Does CSF Aβ Reflect Protofibril Concentration, Rather Than Plaques?

In the Alzheimer’s field, conventional wisdom holds that the growth of amyloid plaques in the brain soaks up Aβ42 peptide, depleting it from cerebrospinal fluid. Now, in the February 12 Nature Aging online, scientists led by Oskar Hansson at Lund University, Sweden, challenge this view. In mouse models, they found that the CSF Aβ42/40 ratio corresponded more closely to Aβ protofibrils in the brain than to plaques. In fact, statistical analyses determined that changes in the brain protofibril concentration fully explained the drop in CSF Aβ, suggesting that plaques have little effect on this marker. “The CSF Aβ42/40 ratio provides different information than does amyloid PET,” co-first author Nils Lindblom told Alzforum.

  • In mice, CSF Aβ42 tracks protofibrils better than plaques.
  • Protofibrils fully accounted for the drop in CSF Aβ42 as disease progressed.
  • Protofibrils also predicted CSF total tau and NfL, markers of neurodegeneration.

“This is a very interesting and potentially important paper, especially if the preferential relationship between protofibrils and [CSF] Aβ42/40 ratios can be confirmed in postmortem human brains,” Eric Reiman at Banner Alzheimer’s Institute in Phoenix wrote to Alzforum.

In related news, other scientists recently reported that phospho-tau isoforms in the CSF mirror protofibrillar forms of tau, called soluble tau assemblies (STA), in the parenchyma. These STAs correlate with tangles in brain and with cognitive decline (Feb 2025 news).

The relationship between increasing plaque load in the brain and falling CSF Aβ42 has been known for more than 20 years (Strozyk et al., 2003; Tapiola et al., 2009). There were always niggling doubts, though. Some data questioned that incorporation of Aβ42 into plaques was to blame. For one thing, Aβ40 also deposits in plaques, but its CSF levels do not change. For another, in people who carry the Arctic or Osaka familial AD mutations, CSF Aβ42 drops even though classic fibrillar plaques do not form (Shimada et al., 2011; Schöll et al., 2012). 

Plaque Growth. In 5xFAD mouse cortex, Aβ40 (aqua, top) deposits in more around the cores of plaque, while Aβ42 deposits more evenly (purple, top). In cores, the ratio of Aβ42/40 does not change from 4 months of age to 12 months (bottom left), but in protofibrils it increases with age (bottom, right). [Courtesy of Andersson et al., Nature Aging.]

To determine what CSF Aβ ratios truly represent, joint first authors Lindblom and Emelie Andersson compared these biomarkers with amyloid in the brain. In 5xFAD mice, CSF Aβ42/40 dropped by a third at 4 months of age, and two-thirds by a year. This corresponded with a rise in the Aβ42/40 ratio in both amyloid plaques and protofibrils in the brain (image above). Protofibril Aβ42/40 more closely tracked the CSF ratio, with a correlation coefficient of 0.66, whereas for plaque Aβ42/40 the correlation was 0.41 (image below). To put it another way, protofibrils explained more of the variance in CSF Aβ than did plaques. The authors isolated protofibrils using a mouse version of lecanemab, which preferentially binds this species.

Protofibril Marker? The relationship between the Aβ42/40 ratio in CSF (y axis) and in brain protofibrils (x axis, left) was stronger than its relationship to the Aβ42/40 ratio in plaques (x axis, right). [Courtesy of Andersson et al., Nature Aging.]

Additional statistical analyses reinforced these findings. When the authors examined all three Aβ42/40 ratios together using multiple linear regression analysis, protofibrils, but not plaques, predicted CSF Aβ42/40. Mediation analysis confirmed that protofibrils fully explained the relationship between parenchymal amyloid and CSF Aβ. The same thing held true in APPNL-G-F knock-in mice.

What about other CSF markers? When the authors correlated t-tau and NfL concentrations with parenchymal amyloid, they again saw that protofibrils explained more of the variance than did plaques. As before, mediation analysis found that protofibrils alone could account for the changes in these CSF markers. The rise in neurodegenerative markers may reflect neuronal injury triggered by protofibrils, the authors speculated. However, they cautioned that these statistical analyses cannot prove causality.

In future work, the scientists will investigate the relationships between protofibrils and phosphorylated tau using a mouse model with human tau knocked in. They also plan to replicate the work in human brain tissue.

Already, their findings have implications not only for understanding changes in biomarkers but also for those evoked by disease-modifying therapies. If CSF Aβ reflects protofibrils rather than plaques, it implies that when the Aβ42/40 ratio rises in someone on amyloid immunotherapy, then protofibrils must have declined. This is not surprising for lecanemab, which targets protofibrils. But why does the Aβ42/40 ratio rise during donanemab treatment, when this antibody only binds plaques? Reiman suggested that it may not matter which species an antibody binds, as long as it triggers microglia to clear plaques. Plaques act as a repository for smaller Aβ species such as protofibrils and oligomers, so plaque clearance should lower these aggregates as well (Feb 2008 conference news).

Andrew Stern at Brigham and Women’s Hospital, Boston, agrees with the paper’s conclusion that soluble Aβ aggregates in the brain could “soak up” stray Aβ42 and lead to a dearth in CSF. However, he believes these soluble species may not be protofibrils per se. He noted that there is no rigorous structural definition for protofibrils; instead, this species is distinguished by the way it is isolated, typically with an aqueous buffer. He previously found that such soluble species in fact contain amyloid fibrils (Stern et al., 2023). “Protofibrils, as distinct from amyloid fibrils, are not necessary to explain the observations in this paper or others,” Stern wrote (comment below).—Madolyn Bowman Rogers

Comments

    • User Profile Image Andrew Stern
      Brigham and Women's Hospital, Harvard Medical School
    • Posted:

    This interesting paper follows 30 years of the highly reproduced finding that the quantity of Aβ extracted from AD cortex in aqueous buffer, and retained in the supernatant after centrifugation (here termed protofibrils), correlates better with cognitive symptoms, tau pathology, and other downstream measures, compared to the amount of Aβ that enters the pellet or the number of amyloid plaques visible histologically. Here the authors nicely complement these findings in a detailed study of mouse models throughout the lifespan, paying particular attention to the Aβ42/40 CSF ratio and the aqueously extracted brain ratio anticorrelate, implying that similar factors control both. I would agree with the authors that a “soaking up” of newly generated monomers into these aqueously extractable aggregates could prevent the monomers from entering the CSF.

    These aqueously extracted Aβ aggregates have been given several names—oligomers, protofibrils, amyloid-derived diffusible ligands (ADDLs), and others. The term “protofibril” in particular has a structural implication—i.e., that the atomic structure of a protofibril is different from that of a fibril, and precedes it during aggregation from monomer. Amyloid fibrils have a strict structural definition: β-sheet-rich planar monomers stacked 4.8 Å apart, held together by intermolecular backbone hydrogen bonding and intramolecular side chain interactions.

    On the other hand, no such rigorous structural definition for a protofibril has been proposed in the literature, let alone one solved—they are so far defined only by molecular weight, not structure. One argument for a “non-fibrillar” structure comes from binding amyloid PET ligands and thioflavin dyes, but these are not necessarily reliable indicators of amyloid fibrils in isolation. Patients with the Arctic mutation have minimal amyloid PET signal, but pathology shows definite amyloid plaques with amyloid fibrils present by electron microscopy (Philipson et al., 2012), and their fibrillar structure has been solved by cryoEM (Yang et al., 2023). 

    We have found that what is extracted in aqueous buffer from human AD cortex is pelleted at sufficient centrifugal force, and possesses amyloid fibrils like in the conventionally insoluble fraction (Stern et al., 2023). Rather than make a structural claim, as the term “protofibril” implies, I would put forward that the Aβ more easily extracted in aqueous buffer reflects more diffusible amyloid fibrils that possess different isoforms/post-translational modifications, or binding partners, or are less compacted by a glial response, or are simply more recently formed, compared to the core. These other factors—modifications, binding partners, diffusibility, accumulation rate, etc.—may explain their toxicity rather than some non-fibrillar structure. This “all-fibril” model would be consistent with the observation that lecanemab, which purportedly targets protofibrils, exhibits approximately the same clinical efficacy and amyloid PET clearance as donanemab, which does not.

    I may be wrong, and a clear, distinct, protofibrillar structure of Aβ may emerge with improvements in extraction and analysis—we can only see what our techniques allow. It is important to look for one. At this moment, however, protofibrils, as distinct from amyloid fibrils, are not necessary to explain the observations in this paper or others.

    References:

    Philipson O, Lord A, Lalowski M, Soliymani R, Baumann M, Thyberg J, Bogdanovic N, Olofsson T, Tjernberg LO, Ingelsson M, Lannfelt L, Kalimo H, Nilsson LN. The Arctic amyloid-β precursor protein (AβPP) mutation results in distinct plaques and accumulation of N- and C-truncated Aβ. Neurobiol Aging. 2012 May;33(5):1010.e1-13. Epub 2011 Nov 26 PubMed.

    Yang Y, Zhang W, Murzin AG, Schweighauser M, Huang M, Lövestam S, Peak-Chew SY, Saito T, Saido TC, Macdonald J, Lavenir I, Ghetti B, Graff C, Kumar A, Nordberg A, Goedert M, Scheres SH. Cryo-EM structures of amyloid-β filaments with the Arctic mutation (E22G) from human and mouse brains. Acta Neuropathol. 2023 Mar;145(3):325-333. Epub 2023 Jan 7 PubMed.

    Stern AM, Yang Y, Jin S, Yamashita K, Meunier AL, Liu W, Cai Y, Ericsson M, Liu L, Goedert M, Scheres SH, Selkoe DJ. Abundant Aβ fibrils in ultracentrifugal supernatants of aqueous extracts from Alzheimer's disease brains. Neuron. 2023 Jul 5;111(13):2012-2020.e4. Epub 2023 May 10 PubMed.

    View all comments by Andrew Stern

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. New Biomarkers Catch Tau Before It Tangles
  2. Popcorn Plaque? Alzheimer Disease Is Slow, Yet Plaque Growth Is Fast

Research Models Citations

  1. 5xFAD (B6SJL)
  2. APP NL-G-F Knock-in

Therapeutics Citations

  1. Leqembi

Paper Citations

  1. Strozyk D, Blennow K, White LR, Launer LJ. CSF Abeta 42 levels correlate with amyloid-neuropathology in a population-based autopsy study. Neurology. 2003 Feb 25;60(4):652-6. PubMed.
  2. Tapiola T, Alafuzoff I, Herukka SK, Parkkinen L, Hartikainen P, Soininen H, Pirttilä T. Cerebrospinal fluid {beta}-amyloid 42 and tau proteins as biomarkers of Alzheimer-type pathologic changes in the brain. Arch Neurol. 2009 Mar;66(3):382-9. PubMed.
  3. Shimada H, Ataka S, Tomiyama T, Takechi H, Mori H, Miki T. Clinical course of patients with familial early-onset Alzheimer's disease potentially lacking senile plaques bearing the E693Δ mutation in amyloid precursor protein. Dement Geriatr Cogn Disord. 2011;32(1):45-54. PubMed.
  4. Schöll M, Wall A, Thordardottir S, Ferreira D, Bogdanovic N, Långström B, Almkvist O, Graff C, Nordberg A. Low PiB PET retention in presence of pathologic CSF biomarkers in Arctic APP mutation carriers. Neurology. 2012 Jul 17;79(3):229-36. PubMed.
  5. Stern AM, Yang Y, Jin S, Yamashita K, Meunier AL, Liu W, Cai Y, Ericsson M, Liu L, Goedert M, Scheres SH, Selkoe DJ. Abundant Aβ fibrils in ultracentrifugal supernatants of aqueous extracts from Alzheimer's disease brains. Neuron. 2023 Jul 5;111(13):2012-2020.e4. Epub 2023 May 10 PubMed.

Further Reading

Primary Papers

  1. Andersson E, Lindblom N, Janelidze S, Salvadó G, Gkanatsiou E, Söderberg L, Möller C, Lannfelt L, Ge J, Hanrieder J, Blennow K, Deierborg T, Mattsson-Carlgren N, Zetterberg H, Gouras G, Hansson O. Soluble cerebral Aβ protofibrils link Aβ plaque pathology to changes in CSF Aβ42/Aβ40 ratios, neurofilament light and tau in Alzheimer's disease model mice. Nat Aging. 2025 Feb 12; Epub 2025 Feb 12 PubMed.

Annotate

To make an annotation you must Login or Register.