water networks in drug design
TRANSCRIPT
Outline
• Introduction to importance of understanding water networks
• Locations of waters and energetics of the waters in drug design – Potency – Selectivity
• Effect of ligand binding on water energetics – implications for ligand design
• Free energy perturbation (FEP) calculations
• Real world examples
Why Is Water Important?
• Water is everywhere in biological systems
• Apo protein binding sites are mostly filled with water
• Water is a direct competitor in ligand binding
• Locations of all waters in active site not provided by crystallography
• Water energetics cannot be determined from structure alone
• Displacement of high energy (unhappy) waters can lead to big potency gains
Thermodynamic Decomposition of Ligand/Protein Binding
Solvated Ligand Solvated Apo
Protein
Desolvated Ligand
Solvated Ligand in Bioactive
Conformation
Ligand-Induced Desolvated Protein
Binding Site
Solvated Protein in Ligand-Induced
Conformation
Protein/Ligand/Water Complex
DGbind = DGi=1
5
å i( )
∆G( 5 ) ∆H reward ∆Srot / t rans penalty
∆G( 1 ) ∆Hconf penalty ∆Sconf penalty
∆G( 3 ) ∆H penalty ∆S reward
∆G( 2 ) ∆Hconf penalty ∆Sconf penalty
∆G( 4 ) ∆H ? ∆S ? WaterMap
WaterMap Background
• Predicts locations and energetics of active site waters –Molecular dynamics simulation with explicit waters
• Protein is rigid
– Identifies regions where water molecules most often reside: hydration sites
• Not limited to crystallographic waters
– Entropic and enthalpic properties, as well as DG, of each hydration site are calculated using statistical methods • Lazaridis T. et al. J. Phys. Chem. B 1998, 102, 3531-3541
Young, T., et al. PNAS 2007, 104, 808 Abel, R., et al. J. Am. Chem. Soc. 2008, 130, 2817
Hydration Sites and Crystallographic Waters
Thrombin (2UUF)
– 1.26 Å resolution
– 41 crystal waters vs.
115 hydration sties
within the active site
– >90% of crystal
waters have hydration
sites within 2.0 Å
= hydration site
= crystal water
WaterMap and Potency
• WaterMap computes the entropy and enthalpy of “hydration sites”
• These can be used to rationalize SAR, drive potency, and tune selectivity – Green = stable – Red = unstable
• Provides a “map”
Stable
(happy)
waters
Unstable
(unhappy)
water
Thrombin
S1 pocket
Abel, R., et al. ChemMedChem 2011, 6, 1049
Hydration Sites and Drug Design: Thrombin
3 stable waters with favorable enthalpy
most unstable water in binding site
A
Abel, R., et al. ChemMedChem 2011, 6, 1049
Thrombin
S1 pocket
Thrombin WaterMap Scoring
B
3 stable waters with favorable enthalpy
most unstable water in binding site
10,000 nM 9,010 nM
A B C
Abel, R., et al. ChemMedChem 2011, 6, 1049
Thrombin WaterMap Scoring
B
3 stable waters with favorable enthalpy
most unstable water in binding site
10,000 nM 9,010 nM
654 nM
A
D
B C
Avoiding a stable water
Abel, R., et al. ChemMedChem 2011, 6, 1049
Thrombin WaterMap Scoring
B
3 stable waters with favorable enthalpy
most unstable water in binding site
10,000 nM 9,010 nM
654 nM 89 nM
A
D E
B C
Partial displacement
Abel, R., et al. ChemMedChem 2011, 6, 1049
Thrombin WaterMap Scoring
B
3 stable waters with favorable enthalpy
most unstable water in binding site
10,000 nM 9,010 nM
654 nM 89 nM 3 nM
A
D E F
B C
Full displacement
Abel, R., et al. ChemMedChem 2011, 6, 1049
WaterMap in Drug Design
• Provides a clear roadmap for ligand design – Which regions of active site are most important for potency – Which regions should be avoided, if possible – What type of functionality to place in each region
• Displace hydration site with hydrophobic group when enthalpic term high
• Replace hydration site when entropic term high and enthalpic term near zero
– Optimal directionality of H-bond donors and acceptors
Water Energetics and Selectivity
• Comparison of WaterMaps of different proteins can provide guidance for the design of selective compounds
McInnes, C., et al., Chem Biol, 2004. 11, 525
CDK2 and CDK4 (homology model)
High-energy water is displaced by MeNH- in CDK4 but not CDK2
N
N
HN
N
HN
S
N
CH3
R1
Kinase Inhibition (uM)
R1 IC50(CDK4) IC50(CDK2) Selectivity
Me- 0.64 1.1 1.7
MeNH- 0.007 0.22 31.4
Apo WaterMap Summary
• Advantages – Single simulation – Predictive for buried, fully displaced waters
• Limitations – Energetics of water rearrangement is not accounted for
• Complex water networks – Protein rigid – Does not include ligand desolvation – Doesn’t rigorously address interactions between ligand and protein
• Holo WaterMaps can be useful when water networks are involved – Holo WaterMaps: simulation run with ligand present
Holo Watermaps: PI3K Selectivity
• Highest energy water molecule in site resides near point of substitution
+6.0
Holo Watermaps: PI3K Selectivity
• Energetics of water similar for beta and delta
• Water molecule is displaced by R-enantiomer
• Equal gain in potency for beta and delta +6.0
R
Holo Watermaps: PI3K Selectivity
• S-enantiomer: Holo watermap for delta isoform exhibits higher energy hydration site than beta isoform with S-Me compound
• Selectivity: 20x beta to delta
+5.9 +6.7
Beta Delta
Holo WaterMaps
• Advantages – Better representation of true solvation thermodynamics upon
binding – Often provides insight into complex water networks and solvent-
exposed regions
• Disadvantages – More computationally intensive
• Separate simulation for each ligand
– Protein rigid – Does not include ligand desolvation – Doesn’t rigorously address interactions between ligand and protein
Free Energy Perturbations (FEP)
DDGbinding = DG1 – DG2 = DGA – DGB
A
B
1 2
Predicts relative free energies of binding
• Over 300 perturbations tested w/ identical protocol
– RMSE ≈ 1.1 kcal/mol
|ΔΔGFEP – ΔΔGExpt.| (kcal/mol)
Pe
rce
nta
ge
46.2%
24.8%
15.4%
7.4% 6.2%
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
< 0.6 0.6-1.2 1.2-1.8 1.8-2.4 >2.4
Schrödinger FEP Retrospective Accuracy
-15
-14
-13
-12
-11
-10
-9
-8
-7
-6
-5
-4
-15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4
BACE
CDK2
JNK1
MCL1
P38
PTB1B
THROM
TYK2
ΔG
FEP
(kc
al/m
ol)
ΔG Expt. (kcal/mol) L Wang, et al. J. Am. Chem. Soc. 2015, 137, 2695
FEP
• Advantages – Takes into account all aspects of binding energy equation – Flexible protein and ligand – Better quantification of ligand-protein interactions – In the majority of cases, allows for rank ordering of compounds
within error range
• Disadvantages – Computationally intensive – Limitation on type of perturbations that are amenable – Not applicable to diverse sets of ligands
Recent Collaboration
Round Number of synthesized
compounds*
Ki of best compound
VS --- 2 µM
1 4 82 nM
2 10 14 nM
*Does not include standards, synthetic intermediates and by-products
• Virtual screen (VS) with Glide SP and WScore
• Two rounds of optimization • WaterMap for design • FEP+ to prioritize for
synthesis
25
Acetyl CoA Carboxylase (ACC): Master Regulator of Fatty
Acid Synthesis & Oxidation
Acetyl
-CoA
ACC
Inhibit
or
ACC1
Malonyl-
CoA
Malonyl
-CoA
Fatty acid synthesis Fatty acid oxidation
ACC2
26
Fatty Liver Spectrum Disorders: Substantial Unmet Medical
Need
% of Population 70-83%1 3-5%2 5-20% of
NASH Patients3
1 in 200 NASH
Patients3
Prevalence
(US)4 9-16M 0.5-3.2M 45-80K6
Annual HCC
Incidence (US)5 15K
1. Arthur J. McCullough, Cleveland Clinic at the AASLD – FDA Workshop on NASH, Sep 5, 2013
2. AASLD Practice Guidelines for NAFLD, 2012 3. Schuppan et al., Liver International, 2010
4 US Census Bureau, 2013 estimate: 317 million 5. Venook et al. The Oncologist 2010;15(suppl 4):5–13
6. Will develop NASH-related HCC over time Graphic: Cohen et al., Science . 2011 June 24; 332(6037): 1519–1523
Normal Liver
Simple
Steatosis
NASH
Cirrhosis Hepatocellular
Carcinoma
12 - 27%1
ND-630
Liver : Muscle 100:1
ND-654
Liver : Muscle 2700:1
27
WaterMap Identified Unexploited Potency Opportunities
High-energy waters in soraphen allosteric binding site
6.0
1.0
2.0
3.0
4.0
5.0
∆G kcal/mol
28
From Binding Mode to Man in 42 Months
0 12
Months
Milestone
Achieved
Confirm Binding
Mode (X-ray) In Vivo Target
Engagement
16
In vivo Efficacy
42
Safety
First in human
5,000 Ideas
175
Total synthesized compounds
8,000 nM <10 nM Potency
ACC1, ACC2
[1.3MM] 10,000
225
<1 nM
Schrödinger Drug Discovery Group
• Ramy Farid • Mark Murco • Leah Frye • Jeremy Greenwood • Mark Brewer • Lidia Cristian • Mats Svensson • Sathesh Bhat • Shaughn Robinson • Josh Kennedy-Smith • Carolyn McQuaw • Shawn Watts • Mee Shelley
• Dragon Cirovic • Sarah Boyce • Markus Dahlgren • Jonathan Gable • Kyle Konze • Nick Boyles • Amy Rask • Mary Beth Grimes • Kyle Marshall • Josh Staker • David Casio • Will Richards
• Robert Abel • Jennifer Knight • Goran Krilov • Lingle Wang • Sayan Mondal • Tyler Day • Jeff Bell • Shulu Feng
• Byungchan Kim
Schrödinger Drug Discovery Group
• Ramy Farid • Mark Murco • Leah Frye • Jeremy Greenwood • Mark Brewer • Lidia Cristian • Mats Svensson • Sathesh Bhat • Shaughn Robinson • Josh Kennedy-Smith • Carolyn McQuaw • Shawn Watts • Mee Shelley
• Dragon Cirovic • Sarah Boyce • Markus Dahlgren • Jonathan Gable • Kyle Konze • Nick Boyles • Amy Rask • Mary Beth Grimes • Kyle Marshall • Josh Staker • David Casio • Will Richards
• Robert Abel • Jennifer Knight • Goran Krilov • Lingle Wang • Sayan Mondal • Tyler Day • Jeff Bell • Shulu Feng
• Byungchan Kim
Acknowledgements Applications Sciences Thijs Beuming Daniel Cappel Roy Kimura Michelle Hall Daniel Robinson Madhavi Sastry Woody Sherman Devleena Shivakumar Thomas Steinbrecher Dora Warshaviak Sanofi (PI3K) Thomas Bertrand Frank Halley Andreas Karlsson Magali Mathieu Herve Minoux Laurent Schio
Leadership Ramy Farid Mark Murcko Scientific Development Yuqing Deng Ed Harder Teng Lin Levi Pierce Justin Xiang Yujie Wu Friesner Lab @ Columbia
Scientific Advisors Bruce Berne John Chodera (new) Rich Friesner Bill Jorgensen Ron Levy David Mobley (new) Vijay Pande (new) D. E. Shaw Research Michael Bergdorf Justin Gullingsrud Ross Lippert Charles Rendleman Huafeng Xu
Acknowledgements: Nimbus
32
BD & Strategy
World-Class
Medicinal
Chemists
Outstanding
Biology
Expertise
Daniel Lynch –
Executive Chairman
BMS, ImClone,
Stromedix, Avila
Jonathan Montagu –
VP, Business
Development &
Operations
Concert, J&J, Chiron,
Biogen Idec
Ron Wester, PhD –
Head of Chemistry
23 years medicinal
chemistry at Pfizer
Donna Romero, PhD –
Head, Nimbus Iris
(IRAK4)
20 years at Pfizer
Craig Masse, PhD –
Head Nimbus Zeus
(Tyk2)
Amgen, Concert
Pharmaceuticals
Gerry Harriman, PhD –
Head, Nimbus Apollo
(ACC)
VP, Chemistry at Galenea,
Millennium
Rosana Kapeller, MD, PhD –
Chief Scientific Officer
Aileron, Millennium, Dana-
Farber
William Westlin, PhD –
Head Preclinical Research
& Early Dev.
Avila, Praecis, Pharmacia
James Harwood, PhD –
Biochemistry
Pfizer
Incorporation of Water Energetics Into Docking: WScore
• Novel scoring function – Integrates water energetics (WaterMap) – Very ‘hard’, empirical scoring function – Designed to improve hit rates – Scoring function based on physics of binding – Employs multi-structure protein ensembles when possible
• Advantages of WScore – Hit identification – Understanding physics of protein-ligand binding – Recovery of a diverse set of chemotypes in screening