GeneFun

1
GeneFun

• Contributions to WP4 5
• 4. Structure-based prediction of function.
• 5. Context-based function prediction Meta-server

2
Prediction Design of Protein-Protein
interactions based on structure
3
What do we need

• Good Structural Templates
• An accurate Force Field
• A Protein design algorithm

4
Protein Design
Target structure sequences

5
FOLD-X A force-field for Automatic Protein
Design
http//FOLD-X.EMBL-Heidelberg.DE
Protein Design Point Mutations (0.6
Kcal/mol) Water Bridge Prediction (95
success) Ion binding Prediction (95 success) and
Kds Protein Complexes binding energies Structure
quality assessment
Unrefined 26.5 5 UnrefinedH20 22.5
2 Refined 3.7 2 RefinedH20 0.9
2 X-ray 0.09
http//model-x.embl.de/
Guerois et al JMB (2002).
DcR1
DcR2
6
Generalisation fo the method to other molecules
Ion binding sites for Ca,Mg,Zn,Co,Cu, with 95
accuracy Ion binding energy for Ca and Mg.
7
Protein-Protein Interactions
Ras with RalGDS-RBD (data from Wittinhofer
col).
Magenta Predicted HOH bridges Red X-ray waters
(only in the binding Interface)
8
Prediction of protein-protein interaction based
on structure
Gregorio Fernandez-Ballester Christina Kiel Luis
Serrano
9
Protein Design Protein-Protein Interactions
PDZ Domain
Could we design proteins to recognize any protein
without knowing its 3D structure? Partly yes
using the C-terminus Of proteins.
ClassI ClassII ClassIII ClassIV
PDZ C-terminus recognition motifs.
Blue Ligand SC Green Interacting SC
Reina et al. Nature Struct.Biol. 2002 In press.
(collaboration with C.Gonzalez)
10
Example Architecture and binding of SH3 domains
(Collaboration with A. Valencia)
11
Homology Modeling of SH3 domains
12
SH3-ligand complex construction
13
model
X-Ray complexes
14
Computation of ligand positions
Ligand positions (7-10)
Models (5 or 7)
Ligand positions are computed individually
15
Check sequence ?
Best complex Best scores
16
Candida SH3 Target
17
PATTERNS derived from the prediction Underlined
consensus positions X(n) Any natural amino acid
n times Type I (includes mod-1 to
mod-3) YFM-APIV-YHFP-KRQHN-HKPY-WYFIV-
PYRH-KPM-HRLYP-YWPRK X(3)-KRQNH-HKPY-
PYRH-KPM-HRLYP-YWPRK (used) TYPE II
(includes mod-4 to mod-6) KRHFNY-PIY-HKY-LK
RIFP-HLPYF-PYT-LKRY-YWH KRHFNY-PIY-H
KY-LKRIFP-HLPYF-X(1)-LKRY
(used) KRHFNY-P-X(1)-LKRIFP-P-X(1)-LKRY
(more restrictive) (used) TYPE III
(mod-7) HKR-HY-MIYL-YFWIVP-HP-MWY-KTH
P HKR-X(1)-MIYL-YFWIVP-X(1)-MWY-KTHP
(used) TYPE different (mod-8) HKQR-PY-KPY-
IKLVM-FKY-KR-KA-NYHKL-KRI HKQR-PY-X
(1)-IKLVM-FKY-KR-X(1)-NYHKL-KRI (used)
18
(No Transcript)
19
Example Ras and Ras binding domains
Ras proteins proteins containing
RA (Ras association) domains
and RBDs (Ras binding domains)
Arl 2 Arl 3
Can we predict which domains bind to Ras proteins
based on the sequence/structure?
20
Structure of Ras in complex with Ras binding
domain
21
(No Transcript)
22
Comparison between experimental and Predicted
Free energies of binding Between Ras or Rap
proteins and WT and mutants forms of AF6, Rin1
and mNORE1
23
lt -4 Kcal/mol /- gt -4 lt -1 gt -1
Kcal/mol
RalGDS RA AF6RA1 RA AF6RA2 RA RASSF1C
RA mNore1 RA Rin1 RA Rin2 RA
PDZGEF RA -/ Rain RA Krit1 RA spByr2
RA scCYR1 RA EpacI RA - - EpacII RA ? Re
pacI RA - PLCeRA1 RA - - PLCeRA2 RA PI3K RA
DAGK_RA2 RA ? MyosinIXARA - - MyosinIXBRA
GRB7 RA C12orf2 RA ? -/ C11orf13 RA ? -
Als2 RA RIAM RA Nexin27 RA
? -/ cRaf RBD mTiam RBD
- -/ RGS12RBD1 RBD ? - RGS12RBD2 RBD ? Ubi
quitin UBQ - - ISG15 UBQ ? - BAG-1 UBQ ? - Ubiquil
in-1 UBQ ? -
24
Domains Currently analyzed

• SH3
• PDZ
• SH2
• Ras associated domains

Under Development
Prediction of DNA and RNA binding
25
S. Cerevisiae SH3 Network
26
Automatic domain-ligand prediction
27
Deliverables D4.1 framework for data
representation and management Month 2 D4.2
protein structure modelling engine coupled to the
function prediction system Month 4 D4.3 database
of enzyme structures coupled to the function
prediction system Month 6 D4.4 linear and
no-linear functional signatures coupled to the
function prediction system Month 20 D4.5
identification of functional site by stability
measures for enzymes Month 6 D4.6 identification
of interfaces of obligate (high affinity)
complexes Month 12 D4.7 identification of
interaction partners for SH3 domains Month
8 D4.8 identification of interaction partners for
other types of peptide-binding modules Months
12-36 D4.9 tailored structure based function
prediction prototype for 5 fold families Month
24 D4.10 prediction of the effects of SNPs
Month 18-36