TIRF imaging

TIRF imaging is now one of the most common fluorescence microscopic imaging to track single particles by selectively illuminating one area to minimize background signals. The innovation of TIRF has greatly enhanced the signal-to-noise ratio by reducing the background fluorescence, resulting in better contrast for detecting single molecules (Schneckenburger, Reference Schneckenburger2005). For TIRF, a solid–liquid interface with different refractive indexes, such as a glass coverslip/slide and a water-based physiological solution, is needed (Fig. 1a) (Shashkova and Leake, Reference Shashkova and Leake2017). The incidence angle with a 90° refraction is called a ‘critical angle’. At larger angles than the critical angle, the excitation light will not transmit through the liquid surface, instead, a total internal reflection of the incident light occurs at the interface and an ‘evanescent field’ is created. The intensity from this field drops off exponentially with the increasing distance from the interface whereas the depth of penetration can actually be adjusted over a wide range. Therefore, TIRF imaging can restrict the excitation to fluorophores that are positioned close to the surface, making this technique particularly suitable for identifying molecules integrated into membranes.

Fig. 1.

Schematic illustrations of the basic concept of SMFI techniques. (a) TIRF uses total internal reflection at the interface where an ‘evanescent field’ is produced and the fluorophores close to the glass surface are efficiently excited by the evanescent field. (b) In FCS method, fluorescent molecules diffuse through a tiny confocal volume and the fluctuations of fluorescence intensity are detected to determine the number of particles in this area. The autocorrelation function provides information about the concentration and the thermodynamic and kinetic parameters. (c) FRET represents a non-radiative energy transfer between a donor and an acceptor chromophore molecule via the long-range dipole–dipole coupling when these molecules are in close proximity (left). The donor fluorescence is quenching while the emission of the acceptor is increasing (right). (d) TCCD is based on oligomers with two distinct fluorophores but cannot exhibit high FRET give coincident fluorescent bursts after emitting the photons (left). The fluorescence intensity derived from the two-color channels (red and blue) presents the numbers of monomers that are labeled in one oligomer (right). (e) FRAP can be used to quantify the recovery of fluorescence intensity over time after a region of interest in a cell or tissue is photo-bleached with a high-intensity laser and subsequently probed by a low-power laser.

TIRF has been applied for the direct visualization of amyloid aggregates. As using covalently labeling proteins might affect the aggregation process (Zheng et al., Reference Zheng, Xu, Yang, Peng, Wang, Yu, Hua, Zhao, He and Hong2018) and it is challenging to tag samples extracted from in vivo sources, Horrocks et al. (Reference Horrocks, Lee, Gandhi, Magdalinou, Chen, Devine, Tosatto, Kjaergaard, Beckwith, Zetterberg, Iljina, Cremades, Dobson, Wood and Klenerman2016) relied on the addition of a dye (Thioflavin-T (ThT)) that non-covalently binds to oligomers and fibrils with β-sheet structures. Small β-sheet aggregates of α-syn, Aβ and tau in vitro are observed with the sizes <250 nm at the beginning of aggregation and the enriched oligomers contain on average 25 monomers. Amyloid fibrillation kinetics can be tracked as well. In the presence of ThT, Yuji's group (Ban et al., Reference Ban, Hoshino, Takahashi, Hamada, Hasegawa, Naiki and Goto2004) clearly observed the real-time individual Aβ40 fibrils growth with a speed-dependent manner and found the constant rate of fibril growth is similar with the steady-state enzymatic kinetics. Another study showed that glucagon fibrils grow exclusively by forming branches along sides of the parent fibril with an angle of 35°–40° (Andersen et al., Reference Andersen, Yagi, Manno, Martorana, Ban, Christiansen, Otzen, Goto and Rischel2009), which differs from Aβ fibrils that extend themselves from the end of the preformed seeds in a linear elongation. By using TIRF, Patil et al. (Reference Patil, Mehta, Jha and Alexandrescu2011) revealed three different human islet amyloid polypeptide (hIAPP) aggregation patterns that demonstrate the role of secondary nucleation mechanisms where the preformed fibrils serve as a new template for fibril formation. All these studies illustrate that TIRF microscopy is an effective tool to visualize entire aggregates and fibril growth kinetics.

TIRF was also applied to depict the permeation of single vesicles or cells filled with fluorescent dye by measuring the Ca²⁺ influx induced by oligomers on the membranes. After the penetration of synthetic Aβ oligomers (down to 200 pM) into the live neuronal astrocyte cells (Drews et al., Reference Drews, Flint, Shivji, Jönsson, Wirthensohn, De Genst, Vincke, Muyldermans, Dobson and Klenerman2016; Flagmeier et al., Reference Flagmeier, De, Wirthensohn, Lee, Vincke, Muyldermans, Knowles, Gandhi, Dobson and Klenerman2017), tetramers or larger oligomers were found as the most toxic Aβ oligomer aggregates across the membranes. This suggests that membrane permeabilization is one of the toxic mechanisms induced by oligomers. The extracellular chaperone clusterin, antibody or nanobody that specifically binds to oligomers could prevent Ca²⁺ influx induced by synthetic Aβ oligomers (Drews et al., Reference Drews, De, Flagmeier, Wirthensohn, Chen, Whiten, Rodrigues, Vincke, Muyldermans, Paterson, Slattery, Fox, Schott, Zetterberg, Dobson, Gandhi and Klenerman2017), whereas these inhibition effects are not obvious under cerebrospinal fluid (CSF) conditions. This implies that these SM assays might be a cost- and time-effective approach before clinical trials. Recently, with TIRF, Tanaka and coworkers showed that Aβ42 oligomers on the surface of live hippocampal neurons impair the ability of learning and memory in the early stage of AD via suppressing the increase of fluorescently labeled (glutamate receptor) GluA1 number and exocytosis frequency in both post-synaptic and extra-synaptic membranes (Tanaka et al., Reference Tanaka, Sakaguchi and Hirano2019). The increase of GluA1 is important for the cellular basis of synaptic plasticity (Jarosz-Griffiths et al., Reference Jarosz-Griffiths, Noble, Rushworth and Hooper2016). Based on this result, a feasible therapy to recover mild memory impairment could be to target the increase of exocytic GluA1 on neuronal cells and TIRF could be an effective imaging tool to validate this potential.

A novel approach, label-free scattering microscopy has been used to image the surface-bound lipids and detect the specific protein-binding to single vesicles without labeling the lipids (Agnarsson et al., Reference Agnarsson, Lundgren, Gunnarsson, Rabe, Kunze, Mapar, Simonsson, Bally, Zhdanov and Höök2015). Scattering microscopy produces the ‘evanescent field’ by optical waveguide instead of using the objective in TIRF. In this method, a more uniform illumination profile is obtained. Recently, Hannestad et al. investigated the kinetics of monomeric α-syn interaction with different lipid vesicles under both fluorescence and light-scattering emission at the single-vesicle level. They found α-syn induces obvious step-wise structural changes and asymmetric membrane deformation of individual phosphatidylglycerol vesicles with a very low α-syn concentration 10 nM (~100 proteins per vesicle) (Hannestad et al., Reference Hannestad, Rocha, Agnarsson, Zhdanov, Wittung-Stafshede and Höök2020). The pore formation mechanism was not concluded to result in membrane leakage when α-syn binds to the vesicles. This finding might enlighten the abnormal behavior of monomeric α-syn in vivo. Scattering microscopy was also used to distinguish misfolded proteins from normal ones in Aβ plaques of an AD mouse brain model (Ji et al., Reference Ji, Arbel, Zhang, Freudiger, Hou, Lin, Yang, Bacskai and Xie2018). Aβ plaques were found to form a typical core–shell structure that is surrounded by a halo-like shell of lipid-rich deposits (Schweikhard et al., Reference Schweikhard, Baral, Krishnamachari, Hay and Fuhrmann2019). This result indicates the ability of scattering microscopy in providing novel perspectives into the biochemical constitution of pathological structures.

FCS method

FCS is a quantitative analysis method with the ultimate SM sensitivity for the detection of fluorescent molecules (Rigler, Reference Rigler2010; Elson, Reference Elson2018). In FCS, the fluctuations of fluorescence intensity, originating from the diffusion of fluorescent molecules (physiological nanomolar to picomolar) through a very tiny confocal volume (~1 μm³) is measured with a high-temporal resolution (nanosecond time scale) (Fig. 1b). The detected fluorescence real-time fluctuations and autocorrelation function provide the thermodynamic or kinetic parameters such as diffusion coefficient (DC) or time, nucleation boundary radius and the free energy barrier for nucleation (Medina and Schwille, Reference Medina and Schwille2002; Garai et al., Reference Garai, Sahoo, Sengupta and Maiti2008; Kitamura and Kinjo, Reference Kitamura and Kinjo2018). Knowing these parameters can assist in developing an exact thermodynamic model of Aβ aggregation. Moreover, understanding the energy landscape of amyloids may estimate the stability of intermediates, which could help the quest for a method to reduce its population (Garai et al., Reference Garai, Sengupta, Sahoo and Maiti2006).

FCS has been used to characterize the amyloids aggregation process and determine the presence of different oligomeric species, both in well-equilibrated systems in vitro and more complicated environments in vivo (Liang et al., Reference Liang, Lynn and Berland2010; Guan et al., Reference Guan, Cao, Cantlon, Elbel, Theodorakis, Walsh, Yang and Shah2015; Tiiman et al., Reference Tiiman, Jarvet, Gräslund and Vukojević2015). Mittag et al. (Reference Mittag, Milani, Walsh, Rädler and Mcmanus2014) fitted FCS data to Gaussian distribution models to clearly distinguish different Aβ oligomeric species in vitro. They estimated the sizes of different amyloid types, including monomer (<1 nm) and low-molecular-weight oligomers (1–5 nm), high-molecular-weight oligomers (5–40 nm), protofibrils (40–4000 nm) and fibrils (4000–20 000 nm). Additionally, Rigler's group observed that Aβ peptides have a highly cooperative polymerization process where the fibril is predominantly built up from monomers or dimers rather than free oligomers (Tjernberg et al., Reference Tjernberg, Pramanik, Björling, Thyberg, Thyberg, Nordstedt, Berndt, Terenius and Rigler1999). The size of small Aβ aggregate particles in vitro is 1.7–2.1 nm and big aggregates have sizes of 100 nm or higher (Garai et al., Reference Garai, Sureka and Maiti2007). Tiiman et al. (Reference Tiiman, Jarvet, Gräslund and Vukojević2015) later suggested that the smallest ThT-visible Aβ aggregates in vitro with a diffuse time ~200 μs, is comprised of 40–70 Aβ monomers. By directly tracking Aβ42 concentration changes, Novo et al. (Reference Novo, Freire and Al-Soufi2018) found that the critical aggregation concentration of Aβ42 to form oligomers is about 90 nM. Based on this, they suggested that the assembly process is spontaneous and cooperative which is similar to the process of surfactants self-assembling and produces the stable micelle-like oligomers formed from 50 monomers (hydrodynamic radius, R _h, 7–11 nm). In terms of complex samples, with FCS, Pitschke et al. (Reference Pitschke, Prior, Haupt and Riesner1998) detected the very large multimers of single Aβ aggregates in all CSF samples from 15 AD patients but in none of the 19 age-matched healthy controls. Funke et al. (Reference Funke, Birkmann, Henke, Görtz, Lange-Asschenfeldt, Riesner and Willbold2007) also easily differentiated three CSF AD samples from the control ones by the presence of Aβ aggregates. FCS is also suitable for identifying the inhibitors of the initial formation of amyloid oligomers by analyzing the average size and number of oligomeric particles. D-enantiomeric amino acid peptides were demonstrated to reduce the average size of Aβ aggregates but increase their number (Wiesehan et al., Reference Wiesehan, Stöhr, Nagel-Steger, Van Groen, Riesner and Willbold2008). These results demonstrate the effectiveness of FCS in determining heterogeneous oligomeric species. FCS can also be used to quantitatively investigate the binding of amyloid proteins on membranes. Fluorescently labeled α-syn has been demonstrated to preferentially bind to acidic vesicles (Miraglia et al., Reference Miraglia, Ricci, Rota and Colla2018). Less bulky head groups of anionic lipids enhance this binding possibly due to the anionic lipids being able to pack more closely together, thus producing a high-charge density (Rhoades et al., Reference Rhoades, Ramlall, Webb and Eliezer2006; Middleton and Rhoades, Reference Middleton and Rhoades2010). The requirment of binding a single α-syn molecule is a minimum of 30 lipid molecules or a 60-molecule bilayer patch (Rhoades et al., Reference Rhoades, Ramlall, Webb and Eliezer2006). Additionally, α-syn prefers to bind small vesicles such as synaptic vesicles, over large multi-lamellar vesicles. It was suggested that the ideal match of vesicle curvature with α-syn would be as found in the N-terminal helix (Middleton and Rhoades, Reference Middleton and Rhoades2010). Another study showed that labeled Aβ40 and Aβ42 oligomers exhibit stronger binding to neuronal and somatic cell membranes than monomers at physiological concentrations (250 nM) (Sarkar et al., Reference Sarkar, Das and Maiti2013). The negligible binding to cell membranes by monomers explains why they are non-toxic. Additionally, on mammalian cell membranes, the size distribution of small Aβ40 aggregates (2.3 nm average R _h) peaks at both tested concentrations of 150 and 350 nM, while only large assemblies of Aβ40 (≫10 nm average R _h) appear at a concentration of 350 nM (Nag et al., Reference Nag, Chen, Irudayaraj and Maiti2010). These results indicate that the formation of large Aβ complexes on living cells relies on Aβ concentration. This discovery might partially explain why the Aβ targeting drugs fail in clinical trials since the much higher physiological concentration of Aβ is used in vitro. Recently, Simpson et al. (Reference Simpson, Szeto, Boukari, Good and Leach2020) identified differences of Aβ aggregation and cytotoxicity between 2D in vitro models (typically culture cells on flat plastic or glass substrates) and 3D environments, such as collagen hydrogels in tissue culture models. Combined with FCS, they found that the size of Aβ aggregates in 3D collagen hydrogel is 25–200 times larger in diameter than 2D culture, suggesting a vastly different aggregation process. It appears that the cytotoxicity of Aβ gets attenuated in 3D compared to 2D environments. The difference between crowding and confinement is the main contribution. This discovery might provide another reason for failed Aβ drug screening models that might miss these two factors.

Fluorescence cross-correlation spectroscopy (FCCS), which is an extension of FCS, was developed by Schwille et al. (Reference Schwille, Meyer-Almes and Rigler1997). In FCCS, there are two differently labeled fluorescent species whose concentration and diffusion characteristics are measured in parallel. In addition, information about whether the two molecules interact can be obtained. The principle is the same with FCS, but the overlapping diffraction-limited confocal volumes exists in FCCS due to two different lasers focusing onto the same spot. FCCS has been used to study the interaction of aggregation inhibitors such as antibodies (Bonito-Oliva et al., Reference Bonito-Oliva, Schedin-Weiss, Younesi, Tiiman, Adura, Paknejad, Brendel, Romin, Parchem, Graff, Vukojević, Tjernberg, Terenius, Winblad, Sakmar and Graham2019) and kinesin-1 (Zheng et al., Reference Zheng, Tian, Peng, Yang, Fu, Jiao, Zhao, He and Hong2016) with Aβ peptide. Two kinesin-1 subunits, KLC-E and E-KLC, were tagged to determine the interaction of kinesin-1 on Aβ peptide at the SM level (Zheng et al., Reference Zheng, Tian, Peng, Yang, Fu, Jiao, Zhao, He and Hong2016). Data revealed both KLC-E and E-KLC have an inhibitory effect on Aβ42 aggregation over a period of time, and KLC-E shows a much higher binding affinity than E-KLC, which might result from the spatial conformation of KLC-E.

FRET, TCCD and FRAP techniques

FRET is one of the most commonly used techniques when studying physical interactions between two molecules within several nanometers (30–70 Å range), a distance sufficiently close for intermolecular interactions to occur (Roy et al., Reference Roy, Hohng and Ha2008; Schuler and Eaton, Reference Schuler and Eaton2008). The principle of FRET is based on the excitation energy transfer from a donor to an acceptor chromophore molecule with the reduced donor fluorescence/lifetime and increased emission of acceptor fluorescence/lifetime. For protein interaction studies, fluorophore-tagged antibodies or fusion proteins are the two common methods to adhere to the fluorophores to the interacting proteins.

Based on the direct observation of coincident bursts of oligomer fluorescence, two distinct structural types of α-syn oligomers were inferred during α-syn aggregation in vitro (Cremades et al., Reference Cremades, Cohen, Deas, Abramov, Chen, Orte, Sandal, Clarke, Dunne, Aprile, Bertoncini, Wood, Knowles, Dobson and Klenerman2012). Oligomers of types A and B, with medium and higher FRET efficiencies, respectively, were found at α-syn aggregation different stages. Type B oligomers with a more compact structure and proteinase-K resistance were monitored in later aggregation stage, indicating type A might undergo a structural reorganization to form type B. Structural conversion has been assumed to be the key step of α-syn aggregation. This result is consistent with the hypothesis of the sequential finding of discrete oligomeric intermediates in the early stage of α-syn aggregation (Conway et al., Reference Conway, Lee, Rochet, Ding, Williamson and Lansbury2000). Additionally, they detected toxic α-syn oligomers originating from disaggregation of mature α-syn fibrils, suggesting that the disaggregation of fibrils can also be a source of toxic species and result in toxicity (Cremades et al., Reference Cremades, Cohen, Deas, Abramov, Chen, Orte, Sandal, Clarke, Dunne, Aprile, Bertoncini, Wood, Knowles, Dobson and Klenerman2012). This implies that attention should also be paid to the disaggregation of fibrils at the final stage beside the toxic forms in the early stage of the aggregation process. Later, Horrocks and co-workers combined fast flow microfluidics with FRET to investigate the oligomerization of α-syn (Horrocks et al., Reference Horrocks, Tosatto, Dear, Garcia, Iljina, Cremades, Dalla Serra, Knowles, Dobson and Klenerman2015). Isolating type A from type B oligomers by dilution, they discovered that type A dissociates but type B stabilizes in the lower ionic strength of the solution. Moreover, type B oligomers from two α-syn mutant forms (A53T and A30P) displayed different FRET efficiencies, suggesting that the variants contribute to the structural diversity of oligomeric ensembles (Tosatto et al., Reference Tosatto, Horrocks, Dear, Knowles, Dalla Serra, Cremades, Dobson and Klenerman2015). By systematic analysis of the FRET efficiency values in a wide range of starting α-syn monomer concentration (from 0.5 to 140 μM), the required number of seeds to affect the initial aggregation was predicted to be in the order of 10⁴ species in a single cell (10 μm)³ (Iljina et al., Reference Iljina, Garcia, Horrocks, Tosatto, Choi, Ganzinger, Abramov, Gandhi, Wood, Cremades, Dobson, Knowles and Klenerman2016b). This demonstrates that FRET is suitable to quantitatively explore the seeding process of amyloid aggregation. The same setup also demonstrated that tau protein has a similar structural trajectory evolving from type A to type B oligomers prior to fibril formation (Shammas et al., Reference Shammas, Garcia, Kumar, Kjaergaard, Horrocks, Shivji, Mandelkow, Knowles, Mandelkow and Klenerman2015). Combining FRET with the FCS method, the fraction of small Aβ oligomer containing <10 Aβ molecules were detectable as low as 2 ± 2% in solution at physiological concentrations (Wennmalm et al., Reference Wennmalm, Chmyrov, Widengren and Tjernberg2015). Additionally, Kaminski Schierle et al. (Reference Kaminski Schierle, Bertoncini, Chan, Van Der Goot, Schwedler, Skepper, Schlachter, Van Ham, Esposito, Kumita, Nollen, Dobson and Kaminski2011) found that labeled α-syn in living worm Caenorhabditis elegans has the slowest decrease in fluorescence lifetime, followed by in vitro samples, then by neuronal cells. It indicates that the aggregation rate of α-syn is environmentally-dependent with the fastest rate of aggregation in cells. This result shows that FRET is an effective tool to track the kinetics of α-syn self-assembly in vitro and in vivo, and the molecular environment should be taken into consideration when revealing the nature of aggregation species.

Recently, FRET has been used to identify the binding of Aβ oligomers to the cholinergic nicotinic ACh receptor α7 (α7nAChR) (Cecon et al., Reference Cecon, Dam, Luka, Gautier, Chollet, Delagrange, Danober and Jockers2019). Loss of activity of α7nAChR is believed to impair cognitive performance in AD (Bertrand et al., Reference Bertrand, Lee, Flood, Marger and Donnelly-Roberts2015). The result showed that the Aβ binding site on α7nAChR overlaps with the orthosteric ligand binding sites comprising loop C with a high-binding affinity (pK _D 10.3 ± 0.2). FRET was also used to evaluate different α7nAChR-related compounds of interfering with the binding abilities of Aβ to α7nAChR (Cecon et al., Reference Cecon, Dam, Luka, Gautier, Chollet, Delagrange, Danober and Jockers2019). In addition, FRET was applied for the screening of various compounds against Zn²⁺–Aβ interaction (Lee et al., Reference Lee, Lee, Kang, Yang, Kim, Ghisaidoobe, Kang, Lee, Lim and Chung2019). Eight natural inhibitors (out of 145 compounds) were demonstrated to have efficient inhibition effects (>80%) and six of them are capable of decreasing the toxicity induced by Zn²⁺–Aβ on living cells. The two studies indicate that FRET is suitable for measuring the binding activity of peptides/metal ions to amyloid proteins and very effective to assess potential drug candidates against neurodegenerative disorders. Zadorsky's group used FRET to demonstrate that the newly found amyloid protein Toh1, which is localized in the yeast cell wall, could interact with other PrPs Rnq1 and Sup35. The result highlighted that the interactions among different amyloid proteins not only occur between molecules with a similar primary structure but also between proteins that have a close secondary structure and conformational folds (Sergeeva et al., Reference Sergeeva, Sopova, Belashova, Siniukova, Chirinskaite, Galkin and Zadorsky2019).

Besides investigating the interactions of two molecules, FRET is also suitable to explore the interaction of membranes and amyloid proteins. Reynolds et al. found that α-syn aggregation appears to extract lipids from the membranes making them thinner. These extracted lipids could be the initial step to a more intense disruption of membrane integrity. Larger damage to membranes was observed at higher α-syn concentrations, demonstrating the concentration dependency of α-syn aggregates on membrane toxicity (Reynolds et al., Reference Reynolds, Soragni, Rabe, Verdes, Liverani, Handschin, Riek and Seeger2011). Terzi et al. (Reference Terzi, Hölzemann and Seelig1995) reported that Aβ preferentially interacts with anionic membranes and Wong et al. found more net negative charge increases the binding affinity of low-molecular-mass species of Aβ40 (<10-mer) with membranes. The very weak interaction of Aβ40 with the pure dipalmitoylphosphatidylcholine neutral liposomes confirms that Aβ40 membrane insertion is highly dependent on the negative charge of the membrane (Wong et al., Reference Wong, Schauerte, Wisser, Ding, Lee, Steel and Gafni2009). The tight binding of Aβ40 to anionic membranes was further confirmed by TIRF imaging but with a much lower dissociation constant of the peptide from the membrane (<470 pM) (Ding et al., Reference Ding, Schauerte, Steel and Gafni2012), compared to Wong's report (38.8 ± 7.0 μM). The main reason is the very low peptide concentration in the latter case (sub-nanomolar), which leads to a very low peptide/lipid ratio. This result indicates that the ratio of peptide/lipid concentration should be considered when calculating the binding affinities of amyloid proteins with lipid membranes.

In addition to the interaction of membranes with oligomers, the fluorescent lipid membranes were found to directly bind to fibrils and accelerate fibril formation (Gorbenko et al., Reference Gorbenko, Trusova, Girych, Adachi, Mizuguchi, Akaji and Saito2015; Lindberg et al., Reference Lindberg, Wesén, Björkeroth, Rocha and Esbjörner2017). The vesicle lipid–fibril interaction reinforces Aβ42 secondary nucleation at the surface of existing Aβ fibrils (Lindberg et al., Reference Lindberg, Wesén, Björkeroth, Rocha and Esbjörner2017). The charged plain lipid membrane has a similar effect on Aβ fibrillation. Additionally, Widenbrant et al. (Reference Widenbrant, Rajadas, Sutardja and Fuller2006) found the Aβ aggregates bind to the giant vesicle membrane, rapidly cross the vesicle bilayer, then incorporate into the interior small vesicle membrane. Meanwhile, the incorporated membrane was found to possess reduced membrane fluidity, which is indicative of the pathophysiology of disease-affected cells. The reason for membrane penetrations and finally cell death was suggested to be due to induced membrane leakage by the localized aggregates, not due to pore formation mechanisms.

TCCD technique uses two fluorophores that cannot exhibit high FRET and which are illuminated by two lasers compared to one laser for FRET (Orte et al., Reference Orte, Clarke and Klenerman2010). Monomers tagged with a single color are only detected by one of the detecting channels while the oligomers with two distinct fluorophores result in coincident fluorescent bursts after emitting the photons (Fig. 1d, left). The signal of fluorescence intensity derived from the two-color channels presents the numbers of monomers that are labeled in one oligomer (Fig. 1d, right). As different oligomers show different coincident bursts of fluorescence when they diffuse through the confocal volume, TCCD is sensitive and quantitative to provide the information about the concentration and the stoichiometries of associated molecules in a heterogeneous environment. Klenerman's group used TCCD to trace the oligomer ensemble of the SH3 domain from bovine phosphatidylinositol-3-kinase (Orte et al., Reference Orte, Birkett, Clarke, Devlin, Dobson and Klenerman2008). The heterogeneous oligomeric species of SH3 with the most cytotoxic properties have a median size of 38 ± 10 molecules. The size distribution of these oligomers remains constant while their stability is gradually increasing during aggregation. The TCCD method was also applied to determine the interaction of Aβ peptide with neuroserpin, a member of the serine protease inhibitor, which can suppress the toxicity of Aβ peptides (Chiou et al., Reference Chiou, Hägglöf, Orte, Chen, Dunne, Belorgey, Karlsson-LI, Lomas and Klenerman2009). It was found that Aβ increases the rate of neuroserpin's polymerization by forming a weak complex with neuroserpin. As Aβ is later released from the complex and not incorporated into the neuroserpin polymers, it was suggested that Aβ40 acts as a catalyst to enhance polymerization. Iljina et al. (Reference Iljina, Garcia, Dear, Flint, Narayan, Michaels, Dobson, Frenkel, Knowles and Klenerman2016a) quantitatively determined the composition of self- or co-oligomer formation of Aβ40 and Aβ42 under physiological conditions. The main oligomers are Aβ40 at the physiologically relevant ratio of 9:1 of Aβ40–Aβ42. Dimers are the predominant oligomers when the total Aβ concentration is ~1 nM. The oligomerization of α-syn, as well as their co-aggregation with Aβ40, Aβ42 and tau, was also determined by TCCD (Iljina et al., Reference Iljina, Dear, Garcia, De, Tosatto, Flagmeier, Whiten, Michaels, Frenkel, Dobson, Knowles and Klenerman2018). At a very low concentration, close to the physiological conditions, co-oligomerization is more favorable than pure α-syn oligomer formation, but co-oligomers also showed lower lipid membrane disruption than α-syn assembled oligomers.

FRAP is another SMFI technique to measure molecular mobility and turnover processes. The basic principle of FRAP is that a region of interest in a cell or tissue is photo-bleached with a high-intensity laser and subsequently probed by a low-power laser to quantify the recovery of fluorescence intensity over time (Fig. 1e). The recovery curve provides information about the mobile fraction and the DC of the labeled component (Axelrod et al., Reference Axelrod, Koppel, Schlessinger, Elson and Webb1976). Nonetheless, specific interactions, membrane topology and viscosity can also lead to differences in DCs (Reits and Neefjes, Reference Reits and Neefjes2001). Thus, great considerations should be taken when performing FRAP experiments in different cell lines or compartments to compare DCs and mobile fractions.

By counting the number of photobleaching steps, FRAP can directly distinguish the labeled Aβ oligomers with a different number of subunits at a very low concentration (1 nM). Combined with confocal microscopy, Ding et al. differentiated labeled Aβ40 oligomers that were deposited onto a cover glass and found the size varies from dimers to hexamers, but mainly the dimers, trimers and tetramers. This result is in agreement with the peak estimation from gel filtration chromatography (Ding et al., Reference Ding, Wong, Lee, Gafni and Steel2009). Nevertheless, some oligomer species such as pentamers and hexamers are only observed by FRAP but not the gel filtration method, indicating the advantage of SMTs on quantitatively measuring oligomer species. Additionally, the determination of Aβ oligomer species on lipid membranes can also be obtained by forming the lipids on the clean cover glass. It was shown that the sizes of formed oligomers highly depend on the concentration of amyloids on lipid membranes. At close to physiological concentration (2 nM), a large fraction of Aβ species (34%) on supported anionic lipid membranes are trimer/tetramer (Ding et al., Reference Ding, Schauerte, Steel and Gafni2012). Similar smaller pore-forming oligomers were observed directly by high-resolution atomic force microscopy (AFM; Quist et al., Reference Quist, Doudevski, Lin, Azimova, Ng, Frangione, Kagan, Ghiso and Lal2005), indicating these species might be the smallest in the membrane and the source of pre-pore forming oligomers and neurotoxicity. At a relatively high concentration (100 nM), faster oligomer formation and larger size distribution with a very wide range from around pentamers to 60-mer were observed. In this situation, the fast forming oligomers might serve as the nuclei for the direct insertion into a membrane or for the initial formation of fibrils and then lead to membrane permeabilization (Ding et al., Reference Ding, Schauerte, Steel and Gafni2012). Additionally, combined with TIRF, they found the binding to anionic lipid membranes starts with the monomers and the subsequent incubation leads to the formation of oligomers in the membranes. To get a more detailed picture of the toxic mechanism at the SM level, Schauerte et al. (Reference Schauerte, Wong, Wisser, Ding, Steel and Gafni2010) combined membrane conductivity measurements with FRAP to identify the structure of amyloid proteins when causing membrane-permeabilization. The ability of the oligomers to permeabilize the membrane was determined by simultaneously measuring the conductance as well as their size distributions. At a low concentration below 10 nM, labeled Aβ monomers uniformly bind to the membrane while they do not induce any detectable conductance. At intermediate concentrations (from 10 to 100 nM), small oligomeric Aβ species (mainly dimers to 14-mer) are identified and the smallest size of oligomers to show a detectable conductivity are hexamers but not dimers. When the concentration is >100 nM, large Aβ structures and large-conductance are detected, indicating the serious interruption of these aggregates on membranes. This observation supports the pore-formation model where monomers interact with the membrane surface followed by surface diffusion and then assemble into pore structures. By analyzing DC curves, the change of membrane fluidity, which is important for biological functions in cell membranes (Vereb et al., Reference Vereb, Szöllosi, Matkó, Nagy, Farkas, Vigh, Mátyus, Waldmann and Damjanovich2003; Marguet et al., Reference Marguet, Lenne, Rigneault and He2006), can also be determined when membranes interact with the amyloid peptides (Sasahara et al., Reference Sasahara, Morigaki, Okazaki and Hamada2012, Reference Sasahara, Morigaki and Shinya2013; Schreiber et al., Reference Schreiber, Fischer and Lang2012). hIAPP shows a more dramatic change of average DC signal before and after adding it into the lipid bilayer (from 1.07 ± 0.2 to 0.004 ± 0.004 μm² s⁻¹), compared to Aβ42 peptide (from 0.99 ± 0.1 to 0.07 ± 0.01 μm² s⁻¹) (Sasahara et al., Reference Sasahara, Morigaki, Okazaki and Hamada2012; Reference Sasahara, Morigaki and Shinya2013). The deposition of hIAPP on the surfaces of bilayers results in the insertion to the bilayers (Sasahara et al., Reference Sasahara, Morigaki and Shinya2014). The membrane mobility decreases considerably when Aβ42 concentration increases from 10 to 20 μM (Meker et al., Reference Meker, Chin, Sut and Cho2018). These results address the involvement of lipids in amyloid aggregation when proteins penetrate the fluid membranes. FRAP has proved also useful for assessing molecular mobility in living cells. α-syn aggregates in the SH-SY5Y human neuroblastoma cells showed a very low fluorescence recovery signal compared with the area without aggregates, which showed complete signal recovery (Roberti et al., Reference Roberti, Jovin and Jares-Erijman2011). The diffusion constant of α-syn is relatively small (0.03–0.04 μm² s⁻¹) in living cells, suggesting that the α-syn aggregates remain predominantly immobile. These results demonstrate that FRAP is powerful to investigate the effects of amyloid protein on the membrane fluidity and function.

SMF–AFM

AFM is a type of scanning probe microscopy with a nanometer resolution, which mainly relies on Van der Waals interactions and Coulomb repulsions resulting in a weak deformation of the AFM micro-cantilever with respect to the sample surface (Binnig et al., Reference Binnig, Quate and Gerber1986). When the probe approaches and retracts from the substrate in the vertical direction, the changes of cantilever position will be systematically recorded and represented in a force–distance curve (Fig. 2a). These forces reflect the surface topography and structure of the biological sample. AFM enables the visualization of the microscopic surface of heterogeneous amyloid oligomers and fibrils under ambient atmospheric and liquid conditions, thereby provide information about parameters such as height, diameter and morphology (Adamcik et al., Reference Adamcik, Jung, Flakowski, De Los Rios, Dietler and Mezzenga2010; Adamcik and Mezzenga, Reference Adamcik and Mezzenga2012; Ruggeri et al., Reference Ruggeri, Benedetti, Knowles, Lashuel, Sekatskii and Dietler2018). For example, AFM imaging has been utilized to observe different shapes of amyloid aggregates under atmospheric conditions, including pore-like and spherical-shaped Aβ oligomers (Connelly et al., Reference Connelly, Jang, Arce, Capone, Kotler, Ramachandran, Kagan, Nussinov and Lal2012; Nirmalraj et al., Reference Nirmalraj, List, Battacharya, Howe, Xu, Thompson and Mayer2020). It was also found that the height of α-syn oligomers varies from 2 to 12 nm and is dependent on salt concentration (Hong et al., Reference Hong, Han, Fink and Uversky2011), whereas the height of primarily formed Aβ oligomers varies from 2 to 8 nm (Mastrangelo et al., Reference Mastrangelo, Ahmed, Sato, Liu, Wang, Hough and Smith2006). The height of α-syn fibrils in the air is 17% lower than those of the same fibrils imaged in liquids and the C-terminal tails of α-syn fibrils in liquids show different conformation in different ion concentrations (Sweers et al., Reference Sweers, Van Der Werf, Bennink and Subramaniam2012). Additionally, AFM allows to investigate the effect of lipid composition on amyloids binding to membranes. In the fluid phase, positively and neutrally charged lipid bilayers have stronger interference with Aβ42, compared to the negative and zwitterionic ones (Hane et al., Reference Hane, Drolle, Gaikwad, Faught and Leonenko2011). Lipid rafts cholesterol and ganglioside GM1 promote the harmful membrane interactions with Aβ (Azouz et al., Reference Azouz, Cullin, Lecomte and Lafleur2019). All these results demonstrate that conventional AFM has an irreplaceable advantage in providing morphological information of amyloids. However, traditional AFM imaging cannot show how two molecules interacting with each other affect the aggregation morphology. By tethering individual amyloid peptides/proteins to the substrate and probe, respectively, SM force AFM (SMF–AFM measures the internal mechanical features of small heterogeneous amyloid oligomers (dimers to tetramers). SMF–AFM mechanically unfolds the native molecule at a SM level with pN resolution (Rief et al., Reference Rief, Oesterhelt, Heymann and Gaub1997; Li et al., Reference Li, Zhang, Zhang, Shen, Liu, Gao and Zou1998; Hugel et al., Reference Hugel, Grosholz, Clausen-Schaumann, Pfau, Gaub and Seitz2001). The obtained force curves can be used to characterize the interaction forces between them or to reveal the unfolding pathway of oligomers (Yu et al., Reference Yu, Malkova and Lyubchenko2008, Reference Yu, Dee and Woodside2013; Trexler and Rhoades, Reference Trexler and Rhoades2010; Kim et al., Reference Kim, Palermo, Lovas, Zaikova, Keana and Lyubchenko2011). Lyubchenko's group has covalently immobilized the C-terminal of a series of cysteine-modified amyloid proteins on both probe and substrate surface (Krasnoslobodtsev et al., Reference Krasnoslobodtsev, Volkov, Asiago, Hindupur, Rochet and Lyubchenko2013; Maity and Lyubchenko, Reference Maity and Lyubchenko2020). The pulling strength/rupture forces of monomers between the probe and substrate can provide information about the spontaneous dissociation and the resulting off-rate constant, indicative of the degree of intermolecular interactions and the self-assembly of an amyloid protein dimer (Kransnoslobodtsev et al., Reference Kransnoslobodtsev, Shlyakhtenko, Ukraintsev, Zaikova, Keana and Lyubchenko2005). It has been reported that α-syn dimer has two different dissociation rates (0.25 and 13.2 s⁻¹), implying there might be two distinct pathways of forming α-syn dimers (Yu and Lyubchenko, Reference Yu and Lyubchenko2009). For Aβ40 dimers, four close rupture forces (~120 pN) with different rupture positions from 25 to 45 nm were observed (Kim et al., Reference Kim, Palermo, Lovas, Zaikova, Keana and Lyubchenko2011). It suggested the formation of different transient misfolded states. Aβ42 and Aβ40 display similar rupture forces and rupture position when probed by SMF–AFM (Kim and Lyubchenko, Reference Kim and Lyubchenko2014). The approach further reveals a set of transients of Aβ dimers with different conformations, representing diverse misfolding pathways of Aβ dimers. In addition, the dimer lifetime was quantified by combining dynamics force spectroscopy. Compared to dynamic monomeric states, the dimer lifespan of α-syn and Aβ is unexpectedly high within the range of seconds (Yu et al., Reference Yu, Malkova and Lyubchenko2008; Kim et al., Reference Kim, Palermo, Lovas, Zaikova, Keana and Lyubchenko2011; Kim and Lyubchenko, Reference Kim and Lyubchenko2014), indicating that the stable dimers may act as nuclei in the aggregation process. Also, the amyloid dimer lifetime value is pH-dependent. α-syn has two different lifetimes at acid pH conditions (4 and 0.8 s at pH 2.7, 1.35 and 0.05 s at pH 3.7) (Yu et al., Reference Yu, Malkova and Lyubchenko2008), demonstrating two different pH-dependent α-syn aggregation processes. As for Aβ40 dimer, the lifetime is 0.46 s at pH 7.0 and 2.5 s at pH 2.7, respectively (Kim et al., Reference Kim, Palermo, Lovas, Zaikova, Keana and Lyubchenko2011). Aβ42 has slightly different lifetimes (0.1 s at pH 7.0 and 2.7 s at pH 2.7) (Kim and Lyubchenko, Reference Kim and Lyubchenko2014). The results suggest that experimental conditions play a vital role in the aggregation process. As protein interactions highly depend on their conformational changes, SMF–AFM is suitable for directly detecting the rupture forces to indicate conformational changes of amyloid proteins. It has been shown that the pulling strength of α-syn and Aβ is the weakest at neutral pH, indicating a random coil structure. At an acidic pH, a higher pulling strength revealed the rapid conformation conversion to a more ordered, predominantly β-sheet structure (McAllister et al., Reference Mcallister, Karymov, Kawano, Lushnikov, Mikheikin, Uversky and Lyubchenko2005). Furthermore, SMF–AFM was applied to study Aβ14–23 trimers, where a larger rupture force is required to unravel trimers, compared to that of dimers (Maity and Lyubchenko, Reference Maity and Lyubchenko2016). This explains the different folding arrangements within trimers and dimers. Yu et al. revealed a dramatic increase in rupture events of α-syn interactions in the presence of Al³⁺ and Zn²⁺, demonstrating the supportive effect of metal ions on α-syn misfolding. In addition, dopamine does not increase the rupture events and forces, which suggests dopamine may not involve in the α-syn dimerization (Yu et al., Reference Yu, Warnke and Lyubchenko2011). Based on these results, SMF–AFM can be used as an efficient tool to screen drugs or biomolecules that interfere with amyloid aggregation in the early stages of the aggregation. Another advantage is that SMF–AFM can be conducted in a broad pH distribution, which is important as it has been shown that acidic pH dramatically facilitates amyloid aggregation and amyloid monomers and oligomers in vivo normally occur in an acidic environment (Pasternak et al., Reference Pasternak, Callahan and Mahuran2004).

Fig. 2.

Schematic examples of the basic principles of SMF techniques. (a) One single amyloid protein is attached to the functionalized AFM tip and a mica surface separately. When the probe approaches and retracts from the substrate in the vertical direction, the force–distance curve that is systematically recorded reveals the protein unfolding pathways. (b) Two single amyloid proteins are tethered to two beads constricted in traps. SMF–OTs exert extremely small forces via a highly focused laser beam that generates a very strong electric field gradient.

As trimers, tetramers and larger oligomers also contribute to the heterogeneity of amyloid oligomers in the aggregation process, flexible nanoarrays (FNAs), an extension of SMF–AFM, were developed to investigate the intermolecular interactions of higher oligomeric structures (Tong et al., Reference Tong, Mikheikin, Krasnoslobodtsev, Lv and Lyubchenko2013; Krasnoslobodtsev et al., Reference Krasnoslobodtsev, Zhang, Viazovkina, Gall, Bertagni and Lyubchenko2015; Maity et al., Reference Maity, Viazovkina, Gall and Lyubchenko2017). In the FNA setup, amyloid proteins (dimer or monomer) are covalently immobilized within a highly flexible, linear polyethylene glycol-type polymer on which the tethered proteins can interact and form the higher complex (Maity et al., Reference Maity, Viazovkina, Gall and Lyubchenko2016). The two functionalized ends of the FNA polymer are linked to the AFM tip and substrate, respectively. As the synthesis process of the polymer can control the incorporation of the monomers, the attached amyloid molecules can be anchored to the desired positions within the polymer (Krasnoslobodtsev et al., Reference Krasnoslobodtsev, Zhang, Viazovkina, Gall, Bertagni and Lyubchenko2015). Then different scenarios can be studied. For example, a dimer construct can interact with another monomer, which is tethered to the flexible polymer, and this creates a trimer model to explore the dimer–monomer intermolecular interaction. Analysis of the dissociation force of Aβ14–23 trimers showed that the first rupture force of the monomer is ~48 ± 2.4 pN, which is slightly lower than the second step of rupturing the dimers (~53 ± 3.2 pN) (Maity et al., Reference Maity, Pramanik and Lyubchenko2018). This demonstrated that the dimers are at a more stable state than the dynamic monomers in the early stage of aggregation. Another example is where a dimer construct forms at the end of the flexible polymer, which allows dimer–dimer interactions to be studied (Maity et al., Reference Maity, Viazovkina, Gall and Lyubchenko2017). The results of the rupture force and lifetime of Aβ14–23 tetramers are close to those of trimers at the neutral pH. However, the acid condition affects oligomer behaviors with different rupture forces and lifetimes. At pH 3.7, the lifetime histogram of the tetramer and trimer has a broad distribution and each fits three Gaussian peaks. Different sets of lifetimes indicate different intermolecular interactions with distinct structures. A similar change due to an altered pH was found for the strengths of the bonds. These results are consistent with the known heterogeneity of amyloid oligomers and suggest they can follow different aggregation pathways that are pH-dependent. Similarly, the intermolecular interaction of higher-order oligomerization such as pentamers and hexamers might be studied. FNA allows dissociating the organized oligomers in a stepwise way and provides a good way to reveal the interaction of different states of oligomers, which can contribute greatly to the understanding of the mechanisms involved in amyloid aggregation.

SMF–OTs

Compared to SMF–AFM that probes biomolecules directly, SMF–OTs indirectly manipulate and trap sub-nanometer to micron-sized dielectric particles by generating a very strong electric field gradient via a highly focused laser beam that exerts extremely small forces. SMF–OTs use a high numerical aperture (>1) microscope objective to create a stable optical trap to stabilize the vicinal dielectric microparticles. The working principle is illustrated in Fig. 2b. A single molecule, such as a DNA strand, a peptide or protein, is attached to a microparticle (Jagannathan and Marqusee, Reference Jagannathan and Marqusee2013). SMF–OTs exert a force of up to 100 pN on particles with a 10⁻⁸ N force accuracy, a 10⁻⁴ s temporal resolution and 10⁻⁹ m spatial resolution. By adjusting the position of the traps, the force on the molecule can be indirectly obtained through the real-time measurement of three-dimensional displacement of the particles. In Woodside's group, the Syrian hamster PrP is linked to DNA handles and connected to the polystyrene beads (Yu et al., Reference Yu, Liu, Neupane, Gupta, Brigley, Solanki, Sosova and Woodside2012). A single PrP molecule is trapped and then the end-to-end extension is traced by exerting the denaturing tension on the PrP. The traps are taken apart at a constant rate to ramp up the force until the monomer PrP is totally stretched out. Under force-extension, PrP shows conformational fluctuations in the folding process with high precision. No partially folded intermediate is found to regulate the misfolding event during the native folding process. It suggests that the native state or the on-pathway intermediate is not an important factor for PrP misfolding. Instead, the unfolded states are found to involve three different misfolding off-pathways, which indicate these unfolded states serve as intermediates in the misfolding process. SMF–OTs allow the observation of the transient transitions from the unabiding structural fluctuations of the SM folding trajectory, which makes SMF–OTs a powerful tool to explore the different aspects of amyloid protein misfolding mechanisms.

Based on the above SMF–OTs setup, Neupane et al. (Reference Neupane, Solanki, Sosova, Belov and Woodside2014) found different intermediates with various sizes and stabilities during α-syn aggregation. They reported that many metastable α-syn structures are formed from small oligomers with a wide range of sizes. By monitoring the folding kinetics of a single α-syn monomer, small parts of discrete unfolding events were analyzed by fitting the force-extension curves. The change of contour length in the conformational transitions upon unfolding varies from 10 to 36 nm. These separate unfolding transitions are also observed in α-syn dimers and tetramers, indicating the formation of unfolding intermediates with several distinct structures. Furthermore, the unfolding forces which reflect the structural stability and the height of the energy barrier to unfolding were compared with the monomers, dimers and tetramers. It shows that the unfolding forces increase with the growing size of the oligomer, confirming that larger oligomers of α-syn have higher structural complexity.

Unlike a single amyloid monomer, it is difficult to manipulate one single fibril by an optical-trapping experiment as amyloid fibrils are highly flexible. Dong and coworkers introduced a new method to tether Sup35 PrP fibrils (Dong et al., Reference Dong, Castro, Boyce, Lang and Lindquist2010). They first attached the N-domain of prion monomer variant (NM), which only contains prion N and M regions, onto the glass. Then these tethered monomers exert the self-recognition feature to capture prion fibrils in which the attachment happens at the end of fibrils. The polystyrene bead was used to capture the other end of the fibrils. After applying normal pulling strength, prion fibrils showed resistance to disruption and maintained fibril integrity. When exerting much higher forces of up to 250 pN, some force-dropping events were monitored before the fibril ruptured and it suggests that dropping events result from the prion intrinsic fold properties rather than from the extrinsic surface attachments. SMF–OTs’ results support the β-helix model structure for prion fibrils rather than the β-pleated-sheet model. Based on this method, the mechanical properties of NM from prion fibrils were explored and showed that the persistence length of NM is much shorter than that of insulin and Aβ fibrils (Castro et al., Reference Castro, Dong, Boyce, Lindquist and Lang2011). This result means that NM fibrils are much less resistant to bending than the other amyloids. The formation of point defects (hinges) of the prion fibril structure greatly reduced the bending stiffness. Goldmann previously reported that actin becomes more susceptible to fragmentation when bending decreases (Goldmann, Reference Goldmann2000). Thus, SMF–OTs' results support the hypothesis that prion fibrils are not rigid enough to be fragmented, producing more seeds to function as protein-based elements of inheritance and leading to stronger prion phenotypes in vivo (Castro et al., Reference Castro, Dong, Boyce, Lindquist and Lang2011). Other mechanical parameters such as the axial and torsional stiffness are also deduced by the SMF–OTs method. These results provide meaningful guidance for designing novel amyloid-based nanomaterials including scaffolds and biosensors. Recently, Fränzl et al. (Reference Fränzl, Thalheim, Adler, Huster, Posseckardt, Mertig and Cichos2019) designed a thermophoretic trap to observe the single amyloid fibril growth over rather long periods of time. Analogous to the electric field gradient in the SMF–OTs, the temperature field gradient was created by heating an optically heated, 10-μm-sized chrome ring via a focused laser beam. The trapped single amyloid fibril showed the rare fibril fracture events, confirming the idea of existing secondary processes during the growth of amyloid fibrils. With a similar concept with SMF–OTs, MTs use the gradient of a magnetic field to manipulate paramagnetic beads (Strick et al., Reference Strick, Allemand, Bensimon, Bensimon and Croquette1996; Gosse and Croquette, Reference Gosse and Croquette2002). There are several reviews detailing the principles of MTs and their application at the SM level (Gosse and Croquette, Reference Gosse and Croquette2002; Neuman and Nagy, Reference Neuman and Nagy2008; Sarkar and Rybenkov, Reference Sarkar and Rybenkov2016). Though the technique of MTs is mature to handle, to the best of our knowledge, there is no report where MTs are applied to characterize amyloid proteins.

Biological nanopores

The transportation of ions or molecules into and out of cells depends on channels, which are protein pores in the cell membrane with a size range of 0.3–10 nm. The typical biological nanopore apparatus is composed of two chambers or reservoirs (cis and trans, as illustrated in Fig. 3a) immersed in an electrolyte solution separated by a lipid bilayer film that is formed across an orifice, and a nanopore set in the lipid bilayer. An electric potential applied between the two chambers generates an electric field across the nanopore, which drives charged ions or molecules to pass through the nanopore (Chen and Liu, Reference Chen and Liu2019). For small unfolded molecules, they are easy to translocate the pore, whereas molecules with large sizes, such as aggregates, fibrils or folded proteins, tend to simply clog the pore or bump into the pore and then diffuse away (Madampage et al., Reference Madampage, Tavassoly, Christensen, Kumari and Lee2012). By analyzing the amplitudes, durations and shapes of the events, the size, conformational changes and interactions with other molecules can be determined at a SM level. Recently, single-nanopore technologies have shown great potential for the characterization of individual amyloid proteins (Houghtaling et al., Reference Houghtaling, List and Mayer2018), which will be further discussed.

Fig. 3.

Illustration of single-nanopore measurements. (a) The process of single amyloid protein translocation the biological nanopore system (using α-HL as an example) (Left). (Right) Typical current trace of a single protein across a nanopore. After applying a constant voltage, an open pore current is recorded (①); once a single molecule has been captured and sliding inside the pore, the current value drops to a lower state (②); the current back to the open pore state when the molecule completely translocates the pore (③). (b) Solid-state nanopore setup with fluid access channels to a nanopore across a silicon nitride membrane. A typical current event is recorded. The components of (b) are from Servier Medical Art.

Zhao et al. (Reference Zhao, Jayawardhana, Wang and Guan2009) studied the translocation of Aβ10–20 and various short peptides through an engineered α-HL nanopore. Both the dwell time and the change in the amplitude of current blockades increase as the length of the sequence increases. A stronger binding affinity was observed between peptides and pores when more aromatic binding sites were engineered to be present inside the lumen of the pore. Based on the characteristic currents caused by different lengths and structures of the peptides, biological nanopores make it possible to study structural changes when Aβ peptides interact with small molecules. Wang et al. (Reference Wang, Ying, Li, Kraatz and Long2011) probed the different aggregation transitions of Aβ42 in the presence of an aggregation promoter (β-cyclodextrin (β-CD)) and an inhibitor (Congo red (CR)) using α-HL nanopores to differentiate between fibrils and oligomers. Two populations were observed of Aβ42-CR, namely, one current amplitude is the same as the events caused by Aβ42 alone, while the other current is slightly higher, indicating that after the incubation with the CR, a new structure emerged in the Aβ42 protein. For Aβ42-β-CD, barely any events were observed, and the current distribution is in a wide range, suggesting that the detected protein structure is bigger than the α-HL and thus cannot be captured effectively. Therefore, only bumping events were recorded as would be expected from large aggregates. Madampage et al. (Reference Madampage, Tavassoly, Christensen, Kumari and Lee2012) also obtained similar results when Aβ42 is aggregated and they investigated the effect of mutation on the conformational changes and aggregation propensities. Aβ40 translocation experiments showed recognizable events corresponding to α-helical structures whereas Aβ40 mutant protein, D23N, has typical folded β-sheet signals. Two mutants (A30P and A53T) but not the E46K mutant of α-syn caused more bumping events than wild-type α-syn, suggesting the two former mutations increase the propensity for misfolding and aggregation whereas the latter does not. Hu et al. (Reference Hu, Ying, Gu, Cao, Yan, Wang and Long2016) characterized the preliminary structural aspects during the aggregation process of Aβ25–35 and its reversed sequence Aβ35–25 in a time-dependent way. Their results showed that Aβ35–25 conforms to a random coil structure while Aβ25–35 comprises both random coil and β-sheet structures, which is consistent with the CD spectra measurements. As the nanopore-based detection is performed in aqueous solution, it is well-suited for studying the kinetic interactions between metals and Aβ proteins. Asandei et al. (Reference Asandei, Schiopu, Iftemi, Mereuta and Luchian2013) compared the Cu²⁺ binding affinity of Aβ16 derived from human and rat tissues. Compared to peptides from humans, Cu²⁺ binding of Aβ16 peptides from rats showed higher association and dissociation kinetics of the peptides reversible interaction with the pore (Asandei et al., Reference Asandei, Iftemi, Mereuta, Schiopu and Luchian2014). In addition, based on single-nanopore experiments the binding affinity of metals to Aβ16 obeyed the following order: Cu²⁺ > Zn²⁺ > Fe³⁺ > Al³⁺.

Gurnev et al. (Reference Gurnev, Yap, Pfefferkorn, Rostovtseva, Berezhkovskii, Lee, Parsegian and Bezrukov2014) utilized α-HL nanopore to explore how α-syn interacts with the lipid membranes at a SM level. It showed the N-terminus of α-syn first binds to the lipid membrane, then the C-terminus enters and finally passes through the nanopore. Additionally, their study of the interaction affinity between α-syn and voltage-dependent anion channels (VDACs) revealed a previously unknown mechanism of mitochondrial outer membrane permeability regulation induced by α-syn (Rostovtseva et al., Reference Rostovtseva, Gurnev, Protchenko, Hoogerheide, Yap, Philpott, Lee and Bezrukov2015). α-syn is able to block VDACs in a concentration and voltage-dependent manner and cross the mitochondrial outer membrane into the intermembrane space when the voltage is above 43 mV. Biological nanopore technologies have been demonstrated to hold the potential for applications in the design of drug screening assays and gaining a deeper understanding of the aggregation mechanisms of amyloid proteins when investigating their interactions with lipid membranes as well as biomolecules such as chaperones and metal ions at the early stage of aggregation.