The main component of Alzheimer’s plaques—fibrillar aggregates of amyloid-β(1-42) (Aβ42)—pose a sticky problem for structural biologists. In the past few years, groups have used various biophysical techniques to solve the molecular structure of Aβ assemblies, revealing a peptide capable of folding into more than one stable format. Now, a cryoEM structure, from the labs of Dieter Willbold and Gunnar Schröder at Heinrich-Heine-Universität, Düsseldorf, and the Forschungzentrum Jülich in Germany, offers the most detail to date for an Aβ fibril. It is also the first to account for the entire peptide, including the elusive N terminus. Different from previous structures, this fibril reveals a unique dimer interface, and a tilt to the subunits that suggests a novel mode of fibril growth. The work appeared September 7 in Science. 

  • CryoEM delivers the most detailed and complete structure of an Aβ42 fibril to date.
  • Structure offers new insight into fibril growth and N terminus functions.
  • Explains toxicity of pathogenic variants of Aβ.

“This is a great paper, showing a really nice structure with high resolution,” said Marcus Fändrich, Ulm University, Germany. Previous studies used cryoEM and nuclear magnetic resonance (NMR) to solve structures of four distinct Aβ fibers. Willbold and colleagues used both techniques, as well, but arrived at a very different result. “This one is different from all of them,” Fändrich said.

Part of the reason for the differences may be the growth conditions of the fibrils. Most of the previous studies grew fibrils in neutral pH and phosphate buffer to mimic conditions inside cells. In contrast, Willbold and colleagues opted for low pH, which may have caused Aβ to adopt yet another fibril configuration. “It’s clear that amyloid proteins form multiple fibril structures, and here we are looking at a different polymorph than before,” Willbold told Alzforum.  

Stacking Up.

A cross section of Aβ42 fiber backbone showing parallel packing of Aβ monomers. [Courtesy of Gremer et al., Science/AAAS.]

“It is amazing that again they observe a different three-dimensional structure of Aβ fibrils,” said Roland Riek, Swiss Federal Institute of Technology in Zurich. “Determining more and more polymorph structures is important to elucidate the structural landscape of Aβ,” he said.

Aβ fibrils adopt a common β-sheet structure, with two chains of stacked peptides intertwining to form fibrils. However, Aβ does not fold into one preferred conformation. Depending on conditions, Aβ subunits can end up in widely varying structures, giving rise to different strains, or fibril forms. That tendency to adopt different structures carries over in vivo, as recently illustrated by researchers in Robert Tycko’s lab at the National Institutes of Health in Bethesda, Maryland, who used NMR to define multiple structural variants of Aβ fibrils from postmortem brains (Jan 2017 news). 

In the new study, first author Lothar Gremer induced recombinant Aβ42 to form fibrils by incubation at low pH with an organic co-solvent for several weeks, yielding stable, highly uniform, unbranched fibrils several micrometers long (see image below). CryoEM analysis, supported by solid-state NMR and X-ray diffraction data, allowed him to solve the atomic structure at a resolution of 4Å, clearly locating each of the 42 amino acids, and their side chains. 

Stable Fibrils.

Electron microscopy of uniform fibers used for cryoEM. [Courtesy of Gremer et al., Science/AAAS.]

The structure revealed the expected: two intertwined protofilaments composed of Aβ monomers stacked like pancakes, with their β-strands running in parallel. Unexpectedly, each peptide was slightly non-planar, and shallowly tilted with respect to the fibril axis. This wrinkle resulted in a staggered arrangement of the subunits in the fibril, rather than straight stacks of dimers seen previously. The staggered rearrangement was also found in the recent tau fibril structure  (Jul 2017 news). 

Seeing the entire Aβ peptide revealed a serpentine “L-S” structure, with an L-shaped N-terminus giving way to the S-shaped C-terminal region in each monomer. In this compact conformation, the N terminal aspartate 1 residue forms a salt bridge with lysine 28 in the opposing peptide, which likely stabilizes the fiber.

The S motif resembles previous NMR structures (Colvin et al., 2016Wälti et al., 2016Xiao et al., 2015May 2015 news). The similarity ends there, however—the previous NMR structures showed a completely different dimer interface and orientation of the Aβ peptides in the two protofilaments. 

Gimme an LS. The “L-S” structure of Aβ42 in fibrils, proceeding from N terminus (blue) to C terminus (red). Left shows backbone and side chains of one dimer pair; right shows ribbon structure of stacked dimers. [Courtesy of Gremer et al., Science/AAAS.]

The new dimer interface revealed a starring role for the final four C terminal residues. Butted up against each other, the carboxyl terminals of opposing peptides formed a hydrophobic core that buried residues 41 and 42. The Asp1-Lys28 salt bridges flanked this core. The role of the hydrophobic terminus suggests why Aβ42 is more prone to aggregation than Aβ40, which lacks those final residues, Willbold explained. In a previous cryoEM structure, Fändrich and Nikolaus Grigorieff of Brandeis University, Waltham, Massachusetts, found the C terminus exposed on the outside, and the dimer contacts closer to the middle of the peptide (Sept 2015 news). 

The new structure suggested an interesting mode of fiber growth. The offset of the subunits to the fiber axis results in exposure of an identical binding surface to each incoming monomer. Because the two protofilaments are tilted with respect to each other, monomers joining one of them makes contact with several monomers on the adjacent protofilament. This implies that fibers grow only by adding monomers to existing fibrils, and that fibril initiation must be a cooperative process. The authors estimate that it may take up to six Aβ monomers to form a seed for this type of fibril.

Because the subunits are not planar, the two ends of the fibril are not identical. One end features a deep groove and the other, a prominent ridge. This could affect directionality of growth, with monomers binding faster at one end than the other. “Now that we know what the ends look like, this could help model fiber growth and how to interfere with it,” said Willbold, adding “For the first time, we can look for compounds that match these grooves and ridges by structure-based rational design.”

For any fibril made in vitro, the question of physiological relevance looms. Several commentators raised the issue, particularly because the fibrils were produced in extreme conditions of very low pH and aqueous-organic solvent mix. But Willbold points out that lysosomes have low pH, and no one really knows where amyloid fibrils form in the cell, or under what local conditions. In fact, recent work from Wim Annaert’s lab in KU Leuven, Belgium, indicates that presenilin 2 forms Aβ mainly in acidic late endosomes, as do some variants of presenilin 1 that cause familial AD  (Jun 2016 news). Still, Gremer acknowledged that they do not know if this polymorph occurs in the brains of affected AD patients. Nevertheless Gremer said he is confident that the new structure is important in vivo because it explains, for the first time, how mutations at the N-terminus of Aβ can be pathogenic, as has long been seen from genetic studies, and cellular, biochemical, and biophysical work. 

Mathias Jucker, University of Tübingen, Germany, wrote in an email that synthetic Aβ seeds fibrils poorly in vivo (Meyer-Luehmann et al., 2006Stöhr et al., 2012). “I can’t wait to see a similar approach used to study the extremely potent Aβ seeds in vivo from mice and AD patients,” he wrote (Ye et al., 2017; Fritschi et al., 2014). 

Others agreed that structural studies of in vivo seeds is where efforts should now focus, noting the promise of the high resolution achieved with cryoEM, especially on the heels of the recent tau amyloid fibril structure solved with the same technique. “This confirms you can solve these structures with cryoEM, and illustrates the applicability to multiple types of amyloid,” said Sjors Scheres of the MRC Laboratory of Molecular Biology in Cambridge, England. Willbold’s team is now working to determine additional structures of other Aβ assemblies at high resolution, he told Alzforum.

David Teplow, University of California Los Angeles, welcomed the detailed structure, stressing this was the first one to clearly show all 42 residues of Aβ. “We infer activity of amino acids based on indirect data, but now we have a structure. We can see how our data make sense.” The structure also offers a jumping-off point for designing inhibitors of fibril formation, Teplow said. “Even if it’s not the physiological structure, we can make inhibitors and test them. It’s a great start,” he said.—Pat McCaffrey


Make a Comment

Comments on this content

  1. This is an important paper, demonstrating that detailed molecular structural information about Aβ fibrils can be obtained from cryoEM. Other labs have attempted this in earlier publications, but with less success. The success of this work by Gremer et al. is due to recent improvements in cryoEM technology as well as their selection of in vitro fibril growth conditions to produce structurally homogeneous Aβ fibrils that are well suited for cryoEM analysis.

    Up to this point, detailed structural information about amyloid fibrils has come primarily from solid-state NMR measurements. The structure determined by Gremer et al. is different from structures of 42-residue Aβ fibrils that have been reported previously by solid-state NMR labs. But the fibril growth conditions are different, so I would not expect the structures to be the same, given the ability of Aβ peptides to form a variety of fibril polymorphs. The new cryoEM-based structure is similar in many respects to solid state NMR-based structures for other 40- and 42-residue Aβ fibril polymorphs that have been reported previously, so the work of Gremer et al. does not change our overall understanding of fibril structures. It is important to recognize that there is no contradiction between the cryoEM results and solid-state NMR results. Different labs are simply studying different polymorphs.

    Gremer et al. also report solid-state NMR spectra of their fibrils. These spectra are not the same as spectra of in vitro 42-residue Aβ fibrils that have been characterized structurally by solid state NMR. The solid-state NMR spectra of Gremer et al. are also not the same as spectra of 42-residue Aβ derived from brain tissue of Alzheimer's disease patients, as reported by Qiang et al. (2017)

    The work of Gremer et al. is an important contribution to progress on molecular structural aspects of amyloid formation, but much work remains to be done on structures of fibrils that develop in the AD brain and on possible connections between structural variations and variations in the disease.


    . Structural variation in amyloid-β fibrils from Alzheimer's disease clinical subtypes. Nature. 2017 Jan 12;541(7636):217-221. Epub 2017 Jan 4 PubMed.

  2. Another example of what the "resolution revolution" in cryo EM is giving us. In combination with rapidly developing image-processing algorithms, the new direct electron detectors allow us to finally get high-resolution views of nearly all disease-causing amyloids. 

    Sample homogeneity and the presence of polymorphic assemblies in any given sample are still stumbling blocks in some cases, but the recent examples of brain-derived PHFs and in vitro assembled Aβ1-42 fibrils lead the way. It took a long time to get to this stage, but the wait was definitely worth it.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Do Palettes of Aβ Fibril Strains Differ Among Alzheimer’s Subtypes?
  2. Tau Filaments from the Alzheimer’s Brain Revealed at Atomic Resolution
  3. Danger, S-Bends! New Structure for Aβ42 Fibrils Comes into View
  4. Electron Microscope Yields Finer Structure of α-Synuclein, Aβ Fibrils
  5. Lodged in Late Endosomes, Presenilin 2 Churns Out Intraneuronal Aβ

Paper Citations

  1. . Atomic Resolution Structure of Monomorphic Aβ42 Amyloid Fibrils. J Am Chem Soc. 2016 Aug 3;138(30):9663-74. Epub 2016 Jul 14 PubMed.
  2. . Atomic-resolution structure of a disease-relevant Aβ(1-42) amyloid fibril. Proc Natl Acad Sci U S A. 2016 Aug 23;113(34):E4976-84. Epub 2016 Jul 28 PubMed.
  3. . Aβ(1-42) fibril structure illuminates self-recognition and replication of amyloid in Alzheimer's disease. Nat Struct Mol Biol. 2015 Jun;22(6):499-505. Epub 2015 May 4 PubMed.
  4. . Exogenous induction of cerebral beta-amyloidogenesis is governed by agent and host. Science. 2006 Sep 22;313(5794):1781-4. PubMed.
  5. . Purified and synthetic Alzheimer's amyloid beta (Aβ) prions. Proc Natl Acad Sci U S A. 2012 Jul 3;109(27):11025-30. Epub 2012 Jun 18 PubMed.
  6. . Aβ seeding potency peaks in the early stages of cerebral β-amyloidosis. EMBO Rep. 2017 Sep;18(9):1536-1544. Epub 2017 Jul 12 PubMed.
  7. . Highly potent soluble amyloid-β seeds in human Alzheimer brain but not cerebrospinal fluid. Brain. 2014 Nov;137(Pt 11):2909-15. Epub 2014 Sep 10 PubMed.

Further Reading


  1. . Aggregation and Fibril Structure of AβM01-42 and Aβ1-42. Biochemistry. 2017 Sep 12;56(36):4850-4859. Epub 2017 Aug 30 PubMed.

Primary Papers

  1. . Fibril structure of amyloid-ß(1-42) by cryoelectron microscopy. Science. 2017 Sep 7; PubMed.