Enantiomeric analysis of pharmaceutical compounds by Ion/molecule reactions

Article abstract of DOI:10.1021/ac001135q

Protonated complexes involving cyclodextrin hosts and guest compounds that are pharmacologically important are produced in the gas phase and reacted with a gaseous amine. The guest is exchanged to produce a new protonated complex with the amine. The reaction is enantioselective and is used to develop a method for determining enantiomeric excess using only mass spectrometry. The pharmaceutical compounds include DOPA, amphetamine, ephedrine, and penicillamine. The presence of more than one reacting species is observed with DOPA and penicillamine. Molecular dynamics calculations are used to understand the nature of the interactions and the possible source of the variations in the reactivities.

Full text of DOI:10.1021/ac001135q

Anal. Chem. 2001, 73, 1684-1691
En a n t io m e ric An a lys is o f P h a rm a c e u t ic a l
Co m p o u n d s b y Io n /Mo le c u le Re a c t io n s
Ga brie la Grigore a n a nd Ca rlito B. Le brilla *
Department of Chemistry, University of California, Davis, California 95616
tides, and crown ethers.1
4-20
The resulting complexes are diaster-
P rotonated complexes involving cyclodextrin hosts and
guest compounds that are pharmacologically important
are produced in the gas phase and reacted with a gaseous
amine. The guest is exchanged to produce a new proto-
nated complex with the amine. The reaction is enantiose-
lective and is used to develop a method for determining
enantiomeric excess using only mass spectrometry. The
pharm aceutical com pounds include DOP A, am phet-
amine, ephedrine, and penicillamine. The presence of
more than one reacting species is observed with DOP A
and penicillamine. Molecular dynamics calculations are
used to understand the nature of the interactions and the
possible source of the variations in the reactivities.
eomeric with either unique ionization efficiencies or fragmentation
patterns that allow enantioselectivity based on the relative
abundances of the respective peaks. A recent example includes
the chiral differentiation of amino acids by the collision-induced
dissociation of protonated trimers composed of the amino acid
and two selector molecules, which are derivatized amino acids
(
N-tert-butoxycarbonylphenylalanine, N-tert-butoxycarbonylproline,
and N-tert-butoxycarbonyl-O-benzylserine).21 The same method
was used to quantify enantiomeric excess in mixtures.22
Metal complexes with chiral coanalytes have also been used
to produce diastereomeric complexes that were probed by
collision-induced dissociation. These analyses were performed by
monitoring the relative abundances of the dissociation products
that varied with the chirality of the analyte. Complexes of cobalt23,24
and copper25,26 have been used in this manner. Recent examples
with the copper complexes show considerable promise as a
method for quantification.26
The determination of enantiomeric excess in mixtures of chiral
drugs and pharmaceutically important compounds is commonly
performed with high-performance liquid chromatography (HPLC)1
and, increasingly, capillary electrophoresis (CE).4,5 However, the
determination of enantiomeric excess strictly by mass spectrom-
etry has several potentially attractive features over existing
techniques. The duty cycle is fast compared to chromatographic
methods. A mass spectrometric method does not necessarily
require derivatization and is capable of high sensitivities. More-
over, mass spectrometry provides structural confirmation that is
important in the analysis of complex, heterogeneous mixtures.
Methods that employ mass spectrometry specifically for the
analyses of enantiomers often require chiral coanalytes to produce
stereomeric complexes that have either unique ionization efficien-
cies or unique fragmentation behavior. Coanalytes that have been
-3
Ion-molecule reactions in the gas phase provide an alternative
method for chiral recognition and quantification. Enantioselectivity
has been achieved with the use of crown ethers19,20,27 and
(
12) Gong, S.; Camara, E.; He, F.; Green, M. K.; Lebrilla, C. B. Int. J. Mass
Spectrom. Ion Processes 1 9 9 9 , 185/ 186/ 187, 401-412.
13) Vekey, K.; Czira, G. Anal. Chem. 1 9 9 7 , 69, 1700-1705.
(14) Sawada, M.; Takai, Y.; Yamada, H.; Kaneda, T.; Kamada, K.; Mizooku, T.;
Hirose, K.; Tobe, Y.; Naemura, K. Chem. Commun. 1 9 9 4 , 2497-2498.
15) Sawada, M.; Okumura, Y.; Yamada, H.; Takai, Y.; Takahashi, S.; Kaneda,
(
(
T.; Hirose, K.; Misumi, S. Org. Mass Spectrom. 1 9 9 3 , 28, 1525-1528.
(16) Sawada, M.; Takai, Y.; Yamada, H.; Hirayama, S.; Kaneda, T.; Tanaka, T.;
Kamada, K.; Mizooku, T.; Takeuchi, S.; Ueno, K.; Hirose, K.; Tobe, Y.;
Naemura, K. J. Am. Chem. Soc. 1 9 9 5 , 117, 7726-7736.
17) Sawada, M.; Okumura, Y.; Shizuma, M.; Takai, Y.; Hidaka, Y.; Yamada, H.;
Tanaka, T.; Kaneda, T.; Hirose, K.; Misumi, S.; Takashi, S. J. Am. Chem.
Soc. 1 9 9 3 , 115, 7381-7388.
(
used in this manner include, but have not been limited to, alkyl
tartrates,6-8 cyclodextrins,9,10 proteins,11,12 amino acids13 and pep-
(
18) Sawada, M.; Shizuma, M.; Takai, Y.; Yamada, H.; Kaneda, T.; Hanafusa, T.
J. Am. Chem. Soc. 1 9 9 2 , 114, 4405-4406.
(
(
1) G u¨ bitz, G. Chromatographia 1 9 9 0 , 30, 555-564.
2) Ameyibor, E.; Stewart, J. T. J. Liq. Chromatogr., Relat. Technol. 1 9 9 7 , 20,
(19) Chu, I. H.; Dearden, D. V.; Bradshaw, J. S.; Huszthy, P.; Izatt, R. M. J. Am.
Chem. Soc. 1 9 9 3 , 115, 4318-4320.
8
55-869.
(20) Dearden, D. V.; Dejsupa, C.; Liang, Y. J.; Bradshaw, J. S.; Izatt, R. M. J. Am.
(
(
(
(
(
3) Tang, Y. Chirality 1 9 9 6 , 8, 136-142.
Chem. Soc. 1 9 9 7 , 119, 353-359.
(21) Yao, Z. P.; Wan, T. S. M.; Kwong, K. P.; Che, C. T. Anal. Chem. 2 0 0 0 , 72,
5383-5393.
4) Blaschke, G.; Chankvetadze, B. J. Chromatogr., A 2 0 0 0 , 875, 3-25.
5) Fillet, M.; Hubert, P.; Crommen, J. J. Chromatogr., A 2 0 0 0 , 875, 123-134.
6) Fales, H. M.; Wright, G. W. J. Am. Chem. Soc. 1 9 7 7 , 99, 2339-2340.
7) Nikolaev, E. N.; Goginashvily, G. T.; Talrose, V. L.; Kostjanovsky, R. G. Int.
J. Mass Spectrom. Ion Processes 1 9 8 8 , 86, 249-252.
(22) Yao, Z. P.; Wan, T. S. M.; Kwong, K. P.; Che, C. T. Anal. Chem. 2 0 0 0 , 72,
5394-5401.
(23) Hofmeister, G.; Leary, J. A. Org. Mass Spectrom. 1 9 9 1 , 26, 811-812.
(24) Dang, T. T.; Pedersen, S. F.; Leary, J. A. J. Am. Soc. Mass Spectrom. 1 9 9 4 ,
5, 452-459.
(
8) Denisov, E. N.; Shustryakov, V.; Nikolaev, E. N.; Winkler, F. J.; Medina, R.
Int. J. Mass Spectrom. Ion Processes 1 9 9 9 , 183, 357-368.
(
9) Ramirez, J.; He, F.; Lebrilla, C. B. J. Am. Chem. Soc. 1 9 9 8 , 120, 7387-
(25) Tao, W. A.; Zhang, D.; Wang, F.; Thomas, P. D.; Cooks, R. G. Anal. Chem.
7
388.
10) Grigorean, G.; Ramirez, J.; Ahn, S. H.; Lebrilla, C. B. Anal. Chem. 2 0 0 0 ,
2, 4275-4281.
11) Camara, E.; Green, M. K.; Penn, S. G.; Lebrilla, C. B. J. Am. Chem. Soc.
9 9 6 , 118, 8751-8752.
1 9 9 9 , 71, 4427-4429.
(26) Tao, W. A.; Zhang, D.; Nikolaev, E. N.; Cooks, R. G. J. Am. Chem. Soc. 20 00,
122, 10598-10609.
(
(
7
(27) Liang, Y.; Bradshaw, J. S.; Izatt, R. M.; Pope, R. M.; Dearden, D. V. Int. J.
1
Mass Spectrom. 1 9 9 9 , 185/ 186/ 187, 977-988.
1684 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001
10.1021/ac001135q CCC: $20.00 © 2001 American Chemical Society
Published on Web 03/16/2001

Ch a rt 1 . P h a rm a c e u t ic a lly Ac t ive Co m p o u n d s
Us e d fo r En a n t io s e le c t ive Gu e s t Ex c h a n g e
Re a c t io n s
the method of Ciucanu and Kerek.30 All alkylamines used in the
study including n-propylamine, (R)-1-amino-2-propanol, ethylene-
diamine, and 1,3-diaminopropane were obtained from Aldrich.
Exchange Reactions. All the experiments were performed
on a home-built external source electrospray FTMS instrument
equipped with a 5.2-T superconducting magnet described in earlier
publications.28,31 The hosts and analytes were dissolved in a 1:1
2
H O/ MeOH solution. The complexes were then prepared by
mixing the 0.01 M host solution with the 0.01 M drug analyte
solution, in a 1:100 ratio, respectively. To introduce solution into
the FTMS, a microspray setup was used. The solution was pumped
through a stainless steel tube, 0.02 in. i.d., 3 in. long, attached to
a Hamilton syringe on one end and the capillary needle at the
other end. The needle was made of fused-silica capillary tubing,
25 µm i.d., 150 µm O.D. The flow rate used was 12 µL/ h. A voltage
of 1.5-2.5 kV was applied to create the charged spray, which then
drifted into a heated stainless steel capillary inside the source.
The alkylamine was purified in the vacuum manifold with several
freeze-thaw cycles. It was introduced into the analyzer chamber
using a variable leak valve. The pressure of the amine was between
9
cyclodextrins. In this report, an enantiomeric analysis method is
developed based on the differences in the reaction rates of gas-
phase host-guest complexes. A diastereomeric complex [HO:A
H] of an oligosaccharide host (HO) with a chiral analyte guest
A) reacts with an alkylamine (B) to produce a guest exchange
product ([HO:B + H]+ ) illustrated in reaction 1.
+
-7
-7
+
1 × 10 and 6 × 10 Torr. Although attempts were made to
32
(
calibrate the ion gauge using efficiency factors, large variations
between the “true” pressure and the recorded pressures are
expected. The variation in the pressures is the source of most of
the error in the absolute rate constants. Although these errors
cancel in the selectivity values, it is suggested that the absolute
rates are treated as those obtained from an uncalibrated ion gauge.
+
+
[
HO:A + H] + B h [HO:B + H] + A
(1)
The reaction is effectively a proton-transfer process mediated
by a host molecule. This reaction has been studied extensively
with amino acids.28 The “three-point interaction” construct was
used to understand the nature of the enantioselectivity. It has also
been shown that the formation of gas-phase inclusion complexes
plays a prominent role in the reaction.29 In this report, we expand
this work and illustrate that the same reaction can be used to
develop a method for enantiomeric determinations of pharma-
ceutical compounds.
Four commonly used chiral drugs were examined:
-DOPA, l- and d-amphetamine (AMP), l- and d-ephedrine (EPH),
and - and -penicillamine (PEN) (Chart 1). DOPA in the l-form
is used for the treatment of Parkinson’s disease. Amphetamine
in the l-form is a central nervous system stimulant. Ephedrine,
also in the l-form, is used as brochodilator and its derivative,
pseudoephedrine, is a decongestant in the d-form. Penicillamine
is used as an antirheumatic. All have a single chiral center except
for the ephedrine, which has two.
+
To obtain the rate constants, the reactant [HO:A + H] and
+
product [HO:B + H] peaks were monitored as a function of time.
Several rf bursts were used to eliminate unwanted masses such
as those belonging to complexes of Na+ and K+ along with any
residual [HO:B + H]+ peaks present at the beginning of the
exchange reaction.
Molecular Modeling. The cyclodextrin and the protonated
analyte structures were constructed and separately optimized fully
using the Insight II builder module. The protonated oligosaccha-
ride-analyte complexes were then formed by merging the
respective oligosaccharide hosts and the analyte guests. Two sets
of calculations were performed corresponding to two types of
initial structures. In one set, the analytes were placed inside the
cavity to produce initially inclusion complexes. In a second set,
the analytes were placed on the outer wall of the host. In both
cases, calculations of the complexes were started with fully
optimized oligosaccharide host and amino acid structures. During
the simulation, the structures of both the analytes and the hosts
were allowed to fully optimize. The simulation was performed by
heating the complexes to 600 K for 400 ps. At 8-ps intervals, a
structure from the trajectory was captured and annealed in steps
of 100 K to 0 K. This resulted in 50 annealing simulations with a
corresponding number of structures. Although only one structure
is shown in each case, all the structures within 5 kcal/ mol of the
lowest energy structure were examined. Unless indicated, these
structures were found to share the same structural features as
the lowest.
L- and
D
L
D
EXPERIMENTAL SECTION
Materials. The pharmaceutical analytes 3,4-dihydroxyphenyl-
alanine (DOPA), amphetamine, and ephedrine were purchased
from Sigma Chemical Co. (St. Louis, MO). Penicillamine was
purchased from Aldrich (Milwaukee, WI). All were used without
further purification. The hosts employed were permethylated
â-cyclodextrin (heptakis(2,3,6-tri-O-methyl)-â-cyclodextrin or CD),
purchased from Sigma, and permethylated maltoheptaose (or
HEP), synthesized in this laboratory from the parent (Sigma) via
(
(
28) Ahn, S.; Ramirez, J.; Grigorean, G.; Lebrilla, C. B. J. Am. Soc. Mass Spectrom.,
in press.
29) Ramirez, J.; Ahn, S.; Grigorean, G.; Lebrilla, C. B. J. Am. Chem. Soc. 2 0 0 0 ,
(30) Ciucanu, I.; Kerek, F. Carbohydr. Res. 1 9 8 4 , 131, 209-217.
(31) Carroll, J. A.; Penn, S. G.; Fannin, S. T.; Wu, J.; Cancilla, M. T.; Green, M.
K.; Lebrilla, C. B. Anal. Chem. 1 9 9 6 , 68, 1798-1804.
1
22, 6884-6890.
(32) Bartmess, J. E.; Georgiadis, R. M. Vacuum 1 9 8 3 , 33, 149-153.
Analytical Chemistry, Vol. 73, No. 8, April 15, 2001 1685

Ta b le 1 . S e le c t ivit ie s fo r t h e CD Ho s t (k ×
Ta b le 2 . S e le c t ivit ie s fo r t h e HEP Ho s t (k ×
-
1 1
3
-1 1
3
1
0
c m /m o le c u le ‚s )
1 0
c m /m o le c u le ‚s )
n-propyl-
amine
(R)-1-amino-
2-propanol
ethylene-
diamine
1,3-diamino-
propane
n-propyl-
amine
(R)-1-amino- ethylene- 1,3-diamino-
2-propanol
diamine
propane
DOPA
0.0047
DOPA
<10-
<10-
15
0.0121
0.0024
4.98
fast: 0.122
slow: 0.031
fast: 0.131
slow: 0.014
fast: 0.93
kL
kD
kL/ kD
<10
-15
0.00982
0.00805
1.22
kL
kD
-
15
<10
15
<10-15
Amphetamine
kL/ kD
kl
kd
kl/ kd
2.22
2.47
0.90
slow: 2.19
Amphetamine
kl
kd
kl/ kd
0.40
0.27
1.46
1.78
1.34
1.33
Ephedrine
1.58
2.02
k(+)
k(-)
Ephedrine
0.53
k(+)/ k(-)
0.78
k(+)
k(-)
k(+)/ k(-)
0.031
Penicillamine
0.64
0.83
kL
kD
4.48
fast: 4.15
slow: 0.42
fast: 1.07
slow: 1.85
Penicillamine
kL
kD
3.4
kL/ kD
fast: 1.80
slow: 0.55
fast: 1.9
slow: 6.18
kL/ kD
RESULTS
A. Rate Constants. Enantioselectivity (S) is defined in this
discussion by the ratio of the rate constants (S ) k / k ). For
convenience, both and l, and similarly and d, will be grouped
L
D
L
D
together in defining selectivity, with the understanding that the
nature of the two sets of nomenclatures is significantly different.
An S value of 1.0 indicates no chiral differentiation between the
enantiomers in question. For the amino acids reported in previous
studies, the values were as large as 5.9,28,29 In some cases, the
ratios were less than 1, which indicated that the D-enantiomers
were favored. It is not yet possible to predict which enantiomer
will be favored in the guest-exchange reaction, albeit for the amino
acids the
L
typically gave the largest rate constants.9,29 Molecular
modeling calculations give qualitative indications when the
selectivity or its inverse will be large based on differences in the
interaction between the host and the guest. However, molecular
modeling also cannot yet be used to predict whether the selectivity
is greater or less than unity. Tables 1 and 2 list the selectivities
of the drugs discussed in this paper.
Amphetamine. To obtain the rate constants, the relative
intensities of each pure enantiomer were monitored throughout
an appropriate length of the reaction period. Figure 1 illustrates
the mass spectra of the reaction mixture at various reacting times.
The series of mass spectra in Figure 1 reflect the differences in
reaction rates for the l- (left spectra) and d- (right) forms of
amphetamine. Time zero was set to the end of the ejection pulse
that eliminated the product peak [CD:NPA + H]+ resulting from
the initial reaction of the drug complex. Therefore, the mass
spectrum at time zero contained primarily the complex ion [CD:
Fig u re 1 . ESI-FTMS spectra of a solution containing â-cyclodextrin
and amphetamines at various reaction times. The complex [CD:
+
Amp+H] is isolated and allowed to react with a background pressure
-
7
of n-propylamine of 3.2 × 10 Torr. This reaction has a relatively
low selectivity corresponding to S ) 1.46.
A rate plot for the protonated amphetamine-cyclodextrin
+
complex [CD:AMP + H] reacting with n-propylamine is shown
+
2
AMP + H] . As the reaction progressed from 9 to 49 s, the [CD:
in Figure 2. For both plots, r ) 0.999. From the rate curves, the
+
AMP + H] reactant peak (m/ z 1565) decreased; the rate of this
rate constants for the two enantiomers were found to be k
L
) 0.40
-
11
3
-11
3
decrease differed between the l- and d-forms, with the latter having
its reactant peak exhausted at a slower rate. When the reactant
peak was exhausted to an intensity of ∼5% of the product peak,
data collection was stopped to ensure accuracy of the intensity
values.
× 10
cm / molecule‚s and k
D
) 0.27 × 10 cm / molecule‚s
giving a selectivity S ) 1.46.
The selectivities of this compound was further examined with
other bases and other host molecules. When a more basic
compound was used for the exchange reaction, such as 1-amino-
1686 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

Fig u re 3 . ESI-FTMS spectra under three reaction times of a solution
+
containing â-cyclodextrin and DOPA. The complex [CD:DOPA+H]
is reacted with 1,3-diaminopropane (DAP) (4.0 × 10 Torr).
-
7
HEP host. The selectivities varied considerably between the two
hosts with a value of 5.0 for CD and 1.22 for HEP. These values
contrast to the selectivity of the phenylalanine, which exhibited a
higher selectivity with the linear HEP host (4.6) than with the
CD host (0.83).
Fig u re 2 . Rate plot of the reaction illustrated in Figure 1. For both
enantiomers, r ) 0.999. (a) d-AMP. (b) l-AMP.
2
2
-propanol, the reactivity of both isomers increased but the
The series of mass spectra in Figure 3 from the reaction of
1,3-diaminopropane and [CD:DOPA+H]+ illustrate the spectral
features of an analyte with relatively large selectivities. The
resulting kinetic plot yielded a break in the curve suggesting the
presence of at least two reactive species in the ion population
selectivity decreased to 1.33. The selectivity reversed when a linear
oligosaccharide (permethylated maltoheptaose or HEP) was used
as host (Table 2).
DOP A. The addition of the hydroxyl group on the phenyl rings
significantly decreased the rate of exchange for DOPA. Exchange
reactions performed with the monoamine bases n-propylamine and
(Figure 4). Both the
L- and the D-forms have fast and slow
components that differed by an order of magnitude. For example,
-
13
3
1
-amino-2-propanol were extremely slow compared to phenylala-
the
L
-form has values of kL,fast ) 12.2 × 10 cm / molecule‚s and
-
11
3
29
-13 3
nine (k ) 10
cm / molecule‚s ). The reaction of the [CD:
kL,slow ) 3.1 × 10 cm / molecule‚s.
+
DOPA+H] complex with n-propylamine was not conducive to
Ephedrine. This compound provides a good contrast to the
obtaining the reaction rate constants indicating that the rate
others because it has two chiral centers. Exchange reactions with
1-amino-2-propanol were performed with both hosts, CD and HEP.
The kinetic plots showed the usual linear behavior with rate
constants that are slightly slower than amphetamine. However,
increasing the number of chiral centers did not apparently increase
the selectivity; poor enantioselectivities were obtained with values
close to 1. For the host â-CD, S equaled 0.83 and for HEP, S
equaled 0.78.
constants were less than 10- cm / molecule‚s. For the reaction
15
3
of the complex with 1-amino-2-propanol, only the
L-form of the
drug exchanged within a time frame where obtaining the reaction
-
13
cm3/
rate constant was still feasible, with k
L
) 0.47 × 10
molecule‚s. Not surprisingly, gas-phase basicity played a role in
the rate of the gas-phase exchange reaction. For this reason, more
basic compounds such as the diamines like 1,3-diaminopropane
(
GB 987 versus 889 kJ/ mol for n-propylamine) were used for the
P enicillamine. The complex undergoes relatively fast ex-
changes, even faster than amphetamine. The kinetic plot of the
guest exchange reactions.
+
The reactions of the complexes of DOPA [CD:DOPA+H] and
complex yielded the usual straight line for the
L-isomer, but the
+
[
HEP:DOPA+H] with ethylenediamine yielded the usual linear
D-isomer showed the two-component behavior when the complex
[CD:PEN+H]+ was reacted with n-propylamine (Figure 5a). The
2
rate plots with r values of 0.999 (data not shown). The reactivity
of the
the exchange reaction rate is k
for the CD host and k
) 0.98 × 10
L
-isomer was nearly of the same magnitude for both hosts;
L
-enantiomer of the drug has a rate constant k
L
) 3.40 × 10-11.
-
13
3
L
) 1.21 × 10 cm / molecule‚s
The D-enantiomer displayed a two-component curve, with kD,fast
-
13
3
-11
-11
L
cm / molecule‚s for the
) 1.88 × 10 and kD,slow ) 0.55 × 10 (Figure 5b). This behavior
Analytical Chemistry, Vol. 73, No. 8, April 15, 2001 1687

Fig u re 4 . Kinetic plot for the reaction of 1,3-diaminopropane with
CD:DOPA+H]+ (Figure 3). Both enantiomers have at least two
reactive components, fast and slow. The rate constant of the slow
reaction is significantly smaller than the fast reaction. (a) L-DOPA.
Fig u re 5 . Rate plots for the reaction of protonated penicillamine
complexed to â-cyclodextrin with n-propylamine. The L-enantiomer
has a linear behavior while the D-enantiomer shows a break in the
plot corresponding to two reacting species. (a) L-PEN. (b) D-PEN.
[
(b) D-DOPA.
mixtures are obtained at the chosen reaction time. We have looked
at the reaction rates as the enantiomer composition changed, and
as expected, the combined reaction rate increases as the percent-
age of the slower enantiomer increases in the analyte mixture.
From the mass spectra, a calibration curve is constructed. The
is unique and is not observed with any other pharmaceutical
compounds or with the amino acids. It suggests not only
differences in the interactions between each enantiomer and the
host but also that the
produce two distinct complexes with different reactivities. The
ratio k / kD,fast ) 1.9, which is similar to the 2.2 value for cysteines
its structural analogue. The ratio involving the slow reaction is
even larger with k
D-enantiomer may interact with the host to
plot has as the ordinate I/ I
sum of the reactant peak intensity and product peak intensity.
The abscissa is / ( ), the -enantiomer mole fraction.
0 0
, with I as product peak and I the
L
D
D
+
L
D
L
/ kD,slow ) 6.2.
The quality of the calibration curve is dependent on the
selectivity value, with larger scatter, hence largest error, as S
approaches unity. For example, Figure 6 shows a curve for
B. Quantitative Analyses of Chiral Drugs by Mass Spec-
trometry. To illustrate the analytical application of the guest
exchange reaction to pharmaceutical compounds, calibration
curves were constructed. To construct calibration curves, a known
set of enantiomeric mixtures is analyzed, beginning with the one
pure enantiomer and ending with the other. The procedure
involves first choosing the proper conditions for the exchange
reaction. Parameters include the alkylamine reagent, its pressure,
and a reaction time. The optimum reaction time for analysis is
the shortest reaction period that produces a spectrum with
sufficiently large differences in relative intensities. A known analyte
is then analyzed in the following manner. The enantiomeric
composition of the known analyte is varied, beginning with one
pure component (100:0) to include several compositions such as
2
amphetamine with an r value of 0.971. To produce this curve,
n-propylamine was used as the reagent gas and cyclodextrin was
used as the host with a reaction time of 39 s. The selectivity of
this system was S ) 1.46.
With a larger selectivity value of 2.19 (cyclodextrin host and
,3-diaminopropane as reagent gas) DOPA yielded r2 ) 0.996
1
(
Figure 7). To produce this calibration curve, a reaction time of
8
8 s was used. Apparently, even with two reacting species it is
possible to construct a valid calibration curve.
DISCUSSION
Chiral differentiation in the gas phase is governed by the three-
80:20, 65:35, 50:50, 35:65, 20:80, and 0:100. Mass spectra for these
point interaction. This model, illustrated in Chart 2, requires three
1688 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

Ch a rt 2 . Th re e -P o in t In t e ra c t io n In vo lvin g t h e
An a lyt e a n d t h e Ho s t (As t e ris k )
illustrates the nature of the interaction (Figure 8). Both enanti-
omers interact similarly with the ammonium interacting with the
lower rim while the phenyl group is constrained by the inner
cavity. This interaction represents the second interaction while
the interaction of the methyl with the inner cavity represents the
third. Thus, amphetamine has a single attractive interaction and
two steric or repulsive interactions. Studies with amino acids
indicate that two attractive interactions provide the largest
selectivity while three and one attractive interactions decrease the
selectivity.28
Fig u re 6 . Calibration curve for a compound, amphetamine, with a
low selectivity (S ) 1.46). The reactant complex is composed of
protonated amphetamine and â-cyclodextrin and reacted with n-
2
propylamine. The value of r is 0.971.
The guest exchange reaction of DOPA is unique for several
reasons. The reaction is extremely slow, compared to compounds
with similar structures such as tyrosine,28,29 requiring strongly
basic amines to perform the exchange reaction. The chiral
selectivity is large, with the cyclodextrin host compared to
structural analogues such as tyrosine (S ) 0.67) and phenylalanine
(
S ) 0.82).10 The two rate constants for each enantiomer further
suggest at least two reacting species. In the complex, this may
be inferred as two species in which the analyte interacts with the
host by different arrangements. DOPA effectively interacts with
three points of attraction. Ion/ dipole and hydrogen-bonding
interactions occur with the protonated ammonium group. The
carboxylic acid and the hydroxyl groups on the phenyl undergo
hydrogen-bonding interactions. Studies with amino acids such as
threonine and methionine indicate that a three-point attraction
does not necessarily increase the selectivity.29 The results of the
molecular modeling of the DOPA complex are summarized in
Figure 9. The upper structures are the result of calculations where
the initial structures involved inclusion of the analyte. The two
lower structures were initiated with the DOPA near the outer wall
of the cyclodextrin cavity. The results indicate that molecule
migrates into the cyclodextrin during the annealing cycles and
further suggest that inclusion is the preferred state of interaction.2
The upper structures in Figure 9 are similar in appearance.
The carboxylic and the ammonium groups in both enantiomers
interact with the lower rim. Both functional groups are constrained
by the phenyl group, which in turn is constrained by the cavity.
We also find that the hydroxyl group interacts with the upper rim.
The similarities in the interaction of the two enantiomers were
also observed with Phe and Tyr, both of which have low
selectivities. We posit that these structures represent the fast-
reacting components as these are predicted to have low selectivi-
ties (Sfast ) 0.93) (Table 1). The other structures (lower) exhibit
large differences in their interactions and may represent the slowly
reacting components, which have a larger selectivity (Sslow ) 2.19).
Ephedrine contains two chiral centers, but the presence of a
second chiral center does not apparently increase selectivity. The
9
Fig u re 7 . Calibration curve for a compound with moderate selectivity
(
2.19). The reactant complex, protonated DOPA and â-cyclodextrin,
2
is reacted with 1,3-diaminopropane. The value of r is 0.996.
interacting points between the host and the guest as the
prerequisite for enantioselectivity. Although this construct is
commonly used in the solution phase, its application in the gas
phase to predict the extent of enantioselectivity has recently been
reported.28
The low selectivity of amphetamine is explained by the three-
point interaction. In the gas-phase complex, amphetamine is
protonated on the amine group. The ammonium group interacts
positively with the methoxyl groups, specifically the narrow rim,
which is composed of the carbon-6 of glucose. The molecular
modeling performed on the protonated complex [CD:AMP+H]+
Analytical Chemistry, Vol. 73, No. 8, April 15, 2001 1689

+
Fig u re 8 . Lowest energy structures produced by molecular modeling of the complex [CD:AMP+H] . The l-enantiomer is shown on the left.
The heat of formation (kcal/mol) is relative to the lowest energy structure.
+
Fig u re 9 . Lowest energy structures produced by molecular modeling of the complex [CD:DOPA+H] . The two top structures were the results
of calculations where the initial geometry was the inclusion complex. The two bottom structures were initiated with the analyte on the outer
cavity of the host.
selectivity of ephedrine is poor with both CD (S ) 0.83) and HEP
S ) 0.78). Taking into account the interactions of the protonated
interactions is between one and two, which would be expected to
yield poor enantioselectivity.
(
analyte with the host, we find one attraction due to the protonated
amine. However, this attraction may be sterically weakened
because the group is a secondary amine. Similarly, the second
attractive interaction due to the hydroxyl may also be attenuated
by the presence of the bulky phenyl group. The third interaction
is repulsive due to the methyl group. The number of net attractive
Penicillamine exhibits good selectivity because it has two
strong attractive interactions. The S-H group is not expected to
interact strongly with the ether groups on the host. The presence
of two reacting species, as evidenced by the break in the kinetic
+
plot of the cyclodextrin complex ([CD:PEN+H] ) reacting with
n-propylamine, is present but only for the
D-enantiomer. The
1690 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

+
Fig u re 1 0 . Lowest energy structures produced by molecular modeling of the complex [CD:PEN+H] . The two top structures were the results
of calculations where the initial geometry was the inclusion complex. The two bottom structures were initiated with the analyte on the outer
cavity of the host.
molecular modeling results are summarized in Figure 10 with the
four most stable structures of the penicillamine complex beginning
with the included structure (upper structures) and the nonin-
cluded structures (lower). The D-isomer shows two structures with
similar energies but with one fully included and the other partially
included with the molecule protruding through the lower rim. The
The prerequisite for this method is that the analyte contains an
amine. The interaction of the ammonium group with the host
provides the stability for observing the host-guest complex in
the gas phase. As with the amino acids, we find that the formation
of inclusion complex is highly favorable and the complex is the
most stable gas-phase species.
L-isomer yields two structures, but the inclusion complex is
ACKNOWLEDGMENT
significantly more stable than the noninclusion structure. We posit
that the high-energy structure converts to the low-energy structure
during ionization so that only a single species is observed for the
Funding provided by the National Science Foundation is
gratefully acknowledged. The authors also thank Seong Hee Ahn
for providing the permethylated maltoheptaose.
L-form while two species are observed for the
D-form.
Received for review September 22, 2000. Accepted
February 5, 2001.
CONCLUSION
The determination of enantiomeric excess by employing ion/
molecule reactions and strictly mass spectrometry is illustrated.
AC001135Q
Analytical Chemistry, Vol. 73, No. 8, April 15, 2001 1691