Resolving the Nanoscale Structure of β-Sheet Peptide Self-Assemblies Using Single-Molecule Orientation-Localization Microscopy

Abstract

Synthetic peptides that self-assemble into cross-β fibrils are versatile building blocks for engineered biomaterials due to their modularity and biocompatibility, but their structural and morphological similarities to amyloid species have been a long-standing concern for their translation. Further, their polymorphs are difficult to characterize by using spectroscopic and imaging techniques that rely on ensemble averaging to achieve high resolution. Here, we utilize Nile red (NR), an amyloidophilic fluorogenic probe, and single-molecule orientation-localization microscopy (SMOLM) to characterize fibrils formed by the designed amphipathic enantiomers KFE8^L and KFE8^D and the pathological amyloid-beta peptide Aβ42. Importantly, NR SMOLM reveals the helical (bilayer) ribbon structure of both KFE8 and Aβ42 and quantifies the precise tilt of the fibrils' inner and outer backbones in relevant buffer conditions without the need for covalent labeling or sequence mutations. SMOLM also distinguishes polymorphic branched and curved morphologies of KFE8, whose backbones exhibit much more heterogeneity than those of typical straight fibrils. Thus, SMOLM is a powerful tool to interrogate the structural differences and polymorphism between engineered and pathological cross-β-rich fibrils.

Keywords: fluorogenic probes; polymorphism; self-assembly; super-resolution microscopy; supramolecular helix.

Figure 1.

Single-molecule orientation-localization microscopy (SMOLM) of Nile red (NR) reveals the nanoscale structure of KFE8 and Aβ42 fibrils. a, (i) Transient binding of fluorogenic probes (red double-headed arrow) to a left-handed (LH) helical bilayer. Upon binding, the transition dipole moment of NR is parallel to the fibrillar backbone, which exhibits a periodic helical twist along its long axis. NR is nonfluorescent in solution (dark arrows). ($x,y,z$), spatial coordinates defined by the microscope. $\tau$, helix phase, which for a helical structure, also corresponds to the height of the fibril above the coverglass. Inset, chemical structure of NR. (ii) A model orientation distribution of NR accumulated across all possible binding sites along the helical bilayer. $(u_x,u_y,u_z)$, orientation coordinates defined by the long axis $u_x$ of the fibril. $\alpha$, wobble cone half angle of NR, quantifying its orientation freedom during a camera frame. ${\delta }_{\mathrm{i}\mathrm{n}\mathrm{n}\mathrm{e}\mathrm{r}}$ and ${\delta }_{\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{r}}$, inner and outer backbone tilt angles of the helical bilayer, respectively. $\tau =0$ corresponds to a minimum value of $u_z$ and increases in the direction of the gray arrow. (iii and iv) Right-handed (RH) helical bilayer and the corresponding orientation distribution of NR. $\tau =0$ corresponds to a maximum value of $u_z$ and increases in the direction of the gray arrow. b-d, (Top) Experimental SMLM images, (Middle) SMOLM images, and (Bottom) their corresponding 3D orientation distributions on the $u_x\ge 0$ hemisphere for one typical (b) KFE8^L (LH), (c) KFE8^D (RH), and (d) Aβ42 (LH)^, fibril. More examples are shown in Figure S2. Each NR localization is depicted as a line segment whose color and orientation are set by the projection of the molecule’s 3D orientation into the $xy$ plane relative to the long axis $u_x$ of the fibril. Scale bars are 100 nm. The orientation distributions are binned using $u_y=\left(-|1|,|-|0.98|,|\dots ,|1\right)$ and $u_z=\left(-1,-0.98,\dots ,1\right)$.

Figure 2.

NR orientation distributions, localization densities, and wobble resolve the helical bilayer structure of KFE8 and Aβ42 fibrils. a, (Top) Experimental and (Bottom) simulated orientation distributions of (Left) KFE8^L, (Middle) KFE8^D, and (Right) Aβ42 with overlaid backbone tilt angles ${\delta }_{\mathrm{i}\mathrm{n}\mathrm{n}\mathrm{e}\mathrm{r}}$ and ${\delta }_{\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{r}}$ calculated using our helix optimization algorithm (see Figure S5). The circles, i.e., the measured backbone tilt angles, are color-coded with respect to helix phase $\tau$ (See Figure 1a). The orientation distributions are binned using $u_y=\left(-1,-0.98,\dots ,1\right)$ and $u_z=\left(-1,-0.98,\dots ,1\right)$. Filter: signal ≥ 500 photons, wobble cone half angle $\alpha \le \text{30}°$. b, NR (i, iii) localization density and (ii, iv-vi) wobble quantified as a half-cone angle $\alpha$ in degrees. Filter: signal ≥ 500 photons. (i and ii) Localization density and wobble are quantified with respect to backbone tilt angle $\delta$. Bin size: 1°. (iii-vi) Localization density and wobble are quantified with respect to helix phase $\tau$. (iv-vi) SMOLM measurements are represented as points; solid lines are from fits to sinusoidal functions. Bin size: $\pi /30$ (6°). Data are accumulated from all measured fibrils. See Figures S10 and S11 for full distributions.

Figure 3.

NR SMOLM quantifies heterogeneity among KFE8^L, KFE8^D, and Aβ42 fibrils. a-c, Median backbone tilt angles $\delta$ of NR measured within 30 fields of view (FOVs) for (a) KFE8^L, (b) KFE8^D, and (c) Aβ42. Hollow circles are median angles within each FOV. Red dotted lines are the backbone tilt angles ${\delta }_{\mathrm{i}\mathrm{n}\mathrm{n}\mathrm{e}\mathrm{r}}$ and ${\delta }_{\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{r}}$ of KFE8^L, KFE8^D, and Aβ42, as estimated by helical optimization of data accumulated across all FOVs shown in Figure 2a. d-f, Median wobble cone half angles $\alpha$ measured within 30 FOVs for (d) KFE8^L, (e) KFE8^D, and (f) Aβ42. Orange circles are the median angles within each FOV. Black dotted lines are median angles over all KFE8^L, KFE8^D, and Aβ42 FOVs in Figure 2b. KFE8^L FOV 26 corresponds to fibril shown in Figure 1b; FOV 4 to Figure S2a, i; FOV 16 to Figure S2a, ii; FOV 12 to Figure S2a, iii; and FOV 21 to Figure S2a, iv. KFE8^D FOV 30 corresponds to fibril shown in Figure 1c; FOV 22 to Figure S2b, i; FOV 15 to Figure S2b, ii; FOV 18 to Figure S2b, iii; and FOV 7 to Figure S2b, iv. Aβ42 FOV 7 corresponds to fibril shown in Figure 1d; FOV 29 to Figure S2c, i; FOV 22 to Figure S2c, ii; FOV 19 to Figure S2c, iii; and FOV 10 to Figure S2c, iv.

Figure 4.

Structural polymorphism of KFE8^L revealed by SMOLM: a-c, Branching and d-f, enclosed loop structures. In a-c, two “parent” branches (ROIs 1–6 and 15–32) and three “child” branches (ROIs 7–11, 12–14, and 33–37) are chosen. (a and d) SMOLM and (Inset) SMLM images. NR orientations in the SMOLM images are represented as colored disks in a (108038 localizations [59% of the data] are randomly selected for plotting in a to avoid oversaturation) and colored lines in d (all data are shown). Each disk/line is color-coded according to the projection of the 3D NR orientation into the $xy$ plane relative to the ${\mu }_x$ axis, which is defined by the microscope instead of the long axis of the fibril. (b and e) Median backbone tilt angles $\delta$ within each region of interest (ROI), which are shown in SMOLM images. Red dotted lines are the backbone tilt angles of the helical bilayer, as estimated by helical optimization of data accumulated across all FOVs shown in Figure 2a, Left. Blue circles lie within the estimated backbone tilt angles of the bilayer; red circles outside the range. (c and f) Median wobble angle $\alpha$ (orange circles) within each ROI. Black dotted lines are the median of the accumulated experimental data in Figure 2b. For more examples, see Figure S14. Scale bars: (a and inset) 1 μm, (d and inset) 100 nm.