ShOULD i STAY OR
SHOULD I STAY:
PROTON TRANSFER
REVISITED
Paul Czodrowski
SPPEXA 2016 // Software for Exascale Computing
Proton transfer?
standard pKa values
protonation state
1
pKa value
0.5
0
1
14
pH
2
amino acid
pKa value
Arg
13.0
Asp
4.0
Glu
4.4
His
6.3
Lys
10.4
Tyr
9.6
1
pKa
G
2.303 kT
3
size of studied systems
pKa calculations?
empirical
implicit
models
molecular
dynamics
quantum
mechanics
speed
pKa calculations?
computed
experimentally determined
Solvent
AH(aq)
Gdiss
G (AH)
A-(aq)+H+(aq)
G (A-)
pKa
Protein
P-AH
P-A-+H+
Not to forget: We get pKa values for all titratable residues!
4
Development of peoe_pb: Our partial charge methodology
solvent (e=80)
Implicit solvent model
Partial charges
PoissonBoltzmann
equation
protein
(e=20)
Electrostatic
potential
Partial Equalisation of Orbital Electronegativities (PEOE)
5
: orbital electronegativity
= 0.5*(IP+EA)
= a+b*q+c*q2+d*q3
distribution of
k
q

 Bk   Ak k



A
Validation of our partial charge methodology peoe_pb
pKa values in proteins
calculated pKa
calculated Gsolv [kcal/mol]
Solvation free energies of small molecules
r2
= 0.78
RMSD = 1.57
6
experimental Gsolv [kcal/mol]
eProtein=20
experimental pKa
Protonation changes detected by ITC
Thrombin
Trypsin
Hmeas  Hbind  nHion
7
Protonation effects for trypsin
Thrombin
Trypsin
1x
1b
4
8
5
1c
1d
trypsin/1b vs trypsin/1c
ITC:
n=+0.90
FDPB:
n=+0.51
*
His57: large pKa shift
*proton
9
uptake
ITC:
n=±0.0
FDPB:
n=±0.0
HIV protease – apo state
1HHP
3HVP
catalytic dyad
Asp25
10
Asp25‘
pKa1
pKa2
Catalytic dyad
Experiment
3.1 – 3.7
4.9 – 6.8
Mono-protonated
Calculation
3.8
6.8
Mono-protonated
DMP-323 bound to HIVP
1QBS
DMP-323
catalytic dyad
Asp25
11
Asp25‘
pKa1
pKa2
Catalytic dyad
Experiment
> 7.2
> 7.2
Doubly protonated
Calculation
5.3
10.7
Doubly protonated
What else can one do with pKa calculations?
12
pKa values & covalent bonds?
Addition Reaction of Model Compounds with Glutathione
taken from:
http://dx.doi.org/10.1021/jm400822z
13
The cysteinome of the kinome
taken from:
http://dx.doi.org/10.1021/jm101396q
14
Set-up of the calculations
• Per target kinase, all public PDB structures are used
• The ligand is not considered in the calculation
• Processing/Calculation is done by OpenEye‘s protein_pka
Amino Acid
Dictionary
PB pKa
Protein
PDB
Ligand
15
Protein
pKas
Analysis of the calculations
Box plots for all CYS pKa values per protein
Box plots for all CYS SASA values per protein
pKa
SASA
CYS pKa value
„out of value“
(i.e. >20)
CYS
model pKa value
16
That‘s my NULL model
EGFR: Covalently attacked CYS-797
3w2p
SASA: 38,96 Å2
(w/o covalent Inhibitor)
pKa
Cys.781
Cys.797
Cys.775
3w2o
SASA: 28,49 Å2
(w/o non-covalent inhibitor)
Cys.797
apo structures
Cys.818
Covalently
attacked
Cys.950
Cys.939
Cys.797
17
Cys797
18
SASA
pKa
EGFR: Covalently attacked CYS-797
EGFR: Covalently attacked CYS-797
Experimental determination of the pKa value
Oral communication at the GordonConference ComputerAided Drug Discovery
pKa EGFR.CYS797 =
Work done
at Pfizer
6.53 +/- 0.05
Site point mutation EGFR/CYS at position 796  pKa = 8.43
Site point mutation EGFR/CYS at position 798  pKa = 8.12
19
Summary
… because
• They tell you something about the
protonation effect of ligand-protein binding
pKa
calculations
are cool
20
• The nucleophilicity of the CYS residue
seems to be related to the predicted pKa
Why not consider protonation changes
in long-scale MD simulations?
Acknowledgment
MERCK
• Carl Deutsch
• Christoph Scholz
21
•
•
•
•
Anthony Nicholls
Mike Word
Jose Batista
Gunther Stahl
•
•
•
•
Gerhard Klebe
Christoph Sotriffer
Frank Dullweber
Ingo Dramburg