Dia 1

1
Virtual Screening of Novel Cannabinergic Ligands
Using a Comparative Model of the CB2
Receptor Outi M. H. Salo,1 Katri Raitio,1 Juha
R. Savinainen,2 Tapio Nevalainen,1 Maija
Lahtela-Kakkonen,1 Jarmo T. Laitinen,2 Tomi
Järvinen,1 and Antti Poso1 1Department of
Pharmaceutical Chemistry, 2Department of
Physiology, University of Kuopio, P.O. Box 1627,
FIN-70211, Kuopio, Finland

• INTRODUCTION
• CB2 receptor is a peripheral G-protein coupled
  receptor (GPCR) located mainly in the immune
  tissues.
• CB2 ligands have potential therapeutic
  applications, e.g., in the management of pain,
  inflammation, and gliomas 1.
• The aim of this study was to build a comparative
  model for the CB2 receptor and utilize it in
  virtual database screening to find novel CB2
  selective lead compounds.

HU-210
HU-308
CP55940

• MATERIAL AND METHODS
• Structure construction, optimization and
  visualization were carried out using the
  molecular modelling packages Sybyl 6.9 2 and
  InsightII 3.
• Molecular dynamics (MD) runs were performed with
  GROMACS 4. GOLD 5 was used for the docking
  studies.
• CScore module of Sybyl and X-Score 6 were
  employed to calculate and rank the docking scores
  for the resulting docking conformations.
• UNITY 4.4 of Sybyl was used for database
  searches.

L759633
L759656
AM1241
Figure 3. Query 3 was based on the
ligand-receptor complex. All but the essential
hydrogens of the amino acids are omitted for
clarity.
Figure 2. Nonselective and selective CB2 agonists.

• Maybridge and LeadQuest databases were searched
  through and the hit molecules were ranked by
  docking and scoring.
• The best-ranked hit compounds that passed the
  "chemist's eye" were purchased from the
  commercial databases and tested for their
  CB2-mediated G-protein activation.

CONCLUSIONS The present study demonstrates that
rhodopsin-based GPCR models can be successfully
used for virtual screening of novel lead
compounds. Even if the receptor model is not as
accurate as a crystal structure, the docking of
known ligands into the putative receptor binding
cavity can be helpful in creating the 3D queries
for database searches.

• 1. CB2 model
• The crystal structure of the bovine rhodopsin
  (PDB code 1hzx) 7 was used as a template for
  the comparative modelling process. The modelling
  alignment is presented in Figure 1.
• We used the same model refinement procedure as
  previously described for the CB1 receptor model
  8 A 500-ps constrained MD simulation at 300 K
  in a water box.
• Additionally, new conformers of the receptor
  binding site were produced in a simulated
  annealing procedure.
• 39 known CB2 ligands were then docked into five
  chosen conformers of the receptor model (models
  1-5) and the docking results were ranked by
  scoring algorithms.
• The best-ranked docking conformations of the
  ligands were used to build rough 3D-QSAR models
  utilizing the CB2 affinity data collected from
  literature (data not shown).
• The receptor conformer that produced the best
  QSAR models was chosen to be used for virtual
  database screening.

• RESULTS AND DISCUSSION
• According to the 3D-QSAR results, model 1 (the
  final structure frame from the MD simulation) was
  chosen for the database searches.
• The overall screening and filtering procedure is
  presented in Figure 4.

ACKNOWLEDGMENTS We thank CSC - Scientific
Computing, Ltd (Espoo, Finland) for providing
software and supercomputer resources for the
docking studies, and National Technology Agency
of Finland, Academy of Finland, and National
Graduate School in Informational and Structural
Biology for supporting the work financially. Toni
Rönkkö (M.Sc.), Ms. Helly Rissanen and Ms. Minna
Glad are acknowledged for their technical
assistance.
REFERENCES
1 Di Marzo, V. Bifulco, M. De Petrocellis, L.
The endocannabinoid system and its therapeutic
exploitation. Nat. Rev. Drug Discov. 2004, 3,
771-784. 2 Tripos Associates, Inc., SYBYL v.
6.9, St. Louis, Missouri,USA 3 Accelrys, Inc.,
Insight II, v. 2000, San Diego, CA, USA 4
Lindahl, E. Hess, B. van der Spoel, D. GROMACS
3.0 A package for molecular simulation and
trajectory analysis. J Mol Mod 2001, 7,
306-317. 5 GOLD v.1.2 Cambridge
Crystallographic Data Centre Cambridge, UK. 6
Wang, R. Lai, L. Wang, S. Further Development
and Validation of Empirical Scoring Functions for
Structure-Based Binding Affinity Prediction. J.
Comput.-Aided Mol. Des. 2002, 16, 11-26. 7
Teller, D. C. Okada, T. Behnke, C. A.
Palczewski, K. Stenkamp, R. E. Advances in
determination of a high-resolution
three-dimensional structure of rhodopsin, a model
of G-protein-coupled receptors (GPCRs).
Biochemistry 2001, 40, 7761-7772. 8 Salo, O.
M. Lahtela-Kakkonen, M. Gynther, J. Jarvinen,
T. Poso, A. Development of a 3D model for the
human cannabinoid CB1 receptor. J. Med. Chem.
2004, 47, 3048-3057.
Figure 4. Database screening and filtering the
hits. Numbers represent the number of compounds
that matched a particular query or passed a
particular filter. Q1 hits found with query 1
Q3 hits found with query 3 Q2-1 hits found
with query 2 from the results of Q1 Q2-3 hits
found with query 2 from the results of Q3.
Figure 1. Modelling alignment of human CB2
receptor and bovine rhodopsin sequences. Boxes
define the aminoacid coordinates taken from the
rhodopsin crystal structure,

• Altogether, about 100 hit compounds were
  purchased and screened for their G-protein
  activation via CB2.
• One of the Maybridge hits was shown to act as an
  agonist at CB2. In the future, this low-potency
  lead structure will be further optimized to
  develop potent CB2 agonists that could be used
  for therapeutic purposes.
• Three hits from LeadQuest database were also
  shown to activate G protein in hCB2 transfected
  CHO cell membranes. These compounds need to be
  better characterized in the future.

• 2. Database Screening
• Three different search queries were built
• Query 1 (Q1) based on six CB2 agonists in their
  docking conformation (Figure 2).
• Query 2 (Q2) based on the structure of the
  receptor binding site of the docked HU-210
• Query 3 (Q3) based on the features of the
  ligand-receptor complex (ligands HU-308, HU-210,
  CP55940). See Figure 3.

Molecular Modelling Group of the University of
Kuopio http//www.uku.fi/farmasia/fake/modelling/
index.shtml