Full atomistic model of prion structure and conversion

RESEARCH ARTICLE Full atomistic model of prion structure and conversion Giovanni SpagnolliID 1 *, Marta RigoliID 1,2 , Simone Orioli 2,3 , Alejandro M. SevillanoID 4 , Pietro Faccioli 2,3 , Holger WilleID 5 , Emiliano Biasini 1 *, Jesu ´s R. RequenaID 6 * 1Department of Cellular, Computational and Integrative Biology (CIBIO)–University of Trento, Povo TN, ITALY, 2Department of Physics, Povo, Trento TN, ITALY, 3INFN-TIFPA, Povo (Trento), ITALY, 4Department of Pathology–University of California—San Diego, San Diego, California, United States of America, 5Department of Biochemistry and Centre for Prions and Protein Folding Diseases–University of Alberta, Edmonton (AB), CANADA, 6CIMUS Biomedical Research Institute & Department of Medical Sciences, University of Santiago de Compostela-IDIS, Santiago, SPAIN *giovanni.spagnol[email protected] (GS); [email protected] (EB); [email protected] (JRR) Abstract Prions are unusual protein assemblies that propagate their conformationally-encoded information in absence of nucleic acids. The first prion identified, the scrapie isoform (PrP Sc ) of the cellular prion protein (PrP C ), caused epidemic and epizootic episodes [1]. Most aggregates of other misfolding-prone proteins are amyloids, often arranged in a Parallel-In-Regis- ter-β-Sheet (PIRIBS) [2] or β-solenoid conformations [3]. Similar folding models have also been proposed for PrP Sc , although none of these have been confirmed experimentally. Recent cryo-electron microscopy (cryo-EM) and X-ray fiber-diffraction studies provided evidence that PrP Sc is structured as a 4-rung β-solenoid (4RβS) [4,5]. Here, we combined different experimental data and computational techniques to build the first physically-plausible, atomic resolution model of mouse PrP Sc , based on the 4RβS architecture. The stability of this new PrP Sc model, as assessed by Molecular Dynamics (MD) simulations, was found to be comparable to that of the prion forming domain of Het-s, a naturally-occurring β-solenoid. Importantly, the 4RβS arrangement allowed the first simulation of the sequence of events underlying PrP C conversion into PrP Sc . This study provides the most updated, experimentally-driven and physically-coherent model of PrP Sc , together with an unprecedented reconstruction of the mechanism underlying the self-catalytic propagation of prions. Author summary Prions are unusual infectious pathogens that do not contain any nucleic acid. They consist of assemblies of misfolded proteins. The scrapie isoform of the mammalian prion protein, PrP Sc , is the most notorious prion, and is responsible for deadly neurodegenerative diseases affecting humans, like Creutzfeldt-Jakob disease, and animals, such as bovine spongiform encephalopathy (“mad cow disease”) and chronic wasting disease affecting elk and deer in North America and, more recently, Europe). Understanding the structure (“shape”) of the PrP Sc prion is critical to understand how it propagates. We have created a very detailed model of PrP Sc , which includes all its atoms, using computational PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 1 / 18 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Spagnolli G, Rigoli M, Orioli S, Sevillano AM, Faccioli P, Wille H, et al. (2019) Full atomistic model of prion structure and conversion. PLoS Pathog 15(7): e1007864. https://doi.org/10.1371/ journal.ppat.1007864 Editor: Surachai Supattapone, Dartmouth College Geisel School of Medicine, UNITED STATES Received: March 1, 2019 Accepted: May 23, 2019 Published: July 11, 2019 Copyright: ©2019 Spagnolli et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: JRR was funded by grants BFU2017- 86692-P and BFU2013-48436-C2-1-P from the Spanish Ministries of Economy and Competitiveness, and Education and Science, respectively, both partially including FEDER funds from the European Union. This work was also supported by a grant from Fondazione Telethon (Italy, TCP14009). GS is a recipient of a fellowship from Fondazione Telethon. EB is an Assistant techniques. The model resembles a 4-rung cork-screw. The model is stable and agrees with all available experimental data on PrP Sc . This structure allowed us to model for the first time the process of prion propagation at high resolution. These data inspire several new hypotheses to elucidate prion biology and design therapeutics for prion diseases Introduction Prion diseases are infectious neurodegenerative disorders characterized by an invariably lethal outcome caused by a proteinaceous infectious agent named “prion” [1]. The central event in these pathologies is the conversion of PrP C , a GPI-anchored protein of unknown function, into a misfolded isoform (PrP Sc ) which accumulates in the central nervous system of affected individuals [6]. While PrP C structure has been widely characterized, and consists of a N-termi- nal disordered tail and a C-terminal globular domain [7], no high-resolution information is available for PrP Sc due to technical challenges posed by its high insolubility and aggregation propensity [8]. In order to fill this gap, different atomistic models based on low-resolution experimental data have been proposed, including a Left-handed-β-Helix (LβH) structure spanning residues 89 to 170 while retaining the two C-terminal α-helices of PrP C [9], and a Parallel In-Register Beta-Sheet (PIRIBS) architecture, characterized by intermolecular stacking of aligned PrP monomers [10]. Recent cryo-EM data obtained using infectious, anchorless PrP Sc fibrils [4] provided strong evidence indicating that PrP Sc fibrils consist of two independent protofilaments, and of the existence of 2 nm structural units repeating along each protofilament axis, suggestive of a 4-rung β-solenoid (4RβS). This would be fully compatible with the LβH model, and much less so with the PIRIBS model, which posits that PrP Sc fibrils are not made up by two protofilaments, but rather, by a single wider filament that features two subdomains or lobes separated by a cleft (Goveman et al., 2014). Moreover, the PIRIBS model fails to accommodate glycosylated residues in PrP Sc , which would result in the introduction of excessive steric clashes [11]. It should be pointed out, however, that the low resolution of available experimental data still does not allow to definitively discard any option (a detailed comparison of PIRIBS and solenoid-based models can be found in [12]). Of note, while consistent with the mentioned experimental constraints, the proposed LβH model is incoherent with a recent reevaluation of previous FTIR data suggesting that PrP Sc does not contain α-helices [8]. Results Construction of an experimentally-driven atomistic model of infectious mouse prion To satisfy current experimental evidence and theoretical structural constraints, we built a new atomistic model of mouse PrP Sc based on the 4RβS conformation, and tested its stability by means of all-atom MD simulations. The construction of the model took into account an array of experimental data, including: (i) cryo-EM [4] and X-ray fiber-diffraction studies [5], which showed that the fold of a mouse glycosylphosphatidylinositol (GPI)-anchorless, infectious PrP Sc is compatible with a 4RβS architecture with L- or T- shaped cross-section [4]; (ii) circular dichroism (CD) and FTIR spectroscopy, which ruled out the presence of α-helices, and suggest that PrP Sc contains approximately 40–50% β-sheet and 50–60% coil/turns [8]; (iii) Mass Spectrometry (MS) analyses indicating the presence of an intact disulphide bond between residues C178 and C213 (mouse sequence) [13], as well as mapping Proteinase K (PK)-sensitive residues, which reflect amino acids likely excluded from the resistant core of the protein [14] Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 2 / 18 Telethon Scientist at the Dulbecco Telethon Institute. HW acknowledges support through grant 201600029 from the Alberta Prion Research Institute. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. [15]; and (iv) the possibility of accommodating complex glycans at positions N180 and N196 [11]. All these constraints were comprehensively included into a 2D threading scheme spanning mouse PrP (moPrP) residues 89–230 (S1 Fig), also considering the structural propensities of different residues: polyglycine tracts and prolines were positioned in loops due to their destabilizing effects on β-strands; charged sidechains were excluded from the inner core of the protein or counterbalanced by salt bridges. This scheme was then modelled onto the 3D arrangement of a naturally-occurring β-solenoid protein (Dickeya dadantii Pectate Lyase; PDB 1AIR). The resulting structure (depicted in Fig 1) features an inner core containing mainly hydrophobic or mildly polar side-chains (T94, T106, L108, V111, Y127, M128, W144, Y149, V165, Y168, I181, I183, V188, F197, T198 and T200), few polar side-chains involved in hydrogen bonding (N142-HB-Y168, H168-HB-T198, Q216-HB-T200 and Q218-HB-S221), and a salt bridge (R147-SB-D166). Conversely, the majority of the highly-polar residues (N and Q) including the glycosylation sites (N180 and N196) and charged side-chains (E, D, K and R) are exposed to the solvent. The structure also encompasses identified PK cleavage sites localized in loops/turns, or at the edge of the β-strands, and the intact disulphide bond between C178 and C213. Importantly, the final model fitted with a previously described, low-resolution cryo-EM map of infectious PrP Sc (S2 Fig). Molecular dynamics simulations of the new 4RβS model To test the physical consistency of the 4RβS model, we challenged its stability by all-atom MD simulations in explicit solvent. First, three independent, 20 ns simulations were performed in the isothermal-isobaric ensemble (T = 300 K, P = 1 Bar) restraining hydrogen bonds distances between atoms involved in β-sheets. This process allowed relaxation of protein loops and side chains of the core (S3 Fig). Next, the imposed restraints were released, and three plain-MD trajectories of 100 ns each were simulated. The stability of the model was then assessed by both Root Mean Squared Deviation (RMSD) of atomic positions relative to the initial frame (t = 0 ns) and secondary structures content. Interestingly, we obtained values in the same range of fluctuation for our 4RβS model and the prion-forming domain of the fungal protein Het-s, a naturally occurring β-solenoid whose structure has been solved by solid-state NMR (ssNMR; PDBs 2KJ3 and 2RNM; Fig 2, and S4 Fig). Conversely, by applying an identical workflow to the previously proposed model of laterally-stacked LβH trimer of PrP Sc , we observed a profound instability of the β-helical domain, which was already evident after few tens of ns of dynamics (Fig 2 and S4 and S5 Figs). Next, we tried to adapt alternative threading schemes to the 4RβS architecture. We obtained two new schemes carrying small deviations (i.e. one or two residue shifts) from the original model but still fully coherent with the experimental constraints (S6 Fig). Every other attempt to model threading schemes with a larger deviation (>3 residue shifts) resulted in violation of experimental constraints. Interestingly, MD simulations revealed that the two alternative models are also significantly stable (S6 Fig). These data suggest that the new threading schemes may represent putative alternative arrangements of the PrP polypeptide into an RML prion fibril. This conclusion implies an intrinsic flexibility of 4RβS architecture to accommodate slightly different arrangements, which could be related to the ability of prions to occur in different conformations specifying prion strains. To mimic the fibrillary behavior of PrP Sc , we built a tetrameric 4RβS model by stacking monomers in a head-to-tail fashion. As expected, this assembly showed a comparable MD stability to the monomer (Fig 3). Moreover, the 4RβS tetramer fits with the two main signals obtained by Fourier transform single particle analysis of cryo-EM data (19.1 Åand the ~40 Å) [4], which reflect distances between the same residues on two contiguous or alternate Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 3 / 18 monomers, respectively, as well as with the observed volume of the protofilaments (Fig 3). Finally, by introducing complex sugar precursors (GlcNAc2Man3Fuc) at positions N180 and N196 of each monomer, we observed the absence of steric clashes, confirming that the 4RβS model accommodates the presence of glycans (S6 Fig). Collectively, these findings indicate that, in contrast to the LβH model, the 4RβS is a solid arrangement for PrP Sc , built on the most Fig 1. View of the 4RβS PrP Sc Model. The structure of PrP Sc modelled as a 4RβS (β-strands represented as red arrows) is depicted in the center of the figure. Residues are displayed in each individual rung (1–4) with different colors. PK cleavage sites identified by mass spectrometry in two different previous reports are colored in red [14] [15]. Glycosylation sites are labelled in green. Prolines are colored in purple. Cystine is indicated in cyan. https://doi.org/10.1371/journal.ppat.1007864.g001 Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 4 / 18 recent experimental data, and showing a conformational stability comparable to that of a natural prion. All-atom reconstruction of prion conversion The 4RβS model allowed us to develop an original scheme to perform for the first time a simulation of the conformational transition from PrP C to PrP Sc . In order to bridge the gap between the computationally-accessible and the biologically-relevant time scales, we employed a Fig 2. Comparison of Stability by MD Simulations of a Het-s Dimer, 4RβS and LβH PrP Sc .Upper graphs report the RMSD deviation from initial configuration for the entire structure (blue lines) or the β-solenoid core (orange lines) of the different proteins. Filled curves indicate standard error of the mean. Results show a comparable stability between the Het-s dimer (A) and the 4RβS model (B), with an average RMSD of the hydrophobic core (calculated as the average of the three trajectories over the last 5 ns, ±standard deviation) of 2.4 ±0.2 Åfor the Het-s dimer (similar results for Het-s stability are reported[35]) and 2.1 ±0.3 Åfor the 4RβS. In contrast, the structural deviation of the LβH (C) hydrophobic core is approximately two-fold higher, reaching a value of 4.3 ±0.6 Å. Lower graphs indicate the α-helical (blue lines) or β-sheet (red lines) content of each protein. The initial and final β-sheet content was calculated as the average of the three trajectories over the last 5 ns of the restrained and unrestrained MD simulations, respectively. The Het-s dimer showed a variation from an initial 41.8 ±2.3% to a final 33.2 ±2.8%. Similarly, the 4RβS model deviates from an initial 41.3 ±2.3% to a final 36.6 ±5.4%. Instead, the LβH model deviates from a starting 18.3 ±1.4% to a 7.5 ±3.1%. These results are illustrated by the structures shown below the graphs, which represent the initial (top) and final (bottom) frames of the MD trajectories. https://doi.org/10.1371/journal.ppat.1007864.g002 Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 5 / 18 specific kind of biased dynamics called ratchet-and-pawl MD (rMD) [16] in the framework of an all-atom realistic force field [17]. rMD-based methods have been successfully applied to simulate protein folding and other conformational transitions [18,19]. However, this scheme provides a sampling of the transition path ensemble only if the biasing force is applied along a reliable reaction coordinate [20]. Therefore, the first step towards developing our rMD simulation was to build a statistical model to identify the reaction coordinate of the process. Using the Markov State Mode formalism, we demonstrate that among all the possible misfolding Fig 3. MD Simulation and PK Restriction Map of a 4RβS Tetramer. (A) 3D representation of a 4RβS tetramer at the beginning (left) and the end (right) of MD simulations. Blue bars indicate the distance between two residues in the same position on two consecutive monomers, which corresponds to 19.2 ±0.4 Å(t = 0 ns) and 19.8 ±2.2 Å(t = 100 ns). Purple bars indicate the distance between a residue in one monomer and the same residue on the second forthcoming monomer, which corresponds to 38.4 ±0.5 Å(t = 0 ns) and 39.6 ±1.8 Å(t = 100 ns). A similar pattern of signals reflecting monomeric and dimeric repeats has previously been observed by cryo-EM studies on Het-s [36]. Both values are in almost perfect agreement with the two main signals obtained by Fourier transform single particle analysis in the cryo-EM experiment (19.1 Åand the ~40 Åsignals). The average monomeric model volume calculated at beginning and at the end of the 100 ns tetramer simulations are equal to (18.5 ±0.7)�10 3 Å 3 and (18.5 ±0.3)�10 3 Å 3 respectively; while the estimated cryo-EM monomeric volume from the protofilament is equal to 18.9�10 3 Å 3 . (B) Upper graph shows the RMSD deviation of the tetramer from initial state for the entire structure (blue lines) or the β-solenoid core (orange lines). Structural deviation over the 100 ns of simulation corresponds to 2.6 ±0.2 Å. Lower graph report the secondary structures percentage, initial β-strand content is 46.2 ±1.2%, while the final 38.6 ±1.7%. Filled curves indicate standard error of the mean. These results indicate a comparable stability between monomeric and tetrameric 4RβS structures. (C) Map of the experimentally observed PK cleavage sites (colored in red) overlapped with the probability of each residue to be in an extended conformation, calculated over the last 5 ns of the MD trajectories. https://doi.org/10.1371/journal.ppat.1007864.g003 Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 6 / 18 patterns from PrP C to PrP Sc the prominent reaction mechanism is the sequential formation of rungs. A biasing coordinate was then built by coupling this dynamical information with the all atom 4RβS structure. The associated rMD simulations yielded a transition pathway with full atomistic resolution in which the C-terminal rung of the solenoid acts as a primary conversion surface for PrP C unstructured N-terminus (residues 89–124). This first event initiates a cascade of conformational transitions in which each newly formed rung acts as a template for the formation of the following one, ultimately leading to the complete conversion of PrP C into PrP Sc (Fig 4 and S1 Movie). This analysis provides a rigorous description of the active role of protofilament ends in the templated propagation of prions, and it is compatible with previous observations suggesting the presence of a structured intermediate conformer of PrP C in its transition to PrP Sc [21] Fig 4. Graphs and Frames Extracted from the rMD Simulation of PrP C to PrP Sc Conversion. Upper graph reports the fraction of native and misfolded contacts of the instantaneous configurations of the PrP C -PrP Sc complex, starting from the initial modelled configuration (depicted in 1) in which PrP C contacts the 4RβS monomer, with respect to the final target configuration (4RβS dimer). Lower graph shows the secondary structure content. Pictures of the evolving complex represent frames extracted from the entire conversion simulation at precise rMD steps, corresponding to the refolding of PrP C as follow: (2) residues 89–115; (3) residues 89–151; (4) residues 89–190; (5) residues 89–230. The process highlights the progressive unfolding and refolding of PrP C onto the 4RβS template, which initially involves the unstructured region, followed by the loss of α-helices and a progressive formation of β-sheets. https://doi.org/10.1371/journal.ppat.1007864.g004 Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 7 / 18 Discussion The elucidation of the structure of PrP Sc at atomic resolution has proven to be a phenomenal experimental challenge, mainly due to its high insolubility and aggregation propensity. Previously generated computational models of PrP Sc have the virtue of providing a plausible 3D structure, but fail to comprehensively accommodate most recent experimental data [8]. Here, we filled this gap by exploiting the information arising from cryo-EM and X-ray fiber-diffrac- tion studies [4,5], which clearly defined the general architecture of PrP Sc as a 4RβS and refined the structure by including experimental constraints obtained by mass spectrometry. The model we have created allowed us to perform the first reconstruction of how the information encoded into the conformation of a protein could be propagated in a directional fashion, a concept directly underlying the infectious nature of prions. Virtues and limits of the 4RβS model The search for effective therapies against prion diseases has so far been mostly unsuccessful [22], partly due to the lack of detailed information regarding the structure of PrP Sc . In fact, defining the structure of infectious prion particles at high-resolution would allow the rational design of molecules that directly target PrP Sc and inhibit its replication and accumulation into the nervous system. Unfortunately, this task is complicated by the wide heterogeneity and high propensity to aggregate of prions, aspects that make canonical high-resolution techniques such as NMR and X-ray crystallography inapplicable [8]. In an attempt to fill this gap, low-resolu- tion methods have provided important experimental constraints illuminating the macroscopic architecture of prion fibrils and laying the groundwork for the application of computational techniques such as those used in this study. While these approaches still suffer from the lack of direct high-resolution information, they are undoubtedly valuable for interpreting the available information and inspire new hypotheses and approaches. Indeed, the new model described in this study possesses several important features. First, it displays an intrinsic coherence with state-of-the-art knowledge about infectious PrP Sc fibrils and it appears to be as stable as the naturally-occurring prion Het-s [3]. Importantly, it also represents a unique workbench for interpreting future structural data or available biological evidence. In fact, it will be possible to test the effect of variations in the PRNP gene on the stability of the model by inferring mutations or polymorphisms known to favor or inhibit prion propagation and analyze the resulting structures by MD. As an example, it is known that the presence of an isoleucine at the polymorphic site 109 (M/I) in Bank vole PrP results in a substantially higher susceptibility to prion disease [23]. Assuming that the structure of prions in mouse and Bank vole are similar, as suggested by the difference in just 9 residues between murine and Bank vole PrP, and in light of the known higher propensity of isoleucine to participate in β-strands as compared to methionine [24], our model may explain the effect of the 109 polymorphism on prion replication in Bank vole. Our 4RβS structure could in principle be adapted to model prions from Bank vole, human, deer, ovine and other species, as well as be employed to test how the effect of known polymorphisms and mutations affect prion stability and/or propagation. Similarly, it will be possible to rationally predict novel amino acid substitutions that could influence the stability of the 4RβS architecture and test their effects in surrogate in vitro (e.g. PMCA) or cell-based assays for prion replication. As a result, these strategies will produce new experimental constraints, which in turn will help to further improve the accuracy of the model, leading to an iterative cycle of prediction, testing and refinement. While waiting for a technological innovation that will definitively solve the structure of infectious prions, this strategy appears as the most feasible and promising among the available options. Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 8 / 18 Using the 4RβS model to address fundamental questions in prion biology Since the original hypothesis of prions as protein particles responsible for transmissible brain pathologies, later evolved into a much wider concept of protein-based inheritance, tens of laboratories have intensively investigated the molecular features of these infectious agents. However, despite enormous efforts worldwide, many important questions of prion biology remain unanswered. For example, we still don’t fully understand the nature of the biological barriers that disfavor or fully impede the propagation of prions among different species. The reconstruction of prion conversion described in this manuscript could provide a unique opportunity to try to address this issue. For example, it will be possible to test the extent by which our templated conversion model tolerates sequence mismatches between the 4RβS template and the PrP substrate. These analyses could be used to interpret available biological evidence, for example the unusual ability of Bank vole PrP molecules to escape species barriers. Interestingly, our model may also help to improve current understanding of the prion strain phenomenon, which until now has been generally conceptualized as the same PrP polypeptide adopting different conformations. We show that the 4RβS scaffold can accommodate slightly different sequence threading schemes while remaining coherent with low-resolution experimental constraints and maintaining structural stability (S6 Fig). This observation suggests surprising flexibility of the 4RβS architecture to the alignment of residues in the β-strands, a feature that may introduce small conformational variability in each PrP Sc monomer and result in large structural effects at the level of fibrils. This conclusion implies that the different biochemical and biological features of prion strains can be encoded in the small differences of PrP sequence alignment along the 4RβS scaffold. The 4RβS model inspiring therapy Computer-aided technologies are among the most promising strategies for the discovery of novel therapeutics against a wide variety of human illnesses, including prion diseases. In most cases, these approaches require the availability of high-resolution information about the target protein, but in absence of experimentally solved structures, as is the case for PrP Sc , may also be applied to structural models. From this standpoint, our 4RβS may represent a unique opportunity to rationally design new anti-prion compounds. These molecules may be directed against two main structural elements of a putative 4RβS fibril: the “sticky” end, which exposes unpaired β-strands to the solvent, or the lateral surface. In the former case, the compound could block the insertion of a new PrP polypeptide into the growing fibril, while in the latter it would act as a stabilizer and prevent breakage into smaller, more diffusible elements. Interestingly, a similar mechanism has already been proposed to explain the anti-prion effects of luminescent conjugated polythiophenes (LCPs), molecules presenting consecutive thiophene rings exposing regularly spaced carboxyl side groups which could promote the interaction with lateral surfaces of prion fibrils [25]. It will undoubtedly be interesting to use our 4RβS structure to model the binding architecture of LCPs, as such information could then be exploited to optimize these compounds and/or rationally design immunological mimics and immunogens giving rise to PrP Sc -specific antibodies or prophylactic vaccines against prion diseases. Materials and methods Template selection & model building In order to accommodate the moPrP 89–230 sequence, we selected a β-solenoid architecture capable of satisfying the following requirements: (i) Approximate number of residues in extended conformation higher than 12 per rung. Requirement needed to fit the secondary Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 9 / 18 ns unrestrained MD simulations. Images (iv) report the initial and final frame of the respective MD simulations. (TIF) S7 Fig. Illustration of the energy minimized 4RβS tetramer carrying complex glycans. A model of energy minimized tetrameric 4RβS structure carrying glycan residues is depicted in two different orientations (A, B). Top view of a laterally stacked trimer of glycosylated 4RβS (C), compatible with the 2D crystals diffraction data. A scheme of the complex glycan precursor added is depicted in (D), blue squares indicate N-acetylglucosamine, green circles indicate mannose and the red triangle indicate fucose. (TIF) S8 Fig. Statistical model for prion propagation mechanism. (A) Schematic representation of the network of transitions leading to the incorporation of PrP C into PrP Sc . The purple arrow describes the formation of a single rung by templating the unstructured region of PrP C (89– 115) on PrP Sc (rate k 0 ). The blue arrows describe the formation of a single by means of a templating process which involves the breaking the native contacts and the docking onto a preformed rung (rate k 1 ). The orange arrows indicate the spontaneous formation and docking of two de-novo rungs (rate k 2 ). Finally, the green arrows indicate the docking of preformed rungs belonging to two adjacent misfolded regions of the same chain (rate k 3 ). (B) Simplified version of the network assuming as a priming reaction step the formation of the first rung. (C) Relative probability of the 6 reaction pathways as a function of the k 1 /k 2 rate ratio. (TIF) S1 Movie. Full-atomistic model of prion propagation. Visualization of the rMD simulation reconstructing the entire refolding events of PrP C onto the 4RβS PrP Sc model. (MP4) S1 File. Atomic coordinates of the model. (PDB) Acknowledgments Calculations were performed using supercomputing facilities Finis Terrae II (CESGA, Spain) and Marconi (CINECA, Italy). We acknowledge the CINECA award under the ISCRA initiative, for the availability of high performance computing resources and support. Author Contributions Conceptualization: Giovanni Spagnolli, Pietro Faccioli, Holger Wille, Emiliano Biasini, Jesu ´s R. Requena. Data curation: Giovanni Spagnolli. Formal analysis: Giovanni Spagnolli, Simone Orioli, Pietro Faccioli, Emiliano Biasini. Funding acquisition: Emiliano Biasini, Jesu ´s R. Requena. Investigation: Giovanni Spagnolli, Marta Rigoli, Simone Orioli, Alejandro M. Sevillano, Pietro Faccioli, Jesu ´s R. Requena. Methodology: Giovanni Spagnolli, Pietro Faccioli, Emiliano Biasini. Project administration: Emiliano Biasini. Software: Giovanni Spagnolli, Pietro Faccioli. Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 16 / 18 Supervision: Pietro Faccioli, Emiliano Biasini, Jesu ´s R. Requena. Visualization: Giovanni Spagnolli, Marta Rigoli. Writing – original draft: Giovanni Spagnolli, Emiliano Biasini. Writing – review & editing: Pietro Faccioli, Holger Wille, Jesu ´s R. Requena. References 1. Prusiner SB. Novel proteinaceous infectious particles cause scrapie. Science. 1982; 216(4542):136– 44. Epub 1982/04/09. PMID: 6801762. 2. Gremer L, Scholzel D, Schenk C, Reinartz E, Labahn J, Ravelli RBG, et al. Fibril structure of amyloidbeta(1–42) by cryo-electron microscopy. Science. 2017; 358(6359):116–9. Epub 2017/09/09. https:// doi.org/10.1126/science.aao2825 PMID: 28882996; PubMed Central PMCID: PMCPMC6080689. 3. Wasmer C, Lange A, Van Melckebeke H, Siemer AB, Riek R, Meier BH. Amyloid fibrils of the HET-s (218–289) prion form a beta solenoid with a triangular hydrophobic core. Science. 2008; 319 (5869):1523–6. Epub 2008/03/15. https://doi.org/10.1126/science.1151839 PMID: 18339938. 4. Vazquez-Fernandez E, Vos MR, Afanasyev P, Cebey L, Sevillano AM, Vidal E, et al. The Structural Architecture of an Infectious Mammalian Prion Using Electron Cryomicroscopy. PLoS Pathog. 2016; 12 (9):e1005835. Epub 2016/09/09. https://doi.org/10.1371/journal.ppat.1005835 PMID: 27606840; PubMed Central PMCID: PMCPMC5015997 alter our adherence to all PLoS Pathogens policies on sharing data and materials. 5. Wille H, Bian W, McDonald M, Kendall A, Colby DW, Bloch L, et al. Natural and synthetic prion structure from X-ray fiber diffraction. Proc Natl Acad Sci U S A. 2009; 106(40):16990–5. Epub 2009/10/07. https://doi.org/10.1073/pnas.0909006106 PMID: 19805070; PubMed Central PMCID: PMCPMC2761340. 6. Prusiner SB. The prion diseases. Brain Pathol. 1998; 8(3):499–513. Epub 1998/07/21. PMID: 9669700. 7. Zahn R, Liu A, Luhrs T, Riek R, von Schroetter C, Lopez Garcia F, et al. NMR solution structure of the human prion protein. Proc Natl Acad Sci U S A. 2000; 97(1):145–50. Epub 2000/01/05. https://doi.org/ 10.1073/pnas.97.1.145 PMID: 10618385; PubMed Central PMCID: PMCPMC26630. 8. Requena JR, Wille H. The structure of the infectious prion protein: experimental data and molecular models. Prion. 2014; 8(1):60–6. Epub 2014/03/04. https://doi.org/10.4161/pri.28368 PMID: 24583975. 9. Govaerts C, Wille H, Prusiner SB, Cohen FE. Evidence for assembly of prions with left-handed betahelices into trimers. Proc Natl Acad Sci U S A. 2004; 101(22):8342–7. Epub 2004/05/25. https://doi.org/ 10.1073/pnas.0402254101 PMID: 15155909; PubMed Central PMCID: PMCPMC420396. 10. Groveman BR, Dolan MA, Taubner LM, Kraus A, Wickner RB, Caughey B. Parallel in-register intermolecular beta-sheet architectures for prion-seeded prion protein (PrP) amyloids. J Biol Chem. 2014; 289 (35):24129–42. Epub 2014/07/17. https://doi.org/10.1074/jbc.M114.578344 PMID: 25028516; PubMed Central PMCID: PMCPMC4148845. 11. Baskakov IV, Katorcha E. Multifaceted Role of Sialylation in Prion Diseases. Front Neurosci. 2016; 10:358. Epub 2016/08/24. https://doi.org/10.3389/fnins.2016.00358 PMID: 27551257; PubMed Central PMCID: PMCPMC4976111. 12. Baskakov IV, Caughey B, Requena JR, Sevillano AM, Surewicz WK, Wille H. The prion 2018 round tables (I): the structure of PrP(Sc). Prion. 2019; 13(1):46–52. Epub 2019/01/17. https://doi.org/10.1080/ 19336896.2019.1569450 PMID: 30646817; PubMed Central PMCID: PMCPMC6422368. 13. Welker E, Raymond LD, Scheraga HA, Caughey B. Intramolecular versus intermolecular disulfide bonds in prion proteins. J Biol Chem. 2002; 277(36):33477–81. Epub 2002/06/26. https://doi.org/10. 1074/jbc.M204273200 PMID: 12082114. 14. Vazquez-Fernandez E, Alonso J, Pastrana MA, Ramos A, Stitz L, Vidal E, et al. Structural organization of mammalian prions as probed by limited proteolysis. PLoS One. 2012; 7(11):e50111. Epub 2012/11/ 28. https://doi.org/10.1371/journal.pone.0050111 PMID: 23185550; PubMed Central PMCID: PMCPMC3502352. 15. Sevillano AM, Fernandez-Borges N, Younas N, Wang F, S RE, Bravo S, et al. Recombinant PrPSc shares structural features with brain-derived PrPSc: Insights from limited proteolysis. PLoS Pathog. 2018; 14(1):e1006797. Epub 2018/02/01. https://doi.org/10.1371/journal.ppat.1006797 PMID: 29385212; PubMed Central PMCID: PMCPMC5809102. 16. Tiana G, Camilloni C. Ratcheted molecular-dynamics simulations identify efficiently the transition state of protein folding. J Chem Phys. 2012; 137(23):235101. Epub 2012/12/27. https://doi.org/10.1063/1. 4769085 PMID: 23267502. Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 17 / 18 17. Lindorff-Larsen K, Piana S, Palmo K, Maragakis P, Klepeis JL, Dror RO, et al. Improved side-chain torsion potentials for the Amber ff99SB protein force field. Proteins. 2010; 78(8):1950–8. Epub 2010/04/ 22. https://doi.org/10.1002/prot.22711 PMID: 20408171; PubMed Central PMCID: PMCPMC2970904. 18. S AB, Fant L, Faccioli P. Variational scheme to compute protein reaction pathways using atomistic force fields with explicit solvent. Phys Rev Lett. 2015; 114(9):098103. Epub 2015/03/21. https://doi.org/ 10.1103/PhysRevLett.114.098103 PMID: 25793854. 19. Ianeselli A, Orioli S, Spagnolli G, Faccioli P, Cupellini L, Jurinovich S, et al. Atomic Detail of Protein Folding Revealed by an Ab Initio Reappraisal of Circular Dichroism. J Am Chem Soc. 2018; 140 (10):3674–82. Epub 2018/02/24. https://doi.org/10.1021/jacs.7b12399 PMID: 29473417. 20. Bartolucci G, Orioli S, Faccioli P. Transition path theory from biased simulations. J Chem Phys. 2018; 149(7):072336. Epub 2018/08/24. https://doi.org/10.1063/1.5027253 PMID: 30134709. 21. Nicholson EM, Mo H, Prusiner SB, Cohen FE, Marqusee S. Differences between the prion protein and its homolog Doppel: a partially structured state with implications for scrapie formation. J Mol Biol. 2002; 316(3):807–15. Epub 2002/02/28. https://doi.org/10.1006/jmbi.2001.5347 PMID: 11866533. 22. Sim VL. Prion disease: chemotherapeutic strategies. Infect Disord Drug Targets. 2012; 12(2):144–60. Epub 2012/03/17. PMID: 22420513. 23. Nonno R, Di Bari MA, Cardone F, Vaccari G, Fazzi P, Dell’Omo G, et al. Efficient transmission and characterization of Creutzfeldt-Jakob disease strains in bank voles. PLoS Pathog. 2006; 2(2):e12. Epub 2006/03/07. https://doi.org/10.1371/journal.ppat.0020012 PMID: 16518470; PubMed Central PMCID: PMCPMC1383487. 24. Chen H, Gu F, Huang Z. Improved Chou-Fasman method for protein secondary structure prediction. BMC Bioinformatics. 2006; 7 Suppl 4:S14. Epub 2007/01/16. https://doi.org/10.1186/1471-2105-7-S4- S14 PMID: 17217506; PubMed Central PMCID: PMCPMC1780123. 25. Herrmann US, Schutz AK, Shirani H, Huang D, Saban D, Nuvolone M, et al. Structure-based drug design identifies polythiophenes as antiprion compounds. Sci Transl Med. 2015; 7(299):299ra123. Epub 2015/08/08. https://doi.org/10.1126/scitranslmed.aab1923 PMID: 26246168. 26. Kajava AV, Steven AC. Beta-rolls, beta-helices, and other beta-solenoid proteins. Adv Protein Chem. 2006; 73:55–96. Epub 2006/12/28. https://doi.org/10.1016/S0065-3233(06)73003-0 PMID: 17190611. 27. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, et al. UCSF Chimera—a visualization system for exploratory research and analysis. J Comput Chem. 2004; 25(13):1605–12. Epub 2004/07/21. https://doi.org/10.1002/jcc.20084 PMID: 15264254. 28. Sali A, Blundell TL. Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol. 1993; 234(3):779–815. Epub 1993/12/05. https://doi.org/10.1006/jmbi.1993.1626 PMID: 8254673. 29. Berendsen HJC, Vanderspoel D, Vandrunen R. Gromacs—a Message-Passing Parallel Molecular- Dynamics Implementation. Comput Phys Commun. 1995; 91(1–3):43–56. https://doi.org/10.1016/ 0010-4655(95)00042-E WOS:A1995TF32200004. 30. Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of Coot. Acta Crystallogr D. 2010; 66:486–501. https://doi.org/10.1107/S0907444910007493 WOS:000275941300018. PMID: 20383002 31. Zhao YL, Wu YD. A theoretical study of beta-sheet models: is the formation of hydrogen-bond networks cooperative? J Am Chem Soc. 2002; 124(8):1570–1. Epub 2002/02/21. PMID: 11853419. 32. Kung VM, Cornilescu G, Gellman SH. Impact of Strand Number on Parallel beta-Sheet Stability. Angew Chem Int Ed Engl. 2015; 54(48):14336–9. Epub 2015/10/13. https://doi.org/10.1002/anie.201506448 PMID: 26457984; PubMed Central PMCID: PMCPMC4715541. 33. Moroncini G, Kanu N, Solforosi L, Abalos G, Telling GC, Head M, et al. Motif-grafted antibodies containing the replicative interface of cellular PrP are specific for PrPSc. Proc Natl Acad Sci U S A. 2004; 101 (28):10404–9. Epub 2004/07/09. https://doi.org/10.1073/pnas.0403522101 PMID: 15240877; PubMed Central PMCID: PMCPMC478584. 34. Bonomi M, Branduardi D, Bussi G, Camilloni C, Provasi D, Raiteri P, et al. PLUMED: A portable plugin for free-energy calculations with molecular dynamics. Comput Phys Commun. 2009; 180(10):1961–72. https://doi.org/10.1016/j.cpc.2009.05.011 WOS:000270628200024. 35. Friedman R, Caflisch A. Wild type and mutants of the HET-s(218–289) prion show different flexibility at fibrillar ends: a simulation study. Proteins. 2014; 82(3):399–404. Epub 2013/09/17. https://doi.org/10. 1002/prot.24402 PMID: 24038616. 36. Mizuno N, Baxa U, Steven AC. Structural dependence of HET-s amyloid fibril infectivity assessed by cryoelectron microscopy. Proc Natl Acad Sci U S A. 2011; 108(8):3252–7. Epub 2011/02/09. https://doi. org/10.1073/pnas.1011342108 PMID: 21300906; PubMed Central PMCID: PMCPMC3044374. Full atomistic model of PrPSc PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1007864 July 11, 2019 18 / 18