Excitation Energy Migration Unveils Fuzzy Interfaces within the Amyloid Architecture

Abstract

Amyloid fibrils are highly ordered nanoscopic protein aggregates comprising a cross-β amyloid core and are associated with deadly human diseases. Structural studies have revealed the supramolecular architecture of a variety of disease-associated amyloids. However, the critical role of transient intermolecular interactions between the disordered polypeptide segments of protofilaments in directing the supramolecular structure and nanoscale morphology remains elusive. Here, we present a unique case to demonstrate that interchain excitation energy migration via intermolecular homo-Förster resonance energy transfer can decipher the architecture of amyloid fibrils of human α-synuclein. Site-specific homo-Förster resonance energy transfer efficiencies measured by fluorescence depolarization allowed us to construct a two-dimensional proximity correlation map that defines the supramolecular packing of α-synuclein within the fibrils. These studies captured unique heteroterminal cross talks between the fuzzy interprotofilament interfaces of the parallel-in-register amyloid spines. Our results will find applications in discerning the broader role of protein disorder and fuzziness in steering the distinct polymorphic amyloids that exhibit strain-specific disease phenotypes.

Significance

Amyloids are misfolded protein aggregates that are known to ravage the brain in devastating neurodegenerative diseases such as Alzheimer’s, Parkinson’s, and prion diseases. The disease phenotype is critically dependent on the precise supramolecular packing of these altered, misfolded protein molecules, giving rise to amyloid polymorphism. The molecular details of the higher-order organization of amyloids via disordered, dynamic, mesh-like networks remain elusive. Here, we show that unique intermolecular cross talks can be readily captured by monitoring the regiospecific nanoscopic proximity via intermolecular energy migration between site-specific fluorescently-labeled protein molecules within the amyloid architecture. Our sensitive methodology of excitation energy migration can serve as a powerful tool to study the amyloids associated with debilitating human disorders.

Main Text

Amyloids are ordered protein aggregates comprising the cross-β-structural motif and are associated with a range of debilitating human diseases (1, 2, 3, 4, 5, 6), as well as important functions in diverse organisms (7, 8, 9, 10). The intricate molecular principles that relate the protein conformation, supramolecular packing, nanoscale organization, and polymorphism are of immense interest in the understanding of both amyloid pathogenesis and function. For instance, pathological inclusions of α-synuclein (α-syn), which is linked to Parkinson’s disease (11, 12, 13), are characterized by the accumulation of conformationally and biochemically distinct polymorphic α-syn fibrils, giving rise to disease-specific strains (14, 15, 16, 17, 18, 19, 20, 21, 22). High-resolution structural studies have provided an incredible wealth of information on the diverse supramolecular packing arrangements of the amyloid core (17,23, 24, 25); however, the nature of critical intermolecular contacts between the individual protofilaments mediated by the flexible disordered regions remains poorly understood. In this work, we captured the intermolecular cross talks within α-syn amyloid fibrils by mapping the intermolecular energy migration efficiency via site-specific homo-Förster resonance energy transfer (homo-FRET). Using this unique and sensitive intermolecular ruler, we constructed a two-dimensional (2D) nanometric proximity correlation map that allowed us to define the supramolecular architecture of α-syn amyloids.

Excitation energy migration via homo-FRET is a special case of FRET that involves a nonradiative energy transfer between the dipoles of two or more chemically identical and proximal fluorophores possessing a small Stokes’ shift, allowing a significant overlap of its excitation and emission spectra (26, 27, 28, 29). Such energy migration events between the different randomly oriented identical fluorophores positioned within the Förster distance depolarize the fluorescence anisotropy because of the apparent orientational randomization independent of rotational dynamics. Such a depolarization via homo-FRET has been previously observed in α-syn amyloid fibrils (30, 31, 32). The energy migration efficiency measured by the observed loss in the fluorescence anisotropy at a given local density of fluorophore is related to the interdye distance. Therefore, we envisioned that site-directed homo-FRET measurements can allow us to characterize the proximal and distal intermolecular locations in complex supramolecular organizations such as amyloid fibrils. Using this conceptual framework, we set out to map the site-specific intermolecular proximities between the different residue positions of α-syn within the amyloid architecture. At a low labeling fraction, the anisotropy represents the characteristic rotational dynamics, whereas at a higher labeling fraction, the excited state dipoles of fluorophores can depolarize via homo-FRET-mediated energy migration to differently oriented adjacent labels within amyloids (Fig. 1). Therefore, the homo-FRET efficiency at a fixed label density at various sites can act as a unique site-specific proximity readout.

Figure 1.

The conceptual framework of homo-FRET-mediated fluorescence depolarization in amyloid fibrils. In the presence of a higher labeling density, a homotransfer causes an extensive depolarization, resulting in the anisotropy loss. The images of α-syn monomer (PeDB 9AAC; structure #150) and amyloid fibrils (Protein Data Bank, PDB: 6CU7) are created using PyMol (Schrödinger, New York). To see this figure in color, go online.

As a prelude, we first characterized the structural conversion of monomeric α-syn into amyloid fibrils (Figs. 2, S1, and S2). Next, to probe the site-specific interactions within the fibrils, we used seven single-Cys constructs of α-syn spanning the entire length (Fig. 2 a). Each Cys variant was then covalently labeled with a prominent, thiol-active homo-FRET dye, namely fluorescein-5-maleimide (F5M), having a Förster distance of ∼40 Å (29). These F5M-labeled single-Cys variants of α-syn were mixed with (unlabeled) wild-type α-syn at different molar ratios, and aggregation reactions were performed to create different local fluorophore density distributions within the fibrils. These mixtures aggregated to form amyloid fibrils similar to the wild-type α-syn (Figs. S1 and S2). The steady-state fluorescence anisotropy that is related to the rotational flexibility of the fluorophore is expected to rise upon the aggregation of sparsely labeled α-syn mixtures because of the formation of amyloid fibrils that exhibit slow tumbling compared to monomers. Therefore, low fractions of labeled protein within the fibrils should represent the inherent dynamic signatures of the fluorophore attached to the fibril. We hypothesized that at higher fractions of labeled protein, homo-FRET should yield a loss in the anisotropy in amyloids because of the increased energy migration between the proximal fluorophore labels despite the slow rotational tumbling in the aggregated state (Fig. 1).

Figure 2.

(a) Amino acid sequence of human α-syn: -N-terminal (1–60), blue; NAC (non-Aβ-component), olive (61–95); C-terminal, red (95–140). The seven residue positions chosen for the single-Cys variants for creating the labeling sites are underscored as shown. (b) Thioflavin-T (ThT) kinetics for aggregation of 200 μM α-syn are shown. (c) The CD spectra of a monomer and amyloid are given. (d) Shown is the vibrational Raman spectrum of the α-syn fibrils showing the characteristic amide I band at 1672 cm⁻¹, which is a hallmark of the cross-β structure. (e) An atomic force microscopy image of a single amyloid fibril is given. See Supporting Materials and Methods for experimental details. To see this figure in color, go online.

To validate our hypothesis, we chose an NAC (non-Aβ-component) location at residue position 90 that is known to be a part of the amyloid core of α-syn fibrils. Upon aggregation of α-syn with low labeling fractions (0.5–1.0%), we observed a sharp increase in the steady-state anisotropy (r) due to the highly dampened rotational mobility in the amyloid state (Fig. 3 a). On the contrary, at higher fractions of labeled protein (5–30%), the amyloid fibrils exhibited a fluorophore-concentration-dependent depolarization in the anisotropy (Figs. 3 a and S3). This loss in the anisotropy indicated the increased homo-FRET efficiencies as a function of fluorophore density on the amyloid fibrils. To confirm that the drop in the anisotropy is indeed due to homo-FRET, we next performed picosecond time-resolved fluorescence anisotropy to directly follow the fluorescence depolarization kinetics.

Figure 3.

(a) Steady-state fluorescence anisotropy and (b) the corresponding picosecond time-resolved fluorescence anisotropy decays of monomeric F5M- Cys90-α-syn and amyloid fibrils formed from increasing ratios of labeled/unlabeled (see Supporting Materials and Methods for details). (c) A site-specific anisotropy map is given of the α-syn fibrils in the absence and presence of homo-FRET. Data are shown as mean ± SEM for three independent experiments. (d) The site-specific anisotropy loss (Δr) between the like-residues is shown. To see this figure in color, go online.

In time-resolved anisotropy measurements, the fundamental (time-zero) anisotropy (r₀) decays biexponentially with a characteristic (local) subnanosecond rotational correlation time and a (global or segmental) nanosecond rotational correlation time. In the monomeric form, F5M-labeled α-syn exhibits local probe dynamics and segmental dihedral mobility of disordered peptide backbones (33,34). Upon amyloid formation using a low labeling fraction, the fluorescence anisotropy decay represents a small amplitude of local motion and a slow rotational component that does not depolarize completely on the nanosecond timescale because of the formation of large, slowly diffusing aggregates (Fig. 3 b). Therefore, at a low labeling fraction, the anisotropy decay captures the inherent dynamic signatures of amyloids. When we increased the ratio of labeled protein from 5 to 30%, we observed increasingly faster anisotropy decay profiles that could only be described by triexponential decay kinetics (Fig. 3 b). These analyses revealed an additional (intermediate) time constant between the local and global correlation times of fibrils. This intermediate nanosecond time constant (∼4 ns) represents the rate of energy migration ∼2.5 × 10⁸ s⁻¹ along with an additional (faster) subnanosecond component (Table S1). The contribution (fractional amplitude) of the energy migration grew as the label density on the fibrils increased (Fig. S4). Taken together, our steady-state and time-resolved anisotropy measurements on α-syn amyloid fibrils comprising a varying extent of labeled fractions established that the loss in the steady-state fluorescence anisotropy (Δr) in fibrils with a higher label density is indeed due to the energy migration via homo-FRET. These results led us to postulate that Δr, if measured in a residue-specific manner, could map the intermolecular distances within the amyloid assembly.

Next, we set out to determine the residue-specific Δr. For these studies, we used all seven single-Cys variants and labeled them with F5M to measure 49 intermolecular proximities involving the like-residues and the cross-residues spanning different segments of α-syn. We carried our anisotropy measurements on amyloid fibrils by varying the ratio of labeled protein/unlabeled protein from 1 to 30%. As described above, 1% of labeled protein did not exhibit a significant homo-FRET, which was used as the control for recording the homo-FRET-free anisotropy. In contrast, higher labeling ratios (>10%) exhibited an extensive homo-FRET due to a higher density of fluorophore within the Förster distance and therefore lacked the location-specific information (Fig. S3). Interestingly, an intermediate labeling ratio (5% F5M-labeled α-syn + 95% unlabeled wild-type α-syn) exhibited location-specific homo-FRET, as indicated by the different values of Δr. Therefore, we chose the 5:95 labeling ratio as an optimal labeling density to map the location-specific energy migration. The site-specific fluorescence anisotropy and the loss in the anisotropy due to homo-FRET are shown in Fig. 3, c and d. The anisotropy without homo-FRET yielded high values for the residues in the pre-NAC and NAC domain at locations 56, 78, and 90, followed closely by the C-terminal residues (residues 124 and 140) and then the N-terminal residues (residues 9 and 18). Under the homo-FRET condition, we observed different extents of the drop in the anisotropy because of the energy migration as a function of residue position (Fig. 3 c). The plot of Δr versus the residue position shows that the pre-NAC and NAC residues are in close proximity compared with the N- and C-terminal residues (Fig. 3 d). Next, to record all possible cross-interactions between the residues within the same and different domains, we performed cross-homo-FRET measurements. In these experiments, we mixed a particular pair of F5M-labeled α-syn variants at an equimolar ratio while maintaining the total labeled protein concentration at 1% (no homo-FRET) and 5% (homo-FRET) (Fig. S5). In this set of cross-homo-FRET experiments, Δr indicated the cross-residue proximity within the fibrils. We then estimated the apparent homo-FRET efficiencies for all the like and cross residues, allowing us to construct a square matrix with the 2D proximity correlation map (Fig. 4).

Figure 4.

2D proximity correlation map of α-syn amyloid fibrils; the mean apparent homo-FRET efficiencies of 49 intermolecular distances are presented as a heat map. See Supporting Materials and Methods for the estimation of the apparent homo-FRET efficiencies and the construction of the heat map. To see this figure in color, go online.

The 2D correlation map depicts comprehensive multiresidue cross talks and elucidates several unique structural features of α-syn amyloid fibrils (Fig. 4). The diagonal entries characterize the interactions between the like residues, whereas the off-diagonal entries represent the cross-residue interactions. A characteristic clustering of diagonal hotspots in the central NAC domain clearly indicates the close proximity between the like residues and therefore is in line with the proposed parallel in-register packing. The correlation map shows that the N-terminal residues are noninteracting and are not recruited into the amyloid spine. These observations are consistent with parallel in-register packing of the central amyloid core primarily comprising pre-NAC and NAC residues from residue 38 to 97. Of note, an antiparallel out-of-register packing would have yielded a very different cluster-free correlation pattern involving low proximities between the like-core-residues. Interestingly, we detected some previously unobserved weak cross talks between the C-terminal residues that could potentially indicate long-range transient contacts between highly disordered C-terminal ends in the higher-order protofilament-protofilament association. Additionally, the off-diagonal entries captured some unique interdomain cross talks within the fibrils. The central NAC domain that constitutes the amyloid core is highly sequestered and therefore is largely isolated from interdomain interactions with a few exceptions. On the contrary, flexible N- and C-termini (residues 9–140, 18–140, and 9–124) that are not recruited into the amyloid core exhibit somewhat significant interdomain interactions (Fig. 4). Interestingly, the cross talk between C- and N-terminal end residues (9–140) is more prominent than the like-terminal residues (residues 9–9, 140–140) and the internal cross-terminal residues, (residues 9–124 and 18–140) clearly demonstrating the interactions between the ends of C- and N-termini. This striking observation might indicate the presence of transient electrostatic contacts between the oppositely charged N- and C-termini, forming a dense, dynamic, fuzzy, and mesh-like network around the fibril core (35). These long-range heteroterminal interactions can potentially act as constraints within the amyloid fold and promote protofilament-protofilament interactions and therefore can critically govern the growth and nanoscale morphology of amyloid assemblies.

In summary, our intermolecular energy migration studies reveal, to our knowledge, novel supramolecular structural features of α-syn amyloid fibrils. In addition to an extensively hydrogen-bonded β-rich amyloid core comprising parallel in-register packing, the flexible disordered segments in protofilament interfaces participate in unique dynamic interactions that assist the interprotofilament assembly into matured amyloid fibrils. These dynamic interactions underscore the importance of a protein disorder and fuzziness in biological assemblies (36,37). Additionally, these interactions can modulate the higher-order packing and nanoscale morphology that govern amyloid polymorphism responsible for strain-specific phenotypes in a wide range of pathological and functional amyloids. This unique and sensitive methodology can serve as a potent tool to study amyloids and a wide range of other biological assemblies.

Author Contributions

S.M. designed the study. A.M., D.D., P.M., and A.A. performed the experiments. A.M. and D.D. analyzed the data. A.M., D.D., and S.M. wrote the article. All authors discussed the results and commented on the manuscript.

Editor: Monika Fuxreiter.

Supporting Citations

References (38, 39, 40) appear in the Supporting Material.