A simplified recipe for prions

As long ago as the 1960s, it was proposed that the infectious agent of the transmissible spongiform encephalopathies (TSEs), or prion diseases, was composed of protein with no essential nucleic acid component. From the beginning, one basic idea was that the causative agent was a corrupted and pathological form of a host protein that could propagate itself by causing its normal homolog to convert to the pathological form (1). This concept, now commonly known as the “prion hypothesis” (2), has been extremely difficult to prove (3). Fungi were shown to have conceptually analogous protein-only prion pathogens (4), but TSE infectivity eluded clear biochemical characterization for decades. Complicating matters were the inherent difficulties in purifying TSE infectivity, which forms sticky, insoluble aggregates that are not amenable to rigorous purification techniques. Consistent with the prion hypothesis, the major component of infectious isolates was found to be an aberrant, partially protease-resistant form of the host's prion protein (PrP^Sc or PrP-res) (2). However, various other molecules, including nucleic acids, could often be found in even the purest preparations (e.g., refs. 5 and 6). Because there were typically ≈100,000 PrP^Sc molecules per infectious unit, it was difficult to exclude the possibility that other essential components of infectivity might be much less abundant. A standard approach to establishing the essential ingredients of any biological activity is the reconstitution of the activity from defined ingredients in vitro. However, despite the seductive simplicity and apparent plausibility of the prion hypothesis, many laboratories were frustrated for years in their attempts to assemble infectivity in defined cell-free reactions. Recently, striking progress has been made toward this goal, and, as reported by Supattapone and colleagues in this issue of PNAS (7), a defined recipe for producing robust mammalian prions is at hand.

A basic characteristic of an infectious pathogen that discriminates it from simple toxins is the ability to propagate itself in the host. Initial indications that PrP^Sc has self-propagating activity came from cell-free reactions in which purified normal PrP (PrP^C) bound to brain-derived PrP^Sc and converted to a PrP^Sc-like partially protease-resistant state (8). Based on these and other observations, such as the ability of PrP^Sc to form linear amyloid-like fibrils, the mechanism of the conversion was described as seeded nucleated or templated polymerization (8, 9). According to these mechanisms, PrP^Sc acts as a seed or template that progressively recruits PrP^C monomers into ordered polymers and alters their conformation. Striking species and strain specificities were seen with various cell-free conversion reactions, consistent with biological hallmarks of TSE diseases (8). However, the yields of new protease-resistant PrP were modest relative to the input PrP^Sc, and the generation of new infectivity was not demonstrable (10).

Numerous experiments, many unpublished, also have found that recombinant PrP^C expressed in Escherichia coli can be converted spontaneously to aggregated, often fibrillar, partially protease-resistant states that lack infectivity. Recently, however, Legname et al. (11) showed that synthetic truncated PrP fibril preparations can induce TSE disease when inoculated into transgenic mice that vastly overexpress the same truncated PrP construct. The effects were striking, but it was not clear whether the inoculated fibrils themselves contained infectivity or accelerated a preexisting TSE disease that arose spontaneously in the mice due to overexpression of the mutant PrP (12). More importantly, the recombinant PrP fibril preparations were completely innocuous in wild-type mice and therefore appeared to be >10⁸-fold less efficient at instigating TSE disease than bona fide scrapie infectivity; thus, something important was lacking in these “synthetic prions.”

The first cell-free amplification of TSE infectivity that was infectious for wild-type animals was demonstrated by Castilla et al. (13) using their protein misfolding cyclic amplification (PCMA) reaction. This reaction was initiated by diluting scrapie-infected brain homogenate into normal brain homogenate, with the former providing PrP^Sc seeds (at least) and the latter providing PrP^C and other potential cofactors for PrP^Sc amplification. The products were diluted into additional normal brain homogenates for subsequent serial amplification cycles (Fig. 1A). The generation of infectivity by PMCA greatly strengthened the argument against the infectious agent being a conventional pathogen that required replication of an exogenous agent-specific genome and, hence, provided the strongest evidence yet for the existence of mammalian prions. However, the complexity of brain homogenates made it difficult to establish which host molecules comprise the basic prion unit.

Fig. 1.

Models of prion multiplication by PMCA and endocytosis-mediated shearing. (A) PMCA flow chart. (B) Seeded polymerization model for PrP^Sc formation. (C) Hypothetical model for shearing of PrP^Sc aggregates in a cell by endocytosis, a process in which cellular machinery drives the inward budding of membrane-bound vesicles. These vesicles (endosomes) can then recycle their contents (e.g., fragmented PrP^Sc seeds) back to the cell surface or transport them to distal subcellular locations for potential transfer to other cells.

In an effort to dissect the molecular requirements for prion propagation in the PMCA reactions, Deleault et al. (14) showed initially that protease-resistant PrP amplification can be supported by using only PrP^C, purified from brain, and polyanionic molecules such as nucleic acids or sulfated glycosaminoglycans (GAGs) (Fig. 1A). In their most recent work with these reactions, this group has shown that moderately titered scrapie infectivity for wild-type hamsters also was generated (7). Furthermore, they found that this infectivity could be generated spontaneously, that is, without requiring the addition of brain-derived PrP^Sc seeds. This in vitro system may be a model of the spontaneous development of infections in human sporadic Creutzfeldt–Jakob disease patients. A key concern, given the ability of PMCA reactions to amplify minuscule quantities of preexisting PrP^Sc seeds (approximately seven PrP^Sc molecules), was the possibility that inadvertent laboratory contamination of reactions with PrP^Sc could have artifactually initiated the amplification of infectivity. However, the authors took major precautions to avoid such contamination, such as the use of new chemicals and equipment with all procedures performed in a formerly prion-free laboratory.

The study by Deleault et al. (7) provides the first clear evidence that robust mammalian prions can be made by using biochemically defined constituents, namely native PrP^C, copurifying lipids, and synthetic polyanions. The results place narrow constraints on the nature of TSE infectivity and its replication mechanisms. Notably, the findings provide even stronger evidence against the infectious agent being a pathogen that is encoded by its own unique nucleic acid genome. However, important questions remain about non-PrP PMCA reaction components: Are the lipids involved? Do any metals or unidentified organic molecules copurify with PrP^C, and, if so, do they play a role? Do the synthetic polyanions become essential prion components, or do they act purely as catalysts or scaffolds in prion assembly? With the new simplified PMCA system, these questions should now be experimentally accessible.

Although Deleault et al. (7) have demonstrated a simple recipe for prion generation in vitro, the challenge now is to understand how the process occurs in vivo. Both PrP^C and PrP^Sc are usually anchored to cellular membranes by their C-terminal glycophosphatidylinositol (GPI) anchors. The PrP^C-to-PrP^Sc conversion usually occurs on the plasma membrane and/or in endocytic vesicles of infected cells (3). One key requirement of prion propagation by PMCA that should be functionally reproduced in vivo is the need for polyanions, such as single-stranded nucleic acids or GAGs. It is not obvious that single-stranded nucleic acids would be routinely available at the cell surface and in endosomes. However, GAGs are plentiful in these sites of PrP^Sc formation and are known to strongly influence (both positively and negatively) PrP^Sc formation in vitro (3), colocalize with PrP^Sc deposits in vivo (15), and serve as PrP^Sc receptors on the surface of cells (16, 17). Thus, it appears more likely that the in vivo counterparts of the polyanions needed for PMCA prion replication are GAGs rather than nucleic acids. Although the mechanistic function of these molecules remains unclear, these long polymers may interact with multiple PrP molecules at once and thereby help to assemble PrP^C–PrP^Sc complexes and/or prime PrP^C for efficient conformational conversion.

Another important feature of the PMCA method is the application of cyclic sonications, which is thought to expedite the seeded polymerization reaction by fragmenting newly made PrP^Sc polymers to increase the concentration of seeds (Fig. 1B) (18). Indeed, small prion particles offer more infectivity per unit mass than do larger amyloid fibrils (19). A key question is what cellular forces are responsible for fragmenting PrP^Sc polymers to facilitate the replication and spread of the infectious particles within the host. The ability to spread within the nervous system and between individuals appears to be a unique feature of TSE/prion diseases with respect to other neurodegenerative diseases that also involve pathological aggregation of host proteins. Unlike the pathological proteins of Alzheimer's, Huntington's, and Parkinson's diseases, PrP^Sc is usually anchored to membranes via the GPI moiety. One possibility is that the membrane attachment of PrP^Sc aggregates, coupled with the mechanical shearing imposed by membrane endocytosis and ruffling, could fragment PrP^Sc aggregates in vivo (Fig. 1C). In this regard, it is notable that in the transgenic mice expressing only the anchorless PrP (which is not tethered to membranes), huge, poorly fragmented extracellular amyloid plaques predominate and clinical disease is greatly delayed (20). Another possibility is that chaperone proteins help to cleave PrP^Sc aggregates as has been seen with yeast prions (21).

In any case, the development of molecularly defined in vitro protocols for prion propagation should greatly potentiate studies of the structure, replication mechanism, and pathological impacts of mammalian prions. These protocols, in turn, should hasten the development of effective diagnostic and therapeutic strategies.