Dissociation of recombinant prion autocatalysis from infectivity

ABSTRACT

Within the mammalian prion field, the existence of recombinant prion protein (PrP) conformers with self-replicating (ie. autocatalytic) activity in vitro but little to no infectious activity in vivo challenges a key prediction of the protein-only hypothesis of prion replication – that autocatalytic PrP conformers should be infectious. To understand this dissociation of autocatalysis from infectivity, we recently performed a structural and functional comparison between a highly infectious and non-infectious pair of autocatalytic recombinant PrP conformers derived from the same initial prion strain.¹ We identified restricted, C-terminal structural differences between these 2 conformers and provided evidence that these relatively subtle differences prevent the non-infectious conformer from templating the conversion of native PrP^C substrates containing a glycosylphosphatidylinositol (GPI) anchor.¹ In this article we discuss a model, consistent with these findings, in which recombinant PrP, lacking post-translational modifications and associated folding constraints, is capable of adopting a wide variety of autocatalytic conformations. Only a subset of these recombinant conformers can be adopted by post-translationally modified native PrP^C, and this subset represents the recombinant conformers with high specific infectivity. We examine this model's implications for the generation of highly infectious recombinant prions and the protein-only hypothesis of prion replication.

A misfolded conformer of the prion protein (PrP) termed PrP^Sc is an essential component of infectious mammalian prions.² Through an as yet unknown mechanism, PrP^Sc interacts with cellular PrP (PrP^C) and templates the conversion of PrP^C into additional PrP^Sc. This converting activity of PrP^Sc is called autocatalysis and is presumed to underlie the infectious nature of the prion diseases.

It is widely believed that autocatalytic PrP^Sc is the sole component of infectious prions; however, attempts to rigorously demonstrate a protein-only mechanism of prion replication have so far been unsuccessful. Interestingly, in the process of testing the protein-only hypothesis, researchers have generated a variety of autocatalytic recombinant PrP conformers in vitro, but only a select few of these conformers possess levels of specific infectivity in vivo approaching that of native prions (reviewed in³). This dissociation between the autocatalytic and infectious activities of recombinant PrP conformers has challenged a key prediction of the protein-only hypothesis – that autocatalytic PrP conformers should be infectious – and remained as yet without a satisfactory explanation.

To better understand the mechanism for the dissociation between PrP autocatalysis and infectivity we recently performed a structural and functional comparison between 2 autocatalytic recombinant PrP conformers generated from the same initial seed but differing >10⁵-fold in specific infectivity for wild-type mice.¹ By comparing this matched pair of autocatalytic recombinant PrP conformers, one fully infectious and the other without detectable infectivity, we sought to identify structural and functional features that correlate with PrP^Sc infectious activity.

Infectious and non-infectious recombinant PrP conformers differ structurally in restricted C-terminal domains

Using hydrogen/deuterium exchange mass spectrometry (DXMS) we found that our infectious and non-infectious autocatalytic PrP conformers were remarkably similar, as measured by solvent accessibility, throughout their C-terminal protease-resistant cores.¹ Both conformers displayed a high degree of protection against solvent exchange, consistent with previous DXMS studies of mammalian prions and with those PrP^Sc models that propose a complete refolding of the PrP C-terminus to β-sheet secondary structure.^1,4-8

Despite the striking structural similarities between the infectious and non-infectious PrP conformers, structural differences were detected in 2 restricted C-terminal domains, including residues 91-115 and 144-163 (mouse PrP numbering).¹ Within these domains, the non-infectious conformer appears to have relatively increased solvent accessibility. Using conformation-specific immunoprecipitation and Raman spectroscopy we were able to confirm that structural differences between the infectious and non-infectious conformers reside within these 2 restricted domains.¹

The first of these 2 domains, residues 91-115, is located within the unstructured N-terminus of normally folded α-PrP.⁹ Not much is known about the role of this domain in PrP misfolding, although it contains 3 residues at which amino acid substitutions lead to familial prion disease,¹⁰ as well as the so-called 'fifth site' for copper binding, which was recently implicated in the PrP conversion process.¹¹

The second domain within which our infectious and non-infectious PrP conformers structurally differ (residues 144-163) corresponds to the first α-helix and second β-strand of α-PrP.⁹ There are many existing lines of evidence to suggest that this α₁- β₂ domain plays a key role in the formation of infectious prions. Several PrP^Sc-specific antibodies have epitopes that reside within the α₁- β₂ domain.^12,13 Small deletions toward the C-terminus of this domain inhibit PrP conversion and, in fact, yield PrP molecules with dominant negative activity against PrP^Sc formation.¹⁴ Furthermore, the complete absence of the α₁- β₂ domain in a redacted 'miniprion' does not prevent PrP^Sc formation, but such miniprions are non-infectious when inoculated into animals expressing full-length PrP.¹⁵ Finally, the α₁- β₂ domain also lies immediately adjacent to the β₂-α₂ loop, a region with notable influence on prion formation and interspecies prion transmission.¹⁶

Infectious and non-infectious autocatalytic recombinant PrP conformers differ in their ability to convert post-translationally modified PrP substrates

To understand the mechanistic basis for the dissociation of autocatalytic and infectious activities in recombinant PrP conformers, we undertook a series of in vitro conversion experiments utilizing our matched pair of infectious and non-infectious autocatalytic conformers to seed differentially modified PrP substrates. Native PrP^C is post-translationally modified by the addition of 2 N-linked glycans and a C-terminal glycosylphosphatidylinositol (GPI) anchor.² We reasoned that perhaps recombinant PrP, free of these bulky modifications, has relatively few conformational constraints and is capable of accessing a greater variety of autocatalytic states than native PrP^C. By this logic, only those recombinant PrP conformations that are compatible with the presence of PrP^C post-translational modifications would be capable of replication in vivo, and this group of recombinant PrP conformers would represent the potentially infectious subset (Fig. 1). \raster="rgFigKPRN_A_1123843_F0001_B"

FIGURE 1.

Model to account for the dissociation of autocatalytic activity from infectivity in some recombinant PrP^Sc conformers. We propose that recombinant PrP, lacking post-translational modifications and associated folding constraints, is capable of adopting a wide variety of autocatalytic conformations. Native PrP^C, which is modified by a GPI anchor and up to 2 N-linked glycans, can only adopt a subset of these conformers. Those recombinant PrP conformers that are compatible with a post-translationally modified PrP^C substrate are highly infectious. G = N-linked glycan; GPI = GPI anchor. Adapted from Noble et al.¹

To test this model, we used our infectious and non-infectious recombinant PrP conformers to seed in vitro conversion reactions containing modified native PrP^C substrates (summarized in Table 1). We found that the non-infectious recombinant PrP conformer was unable to template the conversion of native PrP^C or deglycosylated PrP^C,¹ while the infectious conformer was a functional seed in both settings. Unfortunately, the complementary experiment, in which delipidated native PrP^C is utilized as a conversion substrate is not possible for technical reasons.^17,18 Nonetheless, the results of these conversion experiments are consistent with our proposed model (Fig. 1) and specifically suggest that our non-infectious recombinant PrP conformer fails to function as a seed in vivo because it is unable to template the conversion of GPI anchored PrP substrates.

TABLE 1.

Summary of functional data comparing our infectious and non-infectious recombinant PrP conformers derived from the same initial prion strain.^1,29 Delipidated PrP^C PMCA was not performed due to previously described technical challenges.^17,18 In previous papers, our infectious conformer is referred to as cofactor PrP^Sc and the non-infectious conformer as protein-only PrP^Sc.^1,29

	Functional Experiment
	Recombinant PrP PMCA	Delipidated PrPC PMCA	Deglycosylated PrPC PMCA	Native PrPC PMCA	Bioassay
Infectious Conformer	+	n/a	+	+	+
Non-infectious Conformer	+	n/a	−	−	−

Influence of post-translational modifications on protein folding and misfolding

It is well known that post-translational modifications can influence protein folding pathways. Indeed, N-linked glycans are known to have a chaperone-like activity during protein folding and to enhance overall protein conformational stability.¹⁹ In addition, covalently linked PrP glycans have been shown to alter the rate of amyloid fibril formation,²⁰ limit the prion strain susceptibility of host animals,²¹ and even influence interspecies transmission barriers.²²

The role of the GPI anchor modification in protein folding is less well studied. In the case of PrP, in vitro removal of the PrP GPI anchor interferes with PrP^Sc formation.¹⁷ In vivo, loss of the PrP GPI anchor appears to have only a modest effect on PrP^Sc structure,²³ but significantly alters its biochemical properties, leading to the production of large, fibrillar aggregates.²⁴ Interestingly, the experimental addition of a GPI anchor to the amyloidogenic yeast protein Sup35p appears to have an analogous biochemical influence, promoting the formation of non-fibrillar aggregates with ultrastructural similarity to PrP^Sc.²⁵ Recent work has shown that anchorless PrP^Sc more readily crosses a species barrier.²⁶ Finally, the co-expression of anchorless and wild-type PrP^C molecules in recipient animals appears to allow for the detection of infectious activity in PrP amyloid fibril preparations.²⁷

Although our recent experiments and the model outlined above suggest that a key determinant of in vivo activity for recombinant PrP conformers is the ability to accommodate a native PrP^C GPI anchor, it should be emphasized that the influence of substrate post-translational modifications may be context dependent. For alternative autocatalytic recombinant PrP conformers, one or both host PrP^C glycans may serve as an obstacle to in vivo activity. Furthermore, compatibility between recombinant seed and post-translationally modified substrate may be relative rather than absolute. It is known that non-infectious recombinant PrP amyloids can evolve into infectious prions after repeated in vivo passages, possibly by a deformed templating mechanism (reviewed in²⁸). Our recent results¹ would suggest that perhaps an early step during the deformed templating process involves serial alterations to the input recombinant PrP amyloid structure that more and more efficiently accommodate PrP^C post-translational modifications.

Implications for the production of fully infectious recombinant prions and the protein-only hypothesis

The production of fully infectious recombinant prions has been a long-standing goal of the prion field, as such material is required for the definitive determination of PrP^Sc structure by techniques like solid state NMR. To date, highly infectious recombinant PrP preparations, such as the one utilized in our recent study, have only been obtained in conversion systems that include accessory cofactor molecules, specifically phospholipids.^29,30 These phospholipids appear to maintain the infectious conformation of PrP^Sc,²⁹ but the mechanism by which this occurs is not yet known. One possible explanation, based on the model proposed in our recent study (Fig. 1), is that cofactor molecules favor the formation of autocatalytic recombinant PrP conformers with structural features that accommodate native PrP^C post-translational modifications. Such a model is consistent with the existence of highly infectious recombinant PrP conformers that do not require a cofactor, so long as native PrP^C is able to fold into that particular structure. The identification of such a cofactor-free conformer would represent a definitive proof of the protein-only hypothesis of prion replication.

The existence of autocatalytic recombinant PrP conformers lacking significant levels of specific infectivity has been a persistent obstacle to advancement within many fields of mammalian prion research. Based on our recent work, we propose that the difficulty in producing highly infectious recombinant PrP conformers may be related to the relatively broad conformational space available to recombinant PrP as compared to post-translationally modified native PrP^C. We conclude that matching between a recombinant PrP conformer's structure and the ability of natively modified PrP^C to adopt that structure governs recombinant PrP^Sc infectivity.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

No potential conflicts of interest were disclosed.

Funding

This work was supported by NIH grant RO1 NS046478 to SS.