10 Nov 2022

Some call them soluble oligomers, others prefer the term protofibrils. Regardless, most researchers agree on two things when it comes to free-wheeling aggregates of Aβ: They are more toxic than their plaque-bound counterparts, and their structures are maddeningly elusive. Now, a manuscript reports that at least some of these diffusible forms of Aβ in the Alzheimer's brain are short Aβ fibrils. Led by Dennis Selkoe of Brigham and Women’s Hospital in Boston and posted on bioRxiv on October 18, the study found short, fibrillar species of Aβ lurking within fluid that had passively diffused from AD brain samples. Cryo-electron microscopy revealed that the core structure of these diffusible fibrils matched that of fibrils found within plaques. Notably, the researchers found that lecanemab—an antibody developed to bind soluble Aβ protofibrils—adhered to these diffusible Aβ fibrils and to amyloid plaques.

Short Aβ fibrils detected in AD brain sample buffer.
They adopt the same structure as those found in Aβ plaques.
They also bind lecanemab.

“This may be an extremely important discovery,” wrote Christian Haass of the German Center for Neurodegenerative Diseases in Munich. “The findings of the Selkoe lab suggest that they may have finally identified the neurotoxic Aβ assembly everybody has been looking for, for decades (Haass and Selkoe, 2007).”

While Aβ plaques, the hallmark feature of AD, are clearly visible in postmortem brain samples, the more sinister forms of Aβ are devilishly obscure. Often called soluble oligomers because they are not pelleted by ultracentrifugation of brain homogenates, these enigmatic assemblies are diverse in size and have unclear structure, although some studies have suggested that at least some of them could be fibrillar (Tomic et al., 2009). Complicating matters, which species exist within the living brain remains an open question, because tissue homogenization typically busts apart plaques and larger Aβ aggregates, potentially creating Aβ aggregates that are byproducts of the extraction procedure.

Previously, scientists led by Dominic Walsh, Brigham and Women’s Hospital (Walsh is now at Biogen), honed a gentler extraction method, in which they soaked minced, but not homogenized, pieces of brain in a buffer solution for 30 minutes, allowing any soluble species of Aβ to leak out (Jul 2018 news). Ultracentrifugation of this suspension was thought to pellet all insoluble, fibrillar forms of Aβ, leaving only soluble oligomers in the supernatant. Using the same technique, Selkoe and colleagues were subsequently surprised to find short Aβ fibrils mingling among the supposedly soluble proteins that they immunoprecipitated from the supernatant (Stern et al., 2022).

What were these short fibrils doing in this supernatant? In the current study, first author Andrew Stern and colleagues took a closer look. First, they asked if these suspended fibrils could be pelleted without immunoprecipitation. After gently soaking pieces of postmortem brain samples from people with AD and from controls in buffer, the researchers spun the fluid at a blistering pace of 200,000 g, a speed they had previously assumed was sufficient to pellet all insoluble, fibrillar aggregates of Aβ, Selkoe said. Then, they subjected the supernatant to a second, equally forceful, spin. Examining the contents of the resulting pellet under the electron microscope, the researchers spotted the short Aβ fibrils. Stern found these leached from 13 AD brain tissue samples, but not from three control tissue samples.

“If you had asked me a year ago if insoluble Aβ fibrils could resist pelleting at 200,000 g, I would have said ‘no,’ the Aβ in this supernatant is all soluble,” Selkoe told Alzforum. “Now it turns out that at least some of it is in fibrillar form.”

Using an array of experimental tools including immunoassays, size exclusion columns, denaturation, and different speeds of centrifugation, the researchers ultimately determined that the suspended Aβ fibrils represented a substantial proportion of insoluble Aβ aggregates lingering within the supernatant. These short fibrils were similar in size to species previously described as high molecular weight oligomers or protofibrils, but smaller than fibrils found in plaques. Stern and Selkoe believe that their small size made these fibrils difficult to pellet.

To examine their structure, Stern and colleagues sent samples to Yang Yang, Michel Goedert and Sjors Scheres, MRC Laboratory of Molecular Biology, Cambridge, England, who examined them via cryo-electron microscopy. They identified two core filament structures, which matched the S-shaped Type I and Type II structures previously reported for sarkosyl-insoluble Aβ fibrils extracted from AD brain samples (Jan 2022 news). The finding suggested that diffusible and plaque-bound fibrils in the AD brain shared the same core structure.

Two Types. Cryo-electron microscopy reveals Type I and Type II versions of S-shaped core structures of diffusible Aβ fibrils from one AD brain sample (right two panels) and Type I fibrils from another sample (left panel). These structures match those previously described for insoluble material. [Courtesy of Stern et al., bioRXiv, 2022.]

The researchers next asked whether lecanemab, an antibody developed to bind to soluble protofibrils, might adhere to these aqueous Aβ fibrils. Indeed, they found that lecanemab readily latched onto them. The antibody also bound to Aβ plaques within slices from the same brain samples.

To Stern and Selkoe, the findings imply that at least some of what researchers once thought were soluble oligomeric forms of Aβ are short fibrils that are resistant, but not impossible, to centrifuge out of solution.

**Lecanemab Binds.** Immunogold labeling shows lecanemab decorating the surface of a diffusible Aβ fibril. *[Courtesy of Stern et al., bioRXiv, 2022.]*

Stern proposed that, in the brain, these short fibrils could inhabit diffuse plaques, or congregate loosely around the more tightly packed, dense-core plaques that contain longer fibrils with the same core structure. Perhaps plaques shed these short diffusible fibrils, allowing them to meander into nearby synaptic clefts where they cause trouble, Stern suggested. Selkoe said that it remains to be seen whether the diffusible fibrils are neurotoxic, or perhaps precursors to smaller, more-toxic species that break off from the fibrils.

The authors have not yet tested the toxicity of these diffusible fibrils. While some reports suggest that the smallest soluble species such as dimers are most neurotoxic, others contend that midsize soluble aggregates called protofibrils also pack a synaptic wallop (Sideris et al., 2021; Sehlin et al., 2012).

Colin Masters and Victor Streltsov of the University of Melbourne noted that, along with diffusible Aβ fibrils, Stern's electron microscopy revealed numerous, apparently globular Aβ aggregates. They said the findings leave open the possibility that non-fibrillar Aβ aggregates may be the most neurotoxic state of Aβ in AD brains.

“[One] implication of this research is that it is risky to ascribe the functional characteristics of ‘soluble’ tissue extracts solely to low-molecular-weight oligomers,” commented Lary Walker of Emory University in Atlanta. “How the diffusible Aβ fibrils contribute to the bioactivity of soluble supernatants, and to the pathobiology of AD, remains to be determined,” he wrote. He added that cryo-EM could be applied more broadly to investigate proteopathic Aβ strains in different brain areas and different individuals.

Lars Lannfelt of Uppsala University in Sweden developed BAN2401 with colleagues at BioArctic Neuroscience, Stockholm. He said that the new antibody binding data are largely consistent with prior studies, which found BAN2401, the precursor to lecanemab, bound to both large soluble aggregates of Aβ and to insoluble, fibrillar forms of the peptide (Magnusson et al., 2013). However, Lannfelt emphasized that previous studies indicated that the antibody preferred soluble Aβ aggregates, called protofibrils, latching onto them an order of magnitude more tightly than to insoluble, fibrillar forms of the peptide (Nov 2021 news). The structure of these soluble protofibrils had been unknown, and Lannfelt said that the idea that at least some of them are short, Aβ fibrils is intriguing but unsurprising.

Why would lecanemab bind more tightly to diffusible fibrils than to those woven into plaques if they form the same core structure? Linda Söderberg of BioArctic thinks more lecanemab binding sites maybe be exposed on shorter, diffusible fibrils. She called the findings consistent with the idea that lecanemab preferentially binds to diffuse, loosely packed Aβ aggregates around plaques rather than plaque cores. She and Lannfelt believe that this penchant for diffusible forms of Aβ could explain why lecanemab is less likely than other Aβ antibodies to trigger ARIA-E, which stems from the disruption of densely packed Aβ aggregates such as those within blood vessels. Söderberg proposed that lecanemab bound to these floating fibrils on the plaque surface beckon microglia, which ultimately engulf both diffusible and more tightly packed forms of the peptide.

Along those lines, Stern noted that microglia compact plaques, and that perhaps entails keeping these little fibrils under wraps (May 2016 news; Apr 2021 news). The better microglia are at containing these diffusible fibrils, the better synapses will be protected, he said.

Consistent with this idea, in mice, the TREM2 agonistic antibody 4D9 seems to clear the diffuse halo around plaques, hinting that microglia activated by such antibodies may selectively remove these newly discovered diffusible fibrils, Haass wrote (Mar 2020 news). “This suggests that combinatorial treatments with anti-Aβ antibodies such as lecanemab, together with TREM2-modulating antibodies, may efficiently and selectively clear neurotoxic Aβ fibers,” he wrote.

Though not the focus of this study, Stern et al. detected a second prime culprit in AD pathogenesis within the soaking fluid of the AD brain. Why, hello paired helical filaments of tau! Typically thought to exist within intracellular tangles, these tau PHFs were found along with the aqueous Aβ fibrils within the 200,000 g supernatant. The researchers also used cryo-EM to examine their structure, and found a spot-on match with the back-to-back C-shaped structures that Goedert and Scheres had previously reported for sarkosyl-insoluble tau fibrils extracted from the AD brain (Jan 2017 news; Jul 2017 news).—Jessica Shugart

Comments

- Lary Walker
  Emory University
- Posted: 10 Nov 2022
Although a prominent role of aberrant Aβ in the pathogenesis of Alzheimer's disease (AD) is no longer in reasonable doubt, how the structure of Aβ assemblies is linked to disease remains uncertain. Soluble, low-molecular weight assemblies of Aβ are diffusible and especially toxic, whereas long fibrils of polymeric Aβ, which constitute the insoluble amyloid core of Aβ plaques, are considered less noxious.

The operational definition of soluble Aβ, i.e., the Aβ that remains in the supernatant following ultracentrifugation of brain homogenates, has been a source of ambiguity. What, exactly, are the soluble Aβ entities, and how are they involved in AD? Monomeric and oligomeric Aβ are established components of the soluble fraction, but in this report, Stern and colleagues describe the additional presence of short, diffusible Aβ-amyloid fibrils in ultracentrifugation supernatants from AD brain homogenates. The Aβ in conventionally defined “soluble” fractions of these homogenates thus includes a greater diversity of aggregates than had been previously recognized.

The researchers employed the relatively gentle method of aqueous soaking to extract diffusible Aβ from AD brain tissue samples. Using cryo-EM, they found that the molecular structure of the diffusible Aβ fibrils is the same as that of insoluble fibrils obtained from AD brain homogenates with the aid of the detergent sarkosyl. Sarkosyl thus appears not to alter the basic molecular architecture of the fibrils, a reassuring affirmation of findings from prior studies in which the detergent was used in the fiber-extraction protocol.

Another implication of this research is that it is risky to ascribe the functional characteristics of “soluble” tissue extracts solely to low-molecular-weight oligomers. How the diffusible Aβ fibrils contribute to the bioactivity of soluble supernatants, and to the pathobiology of AD, remains to be determined.

There are translational advantages to investigations of human tissues, but detailed structural and mechanistic analyses of diffusible fibrils might profitably be undertaken in animal models of Aβ deposition (if such fibrils occur in animal models).

The discovery of diffusible fibrils could facilitate the application of cryo-EM more broadly to the question of proteopathic strains in different brain areas and different subjects. For instance, do the fibrils differ in quantity or quality in arctic APP-mutant AD cases, in which most Aβ deposits are diffuse in nature? Finally, is the structure of diffusible fibrils linked to that of pre-fibrillar, low-molecular-weight Aβ oligomers? The molecular architecture of these oligomers, currently a technically intractable problem, remains a salient question.

View all comments by Lary Walker

- Christian Haass
  Biomedizinisches Centrum (BMC), Biochemie & Deutsches Zentrum für Neurodegenerative Erkrankungen (DZNE)
- Posted: 10 Nov 2022
This may be an extremely important discovery. The findings of the Selkoe lab suggest that they finally identified the neurotoxic Aβ assembly that everybody has been looking for, for decades (Haass and Selkoe, 2007). These are not small soluble oligomers, but rather large Aβ assemblies consisting of numerous Aβ molecules. These fibers, which are floating in aqueous solution, can be precipitated by ultracentrifugation. Strikingly, the cryo-EM structure of the fibers resembles that of the previously described Aβ fibers isolated from amyloid plaques (Yang et al., 2022). Even more strikingly, lecanemab, an anti-amyloid antibody that reduces amyloid plaque load and even cognitive decline, selectively binds to this previously unknown assembly. In contrast, bapineuzumab, which binds monomeric and aggregated Aβ, also detects soluble Aβ in the supernatants after ultracentrifugation. These findings may have implications for rational antibody design.

Moreover, perhaps small molecules can be designed that specifically inhibit aggregation of these Aβ assemblies. Finally, the TREM2 agonistic antibody 4D9 seems to clear the halo of Aβ-surrounding plaques (Schlepckow et al., 2020). This could suggest that such antibodies selectively remove the newly identified Aβ fibers, which could be liberated from amyloid plaques and mediate neurotoxicity. This suggests that combinatorial treatments with anti-Aβ antibodies, such as lecanemab, together with TREM2-modulating antibodies, may efficiently and selectively clear neurotoxic Aβ fibers. Finally, one may also test if such Aβ assemblies have seeding activity in mouse models.

References:

Haass C, Selkoe DJ. Soluble protein oligomers in neurodegeneration: lessons from the Alzheimer's amyloid beta-peptide. Nat Rev Mol Cell Biol. 2007 Feb;8(2):101-12. PubMed.

Schlepckow K, Monroe KM, Kleinberger G, Cantuti-Castelvetri L, Parhizkar S, Xia D, Willem M, Werner G, Pettkus N, Brunner B, Sülzen A, Nuscher B, Hampel H, Xiang X, Feederle R, Tahirovic S, Park JI, Prorok R, Mahon C, Liang CC, Shi J, Kim DJ, Sabelström H, Huang F, Di Paolo G, Simons M, Lewcock JW, Haass C. Enhancing protective microglial activities with a dual function TREM2 antibody to the stalk region. EMBO Mol Med. 2020 Apr 7;12(4):e11227. Epub 2020 Mar 10 PubMed.

Yang Y, Arseni D, Zhang W, Huang M, Lövestam S, Schweighauser M, Kotecha A, Murzin AG, Peak-Chew SY, Macdonald J, Lavenir I, Garringer HJ, Gelpi E, Newell KL, Kovacs GG, Vidal R, Ghetti B, Ryskeldi-Falcon B, Scheres SH, Goedert M. Cryo-EM structures of amyloid-β 42 filaments from human brains. Science. 2022 Jan 14;375(6577):167-172. Epub 2022 Jan 13 PubMed.

View all comments by Christian Haass

- Victor Streltsov
  Florey Instotute of Neurosciece and Mental Health
- Colin Masters
  University of Melbourne
- Posted: 15 Nov 2022
Where and when do Aβ and tau interact?

One of the most important questions left to answer in the puzzle of Alzheimer’s disease is when and where do aggregating forms of Aβ and tau interact to give rise to extracellular Aβ amyloid and intracellular tau-reactive neurites and tangles? Does Aβ ever come into direct contact with tau, and if so, in which cellular compartment? Most studies to date have used end-stage AD brain, in which the aggregates are firmly established, and are “cemented” into their respective extra- and intracellular locations.

This report by Stern et al. confirmed the protofilament fold of Aβ fibrils extracted from minced/chopped (not homogenized) end-stage AD brain tissue using 25 mM Tris and 150 mM NaCl, a cocktail of protease inhibitors, and a 30-minute “soaking” technique, followed by sufficient ultracentrifugal force. The extraction method appears important. Previously (Yang et al., 2022), Aβ fibrils were only extracted using a modified sarkosyl protocol (adding sarkosyl following homogenization), which provided “abundant Aβ filaments alongside other amyloids”, whereas the more standard sarkosyl extraction (adding sarkosyl at later stages) yields, surprisingly, “only tau filaments.” Now Stern et al. report that “fibrils account for a substantial proportion of high-molecular weight Aβ aggregates in soaking extracts.” The “soaking” technique was also used recently for soluble Aβ oligomer extraction, “at least some of which displayed fibrillar character (Stern et al., 2022). It seems that the elusive Aβ fibrils are not substantially enriched by any of these extraction techniques. Even though the authors made some attempts to minimize potential sources of ex vivo aggregation, a concern remains that these Aβ fibrillar formations might have been triggered by complex extraction procedures/conditions, while predominantly enriching non-fibrillar Aβ aggregates of various molecular weights and co-aggregates of other polypeptides and macromolecules.

Such non-fibrillar Aβ aggregates (including “oligomers”) may be the most “preferred” and expected neurotoxic state of Aβ in AD brains. For example, recent reports suggest that amorphous aggregates of positively charged polypeptides could sequester aggregation-prone Aβ and suppress the primary nucleation and elongation of Aβ fibrils, thereby modulating Aβ toxicity, despite the local increase in their concentration in biomolecular condensates (Küffner et al., 2021). For example, TDP-43 inhibits Aβ fibrillization at the initial and oligomeric stages of Aβ (Shih et al., 2020). It has also been shown that α-synuclein phase-separates into liquid bio-condensates by electrostatic complex coacervation with positively charged polypeptides, such as tau (Gracia et al., 2022). Condensates undergo either fast gelation, or coalescence followed by slow amyloid aggregation, depending on the conditions, e.g., increasing salt concentration can abolish the inhibitory effect due to charge screening and initiate Aβ fibrillization (Küffner et al., 2021). Importantly in this context, Stern et al. concede that “it remains possible that lecanemab’s target may also include non-fibrillar aggregates because we could not exclude their presence.”

We have recently revisited end-stage AD brain sarkosyl‐extraction to characterize tau and Aβ by biochemical assays and electron microscopy. Immunogold labelling revealed non‐filamentous, globular co‐aggregates of tau and Aβ, along with filamentous tau in the sarkosyl‐insoluble fractions (Mukherjee et al., 2022). Quantitative proteomics showed equimolar amounts of Aβ and R1 tau, while the insoluble fraction was rich in the microtubule-binding region 3R/4R mixed tau isoforms with multiple post‐translational modifications. Aβ and tau may form heterogeneous globular aggregates (whether before or/and during extraction is unknown, but they are decorated by aducanumab and anti-N-terminal tau antibodies), possibly through biomolecular condensation pathways driven by multivalent interactions (hydrophobic, electrostatic, etc.) from which filamentous Aβ may emerge. This fibrilization may be modulated by cofactors and/or even extraction conditions (see figure below). We note that there were also many globular shapes in the negative stain immunoelectron microscopy images of Stern et al.

[Courtesy of Mukherjee et al., J. Neurochemistry, 2022.]

Maybe the time has come to move away from using end-stage AD brain and begin to concentrate on the earliest possible sites of Aβ/tau interactions in the precuneus of persons dying with preclinical AD (Ruwanpathirana et al., 2022)?

References:

Gracia P, Polanco D, Tarancón-Díez J, Serra I, Bracci M, Oroz J, Laurents DV, García I, Cremades N. Molecular mechanism for the synchronized electrostatic coacervation and co-aggregation of alpha-synuclein and tau. Nat Commun. 2022 Aug 6;13(1):4586. PubMed.

Küffner AM, Linsenmeier M, Grigolato F, Prodan M, Zuccarini R, Capasso Palmiero U, Faltova L, Arosio P. Sequestration within biomolecular condensates inhibits Aβ-42 amyloid formation. Chem Sci. 2021 Feb 18;12(12):4373-4382. PubMed.

Mukherjee S, Dubois C, Perez K, Varghese S, Birchall IE, Leckey M, Davydova N, McLean C, Nisbet RM, Roberts BR, Li QX, Masters CL, Streltsov VA. Quantitative proteomics of tau and Aβ in detergent fractions from Alzheimer's disease brains. J Neurochem. 2023 Feb;164(4):529-552. Epub 2022 Nov 22 PubMed.

Ruwanpathirana GP, Williams RC, Masters CL, Rowe CC, Johnston LA, Davey CE. Mapping the association between tau-PET and Aβ-amyloid-PET using deep learning. Sci Rep. 2022 Aug 30;12(1):14797. PubMed.

Shih YH, Tu LH, Chang TY, Ganesan K, Chang WW, Chang PS, Fang YS, Lin YT, Jin LW, Chen YR. TDP-43 interacts with amyloid-β, inhibits fibrillization, and worsens pathology in a model of Alzheimer's disease. Nat Commun. 2020 Nov 23;11(1):5950. PubMed.

Stern AM, Liu L, Jin S, Liu W, Meunier AL, Ericsson M, Miller MB, Batson M, Sun T, Kathuria S, Reczek D, Pradier L, Selkoe DJ. A calcium-sensitive antibody isolates soluble amyloid-β aggregates and fibrils from Alzheimer's disease brain. Brain. 2022 Jan 27; PubMed.

Yang Y, Arseni D, Zhang W, Huang M, Lövestam S, Schweighauser M, Kotecha A, Murzin AG, Peak-Chew SY, Macdonald J, Lavenir I, Garringer HJ, Gelpi E, Newell KL, Kovacs GG, Vidal R, Ghetti B, Ryskeldi-Falcon B, Scheres SH, Goedert M. Cryo-EM structures of amyloid-β 42 filaments from human brains. Science. 2022 Jan 14;375(6577):167-172. Epub 2022 Jan 13 PubMed.

View all comments by Victor Streltsov View all comments by Colin Masters