Designed peptides as nanomolar cross-amyloid inhibitors acting via supramolecular nanofiber co-assembly

Abstract

Amyloid self-assembly is linked to numerous devastating cell-degenerative diseases. However, designing inhibitors of this pathogenic process remains a major challenge. Cross-interactions between amyloid-β peptide (Aβ) and islet amyloid polypeptide (IAPP), key polypeptides of Alzheimer’s disease (AD) and type 2 diabetes (T2D), have been suggested to link AD with T2D pathogenesis. Here, we show that constrained peptides designed to mimic the Aβ amyloid core (ACMs) are nanomolar cross-amyloid inhibitors of both IAPP and Aβ42 and effectively suppress reciprocal cross-seeding. Remarkably, ACMs act by co-assembling with IAPP or Aβ42 into amyloid fibril-resembling but non-toxic nanofibers and their highly ordered superstructures. Co-assembled nanofibers exhibit various potentially beneficial features including thermolability, proteolytic degradability, and effective cellular clearance which are reminiscent of labile/reversible functional amyloids. ACMs are thus promising leads for potent anti-amyloid drugs in both T2D and AD while the supramolecular nanofiber co-assemblies should inform the design of novel functional (hetero-)amyloid-based nanomaterials for biomedical/biotechnological applications.

Introduction

Amyloid self-assembly is linked to numerous devastating cell-degenerative diseases, with AD and T2D being two of the most prominent ones^1,2. The main component of amyloid plaques in AD brains is the 40(42)-residue peptide Aβ40(42), while pancreatic amyloid of T2D patients consists of fibrillar assemblies of the 37-residue IAPP^2,3 (Fig. 1a). IAPP is secreted from pancreatic β-cells and functions as a neuroendocrine regulator of glucose homeostasis³. However, the formation of cytotoxic IAPP assemblies and amyloid fibrils mediates pancreatic β-cell degeneration in T2D³.

Epidemiological studies suggest that T2D patients have an increased risk of AD and vice versa^4,5,6,7. In addition, increasing evidence suggests molecular and pathophysiological links between both diseases^7,8,9,10. Cross-interactions between Aβ and IAPP could be such molecular links^{7,8,9,10,11,12,13,14,15,16}. In fact, polymorphic Aβ/IAPP interactions are able to cross-seed or cross-suppress amyloid self-assembly depending on structures and self-assembly states of the interacting polypeptides^{7,8,9,11,12,13,16}. To this end, IAPP and Aβ fibrils act as reciprocal cross-seeds of amyloid self-assembly, as shown by both in vitro and experimental in vivo studies^8,9,11. On the other hand, nanomolar affinity interactions between early prefibrillar and non-toxic IAPP and Aβ species redirect both polypeptides into initially non-fibrillar and non-toxic co-assemblies, thus delaying amyloid self-assembly^12,13. Importantly, Aβ and IAPP were found to colocalize in AD- and T2D-related amyloid deposits both in humans and in mouse models^8,9,10,14,15. Aβ/IAPP cross-interactions and putative “hetero-amyloids” could thus be highly relevant to the pathogenesis of both diseases^{7,8,9,10,11,12,15,17,18}.

Based on the above, molecules targeting amyloid self-assembly and reciprocal cross-seeding effects of IAPP and Aβ could be promising leads for anti-amyloid treatments in both AD and T2D^7,19. However, so far, only a few inhibitors of amyloid self-assembly of both polypeptides (termed “cross-amyloid” inhibitors) have been reported and none of them suppressed reciprocal Aβ/IAPP cross-seeding^{12,19,20,21,22,23,24}. Moreover, except for a recently approved and controversially discussed anti-Aβ amyloid antibody, no anti-amyloid treatments for AD or T2D have yet reached the clinic.

One reason for the high-affinity IAPP/Aβ40(42) cross-interactions could be the sequence similarity (50%) and identity (~25%) between both polypeptides (Fig. 1a)^11,25. Notably, highest degrees of sequence identity/similarity are observed between their amyloid core segments IAPP(8–28) and Aβ(15–40(42)). In addition, the same IAPP- or Aβ40(42)-“hot segments” within their amyloid core segments were found to mediate both self- and cross-interactions (Fig. 1a)^11,25,26. Strong similarities exist also between their fibril folds and potential cross-seeding interfaces within putative hetero-amyloids were proposed^{13,24,25,27,28,29,30,31}.

Capitalizing on IAPP/Aβ cross-interactions, we have previously designed peptides derived from the IAPP amyloid core IAPP(8–28) as IAPP “interaction surface mimics” (ISMs)²⁰. ISMs effectively suppressed amyloid self-assembly of Aβ40(42) and/or IAPP by sequestering them into amorphous, non-toxic aggregates²⁰.

Here, we explored the idea of designing peptides derived from the Aβ40 amyloid core Aβ(15–40) as Aβ “amyloid core mimics” (ACMs) and inhibitors of amyloid self-assembly and cross-seeding interactions of IAPP and Aβ42. Our inhibitor design concept aimed at distorting the pathogenic fibril fold of Aβ(15–40) and stabilize alternative, amyloid-like but non-amyloidogenic folds¹⁹. These should yield alternative interaction surfaces with IAPP or Aβ42 and redirect them into non-fibrillar and non-toxic aggregates^{12,19,20,32,33}. A series of conformationally constrained peptides was synthesized and studied. In fact, ACMs were non-amyloidogenic and non-cytotoxic, bound IAPP and Aβ42 with nanomolar affinity, and fully blocked their cytotoxic amyloid self-assembly. Furthermore, ACMs effectively suppressed reciprocal cross-seeding effects. Surprisingly, ACMs exerted their inhibitory function by co-assembling with IAPP or Aβ42 into amyloid fibril-resembling nanofibers and their diverse, highly ordered superstructures. For their characterization, a spectrum of biophysical, biochemical, and advanced microscopy methods, including confocal laser-scanning microscopy (CLSM), stimulated emission depletion (STED) imaging, two-photon microscopy (2PM), and fluorescence lifetime imaging microscopy (FLIM)-based Förster resonance energy transfer (FRET) (FLIM-FRET) was applied. In addition, in vitro and ex vivo cell-based assays were used. In strong contrast to IAPP or Aβ42 fibrils (fIAPP or fAβ42), co-assembled nanofibers were “ThT-invisible”, non-cytotoxic, and seeding-incompetent. Moreover, they were thermolabile, easily degradable by proteinase K (PK), and became efficiently phagocytosed in vitro by primary macrophages and cultured microglial cells.

Results

Inhibitor design and concept evaluation

For inhibitor design, Aβ(15–40) was used as a template in the context of the fAβ40 fold suggested by Petkova et al.^31,34, which features a β-strand-loop-β-strand motif with Aβ(12–22) and Aβ(30–40) forming the β-strands and Aβ(23–29) the loop (Fig. 1a, b). Of note, this U-shaped fold has often been applied to model Aβ-IAPP hetero-amyloids^35,36. A minimum number of chemical modifications was made aiming at (a) distorting the loop, (b) stabilizing β-sheet structure, and (c) suppressing intrinsic amyloidogenicity of Aβ(15–40) while maintaining its pronounced self-/cross-assembly propensity in analogy to the ISM concept (Fig. 1b)^12,20,25,32. The modifications were: (a) substitution of loop tripeptide Aβ(24–26) (Val-Gly-Ser) by β-sheet-propagating tripeptides consisting of identical large hydrophobic residues, which were expected to strengthen β-sheet interaction surfaces while being incompatible with localization in turns/β-arcs^37,38,39 and (b) selective amide bond N-methylation of two alternate residues within one of the two Aβ β-strand segments, which should suppress intrinsic amyloidogenicity of ACMs and their co-assemblies (Fig. 1b)^32,40,41. Positions of N-methylations were based on fAβ40 models and previous SAR studies^{31,34,40,41,42}. Finally, Met35 was replaced by Nle to avoid Met(O)-related side effects….