In my view this article is misleading from the very begining, i.e. the title statement on "The emerging role of oligomeric assemblies". We had an opportunity to provide evidence in support of our arguments. Please see the following contributions for further details: Alzheimer's disease and amyloid beta protein Koudinov AR et al. Science online, Published 25 June 2002 [ Full Text ] ; Amyloid hypothesis: summer 2002 and 8th International Conference on Alzheimer’s disease update. Koudinov and Koudinova BMJ 31 July 2002 [ Full Text ] ; The state versus amyloid-beta: the trial of the most wanted criminal in Alzheimer disease. Rottkamp CA et al. Peptides 2002 Jul;23(7):1333-41 [ PubMed ].

View all comments by Alexei R. Koudinov

The recent PNAS paper by Bitan et al. represents the latest in a series of publications by the Teplow group investigating the aggregation of the Aβ peptide. This group has previously employed various biophysical and biochemical approaches to help elucidate the early assembly steps of Aβ. They have demonstrated that protofibrils are key intermediates in amyloid fibril formation, that the assembly pathway for β-sheet-containing amyloid fibrils proceeds through an α-helical intermediate, and that during the prenucleation phase of Aβ assembly, a variety of oligomeric conformations exist in rapid equilibrium.

In this paper, they hypothesize that the differences in biological activities between Aβ40 and 42 result not from the greater fibrillogenic properties of Aβ42, but from a difference in the formation of oligomeric and protofibril intermediates. Using chemical cross-linking (PICUP), they demonstrate a different oligomer distribution between Aβ40 and 42 based on SDS-PAGE, Aβ1-42 oligomers distributed primarily between tetramer and octamer with some higher molecular weight...  Read more

The recent PNAS paper by Bitan et al. represents the latest in a series of publications by the Teplow group investigating the aggregation of the Aβ peptide. This group has previously employed various biophysical and biochemical approaches to help elucidate the early assembly steps of Aβ. They have demonstrated that protofibrils are key intermediates in amyloid fibril formation, that the assembly pathway for β-sheet-containing amyloid fibrils proceeds through an α-helical intermediate, and that during the prenucleation phase of Aβ assembly, a variety of oligomeric conformations exist in rapid equilibrium.

In this paper, they hypothesize that the differences in biological activities between Aβ40 and 42 result not from the greater fibrillogenic properties of Aβ42, but from a difference in the formation of oligomeric and protofibril intermediates. Using chemical cross-linking (PICUP), they demonstrate a different oligomer distribution between Aβ40 and 42 based on SDS-PAGE, Aβ1-42 oligomers distributed primarily between tetramer and octamer with some higher molecular weight species, while Aβ40 oligomers were primarily between monomer and trimer. This basic observation is further supported by dynamic light-scattering (DLS) and size-exclusion chromatography (SEC). Indeed, PICUP/SDS-PAGE analyses of Aβ39 through 43 demonstrate that peptide length correlates with increasing oligomer size. These results also confirm previous observations by several labs that simply using SDS-PAGE for analysis of Aβ42 oligomerization in the absence of cross-linking can generate artifactual results, as the electrophoretic process and the combination of buffer, glycerol, and detergent present in SDS sample buffer may actually induce the formation of SDS-stable oligomers and cause the dissociation of oligomers not stable under these conditions.

The authors conclude with a model proposing the assembly of Aβ42 from monomer to paranuclei to large oligomers to protofibrils to fibrils—a pathway reversible until the protofibril-to-fibril step. Alternatives are also presented whereby monomers and paranuclei can also form protofibrils in a nonreversible reaction. Clearly, additional kinetic studies using a wide range of technical approaches are necessary to validate this or any other assembly mechanism, including identifying differences between the assembly of Aβ40 and 42 beyond simply the rate of assembly. While PICUP is useful as an analytic method to uncover differences in oligomer distribution between Aβ40 and 42, it will be informative if the oligomeric assemblies stabilized using this approach retain biological activity and can be used for further study in cell-based assays.

This paper also serves to illustrate the challenges presented by working with Aβ42 vs. Aβ40, and the unique characteristics associated with each species. As the authors meticulously document, while isolation of low molecular weight Aβ species (LMW) by SEC and filtration results in a comparable, reproducible starting material for Aβ40, LMW species of Aβ42 are not homogenous. Because the Aβ42 LMW starting material varies with isolation conditions, the resulting oligomeric structures also vary. While useful for Aβ40, physical manipulation to remove Aβ42 aggregates from synthetic peptide preparations may simply be incompatible with the rapid aggregation properties of this peptide species. Chemical treatment to generate an unaggregated and homogenous starting material for aggregation studies may be more appropriate for Aβ42.

View all comments by Mary Jo LaDu

In Alzheimer’s disease, the normally soluble protein amyloid β (Aβ) converts to an insoluble form and eventually deposits in the brain in the form of plaques. An understanding of the very earliest events in this conversion from soluble to insoluble aggregates of Aβ is necessary for understanding the pathogenesis of this process, and ultimately the disease. Previous studies have shown that the 42 amino acid form of Aβ (Aβ42) aggregates and forms fibrils in vitro more readily than does the shorter 40 amino acid form of Aβ (Aβ40). David Teplow et al. present interesting findings regarding the process of how these very first aggregation steps might occur.

Bitan and colleagues used a new technique termed Photo-Induced Cross-Linking of Unmodified Proteins (PICUP) to determine key differences in how Aβ40 and Aβ42 form early aggregates in vitro. The technique of PICUP covalently cross-links only proteins that are in very close proximity to each other. Starting with monomeric forms of Aβ40, they demonstrate that cross-linking produces mostly monomers, dimers, trimers, and tetramers as...  Read more

In Alzheimer’s disease, the normally soluble protein amyloid β (Aβ) converts to an insoluble form and eventually deposits in the brain in the form of plaques. An understanding of the very earliest events in this conversion from soluble to insoluble aggregates of Aβ is necessary for understanding the pathogenesis of this process, and ultimately the disease. Previous studies have shown that the 42 amino acid form of Aβ (Aβ42) aggregates and forms fibrils in vitro more readily than does the shorter 40 amino acid form of Aβ (Aβ40). David Teplow et al. present interesting findings regarding the process of how these very first aggregation steps might occur.

Bitan and colleagues used a new technique termed Photo-Induced Cross-Linking of Unmodified Proteins (PICUP) to determine key differences in how Aβ40 and Aβ42 form early aggregates in vitro. The technique of PICUP covalently cross-links only proteins that are in very close proximity to each other. Starting with monomeric forms of Aβ40, they demonstrate that cross-linking produces mostly monomers, dimers, trimers, and tetramers as analyzed by SDS-PAGE and Western blotting. However, when Aβ2 was cross-linked in a similar fashion, the most abundant forms were pentameric and hexameric.

Further analysis using electron microscopy (EM) also revealed interesting differences in the morphology of these early aggregates of cross-linked Aβ. Non-cross-linked Aβ40 examined by EM showed mostly amorphous structures that became slightly larger upon cross-linking. Non-cross-linked Aβ42 examined by EM was mostly in somewhat of a circular form, presumably a spheroid in solution. However, upon cross-linking Aβ42, chains of spheroidal structures with "beads-on-a-string" morphology were seen. These results suggest that even before oligomers or fibrils form, Aβ42 molecules can closely associate with each other (which the authors termed "paranuclei") and, because of this, may be more likely to aggregate than Aβ40. Further experiments will be needed to determine whether or not these structures and these types of processes occur in vivo. In future experiments, it will be very interesting to determine how other Aβ chaperones influence the oligomerization process and "paranuclei" formation. Prevention of paranuclei formation may represent a potential target for therapeutics.

View all comments by David Holtzman

Novel experiments by David Teplow, George Benedek, and their colleagues provide important new information about the nature of oligomeric aggregates that form rapidly in aqueous solutions of the β-amyloid peptides, prior to the formation of protofibrils and full-fledged amyloid fibrils. The significance of this work stems from two sets of prior observations: (1) In vitro, the 42-residue form of β-amyloid (Aβ42) is known to form fibrils more rapidly and at lower concentrations than the 40-residue form (Aβ40). In vivo, Aβ42 is the main component of parenchymal plaques, which are comprised principally of amyloid fibrils. Enhanced production of Aβ42 relative to Aβ40 is associated with early-onset Alzheimer’s disease (AD). These observations would be logically consistent and self-contained if plaques were the primary neurotoxic form of Aβ in AD. (2) Several lines of evidence indicate that smaller Aβ oligomers, rather than full-fledged fibrils, may be the primary neurotoxic form. In particular, correlations between total plaque burden and cognitive impairment and between the locations...  Read more

Novel experiments by David Teplow, George Benedek, and their colleagues provide important new information about the nature of oligomeric aggregates that form rapidly in aqueous solutions of the β-amyloid peptides, prior to the formation of protofibrils and full-fledged amyloid fibrils. The significance of this work stems from two sets of prior observations: (1) In vitro, the 42-residue form of β-amyloid (Aβ42) is known to form fibrils more rapidly and at lower concentrations than the 40-residue form (Aβ40). In vivo, Aβ42 is the main component of parenchymal plaques, which are comprised principally of amyloid fibrils. Enhanced production of Aβ42 relative to Aβ40 is associated with early-onset Alzheimer’s disease (AD). These observations would be logically consistent and self-contained if plaques were the primary neurotoxic form of Aβ in AD. (2) Several lines of evidence indicate that smaller Aβ oligomers, rather than full-fledged fibrils, may be the primary neurotoxic form. In particular, correlations between total plaque burden and cognitive impairment and between the locations of plaques and of neuronal dysfunction are not strong, or at least are controversial. Studies of transgenic mice expressing the human Aβ precursor protein indicate that AD-like symptoms can arise independent of plaque formation. Therefore, something other than the greater propensity of Aβ42 to form amyloid fibrils may underlie its association with early-onset AD.

Teplow and coworkers use the technique of photo-induced crosslinking of unmodified proteins (PICUP) to trap Aβ oligomers that form rapidly (within minutes or less) after preparation of low-molecular-weight (principally monomeric or dimeric) Aβ solutions at 15-40 mM peptide concentration. They find dramatic differences in the oligomerization state of crosslinked Aβ40 and Aβ42, as revealed by electrophoresis. While Aβ40 exhibits only oligomerization numbers N = 1 to N = 6, decreasing in population monotonically with increasing N, Aβ42 shows a maximum population of oligomers at N ª 6 (called "paranuclei"), as well as oligomers with N ª 12 and N ª 18 that are aggregates of paranuclei. These differences show that the behaviors of Aβ40 and Aβ42 in solution are qualitatively distinct even at the earliest stages of aggregation, in addition to the known distinctions with respect to fibril formation. Additional measurements with dynamic light scattering and electron microscopy support the electrophoresis results. The observed differences in paranucleus formation and association may help explain the connection between enhanced Aβ42 production and early-onset AD, if nonfibrillar oligomers are indeed the primary neurotoxic form of Aβ.

Identification of the Aβ paranucleus as a well-defined oligomerization state will certainly lead to additional studies aimed at further characterization of its toxicity and molecular structure. Circular dichroism data presented in this paper indicate that the β-sheet structure characteristic of amyloid fibrils has not yet formed at the paranucleus stage. Paranuclei also do not exhibit the significant α-helix content identified by Teplow and coworkers at later stages of Aβ aggregation in their earlier studies of the time course of fibrillization (Kirkitadze et al., 2001). Recent measurements by solid-state nuclear magnetic resonance and electron microscopy (Antzutkin et al., 2002) and by electron paramagnetic resonance (Török et al., 2002) suggest that the fibrillar forms of Aβ40 and Aβ42 are not significantly different at the level of molecular structure (similar fibril dimensions, morphologies, and mass-per-length; β-sheet structures with the same in-register, parallel alignment of peptide chains). The finding that Aβ40 and Aβ42 can follow qualitatively different routes to similar fibril states leads to new insights into the physicochemical aspects of amyloid fibril formation, as well as new insights into likely AD mechanisms and therapies.

View all comments by Robert Tycko