Prion Propagation

Abstract

The transfer of phenotypes from one individual to another is a fundamental aspect of biology. In addition to traditional nucleic acid-based genetic determinants, unique proteins known as prions can also act as elements of inheritance, infectivity, and disease. Nucleic acids and proteins encode genetic information in distinct ways, either in the sequence of bases in DNA or RNA or in the three dimensional structure of the polypeptide chain. Given these differences in the nature of the genetic repository, the mechanisms underlying the transmission of nucleic acid-based and protein-based phenotypes are necessarily distinct. While the appearance, persistence and transfer of nucleic acid determinants require the synthesis of new polymers, recent studies indicate that prions are propagated through dynamic transitions in the structure of existing protein.

The Protein-Only Mechanism

The prion hypothesis was originally proposed to explain the atypical etiology of the transmissible spongiform encephalopathies (TSEs), a group of progressive and fatal neuro-degenerative diseases including scrapie in sheep, bovine spongiform encephalopathy (or mad cow disease) in cattle, and Creutzfeldt-Jacob Disease (CJD) and kuru in humans.¹ Strikingly, these infectious diseases may also develop spontaneously or through genetic predisposition, suggesting that the genetic determinant is actually host encoded.²

Historically, elements of infection and inheritance are thought to rely on a nucleic acid core to encode genetic information. The TSE agent proved to be enigmatic, however, resisting exposure to manipulations known to destroy nucleic acids, such as nucleases and radiation, while succumbing to treatments that disrupt protein structure, including proteases,¹ detergents,³ denaturants,⁴ chaotropic salts⁵ and organic solvents.⁶ Consistent with these observations, a 27–30 kDa host encoded protein was found to be the major constituent of highly purified brain homogenate preparations that retained infectivity,^7–10 leading Prusiner to postulate that the scrapie agent was an infectious protein (prion) termed PrP.¹⁰ This once heretical model also appears to be applicable to a number of unrelated phenotypes particularly in fungi, where protein-only models accurately describe the inheritance of a wide range of previously inexplicable phenotypes including the use of alternate nitrogen sources,¹¹ the regulation of translation termination efficiency,¹¹ the formation of heterokaryons,¹² the appearance of other prions¹³ and organismal dependence on quality control pathways.¹⁴

While appealing, the idea of protein-directed inheritance is at odds with our classical view of genetics. In the simplest case, the transmission of phenotypes from one individual to another either through heredity or infectivity relies on two key events. First, the genetic determinant of the trait must be able to replicate itself to produce identical copies. Second, a mechanism must exist to ensure that these copies are efficiently partitioned between donor and recipient. As their molecular architecture is particularly well suited to self-templated replication, nucleic acids, specifically the sequence of bases found in these polymers, form an effective repository for genetic information (Fig. 1A). The idea of an infectious protein, however, immediately poses a mechanistic dilemma in the absence of a protein-directed protein synthetic pathway. Thus, protein-only genetic information must be fundamentally distinct from the sequence-based genetic information carried by nucleic acids (Fig. 1B).

Figure 1.

Genetic information can be replicated in two ways. (A) For nucleic acid-based determinants, replication occurs through the polymerization of free nucleotides (light blue), using the sequence of bases in an existing polymer (dark blue) as a template. (B) For protein-based determinants, genetic information is encoded in quaternary structure. For the example shown here, a simple rectangle can be assembled with the subunits aligned along either their short or their long faces, leading to two distinct self-replicating polymers. Adapted from Caughey.¹⁴²

In 1967, Griffith proposed a mechanism whereby a protein could catalyze its own replication.¹⁵ Based on this proposal, self-replicating proteins, such as prions, must adopt at least two distinct conformations, a normal form and a disease-associated form, with the latter serving as a template for the conversion of the former. In line with this prediction, the PrP^C (cellular) and PrP^Sc (scrapie)¹⁶ conformational isomers (conformers) as well as the two forms of the yeast prion proteins are readily distinguished by their sensitivity to proteases,^17–22 their state of oligomerization,^{18,19,22–25} and their secondary structure.²⁶

The idea of a self-replicating protein conformation gained additional support when this principle was directly demonstrated in cell-free systems. In these assays, the abnormal form of a prion protein directs the conversion of soluble prion protein to an oligomerized amyloid state sharing the biochemical characteristics of ex vivo prion-state protein.^24,27–32 In many cases, material generated by in vitro conversion reactions has been shown to promote or accelerate the appearance of prion-associated phenotypes when exogenously supplied to susceptible hosts.^33–39 These observations support the idea that protein-based genetic elements, unlike their nucleic acid-based counterparts, encode information in their quaternary rather than their primary structures (Fig. 1B).

Since the appearance, spread and reversal of prion-associated phenotypes involve changes in protein state rather than the synthesis of new protein components, a clear mechanistic understanding of the process requires insight into prion protein dynamics in vivo. With this idea in mind, the prion mechanism can be broken down into three discrete fundamental steps. First, non-prion state protein must adopt the prion conformation, a process that may be either spontaneous or templated by existing prion-state protein. Second, the prion template must be continually regenerated to provide new surfaces for conversion. Finally, prion-state protein must be transmitted to other cells either by extracellular secretion and uptake for non-dividing cells or via partitioning to daughter cells in actively dividing cultures. Each step of this in vivo prion cycle must be undertaken with high precision to maintain a strong link between protein state and phenotype in order for the prion mechanism to serve as an effective alternate route for the inheritance of traits. In this review, we focus on insights into the mechanisms underlying prion propagation in vivo gleaned from studies of prion protein dynamics.

Prion Conversion In Vivo

The biosynthesis and maturation of proteins in a eukaryotic cell is an intricate process that is regulated by the action of chaperones and proteases and that is influenced by subcellular compartmentalization. ^40,41 While the prion hypothesis predicts that the non-prion conformer can be directly remodeled to the prion form in the presence of a pre-existing prion template,^15,16 the range of quality control pathways regulating protein biogenesis must be considered when extending the predictions of the prion hypothesis to a living cell. For example, protein maturation begins as soon as the polypeptide emerges from the ribosome exit tunnel, as co-translationally acting chaperones⁴¹ and proteolytic mechanisms^42,43 engage the nascent peptide. Indeed, productive folding pathways often lead proteins to their mature folded state by the time synthesis is complete.⁴⁴ With these observations in mind, a key consideration in developing a mechanistic understanding of any in vivo prion cycle is the point at which the alternate biogenesis pathway is initiated, whether it be nascent, non-native or mature prion protein (Fig. 2A).

Figure 2.

In vivo prion conversion. (A) In vivo, a prion protein (dark blue) may convert to the prion form at multiple points during its biogenesis pathway (see text for details). (B) PrP^C (light blue ball and stick) likely converts to the prion form (dark blue ball and loop) at the cell surface (left) or in endocytic vesicles (right). (C) The prion phenotype is transmitted from a [Het-s] prion strain (dark gray) to [Het-s*] non-prion strain (white) through cytoplasmic mixing following hyphal fusion (light gray). This transmission may occur either through the migration of existing [Het-s] complexes (dark blue) into the [Het-s*] recipient and/or via the conversion of existing non-prion state protein (light blue ball and stick) to the prion form (light blue ball and loop). (D) Upon mating, a [psi^-] cell (white) to a [PSI⁺] cell (dark gray), the prion phenotype establishes dominance via incorporation of existing non-prion state protein (light blue) into the introduced prion complexes (dark blue).

For PrP, the conversion process has been extensively studied within the context of a number of scrapie infected cell culture systems, which support conversion of newly made PrP to the prion state at a low frequency (∼10% of total PrP).^20,45,46 Using these systems, several lines of evidence suggest that PrP converts to the prion state after synthesis (Fig. 2B). First, pulse-labeled PrP transits from a protease sensitive to a protease resistant state on the time scale of hours.^45,47 Second, PrP localization appears to impact the conversion process. PrP is a cell-surface glycoprotein,⁴⁸ and retention of PrP in the endoplasmic reticulum by treatment with either brefeldin A⁴⁹ or intracellular antibodies⁵⁰ or removal of PrP from the cell surface by phospholipase treatment^45,51 inhibits the accumulation of protease resistant PrP. Third, exposure of infected tissue culture cells or animals to anti-PrP antibodies inhibits accumulation of protease resistant PrP and infectious prions by inducing PrP degradation or by retaining it at the cell surface, suggesting that conversion may normally occur in endocytic vesicles.^52–55 These observations clearly indicate that the final hallmarks of conversion, protease resistance and infectivity, appear late in the maturation of PrP. A future challenge is to determine if this alternate biogenesis pathway is initiated at the time of synthesis or if mature PrP^C is a direct substrate for this transition.^45,47,49,51

The in vivo conversion process has also been studied in the propagation of two fungal prions, [Het-s], a regulator of heterokaryon compatibility in Podospora anserina,¹² and [PSI⁺], a regulator of translation termination efficiency in Saccharomyces cerevisiae.⁵⁶ In each case, conversion was studied by introducing the non-prion and prion forms of the corresponding proteins into the same cell by fusion. For [Het-s], heterokaryons formed between non-prion ([Het-s*]) and prion ([Het-s]) cells display the prion phenotype: incompatibility with strains harboring the het-S allele.⁵⁷ Upon fusion of [Het-s*] and [Het-s] strains, the prion phenotype can be tracked as it migrates into the non-prion recipient from the site of fusion by dissecting and challenging fragments of the mycelium with a het-S strain.¹² In these experiments, spread of the prion phenotype occurs in the presence of cycloheximide, indicating that the process does not require new protein synthesis.¹² This observation can be explained either by direct conversion of non-prion state Het-s protein to the prion form¹² or alternately by the migration of existing prion state protein from the donor into the recipient cell, as the [Het-s] phenotype is a gain-of-function (Fig. 2C).²⁵ In either case, a dynamic transition in the physical state and/or localization of the Het-s protein is a key contributor to the infectious spread of the [Het-s] prion.

The [PSI⁺] prion, a conformer of the Sup35 protein, is perhaps the most extensively studied of the fungal prions. Conversion of Sup35 from its non-prion [psi^-] state to its prion [PSI⁺] state is accompanied by a partial inhibition of normal Sup35 function in translation termination.^18,19,58 This easily scored link between the physical and functional states of the protein has been a useful tool in dissecting the mechanism of prion conversion in vivo. By mating a [PSI⁺] cell expressing unmarked Sup35 to a [psi^-] cell expressing a Sup35 fusion to the green fluorescent protein (Sup35-GFP), conversion of soluble pre-existing Sup35 to an aggregated state can be directly observed in live cells, and this process occurs on the time scale of minutes (Fig. 2D).⁵⁹ This transition in Sup35 physical state is accompanied by loss of Sup35 activity, as measured by stop codon read-through.⁵⁹ These studies indicate that the prion state does not need to be specified at the time of synthesis but instead can be transmitted to existing, mature non-prion state protein.

Regeneration of the Prion Template In Vivo

Although the original formulations of the prion hypothesis suggested that the template for prion conversion could be a monomer,^15,16 mathematical models of prion infectivity are inconsistent with this idea.^60,61 Rather, such modeling studies suggest that small oligomers would function as the most infectious units.^62,63 Consistent with these predictions, ionizing radiation,^64–66 fractionation,^67,68 and denaturation⁶⁹ studies support a minimum infectious complex of ∼5 PrP subunits, with peak infectivity in the range of 14–28 monomers.⁶⁸ Moreover, the threshold behavior of scrapie appearance, in which the sporadic rate of disease is exceedingly low, can only be explained by a mechanism in which multiple, spontaneously arising PrP^Sc monomers must come together to form a stable complex that functions as an active template.^60,70 In the alternate scenario, every individual in the population would eventually develop a TSE, as each spontaneously arising PrP^Sc monomer would be sufficient to establish disease.⁶⁰ The concept of an oligomeric template is also supported by fungal experiments in which external delivery of protein to live cells and structure-based mutagenesis studies have linked heritable and infectious [Het-s], [PSI⁺], [URE3], and [PIN⁺] prions to ordered amyloid fibers of the Het-s,^39,71 Sup35,^35,37 Ure2,³³ and Rnq1³⁸ proteins, respectively.

Assuming a linear polymer model, the number of free ends (i.e., the templating surfaces) is a limiting factor for the rate of conversion and therefore disease progression.^60,72 Since the number of complexes spontaneously arising or those existing in an infectious inoculum are likely to be insufficient to establish an infection,⁶² early theoretical models of disease predicted a second vital step in the process: the continual generation of additional catalytic surfaces.^60,72 Various mechanistic scenarios are possible to generate these secondary sites of nucleation,⁶⁰ but the continual fragmentation of existing linear templates is the most widely favored pathway.^60,70,72,73

Indirect evidence points to the importance of fragmentation of PrP complexes in vivo. For example, PrP^Sc accumulation is exponential following infection,^74–76 and only mathematical models that consider fragmentation can accurately describe these kinetics.^63,77 Moreover, prion infectivity greatly increases upon partial disruption of ex vivo preparations of PrP by denaturation,⁶⁹ liposome dispersion,^67,78 homogenization⁷⁹ and sonication,^68,80 highlighting the potent effect of fragmentation on prion titre. Finally, the yield of in vitro conversion reactions is greatly improved by cyclic rounds of polymerization and sonication,³¹ a process that increases titres to a level sufficient to establish an infection in vivo.³⁴ Thus, although an endogenous fragmentation activity has not been directly observed in the mammalian prion system, such a process, whether stochastic or catalyzed, has the potential to greatly impact the establishment and progression of disease.⁶²

In S. cerevisiae, template fragmentation is an active process, catalyzed by the molecular disaggregase Hsp104. Hsp104 is a member of the AAA⁺ ATPase family⁸¹ and is required for survival of yeast at high temperatures, provided the organism is first exposed to a more modest heat stress.⁸² At these elevated temperatures, Hsp104's essential function is the resolution of thermally induced aggregates.^83,84 Hsp104 was first identified as a prion modulator in a screen to identify factors stimulating loss of [PSI⁺] when overexpressed.⁸⁵ In this and subsequent studies, an essential role for Hsp104 in the propagation of all fungal prions was uncovered, as inactivation of Hsp104's ATPase activity by deletion, expression of a dominant mutant⁸⁵ or treatment with guanidine hydrochloride (GdnHCl)^86–90 leads to prion loss in all cases.^{13,82,85,91,92}

Insight into the mechanism of Hsp104 action arose from early studies on the kinetics of prion loss or curing by GdnHCl. When [PSI⁺] cultures are grown in the presence of GdnHCl, [psi^-] cells begin to appear after a 4–5 generation lag, and mathematical models of the kinetics of this curing event suggest a two-step process: (1) a failure to replicate or fragment existing prion templates and (2) subsequent dilution of these complexes during cell division.^93,94 Consistent with these ideas, Sup35 complexes increase in size upon Hsp104 inactivation,^95–97 and [PSI⁺] curing only occurs in actively dividing cultures upon Hsp104 inhibition.⁹³ Hsp104 appears to provide similar fragmentation activity in the propagation of [PIN⁺], as Rnq1 complexes also increase in size upon Hsp104 inhibition.⁹⁸ Intriguingly, loss of Sis1, an Hsp40 family member, similarly affects Rnq1 complex size, suggesting that Sis1 and Hsp104 cooperate to generate additional Rnq1 templates in vivo.⁹⁸

Complementing in vitro observations of Hsp104-dependent fragmentation of Sup35 amyloid fibers^99,100 (see also refs. ¹⁰¹ and ¹⁰²), direct proof of Hsp104-dependent fragmentation in vivo was gleaned from observations of prion protein dynamics in live cells. When new protein synthesis is inhibited in cells with wildtype Hsp104 activity, existing prion complexes marked with Sup35-GFP become undetectable by microscopy within hours;^97,103 however, upon Hsp104 inhibition, the same complexes persist.⁹⁷ Since Hsp104 does not alter the metabolic stability of Sup35,⁹⁷ the observed loss of fluorescence in wildtype cells provides an assay for fragmentation. Through the repeated fragmentation of complexes by Hsp104 and the subsequent incorporation of constitutively expressed untagged Sup35, the original pool of Sup35-GFP monomers is redistributed among a greater number of prion complexes, leading to their decreased intensity. Consistent with this idea, quantitative imaging techniques indicate that Sup35-GFP complexes remain the same size despite the progressive loss of fluorescence in wildtype cells.^97,103

Secondary Effects of Fragmentation

Fragmentation is predicted to impact two events during prion propagation: conversion efficiency and prion transmission (Fig. 3).^62,72,73,77 Experimental proof of these predictions has been provided by studies of the Sup35/[PSI⁺] prion in which dynamic transitions in aggregation state, as assayed by fluorescence microscopy and biochemical analyses, were correlated with changes in functional state, as assayed by the efficiency of translational termination. Using these powerful approaches, efficient inactivation of newly synthesized Sup35 by incorporation into existing prion complexes is evident in wildtype cells; however, nascent Sup35 accumulates in a soluble and functional pool upon Hsp104 inhibition, suggesting a defect in the kinetics of conversion likely due to a limitation in the number of prion templates.^{18,97,104,105} This defect is immediately apparent within the first generation of Hsp104 inhibition and can be reversed within the same time frame by reactivation of Hsp104.^97,104–106 Thus, the [PSI⁺] prion cycle is finely tuned in vivo, with the continual generation of new prion templates by fragmentation making a key contribution to this razor's edge balance.

Figure 3.

Fragmentation has multiple consequences. Non-prion state protein (light blue ball and stick) converts to the prion form (light blue ball and loop) upon incorporation into existing prion complexes (dark blue ball and loop), which are subsequently fragmented, either stochastically or catalytically, to generate smaller complexes. This fragmentation facilitates spread of the prion templates, shown here as partitioning to daughter cells in a dividing culture—pathway 1, and increases the efficiency of incorporation of additional non-prion subunits (dark blue ball and stick)—pathway 2.

Despite the importance of fragmentation for conversion efficiency, newly synthesized Sup35 does continue to join the static prion complexes that persist upon Hsp104 inhibition.^97,104 This incorporation gradually decreases the mobility of complexes in the yeast cytosol, as revealed by fluorescence recovery after photobleaching (FRAP) experiments.^97,105 The resulting, largely immobile complexes are inefficiently transferred to daughter cells, leading to a partitioning defect and ultimately loss of [PSI⁺].^97,106

Modulating Prion Dynamics

The role of Hsp104 in fungal prion propagation is a dramatic example of the importance of dynamic transitions in prion protein physical state for the propagation of prion-associated phenotypes. In addition to chaperone effects, homotypic and heterotypic interactions between prion proteins themselves have emerged as potent modulators of in vivo prion cycles.

The prion hypothesis originally predicted that prion proteins can physiologically access two physical states, but studies in both mammals and lower eukaryotes now indicate that this conformational flexibility is much more complex than the model first proposed, with a range of physical and phenotypic states (strains or variants) possible.^56,107–109 These variants have important ramifications for prion biology in vivo. For example, PrP variants differ in both their incubation times and in their patterns of neurodegeneration,¹¹⁰ and variants of the [PSI⁺], [PIN⁺] and [URE3] yeast prions can be distinguished by the severity of their phenotypes.^107,109,111 Intriguingly, recent studies of both PrP and Sup35 variants revealed a difference in the stability of prion complexes, assessed by denaturation either with GdnHCl or SDS,^95,112–114 and this range of stabilities is likely due to conformational differences between the assembled subunits.^{35,37,115–122} In these studies, an increase in the stability of prion complexes, and presumably a decrease in the rate of endogenous fragmentation,⁶² diminished the severity of the prion phenotype (i.e., incubation time for PrP and degree of functional inactivation for Sup35). Thus, conformational differences alone can have profound effects on the physiological consequences of prion propagation.

In addition to sequence-independent effects, the efficiency of prion propagation by wildtype protein can be dramatically altered by co-expression of sequence variants. For example, co-expression of PrP proteins derived from different species interferes with templated conversion of PrP to a protease-resistant form both in vivo¹²³ and in vitro,¹²⁴ and in some cases, inhibition can be linked to single amino acid changes.^125,126 Similar findings have been reported in the yeast system in which fragments of Ure2²³ or point mutations within Sup35 disrupt or diminish prion propagation by wildtype protein.^127–129 Three mechanisms have been proposed to explain these dominant effects. First, sequence differences could alter prion protein interaction with an essential trans regulator.^123,127,130 Second, mutant proteins could interact with and cap existing aggregates, thereby decreasing their templating ability.^{23,127,131,132} Finally, mixed complexes, containing wildtype and mutant proteins, could adopt conformations that are not efficiently inherited.^127,131 Future studies of prion dynamics in the presence of these dominant inhibitors will likely be instructive in revealing their mechanisms of inhibition in vivo.

Inter-prion interactions can also profoundly alter an in vivo prion cycle. A case in point is the de novo appearance of a new prion. In both yeast and mammals, such events are rare, but the frequency of de novo prion induction increases greatly if the prion protein is overexpressed, perhaps by increasing the frequency of a stochastic misfolding event.^{11,12,111,133,134} Whether spontaneous misfolding occurs during or after synthesis is unclear; however, studies in yeast suggest that this initiating event is itself nucleated. In the case of the [PSI⁺] prion, de novo appearance by overexpression of Sup35 depends on the presence of a pre-existing protein template, most frequently the [PIN⁺] prion, encoded by Rnq1.^13,92,133 Recent studies in vitro and in vivo suggest that Rnq1 complexes are likely to effect Sup35 dynamics: Rnq1 complexes heterogeneously nucleate the formation of Sup35 complexes^135,136 and are required for conversion of these nascent Sup35 oligomers to a heritable form.¹³⁷

In addition to positive interactions between prions, co-existence of multiple prion forms, whether they are different sequences or different variants of the same sequence, is often disfavored. For example, co-inoculation of mice with distinct PrP variants extends incubation timing;^138,139 de novo induction of [URE3] by overexpression of Ure2 is diminished in [PSI⁺] strains,¹⁰⁷ and despite their positive interactions under other conditions, some variants of [PIN⁺] destabilize variants of [PSI⁺].¹⁴⁰ These observations suggest a competition for some component common to the propagation pathways. One example of this idea is the competition between different variants of the same prion protein. When two different variants are introduced into the same cytoplasm by mating yeast, only one form persists when the resulting diploids are allowed to grow into colonies, and the protein state is subsequently analyzed.^114,141 This dominance likely results from differences in fragmentation rate, allowing one variant to out compete another for the incorporation of the same pool of nascent prion protein.¹¹²

Concluding Remarks

The prion hypothesis has expanded our view of genetics to include proteins as potential determinants of phenotypic traits. Several examples of this alternate route of inheritance and infectivity now exist, in which protein conformations impart unique phenotypes to an organism. The distinctions between prion and non-prion conformers and phenotypes are, however, extreme points on a continuum. This inherent metastability of protein-based genetics underlies its fascinating biology; that is, the potential for new phenotypes to arise, persist, spread and be lost within the lifetime of an organism. The multi-step prion cycle in vivo provides many exquisitely sensitive points of regulation and potential intervention into this process. Remarkably, slight variations in prion protein dynamics, mediated by either cis or trans effectors, have the capacity to profoundly and swiftly impact the biological consequences of protein-based genetic elements.

Abbreviations

GdnHCl

guanidine hydrochloride

Prp

prion protein

TSEs

Transmissible Spongiform Encephalopathies

CJD

Creutzfeldt-Jacob Disease