Single-molecule approaches to prion protein misfolding

Abstract

The structural conversion of the prion protein PrP into a transmissible, misfolded form is the central element of prion disease, yet there is little consensus as to how it occurs. Key aspects of conversion into the diseased state remain unsettled, from details about the earliest stages of misfolding such as the involvement of partially- or fully-unfolded intermediates to the structure of the infectious state. Part of the difficulty in understanding the structural conversion arises from the complexity of the underlying energy landscapes. Single molecule methods provide a powerful tool for probing complex folding pathways as in prion misfolding, because they allow rare and transient events to be observed directly. We discuss recent work applying single-molecule probes to study misfolding in prion proteins, and what it has revealed about the folding dynamics of PrP that may underlie its unique behavior. We also discuss single-molecule studies probing the interactions that stabilize non-native structures within aggregates, pointing the way to future work that may help identify the microscopic events triggering pathogenic conversion. Although single-molecule approaches to misfolding are relatively young, they have a promising future in prion science.

Keywords: :

• single-molecule
• fluorescence spectroscopy
• force spectroscopy
• optical tweezers
• atomic force microscopy
• aggregation
• oligomerization
• protein misfolding

This article refers to:

Accessing Early Events in PrP Aggregation

It is widely accepted that the scrapie isoform of the prion protein, PrP^Sc, is able to reproduce itself by templated conversion of the native isoform, PrP^C.^1-3 Beyond this general conceptual framework, however, the microscopic details of prion propagation remain unresolved and controversial. The problem of understanding prion replication at a biophysical level can be reduced crudely to two principal questions: (1) what is the structure of PrP^Sc, and (2) how is PrP^C converted to PrP^Sc? A definitive structure for PrP^Sc has proven elusive, although various models have been proposed based on a range of biophysical and biochemical evidence.^4,5 Efforts to elucidate the structural properties of PrP^Sc have been reviewed elsewhere,⁶ and will not be discussed here. Uncertainty about the structure of PrP^Sc has also made the second question difficult to address. Compounding this difficulty, PrP can aggregate into a wide variety of different forms, including various oligomers as well as amorphous aggregates and amyloids, most of which are not infectious and hence do not correspond to PrP^Sc.³

To simplify the task of addressing the early events in PrP aggregation, we can re-frame the question in terms of what folding pathways are available to PrP, which of these lead to misfolding and aggregation, and how these pathways change in response to stimuli (such as protein-protein or ligand-protein interactions) that are known to affect conversion. Several studies have characterized oligomeric species involved in the early stages of prion protein aggregation at the ensemble level.^7-9 However, since the initial events in conversion are likely rare, involving few molecules, it is challenging to resolve key features of this process using traditional methods, owing to ensemble averaging effects. Furthermore, the conversion from PrP^C into aggregates probably involves many different pathways and numerous rare or transient conformations.¹⁰ Such intermediates may not be detectable under the relevant conditions; conversely, the conditions needed for them to accumulate sufficiently to be detectable (e.g., the use of chemical denaturants to stabilize non-native states) may make the results difficult to interpret. These difficulties can be addressed by investigating the misfolding of PrP at the level of single molecules, applying techniques that have already proven very useful in providing insight into the formation of native structure in protein folding.¹¹

Studying PrP at the Single-Molecule Level

In the single-molecule regime, individual PrP monomers, oligomers or fibrils can be distinguished from one another as structural sub-populations and characterized independently. A variety of probes is available to monitor the structure, stability, and dynamics of the protein, based most commonly on fluorescence or force (Fig. 1; Table 1). For example, Förster resonance energy transfer (FRET) between co-localized fluorophores is a very sensitive probe of protein conformation and dynamics,¹² as shown by an elegant FRET study of the prion-determining NM domain of the yeast prion protein Sup35, which found that Sup35 takes on an ensemble of rapidly-interconverting structures.¹³ To our knowledge, however, single-molecule FRET has not yet been applied to PrP. Fluorescence probes can also monitor the formation and growth of oligomeric aggregates, for example by measuring the size-dependent diffusion of oligomers via fluorescence correlation spectroscopy (FCS) or the intermolecular FRET signals from differently-labeled proteins as they aggregate. FCS has been used to follow oligomerization in conjunction with ensemble methods, revealing a multi-step process starting with β-structured dimers.¹⁴ FCS has also been used to study the interaction of PrP with antibodies, with applications for ultra-sensitive prion detection.^15-17

Figure 1. Illustrations of single-molecule techniques. (A) In FRET, the protein is labeled with two dyes, a donor (circle) and acceptor (star), and the donor is excited. Energy transfer to the acceptor is high when the dyes are close (left), low when they are far apart (right). (B) In FCS, fluctuations in the fluorescence of a labeled molecule are measured as it diffuses through a confocal excitation volume and used to determine the diffusion time. (C) In AFM, the tip is used to pull on a molecule tethered to the surface (here, a monomer is pulled out of a fibril). (D) In optical tweezers, beads trapped by laser beams are used to pull on the ends of a protein molecule tethered between them.

Table 1. Single-molecule techniques used in prion research. The basic principles, measurement methods and sample requirements are outlined for FRET, FCS, AFM and optical tweezers

Name	Principle	Measurement	Requirements
FRET	Molecules are labeled with a pair of fluorophores which can exchange energy resonantly. The strength of the energy exchange depends strongly on the distance between the fluorophores.	The FRET efficiency is measured as the conformation of the molecule changes. FRET efficiency is high when the molecule is folded so that the dyes are close together, whereas it is low for structures in which the dyes are far apart. FRET efficiency can be used as a molecular ruler to determine the distance between the dyes.	Specific labeling of protein with donor and acceptor dyes at different locations.
FCS	Dye-labeled molecules diffusing freely through a confocal excitation volume generate a fluctuating fluorescence signal.	Correlations in the fluctuating signal produced by the diffusing molecules are analyzed to determine the diffusion constant for the molecule, which is related to its size and shape.	Specific labeling of protein with fluorescent dye at one location.
AFM	A molecule is tethered between a surface and a sharp tip on the end of a cantilever. The tip is moved to apply force, which acts as a denaturant changing the molecular structure.	The force on the molecule is measured from the cantilever deflection, while the molecular extension is measured from the tip-surface distance. When the structure changes under tension, there is a sudden change in the extension.	Protein attached specifically or non-specifically to surface and AFM tip.
Optical tweezers	A molecule is tethered between micron-sized beads trapped by tightly-focused laser beams. The laser beams are moved apart to apply force to the molecule, denaturing it.	The extension of the molecule is measured from the distance between the beads, while the force is measured from the displacement of the beads in the traps. Conformational transitions change the molecular extension as for AFM.	Protein attached covalently to DNA handles, in turn attached specifically to beads.

Single-molecule studies of PrP have also made use of force spectroscopy, in which a compliant force probe like an atomic force microscope (AFM) or optical trap is used to grab onto a protein molecule, apply tension and induce structural changes. Force spectroscopy offers some particular advantages for studying misfolding and aggregation. For example, extremely stable structures such as amyloids can be disrupted with relative ease, and the use of force as a denaturant allows the solution conditions to be maintained constant while denaturing and renaturing the protein. This last point is especially important for PrP, the misfolding and aggregation of which are sensitive to conditions such as pH, temperature or chemical denaturant:^10,18-20 Since these conditions are typically used to trigger unfolding/refolding, it is challenging to isolate the effects that relate specifically to misfolding and aggregation. Misfolding can therefore be studied under solution conditions which would not normally be considered conducive to misfolding. Conversely, since the single molecules are measured in isolation, the folding dynamics of individual monomers may be studied under solution conditions in which the protein would normally aggregate rapidly.

The first force spectroscopy study of PrP used AFM to pull apart amyloid fibrils made from recombinant human PrP(90–231).²¹ Fibrils were adsorbed to a mica surface, and the gold-coated AFM tip was bonded covalently to the thiol group of a cysteine engineered near the N terminus of the protein, thereby pulling on one of the PrP monomers comprising the fibril. Increasing the force on this monomer, part of the protein was found to stretch like an unstructured, worm-like chain (WLC) polymer before the monomer was ultimately plucked out of the fibril core at ~100–200 pN of applied force. By measuring how the force for removing the monomer changed with the pulling speed of the AFM,²² the unfolding of the PrP monomer from the amyloid was shown to share characteristics typical of β-structured proteins, as opposed to the helical structure of PrP^C. More interestingly, from the length of the unstructured WLC the authors were able to postulate the location of the edge of the β-sheet core of the amyloid. Their results contradicted the predictions of two structural models of PrP^{Sc 4,5} in which substantial native-like helical structure near the C terminus is retained, but were consistent with models of PrP^{Sc 23} and recombinant amyloid²⁴ in which the native C terminus is completely restructured.

Monomeric PrP Explores Several Folding Pathways

Single-molecule studies of fibrils provide useful clues to the structure of PrP amyloid but do not speak to the question of how the native helical structure is able to convert into a β–rich form. To address the latter question, we recently studied the structural dynamics of isolated PrP monomers with force spectroscopy, in this case using optical tweezers¹² as the force probe. Force was applied to the termini of hamster PrP(90–231) (Fig. 2A), and the end-to-end extension of the protein was measured with Å-scale spatial resolution and sub-ms scale temporal resolution,^25,26 either as a function of the applied force while ramping the force (Fig. 2B) or as a function of time at constant force (Fig. 3).

Figure 2. Single-molecule force spectroscopy of a single PrP molecule. (A) Experimental scheme using optical traps. Cysteine labeled PrP is attached to DNA handles linked in turn to beads held by optical traps. (B) Force-extension curves show that as the force increases, the handles stretch until the PrP structure unfolds suddenly as a two-state system. WLC fits²⁶ (dashed lines) to the folded and unfolded parts of the curves reveal a contour length change matching the result expected for PrP^C.

Figure 3. Constant-force trajectories reveal the presence of misfolded states. When PrP is held under constant tension (gray), the extension fluctuates on the second timescale between levels (dashed lines) corresponding to native and unfolded states. On the ms-timescale (black), frequent downward spikes in the extension correspond to formation of misfolded states (dotted lines).

One focus of our work was characterizing the native folding pathway of PrP. A partially-folded on-pathway intermediate has long been proposed to mediate misfolding,²⁷ and partially-native structures are a feature of several models of PrP^Sc.^4,5 However, the experimental evidence for on-pathway intermediates is conflicting: some studies found only two-state folding, e.g., using stopped-flow chemical denaturant²⁸ or temperature jump²⁹ at pH 7–8, whereas others were consistent with the presence of intermediates at both neutral and acidic pH.³⁰ Using the unique ability of single-molecule methods to observe intermediates directly and determine what pathway they are on,³¹ we found that at neutral pH and without chemical denaturant there was no intermediate detectable on the native folding pathway having a lifetime longer than 50–100 μs, the limit of our resolution.²⁶ This observation does not completely rule out the possibility of a partially-native intermediate, since an intermediate involving restructuring that leaves the end-to-end extension unchanged, such as rearrangement of helix 1,³² would not be detected in our measurement (although it could be detected by changing the experimental geometry to pull on helix 1 instead of the terminus). It does suggest, however, that the intermediates observed previously may be a function of the particular measurement conditions, such as the presence of chemical denaturants, rather than a general feature of PrP folding.

Looking at the native pathway in more detail, we reconstructed the profile of the energy landscape governing the folding using pulling curves.³³ We found a single barrier roughly halfway between the folded and unfolded state, indicating a fairly extended transition state²⁵ (the unstable, high-energy species separating the folded and unfolded states), in contrast to the transition state located close to the folded state that was found when pulling PrP from an amyloid.²¹ From the measured extension of the transition state, we estimated that roughly half of the structured part of native PrP was unfolded in the transition state. Based on this result, in combination with the picture of an expanded transition state consisting of native-like contacts between helices 2 and 3 that emerged from phi-value analysis,²⁹ we speculated that the transition state could consist of a structured hydrophobic core such as helix 2, strand 2, and most of helix 3 (with an uncertain degree of native contacts). The same folding barrier is likely being probed with single-molecule force spectroscopy as with the more traditional ensemble approaches, since the folding rates calculated from first principles applying the kinetic theory of Kramers³⁴ to the reconstructed landscape agreed well with the rates found by other methods.²⁵

The most distinctive finding from force spectroscopy of monomeric PrP was the discovery of at least three different types of partially-folded intermediates off the native folding pathway. All were unstable (6–11 kcal/mol less stable than PrP^{C 26}), but two formed with higher frequency than the native state during refolding (Fig. 3). Interestingly, this misfolding did not require low pH. PrP^Sc is often considered to develop in a low pH environment such as endosomes,¹ since PrP^Sc localizes there³⁵ and low pH encourages misfolding.^19,36 However, PrP^Sc also localizes to the cell surface,³⁵ and it is not evident a priori that the environments in which most PrP^Sc is found in the cell—likely those leading to the fastest growth of PrP^Sc—are necessarily the same as those in which the first PrP^Sc seeds form. Given that PrP explores non-native states at neutral pH, it is possible that seed formation could take place in non-acidic environments such as the endoplasmic reticulum (e.g., after being unfolded during membrane translocation), the cytosol (e.g., after retro-translocation³⁷), or the cell surface.³⁸

The misfolding of monomeric PrP originated solely from the unfolded state,^26,31 suggesting that the role of the unfolded state may be under-appreciated.³⁹ Although the misfolded states were quite unstable in wt-PrP, if they are involved in aggregation then they might be stabilized by aggregation-enhancing mutations. This speculation was supported by measurements of the C179A/C214A mutant, which lacks an internal disulfide bond and is highly aggregation-prone in bulk. The most common misfolded states were significantly enhanced in the mutant, suggesting a role as intermediates on the aggregation pathway. Interestingly, these results also showed that, contrary to ensemble folding studies,⁴⁰ the disulfide bond is not required to attain the native fold: isolated molecules of mutant and wild-type PrP formed the same stable structure, whose length was consistent with PrP^C to within ångtröms, with similar rates and stability.²⁶ Indeed, the disulfide bond was almost certainly not formed even for the wild-type PrP, owing to the non-oxidizing conditions of the measurement.²⁶ The mutant’s aggregation propensity is thus likely driven by the differences in the refolding from the unfolded state.

Equally interesting is what was not observed. First, the amyloidogenic part of the unstructured region⁴¹ did not form even transient structure on its own, indicating that any structure in this region must be induced by conditions not present in the measurements, such as interactions with other molecules. Second, we found no evidence for a stable misfolded form of monomeric PrP, as proposed recently.⁴² Since measurements like those in Figure 3 are in equilibrium, all possible structures are explored during the trajectories, but PrP^C was the only stable, long-lived structure observed.²⁶ Hence any alternate structure must either require the truncated residues 23–90 or else specific experimental conditions not found in the single-molecule measurements.

Studying Aggregation in Oligomers at the Single-Molecule Level

While studies of the dynamics of individual PrP molecules give insight into what properties of the protein may drive its unusual behavior, they don’t directly address structural conversion, which requires focusing on the interactions between monomers that lead to aggregates. One approach is to study minimal oligomers such as dimers and trimers. Several lines of evidence support the notion that PrP dimers or trimers play a role in the conversion of PrP^C to PrP^Sc, even though larger oligomers may be more infectious.⁴³ Recombinant PrP forms dimers at low pH^18,20 and upon dilution from 0.2 to 0.05% SDS;^44,45 brain-derived PrP^C can also dimerize in vitro, at least partially.⁴⁶ Synthetic PrP dimers were shown to be toxic to neurons both in vitro⁴⁷ and in vivo.⁴⁸ The crystal structure of human PrP revealed a domain-swapped dimer⁴⁹ which inspired a subsequent model for oligomerization based on domain swapping.⁵⁰ Trimers, on the other hand, are the building-blocks of both the spiral⁴ and β-helix⁵ model of prion fibrils.

Preliminary measurements of single PrP dimers and trimers that we made using optical tweezers show intriguing results. Oligomers formed by joining monomers together end-to-end via a disulfide bridge were unfolded and refolded similarly to the monomers. Such tandem-repeat oligomers have been measured frequently with force spectroscopy,¹¹ typically showing independent native folding of each monomeric domain, with only infrequent misfolding.⁵¹ In the case of PrP, however, dimer unfolding never showed independently-folded native domains. Instead, a single non-native structure unfolded via multiple intermediates, as revealed by the sequence of length changes during the many “rips” in the pulling curves (Fig. 4A). PrP trimers also unfolded from non-native structures via multiple intermediates (Fig. 4B). The fact that the initial structures were non-native could be determined from the total contour length change (ΔL_c) during the unfolding, which relates to the number of structured amino acids. For dimers, ΔL_c ~81 nm, corresponding to ~240 structured amino acids, whereas for trimers, ΔL_c ~150 nm, corresponding to ~430 structured amino acids. Since hamster PrP^C contains only 104 structured amino acids,⁵² this result implies that the domains interact to form new structures incorporating the formerly unstructured part of the monomer. Whether these structures are the same as those formed by the toxic tandem-repeat dimers studied previously⁴⁸ and how they may be connected to pathogenesis remains to be determined, but single-molecule studies of small oligomers evidently provide rich opportunities for probing structural conversion in PrP.

Figure 4. Force spectroscopy of PrP oligomers. (A) Dimers linked at their termini form non-native structures with multiple intermediates. (B) Trimers form even more complex non-native structures. In each case, WLC fits (dashed lines) indicate non-native contour-length changes.

Future Directions

Single-molecule approaches can be technically challenging and have only recently begun to be applied to prion misfolding, but they provide powerful tools for deciphering the microscopic mechanisms driving the structural conversion of PrP. Several opportunities in particular stand out for future work. First, single-molecule experiments deliver a richly-detailed set of constraints that can be used to build or improve structural models of misfolded PrP as well as to guide models or simulations of the conversion process: distance and energy constraints for the possible structures (including intermediates), the pathways available (native and non-native), transition rates, even information about the barriers. Integration of single-molecule and modeling approaches should therefore prove a very fruitful source of future insight. Second, continued study of small oligomers at the single-molecule level promises to yield detailed information about the pathways for conversion into neurotoxic structures. Single-molecule methods can also play an important role in understanding the structural properties of PrP amyloid, for example by extending the initial force spectroscopy measurements to apply the techniques recently used to map out inter- and intra-molecular interactions in yeast prion amyloid.⁵³ Yet another area of potential application is in determining the mechanism of action of anti-prion agents. Observations of the effects of known anti-prion agents (e.g., tetrapyrroles^54,55 or curcumin⁵⁶) on the folding of single PrP monomers and oligomers may provide critical insights into their still-unknown mechanisms of action, potentially leading to more effective anti-prion therapies. Single-molecule approaches have much to offer prion studies, and we foresee their increasing deployment to help solve the central scientific questions posed by prions.

Abbreviations:
FRET	=	Förster resonance energy transfer
FCS	=	fluorescence correlation spectroscopy
AFM	=	atomic force microscope
WLC	=	worm-like chain

Acknowledgments

This work was supported by PrioNet Canada, the Alberta Prion Research Institute, Canadian Institute for Health Research, and the nanoWorks program of Alberta Innovates.