Inhibition of human dhodh by 4-hydroxycoumarins, fenamic acids, and n-(alkylcarbonyl)anthranilic acids identified by structure-guided fragment selection

Article abstract of DOI:10.1002/cmdc.200900454

A strategy that combines virtual screening and structureguided selection of fragments was used to identify three unexplored classes of human DHODH inhibitor compounds: 4-hydroxycoumarins, fenamic acids, and N-(alkylcarbonyl) anthranilic acids. Structure-guided selection of fragments targeting the inner subsite of the DHODH ubiquinone binding site made these findings possible with screening of fewer than 300 fragments in a DHODH assay. Fragments from the three inhibitor classes identified were subsequently chemically expanded to target an additional subsite of hydrophobic character. All three classes were found to exhibit distinct structure-activity relationships upon expansion. The novel N-(alkylcarbonyl)anthranilic acid class shows the most promising potency against human DHODH, with IC50 values in the low nanomolar range. The structure of human DHODH in complex with an inhibitor of this class is presented.

Full text of DOI:10.1002/cmdc.200900454

MED
DOI: 10.1002/cmdc.200900454
Inhibition of Human DHODH by 4-Hydroxycoumarins,
Fenamic Acids, and N-(Alkylcarbonyl)anthranilic Acids
Identified by Structure-Guided Fragment Selection
Ingela Fritzson,*[a] Bo Svensson,[b] Salam Al-Karadaghi,[c] Bjçrn Walse,[b] Ulf Wellmar,[a]
Ulf J. Nilsson,[c] Dorthe da GraÅa Thrige,[a] and Stig Jçnsson[d]
A strategy that combines virtual screening and structure-
guided selection of fragments was used to identify three unex-
plored classes of human DHODH inhibitor compounds: 4-hy-
droxycoumarins, fenamic acids, and N-(alkylcarbonyl)anthranilic
acids. Structure-guided selection of fragments targeting the
inner subsite of the DHODH ubiquinone binding site made
these findings possible with screening of fewer than 300 frag-
ments in a DHODH assay. Fragments from the three inhibitor
classes identified were subsequently chemically expanded to
target an additional subsite of hydrophobic character. All three
classes were found to exhibit distinct structure–activity rela-
tionships upon expansion. The novel N-(alkylcarbonyl)anthra-
nilic acid class shows the most promising potency against
human DHODH, with IC50 values in the low nanomolar range.
The structure of human DHODH in complex with an inhibitor
of this class is presented.
Introduction
Dihydroorotate dehydrogenase (DHODH, EC 1.3.5.2) is a mito-
chondrial enzyme involved in the rate-limiting step of the
de novo biosynthesis of pyrimi-
dine bases.[1] In humans, it has
been demonstrated that resting
T-lymphocytes meet their meta-
bolic requirement of pyrimidines
through a salvage pathway, but
for proliferating T-lymphocytes,
de novo synthesis of pyrimidines
is crucial.[2] Stimulated T-cells
switch on their de novo synthet-
ic route to produce additional
precursors necessary for RNA
and DNA synthesis and other
metabolic activities needed for
clonal expansion. Inhibition of
DHODH decreases cellular levels
of rUMP and arrests proliferating
T-cells in the G1 or early S phase
it cannot be ruled out that tyrosine kinase inhibition also con-
tributes to the in vivo activity of A771726.[8] Thus, it is still of
Figure 1. Structures of known inhibitors of human DHODH.
of the cell cycle, which results in
suppression of the immune
system.[3] Inhibitors of human
[a] I. Fritzson, Dr. U. Wellmar, Dr. D. da GraÅa Thrige
Active Biotech Research AB, Box 724, 220 07 Lund (Sweden)
Fax: (+46)46 19 11 00
DHODH have potential as drugs against autoimmune diseases
and cancer as well as in the treatment of transplant rejection.
The most widely known human DHODH inhibitors brequinar
(1) and A771726 (2, the active metabolite of leflunomide) have
been developed for the treatment of cancer and rheumatoid
arthritis (RA), respectively (Figure 1).[4] However, both com-
pounds have suboptimal properties. Brequinar failed in clinical
trails due to a limited therapeutic window.[5] A771726 has a
disadvantageous plasma half-life of approximately two weeks
in humans and is associated with liver toxicity.[6,7] Furthermore,
E-mail: ingela.fritzson@activebiotech.com
[b] Dr. B. Svensson, Dr. B. Walse
SARomics AB, Box 724, 220 07 Lund (Sweden)
[c] Prof. S. Al-Karadaghi, Prof. U. J. Nilsson
Lund University, Box 124, 221 00 Lund (Sweden)
[d] Dr. S. Jçnsson
MIP Technologies, Box 737, 220 07 Lund (Sweden)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.200900454.
608
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 608 – 617

Inhibitors of Human DHODH
interest to find potent and selective inhibitors of human
DHODH for use as immunosuppressants. The first crystallo-
graphic structures of human DHODH in complex with a brequi-
nar analogue and in complex with the active metabolite of
A771726 were published in August 2000.[9] In these experi-
ments brequinar and A771726 were found to bind to the site
at which the cofactor ubiquinone is believed to bind, and not
to the substrate binding site. After publication of the three-di-
mensional structures, additional compounds that inhibit the
human enzyme in the nanomolar or low micromolar range
were reported (examples 3–6 in Figure 1).[10] Most compounds
share one or more common motifs with brequinar: 1) a bi-
phenyl moiety; 2) a quinoline portion, sometimes exchanged
for naphthalene; or 3) a carboxylic acid group on quinoline or
other aromatic rings. Considering their dissimilarity to the sub-
strate orotate and their structural resemblance to brequinar, all
these compounds probably target the ubiquinone binding
site.
next to the redox cofactor flavin mononucleotide (FMN). Struc-
tural information available from the DHODH–A771726 or
DHODH–brequinar crystal complexes allow the identification
of several electron donor/acceptor and ionisable interactions
between potential ligands and the inner enzyme region.[9] Vir-
tual screening along with visual inspection of the inner subsite
structure was used to guide us in a selection procedure where-
by fragments were selected for the DHODH screen. This proce-
dure allowed us to substantially decrease the number of frag-
ments to be tested relative to the commonly used fragment
screening, in which libraries with several thousand fragments
may be screened against a given target.[12]
Results and Discussion
The general procedure for selection of a small fragment library
targeting the inner DHODH subsite was divided into two parts
(Figure 3): virtual screening and inner subsite-guided extraction
of a scaffold. The fragments identified in this procedure were
then screened at high concentration in the DHODH enzyme
assay.
The aim of our work was to identify, with limited screening
effort, novel human DHODH inhibitors that could serve as lead
structures for subsequent optimisation as drug candidates. The
suggested ubiquinone binding site of DHODH was chosen for
the design of the new compounds (Figure 2). This channel-like
binding site, as described by Liu et al.[9] and others, is located
in a region of the protein that is expected to be involved in
membrane association.[11] The membrane association creates a
highly hydrophobic environment around the entrance to the
binding site. This is also reflected by the hydrophobic charac-
ter of the beginning (entrance) of the channel, which is sur-
rounded mainly by hydrophobic amino acid side chains
(Figure 2). From inspection of the crystal complexes, it can be
concluded that this part of the binding site offers hydrophobic
interactions with nonpolar groups such as biphenyl in brequi-
nar or para-(trifluoromethyl)phenyl in A771726.[9] The channel
ends in a narrow cavity that contains the protein redox site
Figure 3. Schematic summary of the procedure used for the selection of
fragments for the DHODH screen.
Virtual screening
Virtual screening was based on the crystallographic coordi-
nates of human DHODH in complex with brequinar and
A771726 (PDB accession numbers 1D3G and 1D3H, respective-
ly).[9] A notable difference between the inhibitor binding sites
in the two structures is the presence of one buried water mol-
ecule in the case of brequinar and two buried water molecules
in the case of A771726.
Pharmacophores were created by using the structure-based
focusing method.[13] A LUDI interaction map was created for
the binding site in the absence of inhibitor but in the presence
of the buried water molecules.[14] This allowed the identifica-
tion of hydrophobic, acceptor, donor, and ionisable groups
available for interaction with inhibitors. The interaction site
vectors and points were subjected to clustering, and the re-
maining pharmacophore features were selected and combined
into twelve different feature combinations. All pharmaco-
phores created contained a total of four or five features, that
is, hydrophobic groups, hydrogen bond acceptors, hydrogen
bond donors, and a negative ionisable group. The default defi-
nitions of the sizes of the pharmacophore features were used.
The query also included around 200 excluded volume spheres
of 1.2 ꢁ radius. These were positioned on the protein heavy
atoms in the vicinity of the binding site.
Figure 2. The ubiquinone binding site: The binding mode of brequinar
(shown in grey) from crystallographic structure data.[9] Hydrophobic residues
and surfaces are coloured green, and water molecules give grey surfaces.
The inner subsite (i.e., proximal redox site) that was the target for fragment
screening has a high proportion of polar residues and surfaces (coloured
cyan). The Arg136 residue is coloured blue. The redox cofactor flavin mono-
nucleotide (FMN) is shown in magenta.
ChemMedChem 2010, 5, 608 – 617
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
609

I. Fritzson et al.
MED
The Catalyst-formatted databases NCI2000 and May-
bridge2001 from Accelrys were screened using the pharmaco-
phore models created. The conformers that superimposed fa-
vourably with the pharmacophores and that did not overlap
with the excluded volumes spheres were retrieved as hits. The
hits were subjected to consensus scoring based on the scoring
functions LigScore, LUDI, PLP, and PMF using LigandFit soft-
ware.[13,15–18] Visual inspection of the proposed interaction with
the protein was used together with the consensus score to
select fragments for screening and use as scaffolds.
tions and the geometry relative to the benzoic acid scaffold
can be observed from superposition of the DHODH crystal
structures with bound A771726 and brequinar (Figure 4).
Selection of fragments
The inner subsite has a high proportion of polar residues
(Figure 2), and it is expected that interactions in this part of
the enzyme are crucial for the determination of target specifici-
ty and inhibitor affinity. Accordingly, it was decided that one
criterion for a fragment structure should be the presence of a
minimum of two polar functionalities: hydrogen bond donors,
hydrogen bond acceptors, and/or negatively ionisable groups.
The fragment should also fit the inner subsite, having a suita-
ble size and complementary volume. Starting from benzoic
acid, a fragment library was selected by combining up to four
types of structural modification (Figure 4). The first modifica-
tion was the addition of either a functional group or function-
ality included in the aromatic ring of the benzoic acid scaffold
that could create a second polar interaction site. For several
reasons, Tyr356 was identified as the best target for the
second polar interaction. Firstly, it seems likely that ubiquinone
uses this residue to bind to the inner subsite because, as with
Arg136, Tyr356 is invariant in family 2 DHODHs.[9] Secondly,
the short distance (2.8 ꢁ) between one of the oxygen atoms of
bound A771726 and Tyr356 in the crystallographic structure
clearly indicates that the two groups form a hydrogen bond
(see Figure 4). Thirdly, the three-dimensional structure of the
inhibitor-free enzyme shows that Tyr356 is not involved in in-
ternal hydrogen bond interactions.[20] Based on these data and
from visual inspection of the enzyme structure, we could con-
clude that substituents at position 3 of the benzoic acid
moiety should have the best geometry for interaction with
Tyr356. Small groups such as methoxy, hydroxy, nitro, and car-
bonyl were mainly selected to
Extraction of the benzoic acid scaffold from the brequinar
structure
Brequinar is one of the most potent inhibitors of human
DHODH currently known (IC50 =6 nm).[4] It is about 100-fold
more potent than A771726 in vitro,[11b,19] and this justified the
choice of brequinar as the starting point for scaffold extraction.
Benzoic acid, a substructure of brequinar, formed the base for
the selection of the major part of the inner subsite-guided
fragments (Figure 4). Extraction of benzoic acid was based on
analysis of the interactions of the inhibitor with the protein.
The salt bridge between the carboxylic acid of brequinar and
the guanidine group of Arg136 was considered to be an im-
portant feature for inhibition of DHODH by brequinar; thus,
the carboxylic acid was preserved in the scaffold extraction. On
the other hand, the biphenyl portion of brequinar was omitted
because only the inner subsite was the target for the fragment
screen, while the quinoline ring was simplified to a benzene
ring. This extraction procedure resulted in the choice of benzo-
ic acid as a scaffold for fragment selection. This scaffold offered
many diverse and commercially available derivatives and was
expected to generate hits that would, for the most part, be
reasonably easy to synthesise and modify. Two target interac-
probe these interactions. Other
possible interaction targets were
Glu47, Tyr147, and His56.[21]
The second modification in-
volved exchange of the benzoic
acid aromatic ring for naphtha-
lene or a heteroaromatic mono-
or bicyclic system. About 35% of
the compounds selected were
based on a bicyclic ring. To fill
the same volume in the enzyme
as the quinoline ring of brequi-
nar, the second ring was at-
tached mainly at positions 2 and
3 of the benzoic acid.
The third modification in-
volved exchange of the carbox-
Figure 4. Extraction of the benzoic acid scaffold from the brequinar structure and example of modification to
select new fragments. At left is the overlay of bound A771726 (grey) with brequinar (green, with the benzoic acid
portion in dark green) from the superimposed crystallographic structure of DHODH together with the relative po-
sitions of Arg136 and Tyr356.[9] Starting from the benzoic acid scaffold, one of four types of modification is made
in each example to form new fragments. Several of the modifications 1)–4) are then combined to produce a di-
verse set of fragments shown at right.
ylic acid group. A nitro or hy-
droxy group was frequently
used to model the interaction
with Arg136. Also, cyano groups
(as in A771726), carboxamides,
610
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 608 – 617

Inhibitors of Human DHODH
or methyl esters were examined for their ability to form hydro-
gen bonds with Arg136.
Table 1. Selected hits from screening of chemical fragments in a DHODH
membrane assay.
The fourth modification included random addition of one or
several small chemical entities such as methyl, methoxy, hy-
droxy, amine, halogen, and phenyl, including the insertion of
methylene groups between the phenyl and the carboxylic acid
functionality in the benzoic acid scaffold (see Figure 4).
In parallel with the target-orientated selection from benzoic
acid, virtual screening hits and derivatives thereof were select-
ed. The proposed inner subsite binding parts of some of the
hits from the virtual screen were used as scaffolds and treated
according to the scheme used for the benzoic acid scaffold
above. In both cases, the choice of derivatives was restricted
by commercial availability. The final result of the total selection
process was a small, target-directed but diverse set of frag-
ments.
DHODH Inhibition [%][a]
Compound
1 mm
0.1 mm
9
46
18
10
11
12
71
38
85
40
n.d.[b]
101
13
14
15
n.d.[b]
100
68
Evaluation in a human DHODH enzyme assay
A DHODH enzyme assay was set up and validated using mito-
chondrial membrane preparations from the human lymphoma
cell line U937 for the initial screening of fragments.[22] To allow
detection of low-affinity binders, inhibition of DHODH was
measured at concentrations of up to 1 mm, which, in many
cases, were close to the given compound’s solubility limit. In
total, 265 fragments were tested, half of which were selected
from databases with commercially available compounds and
the rest from an in-house collection of compounds. The molec-
ular weight varied between 123 and 384 Da, and the average
of the predicted logP values was 2.2.[23] A list of the most
potent hits is presented in Table 1.
91
n.d.[b]
n.d.[b]
96
16
17
18
79
78
14
106
Three interesting fragment types were identified from the
low-affinity screen: 4-hydroxycoumarin derivatives (compounds
9 and 10), fenamic acid (compound 11), and derivatives of N-
(alkylcarbonyl)anthranilic acids (compounds 12–16, Table 1),
where the N-(alkylcarbonyl)anthranilic acids originate from the
virtual screen, and the others originate from the selection
based on the benzoic acid scaffold. Figure 5 shows the core
fragment structure for each of these three unexplored DHODH
inhibitor classes. The known compound dicoumarol (7,
Figure 1) shares the 4-hydroxycoumarin motif with compounds
9 and 10; dicoumarol was recently found to impair pyrimidine
biosynthesis at the DHODH-catalysed step.[24] The fenamic acid
derivative redoxal (8), originally found in a tumour cell screen,
is the only substance of these classes with a documented
value for DHODH inhibition (IC50 =45 nm in recombinant
human DHODH), although no SAR has been reported for
it.[11b,25,26] Recently, compounds related to fenamic acid have
been patented as DHODH inhibitors.[27]
41
[a] Assays were run as single samples on two different occasions, and
values represent the mean of the duplicate measurements carried out at
the indicated compound concentrations. [b] Not determined.
Figure 5. Fragment structures of three compound classes found to inhibit
human DHODH activity.
Chemical expansion of fragment hits
Besides the three found classes, two additional fragments
from the benzoic acid scaffold selection, 17 and 18, were iden-
tified as hits. The potency of compounds 17 and 18 could be
explained by their close resemblance to the quinoline–carbox-
ylic acid moiety of brequinar. Indeed, a hydroxy derivative of
brequinar analogous to fragment 17 has been reported as an
anti-arthritic and immunosuppressant in vivo.[28]
The objective of the chemical expansion was to determine the
potency of the three inhibitor classes identified (Figure 5)
before initiation of a structure optimisation phase that would
include extensive efforts involving synthesis. The strategy was
to expand the fragments of these compound classes with non-
polar moieties pointing towards the hydrophobic part of the
ubiquinone binding site. An analysis of possible geometries of
ChemMedChem 2010, 5, 608 – 617
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
611

I. Fritzson et al.
MED
bound fragments using the crystallographic structure of the
enzyme was conducted to decide on anchor points for chemi-
cal expansion. A few fragments of each class were then
merged with a phenyl- or biphenyl-containing moiety at the
anchor point, and the derivatives were screened to determine
the IC50 value for DHODH inhibition (compounds 19–32,
Tables 2–4). As expected, successive filling of the hydrophobic
pocket gives higher calculated logP values. Another general
observation was that just expanding the fragment hits with a
biphenyl moiety, as in brequinar, gave markedly lower IC50
values for all three classes (Table 2–4; compounds 23, 26, 27,
and 30).
the lactone oxygen atoms, and either part could interact with
the positively charged Arg136 (possibly mediated by water). It
seems likely that the hydrophobic groups at position 3 of the
coumarins bind in the outer hydrophobic subsite of the
enzyme, and that this restricts the possible binding conforma-
tions of the coumarin portion; this suggests a binding mode
that places the lactone part closest to Arg136 as illustrated for
the particularly potent derivative 23 (IC50 =230 nm; Figure 6a).
If the coumarin part was rotated 1808, placing the hydroxy
groups closest to Arg136, overlap between the biphenyl
groups from brequinar and coumarin 23 would not be possi-
ble. It was observed that 4,5-dihydroxycoumarins 10 and 20
are more potent and have higher ligand efficiencies (see
Table 2) than the corresponding 4-hydroxycoumarins 19 and
21, perhaps due to the possibility that the 5-hydroxy group
may be involved in additional polar interactions, for example
with Tyr356.
Table 2. Inhibition of recombinant DHODH by 4-hydroxycoumarin deriva-
tives.
Compound
IC50 [mm]
LE[a]
0.21
QlogP[b]
HL-60 leukaemia cells treated with the reference compound
dicoumarol (7) were recently found to accumulate in the S
phase of the cell cycle, due to impairment of pyrimidine bio-
synthesis at the DHODH step.[24] We determined the IC50 to be
6.6 mm, which confirms dicoumarol as a DHODH inhibitor.
10
120
1.96
19
20
21
22
>2000
11
0.15
0.25
0.16
0.16
2.42
2.17
2.81
3.81
Fenamic acid derivatives
Fenamic acid derivatives also originate from the fragment se-
lection based on benzoic acid as scaffold. As mentioned
above, the carboxylic acid may form a salt bridge with Arg136.
With such a salt bridge, compound 11 has two possible orien-
tations in the binding site. Either the anilinic phenyl group
points inwards in the binding site or it points in the opposite
direction, towards the outer and more lipophilic subsite. In
compounds 24, 26, 27, and 28 (Table 3), possible hydrophobic
interactions were explored in greater detail. Because the anilin-
ic phenyl group in compound 11 has no polar interaction
points, the phenyl ring was expected to be directed towards
the outer part of the channel. In this binding mode, there is no
interaction with Tyr356; however, it is possible to accomplish
this with a small substituent. A close look at Figure 4 shows
that substituents at position 5 of fenamic acid should have a
good fit for interaction with Tyr356. To investigate this, two
methoxy analogues (compounds 25 and 27, Table 3) were syn-
thesised. The SAR of these compounds supports the hypothe-
sis that the phenyl ring of compound 11 is directed towards
the outer part of the channel. In compounds 27 and 28,
parent fragment 11 was modified with an additional phenyl
group to increase the hydrophobic interaction and with a me-
thoxy group to introduce interaction with Tyr356. This marked-
ly improved the inhibition of DHODH and resulted in IC50
values of 39 and 81 nm for compounds 27 and 28, respectively,
as compared with 38 mm for fragment 11. Subsequently, as
part of this work, human DHODH was crystallised in complex
with compound 28, and the structure was published.[20] The
binding mode of compound 28, shown in Figure 6b, is consis-
tent with the assumed interactions.
860
190
23
7
0.23
6.6
0.27
0.21
3.50
1.79
[a] Ligand efficiencies are calculated as pIC50 divided by the number of
heavy atoms. [b] Calculated logP in octanol/H2O.[23]
4-Hydroxycoumarin derivatives
4-Hydroxycoumarin derivatives originate from the benzoic acid
scaffold. They contain a bicyclic system that can be regarded
as a substitute for the quinoline ring of brequinar. In these
compounds, the carboxylic group of benzoic acid is exchanged
with a hydroxy group. The 4-hydroxycoumarins were expand-
ed at position 3 of coumarin (compounds 19, 20, 21, 22, and
23, Table 2) and evaluated for inhibition of DHODH together
with the known compound dicoumarol (7) as a reference. One
interesting feature is the acidity of 4-hydroxycoumarin, with a
pKa value of 5.1.[29] The acidity indicates that the coumarin
compounds are negatively charged at neutral pH. The negative
charge is delocalised to the enolic hydroxy group as well as
612
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 608 – 617

Inhibitors of Human DHODH
Figure 6. a) Proposed binding mode for coumarin derivative 23. The lactone part is placed close to Arg136 to facilitate polar interactions. The biphenyl
groups from brequinar (pink) and coumarin derivative 23 (green) occupy almost the same positions in the suggested binding mode; van der Waals surfaces
and residues are from the crystallographic structure of DHODH in complex with brequinar.[9] b) Binding site for fenamic acid 28 (blue) in the crystal complex
with human DHODH.[20] c) Binding site for anthranilic acid 30 (yellow) in the crystal complex with human DHODH. d) Binding site for anthranilic acid 30
(yellow) in the crystal complex with human DHODH, shown with electron density. The amino acid residues are shown in cyan, and a 3FoÀ2Fc electron density
map contoured at 1.0 s is displayed in blue mesh. Selected residues within 5 ꢁ of the bound inhibitor are shown in all four panels a)–d); images were gener-
ated with the PyMOL software package (The PyMOL Molecular Graphics System, DeLano Scientific, San Carlos, CA (USA), 2002).
N-(Alkylcarbonyl)anthranilic acid derivatives
Table 3. Inhibition of recombinant DHODH by fenamic acid derivatives.
The N-(alkylcarbonyl)anthranilic acids (compounds 12–16,
LE[a]
QlogP[b]
Table 1) were derived from the virtual screen. The results indi-
cate a binding mode in which the negatively charged carboxyl-
ate group of the anthranilic acid interacts with Arg136, and
the carbonyl oxygen of the amide group serves as a hydrogen
bond acceptor in an interaction with Tyr356. Comparison of
the IC50 values for each of the compounds 12–16 reveals that
position 5 of the anthranilic acid should be a good anchor
point for substitution with a lipophilic group directed towards
the outer hydrophobic subsite. Thus, compounds 29 and 30
(Table 4) were synthesised to fill the hydrophobic subsite with
a considerable gain in potency. Optimisation of the novel N-(al-
kylcarbonyl)anthranilic acid class, in which additional nonpolar
moieties were probed in combination with various alkyl sub-
stituents in the alkylcarbonyl portion, resulted in compounds
with further improved potency. Here, an ethylcarbonyl sub-
stituent on the anthranilic nitrogen atom proved beneficial for
potency. The optimised anthranilic acids 31 and 32 (with IC50
values of 33 and 15 nm, respectively, Table 4) have IC50 values
that are four orders of magnitude lower than that of parent
fragment 12 (190 mm), and compound 31 has the highest
ligand efficiency (0.34, Table 4) found in this study.
Compound
IC50 [mm]
11
24
38
0.28
3.68
19
25
0.24
0.26
5.24
3.57
25
26
0.19
0.31
4.89
27
28
0.039
0.31
0.28
4.99
4.94
0.081[20]
DHODH was crystallised in complex with compound 30, and
a 1.9 ꢁ resolution data set was collected (coordinates deposit-
ed in the Protein Data Bank; PDB ID: 2WV8). The binding
[a] Ligand efficiencies are calculated as pIC50 divided by the number of
heavy atoms. [b] Calculated logP in octanol/H2O.[23]
ChemMedChem 2010, 5, 608 – 617
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
613

I. Fritzson et al.
MED
Conclusions
Table 4. Inhibition of recombinant DHODH by N-(alkylcarbonyl)anthra-
nilic acid derivatives.
A fragment screen was performed with a diverse selection of
fragments that target the inner subsite of human DHODH. The
selection of fragments was guided by structural analysis of
crystallised human DHODH and the binding modes of two in-
hibitors, A771726 and brequinar. The selection process gave a
library of fragments based on either virtual screening hits or a
benzoic acid scaffold. From fragment screening of only 265 se-
lected entities, three unexplored DHODH inhibitor classes were
identified: 4-hydroxycoumarins, fenamic acids, and N-(alkylcar-
bonyl)anthranilic acids.
Compound
IC50 [mm]
LE[a]
0.29
QlogP[b]
12
29
190
1.56
8.4
0.27
0.27
2.69
4.01
The knowledge-based selection of fragments not only de-
creased the number of compounds to screen in the DHODH
assay, but also guided the chemical expansion of the fragment
hits. Upon expansion of the fragments with hydrophobic ele-
ments targeting the outer hydrophobic subsite of the enzyme,
the compound classes identified exhibited distinct SAR with
up to 10000-fold difference in potency between expanded
fragments and the parent fragment. With gradual filling of the
hydrophobic subsite, higher calculated logP values were ob-
served, along with a gain in potency. Of the three DHODH in-
hibitor classes identified, only dicoumarol (7) and the fenamic
acid redoxal (8) are known to affect pyrimidine biosynthesis or
inhibit DHODH (Figure 1). Redoxal (8) is a known DHODH in-
hibitor (IC50 =45 nm toward recombinant human DHODH), al-
though no SAR has been reported.[11b,26] Recently dicoumarol
(7) was found to cause accumulation of treated HL-60 leukae-
mia cells in the S phase due to impairment of pyrimidine bio-
synthesis at the DHODH step.[24] An IC50 value of 6.6 mm for di-
coumarol in our recombinant DHODH assay confirms these
findings. In addition, a ~30-fold more potent coumarin, com-
pound 23, was found in the structure–activity evaluation. The
coumarin derivates, however, were not as potent as the other
two inhibitor classes found; that is, the best coumarin 23 was
about tenfold less potent than the fenamic acids 27 and 28
and the anthranilic acids 31 and 32. A ligand efficiency of 0.3
(calculated as pIC50 divided by the number of heavy atoms) is
sometimes considered as a cutoff value for an efficient binder,
and four compounds (26, 27, 31, and 32) of both the fenamic
acid class and the N-(alkylcarbonyl)anthranilic acid class fulfill
this criterion. The N-(alkylcarbonyl)anthranilic acids represent a
new class of DHODH inhibitors, and within this class the
lowest IC50 value was measured: 15 nm for compound 32.
In this study, where a wealth of structural information could
be found on both target protein and different ligands, it was
most likely a good idea to complement virtual screening with
knowledge-based selection of compounds to increase the
number of screening hits. Because the N-(alkylcarbonyl)anthra-
nilic acids derive from the virtual screen and the fenamic acids
originate from the benzoic acid scaffold, and the classes are
almost equally potent, both selection procedures can be re-
garded as valuable.
30
31
32
0.21
0.033
0.015
0.34
0.30
3.98
4.30
[a] Ligand efficiencies are calculated as pIC50 divided by the number of
heavy atoms. [b] Calculated logP in octanol/H2O.[23]
mode for compound 30 (Figure 6c) can be referred to as bre-
quinar-like in that it occupies basically the same position and
involves interaction with Arg136 with a carboxylic acid group.
Thus, the structure also confirms our fragment-based design
strategy. The carboxylic acid of compound 30 is rotated by
~308 relative to the position of the previously published com-
plex between brequinar and human DHODH; this is possible
because, unlike brequinar, its rotation is not restricted by a
methyl substituent at the ortho position of the aromatic ring.[9]
This facilitates more optimal interactions with Arg136 and
Gln47, which gets closer to the bound ligand by 0.5–1.0 ꢁ. The
distance between the carboxylic oxygen atoms and Arg136
and Gln47 is 2.8 and 2.9 ꢁ, respectively. Additional polar inter-
actions are formed between the carboxylic acid of the ligand
and the bound water molecule, and also, to a lesser extent,
with the carbonyl group of Thr360 (not shown in Figure 6).
One of the key interactions of the anthranilic acids is the hy-
drogen bond that is formed between the carbonyl oxygen of
the amide group and the side chain of Tyr356 (2.6 ꢁ), which
probably contributes to the high potency of the anthranilic
acid class. The interactions between both the carboxylic acid
and the alkylcarbonyl functionalities on compound 30 with the
enzyme give rise to a shift in the position of the biphenyl por-
tion, ~1 ꢁ further out towards the entrance of the binding site
relative to brequinar, but no conformational differences in that
part of the protein are observed.
Although the best compounds found have chemical features
that are similar to those of brequinar, we found a second hy-
drogen bond, to Tyr356, which could be important for target
specificity. The calculated logP value is about one unit lower
614
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 608 – 617

Inhibitors of Human DHODH
([D6]DMSO): d=11.8 (bs, 1H), 8.01 (dd, J=8.0, 1.1 Hz, 1H), 7.62–
7.66 (m, 3H), 7.47 (d, J=8.0 Hz, 2H), 7.39 (m, 2H), 3.98 (s, 2H);
13C NMR ([D6]DMSO): d=163.7, 161.9, 153.0, 145.7, 132.9, 129.8,
127.5 (q, J=32 Hz), 126.0 (q, J=3.8 Hz) 125.3 (q, J=272 Hz), 124.8,
124.2, 117.2, 117.1, 104.1; Anal. calcd for C17H11F3O3: C 63.76, H
3.46, found: C 63.53, H 3.57.
for the best anthranilic acid (ClogP=4.30 for compound 32)
relative to brequinar (ClogP=5.58), which indicates that polar
interactions contribute more to the potency of this class. Both
the second hydrogen bond and the decreased ClogP value are
features that could improve the pharmaceutical properties and
contribute to a wider therapeutic window of our compound
relative to brequinar.
4,5-Dihydroxy-3-(4-biphenyl)coumarin (23): The starting material
2-biphenyl-4-ylmalonic acid diethyl ester was prepared according
to a method already described.[32] Compound 23 was synthesised
from resorcinol and 2-biphenyl-4-ylmalonic acid diethyl ester ac-
cording to the procedure for compound 10, to provide the prod-
uct as an ivory-white powder (170 mg, 5%). 1H NMR ([D6]DMSO):
d=7.68 (d, J=7.2 Hz, 2H), 7.57–7.65 (m, 4H), 7.46 (t, J=7.7 Hz,
2H), 7.33–7.41 (m, 2H), 6.80 (d, J=8.2 Hz, 1H), 6.68 (d, J=8.2 Hz,
1H); 13C NMR ([D6]DMSO): d=166.7, 162.9, 157.6, 154.2, 141.1,
138.8, 133.24, 133.20, 132.1, 129.8, 128.1, 127.4, 126.4, 110.8, 107.8,
105.6, 102.2; Anal. calcd for C21H14O4: C 76.36, H 4.27, found: C
76.02, H 4.22.
The most potent expanded compounds in each class are
promising lead structures for further optimisation as drug can-
didates. The hydrophobic part of the lead compounds were
not subjected to fragment screening, and this part has a par-
ticularly high potential to be modified further, to increase po-
tency and optimise the drug properties.
Experimental Section
Chemistry
5-Methoxy-2-(phenylamino)benzoic acid (25): 2-Bromo-5-methoxy-
benzoic acid (690 mg, 3.0 mmol) was reacted with aniline (560 mg,
6.0 mmol) in the presence of copper powder (20 mg) and K2CO3
(210 mg, 1.5 mmol) in N,N-dimethylformamide (DMF; 5 mL) at
1508C for 2 h. The cooled mixture was added to 5m HCl (5 mL)
and ice. After filtration, the crude product was recrystallised from
The following commercially available compounds were included in
DHODH screening (the bold numbers in parentheses correspond
to compound numbers in Tables 1–4): dicoumarol (7), N-acetylan-
thranilic acid (12), 3-methyl-6-[(2,2,2-trifluoroacetyl)amino]benzoic
acid (13), 2-methyl-6-[(2,2,2-trifluoroacetyl)amino]benzoic acid (14),
3-iodo-6-[(2,2,2-trifluoroacetyl)amino]benzoic acid (15), 4-hydroxy-
3-phenylcoumarin (19), and 3-benzyl-4-hydroxy-2H-1-benzopyran-
2-one (21) were from Maybridge; 2-acetamido-5-bromobenzoic
acid (16) was from Lancaster; 4-hydroxy-3-nitrocoumarin (9), 2-hy-
droxynaphthoic acid (17), and indole-4-carboxylic acid (18) were
from Aldrich; N-(3-trifluoromethylphenyl)anthranilic acid (24) (flufe-
namic acid) was from Acros; and N-phenylanthranilic acid (11) was
from Fluka. 3-Benzyl-4,5-dihydroxycoumarin (20) was prepared as
described previously.[30] The synthesis of compound 28 is described
in reference [20].
1
EtOH/H2O to give a yellow powder (360 mg, 49%). H NMR (CDCl3/
TFA): d=10.9 (bs, 1H), 7.60 (d, J=2.9 Hz, 1H), 7.40 (t, J=7.6 Hz,
2H), 7.32–7.37 (m, 1H), 7.25–7.28 (m, 2H), 7.17–7.25 (m, 1H), 7.15
(dd, J=2.9, 9.1 Hz, 1H), 3.87 (s, 3H); 13C NMR (CDCl3/TFA): d=
173.3, 153.9, 140.9, 140.3, 130.2, 125.7, 124.7, 122.7, 120.1, 115.3,
114.0, 56.4; Anal. calcd for C14H13NO3: C 69.12, H 5.39, N 5.76,
found: C 68.84, H 5.31, N 5.79.
2-[(4-Biphenyl)amino]benzoic acid (26): 2-Bromobenzoic acid was
reacted with 4-bromoaniline according to the procedure for com-
pound 25. The resulting 2-[(4-bromophenyl)amino]benzoic acid
(230 mg, 0.8 mmol) was subsequently coupled using the Suzuki re-
action by stirring it at room temperature for 48 h with phenylbor-
onic acid (100 mg, 0.82 mmol), Pd(OAc)2 (2 mg), and Na2CO3
(250 mg, 2.4 mmol) in H2O/1,2-dimethoxyethane (50 mL:12 mL).
The solution was filtered and then acidified with dilute HCl (1m,
5 mL). The product was collected by filtration as an ivory-white
NMR spectra were recorded on a Bruker 500 MHz spectrometer.
Chemical shifts (d), determined from residual solvent peaks, are re-
ported in parts per million relative to (CH3)4Si. Combustion analysis
was performed using a Fisons EA 1108 CHNS-O instrument for all
compounds synthesised. Nominal molecular weight (M+H+) for all
synthesised compounds was confirmed by electrospray ionisation
LC–MS.
1
powder (150 mg, 65%). H NMR ([D6]DMSO): d=13.1 (bs, 1H), 9.73
(s, 1H), (dd, J=8.0, 1.5 Hz, 1H), 7.66–7.68 (m, 4H), 7.42–7.48 (m,
3H), 7.32–7.36 (m, 4H), 6.83 (t, J=7.5 Hz, 1H); 13C NMR ([D6]DMSO):
d=170.8, 147.4, 140.9, 140.5, 135.4, 135.1, 132.8, 129.8, 128.5,
127.8, 127.0, 122.1, 118.6, 115.1, 113.8; Anal. calcd for C19H15NO2: C
78.87, H 5.23, N 4.84, found: C 76.28, H 5.08, N 4.80.
4,5-Dihydroxy-3-phenylcoumarin (10): Compound 10 was pre-
pared by heating resorcinol (1.1 g, 10 mmol) and 2-phenylmalonic
acid diethyl ester (2.4 g, 10.2 mmol) for 2 h at 2508C. The reaction
mixture was allowed to cool to room temperature, and then EtOH
(10 mL) was added. The mixture was stirred for 2 h before the
crude product was filtered off and washed with a small volume of
cold EtOH. Recrystallisation from acetic acid gave the product as
ivory-white crystals (140 mg, 5%). 1H NMR ([D6]DMSO): d=8–10
(bs, 2H), 7.45 (dd, J=8.1, 1.0 Hz, 2H), 7.39 (t, J=8.2 Hz, 1H), 7.33
(t, J=7.7 Hz, 2H), 7.23 (t, J=7.4 Hz, 1H), 6.80 (d, J=8.2 Hz, 1H),
6.69 (d, J=8.2 Hz, 1H); 13C NMR ([D6]DMSO): d=165.8, 162.8, 157.2,
154.2, 133.7, 133.2, 131.6, 128.2, 127.3, 110.8, 108.0, 105.4, 103.1;
Anal. calcd for C15H10O4: C 70.86, H 3.96, found: C 70.13, H 3.95.
2-(Biphenyl-4-ylamino)-5-methoxybenzoic acid (27): 2-(Biphenyl-
4-ylamino)-5-methoxybenzoic acid was synthesised from 2-bromo-
5-methoxybenzoic acid according to the procedure used for com-
pound 26. Recrystallisation from EtOH/H2O yielded 45% of the title
product as light-green crystals. 1H NMR ([D6]DMSO): d=13.3 (bs,
1H), 9.3 (bs, 1H), 7.59–7.76 (m, 4H), 7.42–7.47 (m, 3H), 7.35 (d, J=
9.1 Hz, 1H), 7.31 (t, J=7.3 Hz, 1H), 7.24 (d, J=8.5 Hz, 2H), 7.13 (dd,
J=3.1, 9.1 Hz, 1H), 3.75 (s, 1H); 13C NMR ([D6]DMSO): d=170.2,
152.3, 142.4, 140.8, 140.7, 134.0, 129.8, 128.4, 127.6, 126.9, 122.4,
120.2, 118.5, 116.0, 115.5, 56.3; Anal. calcd for C20H17NO3: C 75.22, H
5.37, N 4.39, found: C 74.85, H 5.30, N 4.45.
4-Hydroxy-3-(4-trifluoromethylbenzyl)coumarin (22): A mixture
of 4-hydroxycoumarin (240 mg, 1.5 mmol) and 4-(trifluoromethyl)-
benzaldehyde (260 mg, 1.5 mmol) was heated for 3.5 h at 1508C in
triethylammonium formate (2 mL).[31] The reaction mixture was
poured into ice water (20 mL), acidified with dilute HCl (1m, 5 mL),
and the crude product was filtered off. Recrystallisation from EtOH
2-Acetylamino-5-phenylbenzoic acid (29): A mixture of 2-acetyla-
mino-5-bromobenzoic acid methyl ester (540 mg, 2.0 mmol), phe-
nylboronic acid (240 mg, 2.0 mmol), and Pd(PPh3)4 (23 mg,
0.02 mmol) in 1,2-dimethoxyethane (20 mL) and 1m NaHCO3
1
gave the desired product as a white solid (150 mg, 31%). H NMR
ChemMedChem 2010, 5, 608 – 617
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
615

I. Fritzson et al.
MED
(7 mL) was heated at 1008C for 10 min. After cooling to room tem-
perature, 5m NaOH (2.5 mL) was added and stirred overnight for
ester hydrolysis, and then H2O (10 mL) was added before the solu-
tion was acidified with 5m HCl (4 mL). After 1 h the resulting mix-
ture was filtered, washed with H2O, and recrystallised from EtOH/
H2O to provide the product (380 mg, 74%) as a white solid.
1H NMR ([D6]DMSO): d=13.8 (bs, 1H), 11.1 (s, 1H), 8.57 (d, J=
8.7 Hz, 1H), 8.22 (d, J=3.6 Hz, 1H), 7.90 (dd, J=8.7, 2.2 Hz, 1H),
7.66 (d, J=7.6 Hz, 2H), 7.47 (t, J=7.5 Hz, 2H), 7.37 (t, J=7.4 Hz,
1H), 2.17 (s, 3H); 13C NMR ([D6]DMSO): d=170.2, 169.4, 141.0,
139.6, 135.0, 132.9, 129.9, 129.6, 128.4, 127.2, 121.4, 117.9, 25.9;
Anal. calcd for C15H13NO3: C 70.58, H 5.13, N 5.49, found: C 70.00, H
5.03, N 5.51.
ed material was filtered off and hydrolysed in EtOH (120 mL) and
5m NaOH (25 mL) at 708C for 7 h. Most of the EtOH was evaporat-
ed, H2O (100 mL) was then added, and the solution was filtered.
After acidification with 5m HCl (25 mL), the precipitate was filtered
off and dried to provide the intermediate 2-propionylamino-5-hy-
droxybenzoic acid (2.3 g, 55%). A mixture of 2-propionylamino-5-
hydroxybenzoic acid (0.24 mmol, 50 mg) and 2-(trifluoromethyl)-
benzyl bromide (0.38 mmol, 90 mg) in acetone (1 mL) and 0.5m
NaOH (1 mL) was heated for 4 h at 708C. MeOH (0.5 mL) and 5m
NaOH (0.5 mL) were added and heated at 708C for an additional
50 min to hydrolyse the intermediate ester; 5m HCl (0.7 mL) was
added to the warm solution, and the slurry was stirred and allowed
to cool to room temperature. The crude product was filtered off,
dried, and re-crystallised from MeOH/H2O (9:1 v/v) to provide the
title compound (40 mg, 45%) as an ivory-white powder. 1H NMR
([D6]DMSO): d=13.7 (bs, 1H), 10.8 (bs, 1H), 8.38 (d, J=9.1 Hz, 1H),
7.70–7.76 (m, 2H), 7.73 (t, J=7.5 Hz, 1H), 7.60 (t, J=7.5 Hz, 1H),
7.51 (d, J=3.1 Hz, 1H), 7.29 (dd, J=9.1, 3.1 Hz, 1H), 5.26 (s, 2H),
2.38 (q, J=7.5 Hz, 2H), 1.12 (t, J=7.5 Hz, 3H); 13C NMR ([D6]DMSO):
d=172.4, 169.8, 153.6, 135.6 (d, J=2.5 Hz), 133.7, 131.3, 129.7,
127.7 (q, J=30 Hz), 127.0 (q, J=5.7 Hz), 125.2 (q, J=274 Hz), 122.9,
121.7, 119.1, 116.6, 67.4 (d, J=2.4 Hz), 31.3, 10.3; Anal. calcd for
C18H16F3NO4: C 58.86, H 4.39, N 3.81, found: C 57.87, H 4.34, N 3.84.
2-Acetylamino-5-(4-biphenyl)benzoic acid (30): A mixture of 2-
acetylamino-5-bromobenzoic acid (520 mg, 2.0 mmol), 4-biphenyl-
boronic acid (400 mg, 2.0 mmol), and Pd(PPh3)4 (72 mg,
0.062 mmol) in 1,2-dimethoxyethane (10 mL) and 1m NaHCO3
(7 mL) was heated at 1008C for 2 h. The solution was cooled and
then acidified with 1m H2SO4 (10 mL). The resulting precipitate was
filtered off and washed with a small amount of H2O. The crude
product was then dissolved in EtOAc, filtered through a short
column of silica and then concentrated to ~10 mL. Upon standing
at room temperature, crystallisation occurred. The resulting ivory-
white crystals were filtered off and dried to yield the desired prod-
1
uct (230 mg, 35%). H NMR ([D6]DMSO): d=13.8 (bs, 1H), 11.2 (bs,
Biochemistry
1H), 8.56 (d, J=8.7 Hz, 1H), 8.27 (d, J=2.4 Hz, 1H), 7.95 (dd, J=
2.4, 8.7 Hz, 1H), 7.76 (s, 4H), 7.71 (d, J=7.3 Hz, 2H), 7.48 (t, J=
7.7 Hz, 2H), 7.37 (t, J=7.4 Hz, 1H), 2.15 (s, 3H); 13C NMR
([D6]DMSO): d=170.3, 169.4, 141.0, 140.4, 140.0, 138.6, 134.4,
132.7, 129.9, 129.5, 128.4, 128.2, 127.7, 127.4, 121.5, 118.1, 25.9;
Anal. calcd for C21H17NO3: C 76.12, H 5.17, N 4.23, found: C 74.82, H
5.10, N 4.26.
The DHODH enzyme assay was established and validated using mi-
tochondrial membrane preparations from human lymphoma U937
cells for the initial DHODH screening of fragments (Table 1).[22] The
results presented are average values of duplicate samples run on
two different occasions. Relative standard deviations of mean in-
hibition were <40%. IC50 values were determined for expanded
derivatives and the most interesting fragment compounds from
the low-affinity assay (Tables 2–4). For determination of IC50 values,
the recombinant DHODH enzyme was purified, and an in vitro
enzyme assay was established using N-terminally truncated re-
combinant human DHODH.[11b] The assay was based on coupling of
the ubiquinone reduction to the redox dye 2,6-dichloroindophenol
(DCIP), and the reduction of DCIP was monitored photometrically
by decreasing absorption at l 630 nm according to reference [33].
A771726 was used as control, and IC50 values varied between 0.7
and 1.1 mm (reported IC50: 0.77Æ0.08).[11b] The assay was run with
duplicate samples on two to five different occasions. Standard
errors of mean IC50 were usually <20% for tested compounds (see
the Supporting Information).
2-Propionylamino-5-[(E)-styryl]benzoic acid (31): A mixture of
methyl-2-amino-5-bromobenzoic acid methyl ester (6.8 mmol,
1.4 g) and propionyl chloride (12 mmol, 1.1 g) in 1,2-dichloro-
ethane (30 mL) was stirred for 5 h. CHCl3 (30 mL) was added, and
the organic phase was washed with 1m NaHCO3 and H2O. It was
then dried over anhydrous Na2SO4, filtered, and concentrated to
give the intermediate 2-propionylamino-5-bromobenzoic acid
methyl ester (1.6 g, 82%). A mixture of 2-propionylamino-5-bromo-
benzoic acid methyl ester (3.5 mmol, 1.0 g), K2CO3 (3.8 mmol,
0.53 g), tributylamine (3.8 mmol, 0.71 g), Pd(PPh3)2Cl2 (0.05 mmol,
35 mg), and styrene (4.2 mmol, 0.44 g) in dry DMF (20 mL) was
stirred and heated overnight at 1508C. H2O (10 mL) and 5m NaOH
(2 mL) were added, and heating at 1008C was continued for 1 h.
The mixture was allowed to cool to room temperature and was di-
luted with H2O (50 mL), filtered through Celite, and acidified with
5m HCl (4 mL). The precipitate was filtered off, washed with H2O
and recrystallised from EtOH to provide the title compound
(0.35 g, 34%) as an ivory-white powder. 1H NMR ([D6]DMSO): d=
13.8 (bs, 1H), 11.2 (bs, 1H), 8.55 (d, J=8.7 Hz, 1H), 8.17 (d, J=
2.1 Hz, 1H), 7.87 (dd, J=8.7, 2.1 Hz, 1H), 7.62 (d, J=7.3 Hz, 2H),
7.38 (t, J=7.7 Hz, 2H), 7.51–7.64 (m, 2H), 7.23 (d, J=16.5 Hz, 1H),
2.43 (q, J=7.5 Hz, 2H), 1.14 (t, J=7.5 Hz, 3H); 13C NMR ([D6]DMSO):
d=172.8, 170.4, 141.1, 137.8, 132.2, 130.2, 129.6, 128.9, 128.5,
128.0, 127.3, 121.0, 117.4, 31.5, 10.2; Anal. calcd for C18H17NO3: C
73.20, H 5.80, N 4.74, found: C 72.59, H 5.70, N 4.76.
Crystallisation of human DHODH in complex with compound
30
Cloning, expression, and purification of human DHODH and co-
crystallisation with compound 30 as well as data collection and
structure determination were done according to the methods de-
scribed by Walse et al.[20] (see the Supporting Information). Auto-
Dep EBI-41387 has been assigned wwPDB ID 2wv8 for the coordi-
nate entry; we assigned the code r2wv8sf for the structure factors.
Acknowledgements
2-Propionylamino-5-(2-trifluoromethylbenzyloxy)benzoic
acid
(32): 2-Amino-5-hydroxybenzoic acid (20 mmol, 3.0 g) dissolved in
0.5m NaOH (150 mL) was cooled to 0–58C. Propionic anhydride
(50 mmol, 6.5 g) was added, and the resulting solution was stirred
for 10 min and then acidified with 5m HCl (16 mL). The precipitat-
We thank Julia Selmani and Lisbeth Witt for providing biochemi-
cal data, Ann-Charlotte Johansson and Tomas Fristedt for analyt-
ical data, and Anders Sundin for assistance in the preparation of
616
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 608 – 617

Inhibitors of Human DHODH
[18] I. Muegge, Y. C. Martin, J. Med. Chem. 1999, 42, 791.
[19] W. Knecht, M. Lçffler, Biochem. Pharmacol. 1998, 56, 1259.
[20] B. Walse, V. Tamu Dufe, B. Svensson, I. Fritzson, L. Dahlberg, A. Khairoul-
lina, U. Wellmar, S. Al-Karadaghi, Biochemistry 2008, 47, 8929.
[21] R. Baumgartner, M. Walloschek, M. Kralik, A. Gotschlich, S. Tasler, J. Mies,
J. Leban, J. Med. Chem. 2006, 49, 1239.
[22] J. M. Bruneau, C. M. Yea, S. Spinella-Jaegle, C. Fudali, K. Woodward, P. A.
Robson, C. Sautꢂs, R. Westwood, E. A. Kuo, R. A. Williamson, E. Ruuth,
Biochem. J. 1998, 336, 299.
[23] QlogP (octanol/H2O) values were calculated with the Schrçdinger Maes-
tro program QuickProp.
[24] D. Gonzꢃles-Aragꢄn, J. Ariza, J. M. Villalba, Biochem. Pharmacol. 2007,
73, 427.
figures. This work was supported in part by the Swedish Research
Council.
Keywords: dihydroorotate dehydrogenase · drug discovery ·
fragment screening · inhibitors · structure–activity relationships
[1] H. M. Cherwinski, R. G. Cohn, P. Cheung, D. J. Webster, Y.-Z. Xu, J. P. Caul-
field, J. M. Young, G. Nakano, J. T. Ransom, J. Pharmacol. Exp. Ther. 1995,
275, 1043.
[2] L. D. Fairbanks, M. Bofill, K. Ruckemann, H. A. Simmonds, J. Biol. Chem.
1995, 270, 29682.
[3] R. I. Fox, J. Rheumatol. Suppl. 1998, 53, 20.
[4] D. G. Batt, Expert Opin. Ther. Pat. 1999, 9, 41.
[5] J. Gyves, B. Dallaire, R. Sisk, R. Morgan, J. Maroun, J. Ruckdeschel, R. B.
Natale, Cancer Chemother. Pharmacol. 1993, 32, 64.
[6] B. Rozman, Clin. Pharmacokinet. 2002, 41, 421.
[25] E. S. Cleaveland, A. Monks, A. Vaigro-Wolff, D. W. Zaharevitz, K. Paull, K.
Ardalan, D. A. Cooney, H. J. Ford, Biochem. Pharmacol. 1995, 49, 947.
[26] a) W. Knecht, M. Lçffler, FEBS Lett. 2000, 467, 27; b) J. Baldwin, A. M. Far-
ajallah, N. A. Malmquist, P. K. Rathod, M. A. Phillips, J. Biol. Chem. 2002,
277, 41827.
[7] G. W. Cannon, J. M. Kremer, Rheum. Dis. Clin. North Am. 2004, 30, 295.
[8] F. C. Breedveld, J.-M. Dayer, Ann. Rheum. Dis. 2000, 59, 841.
[9] S. Liu, E. A. Neidhardt, T. H. Grossman, T. Ocain, J. Clardy, Structure 2000,
8, 25.
[10] a) J. Leban, M. Kralik, J. Mies, M. Gassen, K. Tentschert, R. Baumgartner,
Bioorg. Med. Chem. Lett. 2005, 15, 4854; b) J. Leban, M. Kralik, J. Mies, R.
Baumgartner, M. Gassen, S. Tasler, Bioorg. Med. Chem. Lett. 2006, 16,
267; c) D. E. Hurt, A. E. Sutton, J. Clardy, Bioorg. Med. Chem. Lett. 2006,
16, 1610; d) H. Kugimiya, M. Inagaki, Patent JP 2007015952, 2007.
[11] a) M. Hansen, J. Le Nours, E. Johansson, T. Antal, A. Ullrich, M. Lçffler, S.
Larsen, Protein Sci. 2004, 13, 1031; b) A. Ullrich, W. Knecht, M. Fries, M.
Lçffler, Eur. J. Biochem. 2001, 268, 1861.
[27] a) J. C. Castro Palomino Laria, M. Erra Sola, M. E. Lozoya Toribio, E. Nav-
arro Romero, PCT Vol. WO2008077639 A1, 2008; b) J. C. Castro Palomi-
no Laria, E. Terricabras Belart, M. Erra Sola, E. Navarro Romero, S. Fon-
querna Pou, A. Cardus Figueras, M. E. Lozoya Toribio, PCT Vol.
2009021696, 2009.
[28] L. H. Sutherland, A. E. Sloboda, R. G. Child, J. F. Poletto, D. W. Powell,
Patent EP-305952, 1989.
[29] A. M. Ferrari, M. Sgobba, M. C. Gamberini, G. Rastelli, Eur. J. Med. Chem.
2007, 42, 1028.
[30] K. N. Trivedi, J. Sci. Ind. Res. Sect. B: Phys. Sci. 1962, 218, 402.
[31] G. Toth, S. Molnar, T. Tamas, I. Borbely, Org. Prep. Proced. Int. 1999, 31,
222.
[12] D. Erlanson, R. McDowell, T. O’Brien, J. Med. Chem. 2004, 47, 3463.
[13] Computer software Cerius2, Accelrys Inc., San Diego, CA (USA).
[14] a) H. J. Bçhm, J. Comput.-Aided Mol. Des. 1992, 6, 61; b) H. J. Bçhm, J.
Comput.-Aided Mol. Des. 1992, 6, 593.
[15] P. S. Charifson, J. J. Corkery, M. A. Murcko, W. P. Walters, J. Med. Chem.
1999, 42, 5100.
[32] M. Carissimi, I. Grasso, E. Gumelli, E. Milla, F. Ravenna, Farmaco Sci.
1963, 18, 705.
[33] E. A. Neidhardt, S. R. Punreddy, J. E. McLean, L. Hedstrom, T. H. Gross-
man, J. Mol. Microbiol. Biotechnol. 1999, 1, 183.
[16] H. J. Bçhm, J. Comput.-Aided Mol. Des. 1998, 12, 309.
[17] D. K. Gehlhaar, D. Bouzida, P. A. Rejto in ACS Symposium Series Vol. 719
(Eds.: A. L. Parrill, M. R. Reddy), American Chemical Society, Washington
DC, 1999, p. 292.
Received: November 3, 2009
Revised: January 21, 2010
Published online on February 22, 2010
ChemMedChem 2010, 5, 608 – 617
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
617