Specific aromatic foldamers potently inhibit spontaneous and seeded Aβ42 and Aβ43 fibril assembly

Amyloid fibrils are self-propagating entities that spread pathology in several devastating disorders including Alzheimer's disease (AD). In AD, amyloid-β (Aβ) peptides form extracellular plaques that contribute to cognitive decline. One potential therapeutic strategy is to develop inhibitors that prevent Aβ misfolding into proteotoxic conformers. Here, we design specific aromatic foldamers, synthetic polymers with an aromatic salicylamide (Sal) or 3-amino benzoic acid (Benz) backbone, short length (four repetitive units), basic arginine (Arg), lysine (Lys) or citrulline (Cit) side chains, and various N- and C-terminal groups that prevent spontaneous and seeded Aβ fibrillization. Ac-Sal-(Lys-Sal)3-CONH2 and Sal-(Lys-Sal)3-CONH2 selectively inhibited Aβ42 fibrillization, but were ineffective against Aβ43, an overlooked species that is highly neurotoxic and frequently deposited in AD brains. By contrast, (Arg-Benz)4-CONH2 and (Arg-Sal)3-(Cit-Sal)-CONH2 prevented spontaneous and seeded Aβ42 and Aβ43 fibrillization. Importantly, (Arg-Sal)3-(Cit-Sal)-CONH2 inhibited formation of toxic Aβ42 and Aβ43 oligomers and proteotoxicity. None of these foldamers inhibited Sup35 prionogenesis, but Sal-(Lys-Sal)3-CONH2 delayed aggregation of fused in sarcoma (FUS), an RNA-binding protein with a prion-like domain connected with amyotrophic lateral sclerosis and frontotemporal dementia. We establish that inhibitors of Aβ42 fibrillization do not necessarily inhibit Aβ43 fibrillization. Moreover, (Arg-Sal)3-(Cit-Sal)-CONH2 inhibits formation of toxic Aβ conformers and seeding activity, properties that could have therapeutic utility.


INTRODUCTION
Protein misfolding can be fatal [1,2]. Proteins misfold from soluble species into highly stable, cross-β amyloid fibrils in Alzheimer's disease (AD) and several other neurodegenerative diseases [1,2]. One strategy to combat these disorders is to develop small molecules that inhibit amyloidogenesis and prevent toxic protein misfolding [3][4][5][6]. Although daunting challenges face potential small molecule inhibitors of amyloidogenesis [7], they are beginning to reach the clinic. Indeed, tafamidis, a small molecule inhibitor of transthyretin amyloidogenesis treats familial amyloid polyneuropathy, a rare but deadly disease [8,9].
Although Aβ peptides longer than Aβ42 are found in AD, they are not a major species and their pathogenic role has been ignored. Recently, this view has changed. Aβ43 is a potent contributor to neurotoxicity in AD [13][14][15]. Aβ43 contains an additional threonine residue at the C-terminal end and fibrillizes more rapidly than Aβ42 [13]. Aβ43 is more abundant in insoluble fractions than Aβ40 in AD and its presence in senile plaques is directly correlated with cognitive decline [13][14][15][16]. Specific inhibitors of Aβ43 misfolding have not been identified and it is unclear whether inhibitors of Aβ42 misfolding will also inhibit Aβ43 misfolding.
Aβ monomers form amyloid via nucleated conformational conversion [22]. First, a subpopulation of Aβ monomers forms molten oligomers, which gradually rearrange into amyloidogenic oligomers that nucleate cross-β fibrils [22,23]. Rearrangement is rate limiting and causes the lag phase of spontaneous fibrillization [22]. During lag phase, Aβ forms diverse oligomeric species, which can be highly toxic [21,[24][25][26][27]. Upon nucleation, fibrils The core foldamer structure is shown in the dashed box, which can be decorated with different moieties at X-, R-, Y-and Z-positions indicated on the periphery. Foldamers possess an aromatic Sal or Benz backbone (Y = OMe or H), Arg, Lys or Cit side chains (R = Arg, Lys or Cit), short length (two to four repetitive units) and various N-(X = NH 2 or Ac) and C-(Z = NH 2 , OH, OMe or β-Ala) terminal groups.
A key challenge is to manipulate Aβ assembly in a manner that depopulates toxic conformers [7]. Agents that inhibit seeded assembly hold promise for preventing the spread of Aβ pathology in AD.

Generation of soluble and fibrillar Aβ42 and Aβ43
To produce monomeric Aβ, synthetic lyophilized Aβ42 or Aβ43 (W.M. Keck Facility, Yale University) was dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, Sigma) at 2 mg/ml. HFIP was removed by drying in a speed vacuum for 30 min. The resulting peptide film was dissolved in DMSO to 1 mM. Aβ42 or Aβ43 fibrils for seeding experiments were prepared by diluting monomerized Aβ42 or Aβ43 in KHMD (150 mM KCl, 40 mM Hepes-KOH pH 7.4, 20 mM MgCl 2 and 1 mM DTT) to 10 μM. This solution was incubated at 37 • C for 3-5 days with agitation (700 r.p.m.) in an Eppendorf Thermomixer. For seeding experiments, preformed fibrils were briefly sonicated or vortexmixed prior to use. We also prepared Aβ42 or Aβ43 using a protocol that avoids DMSO. Thus, Aβ42 or Aβ43 was dissolved in HFIP followed by evaporation of the solvent to dryness [56]. Dry peptide films were dissolved in a minimal volume of 60 mM NaOH followed by dilution with deionized water and sonication for 1 min using a bath sonicator. Peptides were diluted to 0.2 mM by adding an equal volume of 20 mM sodium phosphate buffer (PB, Sigma), pH 8 plus 0.2 mM EDTA (PBE). Samples were centrifuged at 16 000 g for 3 min and subjected to Superdex 75 gel filtration in PBE to remove residual solvent.  Three-letter amino acid nomenclature is used to indicate the side chain (Lys, Arg or Cit) and the Sal or Benz backbone is indicated. N-(Ac) and C-(NH 2 , OH, OMe or β-Ala) terminal groups are also indicated. Foldamers that inhibit spontaneous Aβ42 and Aβ43 fibrillization, (Arg-Benz) 4 -CONH 2 and (Arg-Sal) 3 -(Cit-Sal)-CONH 2 , are boxed in black. Foldamers that inhibit spontaneous Aβ42 fibrillization but not spontaneous Aβ43 fibrillization, Sal-(Lys-Sal) 3 -CONH 2 and Ac-Sal-(Lys-Sal) 3 -CONH 2 , are boxed in grey. concentrated stocks. Subsequent dilutions were made from these stocks to appropriate concentrations in KHMD or PBE.
The first residue was coupled to the resin using 3 equiv. of amino acid, 2.8 equiv. of 2-(6-chloro-1H-benzotriazole-1-yl)-1,1,3,3-tetramethylaminium hexafluorophosphate (HCTU, GL Biosciences) activator and 7.5 equiv. of di-isopropylethylamine (DIEA, CHEM-IMPEX International), shaking for 1 h at room temperature. The resin was washed three times each with DMF, dichloromethane (DCM, Fisher Scientific) and DMF. This step was followed by deprotection (as above). Coupling and deprotection steps were cycled for the remaining residues in each respective peptide sequence. After deprotection of the  final residue the product was rinsed [three times with DMF, three times with DCM, three times with DMF and three times with methanol (MeOH)] and dried with MeOH. This product was split in half. The first half was re-swelled in DMF and acetylated by incubating the resin in 5 % acetic anhydride in 2.5 % DIEA and 92.5 % DMF for 10 min. This acetylated portion was rinsed and dried (as above). Next, both halves (one with a Nterminal acetyl and a second with a N-terminal free amide) were cleaved from the resin using a cocktail of 2:2:2:94 H 2 O/TIS (tri-isopropyl silane)/anisole/TFA (trifluoroacetic acid; Sigma-Aldrich) for 2 h at room temperature. The peptide solution was filtered from the resin and precipitated using 1:1 cold ethyl ether:hexane. The precipitate was dried by lyophilization. The mass and purity of each product was verified by MALDI-TOF MS (Brucker microflex LRF) and analytical HPLC (C18 column). Dried crude foldamer was purified by preparative reverse-phase HPLC, dried by lyophilization and mass and purity was verified as above. All samples were prepared by directly dissolving lyophilized foldamer into TBS buffer to 2 mM.

FUS aggregation
GST-TEV-FUS was purified as described [58]. Aggregation was initiated by addition of tobbaco etch virus (TEV) protease to GST-TEV-FUS (5 μM) plus or minus foldamer (20 μM) in assembly buffer (50 mM Tris/HCl pH 8, 0.2 M trehalose and 20 mM glutathione). Aggregation was for 0-90 min at 25 • C without agitation in a 96-well plate and was assessed by turbidity (absorbance at 395 nm) using a Tecan Infinite M1000 plate reader [58]. No aggregation occurred unless TEV protease was added to separate GST from FUS [58]. SDS/PAGE and Coomassie staining revealed that foldamers did not inhibit cleavage of GST-TEV-FUS by TEV.

Electron microscopy
Reactions were adhered on to 300-mesh-formvar carbon-coated EM grids overnight before being negatively stained with 2 % uranyl acetate for 2 min and rinsed with milli-Q distilled water. Micrographs were acquired using a JEOL 1010 TEM (Jeol USA).

Tracking A11-reactive Aβ42 or Aβ43 conformers
The oligomer-specific A11 antibody was used to detect toxic Aβ42 or Aβ43 oligomers by ELISA as described [21]. Foldamers did not cross-react with A11.

(Arg-Sal) 3 -(Cit-Sal)-CONH 2 antagonizes formation of A11-reactive Aβ42 and Aβ43 oligomers
Could foldamers inhibit the formation of toxic Aβ42 and Aβ43 oligomers? To assess toxic Aβ42 and Aβ43 oligomer formation, we employed the conformation-specific A11 antibody, which specifically recognizes preamyloid oligomers formed by multiple proteins, including Aβ42, but not monomers or fibrils [21]. We assessed formation of A11-reactive species at the start of spontaneous assembly (0 h), at the end of lag phase (0.5 h), and at the endpoint of fibrillization (4 h). In the absence of Aβ42 and Aβ43, no A11 immunoreactivity was observed (results not shown). For Aβ42 and Aβ43, A11-reactive conformers were scarce at the start of the reaction ( Figure 10A, buffer controls, black bars), abundant at end of lag phase ( Figure 10A, buffer controls, grey bars), and declined once fibrillization was complete ( Figure 10A, buffer controls, white bars). Aβ43 exhibited greater A11-immunoreactivity than Aβ42 and appears more prone to accessing this toxic conformation ( Figure 10A).

(Arg-Sal) 3 -(Cit-Sal)-CONH 2 inhibits formation of toxic Aβ42 and Aβ43 conformers
Next, we evaluated the relative toxicity of Aβ42 and Aβ43 conformers formed in the absence or presence of foldamers. We applied Aβ42 and Aβ43 conformers to SH-SY5Y neuroblastoma cells and assessed cell viability using MTT reduction and LDH release. Foldamers and buffer display little toxicity in the absence  3). A one-way ANOVA with the post-hoc Dunnett's multiple comparisons test was used to compare Aβ42 plus buffer to each Aβ42 plus foldamer condition (* denotes P < 0.05). Likewise, a one-way ANOVA with the post-hoc Dunnett's multiple comparisons test was used to compare Aβ43 plus buffer to each Aβ43 plus foldamer condition (* denotes P < 0.05).
Aromatic foldamers could be useful amyloidogenesis inhibitors for various disease-associated proteins. Indeed, another class of aromatic foldamer inhibits amylin fibrillization, which is connected to Type 2 diabetes [66]. Thus, foldamers await further development to antagonize protein misfolding in several settings.