Progress toward an ultimate proof of the prion hypothesis

Early indications that the infectious agent responsible for scrapie did not contain nucleic acid (1) led to several insightful hypotheses to explain this conundrum (2). Eventually considerable data came to support one of these ideas, dubbed the prion hypothesis, also shown to be applicable to related fatal transmissible spongiform encephalopathies including Creutzfeldt–Jakob and mad cow disease (3). According to the prion hypothesis, the prion protein (PrP) can exist in the normal cellular form (PrP^C), or in an “infectious” prion form (PrP^Sc) that causes disease by converting the cellular form into the prion form. Whereas PrP^C is soluble and easily digested by protease K, PrP^Sc is rich in β-sheets, aggregates into fibrils, and has a protease K-resistant core that forms amyloid. The prion hypothesis has now been extended to explain phenomena involving other proteins (4). Using genetic criteria, Wickner (4) identified the non-Mendelian yeast elements [URE3] (5) and [PSI⁺] (6), as prions of the Ure2 and Sup35 proteins, respectively. A great deal of additional genetic and biochemical evidence now supports this hypothesis (reviewed in refs. 7 and 8). Evidence for other yeast prions (9–11), as well as for [Het-s] of Podospora anserina (12–16), is now emerging. Although the evidence for the prion hypothesis is compelling, a direct demonstration that infectious activity is caused by pure protein when in the prion form has been lacking. Using the [Het-s] prion, Maddelein et al. (17) in a recent issue of PNAS have now come tantalizingly close to achieving that aim.

Support from Yeast

Among the earliest and strongest genetic indications that [URE3] and [PSI⁺] were prion forms of the Ure2 and Sup35 proteins were the findings that their de novo appearances were efficiently induced by overexpression of the URE2 (4) and SUP35 (18) genes, respectively. Wickner (4) interpreted these results as evidence for the prion model because an excess of the protein should increase the probability that the prion form of the protein would appear and start a chain reaction establishing the prion state. In further support of the “protein only” hypothesis, the de novo induction of the [URE3] and [PSI⁺] prions was directly shown to be caused by an excess of the proteins and not an excess of the URE2 or SUP35 DNA or mRNA (19, 20). This finding established that the proteins alone were sufficient to cause the appearances of [URE3] and [PSI⁺]. The induction effect could also be obtained when only a portion of the proteins, called the prion domain, was overexpressed (19, 20). The presence of these same prion domain regions was required for prion propagation (19, 21).

Whereas Ure2 and Sup35 are normally soluble proteins, they form protease K-resistant aggregates upon conversion to the prion state in vivo (19, 22–24). Aggregates isolated from extracts of [PSI⁺] but not [psi⁻] cells seed the rapid conversion of purified soluble Sup35 protein into amyloid-like fibers (24). Such fibers also form slowly from unseeded purified soluble proteins containing the Sup35 or Ure2 prion domains and can themselves seed the rapid conversion of soluble Sup35 or Ure2 into fibers, respectively (25–28). In vitro fiber formation appears to be a good model for prion propagation in vivo.

The Ideal Proof of the Prion Hypothesis

In the ideal experiment to prove the protein-only hypothesis, pure protein (preferably recombinant) would be folded into either a prion or nonprion state in vitro. Equivalent amounts of the two forms of the protein would then be introduced into cells, tissues, or animals lacking the prion, and the efficiencies with which they cause prion appearance would be compared. Because, as described above, increased levels of protein in the nonprion form cause the de novo appearance of the prion, such a comparison is necessary to distinguish infection (in vivo propagation of in vitro-made infectious material), from de novo induction (prion form made in vivo) that might be caused by a local increase in the concentration of the protein in a nonprion form.

From Ideal to Experiment

Although PrP^Sc can seed the conversion of PrP^C to a protease K-resistant form in vitro (29, 42), the creation of infectious particles from pure PrP^C has not been demonstrated yet (30). The first breakthrough in using in vitro-made protein to cause prion appearance in vivo came when Sparrer et al. (31) showed that purified Sup35 prion domain, introduced into yeast cells via a liposome transformation procedure, caused the appearance of the [PSI⁺] prion in 1–2% of transformants. However, this technique did not permit a direct comparison of the activities of soluble and aggregated Sup35 prion domain because, whereas aggregated Sup35 could not be directly loaded into the liposomes, the soluble Sup35 that was encased in the liposome spontaneously formed aggregates before being delivered to the cell. Thus, it remained possible that the aggregated Sup35 delivered to the cells by the liposomes was not in the infectious prion state, and that the [PSI⁺] prion appeared de novo in the cell as a result of a high local concentration of noninfectious Sup35, rather than by the growth of a preexisting infectious particle delivered by the liposome. Indeed this possibility would explain why the liposome-mediated appearance of [PSI⁺] in these experiments was substantially reduced in the absence of [PIN⁺] (31), a factor that enhances the de novo appearance, but not the propagation of [PSI⁺] (11, 32, 33).

Properties of the [Het-s] Prion

The [Het-s] prion of P. anserina promotes programmed death when incompatible cells attempt to fuse (12, 34). Podospora grows as a network of filaments (mycelium) divided into articles by incomplete walls that permit the sharing of cytoplasm. When two colonies of P. anserina grow toward each other, their mycelia fuse to form a heterokaryon with mixed cytoplasm. Such fusions spread fungal viruses and prions very rapidly. Podospora has evolved a system to protect itself from the spread of viruses by preventing the fusion of those partners that are not very similar genetically and are therefore likely to carry different viruses (35). If the nuclei of the fusing mycelia differ in any of a number of het loci, the articles at the point of the fusion die, creating an easily visible barrier (“barrage”) to further cell fusion.

The het-s gene, which has two alleles, het-s and het-S (Table 1), is part of this “heterokaryon incompatibility” system. The het-s and het-S alleles encode proteins of 289 aa that differ at 13 residues (15, 36). The protein encoded by het-s can exist in a prion, HET-s, or nonprion, HET-s*, state (Table 1). [Het-s] cells, which bear the prion form of the protein, are incompatible with cells carrying the het-S allele. [Het-s*] (nonprion) mycelia, or mycelia that are missing the HET-s/S protein altogether, are compatible with either [Het-S] or [Het-s] mycelia (12). As expected of a prion, when [Het-s] and [Het-s*] mycelia fuse, the resulting cytoplasmic mixing causes the rapid spread of [Het-s] into the formerly prion-free portion of the mycelium. Also, overexpression of the het-s allele in prion-free mycelia induces the de novo appearance of [Het-s] that, because of efficient spreading of the prion, results in 100% of mycelia becoming [Het-s] within 3 days (12).

Table 1.

Which Het-s is it?

Alleles
het-s	Encodes a protein that can exist in a prion or nonprion state.
het-S	Encodes a protein that can never exist as a prion.
Proteins
HET-s	Prion form of protein encoded by het-s in P. anserina, incompatible with HET-S; also the name of recombinant protein synthesized in Escherichia coli from a het-s gene, whether in the soluble, aggregated or denatured form.
HET-s*	Nonprion form of protein encoded by het-s in P. anserina, compatible with HET-s or HET-S.
HET-S	Protein encoded by het-S, incompatible with HET-s.
Phenotypes
[Het-s]	Incompatible with [Het-S]; converts [Het-s*] to [Het-s] by cytoplasmic mixing.
[Het-s*]	Compatible with [Het-S] and [Het-s]; converted to [Het-s] by cytoplasmic mixing with [Het-s].
[Het-S]	Incompatible with [Het-s].

The biochemical properties of the HET-s prion protein have now been shown to be similar to those of PrP^Sc, [URE3] and [PSI⁺]. HET-s in crude lysates is more resistant to protease K than is HET-s* (12). Purified recombinant HET-s aggregates into high molecular weight amyloid-like fibrils over time and these fibrils seed the more rapid formation of aggregates when added to a solution of freshly renatured soluble HET-s protein (16). HET-s aggregates are also detected in vivo, but only when het-s is overexpressed, suggesting that HET-s prion aggregates are normally small and possibly preferentially degraded (14). Maddelein et al. (17) have now shown that HET-s aggregates isolated from Podospora that are overexpressing het-s are capable of seeding soluble HET-s to fibers in vitro.

Defining the HET-s Prion Domain

The Ure2 and Sup35 prion domains are characterized by unusually high levels of Gln and Asn, and this characteristic has been used to identify other prions in yeast (9, 10). In contrast, HET-s and PrP do not contain a Gln- or Asn-rich region, and the results of deletion experiments designed to define the HET-s prion domain have been difficult to interpret (13). The sequence of a 7-kDa protease K-resistant core of in vitro-made HET-s fibers, which Maddelein et al. (17) now show to be able to cause prion appearance when inserted into mycelia (see below), should help identify the elusive HET-s prion domain.

Using HET-s to Prove the Prion Hypothesis

Maddelein et al. (17) have successfully used microprojectile bombardment to introduce different forms of HET-s protein into P. anserina mycelia (Fig. 1). The structure of mycelia, which allows cytoplasmic mixing between the articles, ensures that prion infection or de novo appearance will spread rapidly. The authors (17) show that the insertion of fibers, made in vitro from renatured recombinant HET-s, into [Het-s*] mycelia, causes the efficient appearance of the [Het-s] prion (Fig. 1). In addition, they find that this activity is maintained by a protease K-resistant core of the in vitro-made fibers composed of 7-kDa fragments, the putative prion domain. When the same bombardment technique was used to introduce soluble (freshly renatured) HET-s into [Het-s*] mycelia, the increase in prion appearance was barely above background levels. This was true even when the amount of soluble HET-s layered on the cells was 100-fold higher than the amount of fibers necessary to cause prion appearance in 75% of bombarded mycelia. Although there is no direct evidence that as much soluble protein as fibrillar protein was actually introduced into mycelia (HET-s in the fibrillar state may be more efficiently inserted into cells by bombardment, and/or with less damage to the molecules), the drastic difference in the efficiency of prion appearance suggests that insertion of fibrilar HET-s is infectious, whereas the insertion of soluble HET-s is not.

Figure 1.

Appearance of [Het-s] after microprojectile bombardment of different forms of recombinant HET-s into [Het-s*] prion-free mycelia. Starting with purified His-tagged HET-s isolated from E. coli under denaturing conditions (Upper), Maddelein et al. (16) produced freshly renatured HET-s that was soluble and monomeric. After 72 h, the renatured HET-s spontaneously formed amyloid-like fibers that were separated from the remaining soluble protein by centrifugation. Limited protease K digestion of these fibers resulted in fibers with a 3- to 5-fold reduction in diameter, composed of 7-kDa fragments. Amorphous aggregates were formed by boiling soluble HET-s. Each of these forms of HET-s was layered on mycelia and then bombarded with tungsten particles after evaporation. Bombardment was directly shown to insert fluorescein-labeled undigested HET-s fibers into mycelia. Although the insertion of the other HET-s forms is shown in the cartoon, their actual insertion is an assumption that has not been demonstrated experimentally. Mycelia were scored for the presence of the [Het-s] prion 2 days after the bombardment by testing samples of them for compatibility with [Het-S].

What Is Next?

Maddelein et al.'s experiment shows for the first time that distinct in vitro-made fibers (the complete HET-s and the protease K-resistant core) can cause in vivo prion appearance (17). The ability to introduce different in vitro-made fibers into cells and the intriguing phenomenon of prion “strains” together may provide the ingredients to prove unambiguously that in vitro-made aggregates are truly infectious. Clinically distinct heritable versions of prion diseases (called strains) can occur within the same inbred mammalian line even though the PrP gene is identical in these inbred animals (37, 38). Curiously, the different disease strains appear to be caused by distinct conformations of the PrP, and these conformation differences seem to be maintained during in vitro propagation (39). Different strains of the yeast [PSI⁺] and [URE3] prions with distinct heritable phenotypes have been isolated in genetically identical cells after overproduction of Sup35 and Ure2, respectively (20, 40), and in vitro-made Sup35 fibers with distinct conformations suggestive of different [PSI⁺] strains have been described (41). If strains of the [Het-s] prion can be identified, it should be possible to ask whether strain differences are determined solely by different prion conformations. To do this, distinct fibers produced in vitro from pure protein and corresponding to different strains would be inserted into prion-free cells by using the procedure established by Maddelein et al. (17). Because infection with the prion particles of one strain should propagate only that strain, whereas de novo induction would cause an array of different strains, the appearance of strain-specific phenotypes corresponding to the different fiber types would lay to rest any doubts that the prion appearance could have been caused by de novo induction rather than infection.

Maddelein et al. (17) have produced a pivotal paper that has brought us to the brink of proving one of the most disputed hypotheses.