How to Bind a Moving Target - Flexibility and Entropy in

How to Bind a Moving Target Flexibility and Entropy in
Protein-Protein Binding.
Johan Leckner
Raik Grünberg & Michael Nilges
Outline
➀ Models for protein-protein association
➁ Conformal sampling
➂ Flexibility before and after binding
➃ Conformal entropy
➄ Recognition between structure ensembles
➅ A unified model for protein-protein association
➀ Models
➀ Models
recognition dominated by ...
Key-lock
... short range forces (vdW, ee, (desolvation))
[Fischer 1894]
Induced fit
... long range forces (ee, (desolvation))
[Koshland 1958, .. Camacho et al. 1999, Selzer & Shreiber 2001, ...]
Conformer selection ... short range forces & conformer equilibrium
[MWC 1965, Foote & Millstein 1994, ..., Kumar et al. 2000]
➀ Models
Problem with model ...
Key-lock
... short range forces (vdW, ee, (desolvation))
... not consistent with free structures
Induced fit
... long range forces (ee, (desolvation))
... not compatible with short-range forces
Conformer selection ... short range forces & conformer equilibrium
... not consistent with the fast
pace of recognition
➀ Time scale of recognition
Northrup & Erickson 1992:
...series of microcollisions, ≈100% recognition success
Window of opportunity:
lower limit: 400 ps -2 ns (Brownian dynamics simulations)
upper limit: 10 ns (exp., absence of unspecific interactions)
How large bound fraction do we need for 50% recognition success?
➡ 4% (400 ps); 1%(10 ns)
➁ Conformal sampling
➁ Molecular dynamics
Sampling strategy:
10 parallell 50 ps trajectories
9 Å layer of explicit water
Two setups:
“normal” MD
principal component restrained (PCR) MD
MD
PCR-MD
➁ Complexes studied
IDa
Receptor / Ligand
c01
c02
c03
c04
c05
c06
c08
c11
c13
c14
c15
c16
c17
c19
c20
c21
c22
Trypsin / Amyloid B-protein precursor inhibitor domain
A-chymotrypsinogen / Pancreatic secretory trypsin inhibitor
Kallikrein A / Pancreatic trypsin inhibitor
Subtilisin BPN / Subtilisin inhibitor
Extracellular domain of tissue factor / Antibody Fab 5G9
Humanized anti-lysozyme Fv / Lysozyme
Anti-lysozyme antibody Hyhel-63 / Lysozyme
Barnase / Barstar
Ribonuclease inhibitor / Ribonuclease A
Acetylcholinesterase / Fasciculin-II
HIVB-1 NEF / FYN tyrosin kinase SH3 domain
Uracil-DNA glycosylase / Inhibitor
RAS activating domain / RAS
Glycosyltransferase / Tendamistat
CDK2 cyclin-dependant kinase 2 / Cyclin A
CDK2 cyclin-dependant kinase 2 / KAP
Heteromeric G-protein / Transductin Gt-A
a Complex
PDB codes with chain / model identifier
rec
lig
com
1BRA
1AAP(A)
1BRC(E:I)
2CGA(A)
1HPT
1CGI(E:I)
2PKA(AB)
5PTI
2KAI(AB:I)
1SUP
3SSI
2SIC(E:I)
1FGN(LH) 1BOY
1AHW(AB:C)
1BVL(AB)
3LTZ
1BVK(AB:C)
1DQQ(AB) 3LTZ
1DQJ(AB:C)
1A19(A)
1A2P(A)
1BSG(A:E)
2BNH
7RSA
1DFJ(E:I)
1VXR
1FSC(A)
1FSS(A:B)
1AVV
1SHF(A)
1AVZ(B:C)
1AKZ
1UGI(A)
1UGH(E:I)
1WER
5P21
1WQ1(R:G)
1PIF
2AIT(1)
1BVM(P:T)
1HCL
1VIN
1FIN(A:B)
1B39(A)
1FPZ(A)
1FQ1(A:B)
1TBG(AE)
1TAG
1GOT(A:BG)
identifier used throughout the paper (retained from www .bmm.icnet.uk/docking/systems.html);
in the bound, but not in the free structure (-, vice versa).
b Size
Size b
rec
lig
223
56
245
56
232
58
275 108
248 211
224 129
424 129
108
89
456 124
532
61
99
59
223
83
324 166
495
74
294 252
290 176
408 314
$res c
0
0
-1
-1
-11
0
0
2
0
0
2
-1
-4
-2
12
13
13
in residues; c number of residues resolv ed
➂ Flexibility before and after binding
➂ Free binding interfaces
...are flexible compared to random surface patches.
15
11
21
11
10
7
2
12
4
6
2
7
7
3
1
14
8
2
4
4
17
4
11
8
5
7
3
6
2
10
3
1
standard
deviation (SD)
0.5
0
24
16
43
16
12
11
27
3
7
15
15
9
10
6
24
4
23
7
5
10
40
8
27
0.5
31
14
1
12
8
23
B backbone atoms (10 x 50 ps)
15
4
1.5
10
3
2
5
0
15
4
flexibility [Å]
1.5
7
2
ligand
receptor
average binding
patch flexibility
A all atoms (10 x 50 ps)
13
5
2.5
c01 c02 c03 c04 c05 c06 c08 c11 c13 c14 c15 c16 c17 c19 c20 c21 c22
random patch
flexibility
(mean and SD)
and number of
random patches
random patch SD
(mean and SD)
➂ Comparing free and bound.
On average binding does not restrict overall flexibility.
average flexibility [Å]
free
2
all
bb
ncs
cs
-
YSD]
bound
all heavy atoms
backbone
non-contact surf.
contact surface
B
Ⓐ 10×50 ps MD
A
Ⓑ 10×1 ns MD
1.5
Longer more
elaborate
simulation for 7
of the complexes.
1
0.5
0
all
bb ncs cs
all
bb ncs cs
➃ Conformational entropy
➃ Conformational entropy
Binding entropy and its decomposition. All values are for bound – free state (in cal mol
conformational ∆S
c02
c06
c11
c15
c16
c17 k
c20
a receptor
lig b
rec lig c
rec × lig d
spurious e
20 ± 37
-196 ± 43
-115 ± 35
-55 ± 24
75 ± 15
-85 ± 44
-149 ± 18
-46 ± 18
31 ± 20
-123 ± 33
-103 ± 22
-144 ± 17
-151 ± 27
104 ± 16
46 ± 1
55 ± 1
55 ± 3
70 ± 2
45 ± 1
52 ± 8
60 ± 2
-11 ± 0
-8.2 ± 1
-8.9 ± 0
-6.6 ± 0
-13 ± 1
-11 ± 1
-14 ± 1
2.4 ± 3
4.3 ± 2
3.6 ± 2
3.0 ± 1
3.0 ± 2
10 ± 6
4.4 ± 7
∼
rec a
1
K
1
at 1M standard state).
total ∆S
∆Scon f f
t rg
sol vh
19 ± 44
-101 ± 34
-157 ± 41
-60 ± 25
-0.7± 13
-137 ± 49
43 ± 23
-100
-105
-101
-98
-103
-108
-109
334
135
242
207
422
587
556
total i
253 ± 44
-71 ± 34
-12 ± 41
49 ± 25
318 ± 13
342 ± 49
490 ± 23
ex p j
(-34 l )
-1m
20n
only; b ligand only; c entropy gain from rigid body motions of receptor against ligand; d entropy loss from motions correlated across the binding interface; e difference between spurious correlations in free and bound state; f total change of vibrational
entropy; g rotational and translational entropy; h solv ent entropy estimated from buried accessible surface; i total entropy change calculated (con f + t, r + sol v); j measured entropy change; k not converged; l measured for a related receptor, Bath et al. 1994, Sundberg et al.
2000; m Frisch et al. 1997; n Arold et al. 1998
➃ Conformational entropy
Binding entropy and its decomposition. All values are for bound – free state (in cal mol
conformational ∆S
c02
c06
c11
c15
c16
c17 k
c20
lig b
rec lig c
rec × lig d
spurious e
20 ± 37
-196 ± 43
-115 ± 35
-55 ± 24
75 ± 15
-85 ± 44
-149 ± 18
-46 ± 18
31 ± 20
-123 ± 33
-103 ± 22
-144 ± 17
-151 ± 27
104 ± 16
46 ± 1
55 ± 1
55 ± 3
70 ± 2
45 ± 1
52 ± 8
60 ± 2
-11 ± 0
-8.2 ± 1
-8.9 ± 0
-6.6 ± 0
-13 ± 1
-11 ± 1
-14 ± 1
2.4 ± 3
4.3 ± 2
3.6 ± 2
3.0 ± 1
3.0 ± 2
10 ± 6
4.4 ± 7
∼
rec a
1
K
1
at 1M standard state).
total ∆S
∆Scon f f
t rg
sol vh
19 ± 44
-101 ± 34
-157 ± 41
-60 ± 25
-0.7± 13
-137 ± 49
43 ± 23
-100
-105
-101
-98
-103
-108
-109
334
135
242
207
422
587
556
total i
ex p j
253 ± 44
-71 ± 34 (-34 l )
-12 ± 41
-1m
49 ± 25
20n
con f
318 ± 13
342 ± 49
490 ± 23
∆S
19 ± 44
± motions
a receptor only; b ligand only; c entropy gain from rigid body motions of receptor against ligand; d entropy-101
34 corloss from
related across the binding interface; e difference between spurious correlations in free and bound state; f total change of vibrational
± 41 calcu-157
entropy; g rotational and translational entropy; h solv ent entropy estimated from buried accessible surface; i total
entropy change
lated (con f + t, r + sol v); j measured entropy change; k not converged; l measured for a related receptor, Bath et al. 1994,±Sundberg et al.
-60 25
2000; m Frisch et al. 1997; n Arold et al. 1998
-0.7± 13
-137 ± 49
Conformational entropy can both rise or fall.
43 ± 23
➃ Conformational entropy
Binding entropy and its decomposition. All values are for bound – free state (in cal mol
1
K
conformational ∆S
c02
c06
c11
c15
c16
c17 k
c20
lig b
rec lig c
rec × lig d
spurious e
20 ± 37
-196 ± 43
-115 ± 35
-55 ± 24
75 ± 15
-85 ± 44
-149 ± 18
-46 ± 18
31 ± 20
-123 ± 33
-103 ± 22
-144 ± 17
-151 ± 27
104 ± 16
46 ± 1
55 ± 1
55 ± 3
70 ± 2
45 ± 1
52 ± 8
60 ± 2
-11 ± 0
-8.2 ± 1
-8.9 ± 0
-6.6 ± 0
-13 ± 1
-11 ± 1
-14 ± 1
2.4 ± 3
4.3 ± 2
3.6 ± 2
3.0 ± 1
3.0 ± 2
10 ± 6
4.4 ± 7
∼
rec a
1
at 1M standard state).
total ∆S
∆Scon f f
t rg
sol vh
19 ± 44
-101 ± 34
-157 ± 41
-60 ± 25
-0.7± 13
-137 ± 49
43 ± 23
-100
-105
-101
-98
-103
-108
-109
334
135
242
207
422
587
556
total i
253 ± 44
-71 ± 34
-12 ± 41
49 ± 25
318 ± 13
342 ± 49
490 ± 23
a receptor
ex p j
(-34 l )
-1m
20n
only; b ligand only; c entropy gain from rigid body motions of receptor against ligand; d entropy loss from motions correlated across the binding interface; e difference between spurious correlations in free and bound state; f total change of vibrational
entropy; g rotational and translational entropy; h solv ent entropy estimated from buried accessible surface; i total entropy change calculated (con f + t, r + sol v); j measured entropy change; k not converged; l measured for a related receptor, Bath et al. 1994, Sundberg et al.
2000; m Frisch et al. 1997; n Arold et al. 1998
In agreement with experimental data.
total
-71 ± 34
-12 ± 41
49 ± 25
ex p
(-34)
-1
20
➃ Conformational entropy
c02
+
inhibitor
α-chymotrypsinogen
➄ Recognition between structure ensembles
➄ Ensemble-Docking
ligand
receptor
Free conformations
Molecular dynamics simulation
Fuzzy clustering
11×ligand
11×receptor
Rigid-body docking
62.000 docking solutions
10 representative
conformers + free
conformation
0.8
Fraction of native atom contacts (fnac)
Glycosyltransferase
and
Tendamistat
(c19)
1.0
Bound docking
A
0.6
0.4
0.2
1.0
0.8
Free docking (traditional docking)
B
0.6
0.4
0.2
1.0
0.8
Best scoring ensemble combination
C
0.6
0.4
0.2
0.0
0
100
200
300
Docking solution
400
500
➄ Ensemble-docking
c19 PCR-MD
fnac solutions
>0
285
>0.1
>0.2
>0.3
75
>0.4
>0.5
>0.6
30
>0.7
>0.8
Ligand conformers
10
9
8
7
6
5
4
3
2
1
free
B
free
C
2
1 2 3 4 5 6 7 8 9 10
Receptor conformers
➄ Recognition of conformers
...does not depend on the bound conformation
All conformer
combinations
25
score - scorefree
20
(interface heavy
atoms only)
15
MD
10
PCR-MD
5
0
-5
0.0
0.5
1.0
1.5
rmsd to bound - rmsdfree to bound [Å]
➅ A unified model
➅ A unified model
We need a model that is compatible with that...
Binding sites are flexible.
Multiple complementary conformers in the free ensembles.
Recognition does not depend on the bound structure.
Conformation entropy can rise or fall.
➅ A unified model
We need a model that is compatible with that...
☹
Multiple complementary conformers.
☹☹☹
Recognition does not depend on the bound structure. ☹ ☹
Conformation entropy can rise or fall.
☹☹☹
Binding sites are flexible.
key-lock? conformer selection? Induced fit? ... or ...
➅ A unified model
➀ Diffusion
➁ Free conformer
selection
➂ Refolding
Rf + Lf
Free energy
Three step model:
Aligned Encounter
Complex
k1
k-1
Rf + Lf
Rf Lf
k2
k-2
Recognition
Complex
R*f L*f
k3
k-3
Native
Complex
RbLb
Rf Lf
R*f L*f
Reaction coordinate
RbLb
long range electrostatic
desolvation
rotational and translational entropy
conformational entropy
short range electrostatic and van der Waals interactions
➅ A unified model
Aligned Encounter
Complex
electrostatic steering/
partial desolvation
Free energy
Diffusion
Rf + Lf
k1
k-1
Rf + Lf
Rf Lf
k2
k-2
Recognition
Complex
R*f L*f
k3
k-3
Native
Complex
RbLb
Rf Lf
R*f L*f
Reaction coordinate
RbLb
long range electrostatic
desolvation
rotational and translational entropy
conformational entropy
short range electrostatic and van der Waals interactions
➅ A unified model
Aligned Encounter
Complex
Free energy
Conformer selection
Rf + Lf
k1
k-1
Rf + Lf
Rf Lf
k2
k-2
Recognition
Complex
R*f L*f
k-3
RbLb
Rf Lf
R*f L*f
Reaction coordinate
fuzzy conformer selection,
(local induced fit)
k3
Native
Complex
RbLb
long range electrostatic
desolvation
rotational and translational entropy
conformational entropy
short range electrostatic and van der Waals interactions
➅ A unified model
Aligned Encounter
Complex
Rf + Lf
Free energy
Refolding
k1
k-1
Rf + Lf
Rf Lf
k2
k-2
Recognition
Complex
R*f L*f
k3
k-3
Native
Complex
RbLb
Rf Lf
R*f L*f
Reaction coordinate
RbLb
long range electrostatic
desolvation
rotational and translational entropy
(induced fit)
conformational entropy
short range electrostatic and van der Waals interactions
➅ A unified model
Aligned Encounter
Complex
Sfree
Free energy
Conformational
entropy
Rf + Lf
k1
k-1
Rf + Lf
Rf Lf
k2
k-2
Recognition
Complex
R*f L*f
k3
k-3
Native
Complex
RbLb
Rf Lf
R*f L*f
Reaction coordinate
Sbound
RbLb
long range electrostatic
desolvation
rotational and translational entropy
conformational entropy
short range electrostatic and van der Waals interactions
Summary
binding sites are more flexible than the remaining surface
multiple complementary conformers exist within the free ensembles
recognition does not depend on the bound conformation
conformational entropy can both rise or fall
Binding may follow 3-step mechanism of:
➀ Diffusion
➁ Free conformer selection
➂ Refolding
Acknowledgements
Collaborators: Raik Grünberg and Michael Nilges,
Institute Pasteur
Funding: Knut and Alice Wallenberg Foundation
For more protein-protein binding and docking visit poster 98