In Alzheimer’s disease, the aggregation of amyloid beta (Aβ) plays a central role in disease progression and memory impairment. Aβ is generated through the cleavage of the amyloid precursor protein (APP), producing peptides of varying lengths, with Aβ40 and Aβ42 being the most common. In addition to Aβ40 and Aβ42, several other Aβ peptide lengths exist, each with distinct aggregation propensities and biological effects. Among these, Aβ42 is particularly prone to aggregation, and is strongly associated with the formation of toxic oligomers and amyloid plaques. The ratio between Aβ42 and Aβ40 is a critical factor influencing the overall likelihood of aggregation. Several familial Alzheimer’s disease-linked mutations, such as the Swedish, Flemish, and Austrian, affect both the cleavage efficiency of the amyloid precursor protein (APP) and the preferred length of the resulting Aβ peptides. Truncations of the N-terminal, such as Aβ3-40/42, are also common and contribute significantly to disease progression (Sims et al., 2023). Generally, higher concentrations of Aβ increase the risk of aggregation. Several factors can impede the clearance or degradation of Aβ, with age being one of the most prominent. Mutations such as the Arctic, Italian, and Dutch mutations (E22G, K and Q) make Aβ more prone to aggregation (Fawzi et al., 2008; Grant et al., 2007; Lord et al., 2009; Yang et al., 2023). Murine Aβ is less prone to aggregation compared to human Aβ, which is why mouse models of Aβ pathology express human APP or a humanized Aβ domain.

The aggregation cascade of Aβ begins when a single monomer misfolds, adopting a beta-hairpin structure that acts as a seed for the misfolding of additional monomers, which leads to the formation of oligomers. There are multiple definitions of Aβ oligomers used in the literature; throughout this manuscript, we define an Aβ oligomer as a small, soluble aggregate that contains the beta-hairpin structure (Larini and Shea, 2012; Ruttenberg and Nowick, 2024). Soluble Aβ oligomers are considered highly neurotoxic due to their ability to disrupt synaptic function, impair long-term potentiation, and induce neuronal dysfunction (Bode et al., 2019; Gallego Villarejo et al., 2022; Huang and Liu, 2020; Kayed and Lasagna-Reeve, 2013; Nguyen et al., 2022; Sengupta et al., 2016). Several clinical trials targeting insoluble Aβ aggregates (e.g., Aducanumab, Gantenerumab) have shown plaque clearance but limited or no cognitive benefit, suggesting that toxic effects arise earlier in the aggregation pathway and may be driven by soluble oligomers (Bateman et al., 2023; Sevigny et al., 2016; Tolar et al., 2021).

These oligomers vary in size and can remain soluble before assembling into larger structures known as protofibrils. Protofibrils are intermediates between oligomers and fibrils and are also associated with neurotoxicity, as they may represent a critical transition stage in Aβ aggregation. Certain therapeutic antibodies, such as lecanemab preferentially bind to soluble protofibrils and have demonstrated moderate slowing of cognitive und functional decline in clinical trials compared to placebo (Lannfelt et al., 2014; Ono and Tsuji, 2020; van Dyck et al., 2023). However, it remains unclear whether protofibrils maintain the beta-hairpin structure or undergo further conformational changes as they mature into fibrils.

As oligomers and protofibrils continue to grow, they gradually adopt more ordered beta-sheet conformation, marking structural transition from beta-hairpins to parallel beta-sheets (Sarroukh et al., 2010; Xu et al., 2005). This conformational shift results in the formation of fibrils, larger aggregates characterized by extensive beta-sheet content. While smaller fibrils can remain soluble, continued aggregation leads to the formation of insoluble fibrils that precipitate as extracellular plaques, one of the pathological hallmarks of Alzheimer’s disease.

Despite their abundance, plaques are generally considered less acutely toxic than earlier Aβ species such as oligomers or protofibrils. All antibodies that have shown clinical benefit, such as Aducanumab (targeting aggregated Aβ), Lecanemab (preferentially binding soluble protofibrils), and Donanemab (binding Aβ plaques via the pyroglutamate-modified N-terminus of Aβ42), also reduce plaque burden and a correlation between plaque clearance and cognitive benefit has been observed (Sevigny et al., 2016; Sims et al., 2023; van Dyck et al., 2023). However, these clinical benefits have been modest to moderate, typically reflecting a slowing of cognitive decline by approximately 20 – 30% compared to placebo. This highlights both the potential and the current limitations of targeting Aβ aggregates. Nevertheless, the precise contribution of individual Aβ species to disease progression has yet to be fully defined, underscoring the need for methods capable of distinguishing between monomers, oligomers, protofibrils, and fibrils in both basic and translational research.

A schematic diagram of this aggregation cascade, showing the stepwise transition from monomers to plaques, can be found in Figure 1.

The oligomeric form of Aβ is widely considered to be the most toxic, potentially due to the hairpin structure and increased mobility, as highlighted in many studies (Celej et al., 2012; Lorenzen et al., 2014; Mrdenovic et al., 2022; Sandberg et al., 2010; Sengupta et al., 2016; Tolar et al., 2021). Understanding and quantifying various types of Aβ aggregates is crucial for improving our knowledge of different transgenic models, accurately characterizing disease types, and assessing the effects of various treatment strategies. Enzyme-Linked Immunosorbent Assay (ELISA) and other antibody-based methods are commonly used to quantify Aβ aggregates. However, designing an ELISA that accurately detects specific forms of Aβ is challenging. This is primarily due to the structural complexity of Aβ aggregates, where portions of the peptide (including the N- and C-terminal regions) may be hidden within the aggregate (Figure 2). Additionally, truncations at the N- and C-termini can alter the accessibility of epitopes, creating potential biases in detection. This structural heterogeneity can lead to misinterpretation of the results, as aggregates of different sizes and conformations may not be equally detected depending on the epitopes that are exposed.

To mitigate these challenges, it is essential to carefully select antibodies that target accessible epitopes specific to the aggregation state of interest. In some cases, using a combination of antibodies that recognize different regions of Aβ (e.g., N-terminal, C-terminal, or mid-region) may enhance the sensitivity and specificity of detection.

Equally important is the preparation of brain homogenates in a manner that enables the effective separation of soluble, membrane-bound, and insoluble Aβ species as the location of the aggregates also affects their toxicity (Gallego Villarejo et al., 2022). It is likely that the initial aggregation occurs inside the cells (Gao et al., 2021; Yu et al., 2018) and the aggregates associated with the cellular membrane are the most toxic (Mrdenovic et al., 2022; Peters et al., 2016). This allows for more accurate quantification of distinct Aβ species.

While ELISA remains a valuable tool, the inherent complexity of Aβ aggregation necessitates further optimization of assay setups to ensure more accurate detection of specific aggregate forms. Incorporating complementary techniques, such as mass spectrometry or FRET (fluorescence resonance energy transfer), could complement the read-outs from the ELISAs.

In this paper, we will outline effective methods for preparing tissue homogenate samples from Alzheimer’s disease mouse models for ELISA. We will focus on the composition and structure of various types of Aβ aggregates, with an emphasis on selecting the most appropriate antibodies for both coating and detection in an Aβ ELISA. Additionally, we will provide useful tips and tricks for optimizing ELISA setups to improve specificity in detecting different Aβ aggregate forms and cellular locations.

Selected ELISA setups from Table 2 were evaluated for their selectivity in detecting specific Aβ forms, using data generated by standard curves made from Aβ1-40 monomers, Aβ1-42 monomers, Aβ42 protofibrils (PF-42), and pre-aggregated Aβ42 as described in “see Section 3.2 Standard curves” (Figure 8).

It is evident in for instance the ELISA setup EAβdopf1-X (Figure 8D), where the monoclonal 3D6 antibody is used as both the capture and detection antibody, that the standard curve generated using Aβ42 monomers also contains an aggregated fraction. The Aβ42 standard curve exhibits a clear signal, suggesting that at least dimers are formed during the incubation process. This interpretation is further supported by the data presented in Figure 8A, which shows results from the ELISA setup EAβm1-X. In this assay, monomer-specific detection is achieved using the m266 antibody for coating and the 3D6 antibody for detection. Only monomers of Aβ40 and Aβ42 are detected under these conditions.

Notably, there is a difference in signal intensity between Aβ40 and Aβ42, indicating that Aβ42 is more prone to aggregation, which reduces its detectable monomeric form (Figure 8A). The aggregates formed by Aβ42 appear to be different from PF-42, as they are not detected in the ELISA setup EAβpf1-X, where RmAb158 is used for coating and 3D6 for detection (Figure 8G). This particular assay is designed to selectively detect Aβ fibrils, protofibrils, and larger assemblies, while excluding Aβ monomers, small Aβ oligomers, and N-terminally truncated Aβ variants.

In contrast, Figure 8H shows the results from the EAβo1-X ELISA, in which the conformation-specific A11 antibody is used for coating and 3D6 for detection. Here, pre-aggregated Aβ42 was included as a standard, and a detectable signal was observed, albeit weaker than that for the PF-42 standard. This suggests that Aβ42 monomers undergo aggregation during the ELISA incubation process, but the extent of aggregation may be limited. The weaker signal implies that longer incubation times may be necessary to allow the formation of larger aggregates or protofibrils comparable to PF-42. For further reference, in our recent publication (Metzendorf et al., 2025), we used the same homogenization procedure and a similar set of ELISA setups to characterize which Abeta variants were affected by the treatment.

Enzyme-Linked Immunosorbent Assay (ELISA) is often perceived as a straightforward and accessible technique; however, achieving accurate and reliable results require careful attention to detail throughout the experimental setup. The success of an ELISA depends not only on the quality of the antibodies used but also on the choice of capture and detection reagents, sample preparation, and the specific characteristics of the analyte being studied. Some assays may be affected by incubation times and temperature, while others may not be sensitive to these conditions. Maintaining strict control over incubation parameters (including temperature) is therefore critical, as even small variations can impact assay performance.

Commercial Aβ ELISA kits are widely used but often lack transparency regarding antibody specificity, epitope recognition, and sensitivity to different Aβ species. This can lead to misinterpretation, particularly when attempting to distinguish between monomeric, oligomeric, and fibrillar forms. Many kits recommend solubilizing samples directly in strong denaturants such as guanidine hydrochloride (GuHCl), which prevents the separation of soluble, membrane-associated, and insoluble Aβ fractions, thereby masking biological relevant distinctions.

Kits that claim to detect specific Aβ lengths often rely on C-terminal antibodies, which may fail to bind larger aggregates in samples prepared with TBS and TBST, leading to preferential detection of smaller species. In addition, commonly used N-terminal capture antibodies often favor the most abundant Aβ form, increasing the risk of underestimating less prevalent or structurally hidden species. These limitations highlight the importance of critically evaluating antibody configurations and sample preparation methods when interpreting Aβ ELISA results, particularly in translational or mechanistic studies.

Differences in Aβ sequence or structural conformation across species may also influence antibody binding and detection efficiency. As a result, quantification of Aβ species in non-human models may not always directly reflect their abundance or forms in human samples. These potential differences should be taken into account when interpreting cross-species data, especially in studies aiming to translate findings from experimental models to human disease contexts.

By tailoring ELISA setups to differentiate between various forms of Aβ, such as monomers, oligomers, aggregates, and different Aβ forms, more detailed and meaningful data can be extracted from the samples. The strategies and protocols outlined above offer enhanced sensitivity and specificity compared to standard ELISA methodologies, providing a more comprehensive approach to studying Aβ and its role in Alzheimer’s disease and other neurodegenerative conditions. These improved setups ensure that data can be interpreted with greater confidence, advancing our understanding of protein aggregation and its pathological consequences.

In addition to their utility in mechanistic studies, these protocols have potential applications in therapeutic development. Specifically, they can be integrated into drug screening pipelines to evaluate the selectivity and efficacy of candidate therapeutics targeting specific Aβ species. Since the method allows for detection of key aggregation intermediates such as oligomers and protofibrils, it can be used to monitor drug-induced shifts in Aβ aggregation states. Furthermore, its compatibility with soluble and insoluble fractions makes it suitable for use in in vitro and in vivo models. These features support the method’s applicability in preclinical drug screening workflows aimed at identifying compounds that modulate Aβ pathology in species-specific and aggregation state-specific manner.