Communication pubs.acs.org/biochemistry
Binding of Thioflavin‑T to Amyloid Fibrils Leads to Fluorescence SelfQuenching and Fibril Compaction David J. Lindberg,† Anna Wenger,†,§ Elin Sundin,‡ Emelie Wesén,† Fredrik Westerlund,† and Elin K. Esbjörner*,† †
Division of Chemical Biology, Department of Biology and Biological Engineering, Chalmers University of Technology, Kemivägen 10, 412 96 Gothenburg, Sweden ‡ Division of Chemistry and Biochemistry, Department of Chemistry and Chemical Engineering, Chalmers University of Technology, Kemivägen 10, 412 96 Gothenburg, Sweden S Supporting Information *
ABSTRACT: Thioflavin-T binds to and detects amyloid fibrils via fluorescence enhancement. Using a combination of linear dichroism and fluorescence spectroscopies, we report that the relation between the emission intensity and binding of thioflavin-T to insulin fibrils is nonlinear and discuss this in relation to its use in kinetic assays. We demonstrate, from fluorescence lifetime recordings, that the nonlinearity is due to thioflavin-T being sensitive to self-quenching. In addition, thioflavin-T can induce fibril compaction but not alter fibril structure. Our work underscores the photophysical complexity of thioflavin-T and the necessity of calibrating the linear range of its emission response for quantitative in vitro studies. isfolding and aggregation of proteins into amyloid fibrils and the subsequent formation of insoluble tissue deposits are pathological hallmarks of some of the most highly debilitating neurodegenerative diseases.1 For this reason, amyloid fibrils have been widely studied, both in vitro and in vivo. Despite large differences in the size and native structure of amyloidogenic proteins, amyloid fibrils formed from different proteins display many commonalities. They are self-associating, highly repetitive polymers consisting of arrays of β-sheets that are held together via hydrogen bonding along the peptide backbone. The β-sheets are arranged parallel to the fibril axis in a so-called cross-β fold. Amyloid fibrils are ∼10 nm in diameter but can extend up to several micrometers in length and typically have twisted and unbranched morphologies.2 Their detection is facilitated by a variety of dyes with amyloid-specific tinctorial properties that arise when the dyes interact with the cross-β fibril core; the benzothiazole derivative thioflavin-T (ThT) (Figure 1A) is the perhaps most prominent and commonly used example. ThT is a so-called molecular rotor. In water, it efficiently dissipates excitation energy via torsional motion of the benzothiazole and aminobenzene rings relative to each other, leading to a nonfluorescent twisted internal charge transfer state.3 The most common explanation to its significant increase in quantum yield upon binding to amyloid fibrils is that the mode of binding restricts rotation around the central carbon−carbon bond. The exact relationships between ThT photophysics (intensity, quantum yield, and spectral shifts) and its putative binding mode(s) are, however, complex and not
M
© XXXX American Chemical Society
Figure 1. (A) Structure of thioflavin-T (ThT). (B and C) Linear dichroism (LD) spectra of shear-aligned insulin amyloid fibrils (concentration of 50 μM) recorded in the presence of the indicated amounts of ThT. Arrows indicate the direction of change in each peak with an increase in ThT concentration. Panel C is a magnification of the lower signal peaks shown in panel B. The shear rate for alignment was 3100 s−1. (D) Change in the degree of alignment (orientation factor, S) of the insulin amyloid fibrils as a function of increasing ThT concentration. The data were retrieved from the peak value at 276 nm in of LD spectra shown in panel C and additional spectra at intermediate concentrations. The change in orientation is presented as normalized data relative to the LD signal of insulin amyloid fibrils in absence of ThT (S0). (E) Atomic force microscopy (AFM) image of 10 μM insulin fibrils deposited on mica. (F) AFM image of 5 μM insulin fibrils mixed with 3 μM ThT. The scale bars are 4 μM.
fully understood.4 Therefore, the magnitude of its fluorescence enhancement in different setups typically remains unexplained. We have for example reported that the emission intensity of ThT differs substantially when it is bound to amyloid fibrils formed by two highly similar isoforms of the amyloid-β Received: January 16, 2017 Revised: April 3, 2017 Published: April 12, 2017 A
DOI: 10.1021/acs.biochem.7b00035 Biochemistry XXXX, XXX, XXX−XXX
Communication
Biochemistry peptide.5 We found that the total emission intensity largely depends on binding site availability, whereas the quantum yield, which also differed, depends on the exact nature of the preferred fibril binding sites. A common assumption is that ThT fluorescence scales linearly with amyloid fibril mass (e.g., with the number of available binding sites). This is indeed the basis for its use as a reporter of increasing fibril mass concentration in studies of amyloid growth; such kinetic studies are important because they give very valuable insight into the molecular mechanisms of amyloid formation.6 Moreover, reductions in ThT emission intensity are often taken as an indication of inhibition. The use of ThT in these types of in vitro assays has many advantages; for example, it has been reported that ThT binding does not significantly affect fibril X-ray diffraction patterns or therefore the cross-β structure of the amyloid fibril.7 In addition, ThT does not appear to perturb aggregation rates, whereas other amyloid stains such as Congo red can act as inhibitors.8 Thus, ThT is a rightful popular choice for in vitro studies, and it is important to fully understand its fluorescence read-out and limitations thereof. Here, we challenge the view of a linear response in emission intensity to fibril concentration by studying in parallel the binding and fluorescence emission of ThT upon binding to amyloid fibrils prepared from bovine insulin. We find that even at substoichiometric dye/monomer ratios, ThT is subject to substantial self-quenching that results in a nonlinear relation between its binding and emission properties. Moreover, we demonstrate that ThT can induce fibril compaction but not alter the structure of the forming fibrils. Amyloid fibrils were prepared by incubation of insulin solutions prepared in 50 mM glycine-HCl buffer (pH 2.2) at 60 °C without shaking. The fibril solutions were diluted to a concentration of 50 μM (based on absorption measurements) and examined under shear alignment in a Couette flow cell made of quartz. The linear dichroism (LD), which is the difference in absorption of linearly polarized light oriented parallel and perpendicular to the orientation axis in an anisotropic (aligned) sample,9 was recorded between 190 and 500 nm and as a function of increasing ThT concentrations (Figure 1B,C). LD scales linearly with the isotropic absorption and reports on the orientation of absorbing transition moments in the aligned sample according to eq 1: LD = A − A⊥ ; LDr =
LD 3 = S(3 cos2 α − 1) A iso 2
The shape of the insulin portion of the LD spectra presented in panels B and C of Figure 1 is in agreement with the results of previous publications.10,16 The major positive peaks at ∼200 nm arise from the π−π* transitions in each aligned peptide bond and are consistent with the perpendicular orientation of β-sheets relative to the fibril axis in the cross-β core.11 The structured positive peaks at 276 nm and the negative dip at 230 nm correspond to the short-axis Tyr (Lb) and long-axis Tyr (La) transitions of insulin’s tyrosine side chains, respectively. The peaks at 440 nm are due to ThT binding and overlap exactly with the excitation maximum of bound ThT (Figure S1). The positive signature of these peaks shows that the longaxis transitions in the bound ThT molecules are preferentially oriented parallel to the fibril axis and is in accord with the results of others.18 Our observation therefore strengthens the view that ThT binds to amyloid fibrils primarily by insertion in the narrow channels formed between the side chains that protrude from the cross-β core.4 Upon titration of ThT to the aligned insulin fibrils, we observed a systematic decrease in the magnitude of all LD bands emanating from absorbing transitions in the insulin fibril (indicated by the arrows in Figure 1B,C), despite the fact that the absorption (concentration) of the sample remained constant and the ThT LD increases. There are two possible explanations for this observation. (1) ThT brings about structural change in the fibril, altering the angles of both βsheets and tyrosines, or (2) the alignment (S in eq 1) of the fibrils decreases. To address the first, we recorded a series of new LD spectra at a final insulin concentration of 10 μM with ThT concentrations from 1 to 10 μM (Figure S2A). This was necessary because in Figure 1B the absorption of the sample below 200 nm is too high for the LD detector to operate in its linear range (see the text in the Supporting Information), and it is thus not possible to quantitatively analyze the β-sheet peak. We find that the 276 nm/202 nm LD ratio is constant (Figure S2B); the small and nonsystematic variation observed translates into an angular variation of the tyrosine short-axis orientation relative to the fibril axis of
