Structural polymorphism and cytotoxicity of brain‐derived β‐amyloid extracts

Abstract

To date, more than 37 amyloidogenic proteins have been found to form toxic aggregates that are implicated in the progression of numerous debilitating protein misfolding diseases including Alzheimer's disease (AD). Extensive literature highlights the role of β‐amyloid (Aβ) aggregates in causing excessive neuronal cell loss in the brains of AD patients. In fact, major advances in our understanding of Aβ aggregation process, including kinetics, toxicity, and structures of fibrillar aggregates have been revealed by examining in vitro preparations of synthetic Aβ peptides. However, ongoing research shows that brain‐derived Aβ aggregates have specific characteristics that distinguish them from in vitro prepared species. Notably, the molecular structures of amyloid fibrils grown in the human brain were found to be markedly different than synthetic Aβ fibrils. In addition, recent findings report the existence of heterogeneous Aβ proteoforms in AD brain tissue in contrast to synthetically produced full‐length aggregates. Despite their high relevance to AD progression, brain‐derived Aβ species are less well‐characterized compared with synthetic aggregates. The aim of this review is to provide an overview of the literature on brain‐derived Aβ aggregates with particular focus on recent studies that report their structures as well as pathological roles in AD progression. The main motivation of this review is to highlight the importance of utilizing brain‐derived amyloids for characterizing the structural and toxic effects of amyloid species. With this knowledge, brain‐derived aggregates can be adopted to identify more relevant drug targets and validate potent aggregation inhibitors toward designing highly effective therapeutic strategies against AD.

1. INTRODUCTION

The misfolding and aggregation of amyloidogenic proteins characterize the progression of numerous protein misfolding diseases including the highly prevalent Alzheimer's disease (AD) (Chiti & Dobson, 2017; Dobson, ²⁰⁰⁴; Knowles et al., ²⁰¹⁴; Lee et al., ²⁰²⁰). AD is an irreversible and progressive neurodegenerative disorder that accounts for the majority of dementia cases (60%–70%) (DeTure & Dickson, 2019; Prince et al., ²⁰¹⁶). AD is clinically described by the progressive deterioration of cognitive functions and the gradual loss of memory which eventually interfere with many bodily functions (DeTure & Dickson, 2019; Jack et al., ²⁰¹¹). Neuropathologically, AD is characterized by neuronal cell loss due to two cellular hallmarks that were identified in the brains of AD patients, including the extracellular accumulation of β‐Amyloid (Aβ) peptide in the form of senile plaques, and the intracellular accumulation of hyper‐phosphorylated tau protein in the form of neurofibrillary tangles (Anand et al., 2014; Jack et al., ²⁰¹⁸; Serrano‐Pozo et al., ²⁰¹¹). Studies have shown that Aβ monomers are produced in healthy brains at low concentrations where they hold potential physiological functions in maintaining neuronal activity and synaptic plasticity (Abramov et al., 2009; Giuffrida et al., ²⁰⁰⁹; Puzzo et al., ²⁰¹⁵). However, excessive Aβ levels detected in AD patients induce the conversion of Aβ monomers into cytotoxic oligomers and fibrils (Puzzo et al., 2015; Resende et al., ²⁰⁰⁸). In fact, Aβ oligomers have been shown to play central roles in inducing neurotoxicity and synaptic loss in AD pathology (Carrillo‐Mora et al., 2014; Huang & Liu, ²⁰²⁰; Kayed & Lasagna‐Reeves, ²⁰¹³).

In vitro aggregation of synthetic proteins has been extensively examined in literature with the aim of elucidating the pathways of protein aggregation (Chiti & Dobson, 2017). Such in vitro systems provide controlled conditions for monitoring and characterizing the formation of aggregated species (Villar‐Piqué et al., 2018). In fact, using in vitro aggregation models enabled substantial advances in understanding key aspects of the synthetic Aβ aggregation mechanism including kinetics of self‐assembly pathways (S. I. A. Cohen et al., ²⁰¹³; Knowles et al., ²⁰⁰⁹), structure of synthetic Aβ aggregates (Ciudad et al., 2020; Colvin et al., ²⁰¹⁶; Gremer et al., ²⁰¹⁷; Hu et al., ²⁰¹⁹; Lührs et al., ²⁰⁰⁵; Paravastu et al., ²⁰⁰⁸; Qiang et al., ²⁰¹²; Schmidt et al., ²⁰¹⁵; Schütz et al., ²⁰¹⁵; Wälti et al., ²⁰¹⁶; Xiao et al., ²⁰¹⁵) and the cytotoxicity of oligomeric and fibrillar species (Bemporad & Chiti, 2012; Campioni et al., ²⁰¹⁰; Cecchi & Stefani, ²⁰¹³; De et al., ²⁰¹⁹; Gonzalez‐Velasquez et al., ²⁰⁰⁸; Hoshi et al., ²⁰⁰³; Lorenzo & Yankner, ¹⁹⁹⁶; Mannini et al., ²⁰¹⁴; Ono et al., ²⁰⁰⁹; Sakono & Zako, ²⁰¹⁰). Although in vitro parameters such as temperature, ionic strength, and pH can be tuned to resemble the in vivo environment, it is highly challenging to accurately mimic the complexity of the cellular microenvironment that, in turn, has profound effects on the actual in vivo aggregation process (Brody et al., 2017; Brody & Gross, ²⁰¹⁴; Li & Stern, ²⁰²²).

In vitro synthetic fibrils have been shown by several studies to have polymorphic structures that are dependent on growth conditions rather than specific amino acid sequences (Diaz‐Espinoza, 2021; Fändrich et al., ²⁰⁰⁹; Petkova et al., ²⁰⁰⁵; Riek, ²⁰¹⁷; Wu et al., ²⁰¹⁰). These findings suggest that the molecular structure of amyloid species grown in the human brain cannot be determined by in vitro synthetic fibrils or fibrils extracted from the brains of transgenic animals (Brody & Gross, 2014). Previous studies reported that brain‐derived fibrils extracted from AD patients are morphologically distinct than purely synthetic fibrils (Lu et al., 2013; Paravastu et al., ²⁰⁰⁹). Particularly, the handedness of the fibrils' twist, key residue distances, residue contacts, and protofilament arrangements are reported to vary between synthetic and brain‐derived fibrils (Kollmer et al., 2019; Lu et al., ²⁰¹³; Paravastu et al., ²⁰⁰⁸; Paravastu et al., ²⁰⁰⁹).

Ongoing research has shown that brain‐derived amyloids have specific characteristics that distinguish them from in vitro‐prepared fibrils. Aβ aggregates derived from brain‐extracts have higher resistance to proteinase K as compared to in vitro synthetic Aβ aggregates (Langer et al., 2011). Furthermore, recent findings suggest that brain‐derived Aβ oligomers have higher neurotoxicity than synthetic oligomers (Brody & Gross, 2014; Li & Stern, ²⁰²²). Importantly, the analysis of Aβ proteoforms present in AD brain tissues reveals the diversity of Aβ post‐translational modifications, particularly N‐terminal and C‐terminal truncations that are absent from synthetic full‐length Aβ peptides (Arai et al., 1999; Cabrera et al., ²⁰¹⁸; Güntert et al., ²⁰⁰⁶; Kummer & Heneka, ²⁰¹⁴; Milton, ²⁰⁰¹; Rijal Upadhaya et al., ²⁰¹⁴; Rostagno et al., ²⁰¹⁸; Sergeant et al., ²⁰⁰³; Wildburger et al., ²⁰¹⁷).

Despite their high relevance to AD progression, brain‐derived Aβ species are less characterized as compared to synthetic aggregates. In this work, we aim to review the literature on brain‐derived Aβ aggregates that have been purified from cerebral cortical tissues or meningeal tissues of deceased clinicopathologically confirmed AD patients. Next, we discuss the polymorphic features of synthetic and brain‐derived fibrils. Given the well‐known pathological roles of Aβ oligomers, we focus on literature findings that describe the neurotoxic roles of natural brain‐derived oligomers which have been extracted at autopsy from brain tissues of confirmed AD patients. Particularly, we highlight studies that utilized in vitro systems for examining the pathological roles of AD brain‐derived Aβ oligomers and dimers.

2. PROTEIN AGGREGATION IN HUMAN DISEASES

The high complexity of the cellular environment, which is crowded with a high concentration (300–400 mg/mL) of various proteins and other macromolecules, necessitates key assistive factors to maintain proteome homeostasis (proteostasis) (Balchin et al., 2016; Dobson, ²⁰⁰⁴; Vendruscolo, ²⁰²²). Molecular chaperones and folding catalysts assist globular proteins to achieve their native states (Balchin et al., 2016; Dobson, ²⁰⁰⁴; Knowles et al., ²⁰¹⁴; Lee, ²⁰⁰⁵). In addition, tight regulations of the synthesis, stability, diversity, and degradation of intrinsically disordered proteins (IDPs) is essential for controlling their self‐ and non‐self‐interactions and preventing their potential pathological effects (Gsponer et al., 2008; Kundra et al., ²⁰¹⁷; Theillet et al., ²⁰¹⁴). Proteostasis is considered quite robust as it tightly regulates protein synthesis, protein turnover, the clearance of misfolded proteins, and the stress‐induced cellular responses (Balchin et al., 2016). Unsurprisingly, the failure of a protein to fold correctly and/or to be cleared if it gets misfolded result in cellular malfunction and eventually lead to disease progression (Dobson, 2004).

The exact physiological causes of protein aggregation and the cellular events that lead to their accumulation are not yet fully understood. However, aging and stress conditions, both genetic and environmental, are key factors that challenge the cellular capacity to maintain proteostasis. As a result, misfolded proteins may escape the cellular quality control checks and transform into intracellular or extracellular intractable aggregates giving rise to a wide range of protein misfolding diseases (Balchin et al., 2016; Dobson, ²⁰⁰⁴; Hipp et al., ²⁰¹⁴; Vendruscolo, ²⁰¹²). Particularly, aging is associated with decreased proteasomal and autophagy functions, which is considered a key risk factor for the development of numerous protein misfolding diseases (Balchin et al., 2016; Yacoubian TA Neurodegenerative Disorders, ²⁰¹⁷). Aging and/or stress conditions could lead to different in vivo events that promote the aggregation of proteins and peptides. These in vivo events include denaturation of a natively folded protein, overexpression of a certain protein that overwhelms proteostasis, peptide fragmentation, and overproduction of a natively misfolded peptide (Eisenberg & Jucker, 2012). In fact, Aβ aggregation results from the fragmentation of its large amyloid precursor protein (APP) (Dobson, 2004). The resulting peptide fragments may not be able to fold correctly in the absence of their precursor proteins which makes them vulnerable to self‐interaction and aggregation (Dobson, 2004).

3. EXTRACTION METHODS AND PROPERTIES OF BRAIN‐DERIVED Aβ AGGREGATES

Amyloid formation in the brains of individuals developing sporadic AD is believed to accumulate for more than a decade before the clinical diagnosis of AD symptoms (DeTure & Dickson, 2019; Dubois et al., ²⁰¹⁶). The senile plaques, containing extracellular Aβ deposits, are found to populate different regions of AD brain patients but their presence with high density is typically observed in the temporal lobe, particularly hippocampus and entorhinal cortex (DeTure & Dickson, 2019; Sideris et al., ²⁰²¹; Smith, ²⁰⁰²). In addition to brain parenchyma deposition, Aβ aggregates are observed to accumulate in the cortical blood vessels with an estimated 85%–95% of AD cases having cerebral amyloid deposition (Attems et al., 2010; DeTure & Dickson, ²⁰¹⁹).

Aβ deposits in brain tissues constitute both oligomeric and fibrillar species that can be experimentally extracted using appropriate extraction protocols. The soluble and insoluble brain extracts can be subjected to specific antibody treatment that differentiate different aggregate species depending on their specific conformations. It has been shown that Aβ oligomers acting as kinetic intermediates during the fibrillation process are recognized by A11 antibody (Kayed et al., 2003) whereas small fibrillar fragments that may act as seeds can be recognized by fibril‐specific OC antibody (Fritschi, Langer, et al., 2014).

The isolation of insoluble Aβ fibrils from senile plaques of cortical brain tissues, including the temporal, frontal, occipital or parietal regions was described by several papers (Fritschi, Langer, et al., 2014; Lu et al., ²⁰¹³; Noguchi et al., ²⁰⁰⁹; Paravastu et al., ²⁰⁰⁹; Pedrero‐Prieto et al., ²⁰¹⁹; Qiang et al., ²⁰¹⁷; Roher & Kuo, ¹⁹⁹⁹; Rostagno et al., ²⁰¹⁸; Shankar et al., ²⁰⁰⁸). Although protocols have slight variations, the overall process is comparable as schematically summarized in Figure 1. Briefly, freshly frozen brain samples of the cerebral cortical tissues are minced and homogenized in an extraction buffer containing protease inhibitors. Next, centrifugation is performed to remove insoluble myelin, cellular debris, and an intermediate soluble supernatant; thus, leaving a pellet which contains the insoluble Aβ aggregates. The pellet can then be enzymatically hydrolyzed with collagenase and DNase I to remove contaminating collagen and DNA. Next, pellet is treated with detergent containing solutions to yield an amyloid‐enriched material that is further subjected in some studies to formic acid treatment to produce solubilized amyloid material (Fritschi, Langer, et al., 2014; Lu et al., ²⁰¹³; Noguchi et al., ²⁰⁰⁹; Paravastu et al., ²⁰⁰⁹; Pedrero‐Prieto et al., ²⁰¹⁹; Qiang et al., ²⁰¹⁷; Roher & Kuo, ¹⁹⁹⁹; Rostagno et al., ²⁰¹⁸; Shankar et al., ²⁰⁰⁸).

FIGURE 1.

Schematic illustration of the overall extraction process of Aβ aggregates from brain tissues of deceased AD patients. Freshly frozen cortical brain samples are homogenized/soaked with extraction buffer and then centrifuged to obtain a soluble brain fraction supernatant and a mixture of insoluble brain fraction pellet. The insoluble pellet (1) is subjected to detergent‐containing buffer and/or formic acid to obtain solubilized amyloid‐enriched fibrils. On the other hand, the soluble brain fraction supernatant (2) is further subjected to ultracentrifugation and/or immunoprecipitation to selectively purify soluble Aβ oligomers and dimers.

The extraction of Aβ fibrils from vascular amyloid deposits of meningeal tissues was described by one paper that followed a fibril‐based water‐extraction protocol (Kollmer et al., 2019). Briefly, preliminary treatment and washing of tissue material with Tris calcium buffer is performed to remove soluble proteins. Next, an overnight collagenase treatment is applied and the resulting pellet is treated with 10 homogenization steps with Tris‐EDTA buffer where in each step, the supernatant is removed carefully and the resulting pellet is resuspended in Tris‐EDTA buffer. Next, 10 water‐based homogenization steps are performed to obtain the supernatant containing soluble amyloids in distilled water (Kollmer et al., 2019; Schubert et al., ¹⁹⁶⁸).

In addition to the purification of insoluble amyloid fibrils, several studies described the purification of soluble AD brain extracts consisting of Aβ dimers (Brinkmalm et al., 2019; Brody & Gross, ²⁰¹⁴; Jin et al., ²⁰¹¹; Lesné et al., ²⁰¹³; Müller‐Schiffmann et al., ²⁰¹⁶; Shankar et al., ²⁰⁰⁸; T. Yang et al., ²⁰¹⁷) and small Aβ oligomers (Esparza et al., 2016; Fritschi, Langer, et al., ²⁰¹⁴; Hong et al., ²⁰¹⁸; Lesné et al., ²⁰¹³; Li et al., ²⁰¹¹; Li & Stern, ²⁰²²; Noguchi et al., ²⁰⁰⁹; Sherman & Lesné, ²⁰¹⁰; Sideris et al., ²⁰²¹). The studies above performed the extraction of soluble brain fractions from supernatants retrieved by ultracentrifugation of AD brain tissues that were homogenized or soaked with an extraction buffer containing protease inhibitors.

Literature findings suggest the high stability and resistance properties of brain‐derived Aβ aggregates. Particularly, formaldehyde treatment of brain‐derived Aβ aggregates did not lead to their inactivation but slightly reduced their seeding activity in vivo (Fritschi, Cintron, et al., 2014). Moreover, brain‐derived Aβ aggregates have high resistance to proteinase K as compared to in vitro synthetic aggregates. A study showed that ultracentrifugation of brain‐extracts resulted in less than 0.05% of soluble Aβ species that had high in vivo seeding activity and Aβ deposition upon their intracerebral injection in APP transgenic mice (Langer et al., 2011). Another distinctive property of brain‐derived Aβ oligomers is their high binding affinity to calcium which enabled their purification using a specific calcium‐sensitive human monoclonal antibody (Stern et al., 2022). Interestingly, the reported Ca²⁺‐specific monoclonal antibody enabled the purification of amyloid fibrils in addition to soluble Aβ oligomers from AD cortical brain tissues (Stern et al., 2022).

The analysis of Aβ proteoforms present in AD brain tissues reveals the diversity of Aβ post‐translational modifications, particularly N‐terminal and C‐terminal truncations (Arai et al., 1999; Cabrera et al., ²⁰¹⁸; Güntert et al., ²⁰⁰⁶; Kummer & Heneka, ²⁰¹⁴; Milton, ²⁰⁰¹; Rijal Upadhaya et al., ²⁰¹⁴; Rostagno et al., ²⁰¹⁸; Sergeant et al., ²⁰⁰³; Wildburger et al., ²⁰¹⁷). Such posttranslational modifications of Aβ were found to impact the spreading, seeding, and neurotoxicity of amyloid deposits (Güntert et al., 2006; Kumar et al., ²⁰¹¹; Kumar et al., ²⁰¹²; Kummer & Heneka, ²⁰¹⁴; Rezaei‐Ghaleh et al., ²⁰¹⁶; Rijal Upadhaya et al., ²⁰¹⁴; Sergeant et al., ²⁰⁰³). N‐terminal truncations were found to be highly prevalent in AD brain extracts with an increase of 20% between Braak stages IV–VI (Güntert et al., 2006; Zampar et al., ²⁰²⁰). Several studies have used immunoprecipitation and mass spectrometry experiments to compare the prevalence of N‐terminal and C‐terminal truncated Aβ species between soluble and insoluble AD brain extracts. N‐terminal truncations were found to constitute 73% of the total identified Aβ proteoforms as opposed to C‐terminal truncations that had a lower prevalence ratio of 30% (Wildburger et al., 2017). In addition, N‐terminal truncations showed high aggregation propensities and were mainly purified from insoluble brain fractions unlike C‐terminal truncations that were mainly purified from soluble brain fractions (Cabrera et al., 2018; Rostagno et al., ²⁰¹⁸; Wildburger et al., ²⁰¹⁷; Zampar et al., ²⁰²⁰).

4. STRUCTURAL POLYMORPHISM OF SYNTHETIC AND BRAIN‐DERIVED Aβ FIBRILS

Fibrils formed by the self‐assembly of misfolded polypeptides share a common molecular subunit known as the cross‐β structure which consists of a repetitive two‐layered intermolecular β‐sheet motif (Diaz‐Espinoza, 2021). The pleated β‐sheets in the fibril core form by intermolecular hydrogen bonds that align parallel to the fibril axis and connect juxtaposed β‐strands of the misfolded polypeptide monomers (Diaz‐Espinoza, 2021; Fändrich et al., ²⁰⁰⁹; Pedersen et al., ²⁰¹⁰; Riek, ²⁰¹⁷; Wu et al., ²⁰¹⁰). The side chains of the self‐assembled polypeptides extend perpendicular to the fibril axis and adopt various intra‐ and inter‐molecular interactions within the fibril (Fändrich et al., 2009). Different polypeptide sequences give rise to variations in the side‐chain spacing which defines the plane of the fibril cross‐section.

In the process of native protein folding, a given protein adopts a unique functional 3D conformation that consists of the same inter‐residue interactions. However, the process of fibril formation is different as the same protein can adopt different inter‐residue interactions or conformations during its self‐assembly giving rise to polymorphic fibrillar structures (Fändrich et al., 2009). In addition, polymorphism in fibrils can be described by other factors including fibril diameter, mass‐per‐length data, amount of fibrillar twisting, and protofilament number. Particularly, the protofilaments that form the substructure of mature amyloid fibrils can have different helical twisting which give rise to polymorphic fibrillar morphologies across different proteins. Electron microscopy (EM) can measure the fibril width and cross‐over distances which vary by the different arrangements of protofilaments (Fändrich et al., 2009; Petkova et al., ²⁰⁰⁵).

In fact, it has been experimentally shown in the literature that Aβ‐fibril polymorphism has a variable effect on neurodegenerative progression in AD (Condello et al., 2018; Tycko, ²⁰¹⁵). An early study showed that different Aβ fibril morphologies are dependent on subtle variations in the fibril growth conditions. The study reported different toxic effect of polymorphic Aβ fibril morphologies on primary rat embryonic hippocampal neurons with Aβ fibrils grown under quiescent conditions having higher toxicity than those grown under agitated conditions (Petkova et al., 2005). At the in vivo level, studies showed differential neuropathological effects in transgenic mouse models injected with distinct polymorphic Aβ fibrils, including Aβ fibrils extracted from familial and sporadic AD cases, as well as synthetically produced Aβ fibrils with distinct C‐terminal variants (Ruiz‐Riquelme et al., 2021; Stöhr et al., ²⁰¹⁴; Watts et al., ²⁰¹⁴; Zhu et al., ²⁰²²). Synthetic Aβ fibrils injected into mice have been shown to be less potent in inducing Aβ deposition compared with brain‐derived fibrils (Stöhr et al., 2012). The different pathologies of AD cases might be correlated with structurally distinct polymorphic Aβ aggregates, as unique oligomeric particles have been detected in the cortical brain tissues of rapidly progressive AD cases (M. L. Cohen et al., ²⁰¹⁵; M. Cohen et al., ²⁰¹⁶). The findings above demonstrate that structural diversity of Aβ fibrils may play a central role in the progression of AD and that synthetic Aβ fibrils may not mimic and structure and toxic effect of brain‐derived Aβ aggregates.

Extensive research efforts have been devoted to elucidate the molecular structure of amyloid fibrils. Studies resolving the structures of in vitro fibrils grown from synthetic Aβ_1–40 and Aβ_1–42 reveal their polymorphic features as will be detailed in this section.

Most synthetic Aβ_1–42 fibrils were reported to have twofold symmetry as two Aβ_1–42 molecules assemble in a single fibril layer (see Figure 2, PDBs: 2NAO, 5KK3, and 5OQV). Nonetheless, the resolved Aβ_1–42 fibrillar structures have different Aβ_1–42 monomeric conformations within each fibril subunit. A study by Xiao et al. reported the structural model of synthetic Aβ_1–42 fibrils having one Aβ_1–42 molecule in each fibril layer which forms a triple parallel β‐motif (Figure 2a, PDB 2MXU) (Xiao et al., 2015). A high‐resolution structure of a disease‐relevant Aβ_1–42 amyloid polymorph was determined using solid‐state nuclear magnetic resonance (ssNMR), and electron microscopy (Figure 2b, PDB: 2NAO). The fibrils had twofold symmetry where each Aβ_1–42 molecule formed cross β‐sheets in a double horseshoe‐like morphology with residues 15–42 forming the fibril core while residues 1–14 appearing to be partially ordered as shown in Figure 2b (Wälti et al., 2016). The twofold symmetry of Aβ_1–42 fibrils observed in previous work was also reported by Colvin et al. that used magic angle spinning NMR to determine an atomic resolution structure of Aβ_1–42 monomorphic fibrils (Figure 2c, PDB: 5KK3) (Colvin et al., 2016). Each Aβ_1–42 monomer had four β‐strands that assembled in an S‐shaped fold with residues 15–42 forming the fibril core. This Aβ_1–42 fibrillar structure (PDB 5KK3) revealed the exposure of two hydrophobic residues (V40, A42) on the fibril surface which could explain the higher secondary nucleation rates observed for Aβ_1–42 in comparison to Aβ_1–40. A recent high‐resolution structure of Aβ_1–42 fibrils was resolved using Cryo‐EM complemented with ssNMR and X‐ray diffraction experiments. In line with previous papers, the fibrillar Aβ_1–42 structure (Figure 2d, PDB 5OQV) had a twofold symmetry consisting of a dimer spanning the fibril subunit. Unlike previous structures, the N‐terminus of each Aβ_1–42 monomer in PDB 5OQV was involved in the fibril cross‐β core resulting in a LS‐shaped fibrillar subunit. Also, the dimer interface is formed by the C‐termini interaction from the two Aβ_1–42 monomers unlike in previous ssNMR structures that elucidated a solvent‐exposed C‐termini (Gremer et al., 2017).

FIGURE 2.

Resolved ssNMR/Cryo‐EM structures of synthetically prepared Aβ_1–42/Aβ_1–40 fibrils. Top view representation was used to demonstrate the fibril symmetry (i.e., conformation and number of molecules per fibril layer). (a–i) Fibrillar models are generated on PyMol using the protein data bank entries and are colored based on secondary structure (yellow for β‐sheets and green for loops). The terms, positive and negative stagger, in panels f–i describe the conformation of Aβ_1–40 which spans two different z‐planes implying that each fibril layer is not occupied by a single Aβ_1–40 molecule.

While for synthetic Aβ_1–40 fibrils, a study revealed their structural polymorphism using solid state NMR coupled with electron microscopy experiments (Paravastu et al., 2008). In this study, synthetic Aβ_1–40 fibrils with untwisted (striated ribbon) and periodically twisted morphologies had twofold and threefold symmetries about their fibril axes, respectively. In fact, the two fibrillar morphologies were produced using the same in vitro conditions except that striated ribbon fibrils were produced via agitation conditions while the twisted fibrils were produced via quiescent conditions (Paravastu et al., 2008; Petkova et al., ²⁰⁰⁶). The protofilaments of striated and twisted Aβ_1–40 fibrils showed differences in the mass‐per‐length and solid‐state NMR data resulting in twofold and threefold symmetric models, respectively. However, both models had in‐register parallel β‐sheets that consist of similar β‐strands and non‐β strands segments but were mainly different in the symmetry, conformation of the non‐β strands segments and the overall quaternary structures (Paravastu et al., 2008; Petkova et al., ²⁰⁰⁶). As seen in Figure 2f–i, four structural models (PDBs 2LMN, 2LMO, 2LMP, and 2LMQ) were identified from synthetic Aβ_1–40 fibrils in this work (Paravastu et al., 2008). The notions, positive and negative stagger describe the conformation of Aβ_1–40 which spans two different z‐planes implying that each fibril layer is not occupied by a single Aβ_1–40 molecule (Chen et al., 2018).

A study by Hu et al. elucidated the structure of synthetic fibrils produced from Ser‐8‐phosphorylated Aβ_1–40 (Hu et al., 2019) (Figure 2e). The resolved structure, having a striated ribbon morphology with a twofold symmetry, is comparable to that of wild‐type Aβ_1–40 fibrils produced using similar in vitro conditions (Hu et al., 2019; Paravastu et al., ²⁰⁰⁸; Petkova et al., ²⁰⁰⁶), but with variations in the fibrillar width (6 nm for wildtype vs. 8.5 nm for Ser‐8‐phosphorylated type). The N‐terminal residues of Ser‐8‐phosphorylated Aβ_1–40 fibrils were involved in strong intra‐strand interactions with the amyloid core unlike the wildtype Aβ_1–40 fibrils that have a highly dynamic N‐terminus that is not involved in the fibrillar core. In addition, the phosphorylated fibrils are shown to possess higher cross‐seeding capacity than non‐phosphorylated Aβ fibrils.

As established earlier, studies reporting the molecular structure of synthetic fibrils, produced by the in vitro aggregation of Aβ_1–40, have shown that these structures are dependent on the growth conditions rather than the specific amino acid sequence. Hence, the molecular structure of amyloid fibrils that grow in AD patients cannot be primarily predicted by the in vitro derived synthetic amyloids.

Given that seeded growth of fibrils was shown to preserve molecular structures, a number of studies have used brain‐derived fibrils as seeds to grow in vitro labeled Aβ fibrils that preserve the molecular details of amyloids derived from human cortical tissues (Lu et al., 2013; Paravastu et al., ²⁰⁰⁹; Petkova et al., ²⁰⁰⁵; Qiang et al., ²⁰¹⁷). In these studies, solid‐state NMR and electron microscopy experiments were employed to characterize fibrillar structures of Aβ as will be discussed next.

In the study by Paravastu et al. (2009), brain‐derived amyloid fibrils from two deceased AD patients were used to seed the in vitro growth of synthetic Aβ_1–40 fibrils. Solid‐state NMR was used to determine the structures of isotopically labeled brain‐seeded Aβ fibrils. The structures were nearly identical between the two AD patients as they both had two co‐existing molecular structures with one structure, corresponding to 6.5 ± 1.0 nm single filament, being more predominant. On the other hand, the structures of the purely synthetic unseeded Aβ fibrils showed higher heterogeneity and were different from the brain‐derived fibrils.

Another study resolved the structures of brain‐derived Aβ fibrils extracted from occipital, parietal and temporal cortical brain tissues of two AD patients having different clinical AD histories, including mild and severe cortical atrophy as well as Lewy body dementia (Lu et al., 2013). For both patients, the resolved Aβ_1–40 fibrillar structure had a threefold symmetry about the fibril axis. However, fibrillar structures of patient I and patient II were different as evident from NMR chemical shifts and TEM fibrillar morphologies that showed a periodic twist in the fibrils of patient II but not in those of patient I. The study also utilized structural calculations to accurately determine the full molecular structure of Aβ_1–40 fibrils from patient I (Figure 3a, PDB 2M4J). Although threefold symmetry was identified for both brain‐derived (PDB 2M4J) and synthetic (PDB 2LMP) Aβ_1–40 fibrils, variations in key residue distances are observed (Lu et al., 2013; Paravastu et al., ²⁰⁰⁸).

FIGURE 3.

Resolved ssNMR/Cryo‐EM structures of natural brain‐derived Aβ_1–42/Aβ_1–40 fibrils extracted from deceased AD cases. Top view representation was used to demonstrate the fibril symmetry (i.e., conformation and number of molecules per fibril layer). (a–e) Fibrillar models of brain‐derived fibrils are generated on PyMol using the protein data bank entries and are colored based on secondary structure (yellow for β‐sheets and green for loops). (a) Threefold morphology of brain derived Aβ_1–40 fibrils, PDB: 2M4J, from occipital, parietal and temporal cortical AD brain tissues. Structural calculations were performed to accurately determine the fibril structure of Aβ_1–40 (PDB: 2M4J) that is fully consistent with the experimental data. (b) Twofold morphology of brain‐derived Aβ_1–40 fibrils, PDB: 6SHS, from the meninges of severe AD and CAA (cerebral amyloid angiopathy). (c) Twofold morphology of brain‐derived Aβ_1–40 fibrils, PDB: 6W0O, from cortical tissues of AD patients with slightly left‐handed twist as revealed by cryo‐EM. (d and e) Twofold morphologies of brain‐derived Aβ_1–42 fibrils, extracted from cortical brain tissues of sporadic (type I in d) and familial (type II in e) AD cases.

A recent paper demonstrated the structural variations in Aβ_1–40 and Aβ_1–42 fibrils that were seeded with brain extracts of patients with different clinical AD subtypes. Aβ fibrils have been extracted from cortical brain tissues of patients having distinct AD phenotypes including typical AD, rapidly progressive AD and the posterior cortical atrophy variant of AD (a visual variant of an atypical AD) (Qiang et al., 2017). The findings of this study indicate that different AD clinical subtypes correspond to different amyloid fibrillar structures in brain tissues. Posterior cortical atrophy and typical variants shared a common predominant Aβ_1–40 fibril structure. On the other hand, the rapidly progressive AD was characterized with increased polymorphism in Aβ_1–40 fibrils that had additional fibrillar structures. Unlike Aβ_1–40 fibrils, Aβ_1–42 fibrils showed increased heterogeneity in all three AD clinical subtypes including the typical phenotype (Qiang et al., 2017). A high‐resolution Cryo‐EM structure was recently determined for the most predominant Aβ_1–40 fibrils described by Qiang et al. (2017). The molecular structure (Figure 3c, PDB: 6W0O) has a twofold symmetry with slight left‐handed twist unlike the right‐handed twist identified for the meningeal Aβ fibril described below (Kollmer et al., 2019).

Recently, Kollmer et al. used Cryo‐EM to identify the structure of brain‐derived fibrils extracted at autopsy from the meninges of three severe AD patients (Kollmer et al., 2019). Although Cryo‐EM showed three major distinct morphologies of brain‐derived Aβ fibrils, a predominant structure was found to occupy brain amyloid extracts of all three AD patients (Kollmer et al., 2019). The predominant Aβ structure (Figure 3b, PDB: 6SHS) had a width of 7.4 nm, relatively short cross‐over distances and consisted of two peptide stacks (i.e., twofold symmetry with two Aβ molecules in each cross‐sectional fibril layer). Two other morphologies identified in the brain samples had larger width and cross‐over distances when compared to the main fibrillar structure described above. In fact, Cryo‐EM reconstruction of the other two morphologies revealed their similarity to the predominant morphology as they all share a common protofilament structure. However, the reported polymorphism in the three brain‐derived fibrillar structures is due to the number of the highly similar protofilaments where fibril morphology I contained one protofilament whereas fibril morphologies II and III contained two and three protofilaments, respectively. Interestingly, brain‐derived fibrils were found to be twisted with a right‐handed orientation unlike the left‐handed twisting observed in vitro synthetic fibrils. Also, brain‐derived fibrils were shown to be resistant to proteinase K (Serine protease) unlike the non‐resistant in vitro fibrils (Kollmer et al., 2019).

As for brain‐derived Aβ_1–42 amyloid, in a recent study, the structures of Aβ_1–42 filaments extracted from cortical brain tissues of 10 patients were resolved using cryo‐EM. Of the 10 patients, five had AD (two familial and three sporadic cases), while the remaining five had other, non‐AD age‐related diseases. Aβ_1–42 fibrils extracted from brain tissues of patients with sporadic and familial AD were characterized by a twisted morphology (Y. Yang et al., ²⁰²²). However, in the sporadic AD cases, a dominant type of AD Aβ_1–42 fibrils was found, termed type I, with a twofold symmetry consisting of two s‐shaped protofilaments packed together to form a predominantly hydrophobic interface. On the other hand, the familial AD cases and the other non‐AD conditions had a distinct Aβ_1–42 fibril morphology, referred to as type II filament. Type II Aβ_1–42 fibrils resemble those of type I in terms of their twofold s‐shaped protofilament arrangement, but the protofilament interface in type II fibrils is smaller, has more exposed hydrophobic residues, and is mainly stabilized by electrostatic interactions between K28 of one Aβ_1–42 unit with the carboxyl group of A42 of the other protofilament (Y. Yang et al., ²⁰²²). The cryo‐EM structure of Aβ_1–42, both type I and II (Figure 3d,e, PDBs: 7Q4B and 7Q4M), exhibits a left‐handed orientation, similar to the left‐handed Aβ_1–40 cortical fibrils (PDB: 6W0O), but different from the right‐handed Aβ_1–40 meningeal fibrils (PDB: 6SHS). In particular, cortical Aβ_1–42 fibrils (PDBs: 7Q4B and 7Q4M) and cortical Aβ_1–40 fibrils (PDB: 6W0O) share a common sub‐structure within the s‐shaped fold that is formed by the residues G25–G37 (Ghosh, Thurber, et al., 2021; Y. Yang et al., ²⁰²²). When comparing brain‐derived and in vitro‐produced Aβ_1–42 fibrils, a common s‐shaped fold is observed, but the side chain interactions and orientations as well as the inter‐fibril arrangements differ between the brain‐derived and synthetic fibrils (Y. Yang et al., ²⁰²²).

Another study by Wickramasinghe et al. reported the structure of Aβ_1–42 fibrils extracted from AD patients, using high sensitivity ssNMR (Wickramasinghe et al., 2021). Interestingly, this study reveals a new polymorphic structure of brain‐derived Aβ_1–42 fibrils grown from bacterially expressed Aβ_1–42 peptide (Wickramasinghe et al., 2021). The fibril structure consists of three β‐strands spanning residues, Tyr10–Asp23, Asn27–Vla36, and Val39–Ile41, which is different from the location of β‐strands of previously identified synthetic Aβ_1–42 fibrils (Colvin et al., 2016; Gremer et al., ²⁰¹⁷; Wälti et al., ²⁰¹⁶; Xiao et al., ²⁰¹⁵). Although Aβ_1–42 has a substantial pathogenic effect in AD patients, only two studies have reported the structural details of brain‐derived Aβ_1–42 fibrils. Future studies highlighting the structural details of Aβ_1–42 fibrils in different clinical subtypes of AD cases would be necessary to investigate whether the polymorphic nature affects disease progression and type, and will be helpful to develop potential therapeutics against AD.

An important aspect of characterizing the structure of brain‐derived Aβ aggregates is to compare them with the amyloid plaques that have been detected in the brain tissues of some non‐demented elderly individuals. Characterizing the structural differences between amyloid deposits from non‐demented and AD individuals would shed more light on the relationship between amyloid deposits and neurodegenerative processes in AD. A study by Ghosh and co‐workers examined the structural details of brain‐derived Aβ_1–40 and Aβ_1–42 fibrils extracted from non‐demented as well as AD individuals (Ghosh, Yau, et al., 2021). The study reported similar ssNMR spectra of isotopically labeled Aβ fibrils seeded by Aβ extracts from cortical tissues of non‐demented and AD subjects, suggesting that similar polymorphs develop in both cases, with Aβ_1–42 fibrils showing stronger variation between the two cases. However, statistical significances of polymorphic populations were observed in the cortical brain tissues of AD and non‐demented individuals indicating higher amounts of seed‐competent Aβ_1–40 and Aβ_1–42 aggregates. Importantly, the observed structural differences in Aβ_1–42 fibrils between AD and non‐demented subjects indicate that fibrillar Aβ_1–42 polymorphs are more predictive of cognitive impairment than Aβ_1–40 polymorphs (Ghosh, Yau, et al., 2021).

Although brain‐derived Aβ fibrils have been shown to have different structures compared with synthetic fibrils, the polymorphic nature of brain‐derived fibrils themselves presents a challenge for screening effective drugs against AD. Structure‐based design of effective inhibitors against AD would require proper identification of ligand‐binding sites on polymorphic brain‐derived Aβ fibrils (Fändrich et al., 2018). It has been found that different amyloid tracers used for in vivo imaging of Aβ can bind to different surface sites on Aβ fibrils extracted from brain homogenates of patients with sporadic and autosomal dominant AD (Ni et al., 2013; Ni et al., ²⁰¹⁷). The presence of multiple binding sites on polymorphic brain‐derived fibrils may also accommodate different ligand structures. However, future studies should further investigate the nature of binding sites on the surface of fibrils to facilitate the discovery of effective therapeutics against AD.

5. CHARACTERIZATION OF SOLUBLE BRAIN‐DERIVED Aβ OLIGOMERS

A growing body of evidence demonstrates that soluble oligomers of Aβ and other amyloidogenic proteins are the principal toxic entities (Li & Selkoe, 2020; Sherman & Lesné, ²⁰¹⁰; Sideris et al., ²⁰²¹; Walsh et al., ²⁰⁰²; T. Yang et al., ²⁰¹⁷). As extracellular species, soluble Aβ‐derived diffusible ligands (ADDLs), have been shown to exert their toxic effects on neurons, astrocytes, microglia, and vasculature through binding to several cell receptors and inducing downstream signaling events that eventually cause neurodegeneration (Carrillo‐Mora et al., 2014; Catalano et al., ²⁰⁰⁶; Fani et al., ²⁰²²; Gong et al., ²⁰⁰³; Huang & Liu, ²⁰²⁰; Kayed & Lasagna‐Reeves, ²⁰¹³; Sengupta et al., ²⁰¹⁶; Wen et al., ²⁰¹⁸). Of interest, Aβ species purified from soluble AD brain extracts were shown to significantly correlate with cognitive impairments than insoluble‐derived Aβ aggregates; a finding that indicates the pathological role of early soluble Aβ aggregates (Gong et al., 2003; Koss et al., ²⁰¹⁶; Koss et al., ²⁰¹⁸). Importantly, impaired synaptic function and inhibited hippocampal long‐term potentiation were observed in mice injected with cell media containing Aβ oligomers but not fibrils (Walsh et al., 2002). In addition, post‐mortem examination of brain tissues of AD and non‐demented elderly individuals indicates the presence of Aβ plaques but not the soluble fibrillar oligomers that are correlated with cognitive dysfunction (Tomic et al., 2009). In non‐demented subjects with Aβ plaques, toxic Aβ oligomers were not localized in postsynaptic densities and therefore did not induce synaptic loss (Zolochevska et al., 2018). In a recent review investigating possible mechanisms preventing non‐demented individuals with Aβ plaques from cognitive impairment, it was found that Aβ oligomers do not bind to synapses and therefore do not induce synaptotoxicity and inflammation through glial activation (Kok et al., 2022).

As demonstrated earlier, the resolved polymorphic structures of brain‐derived Aβ fibrils are markedly different than synthetic in vitro Aβ fibrils. Such findings stress on the need to characterize natural brain‐derived oligomers which could probably differ from their synthetic counterparts. In this section, we focus on literature findings that describe the neurotoxic roles of natural brain‐derived oligomers that have been extracted at autopsy from brain tissues of confirmed AD patients. Particularly, we highlight studies that utilized in vitro systems for examining the pathological roles of AD brain‐derived Aβ oligomers and dimers.

The presence of different soluble brain‐derived Aβ oligomers, including Aβ dimers, Aβ trimers, and Aβ*56 (i.e., 56 kDa Aβ oligomers), has been examined in brain extracts and CSF of cognitively intact and AD probable subjects (Lesné et al., 2013). The levels of Aβ oligomers were found to differ at different ages of cognitively intact individuals. Specifically, children and young subjects had negligible and undetected levels of Aβ*56 and Aβ dimers, but their levels were found to increase significantly in subjects in their 40s and 60s, respectively. On the other hand, Aβ trimers were detected in children and young subjects and their levels were found to gradually increase in subjects in their 70s. Importantly, AD probable subjects had the highest levels of Aβ dimers but not Aβ trimers or Aβ*56 (Lesné et al., 2013).

The synaptotoxicity of Aβ oligomers extracted from AD cortical tissues have been described by several in vivo models (Fritschi, Langer, et al., 2014; Li et al., ²⁰¹¹; Müller‐Schiffmann et al., ²⁰¹⁶; Rush et al., ²⁰¹⁸; Shankar et al., ²⁰⁰⁸). In these studies, brain‐derived Aβ oligomers have been particularly implicated in inhibiting hippocampal long‐term potentiation and long‐term memory in mouse models. It was shown that toxic Aβ dimers, stabilized with an intermolecular disulfide bridge, can be generated by expressing an amyloid precursor protein construct having Aβ‐S8C mutation in a transgenic mouse model. However, these soluble Aβ dimers were unable to generate insoluble Aβ plaques unless a crossbreeding with an Aβ‐plaque generating mouse model was performed (Müller‐Schiffmann et al., 2016). A thorough literature review on the pathological roles of soluble brain‐derived Aβ oligomers, particularly synaptic toxicity and hippocampal synaptic dysfunction, has been very recently published by Li and Stern (2022). In addition to using in vivo models, few studies employed in vitro toxicity assays and reported key pathological roles of natural Aβ oligomers from AD brain extracts we detail next.

An early study by Noguchi et al. showed that soluble Aβ spherical assemblies (10–15 nm, >100 kDa) derived from brain tissues of AD patients induced neurodegeneration of mature human neuronal cells and mature rat hippocampal neurons (Noguchi et al., 2009). Later in 2014, a study by Fritschi et al. showed that soluble Aβ seeds extracted from AD neocortical brain tissues had higher in vitro and in vivo seeding potencies as compared to Aβ seeds extracted from cerebrospinal fluid of the same AD patients (Fritschi, Langer, et al., 2014). For in vivo experiments, soluble Aβ extracts were injected into APP transgenic mice and after 8‐month follow‐up, a high deposition of Aβ was detected in the hippocampus of mice injected with soluble brain extracts but not for those injected with cerebrospinal fluid. Similarly, in vitro experiments of synthetic Aβ aggregation seeded with soluble brain‐derived Aβ extracts resulted in more elongated Aβ fibrils as compared to those formed by seeding with CSF‐derived Aβ oligomers (Fritschi, Langer, et al., 2014). Another recent study characterized Aβ amyloids that were derived from brain extracts of post‐mortem AD brains (olfactory cortex, amygdala, and hippocampus). Such brain‐derived amyloids were shown to internalize in in vitro endothelial cells as well in vivo mice brains that were inoculated with Aβ brain extracts intracerebrally. In this study, Aβ enriched brain‐extracts were characterized using proteomics based‐liquid chromatography–tandem mass spectrometry approach that revealed the exclusive presence of specific proteins, that is, Ring finger protein 213, in brain‐derived extracts of AD patients (Pedrero‐Prieto et al., 2019).

Using neuronal cells derived from human induced pluripotent stem cells and an in vivo mouse model, Hong et al. compared the toxicity of soluble brain‐derived Aβ oligomers obtained by the typical vigorous tissue homogenization method with those obtained by a more gentle buffer‐soaked extraction method (Hong et al., 2018; Jin et al., ²⁰¹⁸). Despite retrieving smaller amounts of Aβ oligomers from the gentle extraction method, the obtained small diffusible Aβ fractions had comparable neurotoxic and LTP impairment effects to those exerted by Aβ homogenized extracts. The paper suggests that majority of soluble Aβ extracts are relatively non‐toxic and that only a small pool of diffusible Aβ oligomers are the main toxic entities (Hong et al., 2018).

A more thorough examination of the distribution of brain‐derived Aβ oligomers in different brain regions at an early stage of AD, that is, Braak stage III, was reported recently by Sideris et al. (2021). In their study, soluble Aβ oligomers, extracted by soaking tissues of eight different brain regions, were analyzed for their size, structure, morphology, and pathological roles including inflammation, lipid membrane penetration, and toxicity. The findings reveal the presence of Aβ oligomers in all examined brain regions that were all capable of inducing inflammation, toxicity, and membrane lipid permeabilization, to varying levels. Particularly, Aβ oligomers of hippocampal extracts, with <100 nm in length and 2 nm diameter, were the most potent neurotoxic and inflammatory species (Sideris et al., 2021).

In addition to the aforementioned studies, which examined brain‐derived high molecular weight Aβ oligomers, other studies have shown that low molecular weight oligomers (T. Yang et al., ²⁰¹⁷), particularly dimers, are more potent species that can potentially induce synaptotoxicity (Brinkmalm et al., 2019; Brody & Gross, ²⁰¹⁴; Jin et al., ²⁰¹¹; Shankar et al., ²⁰⁰⁸). Using an in vitro model of hippocampal neurons, brain‐derived soluble Aβ dimers from cortical tissues of AD patients were shown to induce tau hyperphosphorylation and lead to notable disruption of cellular cytoskeleton at sub‐nanomolar levels, whereas at least two order of magnitudes of synthetic Aβ dimers were required to get the same neurotoxic effects (Brody & Gross, 2014; Jin et al., ²⁰¹¹).

6. CONCLUSIONS AND FUTURE PERSPECTIVES

The field of protein aggregation has largely relied on the use of synthetic proteins as the main species in in vitro aggregation systems. However, the reported polymorphic features of synthetic Aβ aggregates and their dependence on growth conditions suggest that amyloid species grown in the human brain cannot be predicted by fibrils produced in vitro. Indeed, recent work shows important differences between brain‐derived and synthetic aggregates, which are summarized in Table 1. Knowledge of the unique structural features of brain‐derived fibrils is extremely important for making new discoveries in the field of disease‐modifying therapies for AD and other neurodegenerative diseases. However, the polymorphic nature of brain‐derived Aβfibrils presents a challenge in the development of therapeutics for AD. Current knowledge of the polymorphic fibril structures of Aβ_1–40 and Aβ_1–42 in the brain may provide the basis for identifying potential binding sites that could be used in future studies to develop effective structure‐based drugs that either prevent formation of or disaggregate Aβ fibrils. Interestingly, in a recent paper, Seidler et al. used brain‐derived tau fibrils to perform a structure‐based discovery of small molecules that can disaggregate brain‐derived fibrils in vitro (Seidler et al., 2022). This work clearly demonstrates the importance of using amyloids from the brain to advance the discovery of potential inhibitory molecules against amyloid diseases. As the most relevant species, brain‐derived amyloids can be used as a starting material for growing disease‐relevant aggregates, which would enhance future in vitro studies of aggregation inhibition.

TABLE 1.

Brain‐derived versus synthetic Aβ fibrils.

	Brain‐derived Aβ fibrils	In vitro synthetic Aβ fibrils
Fibrillar structural polymorphism	Aβ1–40: Threefold symmetry of isotopically labeled brain‐seeded Aβ1–40 fibrils as revealed by ssNMR and EM (PDB 2M4J) a (Lu et al., 2013; Paravastu et al., 2009) Aβ1–40: Twofold symmetry of brain‐derived cortical Aβ1–40 fibrils with slightly left‐handed twist as revealed by Cryo‐EM (PDB 6W0O) (Ghosh, Thurber, et al., 2021) Aβ1–40: Twofold symmetry of brain‐derived meningeal Aβ1–40 fibril with twisted right‐handed as revealed by Cryo‐EM (PDB 6SHS) (Kollmer et al., 2019) Aβ1–42: Twofold symmetry of brain‐derived cortical Aβ1–42 fibril with twisted left‐handed as revealed by Cryo‐EM (type I: PDB 7Q4B) (Y. Yang et al., 2022) Aβ1–42: Twofold symmetry of brain‐derived cortical Aβ1–42 fibril with twisted left‐handed as revealed by Cryo‐EM (type II: 7Q4M) (Y. Yang et al., 2022)	Aβ1–42: Twofold symmetry of fibrils (Colvin et al., 2016; Wälti et al., 2016; Xiao et al., 2015) Aβ1–40: Twofold (striated ribbon) and threefold (periodically twisted) symmetries of fibrils (Hu et al., 2019; Paravastu et al., 2008; Petkova et al., 2006)
Proteinase K treatment	More resistance to proteinase K (Kollmer et al., 2019)	Less resistance to proteinase K (Kollmer et al., 2019)
Aβ post‐translational modifications	High diversity of post‐translational modifications especially in N‐terminal and C‐terminal truncations. (Arai et al., 1999; Cabrera et al., 2018; Güntert et al., 2006; Kummer & Heneka, 2014; Milton, 2001; Rijal Upadhaya et al., 2014; Rostagno et al., 2018; Sergeant et al., 2003; Wildburger et al., 2017)	No diversity in N‐terminal and C‐terminal truncations
Toxicity of Aβ oligomers	Higher neurotoxicity (Brody & Gross, 2014; Hong et al., 2018; Jin et al., 2011; Noguchi et al., 2009; Sideris et al., 2021)	Lower neurotoxicity (Brody & Gross, 2014; Jin et al., 2011)

^a Structural calculations were performed to accurately determine the fibril structure of Aβ_1–40 that is fully consistent with the experimental data (Lu et al., 2013; Miller et al., ²⁰¹¹).

Future studies of brain‐derived Aβ amyloids should consider the extraction of Aβ aggregates, both oligomers and fibrils, from the parenchyma and cortical blood vessels of deceased AD patients with various AD clinical subtypes. The extracted amyloid material, both soluble and insoluble, can be characterized by high‐resolution mass spectrometry to identify the diversity of Aβ proteoforms, including post‐translational modifications. In addition, determining the percentage of Aβ aggregates in the total isolated protein content is useful to characterize other abundant proteins that may co‐localize with Aβ and influence its aggregation pathways. Next, the structure of the different Aβ proteoforms extracted from AD brain tissues would need to be elucidated and their toxicity assessed both in vitro and in vivo to correlate the toxicity of the aggregates with the different Aβ species, which will be important for the development of effective therapeutics against AD.

AUTHOR CONTRIBUTIONS

Kenana Al Adem: Conceptualization (equal); investigation (lead); methodology (lead); validation (equal); visualization (equal); writing – original draft (lead); writing – review and editing (equal). Sungmun Lee: Conceptualization (equal); funding acquisition (lead); resources (equal); supervision (lead); writing – review and editing (equal).

Al Adem K, Lee S. Structural polymorphism and cytotoxicity of brain‐derived β‐amyloid extracts. Protein Science. 2023;32(5):e4639. 10.1002/pro.4639