Article Information

• PDF (903 kb)
• Full Text with Large Figures
• Supporting Material
• Cited by in Scopus (4)

PubMed

• Articles by M. Jake Pushie
• Articles by Hans J. Vogel

Related Articles

• Modeling by Assembly and Molecular Dynamics Simulations of t...

  Modeling by Assembly and Molecular Dynamics Simulations of the Low Cu Occupancy Form of the Mammalian Prion Protein Octarepeat Region: Gaining Insight into Cu-Mediated β-Cleavage Biophysical Journal, Volume 95, Issue 11, 1 December 2008, Pages 5084-5091 M. Jake Pushie and Hans J. Vogel Abstract The prion protein has garnered considerable interest because of its involvement in prion disease as well as its unresolved cellular function. The octarepeat region in the flexible N-domain is capable of binding copper through multiple coordination modes. Under conditions of low pH and low Cu concentration, the four octarepeats (ORs) cooperatively coordinate a single copper ion. Based on the average structure of the PHGG and GWGQ portions of a copper-free OR model from molecular dynamics simulations, the starting structures of the OR complex could be constructed by assembling the repeating structure of PHGG and GWGQ fragments. The resulting model contains a preformed site suitable for Cu coordination. Molecular dynamics simulations of Cu bound to the assembled OR model (Cu:OR) reveal a close association of specific Trp and Gly residues with the Cu center. This low Cu-occupancy form of prion protein is redox-active and can readily initiate cleavage of the OR region, mediated by reactive oxygen species generated by Cu. The OR region is known to be required for -cleavage, as are the Trp residues within the OR region. The -cleaved form of the prion protein accumulates in amyloid fibrils. Hence, the close approach of Trp and Gly residues to the Cu coordination site in the low Cu-occupancy form of the OR region may signal an important interaction for the initiation of prion disease. Abstract | Full Text | PDF (610 kb)

• Ablation of the metal ion-induced endocytosis of the prion p...

  Ablation of the metal ion-induced endocytosis of the prion protein by disease-associated mutation of the octarepeat region Current Biology, Volume 11, Issue 7, 3 April 2001, Pages 519-523 W.Sumudhu S. Perera and Nigel M. Hooper Summary The neurodegenerative spongiform encephalopathies, or prion diseases, are characterized by the conversion of the normal cellular form of the prion protein PrP to a pathogenic form, PrP . There are four copies of an octarepeat PHGG(G/S)WGQ that specifically bind Cu ions within the N-terminal half of PrP . This has led to proposals that prion diseases may, in part, be due to abrogation of the normal cellular role of PrP in copper homeostasis . Here, we show that murine PrP is rapidly endocytosed upon exposure of neuronal cells to physiologically relevant concentrations of Cu or Zn, but not Mn. Deletion of the four octarepeats or mutation of the histidine residues (H68/76 dyad) in the central two repeats abolished endocytosis, indicating that the internalization of PrP is governed by metal binding to the octarepeats. Furthermore, a mutant form of PrP that contains nine additional octarepeats and is associated with familial prion disease failed to undergo Cu-mediated endocytosis. For the first time, these results provide evidence that metal ions can promote the endocytosis of a mammalian prion protein in neuronal cells and that neurodegeneration associated with some prion diseases may arise from the ablation of this function due to mutation of the octarepeat region. Summary | Full Text | PDF (240 kb)

• A Mechanism for Copper Inhibition of Infectious Prion Conver...

  A Mechanism for Copper Inhibition of Infectious Prion Conversion Biophysical Journal, Volume 91, Issue 2, 15 July 2006, Pages L11-L13 Daniel L. Cox, Jianping Pan and Rajiv R.P. Singh Abstract We employ ab initio electronic structure calculations to obtain two structural models for copper bound in the strongest binding site of the noninfectious form of the prion protein. The models are compatible with available experimental constraints from electron spin resonance data. The bending of the peptide backbone attendant with the copper binding is not compatible with the requisite straight -strand backbone structure for the same sequence contained in two recently proposed models of the prion protein structure in its infectious form. We hypothesize that copper binding at this site is protective against conversion to the infectious form, discuss experimental data that appear to support and conflict with our hypothesis, and propose tests using recombinant prion protein, genetically modified cultured neurons, and transgenic mice. Abstract | Full Text | PDF (96 kb)

• …more

Copyright © 2007 The Biophysical Society. All rights reserved.
Biophysical Journal, Volume 93, Issue 11, 3762-3774, 1 December 2007

doi:10.1529/biophysj.107.109512

Biophysical Theory and Modeling

Molecular Dynamics Simulations of Two Tandem Octarepeats from the Mammalian Prion Protein: Fully Cu²⁺-bound and Metal-Free Forms

M. Jake Pushie and Hans J. Vogel^,  

Structural Biology Research Group, Department of Biological Sciences, University of Calgary, Calgary, Alberta, Canada

Address reprint requests to Hans J. Vogel, Dept. of Biological Sciences, University of Calgary, 2500 University Dr. NW, Calgary, Alberta, T2N 1N4, Canada. Tel.: 403-220-6006; Fax: 403-289-9311.

The prion protein (PrP) is a C-terminal glycosylphosphatidylinosotol-anchored cell-surface protein that is implicated as the causative agent in transmissible neurological conditions such as Creutzfeldt-Jakob disease in humans, chronic wasting disease in deer and elk, and bovine spongiform encephalopathy, more commonly known as mad cow disease 1. The protein contains 208 amino acid residues and a single disulfide bond, and has two oligosaccharides attached 1. Although residues 120–208 of the C-terminal domain are structured, the N-terminal domain has been shown to be flexible and disordered in aqueous solution 2,3,4. The infectious form of PrP arises, via an as-yet enigmatic conformational misfolding process, from its native form, which is relatively high in α-helical character 5, to a conformation with increased β-sheet content. These misfolded molecules can aggregate, resulting in fibril formation, which in turn influences other PrP molecules to misfold and aggregate. The PrP aggregates are known to be protease-resistant 6, but the N-terminal region is not, and it is usually absent in the mature PrP aggregates.

The biological function of PrP is not yet fully understood and although it is expressed in most tissues, the highest concentration is found at the surface of the synapse. In humans, the flexible N-terminal domain contains four tandem copper-binding repeats, each containing eight amino acids, with the sequence (PHGGGWGQ)₄, termed the octarepeat (OR) region, shown in Figure 1A. Although the number of repeats varies among mammals, the sequence of an individual OR is highly conserved across virtually all mammalian species 7, and each domain has a high specificity for binding Cu²⁺ over other divalent metal ions 8,9,10,11.

Display large version of this figure

Figure 1

(A) The octarepeat region of the prion protein, spanning residues 60–91 (human numbering). Underlined residues are required for reproducing the spectral fingerprint of copper binding to the full-length prion protein; bold-face residues constitute the core copper-binding environment. (B) Metal-free OR₂ model. (C) Metal-bound OR₂ model with new copper-containing residue type shown in square brackets.

Copper is used by neurons for signaling, and the cellular form of PrP appears to be finely tuned to bind Cu²⁺ within the concentration range found at the synaptic junction 12. There have been strong suggestions that the native form of PrP may function to buffer the copper concentration at the synapse 13. Within biological systems, the availability and transfer of copper is tightly regulated, where the concentration of free cytoplasmic Cu²⁺ is on the order of less than one atom per cell 14. Organisms have evolved tight control mechanisms over copper transport and utilization, because free copper readily initiates reactions that in aqueous environments generate reactive oxygen species (ROS), such as highly reactive hydroxyl radicals 15. The mechanisms by which ROS may be generated by Cu²⁺ are the Haber-Weiss 16 and Fenton-type (reviewed in 17) mechanisms, which have been extensively characterized. Because of its environmental specificity for Cu²⁺, it has been proposed that PrP is a Cu²⁺ transporter or chaperone 18 and may also have a role in protecting the synapse from the deleterious effects of ROS generation by excesses of copper released from neurons.

Copper binding to the OR region of PrP has been demonstrated to confer proteinase-K resistance to this region of PrP 19. The connection between the involvement of copper and prion disease is not clear-cut, as mice treated with a Cu²⁺-specific chelator (D-penicillamine) demonstrated a delayed onset of disease 20. Although the N-terminal portion of PrP has been shown to be proteolytically removed in the protease-resistant form 21, there is evidence that individuals with additional OR sites, arising from spontaneous gene mutation, for example, have an increased susceptibility to prion diseases 22,23,24,25. The prevalence of OR insertion mutations in disease cases implies that such alterations are a risk factor for developing inherited prion disease (i.e., without extraneous influence from infectious PrP particles). Modification of the number of octarepeats, and its apparent predisposing of individuals to prion disease, is not the rule. Transgenic mice expressing an OR-deletion form of PrP display slowed disease progression and altered disease pathology 26.

In addition to the process of protein misfolding, disruption of metal ion homeostasis is also a hallmark of prion diseases 27, as is PrP oxidation 28. These features are not unique to prion disease, and are also observed in Alzheimer's disease, a neurodegenerative condition involving aggregation of proteinaceous material that is rich in the metal ions copper, iron, and zinc 29. Like Cu²⁺ bound to PrP (at low Cu/PrP ratios), copper bound in Alzheimer plaques is also redox active, and oxidative damage, as well as copper-mediated ROS generation, are hallmarks of this disease 30.

Taken together, the above observations suggest that even though the ORs are not required structurally for propagating PrP misfolding, they may play a role in the development of prion disease.

The core copper-binding region of the fully Cu²⁺-loaded PrP ORs consists of only the HGG amino acid portion. The copper is anchored by the first imidazole-ring nitrogen, N_δ1, of the His residue, and through the deprotonated amide backbone N-atoms of the two Gly residues and the carbonyl oxygen of the second Gly. The following two residues in the OR region, the third Gly and the Trp residue, facilitate in stabilizing the structure of the binding site 31. X-ray crystallographic data for a copper-bound form of the octarepeat fragment HGGGW has been reported 32.

X-ray crystallography and nuclear magnetic resonance (NMR) are the most commonly employed techniques to obtain protein structure information of atomic resolution. Due to the lack of persistent long-range structure, and the flexible nature of the N-terminal domain of PrP, this portion of the intact protein has not been amenable to crystallographic analysis, and it is often removed to aid in crystallization of PrP 33. On the other hand, Cu²⁺ contains one unpaired electron, which increases the relaxation rate of any nearby nuclei, effectively making much of the local chemical information invisible to conventional NMR structure methods.

A significant volume of work has been done to characterize the structure of the OR region in the absence of Cu²⁺. Although the flexibility of the N-terminal repeat region of mammalian PrP has been confirmed by structure methods such as circular dichroism (CD) 34,35,36 and NMR 37,38,39,40, these same experiments show evidence of transient secondary structure within the OR region. For example, NMR experiments by Yoshida et al. 39 and Zahn 40 suggested that the GWGQ region can transiently adopt a β-turn, whereas the HGGGW region adopts a loop structure with the His imidazole and Trp indole rings in close proximity. Based on fluorescence quenching measurements, Gustiananda et al. suggested that either a protonated or neutral imidazole ring may be able to participate in stabilizing a loosely folded structure through cation-π or π-π interactions with the indole ring of Trp 41.

Due to the experimental constraints in characterizing structural elements in the N-terminal domain in the presence of Cu²⁺, mentioned above, our goal has been to design a new copper-containing residue topology using high-level quantum-chemical methods. In this work, we have derived the necessary force field parameters for the unusual backbone-deprotonated and metal-coordinated HGG fragment with Cu²⁺, using models of the copper-binding region of the ORs based on computational studies 42, available crystallographic data 32, and donor ligands observed in electron paramagnetic resonance (EPR) results of the residues making up the binding site 43. The newly designed copper-containing model has been incorporated into a minimal model peptide fragment spanning two OR sites to analyze their structure propensity in molecular dynamics simulations. The same peptide fragment in the absence of Cu²⁺ has also been simulated for comparison. The models used herein are models of a two-OR fragment of the mammalian prion protein in the absence of Cu²⁺ and under full Cu²⁺ occupancy conditions, and they represent the minimum construct necessary to observe potential interactions between adjacent N-terminal copper-bound OR sites.

Unrestricted density functional theory (DFT) calculations were carried out at the B3LYP level of theory 44 using the Gaussian 03 (G03) suite of programs 45. Geometry optimizations were carried out without geometry or symmetry constraints at the B3LYP/6-31G(d) level, followed by frequency calculations (keyword freq). The larger 6-311+G(2df,2p) basis set was used to obtain more accurate relative energy differences for the B3LYP/6-31G(d)-optimized structures. Solvation effects were approximated for the optimized structures using a polarizable continuum model available in G03 (SCRF=IEFPCM) using a united-atom approach with aliphatic H-atoms summed into their parent heavy-atom radii 46,47.

Geometry optimizations were used to determine the lowest energy conformations of the copper-bound peptide fragment: [Cu^II(Ac-HGGG-NH₂)_−2H]⁰, as well as several smaller hydrated Cu²⁺ complexes. In the case of the polypeptide models, the capping acetyl (Ac) and NH₂ groups mimic the peptide bonding found in a larger protein environment and abrogate the undesirable effects of charged termini in the model calculations. All geometries were optimized until the predicted change in energy was typically <0.3×10⁻³kJmol⁻¹. The energy-minimized structures have no imaginary frequencies, confirming that these structures are at a stationary point on the potential energy surface.

New copper-containing residues have been named HGG0 and HGG1. The HGG0 topology contains no water molecules explicitly coordinated to the metal center, whereas the HGG1 model (used in initial trial simulations and not shown) contains only one water molecule above the coordination plane, as is evidenced in the crystal structure 32.

Copper coordination to the HGG fragment requires removal of both of the Gly amide H-atoms. The free bond sites at the amide N-atoms are instead occupied by the central copper atom, approximately along the same bond vector that would have otherwise been occupied by an H-atom. This type of copper-binding observed within the OR region is not accounted for in any of the force fields available due to several factors: 1), the Cu²⁺-binding Gly residues are missing their backbone H-atoms; 2), there are no implemented parameters for Cu²⁺ to dictate acceptable coordination geometries; 3), the redistributed charges on the backbone along the residues about the metal center are not represented; and 4), the Cu²⁺-ligand force constants are not known.

Structure parameters such as bond length, angle, and dihedrals for the new copper-containing residues were assigned based on x-ray 32, and structure data from computational models 42. All bonds, angles, and improper dihedrals are restrained using a harmonic potential. The Ryckaert-Bellemans potential is used for all dihedrals across the copper-complex. Within the copper-bound HGG fragment, harmonic potentials are used for all improper angles that cross through the copper complex, to maintain the correct shape of the binding site (approximately square planar). The dihedral angle restraints include those necessary to maintain correct backbone atom placement within the coordination plane.

To adequately represent the copper-binding HGG portion of the OR region, its physical properties must be derived in the same manner as is used for the parent force field.

Atom-centered charges for the (Cu^II(HGG)_−2H) region of the binding site, with one explicitly coordinated solvent molecule, have been obtained using the CHELPG charge derivation method 48, at the B3LYP/6-31G(d) level of theory, to represent the distribution of charge density at the coordination site. The same CHELPG method was used in the determination of atom-centered charges used in the parameterization of the AMBER force field—from which the charges were derived for the OPLS force field (see below). The calculated CHELPG charges for the new copper-containing topologies are close in magnitude to those of their related atoms in the OPLS force field. Additional details of the charge model and related OPLS charges for comparison with those derived are contained in Fig. S1 in Supplementary Material .

The force constants used to represent the copper-ligand (Cu-L) bonds in the OPLS force field are determined from the output of the G03 frequency calculations, calculated at the B3LYP/6-31G(d) level with the keyword freq. Since it is problematic to isolate vibrations due only to Cu-L stretching modes in large polydentate complexes like (Cu^II(Ac-HGGG-NH₂)_−2H), we instead analyzed the analogous bond stretching modes in model Cu²⁺ complexes containing a number of waters of hydration, as well as small molecule derivatives of the desired protein donor ligands (methylimidazole, water, and various coordination modes of N-methylacetamide). Summarized in Table 1, the OPLS force-constant parameters have been determined from the ν(Cu-L) modes of these small hydrated copper complexes, output in mDyne Å⁻¹, and scaled by 60230 to convert them to harmonic bond force constants (in kJÅ⁻²) for the OPLS force field.

The HGG0 topology includes all of the bond, angle, and dihedral restraints of the parent amino acids His and Gly. In addition to these restraints, additional angle and dihedral angle restraints are included for all possible angles that include the Cu²⁺ center. In addition to the HGG0 restraints, the HGG1 topology uses a set of restraints to maintain placement of the axial solvent molecule. A series of HGG1-containing molecules was also supplemented with a minimal set of restraints (one distance and two dihedral angles) to maintain the placement and orientation of the Trp indole side chain in the orientation observed in the crystal structure 32.

MD simulations were carried out using GROMACS software, version 3.1.2 49,50, with the OPLS-(AA)/L all-atom force field 51. Temperature (300K) and pressure (1 bar) were kept constant by weakly coupling the system to an external temperature bath (τ=0.1 for temperature, and τ=1.0 for pressure) 52. The particle-mesh Ewald (PME) method was employed for long-range electrostatics 53. SETTLE (for water) and LINCS were used to constrain bond lengths 54,55. Simple point charge (SPC) waters, used in all of the simulations, were minimized by steepest descent (md=steep) before simulation. All systems were equilibrated for 5ns before production runs. All simulations employed a 2-fs time step and periodic boundary conditions. As all model peptide fragments are overall charge-neutral, no counterions were added.

The DSSP program was used for secondary structure analysis of all simulations 56. Molecular visualizations were created using VMD 57, and MOLMOL 58. The GROMACS program g_mindist was used to verify that the simulation box sizes were sufficiently large, such that the peptide was not affected by its reflected partners in the periodic array. Cluster analysis was performed using the g_cluster program, with structures generated from the molecular trajectory files every 10ps and grouped according to their respective root mean-square deviations (RMSDs), based on the method of Daura, et al. 59.

Characterization of the structure preference and conformational dynamics of the metal-free OR₂ model (Ac-PHGGGWGQPHGGGW-NH₂ (Figure 1B)) was carried out in a 4.5×4.5×4.5-nm box containing 2938 SPC water molecules for a simulated time of 150ns.

Initial simulations containing a single Cu²⁺-bound HGG site (sequence: Ac-P[HGG0]GW-NH₂) in a 3.0×3.0×3.0-nm box containing 857 SPC water molecules were run for 5ns at simulated temperatures of 300 and 500K to ensure structure stability of the core Cu²⁺-binding region and to monitor its behavior.

Larger simulations that contain two adjacent Cu²⁺-bound HGG fragments representing a portion of the mammalian PrP OR region are depicted in Figure 1C. The metal-bound OR₂ simulations were carried out in a 4.5×4.5×4.5-nm box containing 2938 SPC water molecules.

To aid in the presentation and discussion of the copper-containing OR₂ simulations, a simplified representation of the peptide fragments is used below. The following description presents the portions of the molecule necessary to orient the reader in further discussions. The N- and C-terminal ends (CH₃-C(O)− and NH₂, respectively) are displayed in all such representations. The Pro residues are represented as five-member rings, whereas the copper-binding plane and indole ring are represented by flat discs of sizes that represent their actual structures. Interactions involving the copper-binding regions are described as involving the top or bottom face of the binding plane. These facial distinctions are defined with the N-terminal end (and the imidazole ring of His) on the left of the plane, and the C-terminal end on the right. These relative orientations of the copper-binding site will be used throughout. When referring to relative orientations and structuring between two copper-bound regions, the N-terminal site is noted first, collectively referred to as residue 3 in the sequence (Figure 1C), followed by the second C-terminal copper-bound region, residue 9 in the sequence. For example, a bottom-to-bottom stacked structure (abbreviated b→b) indicates that the bottom face of the first copper-binding site is interacting with the bottom face of the second.

The most stable core N₃O₁-coordination geometries for seven stable conformations found for [Cu^II(Ac-HGGG-NH₂)_−2H]⁰—with coordination via the histidine N_δ-atom and the first two Gly amide N-atoms—differs in the relative orientation of the His side chain and backbone atoms. The published crystal structure 32 shows the C_β group of His to be oriented below the copper plane and the C_α group above, with the imidazole ring lying approximately in the copper-peptide coordination plane. This geometry appears to be the global minimum. The nearest local minimum has the opposite orientation of the C_β and C_α groups and the imidazole ring is twisted out of the coordination plane. Their calculated relative energy difference (ΔH_(g)) at the B3LYP/6-311+G(2df,2p) level is 8.64kJmol⁻¹, favoring the crystal-structure-like geometry (geometry 1). Determination of the relative aqueous free energy (ΔG_(aq) in Eq. (1)) requires calculating the relative gas-phase free energy change and applying a correction for the free energy of solvation upon going from the gas phase to the solvated state (ΔΔG_(solv) in Eq. (1)) for each species.

(1)

Equation (2) shows the expansion of Eq. (1), with the individual terms calculated for each structure, corrected from 1atm to 1M concentration.

(2)

The metal-free form of the two-octarepeat peptide fragment was run for a simulation time of 150ns. Over the course of the entire 150-ns simulation, the GWGQ linker region frequently samples bend-and-turn-type conformations. The bend-and-turn conformations of the linker region have similar overall structure, as shown by the fit of the GWGQ backbone from structures randomly sampled from the simulation depicted in Fig. 2. The GWGQ portion, which starts fully elongated, forms a bend, as identified by secondary-structure analysis (not shown), within the first 2.5ns of the 5-ns equilibration phase (which is not included in the 150-ns analysis). Bends do not have any specific H-bonding pattern and are defined as a direction change of >70° 56. Turns are identified by specific H-bonding and are also sampled by the GWGQ region. The initial bend conformation forms a transiently stable turn from 45 to 48.2ns with an H-bond (2.21±0.65Å) between the Gly-6 amide H-atom and the Gln-9 amide O-atom. The bend conformation continues to persist for the first 90ns before adopting a more defined turn-type structure.

Display large version of this figure

Figure 2

Metal-free OR₂ structures randomly sampled over the 150-ns simulation. Their GWGQ backbone heavy atoms are aligned with the residues of the bend highlighted, shown in two 90°-rotated orientations.

The second turn conformation forms after 90ns and is identified by an H-bond (2.52±0.55Å) between the Gly-6 (residue i in Figure 1B) carbonyl O-atom and the Gln-9 (residue i+4) amide H-atom. The GWGQ region then interconverts between a bend and a turn. The Gly-6 amide H-atom also forms a transiently stable H-bond (2.58±1.18Å) with the amide O-atom of Gly-14 over the course of 95–150ns. The average Φ and Ψ angles in the linker-region turn for residues i+1 (Trp-7) and i+2 (Gly-8) from 90 to 150ns are Φ_i+1=−69±17°, Ψ_i+1=122±11°, Φ_i+2=120±9°, Ψ_i+2=−18±15°. Plots of the Φ and Ψ angles of these residues are displayed in Fig. S5 in Supplementary Material . These Φ and Ψ angles appear to be most consistent with a type II β-turn, with the Gly residue in the i+2 position, allowing for greater flexibility. Before the 90-ns time point, when the GWGQ region is predominantly in a bent conformation, these i+1 and i+2 Φ and Ψ angles are more variable, with an average standard deviation of 58° (decreasing to a standard deviation of 29° over the shorter time spent in the H-bonded turn conformation from 45 to 48.2ns).

The structural flexibility of the backbone region about the metal center of the HGG0 atoms is shown in two different orientations in Fig. 3, generated from 20 randomly selected time points. The RMSD of the HGG heavy atoms in Fig. 3 is ∼0.21Å at 300K. Over the course of the full 5-ns trial simulation at 300K, the average RMSD of the HGG heavy atoms is 0.36±0.06Å, with an only slightly larger variation of 0.38±0.07Å observed at 500K. The 5-ns simulations demonstrate that the HGG0 and HGG1 regions are stable and are suitable for larger simulations (HGG1 data not shown).

Display large version of this figure

Figure 3

Overlay of the backbone heavy atoms from 20 randomly selected MD structures of the copper-containing HGG0 residue topology, shown in two 90° orientations. Flexible regions are related to allowed vibrational modes. The backbone heavy atom RMSD is ∼0.21Å and shows that the copper-binding site is stable throughout the simulation.

A total of 10 simulations were carried out with this model system in two sets of 5×50-ns simulations: the first (series I) contains two dihedral angle restraints—in addition to those defining the copper-binding plane—on the histidine residue to maintain the lowest energy conformation, and the second set of simulations (series II) uses a minimal set of His dihedral angle parameters, potentially allowing for more flexibility of the C_α, C_β, and imidazole groups. In the second series of simulations, although the initial geometry about the metal centers is in geometry 1 (see above), the side-chain flexibility could potentially allow for interconversion between geometry 1 and geometry 2.

The RMSD plots (selected simulations are shown in Figure 4AB, for series I and II, respectively) are generated relative to each simulation's starting conformation before the equilibration step, and the rise in RMSD indicates that the structures have moved away from the starting conformation. In the course of all 10 simulations, the size of the fluctuations in the RMSD plots decreases within the first 15ns in almost all cases, and within 25ns in one instance. The decrease in RMSD fluctuations correlates with the formation of a stable, compact conformation in all simulations.

Display large version of this figure

Figure 4

(A and B) All-atom OR₂ RMSD plots for representative simulations from series I (A), which uses restraints for the His side-chain conformation to maintain rigidity, and series II (B), which does not use any His side-chain dihedral restraints. RMSD plots are generated relative to a single fully extended conformation of the OR₂ peptide. (C and D) Internuclear Cu-Cu separation for the simulations shown in A and B, respectively. Note that the major change in the RMSD occurs at the same time as the collapse of the Cu-Cu distances; this was found in all 10 simulations (five each in series I and II).

In all simulations, the decrease in RMSD fluctuation is accompanied by the GWGQ linker region adopting a bent conformation—as observed in the metal-free simulation (see above)—and the Cu²⁺-binding regions collapsing together. During these simulations the GWGQ linker region forms bend-and-turn-type structures similar to those observed in the metal-free simulation. Additional turn conformations of the linker region are sampled in the Cu²⁺-containing simulations, such as a three-residue turn involving an H-bonding interaction between the Trp-5 amide H-atom and the backbone amide O-atom of Gln-7 (residue numbers are shown in Figure 1C). Although this structure incorporates an H-bonding interaction, it does not fit the normal definition of a turn, as the H-bonding occurs in the reverse direction.

The bending of the linker region facilitates close approach of the HGG regions, with the planar Cu²⁺ sites forming stacked structures that interact via either their top or bottom face. No solvent molecules are present between the interacting faces of the Cu²⁺ sites in the collapsed structures, allowing the binding sites to come within the van der Waals contact range.

All 10 initial OR₂ structures at time 0 begin in a fully-elongated conformation, where the Φ and Ψ angles in the GWGQ region are approximately −140° and 135°, respectively. In this initial conformation, both Cu atoms are maximally separated (∼22Å). At the same time points at which the RMSDs settle in Figure 4AB, the Cu-bound HGG sites collapse and the internuclear Cu-Cu distances fall below ∼7Å or lower, as shown in Figure 4CD. Larger average internuclear Cu-Cu separations are observed in top-to-top complexes (Figure 5C). The top-to-top complexes are less compact and therefore give rise to larger internuclear Cu-Cu separations.

Display large version of this figure

Figure 5

Representations of the copper-bound OR₂ models. Structures shown in cartoon format are used to represent the most relevant structural features of the major conformations observed in all simulations. A single snapshot of the molecule from the simulation is shown below each. The shorthand notation used with each structure indicates which face of the first planar copper site in the sequence is interacting with which face of the second site: (A) Fully extended, with the copper-binding sites maximally separated; (B) bottom-to-bottom (b→b); (C) top-to-top (t→t); (D) top-to-bottom (t→b); and (E) bottom-to-top (b→t).

A simulation of the OR₂ peptide with both of the Cu-HGG regions initially in the higher-energy conformation (geometry 2 in the DFT studies) was run (data not shown). As in the series II simulations, a minimal number of restraints about the His residue and metal center were employed to potentially allow for interconversion between geometry 1 and geometry 2. In this instance, the local geometry does not convert to geometry 1 and the Cu centers do not approach closer than ∼8.5Å in the course of the simulation.

Plots of the Φ and Ψ angles for each of the residues in the linker region, as well as overlays of the GWGQ backbone, are shown in Fig. 6. The overlaid structures in Fig. 6, as well as each point on the Φ/Ψ plots, are generated from 2000 snapshots over the final 20ns of selected simulations where the specific stacking modes shown in Fig. 5 are observed. Secondary-structure analysis from all simulations shows that the GWGQ linker region between copper-binding sites has a high propensity for forming a bend, as shown for various models of stacking interactions. Even in simulations where the peptide is still essentially in an elongated conformation at time 0, the local GWGQ region frequently samples a bent conformation early on. Bending of the short GWGQ region facilitates collapse of the peptide model, allowing close approach and appropriate stacking of the copper-binding sites. These observations are analogous to the metal-free simulation shown in Fig. 2.

Display large version of this figure

Figure 6

Bending of the GWGQ linker region in the copper-bound OR₂ simulations. Overlaid structures are generated from the GWGQ backbone heavy atoms and are highlighted. Each of the structures and the points on the Φ/Ψ plots are generated from the last 20ns of selected simulations. Distinctions are made between sets of data based on the type of copper-copper stacking arrangement observed: (A) bottom-to-bottom; (B) top-to-top; (C) top-to-bottom; and (D) bottom-to-top.

Regardless of the type of stacking between the metal binding sites, the internuclear Cu separations are generally all within the same range, ∼4–7Å (averaged from the last 20ns of all 10 simulations, from series I and II). The bottom-to-top and top-to-bottom conformations are relatively flexible and display a wide spread of copper-copper separations in the collapsed conformations: 5.67±0.42Å and 6.15±0.44Å, respectively. The copper-copper separation in the top-to-top conformation is 5.83±0.67Å, with both copper-binding sites interacting relatively loosely (with one metal-binding site sliding between the top face of the other copper-binding site and its adjacent Pro residue). The copper-copper distance in the bottom-to-bottom conformation is 4.96±0.32Å, demonstrating that this conformation is more stable and compact compared to the other stacking structures observed. No interconversion is observed between different stacking orientations on the timescales of these simulations. Only in the case of the top-to-top structure (from simulation 3 of series II, not shown) are the metal-binding sites seen to separate after initially collapsing. During this separation, the Cu-HGG sites form a transient T-shaped structure, interacting via the top face of one site and the edge of the second (the average Cu-Cu separation is ∼9Å) before re-forming the initial top-to-top geometry.

Close approach of the copper-binding sites does not appear to be an artifact of the representation of the Cu-HGG site. The His imidazole rings of the OR₂ peptide also come into close proximity within the first 10ns of the metal-free simulation. Plotting the internuclear separation between the imidazole N-atoms that would otherwise coordinate Cu²⁺ (the N_δ1-atoms), the closest approach occurs only for those conformations where the GWGQ region is bent. Of the conformations explored over the 150-ns metal-free simulation, 41% have an N_δ1-N_δ1 separation <10Å, and 7% are <5Å—with an approximately linear decrease in this proportion over the 10–5Å range. This data is summarized in Fig. S5 in Supplementary Material .

OPLS was employed in these simulations because it is an all-atom force field specifically designed for the purpose of modeling peptides and organic molecules in the condensed phase 51. The CHELPG charge-calculation method at the B3LYP/6-31G(d) level was used for the original parameterization of the force field and was therefore the logical choice for generating charges for the new residue topologies described herein. The PME method for treatment of electrostatic interactions is currently regarded as a reliable method for calculating such interactions within biomolecular simulations. Comparisons between various methods, including PME, by Baumketner and Shea have shown that although the PME approach is computationally costly, it accurately reproduces the experimentally-derived free-energy minimum of test peptide models 60.

PrP is capable of responding to a range of Cu²⁺ concentrations in its local environment at the exterior membrane surface of the synapse 43. The OR region of PrP can bind Cu²⁺ in three distinctly different coordination modes, represented schematically in Figure 7A. At low [Cu²⁺], each OR cooperatively binds to the metal and involves a minimum of three (likely four) His imidazole side chains. This mode of binding is referred to as a 1:1 (one Cu²⁺ ion/one OR region) coordination in Figure 7A. At intermediate [Cu²⁺], one copper bridges two ORs via two His imidazoles and a single deprotonated backbone amide N-atom, most likely from one of the coordinating His residues or its neighboring Gly—referred to as 2:1 coordination in Figure 7A. When the local [Cu²⁺] is high, each OR site can bind up to one equivalent of copper (Figure 7A, 4:1). These simulations are relevant to the fully-Cu²⁺-bound form of the OR region under conditions of high Cu²⁺ load.

Display large version of this figure

Figure 7

(A) The types of copper-coordination mode presented by the PrP OR region in response to the local copper concentration within the synaptic cleft. (B and C) Examples of the local backbone conformation in the OR region of PrP under various copper loads, e.g., the 4:1 complex (B) and the 2:1 complex (C), based on the conformational preference of the GWGQ residues flanking each HGG portion. (D) The observed stacking structures of the metal-bound OR₂ models are extended to the full-length OR₄ region of human PrP. Each of the stacking orientations is denoted to the left of the interacting pairs and corresponds to one of the metal-bound OR₂ structures observed in the MD simulations.

Both the metal-free and metal-bound simulations indicate that the linker region between each of the PHGG sequences in the OR region adopts a bent or turn structure—which have the same approximate shape and backbone conformation. Of the three potential copper coordination modes available to the OR region, the 4:1 form of the OR sites, modeled herein, likely induces the largest degree of structuring within the N-terminal domain of PrP.

The binding environment of the Cu²⁺-containing HGG models we used correspond to the equatorial N₃O₁ donor atom environment that persists as the dominant species at physiological pH. This is the consensus coordination mode of the fully-Cu²⁺-loaded octarepeat region for the cellular form of PrP, and this model is consistent with EPR, crystallographic, and x-ray absorption data 31,32,43,61. Due to the inflexibility of the peptide backbone residues at the Cu²⁺-binding sites, there appears to be only one relevant stable conformation for this region. Because the Cu²⁺-binding portions maintain only the one conformation, this significantly reduces the conformational space accessible to the OR₂ peptide model we used. Therefore, the bulk of the conformational flexibility in the fully-Cu²⁺-loaded OR region originates primarily in the linker region.

The most significant observations from the metal-free OR₂ simulation are that the HGG region adopts a flexible loop conformation, whereas the GWGQ linker region samples predominantly bend-and-turn-type conformations. Once formed, the bending of the linker region persists in both the metal-free and all of the metal-bound simulations. There are a total of five GWGQ segments spanning the OR region in human PrP, each one flanking a copper-binding site (Figure 1A). Each of these linker regions is likely to form transient bends in the metal-free form of PrP—and may be subsequently stabilized in other copper-bound states, such as those shown in Figure 7BC.

Throughout the simulations of metal-bound peptides, the copper-binding sites arrange themselves into specific orientations after the collapse of the peptide. Of the conformations explored, those of interest are observed to persist for several nanoseconds, and in some cases much longer. There are five major arrangements of copper-coordination sites observed: 1), no interaction, where both copper-binding sites are separated for some finite time, similar to the configuration in Figure 5A; 2), bottom-to-bottom (b→b) orientation, as in Figure 5B; 3), loosely stacked top-to-top (t→t) orientation, as in Figure 5C; 4), top-to-bottom (t→b) orientation, as in Figure 5D; or 5), bottom-to-top (b→t) orientation, as in Figure 5E. In the top-to-top case, the copper-coordination sites slide across one another, allowing the top of the second copper-binding site to also facially interact with the first proline. In addition to the major conformations, there are other transient orientations of copper-binding sites that persist for shorter durations, including side-on interaction of one copper-binding site with one face of another.

The most common structure encountered in these simulations, containing either the HGG1 (results not shown) or HGG0 topology, is the bottom-to-bottom arrangement of copper-binding structures. This arrangement is sufficiently stable that no other conformations are significantly explored within the simulation time once it is formed. In several instances, the bottom-to-bottom stacking persists for the entire 50ns, having already been achieved during the 5-ns equilibration phase. These observations effectively put a minimum folding time limit on the formation of the stacking structures—potentially <5ns.

An EPR investigation of copper ions bound to tandem OR sites, by Chattopadhyay et al. has revealed weak diamagnetic coupling between the copper centers. Because coupling between paramagnetic centers is distance-dependent, it has been possible to derive Cu-Cu distances in the range of 3.0–6.0Å from these experiments 43. The measured Cu-Cu distances in these simulations demonstrate that all of the copper-binding sites may lie within van der Waals contact. The copper separations indicate that each of the collapsed OR₂ structures (shown in Figure 5BE) is compatible with the experimentally derived range of internuclear separation.

Structuring within the copper-bound OR domain has been observed elsewhere as well. The CD measurements of Garnett and Viles, using peptide fragments derived from the OR region, provide direct evidence that once Cu²⁺ is bound to consecutive HGGGW sequences, the metal centers are not isolated from one another, and specific structuring of the linker region occurs 36. This conclusion was inferred from the CD fingerprint of the fully-Cu²⁺-loaded OR₄ peptide fragment, which contains additional elements of structure unaccounted for by the sum of the spectra for four single Cu²⁺-bound OR fragments alone. Yoshida et al. 39, and Zahn 40, using peptide fragments of the OR region in the absence of Cu²⁺, confirmed that the GWGQ region has a propensity to adopt a bent conformation, as collapse of the metal-free peptide is not apparent from their results. Data from all of the simulations carried out here indicate that this same region maintains its propensity to adopt a bent structure in the presence of bound Cu²⁺, and that the bend of the linker region facilitates the collapse of the copper-binding sites.

Several attempts have been made to obtain information about the structure of the copper-bound OR region in solution. Łuczkowski et al. 62 proposed a copper-bound structure that has some differences in the primary coordination sphere of Cu²⁺ that are not compatible with the N₃O₁ coordination environment at physiological pH for the 4:1 Cu-OR complex 43. A structural study by Mentler et al. has used the apparent close association of the Trp ring NH, observed by Łuczkowski et al. 62, as a distance restraint in conformational searches of Cu²⁺ bound to Ac-PHGGGWGQ-NH₂ in concert with a host of other experimentally derived parameters 63. Other models, offered by Bonomo et al., suggested additional coordination of adjacent backbone carbonyl O-atoms to the apical positions of the core N₃O₁ complex 64. Although these structures appear to be isoenergetic with the four-coordinate N₃O₁ complex (data not shown), the current lack of sufficient evidence for such interactions from NMR relaxation experiments 42 or the only available crystal structure 32 has guided us toward explicitly defining only the core N₃O₁ four-coordinate complex.

The published x-ray crystal structure of the copper-bound HGGGW peptide fragment of the OR site shows the indole ring of the Trp side chain stacked above the His imidazole ring at the copper-binding site 32. Other experiments, such as ¹H NMR relaxation rates, give evidence for the close approach of the Trp indole to the bound Cu²⁺ center 62. The persistence of this interaction in solution is not directly evident from the experimental data and could be a transient interaction. The Trp residue has been shown, experimentally, to be necessary for structuring the binding site for Cu²⁺ within the octarepeats and directing a preferred N₃O₁ donor atom count for the metal center 31. Quenching of Trp fluorescence, upon the addition of Cu²⁺, is also often used to monitor Cu²⁺ binding to the OR domains 9. Tryptophan fluorescence quenching indicates recognition of Cu²⁺ binding by the Trp residues. The mechanism of Trp quenching has not been fully elucidated but may occur from transient conformations where the Trp makes a close approach to the paramagnetic Cu²⁺ center and bound imidazole. Such an approach would also increase the relaxation rates of those Trp indole protons closest to the metal ion.

The copper-containing HGG model topologies implemented herein are static structures, with the ligand sphere held in place. For this reason, we cannot address any dynamic structural influence that Trp may have on coordination geometry or donor atom variability of the bound copper ion. However, during these simulations, the Trp side chains make frequent close approaches to Cu²⁺ through the exposed face of the coordination sites that are not involved in stacking interactions.

The Cu²⁺-HGGGW crystal structure implicates a direct role for the Trp at the coordination site, through H-bonding between the indole ring and a water molecule coordinated at the metal center. Bound water molecules within the first solvation shell of aqueous Cu²⁺, generally regarded as [Cu(H₂O)₆]²⁺, undergo rapid exchange (on the order of 1×10⁹ s⁻¹) 65. Within the copper-bound OR regions, the doubly deprotonated peptide backbone makes the overall charge of the metal complex neutral. This decreased charge at the metal center will likely lead to a reduced propensity for axial water interaction with the copper center, over the 2+ ion. Only one coordinated water molecule is observed in the crystal structure of Cu^II(Ac-HGGGW-NH₂)_−2H32, the frequency of solvent exchange from the apical positions is as yet unknown. Close approach (<6Å) of two adjacent Cu atoms in the HGG sites can only occur when the interacting faces of the copper-binding sites are desolvated. This realization is important in reconciling the short internuclear Cu-Cu distances observed by EPR 43 with the consensus N₃O₁ coordination environment of the ORs 32,43.

Considering the bottom-to-bottom structures, the presence of additional copper-bound ORs would allow for more varied interaction between nonadjacent coppers through exchange of facially interacting coordination sites. Analysis of peptide-peptide, peptide-solvent, and solvent-solvent interaction energies indicates that there is little difference in the total energy of the system for the various stacking conformations explored. Larger OR₄ simulations may yield significant energy differences between extended stacking structures, indicating which conformations may be more favored. As all of the observed stacking interactions for the dicopper-bound OR₂ fragment appear to be isoenergetic, mixtures of conformations for this region may exist.

From the metal-free and metal-bound simulations, the propensity for the linker region to adopt a flexible bend is evident. Figure 7BC, shows an overlay of the bent GWGQ sequence with the OR region with the 1:1 and 2:1 copper-coordination modes. Although there may be additional elements of structure within either of these coordination environments, the bend of the linker region observed herein neatly accommodates these other modes of Cu²⁺ binding as well.

Extending the structures of the model OR₂ peptide to the full-length OR region in PrP yields a larger assortment of potential conformations of an OR₄ fragment. Examples of the types of conformations possible for an OR₄ fragment are presented in Figure 7D, and are all related to the observed OR₂ conformations shown in Fig. 5. Because the backbone loops back on itself for each adjacent OR to interact, the extended OR₄ domain could accommodate an extended stacking structure between adjacent ORs.

The short OR₂ fragments used here demonstrate the most relevant features that would be observed for larger copper-containing OR models, specifically the preferred conformations of the flexible GWGQ linker region and the potential stacking interactions of the collapsed copper-binding sites. The internuclear Cu-Cu separations obtained from these simulations are within the experimental range observed by Chattopadhyay et al., determined from EPR measurements of Cu²⁺ bound to tandem ORs 43. Renner et al. reported that the OR region can preorganize through interaction with micelles, and the structuring persists in the presence of Cu²⁺66. The results herein provide an additional level of structure information that may aid in refinement of OR-membrane interaction models. The CD experiments of Garnett and Viles have shown that the OR region, which is normally flexible, contains additional structure elements in the presence of Cu²⁺36. The results of the metal-free simulations show that the GWGQ linker region remains flexible, interconverting between a number of local bend-and-turn-type conformations, whereas the copper-bound form of the same OR₂ peptide demonstrates stabilization of the bent conformation of the GWGQ region. It is clear that although the N-terminal region of PrP, as a whole, may be largely unstructured, these results demonstrate the potential for stabilization of the local structure within the fully-copper-loaded OR sequence under conditions of high [Cu²⁺] in the prion protein's local environment. These simulations only address the potential for intrarepeat stacking of neighboring copper-binding sites; interrepeat stacking between OR regions of neighboring PrP molecules may also be possible.

After synaptic release of copper, any free copper ions that are in the Cu²⁺ oxidation state can readily participate in the generation of ROS 15. Under conditions of high [Cu²⁺], the 4:1 OR coordination mode binds Cu²⁺ in a redox-inactive state, protecting against the deleterious effects of potential Fenton-type or Haber-Weiss chemistry. We propose that the stacking of individual copper-binding sites in the manner presented herein may further enhance the neuroprotective mechanism of PrP by sequestering the redox-inactivated Cu²⁺ sites and limiting access by solvent or other extraneous species, which may promote ROS formation.