Structure-based design, synthesis, and characterization of inhibitors of human and Plasmodium falciparum dihydroorotate dehydrogenases

Article abstract of DOI:10.1021/jm800963t

Pyrimidine biosynthesis is an attractive drug target in a variety of organisms, including humans and the malaria parasite Plasmodium falciparum. Dihydroorotate dehydrogenase, an enzyme catalyzing the only redox reaction of the pyrimidine biosynthesis pathway, is a well-characterized target for chemotherapeutical intervention. In this study, we have applied SPROUT-LeadOpt, a software package for structure-based drug discovery and lead optimization, to improve the binding of the active metabolite of the anti-inflammatory drug leflunomide to the target cavities of the P. falciparum and human dihydroorotate dehydrogenases. Following synthesis of a library of compounds based upon the SPROUT-optimized molecular scaffolds, a series of inhibitors generally showing good inhibitory activity was obtained, in keeping with the SPROUTLeadOpt predictions. Furthermore, cocrystal structures of five of these SPROUT-designed inhibitors bound in the ubiquinone binding cavity of the human dihydroorotate dehydrogenase are also analyzed.

Full text of DOI:10.1021/jm800963t

J. Med. Chem. 2009, 52, 2683–2693
2683
Structure-Based Design, Synthesis, and Characterization of Inhibitors of Human and
Plasmodium falciparum Dihydroorotate Dehydrogenases†
Matthew Davies,‡ Timo Heikkila¨,‡ Glenn A. McConkey, Colin W. G. Fishwick,* Mark R. Parsons,* and A. Peter Johnson*
School of Chemistry, Faculty of Biological Sciences, and Astbury Centre for Structural Molecular Biology, UniVersity of Leeds, Leeds LS2 9JT, U.K.
ReceiVed July 30, 2008
Pyrimidine biosynthesis is an attractive drug target in a variety of organisms, including humans and the
malaria parasite Plasmodium falciparum. Dihydroorotate dehydrogenase, an enzyme catalyzing the only
redox reaction of the pyrimidine biosynthesis pathway, is a well-characterized target for chemotherapeutical
intervention. In this study, we have applied SPROUT-LeadOpt, a software package for structure-based drug
discovery and lead optimization, to improve the binding of the active metabolite of the anti-inflammatory
drug leflunomide to the target cavities of the P. falciparum and human dihydroorotate dehydrogenases.
Following synthesis of a library of compounds based upon the SPROUT-optimized molecular scaffolds, a
series of inhibitors generally showing good inhibitory activity was obtained, in keeping with the SPROUT-
LeadOpt predictions. Furthermore, cocrystal structures of five of these SPROUT-designed inhibitors bound
in the ubiquinone binding cavity of the human dihydroorotate dehydrogenase are also analyzed.
Introduction
The dihydroorotate dehydrogenases (DHODHs) are attractive
chemotherapeutic targets in various pathogens, such as Plas-
modium falciparum and Helicobacter pylori, and for the
treatment of human disease.1-3 DHODH catalyzes the conver-
sion of dihydroorotate (DHO) to orotate by utilizing an FMN
cofactor in the only redox reaction present in the pyrimidine
Figure 1. Brequinar 1, leflunomide 2, and 3 (A77 1726), the active
biosynthesis pathway. The human DHODH (HsDHODH)a and
metabolite of leflunomide.
the P. falciparum DHODH (PfDHODH) both belong to the
group of class 2 DHODHs that use respiratory quinones as the
All potent inhibitors of HsDHODH published to date bind
terminal electron acceptors. This contrasts with the class 1
to the putative ubiquinone binding channel and display beneficial
enzymes which utilize fumarate or NAD as a cofactor.4 The
immunosuppressive and antiproliferative activities, shown to be
most pronounced during T-cell proliferation.11 Brequinar 1 and
leflunomide 2 are two examples of low-molecular weight
class 2 enzymes are membrane-associated and, in eukaryotes,
are localized to the inner mitochondrial membrane.5,6 Kinetic
studies have shown a two-site ping-pong mechanism for the
inhibitors of DHODH that have been in clinical development
conversion of dihydroorotate in class 2 DHODHs.4,5 Structur-
ally, these enzymes consist of a conserved R/ꢀ barrel core and
an N-terminal extension, which folds into a separate R-helical
domain.7-9 Additionally, in eukaryotic organisms, the N-
terminal domain extends into the membrane forming a single
membrane-anchoring helix, although this is usually truncated
for in vitro studies of the enzyme.10,11 The first half-reaction
catalyzed by the enzyme takes place within the core of the
protein and comprises the reduction of dihydroorotate to orotic
acid with concomitant oxidation of the FMN cofactor. This
reaction is mediated by a conserved serine acting as the active
site base.7,10 FMNH2 is subsequently regenerated by respiratory
quinones which, in eukaryotes, are recruited from the inner
mitochondrial membrane and are thought to bind to the helical
N-terminal domain.
(Figure 1).11-13 Leflunomide is now marketed as a treatment
for rheumatoid arthritis.
Additionally, a group of cyclopentene carboxylic acid amides
are in development as HsDHODH-targeting immunosuppressive
agents.14-16 The structure of HsDHODH has been previously
determined in complex with compound 3 (the active metabolite
of leflunomide), compound 1, and some of the cyclopentene
carboxylic acid amides.16,17 Inspection of these structures reveals
that whereas 1 and 3 both appear to show a single mode of
binding to DHODH, a number of the cyclopentene carboxylic
acid amides display an interesting dual binding mode within
the same cocrystal.
The structure of 3 cocrystallized with the P. falciparum
DHODH has been reported recently.8 Although this compound
is a relatively weak inhibitor of the Plasmodium enzyme, this
system provides a good starting point for structure-based
development of more potent inhibitors. The binding mode
observed for 3 in the PfDHODH cocrystal structure differs
significantly from that observed in the HsDHODH cocrystal
structure (Figure 2). In the PfDHODH-3 complex, three direct
hydrogen bonds can be observed (to His185, Arg265, and
Tyr528), whereas in the HsDHODH-3 complex, there are
water-mediated hydrogen bonds to Arg265 and Gln47 in
addition to the direct hydrogen bond to Tyr356. Closer inspec-
tion of the binding cavities also shows that in both cases, the
†
Crystal structural coordinates are deposited in the Protein Data Bank
as entries 3F1Q, 3FJ6, 3FJL, 3G0U, and 3GOX.
* To whom correspondence should be addressed. C.W.G.F.: e-mail,
c.w.g.fishwick@leeds.ac.uk; phone, +44 113 3436510; fax, +44 113
3436530. M.R.P.: e-mail, m.r.parsons@leeds.ac.uk; phone, +44 113
3432574; fax, +44 113 3433106. A.P.J.: e-mail, p.johnson@leeds.ac.uk;
phone, +44 113 3436515; fax, +44 113 3436530.
‡
These authors contributed equally to this work.
a Abbreviations: PfDHODH, P. falciparum dihydroorotate dehydroge-
nase; HsDHODH, human dihydroorotate dehydrogenase; DHO, dihydro-
orotate; CoQ, ubiquinone coenzyme; FMN, flavin mononucleotide.
10.1021/jm800963t CCC: $40.75
2009 American Chemical Society
Published on Web 04/07/2009

2684 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9
DaVies et al.
Figure 3. Hydrophobic tail moiety and hydrogen bond-forming
headgroup of inhibitor 3.
the X-ray crystal structure of the chosen protein to construct
putative ligand scaffolds. The system can then be used to rank
these designed scaffolds using various considerations such as
predicted binding affinity and molecular complexity to aid in
the choice of compound for synthesis. Very recently, we have
developed a variant of this software, SPROUT-LeadOpt, which
can take the structure of a small molecule complexed to a protein
(e.g., a cocrystallized enzyme inhibitor) and generate variants
predicted to have higher affinities. A particular feature of this
software is that the generation of the variant structures is
performed using a retrosynthetic approach utilizing a chemistry
knowledge base and a library of commercially available starting
materials from which alternative “monomers” (which are
structural variants of the monomers found in the original
structure) are automatically selected.
As part of an ongoing program aimed at the development of
inhibitors of class 2 DHODHs using structure-based methods,21
we wished to utilize SPROUT-LeadOpt to rationally improve
the binding affinity of inhibitor 3 that has been previously
cocrystallized with both the human and Plasmodium DHODH
enzymes. In this study, we set out to generate simple compounds
that would utilize the hydrogen bonding network from the
headgroup moiety of compound 3 and have an improved “tail”
(Figure 3) that fills the hydrophobic pocket at the entrance to
the binding cavity.
In addition to the partially filled hydrophobic cavity occupied
by the tail section of inhibitor 3, further analysis of the ligand
binding cavities of both cocrystal structures using SPROUT
revealed the presence of a second, smaller hydrophobic cavity
in the region occupied by the methyl group within the head
section of the inhibitor. As this cavity again appeared to be only
partially occupied, we began by investigating the effect of
replacing the methyl moiety with a number of larger hydro-
phobic groups to fit into the pocket at the end of the ubiquinone
binding channel (Table 1, compounds 4-9). The tail of
compound 3 was extended by replacement of the CF3 group by
a phenyl ring to give a biphenyl tail (Table 1, compounds
10-16). The biphenyl tail was incorporated into the inhibitors
to improve their binding through enhanced hydrophobic interac-
tions within the large hydrophobic region of the PfDHODH
binding cavity.
Figure 2. Overlay of 3 bound in HsDHODH (blue) and PfDHODH
(green), illustrating the different binding mode of the inhibitor in each
case. Note that the “headgroup” of the inhibitor adopts an opposite
orientation. In PfDHODH, hydrogen bonds are made to His185, Arg265,
and Tyr528, whereas in HsDHODH, hydrogen bonds are formed to
Gln47/Arg136 (water-mediated) and to Tyr356 and, significantly, a
hydrogen bond to Tyr147 holds His56 in a conformation where
hydrogen bonding to 3 is no longer possible.
trifluoromethyl phenyl section of the inhibitor only partly fills
a portion of the binding cavity that is mainly lined by
hydrophobic amino acid residues. In the case of the compound
3-derived cocrystal structures, the presence of this unoccupied
space holds promise in the design of suitable variants of these
inhibitors which would better fill this cavity. In addition to the
likely increase in binding affinities, inhibitors designed to more
effectively fill the binding cavities could also exploit differences
in the shapes between the two cavities, possibly leading to
species-specific selectivity. Because of the existence of the
salvage pathway in humans, such selectivity may not be essential
for the development of new antiplasmodial agents targeting
PfDHODH. Of course, such selectivity is highly desirable,
particularly since leflunomide, the pro-drug precursor of com-
pound 3, has been found to exhibit some dose-dependent side
effects in a small number of patients.18
Here, we present the results of a computational structure-
based lead optimization exercise aimed at improving the binding
of compound 3 to the ubiquinone binding cavities in DHODH.
A small number of compounds were synthesized and tested
against the human and P. falciparum enzymes in inhibition
assays. We observed that a number of the compounds were very
potent inhibitors of the human DHODH, and therefore, the
cocrystal structures of five of the most potent compounds bound
to the ubiquinone binding site of HsDHODH were determined
and analyzed.
Results and Discussion
Lead Optimization and Analysis of Inhibitors. Structure-
based computational molecular design is an important tool for
use in medicinal chemistry and chemical biology, and such
methods have been recognized as offering considerable potential
when applied to early stage drug discovery. In particular, the
importance of these methods in therapeutic areas where random
high-throughput screening has proved disappointing, such as
in the discovery of new anti-infective agents, has been recently
underlined.19 We have previously described the application of
the de novo molecular design program SPROUT to the
production of a number of enzyme inhibitors and receptor
antagonists.20 SPROUT is a program that uses a fragment-based
approach applied to a targeted cavity (e.g., active site) within
Following enzymatic screening of compounds 4-16 (see the
Supporting Information) with both the human and Plasmodium
DHODH enzymes, it was evident that substitution of the methyl
group in compound 3 for a cyclopropyl group as in 4 led to an
increase in binding affinity for both Plasmodium and human
DHODH (Table 2), consistent with previous observations for
the rodent enzymes.22 However, further increases in the size of
the hydrophobic substituents at this position (as in compounds
5-9) resulted in an adverse effect upon inhibition of enzymatic
activity compared to that of compound 3 itself (results not
shown). Furthermore, replacement of the trifluoromethyl group
in compound 3 with phenyl as in 10 resulted in a significant
improvement in binding affinities for both of the enzymes

Human and P. falciparum Dihydroorotate Dehydrogenases
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9 2685
Table 1. Analogues of Compound 3
Table 2. Experimentally Defined IC50 and Kiapp Values for Compound 3 and Derivativesa
a For the complete data set of biological activity of compounds 4-16, see the Supporting Information.
studied here (Table 2). A combination of these features,
however, as in compound 11 resulted in good inhibition of
HsDHODH but an 8-fold reduction in activity against PfD-
HODH (Table 2).
Inspection of the SPROUT-generated models of inhibitors
4, 10, and 11 bound within both the human and Plasmodium
DHODH crystal structures revealed some important differences
in terms of the requirements for efficient inhibitor binding
between the cavities for both enzymes. For instance, inhibitor
4, containing a cyclopropyl-based headgroup, is predicted to
fit comfortably into the binding channels of both enzymes. In
the case of inhibitor 10, however, modeling indicated that
acceptable binding within the Plasmodium enzyme to maintain
the predicted H-bonding to the residues in the headgroup region
required movement of the side chain of Met536 which lies
toward the entrance of the binding cavity (Figure 4). Even if

2686 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9
DaVies et al.
that this effect would be most pronounced in the case of
compound 21 where the doubly ortho-substituted benzene ring
would induce a 90° angle between the two biaryl rings, and
indeed, modest selectivity for PfDHODH over the human
enzyme was observed (Table 3).
The potential for inhibitors of PfDHODH to act as antiplas-
modial agents has recently been underlined23,24 in high-
throughput screening studies which led to the identification of
potent and selective inhibitors of this enzyme. Interestingly,
preliminary testing of eight compounds (17-20, 22, 23, 26, and
28) against parasites (data not shown) indicated that compounds
17-19 exhibited activity in the micromolar range, indicating
that these inhibitors merit further and more detailed investigation
into their potential as antiplasmodial agents.
Structures of Inhibitors in Complex with HsDHODH. In
light of the generally good affinity shown by the inhibitors for
the human enzyme, we investigated the details of their interac-
tion with HsDHODH using X-ray crystallography. Cocrystal
structures were determined for five potent inhibitors bound to
human DHODH. The structures were determined to a high
resolution and allow a detailed analysis of the interactions
between the enzyme and inhibitors. In general, the protein
components of the structures of these five cocrystal complexes
are nearly identical to previously published structures of
HsDHODH (CR rmsd < 0.25 Å).7,17
As previously reported, some parts of the structures remain
disordered,7,17 and in particular, the surface loops consisting of
residues 69-71 and residues 216-221 are poorly defined in
the electron density map. Furthermore, the N-terminal polyhis-
tidine tag and the following linker region are fragmented in
density and could not be built into the final model. Analysis of
the membrane association domain also reveals density attribut-
able to the detergent molecules between and at the ends of the
two R-helices comprising the N-terminal domain.7 This density
was also quite fragmented, and the detergent molecules were
not built into any of the models. As expected, the majority of
the differences seen in the side chain conformations appear to
be concentrated in the ubiquinone binding channel and are most
likely induced by the binding of the inhibitors.
In most cases, the head moiety occupies the same part of the
cavity as compound 3, and the water-mediated hydrogen bonds
to Gln47 and Arg136 have been preserved. Additionally, and
consistent with the SPROUT-derived models of these com-
plexes, the biphenyl tail extends toward the surface of the protein
and fills the hydrophobic region of the target binding cavity.
All of the cocrystal structures here clearly show a single mode
of binding for the inhibitors (Figure 5).
However, in the HsDHODH-24 cocrystallized complex, the
headgroup moiety of the inhibitor is rotated 180° compared to
its orientation in the cocrystal structures of the other inhibitors
and that of compound 3 and instead (Figure 6) adopts a
conformation similar to that observed for compound 3 in the
PfDHODH enzyme. Following the nomenclature established
previously,16 this binding mode can be designated as “non-A77
1726-like”. However, it should be noted that despite this change
in the pose adopted by the ligand, and in contrast to the situation
with PfDHODH depicted in Figure 3, the conformation adopted
by His56 still precludes formation of a hydrogen bond to the
ligand, because the alternative hydrogen bond to a water
molecule is apparently preferred (Figure 6).
Figure 4. Movement of the side chain of Met536 (gray, before; green,
after) required for the binding of 10 in PfDHODH.
we allow for this movement, the S-methyl group forming part
of the Met536 side chain is predicted to make unfavorable steric
interactions with the phenyl group of inhibitor 10 which appears
to be reflected in the rather modest inhibition observed for
PfDHODH (Table 2). In the case of HsDHODH, where Met536
is replaced with Pro364, the shape of the cavity is such that
inhibitor 10 is predicted to fit comfortably in the binding cavity,
consistent with the good observed inhibition of HsDHODH.
Additionally, analysis of the modeled complex of inhibitor
10 and PfDHODH revealed the existence of a number of discrete
subcavities and clefts in the protein surface surrounding the large
hydrophobic tunnel containing the biphenyl tail of the inhibitor.
To further increase inhibitor affinity, SPROUT-LeadOpt was
used to identify substituted biaryl-based variants of inhibitor
10 that were predicted to maximize binding of the inhibitor
within the hydrophobic tunnel. SPROUT-LeadOpt’s automated
retrosynthetic analysis of 10 yielded phenylboronic acid and
4-bromoaniline as commercially available monomers. Automatic
substitution of fragments derived from commercially available
superstructure compounds followed by redocking and scoring
gave a series of alternative monomers. This resulted in the design
of a range of mono- and disubstituted biaryl-based systems
[compounds 17-29 (Table 3)] which were predicted (SPROUT
score) to bind with greater affinity to PfDHODH compared to
compound 3 or the biphenyl 10. Following synthesis (see
below), these systems were assayed with both the Plasmodium
and human DHODH enzymes. In all but one case (compound
25), these compounds were found to display a stronger inhibition
of PfDHODH compared to that found for compound 3, in
keeping with the modeling predictions. Additionally, compounds
17-21 showed increased inhibitory activity with PfDHODH
compared to that of inhibitor 10 (Table 3). Almost all of these
biaryls exhibited greater inhibitory action toward HsDHODH
than PfDHODH, the exceptions being compounds 21 and 23
which displayed a marginal inhibitory preference for the
Plasmodium enzyme (Table 3). However, it is noteworthy that
in contrast to the behavior of these inhibitors with PfDHODH,
the introduction of substituents into the biphenyl tail of
compound 10 produced compounds that generally had reduced
potency for the human enzyme. We have recently reported21
a
series of anthranilic acid analogues that were potent PfDHODH
inhibitors, and because of the relative shapes of the binding
cavities in both enzymes, it was found that those inhibitors with
structures able to exist in a nonplanar conformation were shown
to display greater affinity for the parasite enzyme over the human
enzyme. In light of these earlier observations, it was expected
The cocrystal structures of 10, 19, and 29 involve two water-
mediated hydrogen bonds (to residues Arg136 and Gln47) and
a direct hydrogen bond involving the hydroxyl of Tyr356. In
the case of 24, however, although involving similar interactions,

Human and P. falciparum Dihydroorotate Dehydrogenases
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9 2687
Table 3. Experimentally Defined IC50 and Kiapp Values for Compounds Developed from 10 Using SPROUT-LeadOpt

2688 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9
DaVies et al.
Figure 5. Stereo images of the binding modes of compounds 10, 19, and 29. The conserved histidine, His56, appears to form a hydrogen bond
with Tyr147, thus locking His56 in a conformation that allows stacking against the headgroup of the inhibitor and prevents H-bonding to the
inhibitor. Water molecules are shown as orange spheres. The 2FO - FC electron density map is contoured at 2σ around the ligand, with the
compound built in the difference electron density using COOT (see the Supporting Information for details).
here it is the nitrile that forms a hydrogen bond with Tyr356
while contact to Arg136 remains water-mediated (Figure 6).
The torsion angle between the amide moiety and the adjacent
aryl ring of the biphenyl is considerably larger in the HsD-
HODH-24 complex than in the complex with the unsubstituted
compound, 10 (42° vs 20°, respectively). This difference in
overall molecular shape appears to be closely related to the
difference in the binding modes involving the additional
hydrogen bonding interaction of compound 24 with Tyr147.
This is then reflected in the binding affinity, as compound 24
is the most potent of the inhibitors studied here, with a Kiapp of
2.7 nM.
Binding of the cyclopropyl-containing compound 11 closely
resembles that of 10 and 3. The cyclopropyl group occupies a

Human and P. falciparum Dihydroorotate Dehydrogenases
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9 2689
Figure 6. Stereo image of compound 24 bound to the ubiquinone binding site of HsDHODH. The binding mode of 24 shows an additional
water-mediated hydrogen bond to Tyr147 in addition to the direct hydrogen bond to Tyr356 as well as the water-mediated hydrogen bonds to
Gln47 and Arg136. Water molecules are shown as orange spheres. The 2FO - FC electron density map is contoured at 2σ around the ligand, with
the compound built in the difference electron density using COOT.
Figure 7. Stereo image of compound 11 bound to the ubiquinone binding site of HsDHODH. The compound binds in an A77 1726-like binding
mode. Water molecules are shown as orange spheres. The 2FO - FC electron density map is contoured at 1σ around the ligand, with the compound
built in the difference electron density using COOT.
small cavity lined by hydrophobic residues above Tyr356 and
next to the FMN (Figure 7).
extensively compared to that of compound 10. The SPROUT-
generated structures for two representative examples are shown
in Figure 8.
Detailed examination of the structure of the bound inhibitor
within the compound 11-HsDHODH complex revealed the
presence of a conformation in which the cyclopropyl ring is
bisected by the adjacent enolic hydroxyl group which is
consistent with observations reported previously for similar
cyclopropyl-based amides.25
Additionally, to accommodate the binding mode of compound
11, His56 adopts a conformation similar to that found in the
cocrystal structures containing compounds 10, 19, and 29
(Figure 5) and features an H-bond between His56 and Tyr147.
Predicted Binding Mode of Inhibitors in PfDHODH. All
compounds that were found to be inhibitors of PfDHODH were
predicted, using SPROUT, to bind in a manner similar to that
observed for compound 10, with the planar headgroup making
direct hydrogen bonds to residues Arg265, His185, and Tyr528.
The biphenyl tail of each inhibitor was predicted to bind in the
large hydrophobic region of the binding cavity, in a fashion
analogous to that found for these inhibitors in HsDHODH.
Furthermore, the substituted biaryl moiety present in these
inhibitors was predicted to occupy the binding cavity more
It was predicted that all the inhibitors would bind along the
same common axis and make the same hydrogen bonding
interactions (Figure 8). Therefore, differences in the ring
substituents and their pattern of substitution were predicted to
largely dictate the different levels of inhibition of PfDHODH.
It is clear from the inhibitory activity observed for compounds
26-29, that substitution in just one of the aromatic rings did
not lead to improved inhibitory activity with respect to
compound 10. The combination of substituents in both aromatic
rings of the biaryl tail appears to be an important requirement
for maximal inhibitor activity with PfDHODH. Compounds
17-19 and 21 were all more active (by between 3- and 6-fold)
than compound 10 in PfDHODH, and they were the most active
of the series. These compounds all feature substituents in the
aromatic ring, which contains the cyanohydroxybutenamide
moiety, that are ortho to this position of cyanohydroxybutena-
mide attachment (for convenience, here labeled as ortho)
together with substituents in the terminal aromatic ring which

2690 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9
DaVies et al.
Figure 8. SPROUT-generated structures of (a) compound 18 showing the boundary surface of the PfDHODH cavity and (b) compound 17 showing
the boundary surface of the PfDHODH cavity.
are ortho-substituted with respect to the aryl attachment (here
termed ortho′). Other substitutions are labeled in a similar
manner.
Scheme 1
As is apparent from Table 3, the biaryl-substituted compounds
were generally more active against HsDHODH than PfDHODH.
As an example, compound 29 is more than 1000 times more
active against HsDHODH than PfDHODH. As previously
noted,8 in the PfDHODH crystal structure, the opening to the
binding cavity is relatively narrow because of the presence of
Met536, which may act to hinder entry of inhibitors such as
compound 29. However, in the human enzyme, the opening to
the binding cavity is wider than for PfDHODH and therefore
appears to facilitate entry of inhibitors.
Interestingly, as a further illustration of the importance of
molecular shape in enzyme selectivity, compound 21 was found
to be approximately 2-fold more selective for PfDHODH than
for HsDHODH. This appears to reflect the tendency for the di-
ortho′-substituted benzene ring present in 21 to maximize the
dihedral angle between the aromatic rings of the biphenyl tail,
resulting in a rather cylindrical-shaped molecule which would
appear to better occupy the cylindrical-like cavity in PfDHODH
as opposed to the somewhat flattened cavity in the human
enzyme. However, it seems likely that in those cases where
excellent selectivity is observed,23,24,26 this is due to strong
hydrogen bonding interactions to the residues in the headgroup
region of PfDHODH. As noted in Figure 5, one very significant
difference between PfDHODH and HsDHODH in this region
is that in HsDHODH, His56 is hydrogen bonded to Tyr147 and
hence is not readily available for hydrogen bonding to ligands.
This is also true for all the structures reported in this study. We
suggest that for PfDHODH, formation of a strong hydrogen
bond between a ligand and His185 might well be a requirement
for inhibitors to be highly selective for PfDHODH versus
HsDHODH. Some evidence for this hypothesis comes from
mutagenesis studies26-28 which show that mutation of His185
in PfDHODH leads to a very significant loss of inhibitory
activity (500-1000-fold increase in IC50) for some highly
selective inhibitors.
gave intermediate cyanoacetamide 33 in 57% yield, which was then
treated with sodium hydride followed by the appropriate acyl halide
(R1COCl) yielding the target compounds (60-79%).
The biphenyl 10 was synthesized according to the route shown
in Scheme 2.
Curtius rearrangement of acyl azide 35 (obtained from p-
biphenylcarboxylic acid 34) gave isocyanate 36, which was treated
immediately and without further purification with the sodium salt
of cyanoacetone 37 to give compound 10 in 81% yield.
For the synthesis of compounds 11-16, treatment of cyanoac-
etamide 38 with sodium hydride in THF followed by addition of
the appropriate acid chloride (R1COCl) afforded the target com-
pound (Scheme 3).
Compounds 17-29 were synthesized according to the general
scheme in Scheme 4.
In step A, substituted biphenylanilines were obtained via Suzuki
coupling of the appropriate substituted boronic acids 39 with
substituted 4-bromoanilines 40 using catalytic quantities of
Pd(OAc)2 and PPh3 in refluxing toluene for 24 h. The reaction could
also be carried out by stirring the reagents under microwave
irradiation at 160 °C for 40 min. Performing these Suzuki couplings
on a millimole scale under conditions of microwave irradiation was
advantageous over conventional heating due to reduced reaction
times; however, the microwave instrument was not amenable to
larger-scale reactions, and reaction conducted under standard reflux
conditions was the preferred method. The yields of products
(compounds 41a-41l; see the experimental section in the Sup-
porting Information) were similar using either method and were in
the range of 33-100%, with the exception being compound 41b
Chemistry
Compounds 4-9 were prepared using the route shown in Scheme
1.22
Treatment of acid chloride 31 (obtained from cyanoacetic acid
30 and used without purification) with p-trifluoromethylaniline 32

Human and P. falciparum Dihydroorotate Dehydrogenases
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9 2691
Scheme 2
Scheme 3
Scheme 4
Scheme 5
(Figure 9), obtained in only 3% yield. Other conditions were then
explored to improve the yield of compound 41b (see the next
section). In step B, 5-methylisoxazole-4-carboxylic acid 42 was
treated with thionyl chloride to give acid chloride 43, which was
then treated with the appropriate substituted aminobiphenyl
(41a-41l) to yield the corresponding amides in yields of 70-85%.
The O-N bond of the isoxazole moiety was then cleaved via
treatment with NaOH to afford the SPROUT-LeadOpt-designed
compounds in generally excellent yields (90-100%).
We suspected that the steric hindrance around the boron atom
in boronic acid 45 was giving rise to the poor yields observed in
the Suzuki coupling reactions of 45 with 46. Additionally, we
reasoned that for this case, there could be a competing process
Synthesis of Compound 41b. A range of conditions were
explored for the synthesis of compound 41b, and the best results
were obtained using supercritical (sc) CO2 (Scheme 5).

2692 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9
DaVies et al.
a table listing the elemental analyses, details of the experimental
procedures, and spectroscopic data for each compound, details of
the biological assays, and details of protein crystallography. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Tamta, H.; Mukhopadhyay, A. K. Biochemical targets for malaria
chemotherapy. Current Research & Information on Pharmaceutical
Science (CRIPS) 2003, 4, 6–9.
Figure 9. Trichloro-substituted biphenylaniline obtained in only 3%
yield using conventional Suzuki coupling conditions.
(2) Copeland, R. A.; Marcinkeviciene, J.; Haque, T. S.; Kopcho, L. M.;
Jiang, W.; Wang, K.; Ecret, L. D.; Sizemore, C.; Amsler, K. A.; Foster,
L.; Tadesse, S.; Combs, A. P.; Stern, A. M.; Trainor, G. L.; Slee, A.;
Rogers, M. J.; Hobbs, F. Helicobacter pylori-selective antibacterials
based on inhibition of pyrimidine biosynthesis. J. Biol. Chem. 2000,
275, 33373–33378.
(3) Morand, P.; Courtin, O.; Sautes, C.; Westwood, R.; Hercend, T.; Kuo,
E. A.; Ruuth, E. Dihydroorotate dehydrogenase is a target for the
biological effects of leflunomide. Transplant. Proc. 1996, 28, 3088–
3091.
(4) Palfey, B. A.; Bjornberg, O.; Jensen, K. F. Insight into the chemistry
of flavin reduction and oxidation in Escherichia coli dihydroorotate
dehydrogenase obtained by rapid reaction studies. Biochemistry 2001,
40, 4381–3290.
(5) Norager, S.; Jensen, K. F.; Bjornberg, O.; Larsen, S. E. coli
dihydroorotate dehydrogenase reveals structural and functional distinc-
tions between different classes of dihydroorotate dehydrogenases.
Structure 2002, 10, 1211–1223.
(6) Krungkrai, J. Purification, characterization and localization of mito-
chondrial dihydroorotate dehydrogenase in Plasmodium falciparum,
human malaria parasite. Biochim. Biophys. Acta 1995, 1243, 351–
360.
(7) Liu, S.; Neidhardt, E. A.; Grossman, T. H.; Ocain, T.; Clardy, J.
Structures of human dihydroorotate dehydrogenase in complex with
antiproliferative agents. Structure 2000, 8, 25–33.
Figure 10. Proposed N-carboxylated intermediate 47 from 4-bromo-
2-chloroaniline 46 and carbon dioxide.
involving coordination of aniline 46 to the palladium catalyst,
thereby resulting in “catalyst poisoning” and further reducing the
efficiency of the reaction. When a mixture of 45 and 46 along with
diisopropylethylamine, 1,1′-bis(biphenylphosphino)ferrocene, and
palladium bistrifluororoacetate was stirred in scCO2 under pressure
(12 MPa) at 80 °C over a period of 72 h, compound 41b was
obtained in 51% yield. However, when the equivalent reaction was
conducted in toluene, only starting materials were recovered from
the reaction mixture. In the presence of scCO2, it is proposed that
an N-carboxylated intermediate 47 is likely formed from aniline
46, thus preventing deactivation of the palladium catalyst via amine
coordination (Figure 10). Additionally as previously reported,29 the
use of scCO2 may enhance the stability of the palladium catalyst
and prevent decomposition.
(8) Hurt, D. E.; Widom, J.; Clardy, J. Structure of Plasmodium falciparum
dihydroorotate dehydrogenase with a bound inhibitor. Acta Crystallogr.
2006, D62, 312–323.
(9) Loffler, M.; Knecht, W.; Rawls, J.; Ullrich, A.; Dietz, C. Drosophila
melanogaster dihydroorotate dehydrogenase: The N-terminus is
important for biological function in vivo but not for catalytic properties
in vitro. Insect Biochem. Mol. Biol. 2002, 32, 1159–1169.
(10) Hansen, M.; Le Nours, J.; Johansson, E.; Antal, T.; Ullrich, A.; Loffler,
M.; Larsen, S. Inhibitor binding in a class 2 dihydroorotate dehydro-
genase causes variations in the membrane-associated N-terminal
domain. Protein Sci. 2004, 13, 1031–1042.
(11) McLean, J. E.; Neidhardt, E. A.; Grossman, T. H.; Hedstrom, L.
Multiple inhibitor analysis of the Brequinar and leflunomide binding
sites on human dihydroorotate dehydrogenase. Biochemistry 2001, 40,
2194–2200.
(12) Fox, R. I.; Herrmann, M. L.; Frangou, C. G.; Wahl, G. M.; Morris,
R. E.; Strand, V.; Kirschbaum, B. J. Mechanism of action for
leflunomide in rheumatoid arthritis. Clin. Immunol. 1999, 93, 198–
208.
(13) Peters, G. J.; Sharma, S. L.; Laurensse, E.; Pinedo, H. M. Inhibition
of pyrimidine de novo synthesis by DUP-785 (NSC 368390). InVest.
New Drugs 1987, 5, 235–244.
(14) Leban, J.; Saeb, W.; Garcia, G.; Baumgartner, R.; Kramer, B.
Discovery of a novel series of DHODH inhibitors by a docking
procedure and QSAR refinement. Bioorg. Med. Chem. Lett. 2004, 14,
55–58.
(15) Leban, J.; Kralik, M.; Mies, J.; Gassen, M.; Tentschert, K.;
Baumgartner, R. SAR, species specificity, and cellular activity of
cyclopentene dicarboxylic acid amides as DHODH inhibitors. Bioorg.
Med. Chem. Lett. 2005, 15, 4854–4857.
(16) Baumgartner, R.; Walloschek, M.; Kralik, M.; Gotschlich, A.; Tasler,
S.; Mies, J.; Leban, J. Dual binding mode of a novel series of DHODH
inhibitors. J. Med. Chem. 2006, 49, 1239–1247.
(17) Hurt, D. E.; Sutton, A. E.; Clardy, J. Brequinar derivatives and species-
specific drug design for dihydroorotate dehydrogenase. Bioorg. Med.
Chem. Lett. 2006, 16, 1610–1615.
Conclusions
We have applied the computer-aided molecular design
program SPROUT-LeadOpt to cocrystal structures of an existing
inhibitor in both human and Plasmodium DHODH. On the basis
of the resulting designed molecular scaffolds, a small library
of compounds was prepared and these molecules were screened
against both human and Plasmodium enzymes. Essentially all
of the designed inhibitors exhibited enhanced levels of inhibition
for both the Plasmodium and human DHODH enzymes com-
pared to that observed for the existing inhibitor 3. The majority
of inhibitors were selective for HsDHODH which may indicate
that variations in the hydrophobic tail of these ligands can at
best (21 and 23) give modest selectivity for PfDHODH versus
HsDHODH, perhaps by exploiting differences between the
hydrophobic surfaces of the proteins in this region. It is argued
that many (but not necessarily all) highly selective inhibitors
will exploit the possibility of a strong hydrogen bond to His185
in PfDHODH which is much more difficult for the correspond-
ing His56 in HsDHODH, because it would involve loss of an
intramolecular hydrogen bond to Tyr147. Additionally, several
X-ray cocrystal structures of the inhibitors in complex with
HsDHODH have been determined to a high resolution. The
detailed molecular interactions apparent within these structures
were generally in good agreement with those derived from the
design of these structures using SPROUT-LeadOpt.
(18) Kremmer, J. M. What I would like to know about leflunomide.
J. Rheumatol. 2004, 31, 1029–1031.
(19) Payne, D. J.; Gwynn, M. N.; Holmes, D. J.; Pompliano, D. L. Drugs
for bad bugs: Confronting the challenges of antibacterial discovery.
Nat. ReV. Drug DiscoVery 2007, 6, 29–40.
(20) Bostock, J. M.; Chopra, I.; Johnson, A. P.; Hesse, L.; Horton, J. R.;
Fishwick, C. W. G. Macrocyclic Inhibitors of the Bacterial Cell Wall
Enzyme MurD. Bioorg. Med. Chem. Lett. 2003, 13, 1557–1560. Ali,
M. A.; Bhogal, N.; Findlay, J. B. C.; Fishwick, C. W. G. The first de
novo designed antagonists of the human NK2 receptor. J. Med. Chem.
2005, 48, 5655–5658. Besong, G. E.; Bostock, J. M.; Stubbings, W.;
Acknowledgment. We gratefully acknowledge the support
of the EPSRC (studentship to M.D.) and the Wellcome Trust
(studentship to T.H.).
Supporting Information Available: A table listing the degree
of purity for all target compounds (area percent and retention time),

Human and P. falciparum Dihydroorotate Dehydrogenases
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 9 2693
Chopra, I.; Roper, D. I.; Lloyd, A. J.; Fishwick, C. W. G.; Johnson,
A. P. A novel de novo designed inhibitor of D-Ala-D-Ala ligase from
E. coli. Angew. Chem., Int. Ed. 2005, 44, 6403–6406.
Plasmodium falciparum Dihydroorotate Dehydrogenase. J. Biol. Chem.
2008, 283, 35078–35085.
(25) Kuo, P. Y.; Shie, T. L.; Chen, Y. S.; Lai, J. T.; Yang, D. Y. Enzyme
inhibition potency enhancement by active site metal chelating and
hydrogen bonding induced conformation-restricted cyclopropanecar-
bonyl derivatives. Bioorg. Med. Chem. Lett. 2006, 16, 6024–6027.
(26) Baldwin, J.; Michnoff, C. H.; Malmquist, N. A.; White, J.; Roth, M. G.;
Rathod, P. K.; Phillips, M. A. High-throughput screening for potent
and selective inhibitors of Plasmodium falciparum dihydroorotate
dehydrogenase. J. Biol. Chem. 2005, 280, 21847–21853.
(27) Heikkila, T.; Ramsey, C.; Davies, M.; Galtier, C.; Stead, A. M. W.;
Johnson, A. P.; Fishwick, C. W. G.; Boa, A. N.; McConkey, G. A.
Design and synthesis of potent inhibitors of the malaria parasite
dihydroorotate dehydrogenase. J. Med. Chem. 2007, 50, 186–191.
(28) Malmquist, N. A.; Gujjar, R.; Rathod, P. K.; Phillips, M. A. Analysis
of flavin oxidation and electron-transfer inhibition in Plasmodium
falciparum dihydroorotate dehydrogenase. Biochemistry 2008, 47,
2466–2475.
(21) Heikkila, T.; Thirumalairajan, S.; Davies, M.; Parsons, M. R.;
McConkey, G. A.; Fishwick, C. W. G.; Johnson, A. P. The first de
novo designed inhibitors of Plasmodium falciparum dihydroorotate
dehydrogenase. Bioorg. Med. Chem. Lett. 2006, 16, 88–92.
(22) Kuo, E. A.; Hambleton, A. B.; Kay, D. P.; Evans, P. L.; Matharu,
S. S.; Little, E.; McDowall, N.; Jones, C. B.; Hedgecock, C. J. R.;
Yea, C. M.; Chan, A. W. E.; Hairsine, P. W.; Ager, I. R.; Tully, W. R.;
Williamson, R. A.; Westwood, R. J. Synthesis, structure-activity
relationships, and pharmacokinetic properties of dihydroorotate de-
hydrogenase inhibitors: 2-Cyano-3-cyclopropyl-3-hydroxy-N-[3′-meth-
yl-4′-(trifluoromethyl)phenyl]propenamide and related compounds.
J. Med. Chem. 1996, 39, 4608–4621.
(23) Phillips, M. A.; Gujjar, R.; Malmquist, N. A.; White, J.; El Mazouni,
F.; Baldwin, J.; Rathod, P. K. Triazolopyrimidine-based dihydroorotate
dehydrogenase inhibitors with potent and selective activity against the
malaria parasite Plasmodium faliparum. J. Med. Chem. 2008, 51,
3649–3653.
(29) Clarke, D.; Ali, M. A.; Clifford, A. A.; Parratt, A.; Rose, P.; Schwinn,
D.; Bannwarth, W.; Rayner, C. M. Reactions in unusual media. Curr.
Top. Med. Chem. 2004, 4, 729–771.
(24) Patel, V.; Booker, M.; Kramer, M.; Ross, L.; Celatka, C. A.; Kennedy,
L. M.; Dvorin, J. D.; Duraisingh, M. T.; Sliz, P.; Wirth, D. F.; Clardy,
J. Identification and Characterization of Small Molecule Inhibitors of
JM800963T