Amyloid oligomers: spectroscopic characterization of amyloidogenic protein states


M. Lindgren, Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway
Fax: +47 73597710
Tel: +47 73593414


It is assumed that protein fibrils manifested in amyloidosis result from an aggregation reaction involving small misfolded protein sequences being in an ‘oligomeric’ or ‘prefibrillar’ state. This review covers recent optical spectroscopic studies of amyloid protein misfolding, oligomerization and amyloid fibril growth. Although amyloid fibrils have been studied using established protein-characterization techniques throughout the years, their oligomeric precursor states require sensitive detection in real-time. Here, fluorescent staining is commonly performed using thioflavin T and other small fluorescent molecules such as 4-(dicyanovinyl)- julolidine and 1-amino-8-naphtalene sulphonate that have high affinity to hydrophobic patches. Thus, populated oligomeric intermediates and related ‘prefibrillar structures’ have been reported for several human amyloidogenic systems, including amyloid-beta protein, prion protein, transthyretin, α-synuclein, apolipoprotein C-II and insulin. To obtain information on the progression of the intermediate states, these were monitored in terms of fluorescence parameters, such as anisotropy, and quantum efficiency changes upon protein binding. Recently, new antibody stains have allowed precise monitoring of the oligomer size and distributions using multicolor labelling and single molecule detection. Moreover, a pentameric thiophene derivative (p-FTAA) was reported to indicate early precursors during A-beta1-40 fibrillation, and was also demonstrated in real-time visualization of cerebral protein aggregates in transgenic AD mouse models by multiphoton microscopy. Conclusively, molecular probes and optical spectroscopy are now entering a phase enabling the in vivo interrogation of the role of oligomers in amyloidosis. Such techniques used in parallel with in vitro experiments, of increasing detail, will probably couple structure to pathogenesis in the near future.


Alzheimer’s disease


1-amino-8-naphthalene sulfonate




oligomeric LCPs


luminescent-conjugated polymers


pentameric thiophene derivative


thioflavin T




Amyloidosis manifests itself through the extracellular deposition of insoluble protein fibrils, leading to tissue damage and disease. The fibrils form when normally soluble proteins and peptides misfold and self-associate in an abnormal manner [1]. The mechanisms behind the self-assembly of naturally occurring proteins into amyloid deposits remain a mystery, even though it is well known that the final fibrillar structures have a number of structural properties in common, such as the pronounced β-sheet secondary structure and unbranched fibril morphology. Amyloids are associated with serious diseases, including systemic amyloidosis, Alzheimer’s disease (AD), maturity-onset diabetes and the prion-related transmissible spongiform encephalopathies [2,3]. Moreover, the common fibrillar structures also have the ability to bind small molecules such as the widely used amyloid stains thioflavin T (ThT) [4] and Congo red [5], with concomitant alterations of their optical properties in terms of, for example, fluorescence quantum efficiency and their influence of polarized light rendering their appearance birefringent. It is usually anticipated that a multi-order reaction, such as the formation of amyloid fibrils, involves thousands of monomeric protein molecules and that it proceeds through the formation of intermediate smaller structures, at least transiently. There is compelling evidence that the principal component common to many of the related diseases, in addition to the insoluble fibrillar deposits, are in the form of populated prefibrillar small soluble aggregates [6], these days referred to as ‘oligomers’ or ‘protofibrils’. The properties and structures of such oligomeric aggregates are of immense interest to understand amyloid disease and amyloid formation. Furthermore, studies in cell culture have revealed oligomers to show superior toxicity in relation to mature amyloid fibrils. For example, oligomers of amyloid-β are known to impair synaptic plasticity and to be toxic both in vivo and in vitro [7,8]. Similar recent results on transthyretin (TTR) [9] and insulin [10], have corroborated this to be a common property of amyloidogenic oligomers, at least in cell culture. Studies of the above-mentioned proteins and of other proteins linked to human amyloidoses, including lysozyme [11] and prion protein [12], have shown that oligomers transiently populate during fibrillation and, under the correct circumstances, can be exclusively populated without further conversion into amyloid fibrils. Data from several groups have demonstrated that the conformational conversion from a native protein towards a fibrillar state is a phenomenon common to many proteins and peptides, even for proteins of various organisms that are not involved in amyloid disease states, for example the N-terminal domain of HypF from Escherichia coli [13] and human tissue factor [14]. Findings such as these have led to the hypothesis that amyloid fibril formation and amyloidogenic oligomerization is a generic property of the polypeptide backbone [15] and that this is not limited to specific sequences. Hence, it is perplexing why only certain proteins cause amyloid diseases. Nevertheless, despite the sequence independence of misfolding and misassembly in vitro it is evident that sequence does play a major role in dictating amyloid formation, from the perspective both of protein stability [16] and of sequence specificity for modelling aggregation propensity [17]. In conclusion, both sequence-dependent and sequence-independent (i.e. conformation-dependent) mechanisms appear to control the assembly and toxicity of amyloid oligomers.

To clarify the mechanism of fibril formation and the relation of the amyloidogenesis processes to numerous amyloid diseases are challenging tasks. Many studies have addressed these aspects over the past two decades and they have become a major research field. Early studies utilizing turbidity and sedimentation combined with ThT binding provided data on the appearance of high-molecular-weight aggregates [18] or on the disappearance of soluble low-molecular-weight peptides [19]. Using combinations of size-exclusion chromatography, electron microscopy and quasielastic light-scattering spectroscopy, it was possible to distinguish intermediate structures in terms of dimers and protofibrils of typical sizes of a few nm in diameter up to 100–200 nm, respectively [20]. Real-time monitoring of fibril growth is essential, but is also very technically demanding. This article reviews a selection of spectroscopic techniques, predominantly fluorescence techniques, recently used and developed to follow protein misfolding, oligomerization and amyloid fibril growth in real time. We cannot, for the sake of space, cover this entire field, so will merely provide an overview of the exciting progress made.

Structural methods for misfolded proteins and amyloids

The desire to elucidate the structure and properties of proteins has a long history, and techniques for protein-structure characterization cover a very wide range of techniques. As proteins in native and other thermodynamically stable states can be formed in relatively large quantities, and even in crystalline form, NMR and X-ray diffraction are traditionally used to provide detailed atomic resolution 3D information of their individual molecular conformations. However, the studies of protein aggregates and amyloids require the development of other techniques because the final products, as well as intermediate structures, will occur in small quantities and with structural diversity of both conformation and size, and these systems are also present at conditions far from equilibrium, making it difficult to produce reproducible and accurate assessments of structure. The situation is, in some respects, similar to the situation of studying protein folding and dynamic protein–protein interactions, such as in the case of chaperone action. Restricting us to amyoidic structures, techniques such as NMR [21], EPR [22], electron microscopy and atomic force microscopy [23] and X-ray diffraction [24], have been used with some success, as have optical techniques such as CD and light scattering, as mentioned in the Introduction. FTIR spectroscopy revealed that certain vibration-frequency bands can be used to follow the aggregation and to explore environmental properties, such as pH, solvent and temperature effects on the structure [25]. Bovine insulin was recently found to form different aggregated structures controlled by reducing agents; investigations with FTIR showed that one type of filament, consisting of antiparallel beta-sheets, was found to be nontoxic in cell cultures, whereas parallel β-sheet-structured fibrils were found to be toxic with a remarkably lower ThT fluorescence for the filaments [10]. In contrast to methods for structural investigations of mature fibrillar structures, the best-suited biophysical technique for obtaining high-resolution structural information on defined oligomeric species populated during fibrillation is small-angle X-ray scattering. This was recently demonstrated for insulin by Vestergaard et al. [26], where a conformationally defined oligomer was shown to be a building block for mature fibrils. However, there are some practical constraints for small-angle X-ray scattering, namely (a) exclusive equipment is necessary, (b) it requires a high concentration of approximately 0.1 mm of protein, (c) it requires a population of a few species (preferably one to three) at the same time, (d) the sample must be stable for typically several minutes for recording the necessary amount of raw data and (e) it involves a tedious data-analysis process.

Fluorescence methods for capturing the intermediate oligomeric state

To obtain information more rapidly, and to be able to use low protein concentrations, fluorescence techniques, in combination with labeling, offer high sensitivity and measurement of a diversity of structural aspects, which are dependent on the fluorescent probe used. In Fig. 1 the most common techniques are depicted graphically in a cartoon, which are discussed in more detail below. ThT or thioflavin S, and Congo red, are commonly used to detect amyloid deposits in biopsies or in ex vivo postmortem samples [27], as depicted in Fig. 1 (top row). Further modified versions of these dyes have also been used as in vivo imaging agents of amyloid deposits [28]. Derivatives of thioflavins [29], Congo red derivatives [30] and oxazine-derivatives [28] typically bind amyloid fibrils in the nanomolar to micromolar range with multiple binding sites. For more details and extensive references on molecular ligands towards amyloid fibrils the reader is referred to recent reviews on the topic [31,32]. In this article the topic will be restricted to recent work using advanced fluorescence techniques and probes that have been used to detect oligomeric amyloid protein states.

Figure 1.

 Cartoon giving an overview of the fluorescence techniques used to follow and quantify the amyloid fibrillation process.

Small molecular dyes, such as [4-(dicyanovinyl)-julolidine] (DCVJ), as well as derivatives of 1-amino-8-naphthalene sulfonate (ANS, Bis-ANS), have been used for amyloid fibril detection, and are known to bind to the fibrillar or prefibrillar states with dissociation constants typically in the micromolar range. We have successfully used these molecules for recording the kinetics of oligomerization of A-state (molten globule type) TTR, under low-pH conditions [33]. DCVJ and ANS were compared with ThT as probes for the oligomerization of TTR. DCVJ and ANS showed good binding and fluorescence at acidic pH, whereas ThT did not bind oligomers at this pH, requiring a pH shift in the assay buffer. DCVJ proved efficient in the kinetic assay and showed a reasonable binding affinity to preformed early state oligomers of TTR (Fig. 2). DCVJ did not bind to native tetrameric TTR, rendering this an interesting molecule for following the pathway of tetramer dissociation, which was not performed in the cited work. ANS, by contrast, was also found to be a good probe for the kinetic assay and also had a reasonable fluorescence lifetime (16 ns), but ANS binds to the thyroxine-binding pocket of native TTR, which would make measurements of dissociation kinetics impractical. That both DCVJ and ANS responded more efficiently than ThT in both the kinetic and to the isolated TTR oligomers indicates that ThT is more selective towards fibrils than towards oligomers. Nevertheless, ThT can respond also to oligomers, and ThT was recently used to follow the in vitro formation of highly toxic soluble amyloid-β oligomers, which was assisted by the chaperon prefoldin [34].

Figure 2.

 (A) The chemical scheme for TTR amyloid fibrillation from the unfolded monomeric state in low salt and acidic pH. (B) Kinetics, as detected by normalized fluorescence, comparing the fluorescent probes DCVJ, ANS and ThT for detection of early oligomeric species during the fibril formation. From the amplitudes it is evident that ThT also showed a very slow additional phase that is not fully included in the timescale depicted above. (C) The apparent dissociation constants differ equally to their response time towards early oligomers. The figure panels are redrawn from a previous publication [33].

Fluorescence anisotropy is a well-known technique used to obtain information of molecular and protein sizes from the apparent rotation diffusion of fluorescent probes, usually attached to mutated proteins (Fig. 1, second row). The technique has been widely used in studies analysing the changes from unfolded to folded protein structures of sizes typically up to 30–50 kD. It was adopted for use in studies of the aggregation of amyloid-β using a fluorescein marker attached to a cysteine introduced at position 7 [35]. Typical time-resolved anisotropy decays, assayed over time, displayed a trace with fast initial and slow ‘floor’ raising as time progressed during incubation after the initiation of oligomerization (i.e. the contribution of the ‘residual anisotropy’ to the overall anisotropy increased). Similarly, small fluorescent probes, such as DCVJ, ANS and bis-ANS, were tested and compared with ThT for studies of TTR aggregation [33], revealing a clear correlation between the progress of fibril formation and an increase of the slow anisotropic component as a result of the probe binding to increasingly larger fibrillar structures. Recently, it was also demonstrated that from the steady-state polarization the oligomerization process of α-synuclein could be followed, which preceded ThT fluorescence kinetics of the same process [36]. The time-resolved anisotropy studies also demonstrated [33] the shortcoming of the anisotropy technique: it is physically impossible to obtain information of rotational correlation times somewhat longer than (typically three to five times) the decay time of the fluorescent probes used. As fluorescein and ANS have decay times that are shorter than 5 and 20 ns, respectively, only the latter could actually give an accurate determination of the size of the native protein complex if it was larger than approximately 50 kDa, but other information from the changes of the anisotropy can be used quantitatively to follow the initial process. In order to improve the accuracy for larger oligomeric complexes, a single cysteine mutant version of TTR (C10A/A37C) was modified with the long-lived pyrene-methyl iodo-acetamide fluorophore. The decay time of the pyrene-methyl iodo-acetamide fluorophore is up to 150–200 ns, depending on the solvent, and thus permits the determination of rotational correlation times well above 600 ns. This allowed us to determine the kinetics of oligomerization of A-state TTR and was used to determine the size of a toxic oligomer determined to be composed of 20–30 monomer units [9]. Pyrene labelling has also been used for structural packing studies during protein aggregation and amyloid fibrillation. The unusually long lifetime of pyrene can render the formation of an excited state dimer (excimer), given that two pyrene moieties are in proximity. For protein-aggregation studies this approach was first used to map an aggregation interface in the formation of soluble oligomers of carbonic anhydrase [37] and was subsequently used for Sup35 prions [38], glucagon [39] and a-synuclein fibrillation [40]. The major drawback of pyrene attachment has been regarded to be twofold: the necessity of a cysteine-scanning approach; and the intrinsic bulkiness and hydrophobicity of the pyrene label.

Modern fluorescent techniques and novel spectroscopic stains

By using two different labels simultaneously, a multitude of possibilities are available to gain more information on the aggregation processes (Fig. 1, third row). Ryan et al. [41] studied the effect of short-chain phospholipids (DHPC and DHPS) on amyloid fibrillation of human apolipoprotein C-II. Differences in morphology for different fibrillation conditions were found using Alexa488 C5 maleimide and Alexa594 C5 maleimide in conjunction. Fluorescence resonance energy transfer thus gave rise to quenching of the donor probe (Alexa488) during fibril formation. Moreover, the fluorescence anisotropy was found to increase as the fluorescence yield decreased during fibril formation. Previously it had been reported that fibrillation is stimulated by the presence of negatively charged lipid surfaces. Here, it was demonstrated that a similar effect was achieved by interaction with the hydrophobic fatty acyl chain. By using two excitation sources it is possible to detect single and multiple fluorescence events independently, thus discriminating single molecules from small aggregates that are more likely to contain two different fluorophores. Orte et al. studied this phenomenon by using site-specific labelling of two different Alexa dyes (488 and 647) introduced in an experimental amyloidogenic protein composed of the SH3 domain of PI3 kinase [42,43]. By mixing these labelled proteins in a 50 : 50 ratio and following the aggregation process, their data revealed that the formation of oligomeric species proceeded quickly, and then more slowly, and that the oligomers were consumed as longer fibril structures evolved. From the magnitude of the fluorescent signal it was also possible to estimate the size distribution (by comparison with the single molecule cases). The oligomeric aggregate of the SH3-domain was composed of 38 ± 10 monomers. Also, recent work using antibody labels allowed the sensitive detection of amyloid-β oligomers in human cerebrospinal fluid by combining flow cytometry and fluorescence resonance energy transfer [44].

Luminescent-conjugated polymers (LCPs) have been developed over the last few years for use in studies of protein conformations. In contrast to sterically more restricted amyloidotropic dyes, such as ANS, DCVJ, thioflavins and Congo red, LCPs contain a twistable conjugated polymeric backbone whose geometry alters their spectroscopic properties [45–48]. The most widely used LCP in amyloid research has been polythiophenes. Noncovalent binding to proteins, including amyloid, constrains the rotational freedom of LCPs, altering their spectral properties in a conformation-dependent manner. The optical fingerprint obtained from the bound LCP reflects the conformational differences in distinct protein aggregates, as depicted in Fig. 1, lower row. This property has been used to discriminate prion aggregates associated with different prion strains [49,50], conformational heterogeneities in A-beta amyloid plaques in Alzheimer disease mouse models [51] and morphologically different amyloid deposits in systemic AL amyloidosis [52]. LCP luminescent staining technology constitutes an important new tool for the analysis of amyloid and prion complexes when analyzed using modern high-resolution spectral imaging techniques. Importantly, the multiphoton excitation capabilities of a few LCPs provided excellent performance when compared with imaging using conventional ‘single photon’ excitation [53]. The fluorescence quantum yield of LCPs, in the order of 5 – 10%, is lower than that for conventional molecular high-efficiency fluorescence systems based on, for example, fluorescein. However, the nonlinear properties, in terms of two-photon excitation fluorescence efficiency, are comparable owing to a high two-photon absorption cross-section in the range of (4–10) × 1048 cm4·s per photon [53]. This is approximately two orders of magnitude better than fluorescein based on two-photon absorption per molecule. Nevertheless, compensating for the molecular size and normalizing to molecular weight, the number is still larger than or comparable to that of the fluorescein analogs and compensates for the less-efficient quantum efficiency. Interestingly, it was also found that multiphoton excitation schemes in some instances gave complementary spectral signatures in terms of polarization and spectral shifts, in addition to the results obtained using single photon excitation of the very same LCP-stained samples. Multiphoton excitation opens up new possibilities, as excitation in the near infrared wavelength region allows deeper penetration into tissue, and thus better in vivo imaging [54,55]. It has still not been reported that any synthetic molecule can be used to image oligomeric states formed in vivo; however, very recently, oligomeric LCPs (LCO) were reported that can pass through the blood–brain barrier and be used for in vivo studies of plaque distribution in AD mouse models [56]. In vitro studies, which were presented in the same paper using the LCO pentameric thiophene derivative (p-FTAA) (Fig. 3A), showed interesting results with regard to the detection of early prefibrillar oligomeric states that were nonthioflavinophilic (Fig. 3B). It was demonstrated that p-FTAA detects both prefibrillar states as well as mature amyloid fibrils. Further important evidence was the identification of foci within amyloid-β plaques that stained with an anti-oligomer Ig and which were co-stained with p-FTAA, suggesting an irregular distribution of different conformational states (oligomers and fibrils) within the same senile plaque (Fig. 3C) [56].

Figure 3.

 (A) The chemical structure of the LCO p-FTAA. (B) Kinetic traces of A-beta1-40 and A-beta1-42 fibril formation, comparing p-FTAA and ThT fluorescence. p-FTAA responds to early prefibrillar aggregates before ThT. The conversion of A-beta1-42 is faster than for A-beta1-40 and hence ThT and p-FTAA show overlapping kinetic traces (right traces). The asterisks indicate time-points for samples collected for transmission electron microscopy, with the corresponding micrographs above the kinetic traces. (C) Fluorescence micrographs of cerebral amyloid plaques in a transgenic mouse model of AD stained with p-FTAA and conventional antibodies (6E10 anti-A-beta and A11 anti-oligomer) against amyloid components. In the top micrographs, p-FTAA and 6E10 show complete colocalization, and in the bottom micrographs there is partial overlap, indicating that oligomer distribution is localized to certain foci within the plaque. Figure panels are redrawn from a previous publication [56].

There are still several aspects of LCP-like molecules that are well worth exploring. Can the spectroscopic properties of an LCP bound to a prefibrillar oligomer be used as a marker for an on- or off-pathway species? If the spectroscopic property (i.e. conformation) of the bound LCP mimics that of the mature fibril, is it likely that the building block is conformationally related (i.e. a building block for the mature fibril). If, however, the amyloid oligomer requires a substantial conformational change before the formation of an amyloid fibril, the spectral property will be different. Markers of this type will certainly be imperative for the basic understanding of pathways (intermediates) of amyloid fibril formation. These are also crucial for imaging purposes, to irrefutably ascribe endogenous oligomeric states a pathological role also in vivo.


The work was supported by the Swedish Foundation for Strategic Research (to PH), the Swedish research council (to PH) and the Knut and Alice Wallenberg foundation (to PH). P.H. is a Swedish Royal Academy of Science Research Fellow sponsored by a grant from the Knut and Alice Wallenberg Foundation. Part of this work was supported by the European Union FP7 HEALTH (Project LUPAS). We thank Peter Nilsson for valuable comments on the manuscript.