Immobilization of malarial (Plasmodium falciparum) dihydrofolate reductase for the selection of tight-binding inhibitors from combinatorial library

Article abstract of DOI:10.1021/ac070215s

A simple procedure for selection of tight-binding inhibitors of mutant dihydrofolate reductases from Plasmodium falciparum (PfDHFRs) based on preferential binding to the enzyme immobilized on a Sepharose column has been described. PfDHFRs with a cysteine residue at the C-terminal have been prepared in order to immobilize to a thiopropyl-Sepharose gel via S-S linkage. The amount of immobilized DHFRs was estimated to be 4-5 mg/g of dried gel, and the activities of bound DHFRs were comparable to that of free enzymes. The prepared immobilized enzyme has been used for the selection of tight-binding inhibitors from combinatorial libraries, based on the affinities of each ligand with the enzyme. Free ligands were then identified and analyzed quantitatively by high-performance liquid chromatography-mass spectrometry, and the components with high binding affinity of the library could thus be realized. Results could be confirmed by quantitative analysis of the bound ligands released from the enzyme by guanidine hydrochloride treatment.

Full text of DOI:10.1021/ac070215s

Anal. Chem. 2007, 79, 5006-5012
Immobilization of Malarial (Plasmodium
falciparum) Dihydrofolate Reductase for the
Selection of Tight-Binding Inhibitors from
Combinatorial Library
Chawanee Thongpanchang,* Supannee Taweechai, Sumalee Kamchonwongpaisan,
Yongyuth Yuthavong, and Yodhathai Thebtaranonth
National Center for Genetic Engineering and Biotechnology, National Science and Technology Development Agency,
113 Thailand Science Park, Pahonyothin Road, Klong 1, Klongluang, Pathumtani 12120, Thailand
Based on modeling of wild type and mutant PfDHFRs, a number
of inhibitors were synthesized and screened against the enzymes
individually and found to be effective against both wild type and
some mutant PfDHFRs.10-13
A simple procedure for selection of tight-binding inhibitors
of mutant dihydrofolate reductases from Plasmodium
falciparum (PfDHFRs) based on preferential binding to
the enzyme immobilized on a Sepharose column has been
described. PfDHFRs with a cysteine residue at the C-
terminal have been prepared in order to immobilize to a
thiopropyl-Sepharose gel via S-S linkage. The amount of
immobilized DHFRs was estimated to be 4-5 mg/g of
dried gel, and the activities of bound DHFRs were
comparable to that of free enzymes. The prepared im-
mobilized enzyme has been used for the selection of tight-
binding inhibitors from combinatorial libraries, based on
the affinities of each ligand with the enzyme. Free ligands
were then identified and analyzed quantitatively by high-
performance liquid chromatography-mass spectrometry,
and the components with high binding affinity of the
library could thus be realized. Results could be confirmed
by quantitative analysis of the bound ligands released from
the enzyme by guanidine hydrochloride treatment.
The advent of combinatorial chemistry techniques offers the
means for rapid generation of a large number of structurally
related compounds. However, identification of potential leads from
the combinatorial libraries required an effective selection system
that can single out the tight-binding inhibitors for further
characterization and development. Recently, Kamchonwongpaisan
et al. have developed an ultrafiltration method for the selection,
based on the dissociation constant of the enzyme with each
inhibitor.14 In the stoichiometric selection, the limitation of this
technique is its dependence on the solubility of the target enzyme,
which affected the compound concentration in the libraries: i.e.,
the larger the size of the library, the lower the concentration of
each inhibitor.14
It has been reported that immobilized enzymes are useful tools
in many areas such as for enzyme sensors, enzyme reactors, or
enzyme catalysis in organic synthesis.15-19 Dihydrofolate reductase
enzymes were also immobilized onto the solid support in different
Plasmodium falciparum dihydrofolate reductase-thymidylate
synthase (PfDHFR-TS) is a validated target of antifolate antima-
larial drugs1-3 such as pyrimethamine (Pry) and cycloguanil (Cyc),
which are commonly used clinically for the treatment of malaria
infection. Since the emergence of the resistant strain of malaria
parasites to these inhibitors became widespread, there has been
an urgent need to search for new drugs to combat the resistant
malaria. The molecular mechanism for resistance to PfDHFR
inhibitors has been shown to be due to point mutation at various
sites of the parasite gene sequence, leading to the decrease in
binding affinity of inhibitor to the enzyme. Mutation of one or
more residues at amino acid positions 16, 51, 59, 108, and 164 of
PfDHFR were identified to be involved in antifolate resistance.4-9
(6) Foote, S. J.; Galatis, D.; Cowman, A. F. Proc. Natl. Acad. Sci. U.S.A. 1990,
87, 3014-3017.
(7) Gregson, A.; Plowe, C. V. Pharmacol. Rev. 2005, 57, 117-145.
(8) Plowe, C. V. J. Exp. Biol. 2003, 206, 3745-3752.
(9) Mockenhaupt, F. P.; Teun Bousema, J.; Eggelte, T. A.; Schreiber, J.;
Ehrhardt, S.; Wassilew, N.; Otchwemah, R. N.; Sauerwein, R. W.; Bienzle,
U. Trop. Med. Int. Health 2005, 10, 901-908.
(10) Rastelli, G.; Sirawaraporn, W.; Sompornpisut, P.; Vilaivan, T.; Kamchon-
wongpaisan, S.; Quarrell, R.; Lowe, G.; Thebtaranonth, Y.; Yuthavong, Y.
Bioorg. Med. Chem. 2000, 8, 1117-1128.
(11) Yuthavong, Y.; Vilaivan, T.; Chareonsethakul, N.; Kamchonwongpaisan, S.;
Sirawaraporn, W.; Quarrell, R.; Lowe, G. J. Med. Chem. 2000, 43, 2738-
2744.
(12) Tarnchompoo, B.; Sirichaiwat, C.; Phupong, W.; Intaraudom, C.; Sir-
awaraporn, W.; Kamchonwongpaisan, S.; Vanichtanankul, J.; Thebtaranonth,
Y.; Yuthavong, Y. J. Med. Chem. 2002, 45, 1244-1252.
* To whom correspondence should be addressed. Tel.: 662-5646700 Ext 3559.
Fax.: 662-5646632. E-mail: chawanee@biotec.or.th.
(13) Kamchonwongpaisan, S.; Quarrell, R.; Charoensetakul, N.; Ponsinet, R.;
Vilaivan, T.; Vanichtanankul, J.; Tarnchompoo, B.; Sirawaraporn, W.; Lowe,
G.; Yuthavong, Y. J. Med. Chem. 2004, 47, 673-680.
(14) Kamchonwongpaisan, S.; Vanichtanankul, J.; Tarnchompoo, B.; Yuvaniyama,
J.; Taweechai, S.; Yuthavong, Y. Anal. Chem. 2005, 77, 1222-1227.
(15) Iwakura, M.; Kokubu, T. J. Biochem. (Tokyo) 1993, 114, 339-343.
(16) Ahmed, F.; Dunlap, R. B. Biotechnol. Bioeng. 1984, 26, 1227-1232.
(17) Iwakura, M.; Nakamura, D.; Takenawa, T.; Mitsuishi, Y. Protein Eng. 2001,
14, 583-589.
(1) Olliaro, P. L.; Yuthavong, Y. Pharmacol. Ther. 1999, 81, 91-110.
(2) Nzila, A. J. Antimicrob. Chemother. 2006, 57, 1043-1054.
(3) Rathod, P. K.; Phillips, M. A. Nat. Struct. Biol. 2003, 10, 316-318.
(4) Peterson, D. S.; Walliker, D.; Wellems, T. E. Proc. Natl. Acad. Sci. U.S.A.
1988, 85, 9114-9118.
(5) Peterson, D. S.; Milhous, W. K.; Wellems, T. E. Proc. Natl. Acad. Sci. U.S.A.
1990, 87, 3018-3022.
5006 Analytical Chemistry, Vol. 79, No. 13, July 1, 2007
10.1021/ac070215s CCC: $37.00 © 2007 American Chemical Society
Published on Web 05/26/2007

manners and purposes.15,16 In this paper, we report the method
for immobilization of PfDHFR to the Sepharose gel for use in the
selection of tight-binding inhibitors from combinatorial libraries.
By using the immobilized enzyme, the unbound and released
bound compounds can easily be separated from the reaction
mixture for determination and have no limitation on the solubility
of the enzyme.
Scheme 1. Synthetic Route for Trimethoprim
Derivatives
EXPERIMENTAL SECTION
Materials and Methods. For the synthesis of trimethoprim
analogues, reagents were purchased from Fluka, Merck, and
Sigma-Aldrich Ltd. and liquids were distilled before use. For
enzyme studies, chemicals were obtained from Sigma-Aldrich Ltd.,
Merck, and BDH and were used without further purification.
Oligonucleotides were made to order from Bioservice unit at our
research center. Thiopropyl Sepharose 6B was purchased from
Amersham Biosciences. Nuclear magnetic resonance spectra were
recorded in DMSO-d6 on a Bruker AV 500D spectrometer;
chemical shifts are reported in parts per million (ppm). Mass
spectra were recorded on a Micromass LCT using an electrospray
ionization technique.
Enzyme Preparation. Double mutant PfDHFR20 and PfDHFR-
TS21 (C59R+S108N) were cloned and expressed in Escherichia
coli BL21(DE3) pLysS. Two modified PfDHFR enzymes with an
eight-amino acid spacer and a Cys at the carboxylic terminal were
engineered. The spacers of the enzyme were designed based on
its natural amino acid sequences (KMLNEQNC, named K1NCR)
and modified sequences (KMLNGGGC, named K1NGCR) at the
junction region between DHFR and TS domains. These genes
were PCR amplified from the available plasmid carrying the natural
pfdhfr-ts gene21 using 5′-GCC AGC AAG CTT ATG ATG GAA CAA
GTC TGC GAC GTT-3′ as the forward primer and 5′-CTC CGC
GGT ACC TTA ACA ATT TTG TTC ATT TAA CAT TTT A-3′ or
5′-CTC CGC GGT ACC TTA ACA CCC TCC GCC ACC TAA CAT
TTT ATT ATT CGT TTT CTT-3′ as the reverse primer for K1NCR
and K1NGCR, respectively. The PCR protocol was initiated with
95 °C for 5 min, followed by 30 thermal cycles (1-min denaturation
at 95 °C, 1-min annealing period at 50 °C and 2-min extension
period at 72 °C) and 10-min final extension at 72 °C. The PCR
products of 732 base pairs were purified, restriction enzyme cut,
and ligated back into the same modified pET17b. The DNA
sequences of the constructs were verified by DNA sequencing.
Protein expression and purification by Methotrexate-Sepharose
affinity column were carried out as previously described.20
Immobilization of Enzyme to Thiopropyl-Sepharose 6B.
For large-scale preparation, dried thiopropyl-Sepharose 6B gel (1
g) was suspended in 20 mL of buffer A (20 mM phosphate buffer,
pH 7, 0.1 mM EDTA, 50 mM KCl, and 20% glycerol). The mixture
was air-aspirated to remove oxygen and bubbled with nitrogen
through the solution. Enzyme (20 mg) was added to the gel
suspension and mixed gently at 4 °C overnight. The gel-enzyme
suspension was transferred to a column and the unbound enzyme
was removed by the continuous flow of 100 mL of buffer. The
washed enzyme-bound gel was then resuspended in the same
buffer and stored at 4 °C until used.
Determination of the Amount of Enzyme Bound to the
Thiopropyl-Sepharose Gel. Fifteen portions of 10 mg of dried
thiopropyl-Sepharose 6B gel were suspended in 200 µL of oxygen-
removed buffer A. Each gel suspension was combined with 200
µL of the enzyme solution containing different amount of DHFR
enzyme (5-600 µg). The mixtures were mixed gently at 4 °C.
After 5 h of incubation, the suspensions were centrifuged for 10
min at 12 000 rpm at 4 °C, and the supernatant from each tube
was collected for the enzyme assay. The enzyme activity in the
supernatant solution was plotted against the amount of enzyme
added to the gel suspension.
Enzyme Assay. The activity of PfDHFRs was determined
spectrophotometrically at 25 °C according to the method previ-
ously described.20 The reaction (1 mL) contained 10× DHFR
buffer (50 mM N-[tris(hydroxymethyl)methyl-2-aminoethane-
sulfonic acid], pH 7.0, 75 mM â-mercaptoethanol, 1 mg/mL bovine
serum albumin), 100 µM each of the substrate dihydrofolate and
cofactor NADPH, and appropriate amount of enzyme solution. In
the case of immobilized enzyme, there is no â-mercaptoethanol
in the reaction mixture and the gel-bound enzyme was placed
instead of the enzyme solution. Thiopropyl-Sepharose gel was
added to the reaction mixture for blank measurement.
Construction of the Collections of Ligands. Pry and Cyc
were purchased from Sigma-Aldrich Ltd. Trimethoprim (Tmp) and
their derivatives were synthesized (Scheme 1) individually ac-
cording to the method described previously,11,12,22 and their Ki
values were measured by the conventional kinetic experiment.23
Two collections of ligands were prepared for use in the stoichio-
metric selection. The first collection was a mixture of an equimolar
amount of pyrimethamine, cycloguanil, and trimethoprim, which
(18) Cardoso, C. L.; Lima, V. V.; Zottis, A.; Oliva, G.; Andricopulo, A. D.; Wainer,
I. W.; Moaddel, R.; Cass, Q. B. J. Chromatogr., A 2006, 1120, 151-157.
(19) Moaddel, R.; Wainer, I. W. J. Pharm. Biomed. Anal. 2007, 43, 399-406.
(20) Sirawaraporn, W.; Prapunwattana, P.; Sirawaraporn, R.; Yuthavong, Y.; Santi,
D. V. J. Biol. Chem. 1993, 268, 21637-21644.
(21) Chitnumsub, P.; Yuvaniyama, J.; Vanichtanankul, J.; Kamchonwongpaisan,
S.; Walkinshaw, M. D.; Yuthavong, Y. Acta Crystallogr., D: Biol. Crystallogr.
2004, 60, 780-783.
(22) Sirichaiwat, C.; Intaraudom, C.; Kamchonwongpaisan, S.; Vanichtanankul,
J.; Thebtaranonth, Y.; Yuthavong, Y. J. Med. Chem. 2004, 47, 345-354.
(23) Segal, I. H., Ed. Behaviour and analysis of rapid equilibrium and steady state
enzyme systems; Wiley-Interscience: New York, 1975.
Analytical Chemistry, Vol. 79, No. 13, July 1, 2007 5007

Figure 1. Schematic representation of extended DHFRs, which were constructed in this study: (a) DHFR-KMLNEQNC (K1NCR) and (b)
DHFR-KMLNGGGC (K1NGCR).
was used for technique validation. The second collection was a
mixture of an equimolar amount of trimethoprim and its 17
analogues, which was constructed on the basis of the Ki values of
each compound against the double mutant enzyme (C59R+S108N).
The stock solution of each library was prepared in DMSO.
Selection of Tight-Binding Inhibitors. One milliliter of the
collection containing 270 µM ligands and 100 µM NADPH in
buffer A was added into the column of PfDHFR immobilized gel
(1 g of dried gel, contained 0.18 µmol of K1NCR) for the test and
into the column of Sepharose gel for control. After 5-min
incubation at 25 °C, unbound ligands were eluted from the column
and collected for further analysis using HPLC. The column was
briefly washed with buffer and treated with 1 mL of 6 M guanidine
hydrochloride containing 12% DMSO to release bound ligands
from the enzyme. The release compounds were then eluted from
the column for further analysis by HPLC. The peak area of each
ligand was determined.
HPLC and LC-MS Analyses. HPLC analyses were per-
formed on a Waters 600 system equipped with a Waters 996
photodiode array detector, using a reversed-phase column (At-
lantis C18, 5 µm, 4.6 × 250 mm) and acetonitrile and 25 mM
ammonium acetate (pH 4.5) as the mobile phase at a flow rate of
1 mL/min. The separation was started with 15% acetonitrile,
followed by stepwise gradient from 15 to 25, 25 to 35, and 35 to
70% at 5-10, 25-60, and 70-80 min, respectively. Then the
separation continued at 70% actonitrile until the last compound
was collected. The chromatograms were detected through absorp-
tion at 254 nm for peak area determination. Compound identifica-
tion was achieved by LC-MS (Micromass) using the same
column and solvent system. In the case of compounds with similar
mass, standards of individual compounds were used to locate their
peaks by HPLC under the same separation protocol.
constructed with the extension of some amino acids as a spacer
to provide the flexibility for the enzyme. Since the DHFR domain
of the bifunctional enzyme PfDHFR-TS joined to the TS domain
by the junction region (JR) of which the amino acid residue at
position 8 is Cys, double mutant (C59R+S108N) enzyme extended
with eight amino acids of the JR was constructed and named
K1NCR (Figure 1a). Thus, the eight amino acid residues of the
JR acted as a spacer. Alternatively, it is possible that bond
formation of the enzyme to the solid support might be restricted
due to the steric effect of the large side chains of Glu, Gln, and
Asn attached to Cys; therefore, three glycine (Gly) residues were
introduced in place of these amino acid in the modified spacer
and the engineered enzyme was named K1NGCR (Figure 1b).
Following enzyme purification, dithiothreitol (DTT) was re-
moved from the enzyme solution by gel filtration through a small
column of Sephadex G-50 prior to immobilization. This is to
prevent the cleavage of S-S bond between the enzyme and the
solid support in the immobilization process and to keep the S-S
bond intact throughout the study. The specific activities of free
K1NCR and K1NGCR enzymes after DTT removal were 71.8 and
80.3 units/mg of protein, respectively, which were similar to that
of the parent C59R+S108N double mutant enzyme, K1 (90.1 units/
mg of protein). The result indicated that the extended amino acids
and the removal of DTT did not affect the enzyme activity.
Immobilization of DHFRs on Sepharose Gel. The amount
of enzyme bound to thiopropyl-Sepharose gel was determined by
a titration-based experiment.15 Following incubation of 10 mg of
the gel with various amounts of PfDHFR enzymes ranging from
5 to 600 µg, unbound enzyme was separated and the activity of
enzyme was determined. Figure 2 shows the relationship between
the total enzyme added to the gel suspension and the activity of
unbound enzyme detected in supernatant solution. At low amounts
of enzyme added, 5-40 µg, the activity of DHFR enzyme was
undetectable to very low. With the increasing amount of enzyme,
the enzyme activity of the unbound fraction became detectable.
This indicated that, after saturation of the immobilization of
enzyme onto the Sepharose gel, further addition of enzyme merely
increased the amount of unbound enzyme, and therefore, the
enzyme activity in the supernatant solution showed a linear
relationship with the amount of added enzyme. The total amount
of enzyme bound to the gel can then be estimated by extrapolating
the linear line back to zero (x-axis intercept). This estimation is
based on the assumption that the disappearance of enzyme activity
RESULTS AND DISCUSSION
Design and Expression of Mutant DHFRs. The SH group
of the cysteine (Cys) residue is very useful for immobilization of
an enzyme to the solid support15,24 because of its easy bond
formation under mild conditions. For example, modified E. coli
DHFR with a Cys residue at the C-terminal has been successfully
immobilized to a thiopropyl-Sepharose gel.15 In this study, PfDH-
FRs were designed to immobilize to thiopropyl-Sepharose via an
S-S bond. PfDHFRs with a Cys residue at the C-terminal were
(24) Srere, P. A.; Uyeda, K. Methods Enzymol. 1976, 44, 11-19.
5008 Analytical Chemistry, Vol. 79, No. 13, July 1, 2007

Figure 2. Immobilization of enzyme K1NCR (a) and K1NGCR (b) on thiopropyl-Sepharose 6B gel.
Scheme 2. General Protocol for Stoichiometric
Selection
Figure 3. Selective binding of a collection containing pyrimethamine
([), cycloguanil (9), and trimethoprim (2) by an immobilized DHFR.
The total concentration of ligands was fixed at 50 µM while the
concentration of the enzyme was varied from 0, 25, 50, 100, 150, to
188 µM to give the molar ratio of enzyme to total ligands at 0, 0.5,
1.0, 2.0, 3.0, and 3.76, respectively.
following DTT treatment to release the enzymes from the gel.
By this direct determination, the amounts of bound K1NCR and
K1NGCR enzymes on the gel were 4.4 and 3.2 mg/g of dried gel,
respectively, which were similar to the values determined by the
titration method. Upon release from the gel, the specific activities
of the free enzymes (40.7 and 38.3 units/mg of protein) were
comparable to those of the immobilized enzymes.
Technique Validation. The general procedure for the sto-
ichiometric selection of tight-binding inhibitors by an immobilized
enzyme is outlined in Scheme 2. Ligands with high affinity should
bind tightly to the enzyme and the unbound ligands could then
be eluted from the column for further quantitative analysis and
from supernatant solution is immobilized onto the gel support;
thus, the maximum estimate could be obtained from this method.
The calculated specific activity of the enzyme would decrease as
the amount of immobilized enzyme estimated increased. There-
fore, it was not unreasonable to assume that this method gave
the minimum specific enzyme activity. It was estimated that the
amount of bound K1NCR and K1NGCR enzymes on the gels were
5 and 4 mg/1 g of dried gel as shown in Figure 2a and b,
respectively. The calculated specific activities of the immobilized
enzymes were 27.3 and 21.5 units/mg of protein for K1NCR and
K1NGCR enzymes, respectively (using a maximum estimate value
of each enzyme). A more accurate figure was also obtained by
determining the amount of enzyme by protein determination25
(25) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.
Analytical Chemistry, Vol. 79, No. 13, July 1, 2007 5009

Table 1. Structure, Retention Time, Mass, and Ki Value against Double Mutant PfDHFR of Each Ligand in the
Collection
Ki double
molecular
mass
mutant (nM)
retention
time
M + H+
Cpd
peak
R
R′
expt
calcd
calcd
(m/z)
T1
1
2
4-N(CH3)2
3,4,5-(OMe)3
3,4,5-(OMe)3
2,4-(OMe)2
4-OEt
H
12.1 ( 0.3
13.6 ( 0.5
16.8 ( 0.1
20.4 ( 0.1
21.4 ( 0.1
24.7 ( 0.1
28.2 ( 0.1
42.2 ( 0.4
43.3 ( 0.1
46.8 ( 0.1
50.9 ( 0.1
52.8 ( 0.2
54.8 ( 0.1
59.0 ( 0.1
60.3 ( 0.1
68.1 ( 0.1
82.9 ( 0.1
87.1 ( 0.1
449.8 ( 20.1
242.1 ( 40.1a
520.3 ( 122.8a
762.2 ( 37.4
199.8 ( 20.3
285.8 ( 47.0
222.1 ( 80.2
496.5 ( 54.1
5.2 ( 0.3a
410.65
-
243.31
290.32
304.35
260.29
244.30
278.01
250.30
242.32
396.45
440.50
272.35
256.35
306.37
350.42
364.45
364.45
426.52
452.05
244.15
291.14
305.16
261.14
245.14
279.02
251.13
243.16
397.19
441.21
273.17
257.17
307.16
351.18
365.20
365.20
427.21
453.07
Tmp
T2
H
3
Me
H
639.42
890.06
238.12
219.97
169.05
506.48
13.21
12.49
310.47
113.87
22.16
15.82
74.33
15.29
98.98
35.46
T3
4
T4
5
H
T5
6
4-Br
H
T6
7
2,3-(C4H4)-
H
T7
8
4-Pri
H
T8
9
4-OCH2-[3,4,5-(OMe)3]Ph
3-OEt-4- OCH2-[3,4,5-(OMe)3]Ph
3-OBun
H
T9
10
11
12
13
14
15
16
17
18
H
2.2 ( 0.1a
T10
T11
T12
T13
T14
T15
T16
T17
H
288.6 ( 29.2
134.5 ( 18.5
15.1 ( 1.8
4- But
3-OPh
H
Me
H
3-OEt-4-OCH2Ph
3-OMe-4-OCH2Ph
4-OC3H6OCH2Ph
3- OCH2Ph-4-OCH2Ph
4-OC3H6O-[2,4,5-(Cl)3]Ph
9.7 ( 1.4a
Et
H
60.1 ( 8.1a
6.7 ( 0.5
Me
H
80.0 ( 10.3a
39.1 ( 3.1
a Data from ref 22.
identification by HPLC and LC-MS. Ki values of each ligand in
the library could be calculated individually using eq 114 by mixing
the reference compound with known Ki value to the library as an
internal standard.
the amount of free inhibitors was determined. Figure 3 shows
that the concentrations of free ligands decrease sequentially in
the order of Cyc, Pyr, and Tmp as expected. Using eq 1 and data
in Figure 3 at the enzyme/ligand ratio of 2, where all ligands share
the binding to the enzyme at less than 100% binding, Pyr as the
reference with Ki value of 53.9 nM, the calculated Ki values of
Cyc and Tmp were 33.3 and 270.9 nM, respectively. These values
are comparable with the previously reported Ki values of 42.63 (
6.3 and 242.1 ( 40.1 nM, respectively. This demonstrated that
the principle of stoichiometric selection of ligands was applicable
to an immobilized PfDHFR and their ligands and the accurate Ki
value of each compound in the library could be calculated by this
method.
Construction and Identification of Trimethroprim Deriva-
tives in the Collection. A series of Tmp derivatives has been
specifically developed against plasmodial DHFR mutants with
S108N mutation based on its flexible side chain on the pyrimidine
ring that can avoid steric clash with the S108N.22 Tmp and some
derivatives were verified experimentally as competitive binding
inhibitors against the double mutant enzyme (C59R+S108N) using
the classical double-reciprocal plots (data not shown). The Ki
values of these22 and other DHFR inhibitors10-13,20 were therefore
calculated based on eq 226 or using nonlinear least-squares fit of
the data to the Michaelis-Menten equation.23
[li] [Elref]
Ki value of ligand i, Kii )
· Kiref
(1)
[Eli] [lref]
[li], [Eli], [lref], and [Elref] are concentrations of free ligand i,
enzyme-ligand complex, reference, and enzyme-reference com-
plex, respectively, where
peak area of li in the test
peak area of li in control
[li] )
· lti
[Eli] ) lti - [li]
lti ) the total concentration of ligand i
peak area of lref in the test
[lref] )
· ltref
peak area of lref in control
[Elref] ) ltref - [lref]
ltref ) the total concentration of reference
IC50
Ki )
(2)
[S]
Km
1 +
Since the activities of two immobilized enzymes were comparable,
nonmodified enzyme K1NCR was selected for further study. The
theoretical expectation of the binding of immobilized enzyme with
the mixture of inhibitors was tested by titration of the enzyme
with a mixture of equimolar of each Pyr, Cyc, and Tmp.14 Then
where IC50 is the concentration of the inhibitor that inhibits
catalytic activity of the enzyme by 50%.
5010 Analytical Chemistry, Vol. 79, No. 13, July 1, 2007

an excess molar ratio of 1.5/1.0 ligands to the enzyme, i.e., 270
µM total ligands (15 µM each) to 180 µM enzyme. The unbound
ligands were first eluted from the column and then analyzed by
HPLC. The missing compounds from the chromatogram, com-
pared to the control, were determined as bound ligands. These
were confirmed by the analysis of bound ligands, which were later
released from the enzyme by guanidine hydrochloride. Ligands
with lower Ki values (higher affinities) competed for the enzyme
much more readily and bound to the enzyme at higher extent
than those with higher Ki values, as shown by a linear correlation
between percent binding to the enzyme from both unbound and
bound ligands (calculated from eqs 4 and 5, respectively) and Ki
values in Figure 4. It is worth mentioning that percent binding
determined from bound ligands were slightly lower than those
from unbound ligands, which is presumably due to the dissociation
of weak binding complexes (high Koff rate) in the washing step.14
Using trimethoprim as a reference with a Ki value of 242.1 applied
to eq 1, the calculated Ki value of each derivative obtained from
the data of unbound fraction was comparable to that measured
kinetically (Table 1) as shown by the linear relationship between
Ki values from both methods in Figure 5.
Figure 4. Correlation of percentage of binding and Ki value of
ligands upon selection by immobilized enzyme using an excess molar
ratio of enzyme to total ligands of 1.0:1.5; percent binding from bound
ligands (4) and percent binding from unbound ligands (9).
% binding (from unbound) )
peak area of the test
peak area of control
100 1 -
(4)
(5)
(
)
)
Figure 5. Correlation between Ki values of ligands in the collection
from calculated and measured kinetically.
% binding (from bound) )
peak area of the test
peak area of control
100
(
If a tight binding inhibitor was obtained, as defined by an
inhibitor whose IC50 value observed is similar to the concentration
of total enzyme in the sample, the calculation of the Ki value was
based on eq 3, where total concentration of the enzyme [E] was
also taken into account.26
The results indicated that this method could efficiently be used
for the selection of tight-binding inhibitors from the collection,
and the Ki value of each ligand could be predicted reasonably
accurately. It is unfortunate that reuse of the immobilized enzyme
was not possible due to the denaturation of the enzyme by
guanidine hydrocholoride. However, the advantages of this
method over the free enzyme method11 are that a higher amount
of enzyme can be used, leading to a possible screening of a larger
size of collection with higher concentration of ligands and thus
increase in the sensitivity and accuracy of detection by quantitative
HPLC analysis. Thus, this system should be a useful tool for a
primary screening or large-scale selection of tight-binding inhibi-
tors from combinatorial syntheses or libraries of natural products.
[E]
2
IC50
-
Ki )
(3)
[S]
Km
1 +
The collection of Tmp and its derivatives was constructed as
a representative of the combinatorial library. Successful evaluation
of the collection would clearly imply that this method can be used
for the screening of a library of potential inhibitors from combi-
natorial synthesis. In order to generate the collection of the
compounds with a variety of binding affinities, a collection of 17
trimethoprim derivatives together with trimethoprim itself as
reference was constructed based on their Ki values against the
double mutant enzyme (C59R+S108N), which was measured by
a conventional kinetic method.23 This 18-compound-collection was
subjected to HPLC analysis using a reversed-phase column and
stepwise gradient of 15-70% acetonitrile and 25 nM ammonium
acetate (pH 4.5), and each compound was identified by LC-MS.
The amount of each compound was determined from the peak
area at 254 nM. The structures, Ki values, and analysis results of
these ligands in the library are summarized in Table 1.
CONCLUSION
We have shown that PfDHFRs could be immobilized onto
Sepharose gel via an S-S bond with the estimated amount of 4-5
mg/g of dried gel and could be used for the selection of tight-
binding inhibitors from combinatorial libraries. By this method,
both bound and unbound inhibitors were easily separated from
the reaction mixture for further quantitative analysis and identi-
fication by HPLC and LC-MS. Since there was no limitation on
the solubility of the immobilized enzyme, a higher concentration
of the enzyme should be applicable. Thus, the higher concentra-
tion of each ligand could be used and a more accurate detection
can be obtained from quantitative HPLC analysis.
Selection of Lead Compounds from the Collection by the
Immobilized Enzyme. The binding assay was carried out with
ACKNOWLEDGMENT
TRF grant for new researcher from the Thailand Research
Fund to C.T. and supports from Medicines for Malaria Venture
(26) Copeland, R. A. Enzymes: a practical introduction to structure, mechanism,
and data analysis, 2nd ed.; Wiley: New York, 2000.
Analytical Chemistry, Vol. 79, No. 13, July 1, 2007 5011

(MMV) and Wellcome Trust to Y.Y. are gratefully acknowledged.
S.K. is a HHMI international research scholar.
of the PfDHFR-C59RS108N by Tmp and derivative. This material
is available free of charge via the Internet at http://pubs.acs.org.
SUPPORTING INFORMATION AVAILABLE
Received for review February 1, 2007. Accepted April 22,
2007.
Listing of spectroscopic data of Tmp derivatives, HPLC
spectrum of the collection of ligands, and Lineweaver-Burk plots
AC070215S
5012 Analytical Chemistry, Vol. 79, No. 13, July 1, 2007