. Amyloid-β peptide 37, 38 and 40 individually and cooperatively inhibit amyloid-β 42 aggregation. Chem Sci. 2022 Feb 23;13(8):2423-2439. Epub 2022 Feb 7 PubMed.


Please login to recommend the paper.


  1. This is a thorough and thoughtful study that has potentially important implications for our understanding of Aβ aggregation in AD. The data clearly and convincingly demonstrate autocatalytic secondary nucleation as the predominant process by which Aβ37 and 38 form fibrils (similar to what has previously been shown for longer Aβ peptides). More importantly, the study's systematic approach to understanding the rates of fibril formation in 3- and 4-peptide mixtures demonstrates that cross-seeding of Aβ37 and 38 is readily observable using this biochemical approach. The complex interplay within and between Aβ alloforms includes both the inhibition of Aβ42 fibrillization in the presence of Aβ37 and 38 and promotion of Aβ37 and 38 aggregation in the presence Aβ42.

    The authors also demonstrate that the cross-seeding of Aβ fragments leads to a variety of ultrastructurally distinct fibrils. This finding that raises the possibility that the relative proportions of Aβ alloforms in vivo may impact the fine ultrastructure of the β-pleated sheets produced from these fibrils.  In turn, differences in β-pleated sheet ultrastructure based on cross-seeding could potentially impact the detectability of β-amyloid deposits by PET ligands and have important implications for the development and testing of AD therapeutics. 

    Importantly, however, it remains an open question as to whether the cross-seeding seen here in a pure biochemical preparation reflects what occurs in vivo, particularly as the concentrations of Aβ fragments used here are, in some cases, much higher than what is observed under physiologic conditions. 

    Nonetheless, the results here clearly suggest that consideration of Aβ species outside of the heavily studied Aβ40 and 42 fragments may be necessary to comprehensively understand an individual’s level of β-amyloid pathology and how it relates to risk of AD-related cognitive decline.

    The observation that shorter Aβ fragments may inhibit aggregation of longer Aβ fragments (especially Aβ42) fits well with recent, intriguing work from Biofinder and ADNI that suggests having higher levels of shorter Aβ fragments, especially Aβ38, may be associated with lower rates of AD-related cognitive decline (Cullen et al., 2021). 

    To the extent that the presence of greater amounts of soluble Aβ37 and 38 relative to longer Aβ fragments is indicative of successful γ-secretase processivity, these results also bolster the case for using γ-secretase modulators to reduce β-amyloid plaque formation and lower the risk of AD-related cognitive decline. 


    . Association of CSF Aβ38 Levels With Risk of Alzheimer Disease-Related Decline. Neurology. 2022 Mar 1;98(9):e958-e967. Epub 2021 Dec 22 PubMed.

    View all comments by Dennis Selkoe
  2. The study by Braun et al. is a comprehensive analysis of how different Aβ peptides assemble on their own or when mixed together. These data in general support the notion that though the shorter Aβ peptides can assemble into amyloid fibrils, albeit with slowed rate of kinetics, they delay or inhibit Aβ42 aggregation in the mixtures, and that the complex mixtures of the various peptides inhibit aggregation further.

    These data agree with data generated in vivo in preclinical models from my lab and collaborators (Kim et al., 2007; De Mena et al., 2020; Moore et al., 2018) and are also supportive of biomarker studies that suggest that the shorter peptides might be predictive in humans.

    It is pretty clear that these types of test tube aggregation studies really do provide good insight, and have a high predictive value for in-vivo effects of Aβ aggregation kinetics.

    Where I struggle conceptually is the more general issue that we as a field rarely bring up or discuss openly: Aβ prior to deposition is present at low nanomolar levels and is likely bound to many other proteins, so how does Aβ42 ever nucleate in the brain? Obviously, this takes a long time in humans, but even so we must be missing some aspect of phenomena that occurs in the living brain. Appreciable aggregation ex vivo requires near micromolar concentrations or greater. I personally think there must be some other factor or factors that catalyze the seeding. Alternatively, there must be some way Aβ42 is locally concentrated.


    . Abeta40 inhibits amyloid deposition in vivo. J Neurosci. 2007 Jan 17;27(3):627-33. PubMed.

    . Aß40 displays amyloidogenic properties in the non-transgenic mouse brain but does not exacerbate Aß42 toxicity in Drosophila. Alzheimers Res Ther. 2020 Oct 17;12(1):132. PubMed.

    . Short Aβ peptides attenuate Aβ42 toxicity in vivo. J Exp Med. 2018 Jan 2;215(1):283-301. Epub 2017 Dec 5 PubMed.

    View all comments by Todd E. Golde
  3. Braun and colleagues present detailed and important kinetic data on Aβ oligomerization in vitro. These are crucial experiments in complex mixes of different Aβ species—Aβ37, Aβ38, Aβ40 and Aβ42—each of which is able to form aggregates and each of which can affect aggregation of other species.

    The data distinguishes Aβ42 from the shorter peptides and gives credence to the idea that Aβ42 is a more toxic, aggregation-prone species. Aβ37, Aβ38, and Aβ40 seem to behave similarly to one another, with slower aggregation kinetics. Importantly, while Aβ37, Aβ38, Aβ40 seem to be able to co-aggregate, Aβ42 forms fibrils independently. It will be fascinating to investigate other “toxic” species such as Aβ43 in increasingly more complex Aβ mixes and brain-like environments. For example, it is possible to detect at least 33 different Aβ peptides released from iPSC-derived neurons—could depletion of shorter species promote aggregation (Arber et al., 2019)? 

    With respect to familial Alzheimer’s disease, it is a crucial finding at Aβ37, Aβ38, and Aβ40 all reduce the aggregation kinetics of Aβ42. Of importance are two aspects: 1) the mixture of Aβ37, Aβ38, and Aβ40 had a greater inhibitor effect on Aβ2 aggregation than each individual peptide and 2) that of the three alloforms, Aβ38 is the more potent inhibitor of Aβ42 aggregation, suggesting that factors other than just length influence inhibition of Aβ42 aggregation. Mutations in PSEN1 are found to reduce γ-secretase processivity of APP/Aβ, increasing the relative production of Aβ42, and reducing the relative production of Aβ37/Aβ38 (Szaruga et al., 2015; Arber et al., 2019; Arber et al., 2019).  This paper therefore suggests an uninhibited oligomerization propensity behind Aβ42 aggregation in familial AD.

    This then begs the question: What about mutations that increase the entire Aβ spectrum to the same extent? Such as APP duplications or those around the β-secretase cleavage site. Would increasing overall concentration of both Aβ42 and shorter peptides while maintaining the molar ratio promote Aβ42 aggregation?

    Finally, it is tempting to speculate on the formation of plaques in nivo. For example, a core could be composed of Aβ42 rich fibrils which then form a surface accelerating aggregation of shorter Aβ species in halos around the core (Saito et al., 2011). 

    Altogether, this work gives important insights about how Aβ37, Aβ38, and Aβ40 can influence Aβ42 aggregation, and vice versa, and opens a window of opportunities for compounds such as γ-secretase modulators that could favor the production of shorter Aβ alloforms and slow down Aβ42 aggregation.

    View all comments by Rebecca Gabriele
  4. γ-Gamma-Secretase-mediated cleavage of amyloid precursor protein (APP) yields a series of length variants, aka alloforms, of the Aβ peptide, including the physiological—and most abundant—Aβ1-40 and the slightly longer Aβ1-42, whose increased production and accumulation within brain and deleterious effects contribute to Alzheimer disease pathology. Shorter alloforms, such as Aβ1-37 and Aβ1-38, also exist in significant relative amounts, and their roles in AD are often presumed to be fibrillogenic and toxic—a naïve extrapolation from the innumerable characterization studies centered on their longer cousins.

    This paper by Linse and colleagues relies on the thioflavin T (ThT) binding assay to compare the fibrillogenic behavior of Aβ1-37, Aβ1-38, Aβ1-40, and Aβ1-42 peptides either in isolation or in increasingly complex mixtures. In brief, shorter Aβ alloforms, e.g. Aβ1-37 and Aβ1-38, whether present in the reaction mixture as monomers or as seed fibrils, each can slow the rate of aggregation of either Aβ1-40 or Aβ1-42, and their effect is more pronounced with Aβ1-42 than with Aβ1-40; this confirms some of the observations made previously by us (Quartey et al., 2021) and others (Vandersteen et al., 2012). Linse and colleagues then delve deeper, and the story increases in complexity along with the complexity of the mixtures, with different stages of Aβ1-40 or Aβ1-42 fibril progression being influenced by differing combinations of peptides, and with the effect of Aβ1-38 on Aβ1-42 being more evident than that of either Aβ1-37 or Aβ1-40—something we only hinted at using surface plasmon resonance (Quartey et al., 2021). 

    These two studies underpin a critical caveat to the amyloid cascade hypothesis of AD; simply put, not all Aβ peptides are bad. This needs to become a mantra for any group that is considering any anti-amyloid therapeutic (immuno-, pharmaco-, etc.) as a clinical management tool; indeed, the current strategy of indiscriminate targeting of Aβ peptides likely explains most of the reported negative AD clinical trial outcomes using anti-amyloids.

    A possible solution to this unmet clinical need would be the development of γ-secretase modulators (GSMs). Unlike γ-secretase inhibitors, which reduce the yield of all Aβ peptides to the same extent, GSMs simply shift the cleavage of APP away from Aβ1-42 in favor of Aβ1-37 and Aβ1–38 (e.g. see Czirr et al., 2008; Borgegard et al., 2012; Ahn et al., 2020; Rynearson et al., 2021). Notably, GSMs target APP, while sparing other substrates including Notch. Now that we are aware that these shorter alloforms of Aβ may also provide relief from disease progression, these GSMs are looking even more attractive.

    The physiological levels of Aβ1-37 and Aβ1-38 are not inconsequential (~25 percent of total pool of Aβs). Neither is their influence, particularly if one considers (based on the Linse study) that these shorter peptides likely exert most of their effect(s) through reversible—and essentially repeatable—binding. Thus, having any Aβ1-37 and Aβ1-38 will delay Aβ1-42 oligomerization/aggregation sufficiently that Aβ1-42, in equilibrium, would remain predominantly as a small-molecular-weight species. The hope here is that not only would GSMs reduce the effective amount of Aβ1-42, but that any generated, and kept in mono-/dimeric form, would have higher affinity for efflux transport systems/clearance from the brain and/or be in an optimal conformation for enzymatic degradation.

    Parenthetically, it will be interesting to determine whether Aβ1-37 and Aβ1-38 are candidates for the same efflux systems and/or degradative pathways as their longer cousins, or whether they can allosterically modulate recognition of the longer species by transport and/or enzymatic systems, thus potentiating benefit for the AD brain.

    A recent paper found that the higher the concentration of Aβ1-38 in CSF, the lower the likelihood of a diagnosis of dementia on follow-up (four years) (Cullen et al., 2021). Our own work, based on a limited sample size of autopsied brain samples, revealed that a higher ratio of soluble Aβ1-42/Aβ1-38 correlated with earlier age-at-death in older donors with a confirmed diagnosis of AD; this held true for males, but not females (Quartey et al., 2021). An earlier study of ours using the same autopsy samples found a negative correlation between nicastrin (a component of the γ-secretase complex) and insoluble Aβ1-42 in males, but a positive correlation in females (Nyarko et al., 2018). Similar sex-specific observations had previously been made in brain of a mouse model of AD (Placanica et al., 2009). Thus, any GSM-based management options in the clinic will have to consider the very strong likelihood of sex-dependent benefit.

    Clearly, the generation of a pool of Aβ peptide alloforms is not simply a generalized maladaptive response in a diseased brain. The brain is said to be able to fix itself; in the case of AD, it appears to have chosen to pit Aβ peptide against Aβ peptide, and GSMs that target Aβ1-42 while promoting Aβ1-37 and Aβ1-38 may provide the leverage needed in delaying onset of disease in those most at risk.


    . The Aβ(1-38) peptide is a negative regulator of the Aβ(1-42) peptide implicated in Alzheimer disease progression. Sci Rep. 2021 Jan 11;11(1):431. PubMed.

    . Molecular Plasticity Regulates Oligomerization and Cytotoxicity of the Multipeptide-length Amyloid-β Peptide Pool. J Biol Chem. 2012 Oct 26;287(44):36732-43. PubMed.

    . Independent generation of Abeta42 and Abeta38 peptide species by gamma-secretase. J Biol Chem. 2008 Jun 20;283(25):17049-54. PubMed.

    . First and Second Generation γ-Secretase Modulators (GSMs) Modulate Amyloid-β (Aβ) Peptide Production through Different Mechanisms. J Biol Chem. 2012 Apr 6;287(15):11810-9. PubMed.

    . Pharmacokinetic and Pharmacodynamic Effects of a γ-Secretase Modulator, PF-06648671, on CSF Amyloid-β Peptides in Randomized Phase I Studies. Clin Pharmacol Ther. 2020 Jan;107(1):211-220. Epub 2019 Sep 11 PubMed.

    . Preclinical validation of a potent γ-secretase modulator for Alzheimer's disease prevention. J Exp Med. 2021 Apr 5;218(4) PubMed.

    . Association of CSF Aβ38 Levels With Risk of Alzheimer Disease-Related Decline. Neurology. 2022 Mar 1;98(9):e958-e967. Epub 2021 Dec 22 PubMed.

    . Profiles of β-Amyloid Peptides and Key Secretases in Brain Autopsy Samples Differ with Sex and APOE ε4 Status: Impact for Risk and Progression of Alzheimer Disease. Neuroscience. 2018 Mar 1;373:20-36. Epub 2018 Jan 11 PubMed.

    . Gender- and age-dependent gamma-secretase activity in mouse brain and its implication in sporadic Alzheimer disease. PLoS One. 2009;4(4):e5088. PubMed.

    View all comments by Darrell Mousseau

Make a Comment

To make a comment you must login or register.

This paper appears in the following:


  1. “Frustrated Oligomers” Slow Aggregation of Aβ42