The work by Luhrs et al. truly is a tour de force that integrates hydrogen exchange/NMR studies, electron microscopy, thioflavin T-binding, MTT assays, and molecular modeling into studies of the structure of fibrils formed by wild-type Aβ42 and a variety of rationally designed mutants. The result is a provocative model of the organization of Aβ protofilaments, the component structures comprising the mature 10 nm amyloid-type fibrils.

Evaluation of the work requires answers to two questions: Is the model correct, and is the model relevant? Additional experimentation will be required to provide these answers.

The clever single (asymmetric) and multiple (symmetric) amino acid substitution experiments performed by Luhrs et al. yield data consistent with their initial postulation of two β-strand regions connected by a short turn. The molecular modeling studies produce a model with low root-mean-square deviation (RMSD) error, again suggesting consistency and correctness of the derived protofilament structure. However, as with any modeling endeavor, the results depend on the...  Read more

The work by Luhrs et al. truly is a tour de force that integrates hydrogen exchange/NMR studies, electron microscopy, thioflavin T-binding, MTT assays, and molecular modeling into studies of the structure of fibrils formed by wild-type Aβ42 and a variety of rationally designed mutants. The result is a provocative model of the organization of Aβ protofilaments, the component structures comprising the mature 10 nm amyloid-type fibrils.

Evaluation of the work requires answers to two questions: Is the model correct, and is the model relevant? Additional experimentation will be required to provide these answers.

The clever single (asymmetric) and multiple (symmetric) amino acid substitution experiments performed by Luhrs et al. yield data consistent with their initial postulation of two β-strand regions connected by a short turn. The molecular modeling studies produce a model with low root-mean-square deviation (RMSD) error, again suggesting consistency and correctness of the derived protofilament structure. However, as with any modeling endeavor, the results depend on the assumptions. For example, the Luhrs model assumes the correctness of a certain feature of an earlier model, that of Robert Tycko's group, and thus becomes a "model based on a model based on experimental data" (with apologies to Professor Dan Kirschner for stealing his line).
The Luhrs model also integrates experimental data suggesting Met35 is not involved in intersheet side-chain packing. However, substantial experimental data exist showing that Met35 oxidation interferes significantly with Aβ aggregation (e.g., see Hou et al., 2002; Hou et al., 2004; Butterfield and Boyd-Kimball, 2005; Palmblad et al., 2002; Bitan et al., 2003).

These studies emphasize an important structural role for Met35 in fibril formation, one that can be rationalized mechanistically by models such as that of the Tycko group and those derived by molecular dynamics simulations (Urbanc et al., 2004). The contribution of Met35 to fibril formation is consistent with the fact that fibril formation of the [35L]Aβ(1-42) peptides used in the study of Luhrs et al. occurred over a relatively long time period (two months).

The use of oxidized Aβ (Met35 sulfoxide) also raises the question of relevance. What is the primary structure of the Aβ peptide(s) in vivo that gives rise to fibrils? Are the fibrils thus formed structurally equivalent to those studied by Luhrs et al? Evidence suggests that the predominant Aβ species is not oxidized. In addition, a large and increasing body of work supports the hypothesis that the most potent neurotoxins in Alzheimer disease and other diseases (e.g., prion diseases, see Silveira et al., 2005) may be oligomers and not fibrils. It would be exciting and informative if the approach of Luhrs et al. were applied to an analysis of the structure and dynamics of oligomer formation.

In conclusion, the fine work of Luhrs et al. provides a wonderful working model of Aβ protofilament structure that is rich in testable hypotheses.

References:
Hou L, Kang I, Marchant RE, Zagorski MG. Methionine 35 oxidation reduces fibril assembly of the amyloid abeta-(1-42) peptide of Alzheimer's disease. J Biol Chem. 2002 Oct 25;277(43):40173-6. Epub 2002 Aug 26. Abstract

Hou L, Shao H, Zhang Y, Li H, Menon NK, Neuhaus EB, Brewer JM, Byeon IJ, Ray DG, Vitek MP, Iwashita T, Makula RA, Przybyla AB, Zagorski MG. Solution NMR studies of the A beta(1-40) and A beta(1-42) peptides establish that the Met35 oxidation state affects the mechanism of amyloid formation. J Am Chem Soc. 2004 Feb 25;126(7):1992-2005. Abstract

Butterfield DA, Boyd-Kimball D. The critical role of methionine 35 in Alzheimer's amyloid beta-peptide (1-42)-induced oxidative stress and neurotoxicity. Biochim Biophys Acta. 2005 Jan 17;1703(2):149-56. Epub 2004 Nov 20. Review. Abstract

Palmblad M, Westlind-Danielsson A, Bergquist J. Oxidation of methionine 35 attenuates formation of amyloid beta -peptide 1-40 oligomers. J Biol Chem. 2002 May 31;277(22):19506-10. Epub 2002 Mar 23. Abstract

Bitan G, Tarus B, Vollers SS, Lashuel HA, Condron MM, Straub JE, Teplow DB. A molecular switch in amyloid assembly: Met35 and amyloid beta-protein oligomerization. J Am Chem Soc. 2003 Dec 17;125(50):15359-65. Abstract

Urbanc B, Cruz L, Yun S, Buldyrev SV, Bitan G, Teplow DB, Stanley HE. In silico study of amyloid beta-protein folding and oligomerization. Proc Natl Acad Sci U S A. 2004 Dec 14;101(50):17345-50. Epub 2004 Dec 6. Abstract

Silveira JR, Raymond GJ, Hughson AG, Race RE, Sim VL, Hayes SF, Caughey B. The most infectious prion protein particles. Nature. 2005 Sep 8;437(7056):257-61. Abstract

View all comments by David Teplow

1. Before comparing structural studies of Aβ fibrils from different laboratories, it is crucially important to compare the conditions under which the fibrils were grown, as our own solid-state NMR and electron microscopy studies have shown that Aβ fibril structures depend strongly on growth conditions. In the recent cryoEM studies by Zhang et al., the fibrils were grown at 37 C in 10 mM HCl. In our solid-state NMR studies, fibrils were grown at room temperature in pH 7.4 buffer.

2. The Aβ1-40 and Aβ1-42 peptides apparently adopt quite similar molecular conformations in amyloid fibrils, and both form parallel β-sheets, based on solid-state NMR, H/D exchange, and other data. But other aspects of the fibril structures may be somewhat different. Structures of Aβ1-42 fibrils have not yet been characterized completely by solid-state NMR.

3. The most surprising aspect of the cryoEM reconstruction reported by Zhang et al. is the central pore in the Aβ1-42 fibril structure. Structural models for Aβ1-40 fibrils based on solid-state NMR and electron microscopy (especially scanning...  Read more

1. Before comparing structural studies of Aβ fibrils from different laboratories, it is crucially important to compare the conditions under which the fibrils were grown, as our own solid-state NMR and electron microscopy studies have shown that Aβ fibril structures depend strongly on growth conditions. In the recent cryoEM studies by Zhang et al., the fibrils were grown at 37 C in 10 mM HCl. In our solid-state NMR studies, fibrils were grown at room temperature in pH 7.4 buffer.

2. The Aβ1-40 and Aβ1-42 peptides apparently adopt quite similar molecular conformations in amyloid fibrils, and both form parallel β-sheets, based on solid-state NMR, H/D exchange, and other data. But other aspects of the fibril structures may be somewhat different. Structures of Aβ1-42 fibrils have not yet been characterized completely by solid-state NMR.

3. The most surprising aspect of the cryoEM reconstruction reported by Zhang et al. is the central pore in the Aβ1-42 fibril structure. Structural models for Aβ1-40 fibrils based on solid-state NMR and electron microscopy (especially scanning transmission electron microscopy) do not contain such a large pore. Solid-state NMR data for Aβ1-40 fibrils indicate intermolecular contacts that would be inconsistent with the cryoEM results of Zhang et al. But again, the experiments of Zhang et al. were performed on Aβ1-42, rather than Aβ1-40, and the pH and temperature during fibril growth were quite different.

4. Finally, the Aβ1-40 peptide (and possibly also the Aβ1-42 peptide) can probably form five or six different fibril structures. It will be interesting to identify the structure or structures that develop in the human brain. This is one of the goals of our own current work.

View all comments by Robert Tycko

Zhang at al. report a three-dimensional reconstruction of an Aβ1-42 amyloid fibril based on cryoelectron microscopy data. The obtained structure varies very significantly from the fibril structure that our groups have published for Aβ1-40 peptide. This does not only hold for the Aβ1-40 structure quoted by the authors (Sachse et al., 2006; Sachse et al., 2008). It is also true for a very recently published analysis of the structure of 12 Aβ1-40 amyloid fibrils (Meinhardt et al., 2009) None of them are similar to the Aβ(1-42) fibril structure reported here.

The now published Aβ1-42 fibrils were obtained by in-vitro incubation of pure peptide at pH 2.0 for four weeks. Incubation at strongly acidic conditions and for a prolonged time is generally known to lead to peptide fragmentation or other covalent modifications. Furthermore, different pH values can lead to dramatically different fibril structures. Therefore, it is possible that the...  Read more

Zhang at al. report a three-dimensional reconstruction of an Aβ1-42 amyloid fibril based on cryoelectron microscopy data. The obtained structure varies very significantly from the fibril structure that our groups have published for Aβ1-40 peptide. This does not only hold for the Aβ1-40 structure quoted by the authors (Sachse et al., 2006; Sachse et al., 2008). It is also true for a very recently published analysis of the structure of 12 Aβ1-40 amyloid fibrils (Meinhardt et al., 2009) None of them are similar to the Aβ(1-42) fibril structure reported here.

The now published Aβ1-42 fibrils were obtained by in-vitro incubation of pure peptide at pH 2.0 for four weeks. Incubation at strongly acidic conditions and for a prolonged time is generally known to lead to peptide fragmentation or other covalent modifications. Furthermore, different pH values can lead to dramatically different fibril structures. Therefore, it is possible that the analyzed fibrils differ quite substantially from the ones that are present in Alzheimer patients and that are formed, of course, under physiologically relevant pH conditions.

It would be helpful if the manuscript provided more of the technical information that a reader would like to know for judging the reliability of this new Aβ1-42 structure and whether it truly reflects the structure of the analyzed fibrils. For example, comparisons between the raw images obtained in the electron microscope and projections of the structure are not included. No statistical analysis of the different observed fibril symmetries is shown. No mass-per-length measurements were carried out to support the interpretation of the structure with two peptides in cross-section. It is not clear to us why the published structure does not show more structural detail despite its resolution of 10 angstroms. In the light of these concerns, the presented structural model remains speculative at this point, and its relevance for Alzheimer disease remains to be further clarified.

View all comments by Marcus Fandrich
View all comments by Niko Grigorieff

Aβ40 and Aβ42 are 40- and 42-residue peptides produced by the sequential cleavage of amyloid precursor protein by β-secretase and γ-secretase. The peptides have a strong tendency to self-aggregate, initially into soluble oligomers, and eventually into insoluble fibrils and large neuronal deposits. Although the soluble oligomers are considered the major culprit of neuronal toxicity, there is nevertheless strong interest in the structure of the Aβ fibrils. Aβ fibrils have been a longstanding subject of various biophysical studies, including cryoEM. Nevertheless, the cryoEM structure of Aβ42 fiber at 10-angstrom resolution as reported by Lee and colleagues represents a significant step forward in our pursuit of the structural basis of Aβ peptide fibrillization. The new structure reveals the expected two protofilaments twisted along the fiber axis. The novelty of the new structure is that the β-sheets are arranged at the periphery surrounding a hollow core, thus forming a long tube-like structure. This architecture is drastically different from the fiber structure formed by Aβ40...  Read more

Aβ40 and Aβ42 are 40- and 42-residue peptides produced by the sequential cleavage of amyloid precursor protein by β-secretase and γ-secretase. The peptides have a strong tendency to self-aggregate, initially into soluble oligomers, and eventually into insoluble fibrils and large neuronal deposits. Although the soluble oligomers are considered the major culprit of neuronal toxicity, there is nevertheless strong interest in the structure of the Aβ fibrils. Aβ fibrils have been a longstanding subject of various biophysical studies, including cryoEM. Nevertheless, the cryoEM structure of Aβ42 fiber at 10-angstrom resolution as reported by Lee and colleagues represents a significant step forward in our pursuit of the structural basis of Aβ peptide fibrillization. The new structure reveals the expected two protofilaments twisted along the fiber axis. The novelty of the new structure is that the β-sheets are arranged at the periphery surrounding a hollow core, thus forming a long tube-like structure. This architecture is drastically different from the fiber structure formed by Aβ40 peptide, also determined by cryoEM, in Niko Grigorieff’s lab at Brandeis University, Waltham, Massachusetts, and reported previously (Meinhardt et al., 2009; Sachse et al., 2008). In the Aβ40 fiber, the β-sheets are arranged radially, twisting along the helical axis to form the long fiber with a solid core. As Aβ fibers are highly heterogeneous and polymorphic, it will be interesting to find out whether the structural differences observed in these studies merely reflects the peptide constituents (i.e., Aβ40 versus Aβ42) of the particular species of fibers selected for 3D reconstruction, or whether the structural differences represent a true defining feature of two functionally different peptides (i.e., Aβ40 is significantly less toxic than Aβ42).

The new cryoEM map by Zhang et al. fits the cryoEM micrograph well and appears solid. Furthermore, the structure model derived from the cryoEM map is supported by their extensive proteolysis data. Nevertheless, the interpretative model shall be taken with a grain of salt. Since the accurate mass per unit length of Aβ42 fiber is not known in this case, the display threshold for surface-rendering of the cryoEM map has to be somewhat artificial. The choice of threshold would thus have implications in building the structural model. I also want to point out that the proteolysis data, although supportive of their model, is not in conflict with a previous model that involves inter-β-sheets interaction (Sato et al., 2006). In summary, my impression is that there is a need for understanding the structural basis of Aβ peptide fibrillization. The current work might not be the final elucidation of such mechanism, but is a significant step forward in the long quest.

View all comments by Huilin Li

CryoEM-determined structures of Alzheimer’s peptide Aβ1-42 reveal significant differences between the fibrils of this peptide and the other most-studied Alzheimer’s peptide, Aβ1-40. Thus, they extend the known differences in kinetic, thermodynamic, and dynamic properties of these two peptides observed in solution to the supramolecular architecture of fibrils formed by them.

One of the significant points of this study is that fibrils formed by Aβ1-42 have a hollow core in contrast to those formed by Aβ1-40. At a cross-sectional plane, each protofilament accommodates a single molecule of Aβ1-42 in a hairpin-like conformation while two Aβ1-40 peptides are present in extended conformation in their respective fibrils. Structures of both fibrils were determined to the same resolution (~10 angstrom vs. ~8 angstrom); therefore, the differences can’t be attributed to the differences in experimental data collection.

However, fibril morphology is highly dependent on growth conditions. Under a variety of growth conditions, a different conformation from an ensemble of conformations...  Read more

CryoEM-determined structures of Alzheimer’s peptide Aβ1-42 reveal significant differences between the fibrils of this peptide and the other most-studied Alzheimer’s peptide, Aβ1-40. Thus, they extend the known differences in kinetic, thermodynamic, and dynamic properties of these two peptides observed in solution to the supramolecular architecture of fibrils formed by them.

One of the significant points of this study is that fibrils formed by Aβ1-42 have a hollow core in contrast to those formed by Aβ1-40. At a cross-sectional plane, each protofilament accommodates a single molecule of Aβ1-42 in a hairpin-like conformation while two Aβ1-40 peptides are present in extended conformation in their respective fibrils. Structures of both fibrils were determined to the same resolution (~10 angstrom vs. ~8 angstrom); therefore, the differences can’t be attributed to the differences in experimental data collection.

However, fibril morphology is highly dependent on growth conditions. Under a variety of growth conditions, a different conformation from an ensemble of conformations may prevail under a given set of conditions for each peptide. Nevertheless, data shown in this work are consistent with the differences observed between the two peptides in solution studies. They remind us that we still don’t know the nature of molecular interactions that affect these two similar peptides such that they can behave so differently in solution leading to significantly different consequences.

One of the common points between the EM structures of both peptides is that both structures suggest protofilaments are joined through the flexible N-terminal residues of both peptides, which also agree with the solution studies. Thus, it is very likely that dynamic properties of these peptides (i.e., switching between various conformations and their thermodynamic consequences) play a significant role in determining how the individual peptides form the initial complex and extend it to a protofilament and fibril level. This would allow small differences to be amplified, yielding significant kinetic and structural differences in fibrils of the same peptide or between the fibrils of the two peptides.

View all comments by Engin Serpersu