Comparing the electrostatic properties of protein active sites and other cresset research

Comparing the electrostatic properties of protein active sites and other Cresset research

© Cresset

Agenda

> Looking at 3D RISM> Work by Mark Mackey and Paolo Tosco

> Approaches to Protein Fields> Work by Andy Smith, Susana Tomasio

2

© Cresset

New use for the XED force field?

3D-RISM

© Cresset

3D-RISM

> Analytical method for working out where water goes (Ornstein-Zernike equation)

> Conceptually equivalent to running an infinite-time MD simulation on the solvent and extracting the solvent particle densities

ℎ 𝑟𝑟12,𝜔𝜔1,𝜔𝜔2= 𝑐𝑐 𝑟𝑟12,𝜔𝜔1,𝜔𝜔2

+ 𝜌𝜌�𝑑𝑑𝑟𝑟3𝑑𝑑𝜔𝜔3𝑐𝑐 𝑟𝑟13,𝜔𝜔1,𝜔𝜔3 ℎ(𝑟𝑟32,𝜔𝜔3,𝜔𝜔2)

© Cresset

3D-RISM

> Analytical method for working out where water goes (Ornstein-Zernike equation)

> Conceptually equivalent to running an infinite-time MD simulation on the solvent and extracting the solvent particle densities

> Horribly complicated maths> GPL implementation in Amber Tools> Output is grid containing particle densities> Thermodynamic analysis to assign “happiness”

to each water

© Cresset

> Results depend on the potential function from solvent to solute u(r12, Ω1, Ω2)

> In practise, this means vdW + electrostatics> Results only as good as your potential functions> Does the XED description of electrostatics

improve the results?

Problems

© Cresset

Electrostatics from Molecular Mechanics

> XED force field – eXtended Electron Distribution> Multipoles via additional monopoles

> Huckel> separation of π and σ charges – substituent effects> find bond orders and assign hybridization – analogue N atoms

> Full MM Force Field with excellent coverage of organic chemistry and proteins

> Minimization, Conformations etc.> Additional atoms cost more than ACC> Cheaper than other multipole methods> Local polarization

H

0.5

-0.5

+0.9+0.1

© Cresset

Comparing XED with GAFF – Hydrogen Density

XED GAFF

8

© Cresset

Comparing XED with GAFF – Hydrogen Density

XED GAFF

9

© Cresset

Comparing XED with GAFF – Hydrogen Density

XED GAFF

10

© Cresset

Comparing XED with GAFF – Oxygen Density

XED GAFF

11

© Cresset

RISM with XED Conclusions

> Initial results look promising

> Water patterns around small molecules look better with XED

> Does this extend to protein environments?> Does it change the ‘happiness’ factor?

> We’ll keep you posted

12

© Cresset

A quick refresher

Field points

13

© Cresset

Field Points & Scoring

Calculate interaction energy potential with charged oxygen probe. Contour this potential

Field points indicate potential sites where protein atoms want to sit

14

© Cresset

Field Points & Scoring

Minimize Oxygen probe from many starting points onto surface of ligandCharge on probe determines which field we are calculating.

-

-

+

Field points indicate potential sites where protein atoms want to sit

15

© Cresset

Similarity Metric Relies on Field Values

A B

2ABBA

ABEEE →→ +

=BBAA

ABAB EE

ES+

=2

∑ ×=→Afp

ABABA fppositionFfpsizeE ))(()(

EA→B = “Energy”, SAB = Similiarity

A B

16

© Cresset

Example Ligand field – 1FVT

17

© Cresset

Fields and More

Proteins

18

© Cresset

Protein Field Problem

> Can’t use the ligand field points to sample the protein field> Most should be inside the protein surface

> Same problem will exist for protein field points> Protein field points will show where ligand atoms want

to be> Most should be inside the ligand surface

> But we can still compare proteins to proteins> Predict protein similarity

19

© Cresset

Validation

> Find all dissimilar proteins that bind the same ligand> Show that the field for these is similar

> What ligands?> ATP ?

> Yuk!> NADP ?

> Yuk!> Estradiol ?

> Yuk!

Lots of similar proteins with similar ligands, few diverse proteins with synthetic ligands

20

© Cresset

Problems

> Protein preparation a key step> Time consuming> Required automated process

> How to handle highly charged proteins> Scaling?

> How to handle water?> Remove?

> How to handle metals?> Point charge?

> How to define the active site?>Where is it, and where does it stop?

21

© Cresset

Charged Proteins

> Traditional ligand approach – uniformly scaled formal charges (Native)> All formal charges are divided by 8

> Uniformly scaled formal charges with a distance dependant dielectric (Native-DDD)

> Remove or scale formal charges further for solvated charged residues (Varichg)> Fully solvated formal charges: charge scaler=80> Buried in the protein: charge scaler=2

> Varichg with a distance dependent dielectric

22

© Cresset

Some Good Results – 1FVT Native

23

© Cresset

Some Good Results – 4MBS DDD

24

© Cresset

Some Good Results – 1OIT DDD+Scaling

25

© Cresset

Still Work in Progress

> New thinking for protein preparation> Checking of Asn/Gln, Ser/Thr/Tyr orientations> Possible integration with RISM work

> Validate through protein similarity rather than ligand binding ?

> Still need to understand when each field generation method works and why> DDD looks good, but is not best for all proteins

> Need ideas on how to close the ligand-protein field gap> Would be nice to compare ligand and protein fields directly,

but you can’t

26

© Cresset

Some Good Results

Ligand field =

Protein field =

27

© Cressetsmarter chemistry | smarter decisions

[email protected]@cresset-group.com

29