Disulfide mapping reveals the domain swapping as the crucial process of the structural conversion of prion protein

Abstract

Prion diseases are infectious conformational diseases. Despite the determination of many native prion protein (PrP) structures and in vitro production of infectious prions from recombinant PrP the structural background of PrP conversion remains the largest unsolved problem. The aggregated state of PrP^Sc makes it inaccessible to high resolution techniques, therefore indirect methods have to be used to investigate the conversion process. We engineered disulfide bridges into the structured domain of PrP in order to determine the secondary structure elements that remain conserved upon conversion. Rather surprisingly, introduction of disulfides into each or both of the subdomains B1-H1-B2 and H2-H3 of the C-terminal globular domain retained the robust ability to convert into fibrils with increased content of β-structure, indistinguishable from the wild-type PrP. On the other hand disulfide bridges tethering the two subdomains completely prevented conversion, while their reduction reversed their conversion ability. The same conversion propensity was replicated also in prion infected cell lines. Experiments with combinations of engineered cysteine residues further support that domain swapping, centered on the B2-H2 loop, previously associated to species barrier, leads to PrP swapped dimers as the building block of prion fibrils.

Our understanding of the molecular mechanisms of prion diseases recently significantly advanced with the invention of PMCA technique1 and the demonstration that the converted recombinant PrP can induce transmissible disease,2–5 which conclusively fulfills the Koch's postulates of infectivity. Another important development, relevant to the structural background of conformational diseases was the demonstration of the ability of short peptides to form cross-β-structure in many diverse orientations forming the so called dry steric zippers, which might underlay the existence of different prion strains.6 However the main question on the biochemical and structural nature of PrP conversion process remained unanswered.

Prion diseases are characterized by conversion of the native PrP into the form PrP^Sc, which forms amyloid aggregates that are resistant to proteolysis. Tertiary structure of the native form of PrP from more than 15 different species has been determined.7 Their fold is highly conserved, with an unstructured N-terminal half of the protein and a C-terminal structured domain consisting of three α-helices and two β-strands.8 The native form of PrP exhibits high content of α-helical structure, while the converted form is dominated by the β-type secondary structure. The secondary structure content of the PrP^Sc is somewhat controversial. Analyses of infrared or CD spectra suggest that the secondary structure of converted PrP contains between 17–30% of α-helix and 43–50% of β-structure.9,10 This is clearly different from the all-β structure. It is in fact compatible with the conservation of the large part of the secondary structure elements of the C-terminal globular domain and induced formation of the β-structure from the proximal N-terminal segment, disordered in the native state.

The defined tertiary structure of proteins is determined by the multitude of cooperative interactions that provide the sufficient free energy gap between the native and nonnative conformations. The existence of alternative, significantly different global folds of any protein has not been demonstrated yet at the level of a defined tertiary structure. It would be in fact extremely difficult to stabilize the alternative stable fold, where many of the corporative interactions would have to be optimized simultaneously. This proposition is supported by the observation that most of the proteins involved in conformational diseases contain a segment with an intrinsically unfolded structure in the native state.11 Therefore it is much more likely for those unfolded segments to adopt an ordered conformation rather than to completely refold the native globular structure.

Aggregation state of the PrP^Sc hinders determination of high resolution structure. We can however use different biochemical approaches to inquire about the nature of the conversion process and structure of the converted form. Methods, such as antibody mapping,12,13 hydrogen exchange,14–16 binding of fluorescent ligands17 and many others have been used, revealing that both C-terminal and proximal N-terminal segment of the PrP become less accessible to the solvent upon conversion.

In order to unravel the molecular mechanism of PrP conversion, we decided to investigate which of the secondary structure elements or their suprasecondary structure combinations are retained in the converted form. We introduced disulfide tethers into different positions within the globular C-terminal segment of mPrP, connecting different secondary structure elements.18 Several pairs of residues that adhere to the geometric requirements for a disulfide formation were selected.19 Covalent tethers impose a very strong structural constraint as the relative position of the tethered pair needs to remain the same in the converted structure. This approach therefore allowed us to probe the relative position of all secondary structure elements in the converted PrP. We successfully prepared seven disulfide-tethered variants of mPrP. The only variants that we could not prepare were those where the introduced cysteines were in the neighborhood of the existing disulfide and probably led to the heterogeneous disulfide shuffling yielding misfolded products. We demonstrated that in all variants additional disulfides are formed and the secondary structure of the native form of PrP variants is indistinguishable from the wild-type PrP.

The key experiment was in vitro conversion of PrP disulfide mutants. Surprisingly, the majority of the disulfide-tethered variants was able to convert into PrP fibrils. Mutant fibrils had the same morphology, determined by AFM and TEM, pattern of antibody mapping and high content of β-structure as the fibrils prepared from wild-type PrP. The common structural property of the three variants that did not convert and retained the native, α-helical conformation, regardless of the conversion protocol, is that they all tethered the two subdomains B1-H1-B2 and H2-H3 to each other (Fig. 1). The proof that this is indeed an intrinsic structural property rather than a result of serendipitous point mutations is that both variants with single cysteine residues of the disulfide pair retained the ability to convert. Moreover reduction of disulfides rendered the originally non-converting disulfide variants convertible into the β-structured fibrils. Even simultaneous introduction of two new disulfides, one into each of the two subdomains, retained the ability of PrP variants to convert. Introduction of disulfides predominantly improved the stability of the protein, increasing the Tm by 3–12 degrees. However, the conversion ability had no correlation with the thermal stability of the protein as some of the most stable variants, containing two disulfides that increased Tm by more than 16 degrees, readily converted.

Those results provide an exceptionally strong set of constraints to characterize the conversion process and structure of the converted form. Our results are not compatible with most of the current structural models of PrP conversion, which suggest unfolding or significant rearrangements of secondary structure elements of the globular domain.14,16,20,21 Separation of subdomains of PrP implies that this process requires high activation energy or highly unfolding conditions. The loop linking the two subdomains connects B2 to H2 and has also been called “the rigid loop,” named by the increased ordering in the elk PrP in contrast to mouse or human PrP.22 This loop has been implicated in the species barrier23,24 and protective polymorphisms.25 Mice carrying mutations S170N N174T, where residues from mouse PrP are replaced with the corresponding residues from elk, develop spontaneous transmissible prion disease.26

It might be in principle possible that disulfide variants convert off-pathway from the physiologically relevant PrP^Sc form. However we were able to demonstrate the same properties in cell cultures; only in vitro convertible PrP variant was able to replicate prions, while the unconvertible variant did not.

The only structural transition that is compatible with our results is domain swapping of the C-terminal globular domain. Domain swapping represents the mechanism of oligomerization where the monomer and oligomer share the majority of the secondary structure elements. Most of the residues in the swapped-dimer oligomer are in exactly the same type of chemical environment as in the monomer with the exception of residues that represent the hinge of subdomain separation and connection between the monomeric units. Domain swapping requires high activation energy as the monomer has to unfold during conversion. The resulting oligomers are typically extremely stable and often a single protein can form different domain-swapped oligomers.27

In order to confirm domain-swapped model of prion protein conversion we performed additional experiments where we analyzed the conversion products of a mixture of the two single cysteine mutants. Those single cysteine variants were designed in a way that if swapping of the sub-domains B1-H1-B2 and H2-H3 occurs during conversion, cysteines from different single cysteine variants come into contact and can form a disulfide bridge. Indeed proteinase K-resistant covalent dimers were only observed upon conversion of a mixture of both variants.

In conclusion, we present the model of PrP conversion, where the conversion process requires unfolding of the core of the structured C-terminal domain of PrP with separation of the two subdomains, which recombine into a swapped dimer (Fig. 1). It has been demonstrated previously by several different approaches that PrP dimerization is important and a rate limiting step in conversion.28–30 This swapped dimer represents the building block of fibrils and the template for structuring of the unfolded N-terminal segment, which can anneal to the dimer in the form of the β-strands, such as demonstrated in peptide dry steric zippers. We propose that the variability between different strains of prions may originate from differently annealed β-strands of the N-terminal segments and can additionally be affected by posttranslational modifications and the presence of additional molecules, such as nucleic acids or lipids.

Figures and Tables

Figure 1 Mapping of PrP conversion by disulfide tethers. Disulfides engineered within the globular domain of PrP have different effects on its ability to convert into fibrils. Disulfide tethers are schematically represented as straight connectors on mouse PrP structure (1XYX).22 All disulfides (left top), which tether on one side subdomain B1-H1-B2 (gray) and on the other subdomain H2-H3 (black) prevent conversion, while PrP variants with single or even double disulfide tethers within each or both of the two subdomains retain the ability to convert into fibrils (left bottom). Results suggest that the secondary structure of each of the two subdomains is conserved during conversion, which can be accomplished by separation of subdomains (middle) followed by domain swapping. Domain swapped PrP dimer thus represents the building block of fibrils and a template for the annealing of the disordered N-terminal part into β-structure. Monomers within a swapped dimer are shown in gray and black (right).

Acknowledgments

The authors acknowledge the financial support from the Slovenian Research Agency (I.H.B., R.J.), EN-FIST Centre of Excellence (R.J.), and the sixth framework EU project, TSEUR.