Transcript The Molecular Sciences Institute Founded in 1996 by Nobel

The Alpha Project
- Stepping Towards Predictive Biology
Michael B. Gonzales
Senior Research Fellow
Molecular Sciences Institute
Berkeley, CA
The Molecular Sciences Institute
• Founded in 1996 by Nobel Laureate Dr. Sydney
Brenner
• Independent, non-profit research laboratory that
combines genomic experimentation and computer
modeling
• Core research activity - The Alpha Project
• Currently ~20 senior research fellows - molecular
biology, physics, chemistry, mathematics
Goal: To combine genomic and computational research
in order to make predictive models of biological systems.
Magritte
The Alpha Project
• Five-year, multidisciplinary research effort
• The focus of the Alpha Project is to examine
extra/intra-cellular information flow and
processing
• Collaborators include California Institute of
Technology, the Massachusetts Institute of
Technology, the University of California,
Berkeley, University of California, San Francisco
and Pacific Northwest National Laboratory.
Why baker’s yeast?
• S. cerevisiae mating provides a level of system
description greater than that for almost any other
eukaryotic process
• Alpha pheromone signal pathway is GPCR
mediated and analogous to higher eukaryotes
•Yeast are highly tractable experimentally;
facilitating the development of new experimental
methods
• Well-suited to rapid iterative experimental cycles
linking new experimentation to new computation
Sex in the lab
a factor
a
a factor
zygote
a/a
a
Response to Pheromone
2 hour pheromone treatment
Reporter: Prm1-YFP
Bright field image
Fluorescent image
The pheromone response pathway
P
P
P
P
P
Ste2
a/GpaI
b/Ste4 g/Ste18
Ste20
Ste11 S
t
Ste7
MAPK Cascade
e
Fus3 5
Ste12 Transcriptional Activation
G1 Arrest Morphogenesis, fusion
Credits
Ximena Ares
Kirsten Benjamin
Roger Brent
Ian Burbulis
Kirindi Choi
Tina Chin
Alejandro Colman-Lerner
Jay Doane
Michael B. Gonzales
Andrew Gordon
Larry Lok
Andrew Mendelsohn
Orna Resnekov
Eduard Serra
David Soergel
Kumiko Yamaguchi
Richard Yu
UCSF
Matt Jacobson
Brian Shoichet
Kevan Shokat
MIT
Drew Endy (ex MSI)
Ty Thomson (BioEng)
Gerry Sussman (CS)
Tom Knight (CS)
UC Berkeley
Julie Leary (Chem)
Stuart Russell (CS)
Caltech
Shuki Bruck (CS)
Sandia NL
PNNL
Richard Smith (Chem) Steve Plimpton (P, CS)
Danny Rintoul (P, CS)
Robert Maxwell
In search of a Gpa1-specific
inhibitor
Gpa1 Background
• Key regulatory protein in pheromone signalling
pathway
• Tethered to the plasma membrane via interaction
with heptahelical receptor (GPCR)
• No crystal structure
• Several good crystallized homologs
Rat ~66% ID, 45% Sim, 1.5 Angstroms
• Divergent insert aa 130-234 does not include
binding site - removal has no effect on activity
Gtpase sequence conservation in
yeast
Gpa1 Splice
Gpa1
Gpa2
Sar1
Arf3
Cin4
Arf2
Arf1
Arl1
*Arrows indicate GTP binding residues in Gpa1.
Identifying Selective Inhibitor for Gpa1
• Evaluate sequence conservation within S.
cerevisiae
• Evaluate crystal structures for homology model
building
• Build/Evaluate homology model
• Dock small molecules
• Perform small molecule screen
Gpa1 Contact Residues Conserved
Gpa1 model based on 1CIP
Hinge
Gpa1 models with/out
cofactors
Gold = built with cofactors
Aqua = built without cofactors
RMSD = 0.1079
Gpa1 with bound GNP
Initial screen
•
Screened ~500 molecules from Chembank library (thanks
Ilya)
•
Used GTP, GDP, GNP, GTPgS, ATP, ADP as “controls”
•
Glide - standard speed/precision
•
Docked into 2 Gpa1 (spliced) models based on 1CIP
1) Built with Mg cofactor and GNP ligand
2) Built without Mg cofactor and GNP ligand
•
Docked into 1CIP
Cofactors play critical role in
ligand dock scores
Ligand
+Cofactors
-Cofactors
GTP
2
4
GNP
1
92
GDP
5
45
GTPgS
4
17
ATP
3
8
ADP
9
54
Gpa1 pocket built with/out
cofactors
RMSD = 0.159
GTP binding poses nearly
identical
With cofactors
No cofactors
Mg
GTP binding poses in Gpa1 models
built with/without cofactors
Moving forward
• Evaluate the use of multiple (homology) models to
enhance the rank scores
*Dock into multiple representative structures
*Perform simple scoring function across all ranked
molecules - I.e. average score, energy, etc.
• Evaluate the impact of cofactors/ligands on homology
model docking scores
*Build homolgy models of protein with many known
ligands (Cdk2)
- build with and without cofactors/ligands
*dock into several resolved crystal structures as well as
homology models
Small molecule identification
- the old fashioned way
• Perform small molecule screens on S. cerevisiae in the lab
• Powerful genetic tools make assay for inhibitor molecules
very straightforward
• 1000 - 5000 molecules can be screened in ~1month