A strategy on prion AGAAAAGA amyloid fibril molecular modelling

X-ray crystallography and nuclear magnetic resonance (NMR) spectroscopy are two powerful tools to determine the protein 3D structure. However, not all proteins can be successfully crystallized, particularly for membrane proteins. Although NMR spectroscopy is indeed very powerful in determining the 3D structures of membrane proteins, same as X-ray crystallography, it is still very time-consuming and expensive. Under many circumstances, due to the noncrystalline and insoluble nature of some proteins, X-ray and NMR cannot be used at all. Computational approaches, however, allow us to obtain a description of the protein 3D structure at a submicroscopic level. To the best of the authors’ knowledge, there is little structural data available to date on the AGAAAAGA palindrome in the hydrophobic region (113-120) of prion proteins, which falls just within the N-terminal unstructured region (1-123) of prion proteins. Many experimental studies have shown that the AGAAAAGA region has amyloid fibril forming properties and plays an important role in prion diseases. Due to the noncrystalline and insoluble nature of the amyloid fibril, little structural data on the AGAAAAGA is available. This paper introduces a simple molecular modelling strategy to address the 3D atomic-resolution structure of prion AGAAAAGA amyloid fibrils. Atomic-resolution structures of prion AGAAAAGA amyloid fibrils got in this paper are useful for the drive to find treatments for prion diseases in the field of medicinal chemistry.

💡 Research Summary

**
Prion diseases are fatal neurodegenerative disorders characterized by the conversion of the normal cellular prion protein (PrP^C), which is rich in α‑helices, into the pathogenic isoform (PrP^Sc) that is dominated by β‑sheets. The hydrophobic core of the prion protein, especially the palindrome AGAAAAGA (residues 113‑120), has been repeatedly identified in experimental studies as a critical segment for amyloid fibril formation and disease propagation. However, because this segment is intrinsically non‑crystalline and insoluble, traditional high‑resolution structural techniques such as X‑ray crystallography and nuclear magnetic resonance (NMR) cannot be applied.

The present paper proposes a purely computational strategy to obtain atomic‑resolution models of AGAAAAGA amyloid fibrils. The authors start from the high‑resolution crystal structure of a human M129 prion peptide (residues 127‑132) deposited in the Protein Data Bank as entry 3NHC. This structure is a classic steric zipper: two β‑sheets are tightly interlocked by strong van der Waals (vdW) contacts and a network of inter‑strand hydrogen bonds. Because the steric zipper motif is a generic scaffold for many amyloid fibrils, the authors use it as a template to model the AGAAAAGA segment.

The modeling workflow consists of four main steps. First, using Swiss‑PDB Viewer (SPDBV), the authors mutate the AB chains of 3NHC to the AGAAAAGA sequence, generating an initial four‑chain assembly (ABGH). Second, they observe that in this initial configuration the inter‑sheet vdW distances are far larger than those required for a realistic fibril, indicating that the model is not yet physically plausible. Third, they fix the coordinates of the side‑chain Cβ atoms of A‑chain (ALA3) and B‑chain (ALA4) and treat the corresponding Cβ atoms of G‑chain (ALA4) and H‑chain (ALA3) as variables. This reduces the problem to a six‑dimensional optimization (three Cartesian coordinates for each of the two movable atoms). The interaction energy is described by a Lennard‑Jones (LJ) potential, which in reduced units (ε = σ = 1) becomes a simple mathematical function f(x) that can be minimized over ℝ⁶.

To solve this low‑dimensional global optimization problem, the authors employ Simulated Annealing Evolutionary Computation (SAEC), a hybrid algorithm that combines the temperature‑controlled exploration of simulated annealing with the crossover and mutation operators of genetic algorithms. SAEC has been shown to succeed on a wide range of benchmark global‑optimization tasks, and here it efficiently drives the six variables toward a configuration where the vdW contacts between A‑G and B‑H chains become optimal. The resulting coordinates are reported in Equation 9 and visualized in Figures 6‑8, where the two sheets are now closely packed.

After achieving a satisfactory vdW arrangement, the authors expand the model to a full twelve‑chain fibril by replicating the ABGH unit using the symmetry operations described in Equations 6‑8 (producing CD, IJ, EK, FL chains). The complete assembly is then subjected to a conventional molecular‑mechanics refinement using the Amber 11 force field. This step performs energy minimization and a short molecular dynamics relaxation, allowing all atoms to adjust while preserving the steric‑zipper backbone. The final structures—Model 1 (AAAA GA), Model 2 (GAAAAG), and Model 3 (AAAAGA)—are displayed in Figures 9‑11 and exhibit stable total potential energies, robust hydrogen‑bond networks, and realistic inter‑sheet vdW contacts.

The paper’s conclusions emphasize three scientific points. First, a complex amyloid fibril can be modeled accurately by reducing the problem to a small set of key geometric variables, demonstrating the power of dimensionality reduction in protein‑folding calculations. Second, hybrid global‑optimization techniques such as SAEC are effective for systems where experimental data are unavailable, offering a practical route to explore the conformational space of intrinsically disordered or insoluble proteins. Third, the atomic‑level models of the AGAAAAGA fibril provide a structural platform for rational drug design; small molecules or peptide inhibitors that target this hydrophobic core could be screened or designed using the presented structures.

Overall, the study showcases how computational chemistry, combined with clever mathematical formulation and modern optimization algorithms, can fill the gap left by experimental methods in the structural characterization of disease‑relevant amyloid fibrils. The resulting models not only advance our understanding of prion fibrillogenesis but also open new avenues for therapeutic intervention in prion diseases.