W AT E R M A P A N D S O LV AT I O N

EFFECTS FOR IN SILICO

DRUG DISCOVERY
B R YA N C R A M P T O N

BACKGROUND
Increasing use of in silico drug screeningnarrowing

targets for wet lab testing

Requires known crystal structure (with ligand)

Homologous protein analysis can determine structure

and find binding site computationally (loss of accuracy)

Low resolution structures improved using refinement

MANY pieces of software available, varied approaches:
Quantum Mechanics (QM) & Molecular Mechanics (MM)
Statistical (Markov chains) and combinatoric

permutations

SCHRDINGER

Schrdinger: Most reputable molecular modeling software

Many libraries, applications, workflows, distributed-computing
Glide: Ligand docking and scoring software, increasing

resolution in refinement, capable of mixed QM/MM,

combinatoric conformer searching
Rosetta: Open source docking software, increasing resolution in

refinement , (statistical searching), MM, markov models

Many more OpenBabel, AutoDock, Vina, etc.

DRUG SCREENING ISSUES

in silico screening historic issue with false (+) & (-) results
Experimental results align accurately for some proteins

and compoundswildly off for others

Sources of error:
Not ab initio (pure QM) modeling
Software does not account for entropic and

enthalpic effects of the solvent

S C H R D I N G E R ( W AT E R M A P ) T H E O R Y

Hydration sites: Sites with associated water molecules

Gibbs free energy: G = H - TS (enthalpic and entropic effects)
H: Enthalpy due to standard intermolecular interactions
Hydrogen bonds, van der waals, steric effects, electrostatic/ionic

attraction/repulsion, pi-stacking, etc.

S: Entropy due to change in disorder of the system
Ability of water molecule to move randomly in hydration-site

(degree of constraint when associated)

G: Free energy due to sum of both effects

S C H R D I N G E R ( W AT E R M A P ) T H E O R Y
Ligand disruption of hydration-site arrangement contributes

to net change in free energy of the system during binding

Favorable to displace water-molecule with G > 0

Unfavorable to displace water-molecule with G < 0
Why do waters with G > 0 stays protein-associated?
Vacuum (dewetted region) penalty: 6-8kcal/mol
Identified vacuums also critical for docking
H: Directly from the force-field energetic analysis (MM)
-TS: Extremely complicated set of statistical-

thermodynamic equations Inhomogeneous Solvation

Theory

I N H O M O G E N O U S S O LV AT I O N T H E O R Y
Displacement of water molecules

from the active site by the ligand is a

principal (if not dominant) source of
binding free energy (Abel et. al)

Very complicated mathematics used to

quantify the randomness of

associated water-molecules
Se=(-kbw)/ + gsw(r,)ln(gsw(r,)drd +

(-kb2w)/(22) + gsw(r2,2)ln(gsww(r2,2)dr2d2 -

Highly ordered water-molecules more

favorable for displacement

W AT E R M A P A L G O R I T H M P R O C E S S
Stage 1: Simulation
System is simulated for 2ns with full-explicit solvent
Heavy constraints applied on protein backbone, R

groups, and ligand structure

Stage 2: Clustering
Isolates hydration sites in region of interest then

analyzes highest-density regions, continuing until no

clusters with density > 2x bulk water

Stage 3: Thermodynamic Property Generation

Thermodynamic properties (H, -TS, G) calculated

for each hydration site along with additional

calculations and parameters which can be specified

C O N T I N O U S W AT E R M A P
Replace hydration-site model

with a 3D grid of solvent free

energy

Time-lapsed simulation

statistically represents free

energy of water molecules

New feature but gives deep

insight for analysis

Closer to reality of breathing

system

D R U G G A B I L I T Y: G L E E V E C / A B L

Highly druggable target

BCR-abl: Fusion (mutated) genes from two chromosomes

resulting in an always on tyrosine kinase signaling
pathway. Causes uncontrolled cell division (leukemia)

based on WaterMap

Many unfavorable energy

hydration sites in a row

Insight into the shape of

ideal drug-molecule

Continuous Map shows

nearly solid red-region in

active-site

Red Waters: Unfavorable (+G) Green Waters: Favorable (-G)

Numbers represent G values of displaced water molecules

D R U G G A B I L I T Y: C O X - 2

Clear druggable active site

COX-2: Responsible for formation of prostanoids (eg.

prostaglandins) in response to bodily injury. Results in
inflammation and pain. Can be inhibited to prevent both
with NSAIDs (aspirin, ibuprofen)

Tight cluster of unstable

hydration-sites, continuous
high free-energy density

Region without hydration-

sites (vacuum) which yields

high energy reward when
filled by the ligand
Red Waters: Unfavorable (+G) Green Waters: Favorable (-G)
Blue Area: Vacuum (dewetted site)
Numbers represent G values of displaced water molecules

D R U G G A B I L I T Y: P T P - 1 B
Example of an binding site

which is difficult/undruggable

PTP-1B: Negative regulator of the insulin signaling pathway,

promising potential therapeutic target for type II diabetes.
Dephosphorylates activated insulin receptor kinase.

Core of active site has a cluster

of highly stable water molecules

Hydrophobic atoms in this

region completely inhibit

binding
Only ionic atoms are capable of

displacing hydration sites (i.e.

phosphates)
Unstable waters exist in further

regions but need long drug

Red Waters: Unfavorable (+G) Green Waters: Favorable (-G)

Numbers represent G values of displaced water molecules

D R U G D E S I G N & W AT E R M A P
G and H >> 0: Favorable to displace with drug
Methyl/halogens ideal for task (site often hydrophobic)
Explains many of the magic-methyl effects
H << 0 & G ~ 0: Favorable to replace with polar

functional-group on drugnegates enthalpic penalty

Polar group forms H-bonds to replace water which

enters bulk solvent increasing entropy (net benefit)

G << 0 | Conserved: Favorable to interact with drug

Treat site as if it is part of the protein itself, if possible

intermolecular interactions with drug functional groups

A P P L I C AT I O N : S P I R O C Y C L I C S U L F A M I D E S I N
T R E AT M E N T O F A L Z H E I M E R S D I S E A S E

Build up of amyloid- plaques (A) in hippocampal and

cortical regions of brain

A generated by proteolytic cleavage of amyloid

precursor protein (APP) by -Secretase 1 (BACE1)

Research identified target sulfamide with decent binding

affinity and inhibitory efficacy

Structure-activity relationship (SAR) could not be

deduced using a normal docking energetic analysis

Needed higher potency and modification to diffuse

through BBB

A P P L I C AT I O N : S P I R O C Y C L I C S U L F A M I D E S I N
T R E AT M E N T O F A L Z H E I M E R S D I S E A S E

Using WaterMap, discovered

compound 12a (mod of 11a)

Much > binding affinity

8 unfavorable hydration sites

overlapping compound

G for site #2 extremely high

(+4.94 kcal/mol)

Assisted discovery of more

potent inhibitor and SAR

W AT E R M A P : A N T I B I O T I C T R A N S L O C AT I O N
T H R O U G H P O R I N O M P C ( G R A M N E G AT I V E )

Antibiotic translocation mechanism through porins not

understood

Hard to engineer drugs which result in high intracellular

concentrations

WaterMap used to explain entry/exit rates of 6 different

antibiotics

Accurately predicted relative rates of all compounds

Large hydrophobic regions significantly decrease rate of

compound diffusion through porin

W AT E R M A P : A N T I B I O T I C T R A N S L O C AT I O N
T H R O U G H P O R I N O M P C ( G R A M N E G AT I V E )

W AT E R M A P : P 3 8 A M A P K I N A S E I N H I B I T O R S

Direct cellular responses to stress stimuli (mitogens, heat

shock, etc.) and regulate cell functions (proliferation,

differentiation, apoptosis)

Downstream in signaling pathways of many proto-

oncogenic driver mutations

Inhibition could treat some types of cancer

Prevent transduction of signals generated by

constitutively active receptor tyrosine kinases, Ras,

Raf, etc.

W AT E R M A P : P 3 8 A M A P K I N A S E I N H I B I T O R S

kon (association constant) narrowly distributed across many

inhibitor compounds

koff (disassociation constant) widely distributed across

compounds

Hypothesized that desolvation and resolvation of binding

sites RDS for association and/or disassociation of ligand

Experimental results confirmed that koff is correlated to the

free energy change of hydration-sites upon resolvation

W AT E R M A P : P 3 8 A M A P K I N A S E I N H I B I T O R S

Average R2 value of ~0.89

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

EXAMPLE: HIV PROTEASE

D R U G G A B I L I T Y: H I V P R O T E A S E

Catalytic Diad

Moderately druggable site

WaterMap reveals many

unstable hydration-sites

Not continuous as other

examples (stable sites

between)

Stable waters in core of

active site (surrounded by

catalytic Asp-diad)
Red Waters: Unfavorable (+G) Green Waters: Favorable (-G)
Numbers represent G values of displaced water molecules

W AT E R M A P L I M I TAT I O N S
Even with crystal structure, may not have explicit water

molecules

Even with explicit water, typically only ordered,

energetically favorable, water is observed in crystal

Waters can also be in the wrong spatial location (due to

crystallization process)

Software can simulate water but not as accurate

Leads to errors
Not all active sites have important solvation effects, may

be unnecessary work to do this examination

W AT E R M A P A N D F U T U R E
Many promising results from studies utilizing WaterMap
Elucidated SARs of many ligands lacking binding

mechanism

Novel drug targets developed using SBDD

Continuous and Holo WaterMaps (bridging waters)
Specific targeting within large protein classes/families
Unique hydration sites, enthalpy, entropy
Quantum computing and ab initio feasibility

REFERENCES
Abel, R., Young, T., Farid, R., Berne, B. J., and Friesner, R. A. (2008) Role of the Active-Site Solvent in the Thermodynamics
of Factor Xa Ligand Binding. J. Am. Chem. Soc. Journal of the American Chemical Society 130, 28172831.
Brodney, M. A., Barreiro, G., Ogilvie, K., Hajos-Korcsok, E., Murray, J., Vajdos, F., Ambroise, C., Christoffersen, C., Fisher, K.,
Lanyon, L., Liu, J., Nolan, C. E., Withka, J. M., Borzilleri, K. A., Efremov, I., Oborski, C. E., Varghese, A., and ONeill, B. T.
(2012) Spirocyclic Sulfamides as -Secretase 1 (BACE-1) Inhibitors for the Treatment of Alzheimers Disease: Utilization of
Structure Based Drug Design, WaterMap, and CNS Penetration Studies To Identify Centrally Efficacious Inhibitors. J.
Med. Chem. Journal of Medicinal Chemistry 55, 92249239.
Lazaridis, T. (1998) Inhomogeneous Fluid Approach to Solvation Thermodynamics. 1. Theory. The Journal of Physical
Chemistry B J. Phys. Chem. B 102, 35313541.
Pearlstein, R. A., Sherman, W., and Abel, R. (2013) Contributions of water transfer energy to protein-ligand association and
dissociation barriers: Watermap analysis of a series of p38 MAP kinase inhibitors. Proteins: Structure, Function, and
Bioinformatics 81, 15091526.
Robinson, D. WaterMap Theory and Practical Applications. Schrodinger.
Schrodinger. (2016, June 14) WaterMap: Using Water Energetics for Lead Optimization. Schrodinger.
Tran, Q.-T., Williams, S., Farid, R., Erdemli, G., and Pearlstein, R. (2012) The translocation kinetics of antibiotics through
porin OmpC: Insights from structure-based solvation mapping using WaterMap. Proteins: Structure, Function, and
Bioinformatics 81, 291299.
Young, T., Abel, R., Kim, B., Berne, B. J., and Friesner, R. A. (2007) Motifs for molecular recognition exploiting hydrophobic
enclosure in proteinligand binding. Proceedings of the National Academy of Sciences 104, 808813.