Heterotypic amyloid interactions: Clues to polymorphic bias and selective cellular vulnerability?

TIDGN: A Transfer Learning Framework for Predicting Interactions of Intrinsically Disordered Proteins with High Conformational Dynamics

Interactions between intrinsically disordered proteins (IDPs) are crucial for biological processes, such as intracellular liquid-liquid phase separation (LLPS). Experiments (e.g., NMR) and simulations used to study IDP interactions encounter a variety of difficulties, highlighting the necessity to develop relevant machine learning methods. However, reliable machine learning methods face the challenge resulting from the scarcity of available training data. In this work, we propose a transfer learning-based invariant geometric dynamic graph model, named TIDGN, for predicting IDP interactions. The model consists of a pretraining task module and a downstream task module. The pretraining task module learns the dynamic structural encoding of IDP monomers, which is then used by the downstream task module for interaction site prediction. The IDP monomer structure data set and the IDP interaction event data set are constructed using all-atom molecular dynamics (MD) simulations. The transfer learning strategy effectively enhances the model's performance. Both homotypic interactions and heterotypic interactions between two IDPs are considered in this work. Interestingly, TIDGN performs well for the heterotypic interaction prediction. Additionally, the feature ablation analysis emphasizes the importance of invariant geometric graph features. Taken together, our work demonstrates that the integration of transfer learning and the invariant geometric graph network offers a promising approach for addressing data scarcity challenges of IDP interaction prediction.

Interactions between pathological and functional amyloid: A match made in Heaven or Hell?

The amyloid state of proteins occurs in many different contexts in Nature and in modern society, ranging from the pathological kind (neurodegenerative diseases and amyloidosis) via man-made forms (food processing and - to a much smaller extent - protein biologics) to functional versions (bacterial biofilm, peptide hormones and signal transmission). These classes all come together in the human body which endogenously produces amyloidogenic protein able to form pathological human amyloid (PaHA), hosts a microbiome which continuously makes functional bacterial amyloid (FuBA) and ingests food which can contain amyloid. This can have grave consequences, given that PaHA can spread throughout the body in a "hand-me-down" fashion from cell to cell through small amyloid fragments, which can kick-start growth of new amyloid wherever they encounter monomeric amyloid precursors. Amyloid proteins can also self- and cross-seed across dissimilar peptide sequences. While it is very unlikely that ingested amyloid plays a role in this crosstalk, FuBA-PaHA interactions are increasingly implicated in vivo amyloid propagation. We are now in a position to understand the structural and bioinformatic basis for this cross-talk, thanks to the very recently obtained atomic-level structures of the two major FuBAs CsgA (E. coli) and FapC (Pseudomonas). While there are many reports of homology-driven heterotypic interactions between different PaHA, the human proteome does not harbor significant homology to CsgA and FapC. Yet we and others have uncovered significant cross-stimulation (and in some cases inhibition) of FuBA and PaHA both in vitro and in vivo, which we here rationalize based on structure and sequence. These interactions have important consequences for the transmission and development of neurodegenerative diseases, not least because FuBA and PaHA can come into contact via the gut-brain interface, recurrent infections with microbes and potentially even through invasive biofilm in the brain. Whether FuBA and PaHA first interact in the gut or the brain, they can both stimulate and block each other's aggregation as well as trigger inflammatory responses. The microbiome may also affect amyloidogenesis in other ways, e.g. through their own chaperones which recognize and block growth of both PaHA and FuBA as we show both experimentally and computationally. Heterotypic interactions between and within PaHA and FuBA both in vitro and in vivo are a vital part of the amyloid phenomenon and constitute a vibrant and exciting frontier for future research.

Structural commonalities determined by physicochemical principles in the complex polymorphism of the amyloid state of proteins

Advances in solid-state nuclear magnetic resonance (ssNMR) spectroscopy and cryogenic electron microscopy (cryoEM) have revealed the polymorphic nature of the amyloid state of proteins. Given the association of amyloid with protein misfolding disorders, it is important to understand the principles underlying this polymorphism. To address this problem, we combined computational tools to predict the specific regions of the sequence forming the β-spine of amyloid fibrils with the availability of 30, 83 and 24 amyloid structures deposited in the Protein Data Bank (PDB) and Amyloid Atlas (AAt) for the amyloid β (Aβ) peptide, α-synuclein (αS), and the 4R isoforms of tau, associated with Alzheimer's disease, Parkinson's disease, and various tauopathies, respectively. This approach enabled a statistical analysis of sequences forming β-sheet regions in amyloid polymorphs. We computed for any given sequence residue n the fraction of PDB/AAt structures in which that residue adopts a β-sheet conformation (Fβ(n)) to generate an experimental, structure-based profile of Fβ(n) vs n, which represents the β-conformational preference of any residue in the amyloid state. The peaks in the respective Fβ(n) profiles of the three proteins, corresponding to sequence regions adopting more frequently the β-sheet structural core in the various fibrillar structures, align very well with the peaks identified with five predictive algorithms (ZYGGREGATOR, TANGO, PASTA, AGGRESCAN, and WALTZ). These results indicate that, despite amyloid polymorphism, sequence regions most often forming the structural core of amyloid have high hydrophobicity, high intrinsic β-sheet propensity and low electrostatic charge across the sequence, as rationalised and predicted by the algorithms.

A core proteome profile unites mouse models and patients in Alzheimer disease

Levites et al. demonstrate that mouse models of Alzheimer disease (AD), exhibiting amyloid-beta (Αβ) plaque formation, share Αβ responsome proteins with humans. Their work underscores the value of these models in studying Αβ aggregation, cellular vulnerability, and early-stage AD pathology.

Amyloid oligomers and their membrane toxicity - A perspective study

Amyloidosis is a condition involving a disparate group of pathologies characterized by the extracellular deposition of insoluble fibrils composed of broken-down proteins. These proteins can accumulate locally, causing peculiar symptoms, or in a widespread way, involving many organs and. causing severe systemic failure. The damage that is created is related not only to the accumulation of. amyloid fibrils but above all to the precursor oligomers of the fibrils that manage to enter the cell in a very particular way. This article analyzes the current state of research related to the entry of these oligomers into the cell membrane and the theories related to their toxicity. The paper proposed here not only aims to review the contents in the literature but also proposes a new vision of amyloid toxicity. that could occur in a multiphase process catalyzed by the cell membrane itself. In this process, the denaturation of the lipid bilayer is followed by the stabilization of a pore through energetically favorable self-assembly processes which are achieved through particular oligomeric structures.

Local structural preferences in shaping tau amyloid polymorphism

Tauopathies encompass a group of neurodegenerative disorders characterised by diverse tau amyloid fibril structures. The persistence of polymorphism across tauopathies suggests that distinct pathological conditions dictate the adopted polymorph for each disease. However, the extent to which intrinsic structural tendencies of tau amyloid cores contribute to fibril polymorphism remains uncertain. Using a combination of experimental approaches, we here identify a new amyloidogenic motif, PAM4 (Polymorphic Amyloid Motif of Repeat 4), as a significant contributor to tau polymorphism. Calculation of per-residue contributions to the stability of the fibril cores of different pathologic tau structures suggests that PAM4 plays a central role in preserving structural integrity across amyloid polymorphs. Consistent with this, cryo-EM structural analysis of fibrils formed from a synthetic PAM4 peptide shows that the sequence adopts alternative structures that closely correspond to distinct disease-associated tau strains. Furthermore, in-cell experiments revealed that PAM4 deletion hampers the cellular seeding efficiency of tau aggregates extracted from Alzheimer's disease, corticobasal degeneration, and progressive supranuclear palsy patients, underscoring PAM4's pivotal role in these tauopathies. Together, our results highlight the importance of the intrinsic structural propensity of amyloid core segments to determine the structure of tau in cells, and in propagating amyloid structures in disease.

Mechanisms and pathology of protein misfolding and aggregation

Despite advances in machine learning-based protein structure prediction, we are still far from fully understanding how proteins fold into their native conformation. The conventional notion that polypeptides fold spontaneously to their biologically active states has gradually been replaced by our understanding that cellular protein folding often requires context-dependent guidance from molecular chaperones in order to avoid misfolding. Misfolded proteins can aggregate into larger structures, such as amyloid fibrils, which perpetuate the misfolding process, creating a self-reinforcing cascade. A surge in amyloid fibril structures has deepened our comprehension of how a single polypeptide sequence can exhibit multiple amyloid conformations, known as polymorphism. The assembly of these polymorphs is not a random process but is influenced by the specific conditions and tissues in which they originate. This observation suggests that, similar to the folding of native proteins, the kinetics of pathological amyloid assembly are modulated by interactions specific to cells and tissues. Here, we review the current understanding of how intrinsic protein conformational propensities are modulated by physiological and pathological interactions in the cell to shape protein misfolding and aggregation pathology.

Disease-relevant β2-microglobulin variants share a common amyloid fold

β2-microglobulin (β2m) and its truncated variant ΔΝ6 are co-deposited in amyloid fibrils in the joints, causing the disorder dialysis-related amyloidosis (DRA). Point mutations of β2m result in diseases with distinct pathologies. β2m-D76N causes a rare systemic amyloidosis with protein deposited in the viscera in the absence of renal failure, whilst β2m-V27M is associated with renal failure, with amyloid deposits forming predominantly in the tongue. Here we use cryoEM to determine the structures of fibrils formed from these variants under identical conditions in vitro. We show that each fibril sample is polymorphic, with diversity arising from a 'lego-like' assembly of a common amyloid building block. These results suggest a 'many sequences, one amyloid fold' paradigm in contrast with the recently reported 'one sequence, many amyloid folds' behaviour of intrinsically disordered proteins such as tau and Aβ.

Distinct neurotoxic TDP-43 fibril polymorphs are generated by heterotypic interactions with α-Synuclein

Amyloid aggregates of specific proteins constitute important pathological hallmarks in many neurodegenerative diseases, defining neuronal degeneration and disease onset. Recently, increasing numbers of patients show comorbidities and overlaps between multiple neurodegenerative diseases, presenting distinct phenotypes. Such overlaps are often accompanied by colocalizations of more than one amyloid protein, prompting the question of whether direct interactions between different amyloid proteins could generate heterotypic amyloids. To answer this question, we investigated the effect of α-synuclein (αS) on the DNA-binding protein TDP-43 aggregation inspired by their coexistence in pathologies such as Lewy body dementia and limbic predominant age-related TDP-43 encephalopathy. We previously showed αS and prion-like C-terminal domain (PrLD) of TDP-43 synergistically interact to generate toxic heterotypic aggregates. Here, we extend these studies to investigate whether αS induces structurally and functionally distinct polymorphs of PrLD aggregates. Using αS-PrLD heterotypic aggregates generated in two different stoichiometric proportions, we show αS can affect PrLD fibril forms. PrLD fibrils show distinctive residue level signatures determined by solid state NMR, dye-binding capability, proteinase K (PK) stability, and thermal stability toward SDS denaturation. Furthremore, by gold nanoparticle labeling and transmission electron microscopy, we show the presence of both αS and PrLD proteins within the same fibrils, confirming the existence of heterotypic amyloid fibrils. We also observe αS and PrLD colocalize in the cytosol of neuroblastoma cells and show that the heterotypic PrLD fibrils selectively induce synaptic dysfunction in primary neurons. These findings establish the existence of heterotypic amyloid and provide a molecular basis for the observed overlap between synucleinopathies and TDP-43 proteinopathies.