protein dynamics and stability: universality vs....
TRANSCRIPT
Protein Dynamics and Stability:Universality vs. Specificity
Rony Granek
The Stella and Avram Goren-Goldstein Dept. of Biotechnology EngineeringBen-Gurion University of The Negev
Motivation.
Fractal nature of proteins. Fractons – the vibrational normal modes of a fractal (scalar elasticity).
Protein marginal stability – universal “equation of state”.
Dynamics: Vibrational & Random Walk MSD’s, return probabilities, etc.
Tensorial elasticity models.
Dynamic Structure factor – preliminary results.
Force induced unfolding – preliminary results.
Conclusions.
Coworkers
Aviv University-Tel
Shlomi Reuveni
Joseph Klafter
Gurion University-Ben
Marina de Leeuw
Roee Ben-Halevi
Amit Srivastava
RG
Long sequence of amino acids (20 types).
Thousands of different proteins. Differ by sequence and length. Fold in different ways to give different 3-D fold structure.
Conflicting requirements:
Specific folding – leads to a specific function (lock and key…). Large internal motion is needed to allow for biochemical function (enzymatic activity, antibody function, capturing and releasing ions, etc.).
Natural Proteins
–Problem a folded protein has less internal motion than an unfolded
.Protein
Single molecule experiments in proteins
Photo-induced electron transfer (a few angstroms)
FRET: Fluorescence resonant energy transfer (tens of angstroms).
eqXtXtx )()(
W. Min et al., PRL (2005)
FL (fluorescein) and anti-FL complexTyr (tyrosine) – donor, FL – acceptor
Autocorrelation function )0()()( xtxtCx
stt
stttC
x
1
11~)(
2/1
2/1 const.
Small scale motion – VIBRATIONS?
)The Gaussian Network Model (GNM
Scalar elasticity. Springs exist only below a cutoff distance Rc.
All springs have equal spring constant.
I. Bahar and coworkers
normal mode analysis–Vibrations
ll
otlutlumtlu
dt
dm
'
2
2
2
),(),'(),(
Equations of motion
Normal modes (eigenmodes, eigenstates)
tieltlu
)(),(
l
u
m
o
mass
Spring natural frequency
displacement
“name” of anAlpha-carbon
ll
olll
'
22
)()'()(
Density of states (DOS, density of normal modes):
0
)()( dgG
)(g
Cumulated DOS:
Fractals Definition –
“a rough or fragmented geometric shape
that can be split into parts, each of which is (at least approximately) a reduced-size copy of the whole”
Benoît Mandelbrot (1975)
Fractal nature of proteins
Mass fractal dimension :
fd
rM ~
fd
The atoms enclosed in spheres of different radii (pdb: 1OCP )
190N
505N
1184N
D. M. Leitnerand coworkers
r
Fractons
ll
otlutlumtlu
dt
dm
'
2
2
2
),(),'(),(
Vibrations of the fractal
Normal modes (eigenmodes, eigenstates) – Fractons:
tieltlu
)(),(
l
u
m
o
mass
Spring natural frequency
displacement
“name” of a pointmass
Strongly localized eigenstates
Density of states
)( l
S. Alexander and R. Orbach (1982)
Yakubo and Nakayama (1989)
1~)(
s
dg
sd – Spectral dimension
Landau-Peierls Instability
u
– Amplitude of a normal mode )( l
Protein Stability & Unfolding
2
2 3
m
Tku
B
T
Thermal fluctuations of the displacements ( )
)1/2()2(
min
222~~)(
min
ss
o
dd
TTT
Nugduu
ssfddd
gNR
/1/
min~~
N – # of amino acids (“polymer index”)
If , increases with increasing !
Large fluctuations may assist enzymatic/biological activity.
NT
u2
2s
d
2s
d
Equipartition
But should not exceed the mean inter-amino acid distance,
otherwise protein must unfold (or not fold).
2/12
u
Marginal stability. To have large amplitude motion but remain folded:
Proteins can “live” in the “twilight” zone: Folded-Unfolded !
To keep proteins folded, should depend on :
should approach the value of 2 for large proteins.
sd N
sd
Instability threshold: Universal relation between exponents
)1/2(
2
2~
s
d
o
B
T
Nm
Tku
Landau-Peierls Instability:
for unfoldinglike criterion-Lindenman22
~c
surface
Ru
)1/2(222~p)-(1
s
d
bulksurface
Nuupu
Assume
2)1/2(
2~
c
d
o
BRpN
m
Tks
Unfolding:
Take:)/1(
~1
~~f
d
g
NRV
Sp
N
b
ddfs
ln1
12
Tk
Rmb
B
co
22
ln
Unfolding/Melting occurs from the surface inward
Cluster melting analog
Reuveni et al., PRL (2007)
N
ba
ddfs
ln
12
Best fit: CC=0.55, 4,1 53.4,90.0 thethe
baba
Fitting the data of 543 proteins to:
Nln1
fsdd
12
GNM cutoff length Rc=6AReuveni et al., PRL (2007)
Selecting Proteins
Structures in the PDB : 48,638
No ligands
Structures to analyze : 4249
No more than 95% ID
No RNA, no DNA
No Peptides (less than 100 amino acids)
Files with missing data - removed
High accuracy in determining ds and df ,
99.02R
Larger set:
GNM cutoff length Rc=6 A
4249 proteinsColored histogram (100X100 bins)
)ln(
12
N
b
a
fd
sd
)ln(
1
12
N
b
fd
sd
fit to
fit toa=0.95
cc=0.58
M. de Leeuw et al., PLOS-ONE, 2009
4249 proteinsX-axis separated into 100 bins
)ln(
12
N
b
a
fd
sd
)ln(
1
12
N
b
fd
sd
fit to
fit toa=1.065
cc=0.945
M. de Leeuw et al., PLOS-ONE, 2009
Dynamics
I) Time-autocorrelation function of the distance between two alpha-carbons
Motivation: Single molecule experiments in proteins (Xie and coworkers)
eqXtXtx )()(
Autocorrelation function )0()()( xtxtCx
autocorrelation function-Displacement difference time
eqXtXtx
)()(
Two point masses (alpha-carbons), and
Positions in space and
Separation vector
Equilibrium spacing
Displacement difference vector
l
'l
),( tlR
),'( tlR
),'(),()( tlRtlRtX
eqX
),'(),()( tlutlutx
Expansion in normal modes )()(),( ltutlu
+disorder averaging
)0()(|)'(|1
2)0()( utull
Nxtx
Granek & Klafter, PRL (2005)
)0Static fluctuations (t=
)12(
)12(
2
2~
sf
sf
dd
dd
o
Br
b
r
m
Tkx
rwith the real space separation between the two points Diverging
Longer separation distance Larger fluctuations
fractonsoverdampedStrongly
t
T
uutu
2
e)0()(2
Where is the local friction (Rouse-like model).
Propagation length is
m
rξ(t)t
rttxtx
sls
s
ddd
d
: timeslong
)(:sshort timeconst.1~)0()(
)12(
2/1
fd
sd
tt2
~)(
fldd
rlM ~~r - Real space separation between the two points
l - Topological space separation between the two points
ld - Topological space dimension
Granek & Klafter, PRL (2005)
Hydrodynamic interaction (Zimm)
Oseen hydrodynamic interaction tensor
),(),'(),''()',(),(
''''
2tltlutlullOmtlu
dt
d
lll
o
rllO
1~)',(
noise
propagation length
fssf
s
dddd
d
tt2
~)(
timeslong
sshort timeconst.1~)0()(
2
22
2
2
fss
sls
fss
s
ddd
ddd
ddd
d
t
txtx
Granek, Phys. Rev. E, Rapid comm. (2011)
II) Vibrational mean square displacement (MSD)
TT
ututu22
)0()()(
TT
tuN
tu )(1
)(22
A specific alpha-carbon
Average over all alpha-carbons
In the fractal model:
2/12
~)( sd
T
ttu
Dynamics
III) Random Walk on the GNM network
Probability of return to the origin2/
0~)( s
dttP
Mean square displacement (MSD) sfw
ddddttr w 2 ~)(
/22
Mean first passage time (MFPT) fw
dd
r
~MFPT
On Fractals
)(0
tP
)(2
tr
Note equivalence to vibrations: )(2
tr
MFPT
fsdd
tt/2
~)(
)12(2~
sf
dd
rx
)(0
tPT
tudt
d)(
2
Probability of return to the origin2/
0~)( s
dttP
Reuveni et al., PNAS (2010)
1 4 7N
4 9 2N
1 7 6 7N
1 .6 4R W
sd
1 .7 4R W
sd
2R W
sd
0 .0 0 8 0 .9 5D O S R W
s sd d
. 0 .8 6c c
2197N
0.3RW
sd
RW Mean square displacement (MSD) w
dttr
/22~)(
Reuveni et al., PNAS (2010)
1 4 7N
4 9 2N
1 7 6 7N
2 .4 1w
d
2 .4 2w
d
2 .4 0w
d
2 .1 1w
d 2197N
Propagation length (vibrations) w
dtt
/22~)(
Different proteins
Mean first passage timefw
ddr
~MFPT
Reuveni et al., PNAS (2010)
1 4 7N
4 9 2N
1 7 6 7N
0 .7 5
0 .5 1
0 .3 0 5
r
N
MFPTx ~~
2
fwdd
rx
~2
Variance of distance fluctuations
Reuveni et al., PNAS (2010)
?What about specificity
214 Adenylate kinase, N=
but205)A 29( , (a.u.) rMFPT
337)127,53( T
Structure with bound ATP and AMP (substrate)
Dynamic Structure Factor–Sierpinski gasket
)(~
kS
),( tkS
- “Forzen network” static structure factor
- Dynamic structure factor
No translational and rotational diffusion
ps 53.100Reuveni, Klafter, Granek, to be published
Stretched exponential!
Dynamic Structure Factor–Sierpinski gasket
)()0,( kSkS
),( tkS
- Static structure factor
- Dynamic structure factor
No translational and rotational diffusion
ps 53.100Reuveni, Klafter, Granek, to be published
Dynamic Structure Factor–Sierpinski gasket
)()0,( kSkS
),( tkS
- Static structure factor
- Dynamic structure factor
No translational and rotational diffusion
ps 53.100Reuveni, Klafter, Granek, to be published
Dynamic Structure Factor–Proteins
)()0,( kSkS
),( tkS
- Static structure factor
- Dynamic structure factor
No translational and rotational diffusion
Reuveni, Klafter, Granek, to be published
o
A 5 , 5 bkb
ps 22
ps 32
ps 170
N
kji
ikjjkik
N
ji
ijjiijoTEMBruumH
,,
2
,
22
2
1ˆ
2
1
Tensorial Elasticity Model
ikj
k
i
j
Advantage:Allows to compute anisotropic displacement fluctuations.Disadvantage:Complicated.Involves an additional free parameters (B).
Ben-Halevi et al., to be published
Bond-bendingAnisotropic network model (ANM)
Tensorial Model for Methyltransferase (2NQ5).
-22A 1/ Bm
o
Slope=1.508
0
1
2
3
4
5
6
7
8
9
-4.00E+00 -3.00E+00 -2.00E+00 -1.00E+00 0.00E+00 1.00E+00 2.00E+00
ln
)(ln G
Ben-Halevi et al., to be published
Slope=1.63
Unfolding by Pulling Force
Cymotrypsin Inhibitor (CI2)
GNM of CI2
Srivastava & Granek, to be published
force at the end pair
Conclusions:
Novel approach for vibrations in folded proteinsbased on their fractal nature Provides a description
on a universal level.
Folded proteins are marginally stable: they exist in a thermodynamic state close to unfolding, which allows for large scale motion without unfolding.
The above criterion leads to a universal “equation of state”, verified for about 5,000 proteins.
The fractal-like properties of proteins lead toAnomalous dynamics/ strange kinetics:autocorrelation of separation; vibrational MSD; random walk MSD, return probability & mean first passage time, dynamic structure factor.
Questions to address:
Relation to enzymatic/biological activity.
Role of water.
More realistic potentials, e.g.: Tensorial elasticity, distinguishing between backbone bonds, H-bonds, electrostatic bonds, etc.
Response and unfolding under force.