Modulating the aggregation of amyloid proteins by macrocycles

The aggregation of amyloid proteins has been suggested to be the main cause of multiple human disorders; for example, amyloid β aggregates in Alzheimer's disease and α‐synuclein aggregates in Parkinson's disease. In the search for therapeutic medicines, many molecules have been discovered and developed to modulate the aggregation of amyloid proteins. This century has witnessed the flourishing growth of supramolecular chemistry, and some biocompatible macrocycles have been proven to inhibit the aggregation of some amyloid proteins via host‐guest interactions and could thus be used for the prevention or treatment of related diseases. Here, we review the application of macrocycles in modulating the aggregation of amyloid proteins.


INTRODUCTION
Because proteins are the primary participants in cellular activities, protein dysfunction leads to many types of diseases. In addition to genetic factors that may cause the abnormal expression of proteins, the misfolding and/or aggregation of some proteins will also reduce the level of normal proteins and/or increase the level of harmful species. [1] Abnormal protein aggregates are often termed amyloid fibrils, and the proteins that form these fibrils can be defined as amyloid proteins. Although the direct link between the aggregation of amyloid proteins and the pathological processes of some diseases remains elusive, [2,3] it is widely accepted that the major causative factors are the oligomers and fibrils that form during the aggregation process and not the soluble monomers [1] .
The amyloid cascade hypothesis, a theory approximately 30 years old regarding Alzheimer's disease, believes that amyloid β (Aβ) aggregates can not only impair synaptic activity [4] but also trigger the aggregation and downstream toxicity of tau protein. [5][6][7] The accumulation of α-synuclein (α-syn) has been observed in Parkinson's disease and other Lewy body diseases, [8] and α-syn oligomers and fibrils are believed to play key roles in these diseases. Moreover, recent studies suggest that some amyloid proteins, including Aβ and α-syn, have prion-like properties, which means that their aggregates can self-replicate and spread from cell to cell. [9] In addition to the above-described disease-related amyloid proteins, some protein drugs tend to aggregate and can also form amyloid fibrils. These tendencies hinder their formulation and decrease their effective concentration in the body. [10] For example, insulin, which is the cornerstone for the treatment of diabetes, [11] aggregates to form fibrils under various conditions. [12,13] Inhibiting the aggregation of such drugs to form amyloid fibrils and stabilizing them as effective monomers will increase their efficiency in the treatment of related diseases.
Therefore, modulating the aggregation of amyloid proteins has been recognized as a potential therapeutic method for many diseases. Some inhibitors of Aβ, including EGCG (NCT03978052, NCT00951834) and trehalose (NCT04663854), have entered clinical trials for Alzheimer's disease. [14] Tafamidis meglumine, which can stabilize the amyloid protein transthyretin (TTR) in tetramer form and inhibit the formation of fibrils, [15] has been approved by the food and drug administration (FDA) for the treatment of transthyretin amyloid cardiomyopathy. [16] Many other compounds, including small molecules, [17][18][19] peptides, [20,21] and nanoparticles, [22] have also been found to modulate the aggregation of amyloid proteins. Here, we focus on synthetic macrocycles that possess interior cavities and exterior functional groups. [23] The characteristic structural properties of these macrocycles make them suitable for the recognition and hosting of diverse guest molecules, particularly peptides/proteins and other biomolecules. [24][25][26][27] The macrocycles are somewhat similar to antibodies because they exhibit strong and specific binding with target molecules but have lower molecular weights. Compared with small molecules, macrocycles may exhibit better specificity in binding with proteins due to their larger binding surfaces. [28] The rigid skeletons of macrocycles also make them different from flexible peptides. These widely used macrocycles, including cucurbiturils, calixarenes (CAs), and cyclodextrins (CDs), are generally safe and biocompatible, which supports their broad application in biological systems. [24][25][26][27] Amyloid fibrils generally possess β-sheet structures that are stabilized mainly by hydrogen bonds and π-π stacking. [29] Some macrocycles can bind with specific functional groups of amyloid proteins through host-guest interactions and subsequently interfere with the interactions needed for fibril formation or sterically hinder the associations between amyloid proteins. Therefore, these macrocycles can modulate the aggregation of amyloid proteins and may be used for the prevention or treatment of related diseases. Here, we summarize the macrocycles that could function as modulators of amyloid aggregation and offer some personal opinions regarding their future development.

AGGREGATION OF AMYLOID PROTEINS
The detailed aggregation processes of amyloid proteins have not been fully determined due to the difficulties in characterizing and quantifying transient and metastable oligomers. In general, monomers of amyloid proteins first aggregate to form oligomers, and the oligomers are then converted to fibrils, which is a process called primary nucleation. Then, fibrils can subsequently bind monomers and grow in length, which is a process termed elongation. The fibrils can also act as catalysts to generate new fibrils from monomers, and this process is termed surface-catalyzed nucleation or secondary nucleation. Therefore, the modulators (mainly inhibitors) of amyloid aggregation may bind to monomers, oligomers, or fibrils and then interfere with various microscopic steps (Figure 1A). [30,31] Several conventional biophysical methods are widely used to characterize the aggregation of amyloid proteins, including thioflavin T (ThT) kinetics, CD spectroscopy, transmission electron microscope (TEM), and atomic force microscope (AFM). [32] ThT is a fluorescent dye that yields approximately 1000-fold fluorescence intensity when incorporated into the β-sheet structures of amyloid fibrils. [33] Therefore, ThT is widely used to monitor the aggregation process of amyloid proteins. Results from CD spectroscopy reflect the secondary structures of peptides/proteins; thus, it can be used to confirm the formation of β-sheet structures. The morphology and nanostructures of amyloid aggregates (particularly fibrils) can be directly observed by TEM and AFM. These methods do not provide comprehensive information about oligomers and cannot reveal the detailed microscopic aggregation processes.
Other techniques have been developed in recent years to obtain a better understanding of oligomers and aggregation processes. For example, native mass spectrometry and ion mobility mass spectrometry enable the detection and analysis of amyloid oligomers. [34,35] Dot blot has been applied to detect antibody-specific oligomers, [36] and solidstate nuclear magnetic resonance spectroscopy has been used to reveal the structures of amyloid oligomers. [37] In addition, single-molecule fluorescence spectroscopy offers quantitative information about the formation of oligomers at the singlemolecule level. [38] Many molecular probes, including Congo red derivatives and aggregation-induced emission molecules, have been developed to complement ThT in studies of the aggregation of amyloid proteins. [39] Knowles et al. developed kinetic models to describe the microscopic steps of fibril formation. [30,40] Microscopic aggregation processes can be viewed as chemical reactions; thus, chemical kinetics can be applied to determine the rate laws of different steps. With the rate law and reaction time in hand, theoretical predictions of the reaction profiles, including quantities of monomers, oligomers, or fibrils, can be obtained through mathematical calculations. The data obtained from chemical kinetics simulations are consistent with the results from some biophysical experiments, particularly ThT kinetics. The inhibitory mechanisms of different molecules could be confirmed by chemical kinetics because the effects on different microscopic steps will induce different changes in the observed ThT kinetics curves (Figure 1B). [31,41] In fact, a single molecule might interfere with multiple microscopic steps because the molecule may interact with more than one amyloid species during the aggregation process. [42] 3

CUCURBITURILS FOR MODULATING AMYLOID AGGREGATION
Pumpkin-shaped cucurbit[n]urils (CB[n], n = 5-8, 10, 14) [26] are products from the reaction of glycoluril with formaldehyde ( Figure 2A). [25] Increases in n from 5 to 10 can increase the cavity volume from 82 Å 3 to 870 Å 3 , which allows cucurbiturils to bind with guest molecules of various sizes. [25] Cucurbiturils have one nonpolar cavity and two electronegative portals, which facilitates their strong F I G U R E 1 (A) Microscopic steps of Aβ aggregation. Reproduced with permission: Copyright 2018 National Academy of Sciences. [31] (B) Inhibitors may interfere with different microscopic steps, and changes in the observed macroscopic kinetics are plotted based on model simulations. Reproduced with permission: Copyright 2014 Elsevier Ltd. [30] F I G U R E 2 (A) Synthetic route and molecular structures of CB[n]. Reproduced with permission: Copyright 2015 American Chemical Society. [25] (B) Interactions between CB [7] and the N-terminal Phe residue. Reproduced with permission: Copyright 2011 American Chemical Society [46] binding with molecules containing both hydrophobic groups and positively charged groups. Both the hydrophobic interactions inside the cavity and the electrostatic interactions at the exterior contribute to the specific and strong binding of cucurbiturils with some guest molecules. For example, CB [7] binds to adamantylamine with K a = 4.23 × 10 12 M -1 , whereas its binding affinity for 1-adamantanecarboxylic acid decreases to K a = 3.23 × 10 8 M -1 . [43] Because the side chains of aromatic amino acids are hydrophobic, and some basic amino acids are positively charged under physiological conditions, cucurbiturils have the capability to bind with different peptides and proteins. [26,28,44,45] Therefore, it is reasonable to infer that cucurbiturils could interact with some amyloid proteins and may modulate their aggregation.
Anderson et al. developed a method named supramolecular PEGylation to stabilize biopharmaceuticals, including insulin. [10] The researchers first synthesized CB [7]-N 3 using a previously reported method. [50] Then, strain-promoted click chemistry [51] was used to conjugate CB [7]-N 3 with polyethylene glycol (PEG), and CB[7]-PEG was obtained ( Figure 3). CB [7]-PEG still preserved the ability to bind with insulin. Owing to the good solubility of both CB [7] and PEG, the aggregation of insulin was greatly inhibited by CB [7]-PEG under physiological conditions with continuous agitation (1:1, insulin:CB [7]). After aging for 100 d with CB [7]-PEG, no significant aggregation was observed, and the activity of insulin was retained. In addition to insulin, the stabilization of glucagon and an antibody for human CD20 was also achieved with CB[7]-PEG. [10] The researchers also found that the formulation of insulin with CB[7]-PEG could maintain a low blood glucose level in STZ diabetic mice, which suggests a depot effect of CB [7]-PEG on insulin in vivo. The supramolecular PEGylation method has been further extended to insulin analogs [52] and a coformulation of insulin with pramlintide. [53] F I G U R E 3 Molecular structure of CB [7]-PEG [10] Li et al. demonstrated that CB [7] inhibits the fibrillation of another potential protein drug, human calcitonin (hCT). [54] hCT could be a good substitute for salmon calcitonin, which is used for the treatment of some bone-related diseases. However, hCT suffers the same shortcomings as insulin. The researchers found that 25-fold CB [7] is needed to fully suppress the aggregation of hCT because hCT contains no Nterminal Phe residues. Mutating the aromatic amino acids (Tyr and Phe) in hCT to Ala decreases its binding affinity with CB [7], as expected. The researchers injected hCT or a mixture of hCT with CB [7] into rats and observed better calcitonin release in the latter group. Moreover, the authors proved that CB [7] is not immunogenic and could decrease the immunogenicity of hCT.
In addition to CB [7], other cucurbiturils have also been used to modulate the aggregation of amyloid proteins. Kim et al. reported that CB [6], which is barely soluble in water, could form host-guest complexes with Lys-containing amyloid proteins, including insulin, human islet amyloid polypeptide, hen egg lysozyme, and Aβ 1-40/42 . Due to the low solubility of CB [6], the addition of excess CB [6] (1:10, protein:CB [6]) to amyloid protein solutions makes them insoluble, which drives the formation of long and homogeneous fibrils. [55] CB [8] is less soluble than CB [7] but more soluble than CB [6]. [25] CB [8] has a larger cavity than CB [7] and can dimerize peptides/proteins with Phe residues, particularly Nterminal Phe. [26,56,57] Scherman et al. found that CB [8] accelerates the fibrillar processes of Aβ 1-42 (1:3, Aβ:CB [8]), but the cytotoxicity of Aβ 1-42 is decreased by incubation with CB [8]. [58] Li and coworkers focused on the structures and properties of amyloid proteins, particularly proteins with posttranslational modifications. [59][60][61][62][63] In this scenario, Li et al. investigated the effect of CB [7] and CB [8] on the aggregation of N-terminal-truncated Aβ 4-40, [64] which was reported to be abundant in the brains of patients with Alzheimer's disease. Evidence from early studies suggests that Aβ 4-X may exceed Aβ 1-X in amyloid plaques. [65,66] Owing to the N-terminal Phe residues ( Figure 4A), the researchers found that both CB [7] and CB [8] significantly inhibit the aggregation and cytotoxicity of Aβ 4-40 even at a ratio of 1:1, and these effects are better than their effects on Aβ  . Moreover, the authors used the abovementioned chemical kinetics analysis to clarify the inhibitory mechanisms of CB [7] and CB [8]. Mathematical simulations have revealed that the fibril end binding mechanism best fits the results from ThT kinetics. Thus, it is believed that CB [7] and CB [8] may bind with the fibril ends of Aβ and suppress the elongation process ( Figure 4B). The same inhibitory mechanism has also been reported for a protein inhibitor of α-syn. [67] In summary, cucurbiturils generally bind to amyloid proteins with medium to high binding affinities, depending on the sequences of the amyloid protein. Thus, cucurbiturils may exhibit specificity toward different amyloid proteins. The chemical functionalization of cucurbiturils is expected to increase their water solubility and make their in vivo applications possible (Table 1).

CAs FOR MODULATING AMYLOID AGGREGATION
Calix crater-like CAs (CnA, n = 4, 5, 6, 8…) are synthesized by the reaction of phenol derivatives with formaldehyde. [24,68] A greater number of phenol units are associated with a larger cavity size. A large amount of CAs with different structures can be obtained due to the multiple sites available for modification. Different substitutions on both the upper and the lower rims can be easily achieved. CAs also possess different conformational structures, which further extend their structural diversity ( Figure 5). Changing the starting materials from phenol derivatives to aromatic heterocycles leads to heteracalixaromatics. [69,70] The structural diversity of CAs makes it possible for them to recognize and bind to different types of molecules, including cations, anions, organic molecules, and biomacromolecules, which results in an almost unlimited numbers of applications. [24,70,71] Several CAs have been reported to inhibit the aggregation of Aβ and insulin. [24] More biocompatible CAs are expected to interfere with the aggregation of amyloid proteins, which offers potential therapeutic methods for related diseases.
Mohanty et al. reported the inhibitory effects of psulfonatocalix[4/6]arene (SC[4/6]A) on the aggregation of insulin ( Figure 6). [72] Not only was the aggregation process suppressed, but the mature insulin fibrils could also be disin-   [73] Compared with insulin, a higher ratio of SC[n]A is needed to achieve significant inhibition of Aβ 1-42 . Leblanc et al. demonstrated that resorcinarene strongly inhibited the aggregation of Aβ 1-40/42 at low concentrations (1:5, Aβ:resorcinarene). [74] Resorcinarenes, unlike SC[n]A, are the reaction products of resorcinol derivatives with aldehydes. In addition to sulfonate on the aromatic rings, the resorcinarene used is modified by thiomethyl groups at the methylene bridge carbon. The results from ThT kinetics, CD spectra and AFM confirm that the resorcinarene inhibits the fibrillation of Aβ 1-40/42 at a 1:1 ratio. Moreover, cytotoxicity experiments using sea urchin embryos have revealed that the resorcinarene is not toxic at the effective concentrations. The cytotoxicity of Aβ 1-42 is also reduced by the resorcinarene. MD simulations and molecular docking have been performed to study the binding models between Aβ 1-42 fibrils and the resorcinarene. The researchers have found that the thiomethyl groups of the resorcinarene form nonpolar interactions with Ala, Leu, and Val residues on Aβ 1-42 fibrils.
More recently, the same research group studied the effects of different tail-engineered resorcinarenes on the aggregation of insulin. [75] The resorcinarene itself exerts a strong inhibitory effect on insulin at a ratio of 1:0.2 (insulin:resorcinarene). However, changing the tail from F I G U R E 6 Molecular structures of the calixarenes reported to modulate the aggregation of amyloid proteins thiomethyl to propyl or butyl results in disappearance of the inhibitory effect. In fact, butyl derivatives of resorcinarene accelerate the aggregation process. MD simulation and docking results suggest that the resorcinarene may bind to both insulin monomers and dimers and then prevent the dimerization and fibrillation of insulin.
Pappalardo et al. constructed a conjugate of peptide and calix [4]arene, and this conjugate exerts better inhibitory effects on the aggregation and cytotoxicity of Aβ 1-42 than the peptide or calix [4]arene alone (1:5, Aβ:conjugate). [76] The sequence of the peptide is gly-pro-gly, lys-leu-val-phe-phe (GPGKLVFF). The KLVFF (Aβ [16][17][18][19][20] motif is widely used to bind to Aβ and inhibit its aggregation. [20] The GPG motif is added to prevent the aggregation of the conjugate itself because the KLVFF motif has the ability to aggregate. A derivative of p-amino-calix [4]arene has been used for coupling with the peptide at the lower rim. Mass spectrometry and sodium dodecyl sulfate (SDS)-PAGE results suggest that the conjugate could decrease the formation of Aβ oligomers.
Guo et al. synthesized an amphiphilic sulfonatocalixarene and evaluated its effect on the aggregation of insulin. [77] Four dodecyl groups have been added to SC4A to yield amphiphilic SC4CE, which contains hydrophilic sulfonate on the upper rim and hydrophobic alkyl chains at the lower rim. SC4CE has been reported to form micelles at concentrations higher than 0.02 mM. [78] The researchers found that the SC4CE micelle significantly inhibits the aggregation of insulin at a 1:1 ratio, whereas SC4A and sodium dodecyl benzenesulphonate (SDBS) exert weaker effects. The results from a competitive fluorescence titration assay indicate that the binding affinity between SC4CE and insulin is 2.0 × 10 7 M -1 , although direct binding sites have not been revealed.
Guo et al. recently developed a heteromultivalent peptide recognition method to inhibit amyloid fibrillation [79] and fur-ther applied this method to transgenic mice with Alzheimer's disease. [80] The authors constructed a coassembly of CA and CD amphiphiles that formed a hetero-assembly in solution. CA and CD were simultaneously distributed on the outside of the coassembly, exposing two types of recognition sites for different guest molecules. Owing to heteromultivalency and synergistic effects, peptides with multiple side chains that can bind with CA and CD separately are expected to bind selectively and strongly with the coassembly (Figure 7). Specifically, CD binds preferentially to Tyr, whereas CA binds more strongly to Lys. Therefore, the researchers first tested their hypothesis using some model peptides containing Lys and Tyr and proved the importance of heteromultivalency. Afterward, the authors determined that the binding affinity between the CD-CA coassembly and Aβ 1-42 is 7.9 × 10 7 M -1 , whereas no interactions could be observed if the Lys and Tyr residues of Aβ 1-42 were knocked out. The coassembly exerts better inhibitory effects than either CD or CA assembly alone or their mixture. Moreover, the coassembly decreases the cytotoxicity of Aβ and effectively disintegrates preformed Aβ fibrils at a ratio of 1:1. [79] The effectiveness and biocompatibility of the CD-CA coassembly indicates its potential as a therapy for Alzheimer's disease, as was confirmed recently. [80] Researchers have injected the coassembly into 5xFAD mice through intrahippocampal or intranasal administration and found that both the amyloid plaques and the level of Aβ monomers were significantly reduced. The results from multiple behavioral experiments suggest that the administration of the coassembly reduces cognitive decline and restores the memory of 5xFAD mice. Analysis of brain slices revealed that the coassembly reduces neuroinflammation and neuronal apoptosis. These results support the notion that heteromultivalent peptide recognition by the coassembly should be a F I G U R E 7 Illustration of cyclodextrin-calixarene (CD-CA) coassembly and heteromultivalent peptide recognition. Reproduced with permission: Copyright (2018) Springer Nature Limited. [79] good candidate for the treatment of multiple amyloid-related diseases.
The above-described studies show that CAs are significantly effective in inhibiting the aggregation of amyloid proteins. The disintegration of mature fibrils has been reported by different groups. Guo and coworkers have demonstrated that the coassembly of CD-CA amphiphiles is effective in AD model mice. Further in vivo studies of CAs are urgently needed for bench-to bedside translation (Table 2).

CDs
CDs, which were initially separated from the enzymatic reaction products of starch, are cyclic oligosaccharides linked by α-1,4-glucosidic bonds. [24,27] α-, β-, and γ-CDs contain 6, 7, and 8 units of D-glucose, respectively ( Figure 8). CDs are water soluble, and their multiple hydroxyl groups are easily to be modified. Different substituted CDs with different cavity sizes can then be applied to recognize and encapsulate a number of molecules, [27] particularly some therapeutic drugs. [83] β-CD has been approved by the FDA for use in humans as an oral excipient in some drugs. The effects of CDs on the aggregation of amyloid proteins have been extensively investigated, and comprehensive reviews were performed approximately 5 years ago. [84,85] The readers are advised to refer to these papers, and we summarize some recent progress here.
Zheng et al. reported that hydroxypropyl (HP)-β-CD effectively inhibits the aggregation of Aβ 1-42 at a ratio of 1:2 (Aβ:HP-β-CD), [86] which is lower than the concentrations used in previous studies involving β-CD. [84] Fewer fibrils and reduced β-sheet structures have been observed in the presence of HP-β-CD. Occupation of the cavity of HP-β-CD by ferulic acid results in disappearance of its inhibitory effects, which suggests the importance of its hydrophobic cavity. The results from molecular docking and MD simulations have revealed that HP-β-CD interacts with the Phe4, Lys16, Phe19, Ala21, Ile31, and Met35 residues in the Aβ pentamer.
F I G U R E 8 Molecular structures of unmodified cyclodextrins (CDs) and representative molecular structures of the CDs reported to modulate the aggregation of amyloid proteins polymers (pACD) has been constructed. Subsequently, the pACD has been modified with 5-(aminomethyl)-8hydroxyquinoline (AMHQ) to yield pCDHQ, which exerts stronger inhibitory effects than pACD and AMHQ (1:2, Aβ:CD derivatives). In addition, pCDHQ has the capability to chelate copper ions, which in turn inhibits copper-induced Aβ aggregation. [91] Mohanty et al. reported that SBE 7 β-CD, which possesses similar functional groups to SC[n]A, [72] inhibits the aggregation of insulin and lysozyme. [92] The effective ratio is 1:1 (protein:inhibitor). The researchers also proved that SBE 7 β-CD disintegrates insulin and lysozyme fibrils based on circular dichroism, AFM, and dynamic light scattering results. Incubation with SBE 7 β-CD decreases the cytotoxicity of insulin and lysozyme. Based on their studies with SC[n]A and SBE 7 β-CD, the researchers believed that both the sulfonate groups and the cavity of β-CD play important roles in inhibiting the aggregation of amyloid proteins.
Sun et al. synthesized a conjugate of β-CD with peptide Ac-LVFFARK-NH 2 (LK7). [93] The researchers previously found that LK7 partly inhibits the aggregation of Aβ but is itself toxic. [94] The conjugate LK7-β-CD exerts better inhibitory effects on Aβ 1-40 than LK7 alone or the simple mixture of LK7 and β-CD. LK7-β-CD is less toxic than LK7 and reduces the cytotoxicity of Aβ 1-40 at a ratio of 1:1. ITC results have confirmed that LK7-β-CD binds more strongly to Aβ 1-40 monomers than LK7 and β-CD.
In addition to conjugation with other molecules, β-CD has also been mixed with polyphenols and used to inhibit the aggregation of some amyloid proteins. Chowdhury et al. studied the effects of mixtures of β-CD with different polyphenols on α-synuclein aggregation. [95] Curcumin mixed with β-CD exhibits the highest efficiency in inhibiting fibrillation and disaggregating preformed fibrils. Dubey et al. demonstrated that a 1:2 mixture of curcumin and β-CD significantly inhibits the aggregation of silk fibroin. [96] Khan et al. reported that α-CD could reverse surfactantinduced amyloid fibrillation. [97,98] The researchers found that 0.2 mg/ml lysozyme forms aggregates in the presence of 0.3 mM SDS, whereas the addition of 1 mM α-CD completely eliminates the aggregates. Molecular docking results have suggested that α-CD has the capability to bind with both F I G U R E 9 Molecular structures of the crown ethers reported to modulate the aggregation of amyloid proteins lysozyme and SDS. [97] Similar results have been obtained for the SDBS-induced fibrillation of insulin. [98]

Crown ethers
Crown ethers are well known for their ability to bind with different cations, particularly metal ions. [99] Recent studies have suggested that some crown ethers could bind with positively charged Lys, His, and Arg residues, [100][101][102] which expands their applications to biomolecules. Tian et al. reported the effects of 12-crown-4 on the aggregation of Aβ 1-40 . [103] The researchers found that 12-crown-4 could change the surface charges of Aβ 1-40 and partly inhibit its aggregation (1:2, Aβ:12-crown-4). The authors then conjugated PiB, a positron emission tomography ligand for Aβ, to 12-crown-4 and obtained PiB-C (Figure 9), with the aim of achieving specificity for Aβ. The conjugate exerts better inhibitory effects on the aggregation and cytotoxicity of Aβ than PiB. The researchers have proved that the conjugate could effectively cross the BBB and stain amyloid plaques in AD model mice.
Yokoyama and coworkers investigated the effects of a variety of crown ethers and 18-crown-6 (18C6) derivatives on the aggregation of V30M-mutated transthyretin (V30M-TTR). [104] 18-Crown-6 may interact with the Lys residues on TTR. 4′-carboxybenzo-18C6 (Figure 9), which stabilizes the 2 μM TTR tetramers at a concentration of 4 mM, has been identified as the best inhibitor tested. The crystal structure of V30M-TTR complexed with 4′-carboxybenzo-18C6 suggests that the 18C6 derivative should be an allosteric inhibitor.
The easily available and biocompatible CDs are popular drug candidates. However, derivatizations of CDs are needed to achieve higher binding affinity and selectivity toward amyloid proteins. The combination of crown ethers and other macrocycles may yield better results on the aggregation of amyloid proteins (Table 3).

CONCLUSION AND OUTLOOK
In summary, all the above-described studies support the notion that macrocycles efficiently modulate the aggregation of different amyloid proteins. Considering the capabil-ity of macrocycles to stabilize protein drugs or decrease toxic aggregates, these compounds have the potential for further use in the treatment of multiple diseases. In addition to the abovementioned macrocycles, many other synthetic macrocycles, including heteracalixaromatics [70] , calixphyrins, [105] pillararenes, [106] and cyclophanes, [107] have been developed in recent years. Their distinctive structural properties make them potential candidates for modulating the aggregation of different amyloid proteins, although no related studies have been reported thus far.
The conjugation of macrocycles with other molecules increases their capability to inhibit the aggregation of amyloid proteins. [10,76,93] Heteromultivalent recognition methods with two types of macrocycles have also demonstrated great potential for inhibiting fibrillation. [79] Novel derivatives of the macrocycles are expected to be developed for different amyloid proteins. The combined administration of different host molecules is likely to achieve effectiveness and specificity at the same time.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest.