A sensitive probe for amyloid fibril detection with strong fluorescence and early response

We synthesized a new probe, 4-[2-(2-naphthyl)-(E)-ethenyl]-benzyl(triphenyl)phosphonium bromide (NEB), to detect the formation of amyloid fibrils of bovine insulin. The fluorescence intensity of NEB in the presence of insulin fibrils was 30 times higher than that before fibrillation, with the fluorescence quantum yield increased from 2.5% to 78%. In comparison with the commercially available probe, thioflavin T (ThT), NEB exhibits a 10 times stronger fluorescence and a shorter identification lag phase for detecting insulin fibrillation, indicating a higher sensitivity in detection of insulin oligomers and fibrils.


Introduction
Amyloid brils are a denatured brous form of protein aggregates, and have been extensively investigated because their deposition causes several serious diseases, such as Alzheimer's and Parkinson's diseases, type II diabetes and prion diseases. 1 Note that the brils formation is not only limited to the protein related to the neurodegenerative diseases. It has been widely accepted that the bril formation is an intrinsic property of the polypeptide backbones. 2 Potentially, any protein could form amyloid brils under specic conditions. 3 The bril structure appears generic linear laments with a length of several micrometres and a width about 10 nm. Fibril X-ray diffraction studies have revealed an antiparallel b-sheets oriented perpendicular to the long axis of the bril. 4 As these brils and/or amyloid bril intermediates (oligomers) can kill cells or prevent them from functioning properly, detecting these brils and understanding the mechanism of the bril formation are important for the drug development against these neurodegenerative diseases. 5 Recently, different characterization techniques of amyloid brils have been developed, such as absorption, circular dichroism spectra, mass spectrometry and atomic force microscopy as well as uorescence. 6 Among them, uorescence is the most popular one, as it allows a direct observation of the whole brillation process. 6 In the past decades, a great variety of uorescent probes have been developed to monitor the brillation process, including small organic molecules, gold nanoparticles, uorescent proteins and conjugated polymers. 2,7 In particular, thioavin T (ThT) is a well-known probe for amyloid brils detection with several advantages. (i) ThT molecules are water-soluble and are extensively used both in vitro and in vivo. 8 (ii) ThT interacts only with the mature brils and does not interfere with the amyloid brillation process. 9 (iii) When bound to the amyloid brils, the ThT uorescence increases signicantly. 10 Nonetheless, detecting the oligomeric of the proteins which is the main contributor of the neurotoxicity is still a great challenge for the ThT. On the other hand, the low uorescence intensity is another major drawback for the ThT probe. In these regard, great attention have been paid in searching for efficient dyes with better performance. According to the research efforts of scientists, the structure of molecular rotors are particularly interesting for use as amyloid binding agents due to their distinct spectral changes depending on environmental factors and an increase in quantum efficiency upon being geometrically conned. 11 For example, Catherine C. Kitts et al. demonstrated that Michler's hydrol blue is an excellent amyloid bril probe, which exhibits a characteristic red-shi in its excitation spectrum and an increase in the emission quantum yield upon binding to the amyloid brils. 12 Ben Zhong Tang et al. have reported a dye TPE-TPP as a probe to monitor the a-synuclein (a-Syn) brillation process which shows a better performance over the ThT on the emission intensity and the sensitivity towards the oligomeric form of the a-Syn. 13 4-[2-(2-Naphthyl)-(E)-ethenyl]benzyl(triphenyl)phosphonium bromide (NEB) is a typical molecular rotor which has been known as the synthetic intermediates of the carbohelicenes (Fig. 1A). 14 In this work, we will report the optical characterization of NEB to monitor amyloid bril formation. We will demonstrate that NEB is indeed an excellent amyloid bril probe. The NEB exhibits the weak uorescence in the solution of the native protein but exhibits an increase in quantum efficiency when binding to the mature brils of bovine insulin. Compared to ThT, NEB exhibits a 10 times stronger uorescence intensity upon binding to the insulin brils which arise from the increased the k r and decreased k nr . The NEB is utilized to monitor the insulin brillation process and demonstrated a good amyloid bril probe with a shorter identication lag phase for detecting insulin brillation.

Materials and instruments
NEB was synthesized and conrmed by mass spectrometry (GCT-MS Micromass, UK) and 1 H NMR. Thioavin T (ThT) and insulin from bovine pancreas were purchased from Sigma-Aldrich with no further purication.
Insulin amyloid brils were prepared according to the protocols described elsewhere. 12 In the study of using NEB and ThT as an ex situ probe, an aliquot of the insulin solution taken out from the incubation mixture at a dened time was diluted with Tris-HCl buffer, followed by the addition of the probe. In the study of using NEB and ThT as an in situ probe, they were added to insulin solution prior to incubate at 60 C with constant agitation at 600 rpm. The nal concentrations of insulin, NEB and ThT were 5 mM, 2 mM and 2 mM, respectively.
The steady-state absorption spectra were measured on a Shimidazu UV-3600 UV-VIS-NIR spectrophotometer. The stationary uorescence spectra were performed on a Horiba FluoroMax-4-NIR spectrophotometer equipped with an integrating sphere. The relative uorescence quantum yields of solutions and colloidal suspensions were measured by using 9,10-diphenylanthracene as a reference (F ¼ 0.95). Fig. 2A presents the steady-state absorption (black line) and uorescence (red line) spectra of NEB in the dilute ethanol solution. The lowest S 0 / S 1 transition of NEB monomers in the dilute ethanol solution is a broad and structured band with a molar extinction coefficient of 33 263 M À1 cm À1 at 325 nm. The monomer uorescence spectrum exhibits a vibronic progression, with the maximum at 384 nm. To verifying that NEB belongs to a class of molecular rotors, 15 we dissolved NEB in a mixture of ethanol and glycerol. The concentration of glycerol was varied from 0% to 80% to increase the viscosity. It can be seen from Fig. 2B that the emission intensity of NEB in the ethanol/glycerol solution show a linear increase with increasing the viscosity. The low orescence intensity of NEB in 100% ethanol can be ascribed to the torsional relaxation around C-C bond in the excited state. However, upon increasing the viscosity, the intramolecular rotation is greatly restricted. As a result, the NEB molecules become highly emissive in solvent of high viscosity.

Results and discussion
Upon excitation in ethanol, NEB goes from a locally excited state to a non-uorescent TICT state upon internal rotation around C-C bond. This internal rotation which nonradiatively annihilates the excited state is much faster than the uorescence decay, making NEB free in solution exhibit weak uorescence intensity. However, upon increasing the viscosity of the solution, the intramolecular rotation is greatly restricted. The restriction of intramolecular rotation blocks relax into the TICT state and make the excited state remain in the LE state which gives off its energy in the form of uorescence. As a result, the NEB molecules become highly emissive in solvent with high viscosity. Notably, NEB is water-soluble with a saturation concentration of 1 Â 10 À5 M. These unique properties make NEB attractive as a probe for amyloid bril structure determination in vitro and in vivo.
Insulin brils were prepared by dissolving the protein in water solution (0.01 M HCl, 5 mg mL À1 for insulin) and incubated in an orbital thermomixer with constant agitation at 600 rpm at 60 C for 24 h. We used ThT as probe to detect the mature insulin amyloid brils. Aer 24 h incubation, the emission signal from ThT at 487 nm greatly increases, indicating the formation of mature insulin amyloid brils (Fig. 3).  The transmission electron microscopy (TEM) in Fig. S1 † shows that mature brils have a length of several micrometres with an average width of 7-10 nm. When the mature insulin brils were introduced into the solution with the NEB, distinct spectral changes were observed compared to that of NEB in water: (i) The absorption spectrum has a 9 nm red-shi from 324 nm to the 333 nm (Fig. S2 †). (ii) The uorescence spectrum also exhibits a 7 nm red-shi from 384 nm to the 391 nm ( Fig. S2 †). (iii) When the two un-normalized spectra were compared in Fig. 3A, enhanced uorescence emission signal of NEB is observed in the presence of mature insulin brils. The uorescence intensity for the NEB in the presence of insulin brils was 30 times higher than that of NEB with insulin before brillation with a uorescence quantum yield increased from the 2.5% to the 78%. These distinct spectral changes indicate that NEB is capable of binding to the insulin brils and detecting the mature amyloid brils. To make sure that NEB is binding specically to insulin brils and not to native protein, we compared the absorption and uorescence spectra of NEB containing insulin before brillation with that of NEB in water solution and no difference is found (Fig. S3 †). Thus, we can predict that NEB binds specically to the b-sheet structure of the amyloid brils in the same way as ThT. We have measured the absorption and emission spectra of NEB in different solvents. It can be seen that the maximum absorption peak shis from 324 nm (water) to 331 nm (toluene), showing a red-shi of 5 nm. Correspondingly, the uorescence emission maximum shis from 384 to 389 nm with reducing polarity of the solvent (Fig. S4 †). So the red-shi of the absorption and uorescence spectra of NEB from in water to bound with brils can be ascribed to the reduced polarity at the NEB binding site in the insulin amyloid brils.
The dramatic increase of uorescence of NEB when bound to insulin brils is similar to that of when NEB in high viscosity solvent. The increase in emission intensity of NEB upon binding to the brils also arises from the restriction of the NEB's internal rotation around the C-C bond. Excitingly, NEB displays uorescence intensity more than 11 times stronger than ThT under the same conditions. The stronger uorescence intensity is also observed clearly in the photographs of insulin amyloid brils stained by NEB and ThT taken under a UV lamp (365 nm) in Fig. 3.
In order to understand the uorescence enhancement mechanism, we measured the lifetime data of NEB and ThT in the water alone and in solution with mature insulin brils. ThT bound to insulin bril at 452 nm exhibits bi-exponential uorescence decay with an average excited-state lifetime 1.02 ns containing a lifetime of 0.44 ns (A1 ¼ 0.53) and a slower component 1.68 ns (A2 ¼ 0.47). The ThT in the insulin before brillation decays monoexponentially with a faster lifetime < 10 ps (Fig. 4). These are all consistent with earlier reports. 16 However, the lifetime could not be calculated accurately because of the limitation by the resolution of our apparatus. Thus, the three order of magnitude slower lifetime of ThT bound to insulin bril compared to the ThT in water leads to its dramatic increase of the uoresce intensity. 8 The NEB in water follows single exponential decay with s ¼ 0.50 ns. Aer brillation, the time resolved uorescence decays of the NEB bound to insulin brils was represented by a doubleexponential t with a lifetime of 0.42 ns (A1 ¼ 0.43) and a slower component 1.86 ns (A2 ¼ 0.57). The average excited-state lifetime is 1.24 ns. Then we calculated the radiative decay rates (k r ) and nonradiactive decay rates (k nr ) based on equations of k r ¼ F/s and F ¼ k r /(k r + k nr ). It can be seen from Table 1 that the value of k r increases almost one order of magnitude from k r,water ¼ 0.05 ns À1 to k r,brils ¼ 0.62 ns À1 ; meanwhile, the value of k nr decreases one order of magnitude from k nr,water ¼ 1.95 ns À1 to k nr,brils ¼ 0.17 ns À1 . Therefore, the greatly enhanced uorescence can be ascribed to the increase of the k r as well as the decrease of the k nr when bound to the mature insulin brils.
The dissociation constants (K d ) of NEB and ThT were derived by using a xed concentration of the mature insulin brils with varying dye concentrations and tted using a one site binding equation. The resulting K d for NEB and ThT are 3.36 mM and 6.68 mM, respectively (Fig. S5 †). According to reports, ThT binds into channels formed by side chains of the residues of b-strands on the surface of the b-sheet of the bril, and binding is predominantly along the bril axis. 17 To examine whether NEB and ThT have the same mode of binding to amyloid bril, a new experiment was performed. We added ThT into the amyloid brils' solution, and then added the NEB into the solution aer ThT binding to the brils. We dened the sample as NEB @ ThT, the concentration of NEB and ThT in NEB @ ThT are all 2 mM. The uorescence emission signal from ThT is signicantly reduced in the presence of NEB (Fig. 5A), indicating that NEB may compete for the same binding sites with ThT. Considering that this excited wavelength of NEB could not excite the ThT, and the NEB themselves display bluish emissive upon UV excitation (Fig. 3A), with a broad spectrum covering 350-500 nm that overlaps absorption spectrum of ThT (Fig. S6 †), energy transfer may takes place from excited NEB to ThT. Under the excitation wavelength 310 nm of the NEB, we collected the emission signal of 487 nm typical of ThT and the emission signal of 391 nm typical of NEB at the same time. Moreover, the emission signal of the ThT excited at 310 nm exhibits a higher intensity than that at the excitation wavelength 410 nm. This result suggest that efficient energy transfer takes place from excited NEB to ThT in NEB @ ThT bound with mature brils.
The absorption spectrum of the sample NEB @ ThT exhibits the absorption peaks with a strong peak at 325 nm ascribed to the NEB and a relatively weaker peak at 410 from the ThT ( Fig. S7 †). Fluorescence excitation spectrum monitored at the ThT emission (520 nm) agrees well with the absorption spectrum, revealing energy levels including not only the NEB but also ThT. This is another piece of clear evidence for the energy transfer from the donor (NEB) to the acceptor (ThT). Furthermore, the process of energy transfer affected the uorescence decay curve of the donor NEB. The uorescence of NEB bound to insulin brils alone decays double-exponential t with a lifetime of 0.42 ns (A1 ¼ 0.43) and a slower component 1.86 ns (A2 ¼ 0.57) ( Fig. 4A and S7B, † red line). However, the uorescence decay of NEB in the NEB @ ThT bound to mature insulin brils becomes faster with an average lifetime of 0.51 ns, which would remain unchanged if energy transfer occurred via radiative mechanism, indicating that binding to the brils make sure the proximity of the NEB and ThT. This is consistent with the spacing of b-sheets which formed the binding sites for ThT and NEB is about 0.65-0.7 nm. It is reported that the main interactions between ThT and amyloid brils are hydrophobic effect and p-stacking which does not have an effect on the brillation kinetics. Considering that NEB has the similar binding sites with the ThT, we can make a prediction that NEB would not affect the brillation processes.
We then used NEB and ThT to monitor the amyloid brillation processes, respectively (Fig. 6). The maximum emission intensity of the NEB and ThT were utilized. Through following the emission intensity of the NEB at 391 nm, the amyloid brillation can clearly be monitored by the uorescence enhancement. The uorescence intensity of the samples incubated for 12 to 16 h increased about 30 times compared to the sample before incubation, which is in good agreement with the results obtained from the experiment of adding the NEB into the mature brils directly in Fig. 3A. Moreover, the lifetime of the NEB during the incubation process was performed. Aer incubating for 6 h, the time resolved uorescence decays of the NEB bound to insulin brils was also represented by a doubleexponential t with a lifetime of 0.41 ns (A1 ¼ 0.78) and a slower component 1.82 ns (A2 ¼ 0.22). The average excited-state lifetime is 0.72 ns. (Fig. S8 †) Comparing to the NEB in water alone, the value of k r was increased to 0.42 ns À1 with the value of k nr was decreased to 1.0 ns À1 . These results veried that the uorescence enhancement by the restriction of the NEB's internal rotation around the C-C bond when bound to the insulin brils. An initial lag phase, an exponential growth phase, and a nal plateau phase are present in the Fig. 6, which is in good agreement with the results obtained from the similar   experiment using ThT. NEB exhibits a stronger emissive during the whole process of the brillation compared with the uorescence from the ThT (Fig. 6, red line). Closer observation of the kinetics of the insulin bril formation reveals that there is an earlier response to the lag phase of insulin brillation by NEB than ThT. At t ¼ 2 h aer incubation, all the emission of probes only show background signals. Fluorescence enhancements were observed at t ¼ 2 h with NEB and at t ¼ 4 h with ThT. As is reported, the ThT only interacts with the mature brils.
The earlier response of the NEB in Fig. 6 indicates that NEB is capable of detecting the formation of an intermediate which is signicantly different from the monomeric and brillar forms of the insulin. Considering that energy transfer could take place from NEB to ThT, another experiment was performed. ThT and NEB were added into the solution successively with different incubation times. The concentrations of NEB and ThT in the NEB @ ThT system are all 2 mM. The emission intensity of NEB in NEB @ ThT with excitation wavelength 310 nm (391 nm, blue line in Fig. 6A), ThT with excitation wavelength 310 nm (487 nm, green line in Fig. 6A) and 410 nm (487 nm, violet line in Fig. 6A) were utilized respectively with different time points during the formation of amyloid brils of insulin. Compared with NEB alone (black line in Fig. 6), every phase of NEB in NEB @ ThT is in good agreement with the result of NEB alone. However, the uorescence emission intensity of NEB is reduced in the presence of ThT, which is ascribed to the energy transfer from NEB to ThT. For the ThT in NEB @ ThT (Fig. 6, violet line), it also show the similar "S" curve with a decreased uorescence intensity, which agrees well with the result of the competition experiment in Fig. 5B. The uorescence emission signal of ThT (Fig. 6, green line) excited at 310 nm also produced a "S" curve with three different phase. Closer observation of the emission spectra reveals that the uorescence enhancement was observed at t ¼ 4 h. The 2 h lag time compared to NEB alone is ascribed to that the amount of ThT bound with insulin is not enough for energy transfer at this moment, which in accordance with that NEB might detect the oligomeric of the proteins which is still a challenge for ThT. Recently, it is reported that the oligomer form of protein, rather than the brillar form, is the main contributor of neurotoxicity. 18 The earlier response with the stronger uorescence intensity mean the NEB will be a potential and promising candidate uorescent probe for brillation detection and early stage detection of the amyloid diseases.

Conclusions
In conclusion, we have demonstrated a new blue-emissive, high uorescence intensity, earlier response probe NEB for monitoring amyloid brillation process. The NEB exhibits the weak uorescence in the solution of the native protein but exhibits great increase in quantum efficiency when bound to the mature brils of bovine. A thorough photophysical study has shown that the enhanced uorescence arise from the increased the k r and decreased k nr . Compared to ThT, NEB exhibits a 10 times stronger uorescence and a shorter identication lag phase for detecting insulin brillation, indicating a higher sensitivity in detection of insulin oligomers and brils. Considering the short emission wavelength of the NEB, the tailoring of the molecular structures for the red-shi of the emission spectrum was needed. This work is under way in our laboratories and will be published elsewhere.

Conflicts of interest
There are no conicts to declare.