2023 Volume 20 Issue 1 Article ID: e200007
To completely treat and ultimately prevent dementia, it is essential to elucidate its pathogenic mechanisms in detail. There are two major hypotheses for the pathogenesis of Alzheimer’s dementia: the β-amyloid (Aβ) hypothesis and the tau hypothesis. The modified amyloid hypothesis, which proposes that toxic oligomers rather than amyloid fibrils are the essential cause, has recently emerged. Aβ peptides [Aβ(1–40) and Aβ(1–42)] form highly insoluble aggregates in vivo and in vitro. These Aβ aggregates contain many polymorphisms, whereas Aβ peptides are intrinsically disordered in physiological aqueous solutions without any compact conformers. Over the last three decades, solid-state nuclear magnetic resonance (NMR) has greatly contributed to elucidating the structure of each polymorph, while solution NMR has revealed the dynamic nature of the transient conformations of the monomer. Moreover, several methods to investigate the aggregation process based on the observation of magnetization saturation transfer have also been developed. The complementary use of NMR methods with cryo-electron microscopy, which has rapidly matured, is expected to clarify the relationship between the amyloid and molecular pathology of Alzheimer’s dementia in the near future. This review article is an extended version of the Japanese article, Insights into the Mechanisms of Oligomerization/Fibrilization of Amyloid β Peptide from Nuclear Magnetic Resonance, published in SEIBUTSU BUTSURI Vol. 62, p. 39–42 (2022).
To understand and prevent dementia, it is crucial to study its molecular pathology. There are two major hypotheses for Alzheimer’s dementia: the β-amyloid (Aβ) hypothesis and the tau hypothesis. Recently, the modified amyloid hypothesis suggested that toxic oligomers may cause dementia, not amyloid fibrils. Nuclear magnetic resonance (NMR) and cryo-electron microscopy are key tools in studying the structure and fibril formation process of Aβ peptides, which form both insoluble fibrils and soluble oligomers. The combined use of NMR and cryo-electron microscopy will likely clarify the relationship between Alzheimer’s dementia and its molecular pathology.
In June 2022, lecanemab/BAN2401, a recombinant human IgG therapeutic antibody that binds to soluble oligomers of the amyloid-β peptide (Aβ), was approved by the US Food and Drug Administration (FDA) for the treatment of Alzheimer’s dementia (AD). This antibody drug is a humanized version of a mouse monoclonal antibody, mAb158, which is specific for a soluble protofibril of Aβ(1–42) [1]. This antibody was reported to recognize a different epitope of Aβ than that of the conventional antibodies [2]. The approval of lecanemab represents an accomplishment after the many clinical trial failures of therapeutic anti-Aβ antibodies in the past.
Accordingly, the structural polymorphism of Aβ oligomers, fibrils, and protofibrils is attracting renewed attention as a target for anti-AD drug discovery. AD is a progressive dementia reported by Alois Alzheimer in Germany in 1906, characterized by neuronal loss in the cerebral cortex, cerebral atrophy, amyloid deposits called senile plaques (SPs), and the accumulation of neurofibrillary tangles in neurons. Of these, insoluble fiber-like aggregates of Aβ peptides called amyloid fibrils were found in SPs. Aβ is an aggregation-prone peptide of 40 or 42 amino acids [represented as Aβ(1–40) and Aβ(1–42), respectively] [3,4]. These peptides are excised from the amyloid precursor protein (APP), which is originally involved in synapse formation and repair. Despite the fact that only carboxy-terminal two residues (Ile–Ala) are longer, the monomeric Aβ(1–42) peptide is less soluble and more aggregation-prone than Aβ(1–40). The aggregates formed by Aβ(1–40) are also the causative agents of cerebral amyloid angiopathy (CAA), a disease associated with intracerebral capillary hemorrhage, cerebral infarction, and leukoencephalopathy [5]. In both AD and CAA, the formation of amyloid fibrils is implicated in the onset, progression, and severity of the disease. This is the early (original) amyloid hypothesis. The discovery of APP gene mutations in several families with familial AD made the hypothesis persuasive.
However, growing in vitro Aβ studies raised concerns, as the early amyloid hypothesis could not fully explain the expansion of neuronal death in AD. The low solubility of the amyloid fibrils observed in in vitro experiments made it seem difficult for them to diffuse into the bloodstream. This concern has drawn researchers’ attention to other AD hypotheses. Briefly, there are two major hypotheses regarding the molecular pathogenesis of AD. One is the amyloid hypothesis, which proposes that eccentric aggregates (fibrils or toxic oligomers) of Aβ(1–40) or Aβ(1–42) are the cause of AD [4]. The other is the tau hypothesis, which proposes that an increase in tau aggregates as well as loss of function of microtubule-stabilizing tau in neurons are the cause of neuronal cell death [6]. As mentioned above, a series of discontinued clinical trials to develop therapeutic antibodies targeting Aβ raised questions about the early amyloid hypothesis. Subsequently, many oligomers with low molecular weights, such as globulomers [7], Aβ-derived diffusible ligands (ADDLs) [8], and amylospheroids (ASPDs) [9,10], were reported to be soluble, diffusible, and highly toxic. These are included in what is now widely recognized as the toxic oligomer hypothesis or modified amyloid hypothesis (Figure 1) [11–13].
NMR contributions to elucidating oligomer and fibril formation processes of Aβ peptides with relation to the modified amyloid hypothesis. In solution, Aβ monomers undergo rapid exchange (i) and take up multiple local structural elements. In the schematic, arrows represent β-strands and cylinders represent α-helices. It is believed that different pathways slowly form aggregation nuclei for fiber formation or oligomers that are not involved in fiber elongation (ii), and that fiber elongation occurs from the aggregation nuclei (iv). The pathway that the monomer directly forms fibril without forming the aggregation nuclei (iii) is merely observed yet. The latest modified amyloid hypothesis proposes that oligomers contain highly toxic molecular species that induce neuronal cell death.
In the past three decades, nuclear magnetic resonance (NMR) has played an important role in the verification of the amyloid and the modified amyloid hypotheses, as well as in drug discovery research based on these hypotheses. Until the recent development of high-resolution single-particle analysis using cryo-electron microscopy (cryo-EM), solid-state NMR was the sole technique for determining precise atomic models of Aβ protofibrils because Aβ monomers are highly cohesive and difficult to crystallize. In contrast, the solution NMR method seems suitable for the structural characterization of Aβ monomers, which are considered to have large structural fluctuations in aqueous solutions. The method can also be applied to stable and transient Aβ dimers and trimers, which are observed only under special conditions, such as the presence of surfactants. The use of stable isotope labeling further facilitates these two NMR methods [14].
Currently, only low-resolution electron microscope images exist for toxic oligomers, and precise structural information has not yet been obtained. It should be noted that many researchers believe that these toxic oligomers are not intermediates for amyloid fibrilization. The toxic oligomers seem to be “off-pathway” species distinct from “on-pathway” fibrilization nuclei for the growth of amyloid fibrils (Figure 1). Thus, inhibition of the formation and excretion of toxic oligomers is attracting special attention as a novel drug discovery strategy for AD therapy. Because the detailed structural information of these toxic oligomers is believed to be the origin of the modified amyloid hypothesis, further breakthroughs in this area are eagerly awaited.
Historically, solid-state NMR (SSNMR) analysis has been indispensable for the precise structure determination and model building of Aβ fibrils (Figure 1). SSNMR was the only reliable method to analyze the solid aggregates of Aβ until recent high-resolution single-particle analysis with cryo-EM was achieved. Currently (late September 2022), there are approximately 195,000 entries of three-dimensional structures registered in the Protein Data Bank (PDB; http://www.wwpdb.org/), of which only 151 have been determined by SSNMR. About 10% (15 entries) of these 151 entries correspond to the fibril structure of Aβ. From in vitro studies of Aβ aggregation, a nucleation-dependent polymerization model of Aβ fibrils was proposed. In this model, a small number of aggregation nuclei consisting of several monomers are first formed at the earliest step of fibrilization, and then monomeric Aβ peptides are sequentially attached to the nuclei to elongate them into protofibrils. Since SSNMR is one of spectroscopic methods, it enables us to distinguish homogenous and heterogeneous aggregates from the difference of the NMR spectra (i.e., line widths). As a result, SSNMR permits us to estimate the sample quality of Aβ protofibrils for high-resolution structural determination. SSNMR researchers foresaw the existence of structural polymorphisms in Aβ fibrils based on their spectral differences, especially in Aβ fibrillar samples of different origins [15]. Furthermore, a method for sample preparation with low polymorphism has been established by repeating the cyclic process of breaking, nucleating, and growing. For example, fibrils are chopped by sonication and repolymerized by adding a fresh monomer; the overall process is repeated several times [16,17]. By applying this method to samples derived from the brains of AD patients, polymorphisms characteristic of fibers derived from clinical specimens were revealed [15].
Figure 2 shows structures with characteristic molecular arrangements that were selected from the SSNMR-derived PDB entries of Aβ fibrils (Figure 2A, 2B, 2C), with a recently determined high-resolution AD patient-derived Aβ fibril by cryo-EM (2D). Among them, only Figure 2D illustrates side chains in order to visualize specific side-chain packing within and between monomer molecules with two-fold axes of symmetry. In the structure, a single long β-strand formed by the main Aβ chain undulates significantly up and down the vertical direction along the strand axis. Subsequently, the first layer was composed of two or three Aβ molecules, and the next layer aligned precisely with the first layer and repeatedly stacked upon each other to form parallel intermolecular β-sheets. This structural arrangement is called as a cross-β structure, which is characteristic of amyloid fibrils.
Selected structures of amyloid fibrils determined by SSNMR and cryo-EM that show polymorphism. A: PDB ID: 2LMN, Aβ(1–40), SSNMR. B: PDB ID: 2LMQ, Aβ(1–40), SSNMR, unusual structure with 3-fold symmetry axis. C: PDB ID: 6OC9, Ser8 phosphorylated Aβ(1–40), SSNMR. D: PDB ID: 6SHS, Aβ(1–42), reassembled from fibers extracted from an AD patient’s brain as a nucleus and analyzed with cryo-EM.
Interestingly, there are few common side-chain interactions among polymorphisms, and the types of Aβ molecule arrangements are distinct from each other in the different fibril polymorphs. In addition, none of monomers with the similar conformation with the common side-chain packing has been observed by the other method, such as solution NMR (Figure 2, Figure 3). Thus, the fibrilization process seems to only be explained by the nucleation-dependent fibril polymerization model, in which the proteinous fiber elongates using the fibrilization nucleus consisting of a few layers at the start of the fibril as a structural template [18,19]. These observations also support the idea that there are at least two types of Aβ fibrils, one consisting of a basal dimer unit (Figure 2A, C, D) and the other consisting of a basal trimer unit (Figure 2B).
Selected solution and crystal structures of Aβ monomers. A: PDB ID: 1BA4, Aβ(1–40) in the presence of SDS micelles. B: PDB ID: 1HZ3, Aβ(10–35) in neutral aqueous solution. C: PDB ID: 1IYT, Aβ(1–42) in 80% HFIP, often used for MD calculations. D: PDB ID: 1Z0Q, Aβ(1–42) in 30% HFIP. E: PDB ID: 2LFM, Aβ(1–40) in neutral phosphate buffer, F: PDB ID: 6SZF, Aβ(1–42) in 50% HFIP. Aβ(1–42) has low solubility and is highly aggregative. G: PDB ID: 6RHY, Aβ(1–42) in DPC micelles. H: PDB ID: 6WXM, Aβ(16–36) determined by X-ray crystallography.
What, then, is the solution structure of the Aβ monomer that is the component for building these amyloid fibrils? Solution NMR is a suitable method for solving this problem, as long as their molecular weights are not too high. In other words, solution NMR has a critical limit for analyzing Aβ fibrils and Aβ oligomers due to the upper limit of the molecular weight that can be analyzed. However, solution NMR can be combined with stable isotope labeling to capture the three-dimensional structures and equilibrium states of molecules in aqueous solutions, organic solvents, and surfactants. As of September 2022, the PDB has over 15 NMR structure entries for Aβ monomers [including Aβ(1–40), Aβ(1–42), and their partial peptides], more than 80% of which are adopted into α-helical structures throughout or in the middle of the molecule, with a large variety of the peripheral residues outside of the helix (Figure 3A–3F). However, when we assessed each measurement condition in addition to simply looking at the ribbon diagram, we found that most of these structures were not in “pure” aqueous conditions. Instead, there are many solution structures in organic solvents containing trifluoroethanol (TFE) and hexafluoroisopropanol (HFIP) (Figure 3C, 3D, 3F) and micelles containing sodium dodecyl sulfate (Figure 3A). There are few examples of the structural determination of Aβ monomers in neutral or physiological conditions (see Figure 3B).
HFIP is a strong solvent often used in peptide chemistry and a strong inducer of α-helices. In contrast, in the circular dichroism (CD) spectrum of Aβ in a neutral aqueous buffer solution close to physiological conditions, a typical random-coil-like spectrum without either α-helices or β-strands is observed reproducibly. This observation is not consistent with α-helical structures. The only PDB entries for solution structures in aqueous solvents were PDB ID: 1HZ3 (Figure 3B) and PDB ID: 2LFM (Figure 3E). Combined with the fact that the 1H NMR chemical shift of the amide group of Aβ(1–42) in aqueous solution is observed in the region typical of IDPs (7.5–8.5 ppm), it is reasonable to consider that Aβ does not adopt a stable and compact conformation throughout the molecule.
There is another report that Aβ(1–42) in aqueous solution may contain more β-strands than expected. Wälti et al. reported the transient β-rich Aβ(1–42) monomer in a sample mixture of the Aβ monomer and its micellar aggregates after several days of sample incubation at 4°C [20]. They carefully analyzed the 13Ca secondary chemical shifts, exact nuclear Overhauser effects among backbone protons, and three-bond coupling constants between HN and Hα, all of which were subsequently used to determine the secondary structure. In their model, Aβ monomers contained two short β-strands in the N-terminal region of the molecule, followed by two longer β-strands in the middle and C-terminal regions, one of which overlapped with the KLVFF sequence (residues 16–20).
Many researchers have endeavored to understand the mechanism of the earliest step of amyloid fibril formation by employing molecular dynamics (MD) simulations of Aβ peptides with explicit solvent or water models. However, some reports used the α-helix-rich solution structure of Aβ determined in the above-mentioned HFIP/TFE mixed solvent as the initial structure for MD. For this purpose, the peptide was settled in a solvent box by replacing HFIP/TFE with pure water molecules with poorly disclosed protocols. Since HFIP/TFE and water differ greatly in terms of dielectric constant and hydrogen bonding capability, a sufficiently long simulation time is required for the system to reach an equilibrium state. Special care must be taken when simulating of Aβ in aqueous solutions starting from Aβ coordinates derived from NMR experiments in organic mixed solvents. This approach is not recommended by the author.
As mentioned above, HFIP is a strong solvent that solubilizes less-soluble peptides and induces the formation of α-helical structures within them. Historically, HFIP had been believed to be a monomerization reagent for Aβ [21]. However, Shigemitsu et al. overturned this conventional assumption by observing an Aβ dimer in HFIP with NMR [22]. In that report, concentration-dependent NMR signal changes were observed, and its self-association constant in HFIP was determined to be approximately 1 mM. The NMR data also revealed its secondary structure, which is dominantly α-helical, adopted into a dumbbell-like structure consisting of two short helices, which then form a symmetric dimer. In many biochemical experiments, HFIP treatment was used. However, solvent removal with a vacuum concentrator from a small HFIP-dissolved Aβ aliquot will certainly result in a monomer-dimer mixture after concentration. Shigemitsu et al. doubted that this was a cause of low reproducibility in subsequent amyloid or oligomer formation experiments.
Tetramer and Octamer in Dodecylphosphocholine (DPC) Micelles [23]Solution NMR in dodecylphosphocholine (DPC) micelles reveals a unique tetrameric structure consisting of a β-hairpin formed by a single molecule of Aβ combined with a single β-strand formed by half of the molecule (PDB ID: 6RHY) [23]. In this sample, Aβ monomers were adopted into two different conformers, and they were arranged into a symmetrical dimer-of-dimer. The β-hairpin part of the structure seems similar to the other hairpin structure (PDB ID: 6WXM, Figure 2H) formed in the center of the Aβ(16–36) molecule identified with X-ray crystallography [24].
Oligomer Structure—Amylospheroid (ASPD)Amylospheroid (ASPD) is an extensively studied toxic oligomer formed from either Aβ(1–40) or Aβ(1–42). 1H–15N heteronuclear single quantum coherence (HSQC) spectra of ASPD were reported [25]. In solution, the signals originating from the core part of ASPD were not visible, probably because of its high (~150,000) molecular weight. Only a limited number of signals from the N-terminal flanking region of Aβ(1–40) were observed. The signals were in a single state and very similar to those of free Aβ monomers in terms of their chemical shifts. The outline of these visible and invisible signals were illustrated in Figure 4. Accordingly, the authors concluded that the visible region was highly mobile and in a chemical exchange state between free monomers and surface-attached ASPD molecules [25]. On the other hand, the residues of invisible signals may be attributed from the contact residues for the monomer to the oligomer. Note that this observation (that approximately 10 N-terminal residues were flexible) is similar to the model of the contact residues of the Aβ monomer to the amyloid protofibril surface, which were determined by the dark-state exchange saturation transfer (DEST) NMR experiment described later [26].
Comparison of the Aβ(1–40) monomer to oligomer/protofibril contact residues determined with the different NMR methods. ASPD: Migrated HSQC signals of surface-attached molecules to amylospheroid [25]. DEST: Residues showing high spin diffusion rate of surface-attached molecules to protofibrils determined with DEST spectroscopy [28]. WSTDHSQC: Residues with the reduced signal intensities of surface-attached molecules to protofibril analyzed with WSTDHSQC [29].
To elucidate the early stages of the fibril formation process, additional advanced NMR techniques have been developed and applied to the Aβ system. In particular, saturation transfer and spin diffusion techniques have been the focus. It is known that the efficiency of magnetization transfer for cross-relaxation is high for molecular aggregates with high molecular weights. At the same time, high molecular weights hamper the detection of NMR signals originating from such large aggregates. Therefore, sample conditions should be carefully set to the state in which chemical exchange between the monomer molecule and the protofibril occurs. Then, only NMR signals derived from the monomeric components in the sample can be observed. As a result of magnetization transfer between the monomer and the aggregates, we can now observe a certain attenuation of signal intensity. The data can semi-quantitatively determine the interfacial contact residues formed from the interaction between the monomer and the aggregate. The amyloid formation processes studied with saturation transfer or chemical exchange-based NMR techniques were comprehensively reviewed by Ahmed and Melacini [27].
Dark State Exchange Saturation Transfer (DEST) Spectroscopy [28]DEST spectroscopy was developed to investigate the residue-specific contact between monomeric Aβ and protofibril in a chemical-exchange state. The signals were detected from the free monomer species; however, they were in equilibrium with the protofibril, the signals of which were invisible because of its large molecular weight. During the pulse sequence, a spin-locking pulse was applied to 15N, and the transverse relaxation was allowed to occur. Note that T2 relaxation of large molecules is 103–105 times faster than that of monomers, and the difference in T2 relaxation times is quantified by changing the offset frequency of 15N spin-lock. In Aβ/protofibril interaction, DEST experiments showed two broad peaks, the tops of which were residues 18 and 32. The outline of the result from the DEST experiment was summarized in Figure 4. The results were interpreted as indicating that the KLVFFA core sequence (residues 16–21) and the residues of the C-terminal half of Aβ(1–42) were in contact with the protofibril. In contrast, approximately 12 residues from the N-terminus were flexible and separated from the protofibril surface. Fawzi et al. further interpreted these contact sites as corresponding to the two β-strand portions located at the center of the molecule toward its C-terminus.
Water STD-HSQC/Methyl STD-HSQC [29]Ahmed et al. developed a series of saturation transfer difference (STD)-based NMR techniques for investigating the mechanism of amyloid inhibition by epigallocatechin gallate (EGCG): water STD-HSQC and methyl STD-HSQC (WSTDHSQC and MeSTDHSQC, respectively). In the methods, first, bulk magnetization from water to the fast-exchanging amide protons of Aβ monomer, and then propagated through multiple spin diffusion steps through the contact site to the aggregates. Finally, the remaining magnetization on the Aβ monomers is detected using an HSQC detection block. As mentioned above, the spin diffusion within the structured Aβ aggregates is more efficient than the flexible monomers, similar to the basic STD experiment. These experiments are useful to distinguish solvent-exposed residues and the contact residues of the Aβ monomer attaching to the oligomer surfaces. These interpretations are summarized in Figure 4.
Possible Early Events in Aβ FibrilizationTogether with other findings, I propose the following molecular processes in the early stages of Aβ fibrilization.
(1) Intrinsically disordered Aβ monomers are partially and transiently adopted into α-helical conformations, probably on cell membranes.
(2) With increasing local concentration, α-helical-rich conformations are stabilized, and then form helical dimers.
(3) Aβ is transformed into β-hairpin.
(4) The β-hairpin is further transformed into an intermolecular cross β-strand structure.
(5) Finally, amyloid fibrils begin to elongate from the aggregation nuclei.
Although all of the above processes were NOT completely observed by NMR and other spectroscopic experiments, the several lines of experiments supported the hypothesis. For example, the presence of α-helical conformations of Aβ monomer in membrane mimicking conditions (HFIP, TFE, and MeOH) were observed, described in the above section of “Structure of Aβ monomer Observed with Solution NMR”. By CD experiments, α-helical conformations of Aβ(1–40) was observed in the presence of phosphatidylglycerol vesicles [30]. The formation of α-helical dimer in HFIP was also observed by NMR [22]. Accordingly, under the condition mimicking fibril formation with the increasing temperature, Yamaguchi et al. observed the increased β-strand propensity at the several residues by NMR [31]. These residues were consistent with the β-hairpin structures in several crystal structures (Figure 3). Thus, taking all these indications into account, I propose a possible scenario for the early events in Aβ fibrilization.
Stable isotope incorporation into protein samples is an essential technique for fully exploiting the strengths of NMR [14]. It should be noted that the conventional chemical peptide synthesis is the first choice of sample preparation for Aβ studies in the other biophysical method. However, due to cost issues, chemical preparation of isotopically-labeled Aβ peptides was not always preferred. Only in limited cases, the methods were used for residue-specific/atom-specific isotope labelling of Aβ peptides for SSNMR experiments, with the accurate measurement of distances using 13C-13C dipolar coupling [32]. Accordingly, many researchers preferred stable isotope incorporation into Aβ peptides by bacterial expression.
In bacterial cytoplasm, Aβ(1–42) is easily degraded by endogenous bacterial proteases because of its nature of intrinsically disordered proteins (IDPs), although Aβ(1–42) is extremely insoluble and aggregation-prone in vitro. Many efforts have been made to prepare sufficient amounts of isotope-labeled Aβ samples in the E. coli expression system. Selected methods are summarized in Table 1. All of the reported Aβ expression systems are designed as fusion proteins with tag(s) attached to the N-terminus to facilitate affinity purification and stable protein accumulation. In addition, many researchers have decided to express these fusion proteins in bacterial inclusion bodies (IBs). After the fusion proteins were expressed in the IBs, they were washed out and solubilized with high concentrations of urea, guanidinium hydrochloride, or surfactants.
| Strategy1 | Target | Purification/solubility/delivery tags | Tag location2 | Purification | Tag–removal | Ref. |
|---|---|---|---|---|---|---|
| SF | Aβ(1–42) | [His]63–[NANP4]19–[RSM5] | NT | IMAC6 | CNBr7 | [33] |
| SF | Aβ(E3Q–40) Aβ(E3Q–42) | [His]6–[NANP]19–[TEV]8 | NT | IMAC | TEV protease | [34] |
| SF | Aβ(1–42) | MBP9–[NANP]3–[TEV] | NT | amylose resin | TEV protease | [35] |
| IB | Aβ(11–26)hs10 | KSI11–[MET]–Aβ(11–26)M x3–[His]6 | NT + CT | IMAC | CNBr | [16] |
| IB12 | Aβ(1–40) | GST13–[thrombin]14 | NT | glutathione resin | thrombin | [36] |
| IB | Aβ(1–40) | [His]6–yeast Ub15 | NT | IMAC | YUH116 | [37] |
| IB | Aβ(1–40) Aβ(1–42) | [His]6–human Ub | NT | IMAC | YUH1 | [22] |
| IB | Aβ(1–40) Aβ(1–42) | [His]6–IFABP17–[FXa]18 | NT | IMAC | Factor Xa | [38,39] |
| SF | Aβ(1–40) Aβ(1–42) | [His]6–Spidroin NTD19–[TEV] | NT + CT | IMAC | TEV protease | [40] |
| SF | Aβ(1–42) | [His]6 | CT | IMAC | not removed | [41] |
| SF | Aβ(1–42) | [His]6–[thrombin]–Aβ(1–42)–[Lys]6 | NT + CT | IMAC | not removed | [42] |
(1) SF: soluble fraction, IB: inclusion body, (2) NT: N–terminal, CT: C–terminal, (3) [His]6: hexahistidine tag, (4) NANP: repetitive tetrapeptide from malaria (Plasmodium falciparum) surface protein, (5) RSM: cyanogen bromide cleavage site, (6) IMAC: immobilized metal affinity chromatography, (7) CNBr: cyanogen bromide, (8) [TEV]: tobacco edge virus (TEV) protease cleavage site, (9) MBP: maltose binding protein, (10) Aβ(11–26)hs: Aβ(11–26) with a C–terminal additional homoserine, (11) KSI: ketosteroid isomerase (as inclusion body delivering tag), (12) this fusion protein was refolded by sodium sarcosinate – Triton X100 method, (13) GST: glutathione S–transferase, (14) [thrombin]: thrombin cleavage site, (15) Ub: ubiquitin, (16) YUH1: yeast ubiquitin hydrolase 1, (17) IFABP: intestinal fatty acid binding protein (as expression accumulation tag), (18) [FXa]: Factor Xa cleavage site, (19) Spidroin NTD: Nephila calvipes spidroin N–terminal domain (as solubility tag)
Immobilized metal affinity chromatography is commonly used to purify Aβ that is fused to a hexahistidine tag in the presence of solubilizers. However, these high concentrations of solubilizing agents should be removed with sequence-specific proteases, such as TEV or Factor Xa, before the tags are removed. In addition, the processed Aβ peptides are usually subjected to a final purification step using reversed-phase HPLC. The process requires two or more additional days than the standard protocols for soluble globular proteins. It should be noted that the addition of several repetitive amino acids to the C-terminus of Aβ, for example, His x6 or Lys x6, has been shown to drastically improve the solubility of the fusion peptides. However, in these cases, the additional C-terminal residues were not designed to be removed.
As described above, amyloid fibril formation occurs in a nucleus-dependent manner. Since Aβ monomers are intrinsically disordered in aqueous solutions under physiological conditions, the monomer must be locked into a specific conformation to form either toxic oligomers or protofibrils with a certain entropy loss. Therefore, attempts have long been made to explore low-molecular-weight compounds that inhibit this structural transition. However, the Aβ monomer does not have a clear drug-binding pocket. It appears to be difficult to apply the conventional “a key and a keyhole” drug discovery model. As a result, a highly potent inhibitor has not yet been found.
Instead, the authors evaluated polyphenols and osmolytes (sugars), which have been reported to have aggregation inhibitory activity, using NMR titration experiments while observing the two-dimensional NMR spectra of 15N-labeled Aβ(1–42) as an indicator [33]. Very high (>0.5 M) concentrations of sugars, such as trehalose and sucrose, are known to inhibit nucleation and subsequent fibril growth [33]. These conditions also resulted in slight chemical shift changes of the Aβ monomers in NMR. Phenolic food components, such as curcumin and EGCG, inhibit Aβ(1–42) fibrosis at much lower concentrations than those of trehalose and sucrose. Under nearly equivalent conditions, chemical shift changes in the amide group of Aβ(1–42) upon polyphenols were observed, but we found that the residues showing the signal changes were different from those given by the osmolytes. In addition, the chemical shift changes of the Aβ(1–42) monomer between 5°C and 37°C were also observed. Note that at 5°C, Aβ(1–42) at the NMR-measured concentrations and conditions remained in a monomeric state for several days, but it quickly started to form aggregates and to turn cloudy when the temperature was increased to 37°C.
To visualize these chemical shift changes, we employed principal component analysis of the chemical shift of the 2D NMR signal in an attempt to explain the mechanism of inhibition of amyloid formation by the equilibrium shift of the structural ensemble of the Aβ monomer [33]. In general, the results of NMR titration experiments with drug targets and candidate compounds are often visualized by mapping the magnitudes of chemical shift changes with different color gradients on the protein surfaces (to the left of the dashed line in Figure 5). However, this method assumes that the target protein is rigid and does not change its conformation upon binding with the compound. The Aβ monomer is an IDP, and this is not the case. In contrast, if the target is an IDP, both compound and solvent conditions may change the equilibrium among multiple transient local conformers of the target IDP (to the right of the dashed line). Thus, we propose that principal component analysis is expected to be an alternative to chemical shift mapping, while chemical shift mapping representation is somehow misleading.
Comparison of complementary methods to visualize NMR titration experiments. The rigid drug target, which is expected not to change its conformation upon compound binding (left), and the NMR-based drug screening against IDP (right). Only when the target is rigid enough can the chemical changes of the amide signals be interpreted as an effect of the proximity of the compound. In this case, the binding site can be conventionally visualized by mapping the amount of change onto the 3D structure (lower left). In the case of IDP, not only direct binding of the compound but also certain changes in liquid conditions caused by the compound may induce an equilibrium shift between different local conformations. In this case, color mapping to the hypothetical structure is not appropriate.
Because society is aging, both the prevention and treatment of dementia are becoming increasingly important, not only in Japan but also in China, Europe, and the United States. In addition to the modified amyloid hypothesis, many other hypotheses have been proposed regarding the pathogenesis of AD, including the tau hypothesis, the periodontitis bacteria hypothesis [34], and the virus-infection hypothesis [35]. These confusing hypotheses make it difficult to narrow down to a definitive anti-AD drug discovery strategy. The approval of lecanemab/BAN2401 is good news, but given the high cost of antibody therapies, there is an urgent need to develop small-molecule therapies. However, it is relatively difficult to handle the Aβ(1–42) peptide and its oligomer in the laboratory due to its low solubility and high aggregation properties. In addition, a lack of three-dimensional structures of toxic Aβ oligomers hampers the application of modern structure-guided drug discovery strategies. The author hopes that the rapid progress of new technologies such as cryo-EM and ultra-high-field NMR spectroscopy will be utilized extensively in future Aβ research.
H.H. is the founder of a Nagoya University-based spinoff startup company, BeCellBar LLC.
Conceptualization, H.H.; writing—original draft preparation, H.H.; writing—review and editing, H.H. All authors have read and agreed to the published version of the manuscript.
The evidence data analyzed during the current study are available from the corresponding author on reasonable request.
This work was partly supported by Salt Science Research Foundation (Grant No.1222) and the Mishima Kaiun Memorial Foundation (Grant No.H25-153) in Japan. The authors would like to thank Scribendi Editing Services (https://www.scribendi.com/) for the English language review.
