Synthesis of High Titer Infectious Prions with Cofactor Molecules*

Recently, synthetic prions with a high level of specific infectivity have been produced from chemically defined components in vitro. A major insight arising from these studies is that various classes of host-encoded cofactor molecules such as phosphatidylethanolamine and RNA molecules are required to form and maintain the specific conformation of infectious prions. Synthetic mouse prions formed with phosphatidylethanolamine exhibit levels of specific infectivity ∼1 million-fold greater than “protein-only” prions (Deleault, N. R., Walsh, D. J., Piro, J. R., Wang, F., Wang, X., Ma, J., Rees, J. R., and Supattapone, S. (2012) Proc. Natl. Acad. Sci. U.S.A. 109, E1938–E1946). Moreover, cofactor molecules also appear to regulate prion strain properties by limiting the potential conformations of the prion protein (see Deleault et al. above). The production of fully infectious synthetic prions provides new opportunities to study the mechanism of prion infectivity directly by structural and biochemical methods.

Recently, synthetic prions with a high level of specific infectivity have been produced from chemically defined components in vitro. A major insight arising from these studies is that various classes of host-encoded cofactor molecules such as phosphatidylethanolamine and RNA molecules are required to form and maintain the specific conformation of infectious prions. Synthetic mouse prions formed with phosphatidylethanolamine exhibit levels of specific infectivity ϳ1 million-fold greater than "protein-only" prions ( Prions are the unconventional infectious agents of spongiform encephalopathies, a group of deadly neurodegenerative diseases that can be transmitted among humans and other mammals. Although these infectious diseases can be propagated indefinitely by serial passage in wild type hosts, prions do not contain a replicable genome capable of directing self-synthesis, in contrast to all other known infectious agents (2). Instead, infectious prions appear to be formed by the conversion of a host-encoded, membrane-bound glycoprotein termed PrP C 2 into several misfolded conformation(s) collectively termed PrP Sc . Nonetheless, like other infectious agents, prions can replicate to high titer and exist as self-propagating strains within the same host species, which can be characterized by distinct clinical phenotypes and selective patterns of neurotropism (3,4).
Several lines of evidence indicate that prion infectivity and strain properties are encoded by specific PrP Sc conformation(s), but at present neither the tertiary structure of infectious PrP Sc nor the molecular mechanism responsible for generating infectious PrP Sc conformation(s) has been fully elucidated, despite decades of intensive investigation. Historically, several factors have conspired to impede attempts to study these questions directly, namely: 1) PrP Sc molecules exist as insoluble aggregates, precluding structural determination by conventional methods such as x-ray crystallography; 2) the most accurate method to measure prion infectivity, end-point titration in wild type hosts, is expensive and time-consuming; and 3) no model of prion infection exists that would be amenable to genetic screening.
In recent years, we have learned much about the mechanism of prion infectivity by taking a reductionist biochemical approach, particularly the critical role played by cofactor molecules. Moreover, the insights provided by these studies have enabled the efficient synthesis of high titer infectious prions from chemically defined components in vitro, an advance that should facilitate experimental determination of the tertiary structure of PrP Sc by biophysical techniques such as solid-state nuclear magnetic resonance.

In Vitro Formation of PrP Sc and Prion Infectivity
Pioneering studies of seeded PrP C -to-PrP Sc conversion in vitro using purified substrates indicated that this process recapitulated many of the specific features of in vivo prion transmission, but was relatively inefficient (5-7). Indeed, it was observed that a stoichiometric excess of input PrP Sc was required to drive the conformational change of radiolabeled PrP C substrate in this purified system. Interestingly, the efficiency of PrP Sc formation in this system could be increased by glycosaminoglycans, providing an early clue that prion conversion might utilize factors other than PrP (8).
Soto and co-workers (9) showed that efficient amplification of PrP Sc was achievable through their development of the protein misfolding cyclic amplification (PMCA) technique. In this technique, crude brain homogenates containing PrP C and PrP Sc are mixed together and incubated for 1-3 days with intermittent sonication, amplifying both PrP Sc and prion infectivity (10). The mixture of crude brain homogenate substrates without sonication also resulted in PrP Sc amplification, suggesting (by comparison with the relative inefficiency of converting purified radiolabeled PrP C substrate) that factors present in the brain other than PrP molecules enable efficient PrP Sc amplification (11).

Identification of Endogenous PrP Sc Propagation Cofactors
In an effort to identify endogenous PrP Sc amplification cofactors in crude hamster brain homogenates, my colleagues and I subjected the homogenate substrate to various specific treatments known to degrade specific classes of molecules and assessed the effect of such treatments on PrP Sc amplification (12). Remarkably, these experiments indicated that selective degradation of single-stranded RNA within homogenates inhibited the amplification of PrP Sc molecules derived from several strains of hamster prions. Moreover, reconstitution of * This work was supported, in whole or in part, by National Institutes of Health RNA-depleted homogenates with single-stranded nucleic acids restored PrP Sc amplification, suggesting that RNA might serve as prion propagation cofactors for various hamster prion strains. Further characterization studies showed that no specific nucleotide sequence was required for cofactor activity, confirming that prions do not contain specific genomes. Interestingly, similar enzymatic treatment and reconstitution studies using several strains of mouse prions indicated that in vitro PrP Sc propagation in this species is generally much less reliant (than hamster) upon RNA as a cofactor (13). This discrepancy suggested that it might be fruitful to search for the presumably novel endogenous factor that facilitates the propagation of mouse prions in vitro. Using a reconstitution PMCA assay in which purified PrP C is mixed with fractions containing cofactor activity to produce a substrate mixture, my colleagues and I were able to purify by classical biochemical methods the novel cofactor activity and identify it as phosphatidylethanolamine (PE) (14). Additional experiments showed that PE could facilitate the propagation of PrP Sc molecules derived from multiple mouse strains and even some hamster prion strains, albeit less potently.
Because RNA and PE molecules are so different chemically, it seems likely that there are other major classes of molecules that serve as PrP Sc propagation cofactors. Novel classes of cofactors might be required for the efficient propagation of prions derived from different strains or animal species.

In Vitro Generation of Chemically Defined, Highly Infectious Prions
The identification of RNA and PE as prion propagation cofactors provided a unique opportunity to produce chemically defined infectious prions in vitro. Using native PrP C molecules immunopurified from normal hamster brain (note that this preparation was devoid of contaminating proteins, nucleic acids, and metals, but did contain stoichiometric quantities of a co-purified lipid) and synthetic polyadenylic acid as a substrate mixture, my colleagues and I were able to produce and propagate bona fide infectious prions from a minimal set of components (15). End-point titration bioassay in wild type hamsters confirmed that these prions possessed a significant level of specific infectivity (ϳ2 ϫ 10 5 LD 50 units/g of PrP). Parallel dropout controls demonstrated that the polyadenylic acid (poly(A) RNA) cofactor was essential for prion propagation. Moreover, using this simple system, it was possible to demonstrate for the first time that PrP Sc and prion infectivity could be reproducibly generated de novo (i.e. in the absence of infectious PrP Sc seed) in a completely contamination-free (prion-naive) environment (15) (an essential condition for demonstrating de novo prion formation because PMCA can be subject to cross-sample contamination (16)). Thus, the interaction between PrP C and cofactor molecules may provide a useful model for studying the spontaneous formation of infectious PrP Sc molecules in the brains of patients with sporadic Creutzfeldt-Jakob disease. Similar results were later obtained using Escherichia coli-expressed recombinant mouse (rec)PrP (rather than hamster PrP C ) substrate together with polyadenylic acid and the bacterial lipid phosphatidylglycerol (PG) (17); again, PrP Sc and prion infectivity could be produced de novo, albeit with a much lower conversion efficiency, presumably because the ability of PG to substitute for endogenous lipid cofactor(s) such as PE is relatively weak (14) and/or RNA interacts less well with mouse PrP C than with hamster PrP C (13).
To test the ability of endogenous PE to produce high titer prions, my colleagues and I generated synthetic prions using only mouse recPrP and chemically synthesized PE. In these reactions, the PrP conversion efficiency was nearly 100% (14) (as compared with ϳ2-10% previously obtained for PG (17) or crude mouse brain homogenates (18)). More significantly, endpoint bioassay showed that PE-PrP Sc molecules (PrP Sc molecules produced by serial PMCA propagation in recPrP and PE mixture) displayed a level of specific infectivity (ϳ2 ϫ 10 6 LD 50 units/g of PrP) (1) that is similar to that of brain-derived hamster Sc237 PrP27-30 molecules (ϳ6.5 ϫ 10 6 LD 50 units/g of PrP) (15).
A comparison of the levels of specific infectivity associated with various preparations of synthetic prions shows the unique ability of PE to facilitate the production truly high titer prions (Table 1). Notably, all preparations made to date without the addition of cofactor molecules have specific infectivity levels Ͻ0.1 LD 50 units/g of PrP (19 -22). Because the infectious titers of such preparations are so low (i.e. ϳ10 7 -fold lower than brain-derived prions or PE-PrP Sc molecules), it is unclear whether these minimal amounts of infectivity observed are due to either the presence of a small amount of bacterially derived cofactor(s) in the purified recPrP preparation or the creation of a minimally infectious conformation that can gradually adapt after inoculation into animals, possibly by interacting with endogenous cofactor molecules (23)(24)(25). In the case of "protein-only" samples produced by PMCA (as opposed to amyloid fibrils produced by chemical denaturation), it is also possible that either a correspondingly small fraction of PrP Sc molecules truly formed without cofactors or an accidental contamination of the sample with very small quantities of infectious prions occurred.

Essential Role of Cofactor for Maintaining Infectious PrP Sc Conformation
Collectively, the results from many different laboratories attempting to make synthetic prions (Table 1) suggest that cofactor molecules are required to produce prions with significant levels of infectivity (1, 15, 17, 19 -22). A unique opportunity to test this hypothesis directly in a controlled experiment arose once PE-PrP Sc prions were produced (1). Fully infectious PE-PrP Sc prions were used as the template seed for two lines of serial PMCA propagation processed in parallel, one using a substrate mixture containing recPrP and PE and the other using a substrate containing recPrP alone (Fig. 1). Interestingly, upon serial PMCA propagation in recPrP alone, autocatalytic protein-only PrP Sc molecules were formed exhibiting a conformation distinct from PE-PrP Sc , as assessed by the size of its protease-resistant core.
End-point titration bioassay showed that protein-only PrP Sc molecules were completely noninfectious to wild type mice, and therefore were ϳ10 6 -fold less infectious than PE-PrP Sc molecules produced in parallel (1). Thus, this internally controlled experiment confirmed that protein molecules alone are unable to maintain the infectious conformation of mammalian prions. Another critical corollary of this experiment is that autocatalytic protein misfolding does not by itself encode infectivity. It is worth noting that the large differences in specific infectivity between various preparations made either with or without cofactor (Table 1) correlate with their ability to seed homogenate PMCA reactions. Therefore, it is unnecessary to invoke other potential mechanisms, such as differential clearance or strain effects, beyond the lack of an infectious confor-mation to explain the minimal in vivo infectivity of protein-only misfolded PrP preparations.
Interestingly, the infectious conformation of PrP Sc appears to be difficult to create and maintain in vitro. Attempts to regenerate the PE-PrP Sc conformation by subsequent serial propagation of protein-only PrP Sc molecules in a substrate mixture containing both recPrP and PE were not successful (1). These negative results suggest that cofactor molecules do not simply provide a "selective pressure" for the evolution of PrP Sc conformers during serial PMCA propagation (26). Rather, it appears more likely that they play a more fundamental role in maintaining the infectious conformation of PrP Sc , which is lost irreversibly when cofactor is withdrawn.
Two groups have recently reported that noninfectious PrP Sc conformers resembling protein-only PrP Sc biochemically could be produced de novo in the presence of PG and RNA (27,28). However, cofactor withdrawal studies were not performed to distinguish whether these conformers are actually cofactor-dependent or simply represent the formation of protein-only PrP Sc in a marginal cofactor mixture.

Strain Properties of Synthetic Prions
A key study by Castilla et al. (29) showed that the biological strain properties of mouse prions could be maintained by serial PMCA propagation in crude brain homogenates. Piro et al. (30) later showed that a reconstituted system containing PrP C and brain homogenates from knock-out mice lacking PrP could also propagate the biological properties of several mouse prion strains in serial PMCA reactions.
Because cofactor molecules are required to maintain the infectious conformation of PrP Sc molecules and prion strain properties are potentially encoded by distinctive PrP Sc conformations (1), it was logical to employ a chemically defined serial PMCA system to test whether cofactor molecules might play a role in determining strain properties (by directing PrP Sc conformation). My colleagues and I serially propagated three different mouse prion strains in a substrate mixture containing only recPrP and PE, inoculated the resulting set of PE-PrP Sc molecules into wild type mice, and performed standard strain typing assays to determine whether biological strain properties were maintained or altered by in vitro propagation in the presence of only a single cofactor (Fig. 2) (1). The results showed that propagation in the presence of PE as the sole cofactor caused all three input strains to adapt into a single, apparently novel PE-PrP Sc output strain, as judged by incubation time, neurotropism, and biochemical assays. Notably, the dramatic strain shifts observed could not simply be attributed to the use of recPrP substrate lacking post-translational modifications (although glycosylation may influence the biological properties of some mouse prion strains (31)) because one of the input strains used in the experiment consisted of recPrP Sc made with different cofactors.
Thus, when only one cofactor is available during prion replication, nascent PrP Sc appears to be conformationally constrained, restricting strain diversity (1). These results contrast with those obtained when crude brain homogenate (containing all of the potential endogenous cofactors in brain) is used as substrate, in which case strain diversity was maintained. It is tempting to speculate that prion strain diversity might be encoded by the selective use of different sets of cofactors during the formation of different PrP Sc molecules associated with distinct strains. Consistent with this idea, it has been reported that, under specific experimental conditions, PMCA assays of different hamster and mouse prion strains exhibit varying levels of RNA dependence (32,33). Ultimately, to test this hypothesis rigorously, it will be necessary to isolate and identify additional prion conversion cofactors, especially the components of crude brain homogenate that permit faithful propagation of naturally occurring strains in vitro.

Conclusions and Future Prospects
In summary, studies of infectious prion propagation in vitro have highlighted the important role played by cofactor molecules in this process. Cofactor molecules appear to help maintain the infectious conformation of PrP Sc and may also influence strain properties by facilitating specific PrP Sc conformation(s).
By using chemically defined substrates to produce purified wild type prions with levels of specific infectivity similar to brain-derived prions, these biochemical studies have formally fulfilled all of Koch's postulates for prions as infectious agents without a nucleic acid genome (1). On the other hand, these studies also raise the possibility that infectious prions may not be composed solely of PrP, i.e. host cofactor molecules may be integral informational components that determine the infec-tious conformation of the protein scaffold. This possibility could be tested directly by determining the detailed structure of in vitro-generated PrP Sc molecules, a scenario that has become more feasible due to our ability to produce high titer recombinant prions. It may be particularly instructive to compare the structures of infectious PrP Sc molecules (such as PE-PrP Sc ) with noninfectious samples prepared without cofactor (such as protein-only PrP Sc ) to identify the crucial structural determinants of infectivity. It is important to note in this regard that infectious prions formed with cofactors differ in their ultrastructure and seeding specificity from amyloid fibers (18) and that PrP amyloid fibers lack significant prion infectivity (19 -21).
Other future prospects include the identification of additional cofactors required for the faithful propagation of various prion strains (29,30). The selective susceptibility of neuronal populations in vivo (3,4) and in cultured cells (34) to specific prion strains suggests that strain-specific cofactors may be differentially expressed in various cell types, although other explanations for these phenomena are also possible. Such cofactors could potentially be used as novel therapeutic targets or diagnostic reagents. A prerequisite for the identification of strainspecific cofactors would be the development of a simple and rapid method to confirm the maintenance of specific strain properties during in vitro propagation. Without a doubt, there will be many significant opportunities and challenges ahead in our attempts to understand the unique and remarkable mechanism of prion infectivity.