1
Crystax Pharmaceuticals SLBarcelona Science Park
Josep Samitier 1-5, E-08028 Barcelona Tel: +34 93 403 4703 Fax : +34 93 403 4788 www.crystax.com
Fragment Screening in Drug Discovery
SEQT, Sitges, 19th-20th October 2006
Marc Martinell
SEQT, 19-20th October 2006
• The Company
• Fragment Screening
• Fragment Library
• Detection of Fragment Binding
• Structures of fragment-protein complexes
• Hit Selection and optimization
• Summary
Overview
2
SEQT, 19-20th October 2006
The Company
• CrystaX is a Structure-based Drug Discovery company.
• Founded in 2002 by recognised scientists J. Aymamí and M. Coll.
• Current team of 22, among them 12 PhD with international experience, additional technical staff.
• Advisory agreements with experts in complementary areas (Computational Chemistry, Organic Synthesis, etc.)
• Our main strength is the combination of structural biology and chemistry into a team that can address any issue in the lead-finding process.
Barcelona Science ParkAccess to top technology equipment
and labs
X-ray CrystallographyNMR for Biomolecules
Fine ChemistryCombinatorial Chemistry
MicrocalorimetrySurface Plasmon Resonance
Genomics and TranscriptomicsProteomics
SEQT, 19-20th October 2006
Business Model
Fee for ServiceCo-Development of new drugs with pharma partners using our platformtechnologyDevelopment of own pipeline
Crystax’s approach to business
B2D2Common technology platform
Fee-for-service business and R&D
collaborations
Product pipeline from own drug
discovery • Short term: in collaboration
with other companies• Mid term: own licensing
opportunities
• Deliver value to clients• Develop relationships for
licensing opportunities
3
SEQT, 19-20th October 2006
Collaborations
Fragment Screening
4
SEQT, 19-20th October 2006
1060 – 10180 Drug like molecules with Mw < 800
Fragment Screening
There is a massive amount of drug like molecules with a suitable molecular weight for drug discoveryHTS and traditional discovery techniques often start with relatively large and complex molecules.The main disadvantage of traditional approaches is that finding one right molecule amongst such a vast number is quite difficult and, moreover, it is indeed hard to, once found, jump from one chemical “branch” to another.
SEQT, 19-20th October 2006
Fragment Screening
Fragment screening allows to start with a smaller molecule and then add as much complexity as needed.
The difficulty is that you start at a pre-hit stage, where functional activity is difficult/impossible to measure.
5
SEQT, 19-20th October 2006
Fragment Screening
Adapted from: M.H. Hann et al, J. Chem. Inf. Comput. Sci. 2001, 41, 856-864
Drug-likeFragment-like
0 2 4 6 8 10
Ligand complexity
Prob
abili
ty
Probability of detection Probability of binding
KDmM µM nM pM
SEQT, 19-20th October 2006
Techniques able to detect and develop low
affinity binders are needed
NMRX-Ray CrystallographyBiophysical techniques
Fragment Screening
Drug-likeFragment-like
0 2 4 6 8 10
Ligand complexity
Prob
abili
ty
Probability of detection Probability of binding
KDmM µM nM pM
Adapted from: M.H. Hann et al, J. Chem. Inf. Comput. Sci. 2001, 41, 856-864
6
SEQT, 19-20th October 2006
Fragment Screening
R. Carr and H. Jhoti, DDT, 2002, 7, 522-527
SAR-by-NMR (S.B. Shuker et al. Science 1996, 274, 1531-1534)
SEQT, 19-20th October 2006
Fragment Screening - Examples
Hit
NO
OH
N
N
O
OH
N
FIC50 = 1.1mMIC50 = 200nM
N
OH
NH2
SN
OO
CO2H
Ki = 56µM
Ki = 1000µM
N
NH
NH2
NN
Ki = 370nM
SN
OO
HN
O
CO2H
CO2H
O
CO2H
Ki = 33nM
Lead
p38 kinase
Urokinase
ThymidylateSynthase
D. A. Erlanson et al, PNAS, 2000, 97, 9367-9372
V.L. Nienaber et al, Nat. Biotech., 2000, 18, 1105-1108
J. Fejzo et al, Chem.Biol, 1999, 6, 755-769
7
Fragment Library
Fragments are organic moleculeswith a low degree of complexityand non-reactive
SEQT, 19-20th October 2006
CrystaX’s Fragment Library
Selection process based on the newest criteria for fragment libraries.
The balance between chemical space exploration and efficiency of the hit to lead process is optimized.
commercial compounds
3,000.000
≈1000
CrystaX’sFragment
Library
8
SEQT, 19-20th October 2006
CrystaX’s Fragment Library
Molecular properties
(fragment-like)
Unwanted reactivity
commercial compounds
3,000.000
280.000
80.000
Clustering and
selection
≈1000
CrystaX’sFragment
Library
SEQT, 19-20th October 2006
CrystaX’s Fragment Library – Quality Control
Constant monitoring of false positive and/or promiscuous
binders
615 compounds ready for Fragment Screening in 71 mixtures
• Solubility, identity, purity and stability
Quality Control of individual compounds
• Designed to obtain the minimum signal overlap among compounds• Solubility, identity, purity and stability
Quality Control of mixtures of 7-9 compounds
9
Detection of fragment binding
Due to their low degree ofcomplexity, fragments are lowaffinity binders
SEQT, 19-20th October 2006
Target protein
NMR screening
Positive Fragments
• STD• Relaxation edited spectra• WaterLOGSY• TrNOE• 19F-NMR• Chemical Shift Mapping (CSM)
Continuousdevelopment
Fragment Screening by NMR
10
SEQT, 19-20th October 2006
Fragment Screening by NMR - STD
Each molecule has characteristic signals on a 1D 1H spectra
I0
B. Meyer and T. Peters, Angew. Chem. Int. Ed., 2003, 42, 864-890
SEQT, 19-20th October 2006
Fragment Screening by NMR - STD
I0 In a solution of a protein with a large excess of these molecules, their spectra is almost not affected
B. Meyer and T. Peters, Angew. Chem. Int. Ed., 2003, 42, 864-890
11
SEQT, 19-20th October 2006
I0
ISAT
When the protein is saturated with a selective irradiation, this saturation is transferred to the binding molecules. This saturation produces an attenuation of its NMR signal.
Fragment Screening by NMR - STD
B. Meyer and T. Peters, Angew. Chem. Int. Ed., 2003, 42, 864-890
SEQT, 19-20th October 2006
By subtracting both spectra, an NMR difference spectrum is obtained in where ligand molecules that bind to the target can be identified
I0
ISAT
ISTD = I0 - ISAT
δ (ppm)
Fragment Screening by NMR - STD
B. Meyer and T. Peters, Angew. Chem. Int. Ed., 2003, 42, 864-890
12
SEQT, 19-20th October 2006
Saturation Transfer Difference (STD)
Fragment Screening by NMR
1D 1H mixture
Relaxation Edited Experiment The signals of fragment binders disappear
The signals of fragment binders appear
Direct deconvolution from mixtures of fragments
The complete library is screened by NMR
SEQT, 19-20th October 2006
1D 1H spectrum
STD spectrum
Fragment Screening by NMR – Examples
13
SEQT, 19-20th October 2006
Direct Deconvolution
STD
CXL-23
CXL-20
CXL-18
CXL-12
CXL-11
CXL-9CXL-7
CXL-8
CXL-28
CXL-24
Fragment Screening by NMR – Examples
SEQT, 19-20th October 2006
Direct Deconvolution
Positives fragments: CXL-20 and CXL-23
STD
CXL-23
CXL-20
Fragment Screening by NMR – Examples
14
SEQT, 19-20th October 2006
STD
STD upon addition of known active-site ligand
Fragment Screening by NMR – Examples
SEQT, 19-20th October 2006
CXL-23 CXL-23
STD
STD upon addition of known active-site ligand
Only compound CXL-23 interacts
with the active site
Fragment Screening by NMR – Examples
15
SEQT, 19-20th October 2006
1D 1H mixture
Relaxation Edited Experiment
STD
1D 1H compound CXL-212
The signals of fragment binders disappear
The signals of fragment binders appear
Compound CXL-212 interacts with the
protein
Fragment Screening by NMR – Examples
SEQT, 19-20th October 2006
addition of known active-site ligand
+ Competitor
1D 1H mixture
Relaxation Edited
Experiment
STD
1D 1H compound CXL-212
Fragment Screening by NMR – Examples
16
SEQT, 19-20th October 2006
Compound CXL-212 interacts with the active site of the
protein target
Using competition studies by NMR ligands for specific binding sites can be
identified
1D 1H mixture
Relaxation Edited
Experiment
STD
1D 1H compound CXL-212
Fragment Screening by NMR – Examples
SEQT, 19-20th October 2006
3%Oncology5
1%-6
5%Oncology3
4%Auto-immune4
2
1
Project
Oncology
Inflammation
Field
5%
10%
Hit rate
Projects at CrystaX
17
Structures of fragment-protein complexes
Due to their low affinity and smallsize fragments are more difficult tostudy by Xray crystallography
SEQT, 19-20th October 2006
Fragment Screening by Xray
Positive Fragments
Crystallization (co-crystallization or soaking)
Structure determination
Fragment Hits
18
SEQT, 19-20th October 2006
Sparse matrix screening forinitial crystallization conditions
Optimization of conditionsfor crystal growth
Characterizationof crystals
Collection of“apo”-datasets
Large scale productionof crystals
Soaking of crystalswith ligands
Collection of diffraction datain the presence of inhibitor
Co-crystallizationunder analogous conditions
Ab-initio screening forcrystallization conditions
in the presence of inhibitors
Reproduction ofknown crystallization conditions
High-Throughput Crystallography
> 1000 > 1000 conditionsconditions
SEQT, 19-20th October 2006
Automatic Data Processing & Analysis of Results
Data are processed automatically using commercial software and a modular suite of proprietary scripts.
Resulting electron densities are inspected individually, analyzed and classified. Models of the protein-ligand complex are partially or completely refined, depending on the needs of the individual project.
19
SEQT, 19-20th October 2006
Structures of fragment-protein complexes
Fragment Screening renders a high amount of structural data, thus increasing the efficiency of the hit to lead process
SEQT, 19-20th October 2006
ongoing5%Oncology3
45%5%Oncology2
40%
Hit confirmation
(Xray)
1
Project
Inflammation
Field
10%
Hit rate (NMR)
Projects at CrystaX
20
SEQT, 19-20th October 2006
NMR (ligand-based detection) Xray
NMR (protein-based detection) Xray
SPR Xray
Xray
Biochemical assays Xray
-
Alternative approaches
The combination of ligand-based NMR methods and Xraycrystallography renders the most general approach with the
minimum consumption of protein sample
Primary screening method Hit confirmation
Virtual Screening Xray
Hit Selection and Optimization
Several prioritization criteria are needed in order to select the most interesting hits
21
SEQT, 19-20th October 2006
Hit validation and selection
Evaluation of other molecules within its cluster and/or molecules that contain the same binding motif
Hit selection
Selected HITS
Fragment Hits Preliminar SAR
Validation of Binding-mode
New chemical structures
SEQT, 19-20th October 2006
Biophysical methods (B2D2TM)• SPR Kon and Koff• Microcalorimetry ∆Hº, ∆Sº
………
Synthesis
Molecular Modeling
Fragment Screening!!
Activity Assays
LEADS
Optimization
Selected HITS
Hit Optimization
Fragment Hits
22
SEQT, 19-20th October 2006
mM
µM
nM
pM
Fragment Screening
Activity Assays
Lead
Hit
?
Biophysics-Based Drug Discovery (B2D2TM)
SEQT, 19-20th October 2006
mM
µM
nM
pM
Fragment Screening
Activity Assays
ADMEToxicity
SelectivityPatentability
B2D2TM
Lead
HitBiophysical
characterization renders high quality data and increases
the efficiency of Hit2Lead process
Biophysics-Based Drug Discovery (B2D2TM)
23
SEQT, 19-20th October 2006
Biophysics-Based Drug DiscoveryB2D2TM
NMR X-Ray crystallography
Biophysics-Based Drug Discovery (B2D2TM)
SEQT, 19-20th October 2006
Biophysics-Based Drug DiscoveryB2D2TM
NMR X-Ray crystallography
• Unique technique for thermodynamic data (KD, ∆Hº, ∆Sº)
• Low throughput• Label free• High protein consumption
Calorimetry
∆G ∆H -T∆S ∆G ∆H -T∆S
Same KD but differentthermodynamics
Biophysics-Based Drug Discovery (B2D2TM)
24
SEQT, 19-20th October 2006
Use of ITC in Fragment Evolution
10 < nKA[M]T < 100
ITC requirements
KA ~ 104 – 108
KD ~ 100µM – 10nM
Soluble compounds
High amounts of sample
Fragment Hits: mM - µM
Lead compounds: nM - pM
SEQT, 19-20th October 2006
Use of ITC in Fragment Evolution
25
SEQT, 19-20th October 2006
Biophysics-Based Drug DiscoveryB2D2TM
NMR X-Ray crystallography
• Kinetic data (KA, kon, koff)• Medium-high throughput• Immobilization needed• Low protein consumption
Calorimetry
Biacore (SPR)
Time (seconds)
0
500
0 120 240 360 480
RU
Same KD butdifferent kinetics
Biophysics-Based Drug Discovery (B2D2TM)
SEQT, 19-20th October 2006
Biophysics-Based Drug DiscoveryB2D2TM
NMR X-Ray crystallography
Calorimetry
Biacore (SPR)
Fluorescence spectroscopy
• Affinity constant (KA)• Medium-high throughput• Fluorescent-label needed• Low protein consumption
Biophysics-Based Drug Discovery (B2D2TM)
26
SEQT, 19-20th October 2006
Fragment ScreeningB2D2™
Summary
Novel Structures
Renders….
… but also….
SEQT, 19-20th October 2006
Summary
Novel Structures
Lead Optimization
Structural watersProtein Hot-spots
New binding modes
Protein flexibility
New binding sites
Fragment ScreeningB2D2™
Biophysical characterization
27
Joan AymamíMiquel Coll
Maria KontoyianniIngo KorndoerferMontse SolerXavier BarrilIsabel NavarroFranck ChevalierTeresa LuqueIrena Bonin
Unitat RMN (SCT-UB)Unitat Químic Fina (SCT-UB)Unitat Citometria (SCT-UB)Unitat Química Combinatòria (PCB)Plataforma Raigs-X (PCB)A. Llebaria (RUBAM, IIQAB-CSIC)R. Gutierrez (IMIM-UPF)F. J. Luque (UB)
Carolina MoralMarta MasipSarah SotilVerónica ToledoLaura QuintanaSonia SorianoAnja LeimpekMarian DomínguezMarta Martín
Acknowledgements
Thank you for your attention