Alzheimer’s disease (AD) is one of the prevalent neurodegenerative disorders, and is primarily caused by the misfolding of amyloidogenic proteins (1). Globally, approximately 32.3 million individuals suffer from dementia by AD, while the United States alone spends around a trillion dollars annually on social and economic costs associated with AD patients (2). While the exact causes of AD remain unclear, the amyloid cascade hypothesis (ACH) has been widely investigated to elucidate the etiological mechanisms of AD mediated by amyloid-β (Aβ) proteins (3-5). The hypothesis states that the aggregation of Aβ proteins of which the unstructured monomeric forms are converted to insoluble amyloid fibrils is central to AD pathogenesis.

Unstructured Aβ proteins self-assemble in a range of protein aggregates, spanning from small oligomeric intermediates (< 10 nm) (6, 7) to larger amyloid fibrils (> 50 nm) (Fig. 1A) (8, 9). These aggregates with varying morphologies are characteristic of AD manifesting as amyloid plaques in the brain tissues of AD patients (10), while also highly cytotoxic, causing membrane disruption (11), neuronal dysfunction (12), mitochondrial dysfunction (13), and ultimately, cell death (14). Furthermore, amyloid fibrillation of Aβ in AD progression is synergistic with the pathological aggregation of microtubule-associated protein tau (Tau) (15, 16). Aβ fibrils accelerate fibrillar aggregation of Tau, resulting in the rapid spreading of neurotoxic Tau aggregates in the brain of AD patients (17-19). Such Aβ-mediated tau pathology mechanism follows either indirect pathways through the impact of Aβ fibrils on neuronal physiology or direct pathways through Aβ fibril-mediated heterotypic seeding of Tau (20). Since the onset and progression of AD is closely associated with Aβ aggregation, understanding the nature of Aβ aggregation at the molecular level has been crucial to develop diagnostic and therapeutic strategies of AD. Molecular behaviors of Aβ peptides originate from multiple isoforms with different length of the primary structures by the cleavages of N-terminal and C-terminal regions. Given that the general significance of Aβ in AD has been widely described in other perspectives and reviews, in the current mini-review, we focus intensively on recent molecular studies of pathogenic Aβ isoforms (i.e., Aβ 1-42 [Aβ42] and Aβ 1-40 [Aβ40]) that are indispensable biomarkers of AD diagnosis.

Quantitative analysis of Aβ42, phosphorylated Tau (pTau), and amyloid aggregate in clinical samples (e.g., human cerebrospinal fluid [CSF] (21, 22), plasma (23, 24), and amyloid plaque (25), was utilized in the diagnosis of AD-mediated mild cognitive impairment (MCI) or dementia. Their significant association as a biomarker of AD was typically marked with the decrease of soluble species in fluid samples, and the increase of insoluble species in amyloid plaques of post-mortem patients. It has been suggested that an abnormality of a biomarker begins with the changes of Aβ concentrations in CSF and plasma (Fig. 1B) (26, 27); the change of the biomarker implies the accumulation of Aβ aggregates, thereby preceding the progression of AD. Using stable isotope labeling kinetics approach in conjunction with immunoprecipitation mass spectrometry, the quantitative analysis of Aβ isoforms was performed (28). The concentrations of Aβ42 in human plasma were found to be 30.13 pg/ml in the amyloid-positive group and 37.13 pg/ml in the control group. By contrast, Aβ40 concentrations of the amyloid-positive group were 272.4 pg/ml while those of the amyloid-negative group were 288.0 pg/ml. Likewise, in enzyme-linked immunosorbent assay (ELISA), the CSF levels of Aβ42 were 614.5 pg/ml (AD-MCI group) and 1,108 pg/ml (Control) while Aβ40 concentrations were 16,631 pg/ml (AD-MCI group) and 14,622 pg/ml (Control) (29). Thus, to improve the accuracy of the assessment for AD-mediated MCI and dementia, the Aβ42/Aβ40 ratio in human CSF (21) has been proposed as a new biomarker for AD. A growing body of evidence suggests that the diagnostic performance of the Aβ42/40 ratio in CSF is better than that of CSF in Aβ42 alone (22). Thus, when analyzing AD biomarkers in CSF, the measurement of relative Aβ42/40 ratio in CSF is currently widespread, rather than the absolute quantitation of Aβ42.

In addition to the diagnosis of amyloidosis in AD, the ratio of Aβ42 to Aβ40 is utilized as one of the indices to monitor the therapeutic efficacy of antibodies in clinical trials of AD. The anti-amyloid antibody approach is one of the promising therapeutic strategies for Aβ clearance in human brain through passive immunotherapy. Aducanumab (Aduhelm) (30, 31), Lecanemab (Leqembi) (32), and Donanemab (TRAILBLAZER-ALZ 2) (33), approved or being examined by the US Food and Drug Administration (FDA), are human IgG1 monoclonal antibodies targeting Aβ aggregates. To monitor the progression of amyloid status during the administration of the antibodies to patients, Aβ42/40 ratios in human plasma or CSF are measured as supportive evidence of efficacy. As described, the ratio of Aβ42 to Aβ40 has recently been highlighted as an important biomarker of AD diagnosis and treatment. Thus, we discuss the origin of Aβ42 and Aβ40 secretion in the next section.

Production of Aβ isoforms (Table 1) originates from the enzymatic cleavage of amyloid-β precursor protein (APP), yet the mechanism involved in determining the ratio of the isoforms remains unclear (34). The transmembrane domain of APP, which is embedded in the plasma membrane of human neuronal cells, contains multiple cleavage sites targeted by α-, β-, and γ-secretases (35). Orchestration of the secretases produces peptide fragments that are released into the extracellular space, involving neurotrophic activities, synaptic plasticity, and intracellular signaling (Fig. 1C) (36). The sequential cleavage of APP by β- and γ-secretases (i.e., amyloidogenic pathway) generates amyloidogenic Aβ peptides with 37-43 amino acid residues, whereas the combination of α- and γ-secretases guides the non-amyloidogenic secretion pathway that forms soluble P3 fragments. In the amyloidogenic secretion pathway, β-secretase generates soluble APP beta peptide (sAPPβ) and the APP C-terminal fragment (C99) by the cleavage of APP; then, γ-secretase splits C99 into the Aβ peptide and the APP intracellular domain (AICD). The length of Aβ proteins released to the extracellular region of neuronal cells typically terminates at either the Aβ40 or Aβ42 position, while AICD starts at the 49th or 50th position of C99. Although proteolytic cleavage mediated by the multiple secretases is the primary mechanism that determines the lengths of Aβ peptides, N-terminal truncated isoforms by non-conventional mechanisms (e.g., Cu^II-mediated self-hydrolysis, metalloproteases) have been reported as well (37). These isoforms share most of the primary structure with Aβ42 or Aβ40; however, the deletion of several N-terminal amino acid sequences significantly alter the aggregation behaviors of the truncated Aβ peptides.

The different locations of the cleavage sites in Aβ and AICD indicate that γ-secretase sequentially processes C99 at ε-cleavage (Aβ49 and 48), ξ-cleavage (Aβ46 and 45), and γ-cleavage (Aβ37, 38, 40, 42, and 43) sites. It has been proposed that Aβ40 production follows a tripeptide trimming pathway (Aβ49→46→43→40→37), while Aβ42 production follows a tri/tetrapeptide trimming pathway (Aβ48→45→42→38) (43, 44). The mechanism of the promiscuous hydrolysis by γ-secretase (45) remains unclear, but may involve structural dynamics/allosteric regulation of trimmed peptides affecting the sequential cleavages of C99 and the affinity of C-terminal motifs that determine the trimming pathways (46-48). Trimmed Aβ peptides are released from γ-secretase to the extracellular environment, when their interactions are destabilized. The two predominant forms of Aβ peptides are Aβ42 and Aβ40; Aβ42 is less abundant than Aβ40 (CSF Aβ40/Aβ42 = [9.6 ± 5.6] in normal control group, [14.2 ± 7.5] in patients with MCI, and [16.1 ± 6.7] in patients with AD) (49, 50). Aβ42 primarily leads to the formation of fibrillar aggregates, because the fibrillation rate of Aβ42 is much faster than that of Aβ40 (51, 52). Hence, the ratio of Aβ42 to Aβ40 in CSF and plasma is one of the common biomarkers to assess AD progression (22). A reduced Aβ42/Aβ40 ratio indicates the conversion of Aβ42 in CSF samples into aggregate species (53).

The presence of two additional C-terminal residues (Ile41 and Ala42) dramatically alters the aggregation propensity of Aβ42, compared to Aβ40. It is important to note that all Aβ isoforms are classified as intrinsically disordered proteins (IDPs), due to the lack of strong electrostatic/hydrophobic intramolecular interactions for a globular structure (51). Although the flexible conformations of Aβ42 and Aβ40 make them biophysically undistinguishable in the monomeric state, the slight difference at the C-terminus leads to significances in the aggregation kinetics and fibril structures of Aβ42 and Aβ40. This fact indicates that the structural dynamics of Aβ isoforms with the small change in the primary structures can influence the aggregation mechanism (54, 55). Hence, the impact of the two additional residues of Aβ42 should be emphasized to describe AD pathogenesis from the viewpoint of Aβ molecules. Thus, the molecular details of Aβ42 and Aβ40 in the aggregation are discussed in the next section.

It has been a challenging issue to determine what type of Aβ aggregates is central to neurotoxicity in AD, because the molecular mechanism and toxicology studies of Aβ aggregates (i.e., small oligomer, protofibrils, mature fibrils) indicated that the neurotoxic Aβ species were not limited to a single form (56). Although extensive research has focused on understanding the assembly mechanisms and neurotoxic effects of Aβ aggregates during the last decades, our understanding of AD and Aβ aggregates has remained shallow. However, the recent advancements in immunotherapy targeting fibrillar Aβ aggregates have identified the importance of fibrillar Aβ aggregates as a main target to alleviate AD-mediated MCI and dementia (32). The fibrillar Aβ aggregates are deposited in the amyloid plaque, a pathological hallmark of AD present in the extracellular region of neuronal cells (57). The fibril structure of Aβ aggregates is composed of β-sheet rich, unbranched, unidirectional protein assemblies (58). The peptide backbone and side chains of Aβ monomers are tightly packed following the spine of the fibril structure, and the monomers in the spine are repeated at ∼4.8 Å intervals (59). The peptide backbone of Aβ monomer forms intermolecular hydrogen bonds that strengthen β-sheet alignment. Stacking of aromatic/polar side chains and salt bridges of acidic/basic side chains further stabilizes the fibril structure through hydrophobic/electrostatic interactions and hydrogen bonds. A single stack of the fibril structure is defined as a protofibril, and multiple protofibril bundles are laterally assembled to a mature fibril. Lateral assemblies of the protofibrils are induced when the hydration shell surrounding the protofibril is liberated due to hydrophobic and electrostatic interactions between side chains on the fibril surface.

The fibril structures of Aβ42 and Aβ40 share the characteristics of non-covalent interactions due to the similarity of the fibril structure, yet different topological alignments are observed in the cross-section of the fibrils (Fig. 2). Cryo-electron microscopy (Cryo-EM) structures of Aβ42 fibrils extracted from the brain tissues of sporadic AD patients predominantly form Type I/II fibrils made of two identical S-shaped protofibrils (8). The β-sheet rich core region of the Type I protofibrils extends from Gly9 to Ala42 (Fig. 2A). The N-terminal arm (residues 9-18) and the S-shaped region (residues 19-42) constitute the cross section of the fibril spine of the Type I case. The interfacial spaces of the two protofibrils in the mature fibril are stabilized by tight packing of hydrophobic residues (Val, Leu, Phe) on the internal surface of the protofibril, while positively/negatively charged side chains (Glu, Lys, Asp) are oriented toward the outward direction on the fibril surface. The Type II protofibril structure extends from Val12 to Ala42 with four β-strands (Fig. 2B). The cross section of the spine is similar to the Type I structure with a shorter N-terminal arm. In addition, the interfaces between two S-shaped protofibrils are stabilized by salt bridges between the side chain of Lys28 and C-terminus of Ala42. Hydrophobic residues that are tightly packed in the interspace of protofibrils in Type I structure are exposed to the outside, forming a wide hydrophobic patch. The S-shape conformation of Aβ42 fibril structures is the common feature in other cryo-EM (60) and nuclear magnetic resonance (NMR) (58, 61, 62) structures of Aβ42.

In contrast to Aβ42 fibrils, the cryo-EM structure of Aβ40 fibrils extracted from the meninges of AD patients span the residues from Asp1 to Val40 (9). The topology of the Aβ40 protofibril adopts a C-shaped conformation with the N-/C-terminal arches (Fig. 2C). These arches fold toward the central hydrophobic domain, shielding the core region of the fibrils. Most of the positively/negatively charged side chains are solvent-exposed, except for Glu11 and Lys16 buried within the N-terminal arch, but stabilized through a salt bridge. Two protofibrils are contacted around ²⁴VGS²⁶, forming a cross-stack heterotypic zipper with two small cavities found in the overall structure of the mature fibril. The C-shaped conformation of the core region in Aβ40 fibrils is commonly observed in other solid-state NMR (63, 64) structures. The core residues shared in the C-shape extend from Tyr10 to Val40, and the hydrophobic residues (residues 30-40) involve the inter-protofibril interaction in the mature fibril. The topology of Aβ40 fibrils significantly differs from that of Aβ42 fibrils, but the Arctic mutation (E22G) (65) and the Osaka mutation (E22∆) (66) allow Aβ40 to form Aβ42-like fibrils with the N-terminal arm and the S-shaped conformation. Although the effect of Glu22 mutation on the fibril topology has not yet been fully investigated, a repulsive charge-charge interaction of Glu22 and Asp23 may regulate the folding of the C-terminal hydrophobic residues of Aβ40. Another conformation reported as one of the Aβ40 fibril structures is a parallel alignment of two Aβ40 monomers stacked from Tyr10 to Val40 in cryo-EM analysis (67). The fibril structure with the parallel conformation was produced by seeding fresh Aβ40 using sonicated cortex tissue extract of an AD patient. In addition, recent cryo-EM structures have reported the parallel stacking of two Aβ40 monomers (68-70). The fibril structures of Aβ42 and Aβ40 vary in the cross section of the protofibril and the interfibrillar contact area of the protofibril, implying that the additional two residues regulate considerable changes in the aggregation processes. In the next section, the mechanistic changes of Aβ42 and Aβ40 fibrillation are reviewed with the kinetic modelling of protein aggregation.

As the topologies of Aβ42 and Aβ40 fibrils are differentiated, the fibrillations of the two isoforms follow their independent aggregation pathways. At the initial stage of the fibrillation, the fibrillation of amyloid proteins begins with the primary nucleation of protein monomers (Fig. 3A). The nuclei are then elongated to amyloid fibrils by capturing protein monomers. In addition to the primary nucleation/elongation steps, the secondary nucleation on the aggregate surface (major) and the fragmentation of elongated fibrils (minor) catalyze the proliferation of active nuclei, exponentially accelerating the fibrillation due to the positive feedback between the fibril formation of nuclei and the secondary nucleation on the fibril surface. Aβ42 and Aβ40 have been the subject of systematic investigation of the aggregation process using the mechanistic models based on the primary/secondary nucleation, elongation, and fragmentation pathways (Fig. 3B). These microscopic pathways were demonstrated by mathematical modelling of in situ fibrillation kinetic traces in thioflavin T assay (71), a sensitive fluorescence dye to β-sheet rich assemblies (72). In the kinetic analysis (73), the primary nucleation of Aβ42 (3 × 10⁻⁴ M⁻² s⁻¹) is 150-fold faster than that of Aβ40 (2 × 10⁻⁶ M⁻² s⁻¹), while the elongation of Aβ42 (3 × 10⁶ M⁻¹ s⁻¹) is 10-fold faster than that of Aβ40 (3 × 10⁵ M⁻¹ s⁻¹). By contrast, the secondary nucleation of Aβ42 (1 × 10⁴ M⁻² s⁻¹) is only 3-fold faster than that of Aβ40 (3 × 10³ M⁻² s⁻¹). These results indicate that the two additional residues have a significant impact on the primary nucleation and the elongation, rather than on the secondary nucleation. The changes in the structural dynamics by the two hydrophobic side chains at the C-terminus of Aβ42 would be critical to reduce the activation energies of those molecular pathways. Thus, the molecular dynamics of Aβ42 and Aβ40 during the fibril growth are discussed in the next sections.

The self-assembly of Aβ monomers is guided by the structural transition of the proteins that promotes the conversion of intramolecular interaction to intermolecular interaction. This process leads to the formation of Aβ nuclei during the primary nucleation. The structural dynamics of Aβ42 and Aβ40 monomers have been thoroughly characterized through versatile biophysical approaches, such as two-dimensional infrared spectroscopy (2D-IR) (74), NMR spectroscopy (75), solution small-angle X-ray scattering (SAXS) (51), and molecular dynamics (MD) simulations (76). These approaches in common point out that (i) Aβ has weak intramolecular interactions, and (ii) the intermolecular hydrophobic interactions around residues 17-21 outcompete the intramolecular interactions, triggering the protein self-assembly above the threshold for spontaneous protein aggregation. The intramolecular interactions controlling the structural dynamics of Aβ are mediated by hydrophobic motifs within residues 10-35 (Fig. 3C). These hydrophobic motifs induce the formation of partially compact local structures by weak transient intramolecular interactions. Despite the flexible conformations of Aβ, local intramolecular interactions of Aβ in the central hydrophobic region delay the self-assembly of Aβ by intermolecular interactions. Compared to Aβ40 without Ile41 and Ala42, the C-terminus of Aβ42 disrupts the intramolecular interactions of the central region. The two additional residues preferably form a turn motif through frequent contacts with the hydrophobic region near residues 31-34 (75). This mode of action reduces the frequency of the intramolecular interactions that disturb the exposure of the core residues (Gln15-Gly25) and increases the possibility of intermolecular interactions in the core regions. If Ile41 and Ala42 are substituted to hydrophilic Asn residues at the same time, the hydrophilic variant of Aβ42 exhibits a slower aggregation rate, compared to the wild-type Aβ42 (77). Thus, the central hydrophobic regions of Aβ that are shielded by transient intramolecular hydrophobic interactions are attenuated in Aβ42. The hydrophobic effect of Ile41 and Ala42 also agrees well with the S-shaped conformation of Aβ42 fibrils being stabilized by the hydrophobic clusters in residues 30-42.

The nucleation of Aβ42 and Aβ40 is modulated by various environmental factors, such as pH (78), metal ions (79, 80), ionic strength, lipid membranes (81), small ligands (82, 83), peptides (76, 77), and proteins (84, 85). Because of the similarity of the primary structure, binding partners of Aβ42 and Aβ40 interact with similar regions, regardless of the two additional residues. Thus, the relative order of the nucleation rates (Aβ42 > Aβ40) is not affected. For example, the aggregation of Aβ peptides is promoted by lowering the pH in a neutral aqueous solution, because repulsive electrostatic interactions of Aβ peptides with negative charge states are attenuated by neutralization of total charge state through pH drop. However, the nucleation of Aβ42 is faster than that of Aβ40 regardless of pH changes, in that the C-terminal hydrophobic regions are not protonated/deprotonated. The increase of ionic strength shows a similar effect to the decrease of pH (86). As the ionic strength increases, the electrostatic repulsive interactions between Aβ peptides dissipate by stabilizing the charged side chains, and thereby, the nucleation rate increases.

The variation in aggregation kinetics of Aβ42 and Aβ40 is important to explain the benefit of a higher ratio of Aβ40 in human fluid that suppresses Aβ42 nucleation. Cross-interaction of different amyloid proteins is unconventional, due to the sequence-specificity in the tightly packed protein-protein interface of the aggregates. However, the similarity of Aβ42 and Aβ40 sequences enables the cross-interaction, facilitating hetero-oligomerization and fibrillation. Understanding the molecular behaviors of Aβ42 and Aβ40 when they coexist in a system has been challenging, because of the disordered protein structures and variable assembly states of Aβ. The average radius of gyration (R_g) distributions of Aβ42 (∼20.6 Å) and Aβ40 (∼20.1 Å) conformations in solution are similar (51). In the system where Aβ42 and Aβ40 coexist, Aβ40 competes with Aβ42 to form hetero-oligomers in the early stage of the aggregation, thus interfering with the self-assembly of Aβ42, and slowing the aggregation rate (73). Although the Aβ42 is more prone to aggregation in the monomeric state, Aβ40 effectively reduces the collision frequency of Aβ42 molecules, thereby delaying their self-assembly (51). Since two Aβ isoforms share the identical sequence from Asp1 to Val40, Aβ40-mediated suppression in the early stage of oligomerization would originate from the identical sequence. Molecular details of Aβ42-Aβ40 complexation remain elusive due to the structural flexibility and aggregation propensity of Aβ proteins. To overcome the limitation in the characterization of Aβ42-Aβ40 interactions, peptide design approaches mimicking the sequence of Aβ and MD simulations would be a breakthrough for understanding the remaining question.

Other isoforms with shorter AA lengths (Aβ38/37) also delay the nucleation rate of the isoforms with longer AA lengths (54). Slowing the nucleation of Aβ42 by Aβ40 is beneficial to lowering the possibility of forming cytotoxic protein aggregates. Nevertheless, note that when Aβ42 coexists during the aggregation, the self-assembly of Aβ40 is accelerated (51). Accelerated fibrillation of Aβ40 indicates that Aβ42 aggregates behave as preformed nuclei, catalyzing the aggregation of Aβ40, despite the low aggregation propensity of Aβ40. Although Aβ40 aggregates are generally less cytotoxic than Aβ42 aggregates (87, 88), Aβ40 aggregates would induce the propagation of Aβ self-assembly (including Aβ42, Aβ40, Aβ38, Aβ37) by the elongation or the secondary nucleation process. Thus, the inhibitory effect of Aβ40 on Aβ42 aggregation is limited to the primary nucleation at the initial stage, and rather, Aβ40 participates in the overall aggregation.

The nucleation/elongation mechanism of Aβ under in vitro condition is not disturbed by regulatory mechanisms of neuroglial cells. However, in human brain, toxic Aβ species generated during fibrillation trigger the activation of neuroglial cells, initiating inflammatory responses and ultimately leading to cell death. This activation is initiated by the binding of Aβ aggregates to specific receptors (56, 89). Once activated, microglial cells migrate towards the plaques and engulf Aβ aggregates through phagocytosis (90, 91). The phagocytosis by the microglial cells is induced through the recognition of the Aβ aggregates by TAM receptors (92). Consequently, this process results in the formation of dense-core plaques and a reduction in toxic Aβ species, suppressing additional aggregation processes. Understanding these regulatory mechanisms by neuroglial cells would be essential for comprehending in vivo Aβ nucleation/elongation mechanisms and developing effective strategies to control AD progression.

Structural dynamics mediated by the additional hydrophobic side chains (Ile41 and Ala42) of Aβ42 (i) accelerates the nucleation and elongation steps, and (ii) induces the formation of an S-shaped fibril topology that is distinct from Aβ40. Since Aβ42 is more prone to aggregation than Aβ40, Aβ40 and shorter isoforms abundant in human fluids play a crucial role in the suppression of Aβ42 aggregation. If the ratio of Aβ42 was higher, the aggregation of Aβ42 would be severe due to the lack of inhibitory actions of the isoforms, and the aggregation of other Aβ isoforms would be promoted through the secondary nucleation by Aβ42 aggregates. Monitoring abnormal changes of Aβ42/40 ratio in AD diagnosis and therapeutic approach is correlated with the different aggregation propensities of Aβ42 and Aβ40. In addition to Aβ40, shorter Aβ isoforms (i.e., Aβ38, Aβ37) with low aggregation propensity are recently highlighted, due to their potential as novel biomarkers for AD diagnosis (50), and their inhibitory effects on Aβ42 aggregation (54). Given that the observation of the biomarker abnormalities in AD is highly relevant to the molecular role of Aβ isoforms, the molecular characterization of the shorter Aβ isoforms (amino acid length < 42aa) and their formation mechanisms by γ-secretase would be crucial for future studies with regard to the ratio of short isoforms and Aβ42. Several familial mutations in AD cases involve the region of APP close to the cleavage sites of γ-secretase, thereby affecting the ratio of Aβ42 to Aβ40 (93-95). Due to the importance of γ-secretase activity, attempts have been made to reduce Aβ42 production using chemical modulators of γ-secretase. However, the modulation strategy of γ-secretase inevitably leads to side effects (e.g., cognitive deterioration), because the γ-secretase hydrolyzes other transmembrane proteins besides APP (96, 97). Thus, high specificity in the regulation of the enzymatic cleavage of APP would be required to develop the next generation of the γ-secretase modulator, to reduce the likelihood of side effects. Modulating γ-secretase activity may not be optimal to removing accumulated amyloid plaques but would be effective to maintain the low concentration of pathogenic Aβ isoforms in a subsequent therapeutic strategy.

In the context of a therapeutic strategy, regulation of Aβ42 aggregation at the molecular level, rather than Aβ40 or shorter isoforms, would be at the core of suppressing the initiation or propagation of Aβ deposition. The thermodynamic stability of Aβ42 aggregates is extremely high, despite the short distance of the primary sequence, compared to other amyloidogenic proteins (59). Such high stability of the fibril structure hinders the resolubilizing of the formed Aβ42 aggregates into a monomeric state while the aggregates propagate pathogenic aggregation through catalytic centers on the fibril surface. Hence, general strategies of the conventional Aβ42 inhibitors were limited to delaying the primary nucleation or isolating/depleting residual monomers to prevent additional aggregations. However, as shown in Fig. 1, the changes of biomarkers are not parallel with the onset of AD symptoms, implying that Aβ aggregates are already dominant in AD patients when the symptoms are observed in the late stage of AD. For this reason, passive immunization approaches using Aβ aggregate-targeting antibodies are focused on the activation of spontaneous fibril disaggregation/degradation by microglial cells. Thus, to facilitate the disaggregation/degradation pathways of Aβ aggregates by the antibodies, studies to overcome the thermodynamic stabilities of Aβ42 aggregates at the molecular level would be crucial.

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. RS-2023-00213155 and RS-2023-00221182 to T.S.C. & RS-2023-00274504 to D.I.), and the Korea Basic Science Institute (KBSI) National Research Facilities & Equipment Center (NFEC) funded by the Korea government (Ministry of Education) (2019R1A6C1010028 to T.S.C.). Figures were created with Biorender.com and all graphics related to fibril structures were produced using PyMOL (98).

The authors have no conflicting interests.