Discovery of inhibitors of the pentein superfamily protein dimethylarginine dimethylaminohydrolase (DDAH), by virtual screening and hit analysis

Article abstract of DOI:10.1016/j.bmcl.2007.04.095

An efficient process for the discovery of inhibitors of DDAH enzymes, without the requirement for high throughput screening, is described. Physicochemical filtering of a 308,000-compound library according to drug likeness followed by reciprocal nearest neighbour selection produced a representative subset of 35,000 compounds. Virtual screening on a dual processor PC using FlexX, followed by biological screening, identified two hit series. Similarity searches of commercial databases and chemical re-synthesis of pure compounds resulted in SR445 as an inhibitor of Pseudomonas aeruginosa DDAH at 2 μM.

Full text of DOI:10.1016/j.bmcl.2007.04.095

Bioorganic & Medicinal Chemistry Letters 17 (2007) 3953–3956
Discovery of inhibitors of the pentein superfamily protein
dimethylarginine dimethylaminohydrolase (DDAH), by
virtual screening and hit analysis
Basil Hartzoulakis,a, Sharon Rossiter,a,à Herpreet Gill,b Bernard O’Hara,c
Emily Steinke,a Paul J. Gane,a Ramon Hurtado-Guerrero,b James M. Leiper,b
Patrick Vallance,b Judith Murray Rustc and David L. Selwooda,*
aWolfson Institute for Biomedical Research, University College London, Gower Street, London WC1E 6BT, UK
bDivision of Medicine, BHF Laboratories, Rayne Building, University College London, 5 University Street, London WC1E 6JF, UK
cSchool of Crystallography, Birkbeck, Malet Street, London WC1E 7HX, UK
Received 4 January 2007; revised 27 April 2007; accepted 28 April 2007
Available online 3 May 2007
Abstract—An efficient process for the discovery of inhibitors of DDAH enzymes, without the requirement for high throughput
screening, is described. Physicochemical filtering of a 308,000-compound library according to drug likeness followed by reciprocal
nearest neighbour selection produced a representative subset of 35,000 compounds. Virtual screening on a dual processor PC using
FlexX, followed by biological screening, identified two hit series. Similarity searches of commercial databases and chemical re-syn-
thesis of pure compounds resulted in SR445 as an inhibitor of Pseudomonas aeruginosa DDAH at 2 lM.
Ó 2007 Elsevier Ltd. All rights reserved.
The pentein superfamily of proteins (Fig. 1a) is charac-
terized by a propeller like arrangement of five abbab
units forming a narrow channel with a central negatively
charged core.1,2 This elegant structure is ideal for recog-
nition of the guanidinopropyl side chain of the amino
acid arginine: its main biological function in the guani-
dine modifying enzyme (GME) branch of the superfam-
ily. GMEs contain a conserved triad of His, Asp and
Cys amino acids which serve to catalyze a range of
hydrolysis and transferase reactions on the arginine or,
in the case of DDAH, the methyl arginine side chain
(Fig. 1b). The biological functions of this family are very
diverse,2 not only allowing metabolic reactions on sim-
ple arginine derivatives but also, in the case of the pro-
tein arginine deiminases (PADs), able to function as
molecular switches for transcription by citrullination
of arginine side chains on histones.3 Mammalian
DDAH functions to control ADMA levels (a cardiovas-
cular risk factor) and can modulate the nitric oxide
pathway with its pivotal role in endothelial cell func-
tion.4,5 Significantly DDAH may also participate in pro-
tein regulation and DDAH 2 has been shown to increase
transcription of the cancer promoting cytokine VEGF6
(Fig. 1). Pseudomonas aeruginosa (Pa) is a pathogen
linked to chronic lung infections in cystic fibrosis7 and
PaDDAH is a possible factor involved in pathogenic-
ity.8 The discovery of inhibitors of DDAH and the wider
pentein family is therefore of considerable interest
(Fig. 2) and both competitive 1, 2 and irreversible 3, 4,
5 inhibitors have been reported.3,9–12
We have previously described substrate based inhibi-
tors9 of mammalian DDAH such as 1 (SR291), however
we required access to inhibitors of better potency and
different chemical structural types. The availability of
the PaDDAH crystal structure allowed us to consider
virtual screening as an approach to the discovery of
new inhibitors. Virtual screening has many potential
advantages: lower cost, higher speed and lower resource
requirements make it ideally suited to an academic re-
search group. We therefore initiated an approach based
on physicochemical filtering, virtual screening and hit
Keywords: Pentein; Dimethylargininedimethylaminohydrolase; DDAH;
Virtual screening; Hit analysis.
*
Corresponding author. E-mail: d.selwood@ucl.ac.uk
Present address: NCE discovery Ltd., 418 Science Park, Milton
Road, Cambridge CB4 0PZ, UK.
à
Present address: School of Life Sciences, University of Hertfordshire,
College Lane, Hatfield, Herts AL10 9AB, UK.
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.04.095

3954
B. Hartzoulakis et al. / Bioorg. Med. Chem. Lett. 17 (2007) 3953–3956
308,000 structures
a
Drug likeness: ClogP -2 to +5, MWt <650
H-bond donors <5, H-bond acceptors <10
Rotatable bonds <10, low complexity
260,000 structures
35,000 structures
109 compounds
Reciprocal Nearest Neighbor packing
selection. Radius 0.6, cl_radius 0.85
FLexX fragment based docking, 10 days
on 2× 2.8 MHz Xeon Processors
DDAH biological screen (500 µM or 50 µM
if insoluble)
3 actives
Similarity searches, analogue purchase
(20 compounds), confirmation of structure,
retesting.
2 inhibitors
NH2
O
H
N
b
N
OH
SP1
NH
Figure 3. Filtering, virtual screening, biological testing and hit analysis
lead to the discovery of micromolar inhibitors of DDAH.
PKA
a diverse, subset of compounds such that clusters of sim-
ilar compounds in the original database were repre-
sented by small groups of compounds (not singletons)
in the subset. In this way a cluster containing actives is
better represented in the subset.22 The subset of 35,000
two-dimensional structures was then converted into
3D using Concord23 and minimized with the MMFF94
force field. In the final step the MMFF94 charges were
replaced with formal charges for input to the docking
program FlexX. The PaDDAH structure (pdb: 1h70)
was manipulated by restoring the Cys249, adding hydro-
gens and Kollman atom charges then minimized using
the Amber Force Field (version 7.0). Citrulline was left
in place during the minimization to prevent collapse of
the active site and was then removed. Virtual docking
was conducted using FlexX.24 FlexX cuts ligands into
fragments at rotatable bonds and places a base fragment
into the active site, the ligand is then reformed and the
energy calculated using Bohm’s function. The dimethy-
larginine binding pocket site was defined accurately
using the Sybyl/Flexx interface (Leu18, Asp 60, Phe63,
DDAH2
DDAH
NH2
H
N
P
O
O H
SP1
VEGF
NH2
O
Figure 1. (a) Ribbon diagrams of pentein family proteins PaDDAH
and PAD4. (b) DDAH hydrolyses methylated arginines including
ADMA a cardiovascular risk factor. DDAH2 can increase the
expression of VEGF through phosphorylation of its transcription
factor SP1.
analysis to identify inhibitors of PaDDAH utilizing the
crystal structure.13
Starting with a database of 308,000 commercially avail-
able compounds,14 we initially employed a series of fil-
ters to remove about 48,000 non-druglike15,16
compounds and reactive molecules (Fig. 3). This first
step is divided into a structural filter17,18 (see Supple-
mentary data) which removes non-drug-like compounds
and those with undesirable atoms/functional groups and
a properties filter19,20 that passes only lead-like com-
pounds. All filtering was implemented with scripts writ-
ten in the Sybyl programming language. After this first
step the number of compounds in the databases was re-
duced to ꢀ260 k. A reciprocal nearest neighbour21
(RNN) packing algorithm was then utilized to generate
O-
N+
S
O
O
O
N
O
S
N
N
S
N
NH
NH
H
6 (28%)a
N
O
O-
N+
O
S
O
N
8 (59%) IC50 16.9 μM
NH
O
HO
7 (26%)a
NH2
PFPO SO2
H
N
H
N
O
O
O
OMe
O
NH
NH
N
O
S
S
NH
O
N
O
Me
O
NH
NH
1 (SR291) huDDAH 20 μM
Br 2 PaDDAH 16 μM
O
N
PFP = pentafluorophenyl
O
N
H
O
N
NH2
NHBz
H
NH
NH2
OR
N
OR
Cl
S
F
9 IC50 6.4 μM
10 (SR445) IC50 2 μM
O
NH
O
Figure 4. Structures and activity (500 lM) of screening hits (blue) and
similarity search hits (black). IC50S are of re-synthesised pure samples.
aInsoluble in buffer.
3 F-amidine, PAD4
4 bovine DDAH
5 bovine DDAH
Figure 2. Inhibitors of DDAH and PAD4.

B. Hartzoulakis et al. / Bioorg. Med. Chem. Lett. 17 (2007) 3953–3956
3955
is shown for compounds 9 and 10 in Scheme 1. If re-
quired the indole carbaldehydes were N-alkylated fol-
lowed by condensation with the substituted barbituric
acids26 to provide the target molecules.
R
NaH/DMF
NH
N
O
(For R not = H)
O
O
The indolylthiobarbituric acid scaffold proved the most
interesting; the compounds resulting from the similarity
searches, 9 and 10 (SR445), demonstrated much im-
proved activity in the re-synthesised samples with 10
having an IC50 of 2 lM. This is the most potent PaD-
DAH inhibitor identified to date and is structurally dis-
tinct from the other inhibitors of both PaDDAH and
mammalian DDAH (Fig. 2). The predicted binding
mode of 10 (Fig. 5) shows insertion into the active site
of DDAH with the indole moiety taking the place of
the arginine side chain.
O
6M HCl
EtOH
NR'
HN
S
R
N
O
O
N
9, R =
R' = H
O
H
O
NR'
HN
10, R = H, R' = 4PhOMe
S
This study highlights the advantages of virtual screen-
ing, allowing academic laboratories with limited man-
power and resources to discover protein ligands for
biological studies. Pre-filtering the database produced
a representative set amenable to screening on a dual pro-
cessor PC. We cannot yet consider the docking process
to be very precise.27 Rather docking enriches the final
biological screening set to increase the chances of
obtaining an active. The recent disclosures of the crystal
structures of bovine and human DDAH5,28 should allow
similar methodology to be applied to the discovery of
inhibitors of the mammalian and human isoforms.
Scheme 1. General synthesis of the indolyl barbiturates.
Asp66, Glu65, Arg85, Glu114, Arg132, Leu161, His162,
Ile243, C249).
The top 2000 successfully docked compounds were
ranked with the CScore scoring function. The highest
scoring 200 molecules were visually inspected to ensure
that the docking algorithm had placed them in the bind-
ing pocket with at least one hydrogen bond between the
protein and the ligand. Of these 109 structures, 90 were
available for supply and these were screened, using a
simple colorimetric assay (see supporting information).
Three actives were identified, compounds 6, 7 and 8
(Fig. 4), representing two chemical scaffolds.
Acknowledgment
This work was funded by the Wellcome Trust (Grant
065612) and UCL Biomedica.
About 30% of the original screening set exhibited poor
solubility precluding accurate IC50 determination. The
activity around the active scaffolds was probed by the
purchase of 20 additional analogues identified using
similarity searches on the active hits based on Unity fin-
gerprints and a Tanimoto score of >80%.
Supplementary data
Details of computational methods, chemistry syntheses
and biological screening. Supplementary data associated
with this article can be found, in the online version, at
doi:10.1016/j.bmcl.2007.04.095.
Re-synthesis to provide pure authentic samples was a
key part of the hit analysis and the general synthesis25
Figure 5. Top and side views of predicted binding mode of 10.

3956
B. Hartzoulakis et al. / Bioorg. Med. Chem. Lett. 17 (2007) 3953–3956
13. Murray-Rust, J.; Leiper, J.; McAlister, M.; Phelan, J.;
References and notes
Tilley, S.; Santa, M. J.; Vallance, P.; McDonald, N. Nat.
Struct. Biol. 2001, 8, 679.
14. The Asinex database was chosen as the compounds were
easily available and each was provided with analytical
data (NMR, LCMS).
1. Teichmann, S. A.; Murzin, A. G.; Chothia, C. Curr. Opin.
Struct. Biol. 2001, 11, 354.
2. Shirai, H.; Mokrab, Y.; Mizuguchi, K. Proteins 2006, 64,
1010.
15. Lipinski, C. A. J. Pharmacol. Toxicol. Methods 2000, 44,
235.
16. Veber, D. F.; Johnson, S. R.; Cheng, H. Y.; Smith, B. R.;
Ward, K. W.; Kopple, K. D. J. Med. Chem. 2002, 45, 2615.
17. Muegge, I.; Heald, S. L.; Brittelli, D. J. Med. Chem. 2001,
44, 1841.
18. Rishton, G. M. Drug Discov. Today 2003, 8, 86.
19. Clark, D. E.; Pickett, S. D. Drug Discov. Today 2000, 5, 49.
20. Walters, W. P.; Ajay; Murcko, M. A. Curr. Opin. Chem.
Biol. 1999, 3, 384.
21. Ward, J. H. J. Am. Stat. Soc. 1963, 58, 236.
22. Leach, A. R.; Bradshaw, J.; Green, D. V.; Hann, M. M.;
Delany, J. J., III. J. Chem. Inf. Comput. Sci. 1999, 39,
1161.
3. Luo, Y.; Knuckley, B.; Lee, Y. H.; Stallcup, M. R.;
Thompson, P. R. J. Am. Chem. Soc. 2006, 128, 1092.
4. Knipp, M. Chem. Biochem. 2006, 7, 879.
5. Leiper, J.; Nandi, M.; Torondel, B.; Murray-Rust, J.;
Malaki, M.; O’Hara, B.; Rossiter, S.; Anthony, S.;
Madhani, M.; Selwood, D.; Smith, C.; Wojciak-Stothard,
B.; Rudiger, A.; Stidwill, R.; McDonald, N. Q.; Vallance,
P. Nat. Med. 2007, 13, 198.
6. Hasegawa, K.; Wakino, S.; Tanaka, T.; Kimoto, M.;
Tatematsu, S.; Kanda, T.; Yoshioka, K.; Homma, K.;
Sugano, N.; Kurabayashi, M.; Saruta, T.; Hayashi, K.
Arterioscler. Thromb. Vasc. Biol. 2006, 26, 1488.
7. Agarwal, G.; Kapil, A.; Kabra, S. K.; Das, B. K.;
Dwivedi, S. N. BMC. Microbiol. 2005, 5, 43.
8. Stone, E. M.; Person, M. D.; Costello, N. J.; Fast, W.
Biochemistry 2005, 44, 7069.
23. Pearlman, R.S. ‘‘Concord’’ distributed by Tripos Inc., St
Louis, Missouri, 63144, USA. 2006.
24. Rarey, M.; Wefing, S.; Lengauer, T. J. Comput. Aided
Mol. Des. 1996, 10, 41.
25. Nightingale, D.; Alexander, C. H. J. Am. Chem. Soc. 1936,
58, 794.
26. Dox, A. W.; Plaisance, G. P. J. Am. Chem. Soc. 1916, 38,
2164.
9. Rossiter, S.; Smith, C. L.; Malaki, M.; Nandi, M.; Gill,
H.; Leiper, J. M.; Vallance, P.; Selwood, D. L. J. Med.
Chem. 2005, 48, 4670.
10. Vallance, P.; Bush, H. D.; Mok, B. J.; Hurtado-Guerrero,
R.; Gill, H.; Rossiter, S.; Wilden, J. D.; Caddick, S. Chem.
Commun. (Camb.) 2005, 5563.
27. Marsden, P. M.; Puvanendrampillai, D.; Mitchell, J. B.;
Glen, R. C. Org. Biomol. Chem. 2004, 2, 3267.
28. Frey, D.; Braun, O.; Briand, C.; Vasak, M.; Grutter, M.
G. Structure 2006, 14, 901.
11. Stone, E. M.; Schaller, T. H.; Bianchi, H.; Person, M. D.;
Fast, W. Biochemistry 2005, 44, 13744.
12. Knipp, M.; Braun, O.; Vasak, M. J. Am. Chem. Soc. 2005,
127, 2372.