Mechanism of copper(II)-induced

misfolding of Parkinson’s disease protein

SUBJECT AREAS:
Frisco Rose, Miroslav Hodak & Jerzy Bernholc
COMPUTATIONAL
BIOLOGY
NEURODEGENERATION Center for High Performance Simulation and Department of Physics, North Carolina State University, Raleigh, North Carolina
27695-7518, USA.
PROTEINS
DISEASES
a-synuclein (aS) is a natively unfolded pre-synaptic protein found in all Parkinson’s disease patients as the
major component of fibrillar plaques. Metal ions, and especially Cu(II), have been demonstrated to
Received accelerate aggregation of aS into fibrillar plaques, the precursors to Lewy bodies. In this work, copper
11 February 2011 binding to aS is investigated by a combination of quantum and molecular mechanics simulations. Starting
from the experimentally observed attachment site, several optimized structures of Cu-binding geometries
Accepted are examined. The most energetically favorable attachment results in significant allosteric changes, making
06 May 2011 aS more susceptible to misfolding. Indeed, an inverse kinematics investigation of the configuration space
uncovers a dynamically stable b-sheet conformation of Cu-aS that serves as a nucleation point for a second
Published
b-strand. Based on these findings, we propose an atomistic mechanism of copper-induced misfolding of aS
14 June 2011 as an initial event in the formation of Lewy bodies and thus in PD pathogenesis.

P
Correspondence and arkinson’s disease (PD) is the most prevalent neurodegenerative movement disorder affecting more than
requests for materials
0.1% of the population older than 40 years of age1. At the cellular level, PD is characterized by the accretion
of Lewy body plaques in neurons of the substantia nigra2. Lewy bodies are accumulations of amyloid-like
should be addressed to
fibrils, which are comprised primarily of an oligomerized form of a-synuclein (aS)3, a natively unfolded protein
J.B. (bernholc@ncsu. consisting of 140 amino acids4. In addition to being implicated in PD, aS also plays a key role in other neuro-
edu) degenerative disorders, such as dementia associated with Lewy body disease, diffuse Lewy body disease and
multiple system atrophy4. Furthermore, Lewy bodies are found in a significant fraction of Alzheimer’s disease
(AD) patients5, suggesting that aS is also important in the pathogenesis of AD.
An increasing amount of evidence now points to the significance of metal ions in PD. Elevated levels of iron and
zinc were found in PD brains6 and high levels of copper were detected in the cerebrospinal fluid7. Furthermore, aS
binds metal ions which accelerate its oligomerization into fibrils6. The link between metals and PD is similar to
that in AD, where altered levels of metals in the brain are also observed8. For the case of AD, Bush and Tanzi9 have
postulated the metal hypothesis, which argues that neurotoxic effects of proteins are promoted and/or caused by
their interactions with metal ions. Based on this hypothesis, metal chelation therapies for AD were developed and
are presently being tested in clinical trials10. Similarities between the roles of metals in AD and PD suggest that
metals represent a promising therapeutic target for PD.
Copper is one of the most studied metals in neurodegenerative diseases. It appears to have distinctly different
biological roles in different diseases: while it is harmful and causes structural damage to proteins in AD and PD11,
it is thought to have neuroprotective effects in prion diseases 12,14,15. In fact, our previous investigation12 of copper
attachment to prion protein (PrP) showed that copper binding is a part of PrP’s normal function and that this
protein performs a buffering role, thereby protecting proteins such as aS from damaging effects of free copper
ions. However, it appears to have detrimental effects in PD, since it was found to be the most effective metal in
causing the in vitro oligomerization of aS16.
In this report, we use a combination of quantum and classical simulations to study Cu(II) attachment to
aS. Quantum simulations are employed to determine the local binding geometry of the Cu-aS complex, while
classical molecular dynamics is applied to probe the structure of the full-length copper-bound aS. These calcula-
tions reveal that copper attachment to aS causes development of secondary structure in the otherwise unstruc-
tured protein, which accelerates the misfolding of aS. Based on this finding we propose an atomistic mechanism of
copper-induced misfolding of aS.

SCIENTIFIC REPORTS | 1 : 11 | DOI: 10.1038/srep00011 1

 www.nature.com/scientificreports

Table 1 | Relative energies of copper binding modes in aS

Structure DE (eV) DE (kJ/mol)

3N1O 0.00 0.00

NNOO 11.77 170.78
NONO 12.51 242.18

functional theory (KS/OF DFT) method20 and the relaxed geometries

are shown in Figs. 1 and 2. For all examined structures, the binding
is mostly planar, with the imidazole ring slightly tilted from the
coordination plane. Note that for all the binding modes, the
attachment of copper causes the protein backbone to deflect at
the binding site.
Since the 3N1O and 2N2O binding modes differ in the total
Figure 1 | Relaxed 3N1O binding geometry of copper to a-synuclein. charges (0 and 11, respectively), we include a hydronium ion in
The bond distances are: Cu-Nd 1.95 Å, Cu-N1 2.01 Å, Cu-N2 1.93 Å, and the solvation shell of the 3N1O model to facilitate a direct compar-
Cu-O 2.08 Å. ison of energies. The relative energies with respect to the 3N1O mode
are listed in Table 1, which shows that the 3N1O type of binding is
Results energetically favorable over the best of the 2N2O structures by
We consider the attachment of Cu at the experimentally observed 1.77 eV.
site His-50, identified by Rasia et al.17 using a combination of electron The properties of full-length aS are studied using classical mole-
paramagnetic resonance (EPR) and nuclear magnetic resonance cular dynamics (MD) via NAMD26. The model of free aS, whose
(NMR) techniques. Their data showed that the neighboring residues structure has not been determined yet due to its unfolded character27,
(49–52) are also involved in copper binding. A contribution of the is built and equilibrated as described in the Methods Section.
N-terminal residues (3–9) was also detected, but a more recent Extended molecular dynamics simulations show that the free aS is
study18 found that the N-terminal residues form an independent mostly linear and extended. No propensity for forming secondary
binding site not related to the primary binding at His-50. While structure is observed with the exception of several intermittent turns.
the NMR data indicate good correlation with Cu(II) attachment to A snapshot of the equilibrated free aS is shown in Fig. 3 (top). Note
two nitrogen and two oxygen atoms (2N2O), other binding types are that the imidazole group in His-50, which anchors the copper ion in
not precluded. In particular, a 3N1O type of attachment is plausible, copper-bound aS, is extended away from the backbone and can
due to the similarity of its nuclear spin interaction coefficients to readily bind metal ions.
those of 2N2O19. Therefore, we consider both 2N2O and 3N1O types The structural model of the full copper-bound aS is based on
of attachment in this work. the equilibrated free aS, onto which the quantum-mechanically-
To investigate the properties of the copper-aS complex, we con- calculated 3N1O copper binding geometry is imposed. Within
struct multiple binding models of 2N2O and 3N1O attachments. The classical molecular dynamics we represent the copper binding by a
aS is represented by residues 49–54 (VHGVAT) and the copper ion is set of harmonic springs. A detailed description of this model is given
assumed to be coordinated by Nd in the histidine imidazole and by a in the Methods section. The initial structure of copperbound aS is
combination of deprotonated amide nitrogens and carboxyl oxygens shown in Fig. 3 (bottom). Note that copper binding causes curving of
in the protein backbone. Several combinations exist for each binding the aS backbone at the binding site, creating a turn-like structure that
mode, but many can be ruled out because they distort the fragment brings fore and aft portions of the backbone into proximity. These
too severely. Using insight from our previous investigations of Cu- regions stay proximal during the entire simulation (25 ns). This
bound proteins12,20,21 and results from Ref.13, where many possible conformation is further stabilized by contacts between side chains,
copper attachments modes were investigated, we narrow our search the most stable of which are between Glu-46 and Lys-60. A snapshot
to three plausible candidates: Two 2N2O structures, denoted as exhibiting this contact is displayed in Fig. 4 (top). The simulation also
NNOO and NONO according to the order of copper ligands in shows that the proximal parts of the protein start to align, indicating
counterclockwise direction starting from the imidazole Nd (see the development of a b-sheet around the binding site and the cre-
Fig. 2), and one 3N1O structure. In the calculations, we use neutral ation of a b-hairpin there. Unfortunately, due to the large time scales
ends, i.e., the protein backbone is continued in both directions so that involved, the formation of b-sheets in full-length proteins cannot be
it ends with neutral CH3 groups. Each binding model is geometric- studied in atomistic simulations at present. Instead, we develop mul-
ally optimized using the hybrid Kohn-Sham/orbital-free density tiple plausible candidates for a b-sheet located in the area where

Figure 2 | Relaxed 2N2O binding geometries of copper to a-synuclein: NONO (left) and NNOO (right). The bond distances in counterclockwise order
are: Cu-Nd 1.99 Å, Cu-O 2.02 Å, Cu-N 1.92 Å, and Cu-O 2.18 Å (NONO); Cu-Nd 1.98 Å, Cu-N 1.97 Å, Cu-O 1.93 Å, and Cu-O 2.08 Å (NNOO).

SCIENTIFIC REPORTS | 1 : 11 | DOI: 10.1038/srep00011 2

 www.nature.com/scientificreports

Figure 3 | Misfolding events in a-synuclein due to copper binding. Top: Snapshot of the free a-synuclein showing the location of His-50 (upper).
Bottom: Initial relaxed structure of Cu(II)-bound a-synuclein. The inset shows the details of copper binding.

alignment is observed, using inverse kinematics28 and testing their aS in Lewy bodies4 and this transformation may constitute the first
dynamic stabilities in extended MD simulations. Among all the step in the development of PD.
candidate structures, only one turns out to be stable during the
MD simulations. The other ones quickly relax to the unstructured Discussion
form. The dynamically stable b-sheet conformation involves con- Our calculations show that 3N1O type of copper attachment is pre-
tacts between residues 43–45 and 57–59, see Fig. 4 (bottom). It ferred over 2N2O by 1.77 eV. This conclusion seemingly contradicts
remains stable for the entire duration of the simulation (23 ns). In the work of Rasia et al.17, who proposed that their EPR measurements
addition to the dynamical stability of this b-sheet, an additional indicate a 2N2O copper attachment. However, their analysis of EPR
structural development at residues 40–42 is also observed during data does not include the 3N1O mode; only 4N, 2N2O and 4O were
the simulation. These residues align with the already existing b-sheet, considered. Furthermore, since the values of tensor coefficients Ajj
indicating the development of a second b-strand parallel to the ori- and gjj for 2N2O and 3N1O modes are interspersed19, distinction
ginal b-sheet. The observed onset of the formation of a b-sheet has between these modes is difficult. To facilitate comparison with the
important implications, since b-sheets are a defining characteristic of experimental results, we calculate the EPR tensor coefficients for

Figure 4 | Top: Snapshot of evolved Cu-bound a-synuclein, showing hydrogen bonds between side chains that stabilize the turn-like conformation.
Bottom: Snapshot of Cu-bound a-synuclein with dynamically stable b-sheet. For visual clarity, non-participating side chains are not shown. Insets
show close-ups at the b-sheet nucleation point.

SCIENTIFIC REPORTS | 1 : 11 | DOI: 10.1038/srep00011 3

 www.nature.com/scientificreports

both 2N2O and 3N1O attachments using ORCA22. Unfortunately, corresponding to a kinetic energy cutoff of 48 Ry. Ultrasoft pseudopotentials31,33 and
generalized gradient approximation in the PBE form34 are employed.
the ORCA EPR calculations do not have sufficient accuracy to dis-
An attempt to differentiate conformational binding modes based on EPR data was
tinguish between configurations with similar EPR tensors, because made using the DFT-based code ORCA22. Due to a high computational cost, only the
the calculated tensors typically differ by about 10% from the experi- immediate binding atoms and their nearest neighbors were included. The dangling
mental results25. Our ORCA calculations obtain similar results for bonds were passivated by hydrogen. Orbitals were computed via the unrestricted
the 2N2O and 3N1O configurations: jAjjj5473.8 MHz, gjj52.20 for Kohn-Sham (UKS) method using the B3LYP23 exchange-correlation functional with
the Ahlrichs-TZV basis set24, which produced the best results in previous investiga-
2N2O (NNOO) and jAjjj5467.7 MHz and gjj52.17 for the 3N1O tion of EPR coefficients of copper complexes25. The EPR tensor was calculated for the
structure. These values differ from the experimental data of Rasia copper and contributions from the isotropic, dipole and quadrupole hyperfine
et al.17, jAjjj 5 557.6 MHz and gjj 5 2.22, by about 15 %, which is interactions were all included.
similar to the differences between theory and experiments found in The simulations of the full-length aS protein are performed with the molecular
dynamics package NAMD26 with CHARMM2735 force-field parameters. As is well
other works25. known, copper and other transition metals are difficult to describe with force fields,
The classical molecular dynamics investigation of the full copper- due to the importance of quantum effects. In this work, the effects of copper were
bound aS indicates that the copper attachment in the 3N1O mode modeled in two ways: In the initial model, the copper ion was not present and its effect
leads to the development of b-sheets in aS by bringing the portions of is modeled by fixing its ligands, while we later developed a more sophisticated model
in which the copper was represented by a set of springs which kept the copper
the protein around the binding site into close proximity. Due to the geometry close to that of our quantum investigation. Four springs were used to
high computational cost, we did not investigate the 2N2O attach- represent copper-ligands bonds and two additional springs were used to enforce
ments. However, since the NNOO mode, which is preferable over the Nd-Cu-N2 and O-Cu-N1 angles, where the atom labels are from Fig. 1. The angular
NONO structure, has a backbone deflection very similar to that of springs are necessary to preserve the planar character of the copper attachment. A
3N1O, it will likely follow a very similar structural development path. value of 250 was used as a spring constant for all bond and angular springs. The spring
constants’ units are kcal/(mol Å2) and kcal/(mol Å) for bond and angular springs,
On the other hand, the NONO mode induces a much smaller back- respectively. This value reproduced the vibrational amplitudes of the quantum cal-
bone deflection in the full protein and thus its structural develop- culations. The springs were implemented using the tclforces framework available in
ment, if any, will likely be different. NAMD, and the calculated forces were added to those evaluated from the CHARMM
Based on findings in this work, we propose a mechanism for force field. Since the overall charge of the 3N1O copper attachment is 0, no charge was
assigned to the copper ion. The differences between the b-sheet formation processes
copper-induced misfolding of aS: In the first step, copper attachment obtained from the models were very minor.
causes deflection at the binding site and creates a stable turn-like The calculations use periodic boundary conditions and solvent is added to the
structure. This feature is crucial for further structural development, simulation cell so that the distance between any protein atom and an edge of the water
since turns play an essential role in the initial steps of protein fold- box is at least 7 Å. This results in a total atom count that varies between 275,000 and
350,000. The solvent is ionized with 0.15 mM/L Na1 and Cl2 to approximate the
ing29. Its presence creates favorable conditions for b-sheet nucleation human intraneuronal ionic concentration36 and normalize the system to zero net
as adjacent parts of the protein are brought into proximity. Hydrogen charge. The temperature is maintained at 310 K by means of Langevin dynamics. The
bond contacts between the proximal side chains are formed and Nose-Hoover Langevin Piston algorithm37 is used to keep the pressure constant at
further stabilize this conformation. Subsequently, a b-sheet develops 1 atm. The Verlet-I/r-RESPA38 method is employed with the following time steps: 4 fs
for the long-range electrostatic forces, 2 fs for short-range nonbonded forces, and 2 fs
around the binding site, serving as a template for the formation of for bonded forces. Bonds involving hydrogen are constrained using SHAKE39 to
additional b-sheets. This process continues until the protein is fully enable the use of large time steps.
misfolded. The initial structural model of aS is built according to its amino acid sequence as a
In summary, we performed computer simulations to investigate b-strand using PyMol40. The protein is first equilibrated by running molecular
interactions between a copper ion and aS at an atomistic level. Our dynamics during which the initial structure rapidly degenerates to a conformation
with almost no secondary structure, with only random coils and transient turns
work, which builds on recent experimental investigations 17, con- present. This equilibration is computationally demanding due to the unfolded nature
sidered 2N2O and 3N1O types of attachment and found that the of aS, and requires a large number of solvent molecules. During the 21 ns of simu-
latter is energetically preferred. Further investigation of the preferred lation time that could be afforded, the root mean square deviation (RMSD) of the
3N1O mode shows that the copper attachment induces deflection of protein backbone does not completely flatten out, but other quantities, such as the
kinetic and potential energies, solvent accessible surface and radius of gyration
the protein backbone at the binding site, leading to the creation of a achieve stable values, indicating that the structure of the protein is not changing
turn-like feature and backbone alignment as a precursor in the significantly and is sufficiently equilibrated for the study of Cu binding.
development of a b-sheet. This indicates that the role of copper in The exploration of possible b-sheet registrations is aided by inverse kinematics as
the fibrillation of aS is both to initiate misfolding via conformational implemented within ProteinShop28 and based on the AMBER41 force field. By using
inverse kinematics we are able to reduce the sample set to contain only energetically
modification and to stabilize the partially folded intermediate once favorable b-sheet conformations.
formed. After the formation of the initial b-sheet, the folding process
continues by aligning adjacent parts of the protein to the existing
b-sheet, leading to a full misfolding. 1. Siderowf, A. & Stern, M. Update on Parkinson disease. Ann of Intern Med 138,
651–658 (2003).
2. Lewy, F. H. Paralysis agitans. I. Pathologische anatomie. Handbuch der neurologie
Methods 3, 920–933 (1912).
The copper binding geometries are investigated at the ab initio level using a hybrid 3. Spillantini, M. G., Crowther, R. A., Jakes, R., Hasegawa, M. & Goedert, M.
DFT/DFT method20. In this approach, the chemically active parts of the system, a-Synuclein in filamentous inclusions of lewy bodies from Parkinson’s
including the first solvation shells, are treated at a Kohn-Sham DFT level, while an disease and dementia with lewy bodies. Proc Natl Acad Sci U S A 95, 6469–6473
approximate orbital-free DFT is used for the remaining solvent molecules. The (1998).
compatibility between the KS and OF DFT methods enables seamless integration 4. Spillantini, M. G. & Goedert, M. The a-Synucleinopathies: Parkinson’s disease,
between the two. In particular, the flow of solvent molecules across the KS/OF dementia with lewy bodies, and multiple system atrophy. Ann N Y Acad Sci 920,
interface is allowed and the total energy is conserved. The implementation of the 16–27 (2000).
hybrid KS/OF DFT approach is based on the RMG30–32 code, in which Kohn-Sham 5. Hamilton, R. L. Lewy bodies in Alzheimer’s disease: A neuropathological review of
equations are solved in real-space and the multigrid technique is used to accelerate 145 cases using a-Synuclein immunohistochemistry. Brain Pathol 10, 378–384
convergence of the ground state wavefunctions. This method was already successfully (2000).
used to study the binding of copper to the prion protein12,20,21. 6. Uversky, V. N., Li, J. & Fink, A. L. Metal-triggered structural transformations,
In the KS/OF DFT simulations, aS is represented as a fragment consisting of aggregation, and fibrillation of human a-Synuclein. a psossible molecular link
residues 49–54 (VHGVAT). Neutral ends are applied to the fragment, resulting in a between Parkinson’s disease and heavy metal exposure. J. Biol. Chem. 276, 44284–
total of 91 atoms. In each calculation, the protein fragment is fully solvated and 44296 (2001).
minimized using classical molecular mechanics prior to performing KS/OF DFT 7. Pall, H. S. et al. Raised cerebrospinal-fluid copper concentration in Parkinson’s
simulations. This is followed by minimization within the KS/OF DFT method, in disease. Lancet 2, 238–241 (1987). PMID: 2886715.
which the protein fragment, the copper ion and 12 water molecules closest to the 8. Lovell, M. A., Robertson, J. D., Teesdale, W. J., Campbell, J. L. & Markesbery, W. R.
binding site are treated by KS DFT, while 1788 waters are enclosed in the orbital-free Copper, iron and zinc in Alzheimer’s disease senile plaques. J Neurol Sci 158,
DFT region. A grid with spacing of 0.32 Bohr is used to represent the wavefunctions, 47–52 (1998).

SCIENTIFIC REPORTS | 1 : 11 | DOI: 10.1038/srep00011 4

 www.nature.com/scientificreports

9. Bush, A. I. & Tanzi, R. E. Therapeutics for Alzheimer’s disease based on the metal 29. Lewis, P. N., Momany, F. A. & Scheraga, H. A. Folding of polypeptide chains in
hypothesis. Neurotherapeutics 5, 421–432 (2008). proteins: A proposed mechanism for folding. Proc Natl Acad Sci U S A 68, 2293–
10. Biran, Y., Masters, C. L., Barnham, K. J., Bush, A. I. & Adlard, P. A. 2297 (1971).
Pharmacotherapeutic targets in Alzheimer’s disease. J Cell Mol Med 13, 61–86 30. Briggs, E. L., Sullivan, D. J. & Bernholc, J. Real-space multigrid-based approach to
(2009). largescale electronic structure calculations. Phys Rev B 54, 14362–14375 (1996).
11. Gaggelli, E., Kozlowski, H., Valensin, D. & Valensin, G. Copper homeostasis and 31. Hodak, M., Wang, S., Lu, W. & Bernholc, J. Implementation of ultrasoft
neurode-generative disorders (Alzheimer’s, prion, and Parkinson’s diseases and pseudopotentials in large-scale grid-based electronic structure calculations. Phys.
amyotrophic lateral sclerosis). Chem. Rev 106, 1995–2044 (2006). PMID: Rev. B 76, 085108–8 (2007).
16771441. 32. Bernholc, J., Hodak, M. & Lu, W. Recent developments and applications of the
12. Hodak, M., Chisnell, R., Lu, W. & Bernholc, J. Functional implications of real-space multigrid method. J Phys Condens Matter 20, 294205–294205 (2008).
multistage copper binding to the prion protein. Proc Natl Acad Sci U S A 106, 33. Vanderbilt, D. Soft self-consistent pseudopotentials in a generalized eigenvalue
11576–11581 (2009). formalism. Phys Rev B 41, 7892–7895 (1990).
13. Pushie, J.M. & Rauk, A. Computational studies of Cu (II)[peptide] binding motifs: 34. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation
Cu [HGGG] and Cu [HG] as models for Cu (II) binding to the prion protein made simple. Phys Rev Lett 77, 3865–3868 (1996).
octarepeat region. J. Biol. Inorg. Chem. 8, 53–65 (2003). 35. Feller, S. E. & MacKerell, A. D. An improved empirical potential energy function
14. Cox, D. L., Pan, J. & Singh, R. R. P. A mechanism for copper inhibition of for molecular simulations of phospholipids. J Phys Chem A 104, 7510–7515
infectious prion conversion. Biophys. J 91, L11–13 (2006). PMID: 16698781. (2000).
15. Bocharova, O. V., Breydo, L., Salnikov, V. V. & Baskakov, I. V. Copper(II) inhibits 36. Purves, D. G., Fitzpatrick, A. D., Katz, L. C., Mantia, A. S. L. & McNamara, J. O.
in vitro conversion of prion protein into amyloid fibrils. Biochemistry 44, 6776– Neuroscience. Sunderland Mass: Sinauer Assoc (Inc , 1997).
6787 (2005). 37. Feller, S. E., Zhang, Y., Pastor, R. W. & Brooks, B. R. Constant pressure molecular
16. Paik, S. R., Shin, H. J., Lee, J. H., Chang, C. S. & Kim, J. Copper(II)-induced self- dynamics simulation: the langevin piston method. J Chem Phys 103, 4613 (1995).
oligomerization of alpha-synuclein. Biochem J. 340, 821–828 (1999). 38. Tuckerman, M., Berne, B. J. & Martyna, G. J. Reversible multiple time scale
PMC1220316. molecular dynamics. The Journal of Chemical Physics 97, 1990 (1992).
17. Rasia, R. M. et al. Structural characterization of copper(II) binding to a-synuclein: 39. Ryckaert, J. P., Ciccotti, G. & Berendsen, H. J. C. Numerical integration of the
Insights into the bioinorganic chemistry of Parkinson’s disease. Proc Natl Acad cartesian equations of motion of a system with constraints: molecular dynamics of
Sci U S A 102, 4294–4299 (2005). n-alkanes. J. comput. Phys 23, 327–341 (1977).
18. Sung, Y., Rospigliosi, C. & Eliezer, D. NMR mapping of copper binding sites in
40. DeLano, W. The PyMOL molecular graphics system. http://www.pymol.org/
alpha-synuclein. Biochim. Biophys. Acta 1764, 5–12 (2006).
41. Weiner, S. J. et al. A new force field for molecular mechanical simulation of nucleic
19. Peisach, J. & Blumberg, W. E. Structural implications derived from the analysis of
acids and proteins. J Am Chem Soc 106, 765–784 (1984).
electron paramagnetic resonance spectra of natural and artificial copper proteins.
Arch. Biochem. Biophys 165, 691–708 (1974). PMID: 4374138.
20. Hodak, M., Lu, W. & Bernholc, J. Hybrid ab initio Kohn-Sham density functional
theory/frozen-density orbital-free density functional theory simulation method
suitable for biological systems. J. Chem. Phys. 128, 014101–9 (2008). Acknowledgments
21. Hodak, M. & Bernholc, J. Insights into prion protein function from atomistic This work was mainly supported by DOE DE-FG02-98ER45685, with the petascale code
simulations. Prion 4, 13–19 (2010). development supported by NSF OCI-0749320. The calculations were carried out at the
22. Neese, F. ORCA - an ab initio, Density Functional and Semiempirical program National Center for Computational Sciences at ORNL. Additionally, we would like to thank
package, Vers. 2.7 (2008). Robin Chisnell for his efforts in enhancing the CHARMM force modifications.
23. Becke, A.D. Density-functional thermochemistry. III. The role of exact exchange.
Chem. Phys 98, 5648–5652 (1993).
24. Schiafer, A. & Horn, H. & Ahlrichs, R. Fully optimized contracted Gaussian basis
Author contributions
FR, MH and JB wrote the main manuscript text and FR prepared figures 1–4. All authors
sets for atoms Li to Kr. J. Chem. Phys 97, 2571 (1992).
reviewed the manuscript.
25. Ames, W. M. & Larsen, S. C. Density functional theory investigation of EPR
parameters for tetragonal Cu(II) model complexes with oxygen ligands. J Phys
Chem 113, 4305–4312 (2009). Additional information
26. Phillips, J. C. et al. Scalable molecular dynamics with NAMD. J. Comput. Chem. Competing financial interests: The authors declare no competing financial interests.
26, 1781–1802 (2005). PMC2486339.
27. Weinreb, P. H., Zhen, W., Poon, A. W., Conway, K. A. & Lansbury, P. T. NACP, a License: This work is licensed under a Creative Commons
protein implicated in Alzheimer’s disease and learning, is natively unfolded. Attribution-NonCommercial-NoDerivative Works 3.0 Unported License. To view a copy
Biochemistry 35, 13709–13715 (1996). of this license, visit http://creativecommons.org/licenses/by-nc-nd/3.0/
28. Crivelli, S., Kreylos, O., Hamann, B., Max, N. & Bethel, W. ProteinShop: a tool for How to cite this article: Rose, F., Hodak, M. & Bernholc, J. Mechanism of
interactive protein manipulation and steering. J. Comput. Aided Mol. Des. 18, copper(II)-induced misfolding of Parkinson’s disease protein. Sci. Rep. 1, 11; DOI:10.1038/
271–285 (2004). srep00011 (2011).

SCIENTIFIC REPORTS | 1 : 11 | DOI: 10.1038/srep00011 5