6.12. This is especially easy since it requires simple
reprogramming of the analysis sequence in automated equipment.
127
CHAPTER6
0.012 .-----.---------------------,
O.oJO
50°C
0.008
s ~
~ 0.006 -ê Cl
li <
0.004
0.002
25°C
0.000 ._ _____ __... ______ __._ _____ __.
0 5 10 15
migration time [min]
Figure 6.12 Separation of a racemic mixture of ibuprofen in 0.01 M sodium/acetate,
pH 4.47 with 2.5 mM f3·CD in a coated capillary at different temperatures.
128
SlMULATIONS AND THERMODYNAMICS
6.5 Conclusions
The chiral parameters, mentioned in the first part of this chapter, together with
the original features of the steady state simulation software, provide a very flexîble
training tool for chiral separations in capillary electrophoresis. For example,
determination of the optimum chiral selector concentration can be done in seconds,
using the simulation software. Although the software is intended mainly for training
purposes, in cases where sufficient details of existing chiral selector-analyte
combinations are available, the program is also suitable for method development of
chiral separations in CE.
Specifically for the chiral separation parameters, it was observed that K values
decrease with increasing temperature, with negative values for both free energy and
entropy changes. This means that when optimizing a chiral separation, using the model
presented in chapter 2, a different operating temperature may lead to different
optimized conditions: selectivity wil! gencrally be lower at elevated temperatures. On
the other hand, temperature affects all mobilities as well. Consequently, increasing both
the temperature and the concentration of the chiral selector may result in shorter
analysis times, without affecting selectivity.
In conclusion one can observe that in principle temperature effects can be
predicted by extending existing models. In addition, changing the temperature may
sometimes be used to fine-tune separations, also in chiral CE applications.
129
CHAPTER6
References
J.C. Reijenga, J.H.P.A. Martens and F.M. Everaerts, Electrophoresis,
16 (1995) 2008
2 J.C. Reijenga, J. Chromatogr., 588 (1991) 217
3 J.C. Reijenga and M. Hutta, J. Chromatogr. A, 709 ( 1995) 21
4 J.C. Reijenga and E. Kenndler, J. Chromatogr. A, 659 (1994) 403
5 J.C. Reijenga and E. Kenndler, J. Chromatogr. A, 659 (1994) 417
6 H. Nishi and S. Terabe, J. Chromatogr. A, 694 (1995) 245
7 S. Fanali, J. Chromatogr., 545 (1991) 437
8 Y.Y. Rawjee, D.U. Staerk and G.Vigh, J. Chromatogr., 635 (1993) 291
9 Y.Y. Rawjee, R.L.Williams andG.Vigh, J. Chromatogr" 652 (1993) 233
10 F.M. Everaerts, J.L. Beckers and Th.P.E.M. Verheggen, Isotachophoresis,
J. Chromatogr. Library 6, Elsevier, Amsterdam, 1976
11 P. Bocek, M. Deml, P. Gebauer and V. Dolnik, Analytica! Isotachophoresis,
Electrophoresis Library 1, B.J. Radola (editor), VCH, Weinheim, 1988
12 R.A. Robinson, R.H. Stokes, Electrolyte Solutions, Butterworth, London,
1959
13 J.C. Reijenga, G.V.A. Aben, Th.P.E.M. Verheggen and F.M. Everaerts,
J. Chromatogr., 260 (1983) 241
14 B.A. Williams and G. Vigh, Anal. Chem., 68 (1996) 1174
15 W. Schützner, S. Fanali, A. Rizzi, E. Kenndler, J. Chromatogr. A,
719 (1996) 411
16 M.E. Biggin, R.L. Williams and G. Vigh, J. Chromatogr. A,
692 (1995) 319
17 A. Nardi, A. Eliseev, P. Bocek and S. Fanali, J. Chromatogr.,
638 (1993) 247
18 S. Fanali, J. Chromatogr" 545 (1991) 437
19 C.D. Herzfeldt and R. Kummel, Drug Dev. Ind. Pharm" 9 (1983) 767
20 W. Friedl, J.C. Reijenga and E. Kenndler, J. Chromatogr. A, 709 (1995) 163
21 J. Szejtli and T. Osa (editors), Comprehensive Supramolecular Chemistry,
volume 3, Elsevier, New York, 1996
22 B.A. Williams and G. Vigh, Anal. Chem" 68 (1996) 1174
23 S.M. Han and N. Purdie, Anal. Chem" 56 (1984) 2825
24 K.B. Lipkowitz and C.M Stoehr, Chirality, 8 (1996) 341
25 A. Guttman, A. Paulus, A.S. Cohen, N. Grinberg and B.L. Karger,
J. Chromatogr" 448 (1988) 41
130
APPLICA TIONS
7. APPLICABILITY OF CE IN CHIRAL SEPARATIONS
Abstract
This chapter shows the potential of CE for chiral analyses of various compounds in
different matrices. The applicability of CE for the detemzination of the optica/
isomers of some herbicidal compounds is shown. Furthemwre, it is demonstrated that
CE can be appliedfor the chiral analyses of pharmaceutical and clinical samples.
7.1 lntroduction
The introduction of automated equipment for capillary zone electrophoresis
(CZE), applying fused silica capillaries, in the early eighties has strongly increased the
number of applications of CZE. Chiral analysis has become one of the main areas of
interest, resulting in some excessive reviews, listing many applications and hundreds of
references [1-3]. The ease of method development might be considered as one of the
main advantages of CE over HPLC. This chapter discusses some applications of CE in
chiral separations. The applicability of ergot alkaloids, discussed in chapter 4, for the
chiral separation of some racemic herbicides is shown. It is shown that CE can be
applied for the determination of the optica! impurity of drugs. Furthermore, it is shown
that CE can be applied to determine the metabolism of thiopental and pentobarbital
enantiomers, by analyzing plasma samples.
7.2 Herbicides
7.2.1 lntroduction
Chloro-2-phenoxypropionic (Cl-APA) and halogen substituted 2-aryloxyphenoxy
propionic (APPA) acids, as well as N-benzoyl-N-(3-chloro-4-flurophenyl)amino
propionic acid (flamprop) are (structurally related to) herbicides. These compounds
have a stereocenter in position 2 of the propionic acid functional group. It has been
shown that the (R)-(-)-isomers of Cl-AP As and APPAs, and the (R)-( + )-isomer of
flamprop exhibit the strongest herbicîdal activity [4-6]. However, both optica] isomers
of these compounds are toxic [7], and their use should therefore be minimized.
131
CHAPTER7
Consequently, recent legislation in several European countries bas resulted in the
marketing of pure enantiomers. Analytica) methods are needed in order to determine
the optica! purity of these formulations. Liquid chromatography, using a Pirkle-type
chiral stationary phase (CSP) can be applied for the above purpose [5,7]. The capillary
electrophoretic separation of phenoxy acid herbicide enantiomers, applying a-CD and
DIME-~ as chiral selector, is shown by Nielen [8]. Recently, Padiglioni et al. showed
the enantioseparation of some herbicides applying a CSP derived from terguride [9}.
Earlier, this terguride CSP showed high selectivities for the enantiomers of
several organic acids [10]. Chapter 4 of this thesis describes the use of terguride and
structurally related compounds, as chiral selectors in CE. In this study, CE using the l -
allyl derivative of (5R,8S,10R)-terguride (allyl-TER) as chiral selector was applied for
the chiral separation of some herbicidal compounds. The same analytes as those used
in the HPLC-study [9] were chosen in order to make a fair comparison between both
separation techniques.
7.2.2 Experimental
A P/ACE 2200 (Beckman, Fullerton, CA) was used for all electrophoretic
experiments. The instrument used uncoated and polyacrylamide coated capillaries [11]
of 37 cm, with an effective length of 30 cm and an I.O. of 50 µm.. The UV-detector
was operated at 230 nm. The capillary cartridge was thermostated at 25°C.
~-Alanine and acetic acid were purchased from Merck (Darmstadt, Germany).
Allyl-TER was synthesized by the method published earlier [12]. Fluazifop (2-(4-{[5-
(trifluoromethyl)-2-pyridinyl]oxy }-phenoxy)propionic acid), halossifop (2-(4-{ [3-
chloro-5-(trifluoromethyl)-2-pyridinyl]oxy} phenoxy)propionic acid), fenoxaprop (2-
[4-( 6-chloro-2-benzoxazolyl)oxy ]phenoxypropionic acid) and flamprop (N-benzoyl-N
(3-chloro-4 fluorophenyl)-DL-alanine were donated by Mirko Flieger from the
Academy of Sciences of the Czech Republic (Prague, Czech Republic). The structure
of these compounds is shown in Figure 7 .1.
All samples were diluted in MeOH : H20 = 1 :5 to a concentration of 104 M
and injected hydrodynamically (5 seconds, 3. HY Pa). The BGE was prepared by
adjusting a 200 mM ~-alanine solution with acetic acid to pH 4.0. Subsequently, 1 part
of this electrolyte solution was diluted with 1 part of MeOH. This resulted in a BGE
consisting of 100 mM ~-alanine/acetate, 50% MeOH, pH* 5.3. The applied voltage
was 30kV.
132
APPLICA TIONS
Fluazifop Halossifop
Fenoxaprop Flamprop
Figure 7.1 Chemica[ structures of the examined herbicides.
7.2.3 Results and discussion
According to literaturc, a phcnoxy substituent at the a-position of propionic
acid decreases the pK.-value of the analyte from 4.9 to 3.1 [13]. Therefore, it can be
assumed that fluazifop, halossifop, and fenoxaprop have a high degree of dissociation,
whereas flamprop is assumed to have a relatively low degree of dissociation, at the
selected pH-value. This is confirmcd by the migration bchavior, in case no chiral
selector was added to the BGE. The phenoxy substitutcd analytes pass the detection
window well within 9 minutes in the order of their molecular mass (m) ( l si fluazifop; m
= fenoxaprop; m = 333.5, 31d halossifop; m = 361.5), whereas flamprop (m =
321.5) passes the detection window after approximately 12 minutcs
In order to separate the optica] isomers of the herbicidal analytes, the buffer
was supported with 25 m.V1 allyl-TER, which resulted in a slight increase of the pH of
the BGE. Before injecting the racemic analytes, the capillary was rinsed with BGE
containing allyl-TER. The in- and outlet consisted of pure BGE, without chiral
selector.
133
CHAPTER 7
IA
IB
20 ...
imp.
~ Û,_ H
l
8 JO
2A
28
3A
3B
l
12
time [min]
4
1
14 16
Figure 7.2 Electropherogram of the chiral separation of some herbicidal
compounds. JA, JB = fluazifop; 2A, 28 = halossifop,· JA, 3B = fenoxaprop; 4 =
flamprop; imp.= impurity. BGE: 100 mM {3-alanine/acetate, 50% MeOH, pH• 5.3,
supported with 25 mM allyl-TER. Separation voltage 30 kV. Coated capillary: 30-37
cm, l.D. 50 µm.
An electropherogram of the separation of the optica! isomers of the herbicidal
compounds is shown in Figure 7 .2. The enantiomers of the phenoxy substituted
propionic acids are well separated in approximately 13 minutes. No resolution was
observed for the flamprop enantiomers. A possible explanation of the limited
enantioselectivity of allyl-TER towards flamprop is the lower degree of dissociation of
this compound. In chapter 4 of this thesis, it is shown that only the dissociated acids
interact stereoselective with the chiral selector.
An impurity, originating from the fenoxaprop sample, is visible as two small
peaks after approximately 8 minutes. The impurity seems to be a racemate since only
one small peak is visible without the addition of the chiral selector to the BGE. lt is
possibly a degradation product of fenoxaprop: 2-( 4' -hydroxyphenoxy)propanoic acid.
Partial resolution of flamprop enantiomers can be obtained by increasing the
concentration of the chiral selector or by increasing the degree of dissociation of the
134
APPLICA TIONS
analyte. Therefore, a BGE was applied consisting of pure MeOH containing 100 mM
acetic acid and 50 mM triethanolamine (TEA), supported with 100 mM allyl-TER.
Approximately the first 28 cm of an uncoated capillary were filled with BGE
containing the chiral selector. The rest of the capillary, including the in- and the outlet
via) contained 100 mM acetate and 50 mM TEA in 100% MeOH. No rnigration of the
boundary between the zones with and without allyl-TER was observed under these
experimental conditions. Apparently, the electrophoretic mobility of the ergot alkaloid
is largely compensated by the residu al electroosmotic flow. The Jatter was reversed
due to the presence of TEA in the BGE. The resulting electropherograrn, applying 20
kV, is shown in Figure 7.3. Partial resolution is obtained for the enantiomers of
flamprop (R, 0.7), whereas high resolutions were obtained for the optica! isomers of
the other compounds.
-20 flamprop
E3' -40 < .§. hal ossi fop ... "' = ... i: -60 j <
fluazifop
-80
fenoxaprop
-100 i::====::::::::___:::i==:::_====---'------__J 10 12 14 16
time [min]
Figure 7.3 Electropherogram of the chiral separation of some herbicidal
compounds. BGE: 100 mM acetate, 50 mM TEA in 100% MeOH supported with 100
mM allyl-TER. Separation voltage 20 kV. Uncoated capillary: 30-37 cm, T.D. 50 µm.
Similar resolutions as shown in Figure 7.3 could be obtained applying HPLC
[9]. The selectivities obtained in the HPLC experiments (as defined by k1lk2) however,
were much higher than those obtained in the CE experiments (as defined by the ratio of
135
CHAPTER 7
the effective mobilities of the optica) isomers). Equal resolutions must be explained by
the much higher efficiencies, usually obtained in CE. The separation time in CE is
shorter than in HPLC; e.g., separation of the phenoxy substituted enantiomers takes
approximately 90 minutes, using the terguride packing in HPLC whereas only 15
minutes are needed applying CE with allyl-TER as chiral buffer additive (see Figure
7.3).
In this section it is shown that CE can be successfully applied for the separation
of the herbicidal optica! isomers. The method can be useful for the analysis of real
production samples and the determination of their enantiopurity.
7 .3 Quality control of fenfluramine enantiomers
7.3.1 lntroduction
Meta-fenflurarnine (Figure, 7.4, N-ethyl-a-methyl-(nHrifluorornethyl)-
phenethylamine) is a basic compound which has been applied extensively as an anti
obesity drug. The metabolism as well as the pharmacokinetic properties of this
compound are well documented. From these studies it was shown that the d-isomer is
the biologically active compound [14]. In the Netherlands, it is commercially available
both as a racemic mixture (Ponderal®) and as an optically pure d-isomer
(lsomeride®).
Figure 7.4 Chemical structure of meta
! enfluramine
136
The optica! purity of Isomeride was
investigated by Porrà et al. [15], using
capillary electrophoresis with TRIME-~ as
chiral selector. In this section, similar
experiments were performed to calculate the
optica! purity. Moreover, the chiral
parameters, mentioned in section 6.3.1.,
were determined. Using these parameters it
was possible to simulate the determination of
the chiral purity of m-fenfluramine [16]
APPLICATIONS
7.3.2 Experimental
Simulated and real experimental equipment conditions correspond to a usual
commercial CE instrument (P/ACE, Beckman, Fullerton, CA). Equipment conditions
were: coated capillary [11] length 400 mm to detection, 470 mm overall length, 50 µm
I.D" voltage 30 kV. Samples were injected hydrodynamically (3.3. l 03 Pa) during 5 s.
The instrument was thermostated at 25°C. The detection wavelength was 2 l 4 nm.
Ionic strength of the background electrolyte (BOE) was 10 mM. BGE's were prepared
with analytica] grade chemicals from the usual sources (Merck, Darmstadt, Germany;
SIGMA, St. Louis, MO). Heptakis-2,3,6-tri-O-methyl-P-cyclodextrin (TRIME-P) was
from Cyclolab (Budapest, Hungary). m-Fenfluramine as Ponderal® (racemate) and as
Isomeride®, both from Servier Technologie (Paris, France), were a kind gift of Mrs.
B.T. van Tuijl-Knipscheer (Blixembosch Pharmacy, Eindhoven). The concentration of
the sample solutions was 104 M. The Isomeride® sample, applied for the
determination of the optica! purity, had a concentration of 1 gil.
7.3.3 Results and discussion
First, the pK and mobility of m-fenfluramine were determined at 25°C applying
a buffer with an ionic strength of 10 mM. Different BGE's were prepared by adding
boric acid to a 10 mM sodium hydroxide solution. Mobilities were determined by
injection of mesityloxide as neutra! marker, according to the recently published dual
rnarker method by Williams et al. [14]. The inverse values of the resulting effective
mobilities were plotted vs. the hydroxide ion concentration in the pH range 8.6 to 9.4,
as described in section 6.3.6. L The pKa was determined as 10.1, significantly higher
than the literature value of 9. 1. The µ 0 value was 19.4.10·9 m2/Vs.
Subsequently, using an e-arninocaproic acid/acetate buffer of pH 5 with an
ionic strength of 10 mM, the effective mobilities of the optical isorners were
determined at different concentration levels of TRIME-P in the range 0-80 miVI. Under
these conditions, m-fenflurarnine was completely ionized, so that only K2 values were
relevant. K2 was determined as described in section 6.3.6.2. All measured mobilities
were corrected for viscosity effects, as described in section 6.3 .1.
Fînally, using 10 mM sodium/borate pH 9.46 as BOE, effective mobilities were
measured at different TRIME-~ concentrations. ünder these conditions, both the ionic
and the non-ionîc form of m-fenfluramine were present. From these experiments, K1
could be determined as described in section 6.3.6.2. All resulting chiral parameters are
presented in TABLE 7.1. The results indicate ionoselective interaction.
137
CHAP1ER 7
TABLE 7.1 EXPERIMENT AL VALUES OF ELECTROPHORETIC AND
CHIRAL PARAMETERS OF m-FENFLURAMINE (25°C, i = 10 mM).
parameter dimensions value standard deviation
µo m2Ns 19.38.10-9 0. 1.10-9
pK. 10.l 0.05
µc m2Ns 8.2.10-9 0. 1.10-9
M' 41% 0.5%
K1 56 2.1
AK1 % 0%
Kz 19.2 0.2
AK2 % 6.3% 0.1%
The separation of a typical Isomeride® sample was now simulated, using the
chiral parameter values in T ABLE 7 .1 as model input. The simulation is shown in
Figure 7.5b. Figure 7.5a shows the actual separation, under the same conditions. The
comparison is satisfactory, although selectivity is underestimated in the simulated
electropherogram. The optica! purity of the Isomeride® sample was calculated from
the ratio of the peak areas. Accordingly, an 2 % impurity of the 1-m-fenfluramine
isomer was found. This is in agreement with the data obtained in Ref. [15].
138
APPLICA TIONS
a
0 2 3 4 5 6 7 8
b
0 2 3 4 5 6 7 8
migration time [min]
Figure 7.5 Chiral impurity determination (a) of Isomeride. Coated capillary 400-470
mm, I.D. 50 µm. BGE: 100 mM /3-alanine!acetate, pH 4.2 supported with 40 mM
TRIME-/3. Separation voltage 30 kV. Simulated electropherogram in (b).
139
CHAP1ER 7
7 .4 Determination of thiopental enantiomers in plasma
7.4.1 Introduction
Thiopental (5-ethyl-5-( 1-methylbutyl)-2-thiobarbituric acid) is commonly
employed as an intravenous anesthetic agent, administered by injection for induction of
anesthesia or by continuous intravenous infusion for maintenance of anesthesia. High
doses, administered for several days, are used to decrease the intracranial pressure to
prevent the brains against the consequences of hypoxic ischaemia [ 17]. Thiopental is
marketed as a racemate although the S-(-)-isomer is more potent. The anesthetic
activity and acute toxicity of the optically pure enantiomers of thiopental and other
barbituric acid derivatives (e.g., pentobarbital) was studied in mice, and the S-(-)
isomer was found to be significantly more toxic and potent in all cases [18]. Possibly,
the behavior of the individual enantiomers in biologica] fluids is also different. In that
case, it is inappropriate to relate pharmacodynamic effects to racemic plasma
concentrations. Therefore, it was decided to determine the enantiomers of thiopental in
human plasma from a patient receiving the drug by intravenous infusion.
The analysis of plasma samples is comrnonly a difficult task since analytes are
dissolved in a very complex matrix. The use of CE for the separation of optical isomers
in plasma samples has been demonstrated by Prunonosa et al. [19] and D'Hulst and
Verbeke [20]. This section presents CE as an enantioselective method for the
determination of thiopental and its metabolite pentobarbital in human plasma.
7.4.2 Experimental
A PI ACE 2200 (Beckman, Fullerton, CA) was used for all electrophoretic
experiments. The instrument used a polyacrylamide coated capillary [11) of 57 cm,
with an effective length of 50 cm, and l.D. 50 µm. The capillary was thermostated at
20°C. Samples were injected for 20 seconds, applying 3.3.Jü3 Pa pressure. The
detection wave\ength was 214 nm to enable simultaneous measurements of both drug
and metabolite. The applied voltage was 25 kV.
TRIME-~ was purchased from Cyclolab (Budapest, Hungary). Thiopental
sodium was a kind gift from Eveline Wuis of the University Hospita! Nijmegen St.
Radboud (Nijmegen, the Netherlands). Pentobarbital, allobarbital, tris(hydroxymethyl)
arninomethane (TRIS), boric acid, diethylether and Extralut were purchased from
140
APPLICA TIONS
Merck (Darmstadt, Germany). The plasma samples were obtained from a patient
treated with high doses of thiopental for 160 hours.
The BGE consîsted of 100 m.\1 TRIS, adjusted to pH 9.0 with borîc acid. For
the chiral experiments, 40 mM TRIME-~ was added to the BGE. Only the capillary
was filled with BGE containing TRIME-~. The chiral selector was not present in the
in- or outlet via!.
Allobarbital was used as an internal standard (LS.). Pentobarbital and
thiopental were dissolved in demineralized water at a concentration of 1 mg free
acid/ml. Allobarbital was dissolved in a sodium hydroxide solution (pH 9) at the same
concentration. Reference plasma was prepared by the addition of thiopental and
pentobarbital solutions (5, 10, or 15 µI) and allobarbital solution (5 µl) to 500 µl blank
plasma. Equal amounts of the I.S. were added to all samples. A 500 µ! amount of the
reference samples and the real samples were pretreated applying a liquid-solid
extraction with laboratory-made Extralut columns. The samples were eluted with
approximately 10 ml of diethylether. The ether fraction was evaporated to dryness,
under a gentle stream of nitrogen. The residuals were dissolved in 100 µl
demineralized water.
7.4.3 Results and discussion
Allobarbital, thiopental, and pentobarbital (see Figure 7.6 for chemica!
structures) were well separable at pH 9.0 without the addition of TRIME-~.
allobarbital pentobarbital thiopental
Figure 7.6 Chemica/ structures of allobarbital, pentobarbital and thiopental.
Calibration graphs were obtained by plotting the ratio of the normalized peak areas of
thiopental and allobarbital or pentobarbital and allobarbital. The normalized peak area
is defined as the ratio of the peak area and the migration time. The regression
correlation coefficient of the calibration graph was 0.9993 for thiopental and 0.9995
141
CHAPTER 7
for pentobarbital. The recovery of the extraction procedure was determined by adding
the analytes to blank plasma. The recovery measured > 95% (s.d.= 8%) for thiopental,
pentobarbital, and allobarbital.
50 ~
t:. t:. t:. t:.
1 t:.
10 >-t:.
s = .t ij .J
~ t:. t:.
1~~~--·~~~---·~~~----·~~---·~~~--0 40 80 120 160 200
time [hours]
Figure 7.7 Racemic thiopental plasma level in a patient treated with continuous
intravenous thiopental. The infusion was stopped after 160 hours.
The racemic plasma concentration-time profile of thiopental in a patient is shown in
Figure 7.7. The measured concentration of pentobarbital in plasma never exceeded 6
mg/l.
The capillary electrophoretic chiral separation of thiopèntal and pentobarbital
was shown by Nishi et al. [21 ]. They applied a BGE supported with 30 mM y-CD, 50
mM sodiumdodecylsulphate (SDS), 60 mM 1-methoxyacetic acid and 40 mM d
camphor-10-sulphonate (d-cam) (final pH 9). Tanaka et al. [22) showed full separation
of the optica! isomers of pentobarbital applying a BGE at pH 9 su pported with 10 mM
of either TRIME-~ or TRIME-a.. In the current study, good separation of the
enantiomers of thiopental and pentobarbital could be obtained with a BGE supported
with 40 mM d-cam, 50 mM SDS and 40 mM y-CD at pH 8.9. The analysis time was
about 18 minutes. Similar results could be obtained using a BGE of 100 mM
TRIS/borate at pH 9.0 supported with 40 mM TRIME-B and coated capillaries. The
142
APPLICA TIONS
Jatter method was more straightforward and cheaper since only the capillary had to be
filled with the chiral selector. Therefore, this method was chosen to analyze the plasma
samples. As an example, Figure 7.8 shows the electrophoretic chiral separation ofboth
barbiturates in a plasma sample.
0.008
0.006
0 s 10
migration time [min]
tp
1
1\ pb
11 /\
CJuv 15
Figure 7.8 Electropherogram of a plasma sample at time = 162 hours (see Figure
7.7). ab allobarbital (1.S.) tp = thiopental enantiomers, pb = pentobarbital
enantiomers. BGE: 100 mM TRIS/borate pH 9.0 + 40 mM TRIME-/3. Coated
capillary 50-57 cm, 50 µm I.D" Separation voltage 25 kV. Detection 214 nm.
In all samples measured in this patient, with concentrations below 60 mg/I, no
difference in concentration of the enantiorners of both thiopental and pentobarbital was
found. This was confirmed by Nguyen et al.; they showed that the net uptake of
thiopental by all human body tissues is not stereoselective [23]. However, this does not
exclude the possibility of stereoselective metabolism or stereoselective uptake by
individual organs. Recently, Nguyen and Morgan showed that the tissue accumulation
of thiopental enantiomers in an isolated perfused rat heart was not stereoselective [24].
Current study shows that CE can be a suitable technique for the determination
of drugs in human plasma. The technique can be applîed to determine stereoselective
143
CHAPTER 7
drug metabolism. The limited sensitivity of CE might give rise to detection problems.
This problem could be partly resolved by concentrating the sample after extraction.
7.5 Conclusions
The results presented in this chapter show that CE can be a suitable technique
for the separation of optica] isomers in pharmaceutical preparations and in serum.
Moreover, it is shown that CE can be applied for the separation of herbicidal optica!
isomers. The latter method can be useful for the analysis and determination of the
enantiopurity of real production samples.
It can be concluded that CE can be applied for the separation of enantiomers,
even in difficult sample matrices. The limited sensitivity is the main drawback of the
technique. In this respect, CE is especially suitable for the analysis of production
samples and phannaceutical preparations.
References
H. Nishi and S. Terabe, J. Chromatogr. A, 694 (1995) 245
2 H. Nishi, J. Chromatogr. A, 735 (1996) 57
3 S. Fanali, J. Chromatogr. A, 735 (1996) 77
4 B. Blessington and N. Grabb, J. Chromatogr" 454 (1988) 450
5 W.A. Koning, D. lchein, T. Runge, B. Pfaffenberg, P. Ludwig and
H. Huhnerfuss, J. High Res. Chromatogr., 14 (1991) 530
6 D.W. Bewick, Pestic. Sci., 17 (1986) 349
7 B. Blessington, N. Crabb and J. O'Sullivan, J. Chromatogr., 396 (1987) 177
8 M.W.F. Nielen, J. Chromatogr" 637 (1993) 81
9 P. Padiglioni, C.M. Polcaro, S. Marchese, M. SinibaJdi and M. Flieger,
J. Chromatogr. A, 756 (1996) 119
10 M. Sinibaldi, M. Flieger, L. Cvak, A. Messina and A. Pichini,
J. Chromatogr. A, 666 (1994) 471
11 M.J. van der Schans, J.L. Beckers, M.C. Molling and P.M. Everaerts,
J. Chromatogr. A, 717 (1995) 139
12 A. Messina, A.M. Girelli, M. Flieger, P. Sedmera, M. Sinibaldi and L. Cvak,
Anal. Chem., 68 (1996) 1191
13 H.A. Sober (Ed.), Handbook of Biochemistry, znd ed" CRC, Cleveland, 1970
144
APPLICA TIONS
14 J.E. Myers, D.J. Buysse, M.E. Thase, J. Perel, J.M. Miewald, T.B. Cooper,
D.J. Kupfer and J.J. Mann, Biol. Psychiatry, 34 (1993) 753
15 K. Porrà, M.G. Quaglia and S. Fanali, Chromatographia, 41 (1994) 383
16 Jetse C. Reijenga, Benno Ingelse and Frans M. Everaerts, J. Chromatogr. A,
in press
17 H. Russo, F. Bressole, J. Brès and M-P. Duboin, Clin. Drug lnvest" 11 (l 996)
32
18 H.D. Christensen and 1. S. Lee, Toxicol. Appl. PharmacoL, 26 (1973) 495
19 J. Prunonosa, R. Obach, A. Diez-Coscon and L. Gouesclou, J. Chromatogr.,
574(1992) 127
20 A. D'Hulst and N. Verbeke, Chirality, 6 (1994) 225
21 H. Nishi, T. Fukuyama and S. Terabe, J. Chromatogr., 553 (1991) 503
22 M. Tanaka, S. Asano, M. Yoshinago, Y. Kawaguchi, T. Tedsumi and
T. Shono, Fresenius J. Anal. Chem" 339 (1991) 63
23 K.T. Nguyen, D.P. Stephens, M.J. McLeish, D.P. Crankshaw and
D.J. Morgan, Anesth. Analg., 83 (1996) 552
24 K.T. Nguyen and DJ. Morgan, Chirality, 8 (1996) 477
145
CHAP'IER 7
146
ABS1RACT
ABSTRACT Enantiomers have identical physical and chemical properties, in an isotropic
environment. lts chirality is only observed when the molecule is subjected to a chiral
influence. Consequently, chiral analyses cannot be executed by separation techniques
based upon differences in above mentioned properties.
Capillary electrophoresis (CE) is a highly efficient analytica! separation
technique. Separation is based upon differences in mobility, which mainly depends on
the charge to mass ratio of the molecule. The most common ways to influence
mobilities are ehanging the pH of the buffer or adding an appropriate complexing agent
to the buffer (Chapter 1).
The mobilities of optica! isomers are identical, irrespective of the pH-value of
the buffer. However, the addition of an enantioselective complexing agent, or chiral
selector, can change the charge to mass ratio of the enantiomers in a selective way.
This can result in a mobility difference between the optica! isomers which is a
prerequisite for (chiral) separation. Optica! isomers can be successfully separated if the
mobility of the complex differs from the mobility of the free enantiomers and if the
complex stability of the diastereomeric complexes is not identical.
The stability of complexes between enantiomers and chiral selectors is
expressed by their formation constants. Acidic and basic compounds can exist as
charged or neutra! molecules. The stability of the resulting complexes with the chiral
selector is generally not alike. Moreover, formation constants can be identical for the
two charged enantiomers, and different for the non-charged enantiomers, or vice versa.
Consequently, the pH is major parameter to optimize chiral separation. The optimum
concentration of the chiral selector depends on the magnitude of the formation
constants. The value of the formation constants can be obtained by measuring
the mobility of the enantiomers at different concentrations of the chiral selector
( Chapter 2).
Cyclodextrins are the most applied chiral selectors in CE. The separation of the
enantiomers of several basic compounds of pharmaceutîcal interest was investigated by
CE, employing a soluble neutra] ~-cyclodextrin polymer. Both selectivity and
resolution were influenced by the concentration of the ~-cyclodextrin polymer. Also, it
is shown that increasing the ionic strength of the background electrolyte could lead to
increased resolution. The addition of different organic additives to the background
electrolyte (BGE) generally resulted in a decrease of resolution ( Chapter 3).
Ergot alkaloids are a novel group of chiral selectors. Stereoselectivities of
several ergot alkaloids towards a number of racemic acidic compounds were
compared. The effects of pH and MeOH added to the background e!ectrolyte (BGE)
147
ABSTRACT
were investigated. Low pH proved to have an adverse effect on enantioseparation,
indicating ionoselective complex formation. These observations were confirmed by
determining the formation constants with the dissociated and the non-dissociated acid.
The addition of 50% MeOH to the BGE altered stereoselectivity and increased the
solubility of the chiraJ selector. The optica! isomers of all test compounds could be
baseline resolved ( Chapter 4).
The influence of the nature of the buffer on the enantiomeric separation of
some sulfonamides was studied. For this purpose, the enantioselectivity and the
complex stability of a range of native and modified cyclodextrins in different electrolyte
systerns was examined. Results showed that the nature of the co-migrating buffer anion
may significantly influence the magnitude of equilibrium constants, depending on the
type of modification of a specific cyclodextrin. It is shown that the use of benzoate in
the BGE generally has an adverse effect on chiral interaction due to competitive
inclusion in the cyclodextrin cavity. This may also strongly influence the optimum
cyclodextrin concentration fora particular separation (Chapter 5).
Detailed knowledge of all parameters influencing the mobility of an optica!
isomer makes it possible to simulate chiral separations, applying a steady-state
simulation program. The performance of the program is illustrated with simulations,
using chiral parameters obtained from literature and from own experiments. The
program can be used as an aid for method development and provides a flexible training
tool for chiraJ separations in CE. In order to gain more insight in the effect of
temperature on chiral separations, free enthalpy and entropy changes were determined
for the interaction between P-cyclodextrin and ibuprofen enantiomers. For the chosen
example, it is shown that an increase of the temperature always results in an increase
of the optimum concentration of p-cyclodextrin, resulting in maximum selectivity
(Chapter 6).
The potential of CE for chiraJ analyses of various compounds in different
matrices is illustrated in three examples. Firstly, the separation of the optica! isomers of
sorne herbicidal compounds is shown, applying the 1-allyl derivative of terguride as
chiral selector. Secondly, CE is applied to determine the optical purity of a
pharrnaceutical formulation. Finally, it is shown that CE can be applied for the
determination of drug enantiomers in human plasma ( Chapter 7).
148
SAMENVAmNG
SAMENVATTING Enantiomeren hebben identieke fysische en chemische eigenschappen in een
isotropische omgeving. Hun chiraliteit kan slechts geobserveerd worden wanneer het
molecule in een chirale omgeving geplaatst wordt. Ten gevolge hiervan kunnen chirale
analyses niet uitgevoerd worden met scheidingstechnieken die gebaseerd zijn op
verschillen in de bovengenoemde eigenschappen.
Capillaire electroforese is een efficiënte scheidingstechniek. De scheiding is
gebaseerd op verschillen in mobiliteit. De mobiliteit hangt voornamelijk af van de
lading-massa verhouding van het molecuul. Gewoonlijk wordt de mobiliteit beïnvloed
door verandering van de pH van het achtergrondelectrolyt of door geschikte complex
vormers toe te voegen aan de buffer. (Hoofdstuk 1).
De mobiliteit van enantiomeren is identiek, ongeacht de pH van de buffer.
Echter, de toevoeging van een enantioselectieve complexvormer, ofwel .chirale
selector, kan de lading-massa verhouding van de enantiomeren selectief veranderen.
Dit kan vervolgens leiden tot een mobiliteitsverschil tussen de optische isomeren,
hetgeen een eerste vereiste is voor een (chirale) scheiding. Enantiomeren kunnen
succesvol gescheiden worden indien de mobiliteit van het gevormde complex verschilt
van de mobiliteit van de vrije enantiomeren en indien de stabiliteit van beide
diastereomere complexen verschillend is.
De stabiliteit van een complex tussen een enantiomeer enerzijds en een chirale
selector anderzijds wordt uitgedrukt door de formatieconstante. Zure en basische
componenten kunnen zowel in geladen als in ongeladen vorm voorkomen. De
stabiliteit van de resulterende complexen is over het algemeen verschillend. Bovendien
kunnen de formatieconstanten identiek zijn voor de twee geladen isomeren (geen
selectiviteit) en verschillend voor de ongeladen isomeren (wel selectiviteit), en vice
versa. De pH is dientengevolge een belangrijke parameter om de scheiding te
optimaliseren. De optimale concentratie van de chirale selector hangt af van de
stabiliteit van de gevormde complexen. De waarde van de formatieconstante kan
verkregen worden door de mobiliteit van het enantiomeer te meten bij verschillende
concentraties van de chirale selector (Hoofdstuk 2).
Cyclodextrines zijn de meest toegepaste chirale selectoren in CE. In Hoofdstuk
3 is de electroforetîsche scheiding van de enantiomeren van verschillende basische
farmaceutica onderzocht, met gebruikmaking van een neutraal (3-cyclodextrine
polymeer. De resultaten laten zien dat zowel de selectiviteit als de resolutie worden
beïnvloed door de concentratie van het (3-cyclodextrine polymeer. Bovendien wordt
aangetoond dat een toename van de ionsterkte van de buffer kan leiden tot een
149
SAMENV A TIING
toename van de resolutie. De toevoeging van organische oplosmiddelen aan de buffer
leidt in deze studie over het algemeen tot een afname van de resolutie.
Ergot alkaloïden vormen een nieuwe groep van chirale selectoren in CE. De
stereoselectiviteit van een aantal ergot alkaloïden ten aanzien van enkele racemische
zuren is vergeleken in Hoofdstuk 4. Tevens is onderzocht wat het effect op de
stereoselectiviteit is van de pH van de buffer en van de toevoeging van methanol aan
de buffer. Aangetoond is dat een lage pH een nadelig effect heeft op de chirale
scheiding. Dit duidt op ionoselectieve complexvorming. Deze observatie wordt
bevestigd door middel van het meten van de formatieconstanten van het ergot
alkaloïde met het gedissocieerde en het niet-gedissocieerde zuur. De stereoselectiviteit
wordt beïnvloed door het toevoegen van 50% methanol aan de buffer. De toevoeging
van methanol resulteert bovendien in een betere oplosbaarheid van de chirale selector.
Het blijkt mogelijk om de optische isomeren van alle testcomponenten te scheiden.
De invloed van de aard van de buffer op de chirale scheiding van enkele
sulfonamiden is onderzocht in Hoofdstuk 5. Hiertoe is de selectiviteit en de complex
stabiliteit van enkele natuurlijke en enkele gemodificeerde cyclodextrines bestudeerd,
in verschillende buffer systemen. De resultaten tonen aan dat, afhankelijk van de
modificatie van de cyclodextrine, de aard van het co-migrerende ion een belangrijke
invloed kan hebben op de grootte van de formatieconstante. Het gebruik van benzoaat
blijkt over het algemeen een negatief effect te hebben op de chirale interactie, vanwege
competieve complexvorming. Dit laatste kan de optimale cyclodextrine concentratie
voor een specifieke scheiding sterk beïnvloeden.
Gedetailleerde kennis van alle parameters die de mobiliteit van optische
isomeren beïnvloeden, maakt het mogelijk om chirale scheidingen te simuleren, gebruik
makend van een 'steady state' simulatie programma. In Hoofdstuk 6 is de werking van
dit programma ge11lustreerd. Hiertoe zijn chirale parameters gebruikt die verkregen zijn
uit de literatuur en uit eigen experimenten. Het programma kan gebruikt worden als
hulpmiddel bij methode-ontwikkeling, en biedt bovendien een flexibel
trainingsgereedschap voor chirale scheidingen in CE. Om meer inzicht te krijgen in het
effect van temperatuur op chirale scheidingen werden de vrije enthalpie- en
entropieveranderingen gemeten van de interactie tussen ~-cyclodextrine en de
enantiomeren van ibuprofen. Voor het gekozen voorbeeld is aangetoond dat een
temperatuurtoename altijd resulteert in een toename van de optimale cyclodextrine
concentratie welke resulteert in maximale stereoselectiviteit. Temperatuur
optimalisatie kan gebruikt worden om de scheidingstijd te minimaliseren.
De mogelijkheden van CE om chirale analyses uit te voeren in verschillende
matrices worden gei1lustreerd in Hoofdstuk 7. Allereerst wordt de scheiding getoond
150
SAMENV A TIING
van de optische isomeren van enkele herbiciden. Hierbij wordt 1-allylterguride
toegepast als chirale selector. Vervolgens is CE toegepast voor de bepaling van de
optische zuiverheid van een farmaceutisch preparaat. Tenslotte wordt aangetoond dat
CE toegepast kan worden voor de bepaling van drug enantiomeren in plasma.
151
SAMENV A TilNG
152
SYMBOLS AND ABBREVIA TIONS
SYMBOLS
a1, a2, a3 constants [-]
ARJs selectivity [-]
c concentration [mo1.r1]
D diffusion coefficient [m2.s-1]
E electric field strength [V.m-1]
F Faraday constant [C.mor1]
Fe1 electric force [N]
F1 friction force [N]
L1G Gibbs free energy change [J.mor1]
Af{ enthalpy change [J.mor 1]
ionic strength [moJ.r1]
Ka, Kb acid/base dissociation constant [-]
K1, K1, Kc equilibrium constant of complex formation [-]
ld effective capillary Iength [m]
l, total capillary length [m]
m molecular mass [g.mor1]
M' relative complex mobility [-]
N theoretica! plate number [-]
q net charge [C]
r radius [m]
regression correlation coefficient [-]
R gas constant [J.mol1 K 1]
R, resolution [-]
s selectivity [-]
SF separation factor [-]
L1S entropy change [J.mor1 .K1]
fm migration time [s]
T temperature [K] or [0 C]
v velocity [m.s- 1]
v voltage [V]
z charge number [-]
Greek
a degree of dissociation [-]
E dielectric constant [C2.r1.m-1]
1J viscosity [N.s.m-2]
153
SYMBOLS AND ABBREVIA TIONS
(j
Subscript
HR,HS HRCD,HSCD
R,S R,S
RB RCD
RCff, SCff
Superscript
eff
154
electrophoretic mobility
mobility at infinite dilution
mobility of fully dissociated acid/base
apparent mobility
complex mobility
electroosmotic mobility
effective mobility
mobility of free enantiomer
variance
zeta potential
non dissociated acidic enantiomer
[m2. y-1.s-1]
[m2.v-1.s-1]
[m2.v-1.s-1]
[m2.v-1.8-1]
[m2.v-1.s-1]
[m2.v-1.s-1]
[m2.v-1.s-1]
[m2.v-1_8-1]
[s] or [m]
[V]
complex between cyclodextrin and non dissociated acidic
optical isomer
complex between cyclodextrin and protonated basic
optical isomer
enantiomer
dissociated acidic enantiomer
rear boundary
complex between cyclodextrin and non protonated basic
optica! isomer
complex between cyclodextrin and dissociated acid enantiomer
refers to enantiomer interacting with chiral selector
effective
ABBREVIATIONS allyl-TER
AMP-TER
a.u.
BGE
BS CD
CE
CSP
CZE
DIME-~
EOF
EP-P-CD
eq.
HPLC
HP-P-CD
LD.
ITP
MEKC(MECC)
MeOH
MES
MOPS
MBS
MPS
NMR
opt.
PA
Ref.
s.d.
SDS
TBA
TBE
TEE
TME
TRIS
uv ZE
SYMBOLS AND ABBREVIATIONS
1-allyl-( +)-(5R, 8S, 1 OR)-terguride
1-(3' -aminopropy 1)-( + )-( 5R, 8S, lOR)-terguride
arbitrary units
background electrolyte
2-butylsulfonamide
cyclodextrin
capillary electrophoresis
chiral stationary phase
capillary zone electrophoresis
heptakis(2,6-di-O-methyl)-P-cyclodextrin
electroosmotic flow
soluble neutra! P-cyclodextrin polymer
equivalent
high performance liquid chromatography
2-hydroxypropylated P-cyclodextrin
capillary inner diameter
isotachophotesis
micellar electrokinetic (capillary) chromatography
methanol
2-(N-morpholino )ethanesulphonic acid
morpholinopropanesulphonic acid
2-methylbutylsulfonamide
P-methylphenethylsulfonamide
nuclear magnetic resonance
optimum
polyacrylamide
reference
standard deviation
sodium dodecylsulphate
tetrabutylammonium
tryptophan butyl ester
tryptophan ethyl ester
tryptophan methyl ester
tris(hydroxymethyl)aminomethane
ultra violet
zone electrophoresis
155
SYMBOLS AND ABBREVIATIONS
156
DANKWOORD
DANKWOORD Veel mensen hebben bijgedragen aan de totstandkoming van dit proefschrift.
Natuurlijk ben ik dank verschuldigd aan Frans Everaerts, mijn eerste promotor, die
altijd bereid was mij te helpen en altijd het onmogelijke gedaan lijkt te kunnen krijgen.
Van zeer groot belang was ook de inzet van Jetse Reijenga van wie ik heel veel geleerd
heb en met wie ik erg prettig heb samengewerkt. Ik dank Salvatore Fanali, die mij
heeft ingewijd op dit vakgebied en Mirko Flieger met wie ik een heel plezierige
samenwerking gehad heb. Joost van Dongen wil ik bedanken omdat hij altijd bereid
was te helpen. Ook Henk Claessens heeft mij erg geholpen en ik bewaar goede
herinneringen aan onze samenwerking met Sjoerd van der Wal en Lucien Duchateau.
Verder wil ik Eveline Wuis bedanken voor de zeer plezierige samenwerking en voor
het corrigeren van een gedeelte van dit proefschrift. Ik bedank ook Ernst Kenndler en
Karim Sarmini en natuurlijk Jurai Sevcik voor hun bijdrage aan gemeenschappelijke
projecten. Ook wil ik Beckman Instruments (Nederland) B.V. bedanken voor het
beschikbaar stellen van apparatuur en voor de financiële ondersteuning van het
drukken van dit proefschrift. Heel belangrijk waren ook de discussies met mijn
vrienden en collega's. Vaak kwamen goede ideeën 's avonds Iaat in de bar van de AOR
tot stand. Zij hebben, inclusief alle hierboven genoemde mensen, ervoor gezorgd dat ik
een hele leuke tijd gehad heb. Speciaal de mensen die mijn proefschrift kritisch hebben
doorgenomen ben ik zeer erkentelijk. Tenslotte wil ik nog graag mijn ouders bedanken.
Zij hebben me altijd gesteund en hebben altijd interesse getoond in mijn activiteiten.
157
DANKWOORD
158
CURRICULUM VITAE
CURRICULUM VITAE
Benno Allard Ingelse werd geboren op 2 augustus 1968 te Breda. In 1986 behaalde hij
het atheneum-B diploma aan het Willem van Oranje college in Waalwijk. Aansluitend
hierop begon hij met de studie Scheikundige Technologie aan de Technische
Universiteit Eindhoven. Het afstudeeronderzoek "Klonering en expressie van Horse
Liver Alcohol Dehydrogenase in E. coli" werd verricht bij de vakgroep Organische
Chemie onder leiding van prof. E. :\1eijer. In 1992 behaalde hij het ingenieursexamen.
Vanaf 1992 is hij werkzaam geweest in de vakgroep Instrumentele Analyse aan de
Technische Universiteit Eindhoven. Hier werd het promotieonderzoek uitgevoerd
onder leiding van prof.dr.ir. F.M. Everaerts. In dat kader werd in 1993 een aantal
maanden onderzoek gedaan aan het Istituto di Cromatografia del CNR in Rome, onder
begeleiding van Salvatore Fanali. De resultaten van het promotieonderzoek zijn
beschreven in dit proefschrift.
159
CURRICULUM VITAE
160
BIBLIOGRAPHY
BIBLIOGRAPHY
Enantiomeric separation by capillary electrophoresis using a soluble neutra] ~-cyclodextrin
polymer, B.A. Jngelse, F.M. Everaerts, C. Desiderio and S. Fanali, Joumal of
Chromatography A, 709 (1995) 89-98.
A further study on the chiral separation power of a soluble neutral ~-cyclodextrin polymer,
B.A. Inge/se, F.M. Everae11s, J. Sevcik, Z. Stransky and S. Fanali. Journal of High
Resolution Chromatography, 18 (1995) 348-352.
Capillary electrophoretic enantioseparation of selegiline, methamphetamine and ephedrine
using a neutra! ~-cyclodextrin epichlorhydrin polymer, J. Sevcik, Z. Stransky, B.A. lngelse
and K. Lemr, Joumal of Pharmaceutical and Biomedical Analysis, 14 (1996) 1089-1094.
Ergot alkaloids as novel chiral selectors in capillary electrophoresis, B.A. lngelse, J.C.
Reijenga, H.A. Claessens and F.M. Everaerts, Journal of High Resolution
Chromatography, 19 (1996) 225-226.
Influence of the nature of the buffer on chiral separation in capillary electrophoresis, B.A.
Jngelse, H.A. Claessens, Sj. van der Wal, A.L.L. Duchateau and F.M. Everaerts, Joumal
ofChromatography A, 745 (1996) 61-71.
Ergot alkaloids as chiral selectors in CE. Determination of the separation mechanism,
B.A. /ngelse, M. Flieger, H.A. Claessens, F.M. Everaerts, Journal of Chromatography
A, 755 (1996) 251-260.
Training software for chiral separations in CE, J.C. Reijenga, B.A. Ingelse and F.M.
Everaerts, Joumal of Chromatography A, in press.
Reversed determination of the formation constants of 1-allyl terguride with mandelic acid
optical isomers using CE, B.A. Inge/se, J.C. Reijenga and F.M. Everaerts, Joumal of
Chromatography A, in press.
Chiral interactions in capillary zone electrophoresis. Computer sirnulation and comparison
with experiment. B.A. lngelse. K. Sannini, J. C Reijenga, E. Kenruller and F.M. Everaerts,
in press
161
BIBLIOGRAPHY
Thermodynamics of chiral interaction in CE. Separation of ibuprofen enantiorners with · ~
CD. J.C. Reijenga, B.A. Inge/se and F.M. Everaerts, submitted for publication in Joumal
of Chromatography A ·
Determination of herbicidal enantiorners by CE using ergot alkaloids, B.A. Inge/se, J.C.
Reijenga and F.M. Everaerts, manuscript in preparation
162
STELLINGEN
1. Het overnemen van pK.-waarden voor zure ofbasische componenten,
zoals gevonden in de literatuur, kan grote fouten introduceren bij de bepaling
van de stabiliteitsconstante van complexen waarbij dergelijke componenten
betrokken zijn.
Dit proefschrift, Hoofdstuk 6
2. Het wekt wellicht verbazing dat de effectieve mobiliteit van een ion kan
toenemen door interactie met een neutrale complex-vormer.
F. Wang andMG. Khaledi,Anal. Chem" 68 (1994) 3460
3. Het succes van 'combinatorial chemistry' is voor een belangrijk deel toe te
schrijven aan de doorbraak van massaspectrometrische technieken als
electrospray en MALDI-TOF.
4. Onvoldoende kennis van elementaire basischemie kan leiden tot overschatting
van het belang van de pH op de selectiviteit van de scheiding van sterke ionen
in capillaire electroforese.
C-E. Lin, C-C. Chang, W-C. Lin, E.C. Lin, J. Chromatogr. A, 753 (1996) 133
5. Het gebruik van de temperatuur als scheidingsparameter in capillaire
electroforese, verdient meer aandacht.
Dit proefschrift, Hoofdstuk 3 en 6
6. Sommige analytisch chemische tijdschriften hebben klaarblijkelijk niet de
deskundigheid in huis om manuscripten over numerieke simulaties op niveau te
beoordelen.
C. Schafer-Nielsen, Electrophoresis, 16 (1995) 1369
7. Het gebrek aan TBS-plaatsen in Nederland zorgt ervoor dat misdaad loont.
8. Een artiest als Serrano doet de kunstconsument hevig verlangen naar "eenvoud
verlichte wateren" van Lucebert.
9. Slechte films kunnen nog altijd waardering krijgen middels het predikaat cult.
10. Kennis over het privéleven van topsporters lijkt een grotere prioriteit te hebben
bij sportverslaggevers als kennis van de sport zelf
11. Zelfs in wetenschappelijke publicaties valt een niet-wetenschappelijke huiver te
bespeuren voor het gebruik van 'beter als'-constructies.
12. De grondigheid waarmee Frankrijk zijn nationale belangen nastreeft staat in
schril contrast met zijn Franse slag methode ten aanzien van de bescherming
van de internationale mensenrechten.
Stellingen behorende bij het proefschrift: "Chiral separations using capillary
electrophoresis".
Eindhaven, 19junil997
Benno A. Inge/se