Copper and Zinc Coordination
mode in b-Amyloid Peptides
A XAS and ab initio study
V. Minicozzi
Phys. Dept.- University of Rome
Tor Vergata
ABR2008, 10-11 Aprile 2008, Roma
Biophysics Group in Rome
“Tor Vergata”
Experiments
CNR- Trento
M. Dalla Serra
C. Potrich
S. Morante
G. C. Rossi
V. Minicozzi
F. Stellato
S. Alleva
A. Maiorana
Calculations
DESY -Berlin
K. Jansen
N. Christian
EMBL-DESY-Hamburg
Computations have been performed at
W. Meyer-Glauche
Fermi
BEN
Altix
Summary
• Ab peptide
• XAS results
• ab-initio simulations
• Conclusions and outlook
Alzheimer Disease (AD)
AD brains show two lesions
Neurofibrillar
Tangles
1- Amyloid Plaques:
Extracellular
peptide
deposits
of
Ab
almost spherical with a 10-100
mm diameter
2- Neurofibrillar Tangles:
Intracellular abnormal elicoidal
fibers mainly composed by tau
protein
These two lesions can occur
independently of each other
Neuron
Amyloid
Plaques
Amyloid b-peptide
•- & -secretases
cleavage  nonpathological peptide P3
• Ab is derived from
proteolitic cleavage of APP
protein (Amyloid Precursor
Protein).
•APP: 770 trans-membrane
protein coded in
chromosome 21
•b- & -secretases
cleavage  pathological
peptides Ab1-40, Ab1-42
APP
-secretase
P3
17
40-42
-secretase
APP
b-secretase
Ab
1
40-42
Cu2+
Cu2+
EPR
NMR
J. Danielsson et al. (2007) FEBS 274:46
A.K. Tickler et al. (2005) JBC 280:13355
Zn2+
Zn2+
NMR
NMR
S. Zirah et al. (2006) JBC 281:2151
Syme & Viles (2006) BBA 1764:246
Experiments on…
DAEFRHDSGY EVHHQKLVFF AEDVGSNKGA IIGLMVGGVV
Cu2+/Zn2+-
Ab1-16
minimal fragment containing His6, His13, and
His14, suggested to be involved in metal
binding
Cu2+/Zn2+- Ab17-40
complementary sequence where none of these
His’s is present
Cu2+/Zn2+-
besides the presence of these three His’s, a
long hydrophobic region believed to be
relevant in the aggregation process
Ab1-28
Cu2+/Zn2+- Ab5-23
the N-terminal region of the Ab-peptide can
play any role in the metal binding process?
• Stellato et al., Eur Biophys J (2006) 35: 340
• Minicozzi et al., (2008) J Biol Chem in press
EXAFS
Cu2+/Zn2+-Ab17-40 = Cu2+/Zn2+- buffer
Cu2+-Ab1-16 = Cu2+- Ab1-28 = Cu2+-Ab1-40
Cu2+-Ab5-23 ≠ Cu2+-Ab1-16
Zn2+-Ab1-16 = Zn2+-Ab1-28 = Zn2+-Ab5-23 = Zn2+-Ab1-40
Cu-Ab1-16
EXAFS
|FT|
Data
Fit
0
1
2
3
•3 Histidines
•1 Tyrosine
•1 O
4
r(Å) 5
Cu-Ab5-23
EXAFS
|FT|
Data
Fit
0
1
•2
•1
•1
•1
2
Histidines
N-term
Tyrosine
O
3
4
r(Å) 5
Zn-Ab1-16
EXAFS
|FT|
Data
Fit
0
1
2
•4 Histidines
•1 O
3
4
r(Å) 5
XAS Conclusions
• Metal binding site lies within the first 16 aminoacids
• Cu2+ and Zn2+ have different binding geometry
• Zn2+-Ab less rigid geometry, sensitive to solution condition
• Cu2+-Ab very stable binding mode
• Zn2+-Ab inter-molecular binding suggests aggregation
• Cu2+-Ab intra-molecular binding
Questions:
1. precise location of metal binding site along the sequence
2. different zinc and copper role in aggregation processes
A promising tool is ab initio molecular dynamics
Car-Parrinello Molecular Dynamics simulations (CP-MD)
Classical MD atoms move in the chosen force field
ab-initio MD electrons are active quantum mechanical DOF
CP method is based on DFT theory
ℒ
     i r  
2
i
1
M I R I2  E  i , RI     ij [  i  j   ij ]

2 I
ij
Enforcing the orthonormality of KS wave functions
Nuclei move experiencing both the force due to electrons, EDFT , and the force due
to electrostatic nuclear interaction, EN
Fictitious dynamics for electrons
i  
⇨
electronic degrees of freedom  i
E
   ij j
 i
j
parallel version of Quantum-ESPRESSO package ⇨ CP-MD
• Vanderbilt ultrasoft pseudopotentials
• Perdew-Burke-Ernzerhof (PBE) exchange-correlation (xc) functional
Ab-peptide systems
Classical MD simulations of Cu+2-Ab1-16 in water:
- Cu+2 bounded to His6, Tyr10, His13, His14
- Cu+2 bounded to Nterm, His6, His13, His14
S2  Cu+2(D1-2-3-4-5-H6-7-8-9-Y10-E11-12-H13-H14-cap) + 180 H2O
Classical MD simulations of Zn+2-Ab1-16 in water:
comes from classical MD of Cu+2-Ab1-16 with Cu+2 bound to His6, Tyr10, His13, His14
- Zn+2 bounded to His6, Tyr10, His13, His14
- Zn+2 bounded to 4 Histidines  2 Ab1-16
CP-MD simulations of Cu complexes
S3  Cu+2(D1-2-3-4-5-H6-7-8-9-Y10-E11-12-H13-H14-cap) + 158 H2O
+2(D -2-E -4-5-H -cap cap-H -H -cap) + 125 H O
S1comes
 Cu
1
3 MD of Cu
6 +2-Ab1-16 with13
2 6, His13, His14
from
classical
Cu+2 14
bound to Nterm, His
S2  Cu+2(D1-2-3-4-5-H6-7-8-9-Y10-E11-12-H13-H14-cap) + 180 H2O
S3  Cu+2(D1-2-3-4-5-H6-7-8-9-Y10-E11-12-H13-H14-cap) + 158 H2O
S1  494 atoms and 1369 electrons
S2  703 atoms and 1951 electrons
S3  628 atoms and 1776 electrons
S1: Distances from Cu2+
S2: Distances from Cu2+
Conclusions and Outlook
• We can discriminate via ab-initio simulations among the
structural models extracted from XAS experiments
• QM/MM simulations of the whole hydrated Ab peptide
based on structural XAS data expecially relevant for Zn+2
structures where pairs of peptides are involved
• New XAS experiments with Aluminium and Zinc
Thanks for your
attention!
XANES
Cu2+-Aβ1-16 = Cu2+-Aβ1-28 = Cu2+-Aβ1-40 ≠ Cu2+-Aβ5-23 ≠ Cu2+-Aβ17-40
Zn2+-Aβ1-16 = Zn2+-Aβ1-28 ≠ Zn2+-Aβ1-40 ≠ Zn2+-Aβ5-23 ≠ Zn2+-Aβ17-40
Feasibility studies for S1 system
Platform
CPU hours for 104 steps
Fermi1 in a 16 node configuration
1650
BEN in a 16 node configuration
1300 on S1
Altix scaling
Altix in a 16 node configuration
300
psbased
at 300
K GHz
- Fermi1 Linux-clusters (E. Fermi Institute, Rome -3.6
Italy)
on 1.7
Pentium IV processors
64 nodes ~ 1 month
- BEN Linux-cluster (ECT∗ Institute, Trento - Italy) based on Intel/XeonV.Minicozzi, et al. (2008) International Journal of Quantum
2.8 GHz processors
Chemistry in press
- ALTIX 4700 (LRZ, Munich - Germany) based on Intel Itanium2 Madison
9M 1.6 GHz processors
S1: Dihedral angle Nδ(H6)–Cu–Ne(H13)–Nδ(H14) as
a function of the CP simulation time