Creationism and Evolutionism in Prions

Abstract

This Commentary highlights the article by Ghaemmaghami et al, who investigated the nature of synthetic prion transformation and demonstrated that these can assume multiple intermediate conformations before converting into one conformation optimized for in vivo propagation.

See related article on page 866.

Prion diseases, earlier referred to as transmissible spongiform encephalopathies, are rare conditions that have nonetheless drawn considerable attention because they are transmissible, affect both humans and animals, and are remarkably heterogeneous in their phenotypic and molecular features. The transmissibility to several mammalian species makes prion diseases hazardous to public health, owing to their propensity to spread not only between humans but also from animals to humans. Transmissibility uniquely confers on prion diseases three etiologies: sporadic, genetically determined, and acquired by infection.

Given these challenging features, it is not surprising that the pathogenesis of prion diseases has been controversial. The prion hypothesis, formally proposed by Stanley Prusiner in 1982, posits that the agent that causes the disease and makes it transmissible consists mostly, if not exclusively, of a protein that he named proteinaceous infectious particle or prion.¹ The prion, now also identified as scrapie prion protein (PrP^Sc), is derived from a normal protein well represented in mammals (cellular prion protein, or PrP^C) through a conformational transition whereby PrP^Sc interacts with and templates PrP^C, forcing it to adopt the PrP^Sc abnormal conformation. Initially, the prion hypothesis was intensely criticized by the proponents of the viral etiology, but it has since been definitively confirmed; a series of experiments has demonstrated that a typical prion disease can be engendered in wild-type animals after inoculation of PrP^Sc generated in vitro from recombinant PrP.^2–5

Strains

It has been known for several decades that prions are formed in distinct kinds, called strains, that can be experimentally distinguished by the length of the incubation time and the pattern of the brain lesions caused in the host after transmission.⁶ Recently, prion strains have revealed more complex and dynamic features.^6–8 Strain variation is commonly related to differences in the amino acid sequence between the exogenous PrP^Sc and the PrP^C of the host; such variation is commonly found in transmission between species, but can also occur within the same species when PrP^C heterogeneity is due to polymorphism in the PrP gene.⁹ Even more revealing mechanisms of strain diversity play out when PrP^Sc and PrP^C share amino acid sequence and the strain diversity is attributable merely to conformational differences.¹⁰

Prion strains are currently viewed as combinations or ensembles of substrains (ie, a constellation of different kinds of PrP^Sc characterized by distinct conformations and, possibly, other physicochemical attributes), one of which is dominant and imparts its characteristics to the strain as a whole.⁷ When on inoculation such a strain cannot readily template the host’s PrP^C (owing to mismatch in amino acid sequence, or to only limited conformational compatibility), it must undergo a process of adaptation for the PrP^C-to-PrP^Sc conversion to occur and become more efficient.^6–8 This adaptation generally requires two or more serial transmissions through the hosts, resulting in different outcomes, depending on the degree of heterogeneity between the exogenous PrP^Sc and the host PrP^C. The presence in the strain of substrain components capable of templating the host PrP^C may promote the emergence of a new dominant substrain. If none of the substrains is compatible with the host’s PrP^C, the strain may change drastically, or mutate, resulting in a new strain. The process of strain formation and selection becomes fairly stereotyped after serial passages in syngeneic hosts as the strain reaches the optimal combination of substrain components to cause disease. This selection is commonly revealed by the progressive reduction of the incubation time required to reach the steady state.

Synthetic Strains

In 2004, Prusiner and colleagues² published the first evidence that infectious mammalian prions could be generated in vitro by refolding recombinant PrP into amyloid conformations. An article by Ghaemmaghami et al¹¹ in this issue of The American Journal of Pathology reports the latest study from the Prusiner research group, building on earlier research into the design and construction of synthetic prion strains.¹² The study focuses on the process of synthetic strain adaptation after multiple serial transmissions to mice. Using in vitro manipulations, Ghaemmaghami et al¹¹ generated four synthetic prion preparations with amyloid-like characteristics by inducing refolding of purified recombinant mouse PrP. The four preparations (identified as MoSP5, MoSP6, MoSP7, and MoSP9) were obtained with different procedures, and display distinct ultrastructural and conformational stability features, and can therefore be viewed as distinct prion strains.

After the first inoculation or passage (P1) to transgenic mice expressing four to eight times the normal amount of full-length mouse PrP^C, each preparation caused a prion disease after average incubation times ranging from 500 to nearly 800 days, which is quite lengthy incubation, compared with the 50-day incubation required of the native prion strain RML (Figure 1).¹² Six PrP^Sc strains were recovered at P1 from the transgenic mice inoculated with the four MoSPs; these strains could be distinguished on the basis of PrP^Sc electrophoretic banding patterns, incubation time, conformational stability, and amyloid seeding capacity. The banding patterns observed were similar to those described in human PrP^Sc and were also similarly identified.^13,14 As for human PrP^Sc, types 1 and 2 were defined by electrophoretic mobility at 21 or 19 kDa, respectively, of the unglycosylated (or lowest) band of PrP^Sc after protease treatment.^13,14

Figure 1.

In-pathway intermediate states and final forms of the four synthetic prion strains, along with changes in incubation time, banding pattern, and conformational stabilities after serial passages in transgenic mice. MoSP7 generated all three strain types (indicated as a, b, and c), which traversed different intermediate states before reaching the optimal condition. Strain MoSP7c split into two distinct strains (1 and 2) at the second passage. Strain characteristics in the adapted state are highlighted in red.

Data are summarized from Figure 4 of Ghaemmaghami et al.¹¹

At P1, Ghaemmaghami et al¹¹ recovered PrP^Sc type 2 in MoSP5, MoSP6, and MoSP7a, type 1 in MoSP7b, and type 1/2 (a mixture of the two types) in MoSP7c and MoSP9. Of note, MoSP7 generated all three banding patterns (Figure 1). After P1, the strains took seemingly unpredictable paths through the various passages, generating intermediate forms that often differed in the type of banding pattern, incubation time, and stability. Interestingly, in MoSP7b, the banding pattern that was type 1 at P1 switched to and remained type 2 through two more passages, and then reverted to type 1 at P4; although both MoSP7c and MoSP9 were type 1/2 at P1, only MoSP7c split into types 1 and 2. Nonetheless, all of the P1 prion strains eventually converged into an apparently common strain that was type 1, exhibiting similar low conformational stabilities and amyloid seeding activities (although some differences remained for incubation time).

On the basis of these findings, the authors conclude that prion adaptation is not a straight process.¹¹ Rather, it appears to rely on a trial-and-error strategy, in which errors are progressively corrected by Darwinian selection until the most pathogenic strain is ultimately selected.⁶ The study by Ghaemmaghami et al¹¹ suggests additional considerations concerning prion strain formation and evolution.

It should be emphasized that the various pathways leading to the selection of the optimal strain were individually reproduced in sets of the same transgenic mice, with data showing relatively narrow ranges (see Figure 4 of Ghaemmaghami et al¹¹). Furthermore, the selection process could be replicated in cells, at least for one synthetic strain. Thus, the selection pathways of the individual strains, although quite different, were reproducible (and therefore not random), presumably reflecting the presence of consistent constraints in what the authors call a conformational landscape.¹¹

It is widely accepted that the length of incubation time (which reflects the pathogenicity) and conformational stability are directly related. This relationship is expected, given that unstable strains are more likely to fragment, which makes the interaction with the PrP^C substrate more efficient. On the whole, this observation is confirmed by the study of Ghaemmaghami et al.¹¹ However, as the authors also point out, these two characteristics became inversely related in some of the passages (Figure 1). For example, in the P2 conversion, in which PrP^Sc type 1 of MoSP7b and type 1/2 of MoSP7c are each converted to PrP^Sc type 2, the incubation time was drastically reduced but the stability actually increased, compared with P1. This finding suggests that pathogenicity and stability may be uncoupled under certain conditions. In both MoSP7b and MoSP7c, the newly formed and more stable PrP^Sc was type 2, but it remains to be determined whether this result was coincidental or indicates that the uncoupling of pathogenicity and stability is a characteristic of the mouse PrP^Sc type 2 strain.

The study by Ghaemmaghami et al¹¹ also leaves a few issues to be addressed in future research. For instance, it would be interesting to extend the characterization of the strains to include the amount of total PrP^Sc and the size distribution of the PrP^Sc aggregates. These additional data, combined with those already collected, would provide a more complete picture of the physicochemical characteristics that the strain adopts while evolving toward the final isoform. These data would also expand the correlation of the incubation time and stability to include the size of the aggregates and the role, if any, played by the amount of PrP^Sc in determining the increasing pathogenicity through the serial passages. Infectivity has been shown to correlate with aggregate size,¹⁵ and so prion evolution might include shifts in aggregate size. Finally, it is surprising that no apparent variation was detected in the type, topography, or severity of the brain pathology throughout the various passages, despite significant variations in the banding pattern, stability, or both. Perhaps detailed analysis of lesion topography and severity may settle this issue.

Adaptation of synthetic prions through serial passages has also been examined in hamsters by Baskakov and colleagues.³ Although they did not study their panel of synthetic prions in passages that are parallel to or easy to compare with those of Ghaemmaghami et al,¹¹ interesting differences emerge between the two sets of studies. The synthetic prions developed by Ghaemmaghami et al¹¹ seemingly displayed fairly conventional banding patterns even at P1 and evolved into stabilized strains with incubation times similar to those of native adapted strains. However, Baskakov and colleagues³ observed atypical banding patterns at early passages, and the incubation times remained extended even after strain stabilization; furthermore, in their experiments, animals were asymptomatic at P1. Because, after careful search, no authentic PrP^Sc component was detected in their synthetic prion preparation, Baskakov and colleagues³ attributed the lack of clinical disease at P1 and the gradual appearance of typical PrP^Sc to the strong conformational barrier (ie, the conformational constraints encountered by the synthetic strain in templating the host PrP^C). At variance with the study of Ghaemmaghami et al,¹¹ Baskakov and colleagues exposed their hamster recombinant PrP amyloid fibers to heat (ie, annealing). Furthermore, they used wild-type animals, not animals overexpressing PrP^C. These methodological differences might explain the variation in the adaptation process between the sets of strains developed by the two research groups.

Native Human Strains

How applicable are the data of Ghaemmaghami et al¹¹ (as well as other current notions on strains) to human prion diseases and to other neurodegenerative conditions? Of the three etiological forms of human prion diseases (ie, sporadic, inherited, and acquired by infection), current data on prion strains most readily apply to the form acquired by infection, especially to the best-known infection, variant Creutzfeldt–Jakob disease (vCJD), which is acquired from the consumption of meat or byproducts from prion-contaminated cattle. However, nearly 90% of human prion diseases are sporadic (Table 1), and the nature and selection process of the prion strains associated with the sporadic form are poorly defined, in part because the mode of formation of PrP^Sc in the sporadic form remains enigmatic, owing to the lack (at least until very recently) of robust experimental models of sporadic prion disease.¹⁶

Table 1.

Classification of Human Prion Diseases

Phenotype, by form or etiology
Sporadic
Sporadic Creutzfeldt–Jakob disease (sCJD)
sCJDMM(MV)1∗
sCJDMM2
sCJDVV1
sCJDVV2
sCJDMV2
Fatal insomnia
Variably protease sensitive prionopathy (VPSPr)
Familial
Familial Creutzfeldt–Jakob disease (fCJD)
Fatal familial insomnia (FFI)
Gerstmann–Sträussler–Scheinker disease (GSS)
Acquired by infection
Variant CJD
Kuru
Iatrogenic CJD

Forms, types, and subtypes have been simplified for clarity.

^∗The letters MM, MV, and VV refer to the genotype at codon 129 of the prion protein gene; the numbers 1 and 2 refer to the PrP^Sc banding pattern or type.

Human sporadic prion diseases comprise three types (CJD, fatal familial insomnia, and variably protease sensitive prionopathy), and include five well-defined subtypes of sporadic CJD (sCJD) (Table 1). The five sCJD subtypes are consistently associated with distinct clinical and histopathological phenotypes, which are thought to be determined by the pairing of the PrP genotype at codon 129 and the features of the associated strain.^10,17 Human codon 129 is the site of a common methionine/valine (M/V) polymorphism that exchanges the amino acid sequence of the PrP^C encoded by each PrP allele, thereby generating three PrP^C isoforms: 129MM, 129MV, and 129VV. The major strain characteristic codistributing with the phenotype is the banding pattern that distinguishes the PrP^Sc in types 1 and 2 in precisely the way described by Ghaemmaghami et al.¹¹ Altogether, at least five distinct strains are thought to be associated with sCJD.¹⁰

Strain formation and selection in sporadic prion diseases poses several new challenges. Most intriguing are the mechanisms of the initial PrP^C-to-PrP^Sc conversion. Compelling evidence points to the failure (likely to be age-related) of the quality-control complex known as the proteostasis network to clear misfolded PrP^Sc-like species that are produced under normal cellular conditions as a result of intrinsic errors in protein biosynthesis.¹⁷ The initial PrP^Sc would be formed in a limited region of the brain, or even in a single cell; from the site of formation, the PrP^Sc would then propagate to other brain regions, where it would cause tissue damage.¹⁰

Little is known about strain formation when PrP^Sc, as described above, is formed de novo in the absence of an exogenous strain to template the PrP^C. It is unclear whether the initial conversions generate substrains that then undergo a selection process culminating in the rise of a dominant isoform, as happens after inoculation of an exogenous strain, or whether these events happen before or during propagation.

De novo prion strain formation and evolution must also accommodate the repetitiveness of the individual sporadic phenotypes and, thus, of their associated strains. The consistency of individual strains indicates that the de novo PrP^Sc isoforms follow a relatively limited number of formation and evolution pathways. This view is fully consistent with the findings of Ghaemmaghami et al.¹¹

The sCJD subtypes are often associated with both PrP^Sc types 1 and 2 (what Ghaemmaghami et al¹¹ call type 1/2). Each strain type appears to be fully mature and competent to determine its own specific phenotype, a condition that leads to the formation of phenotypes with hybrid characteristics. Furthermore, when the two strain types coexist in the same brain region, they reciprocally affect some of their characteristics (eg, their conformational stability). The co-occurrence of the two strain types raises the question of whether one results from the mutation of the other or whether they are formed de novo concurrently.

Finally, if strains determine the phenotype, then they must not form randomly in sCJD, because the prevalences of the CJD phenotypes are markedly different. Although sCJDMM(MV)1 accounts for approximately 60% of sporadic cases, sCJDVV1 is observed in only approximately 2% of cases; other subtypes have intermediate prevalences.¹⁷ Whether this lopsided strain occurrence is established at the time of the initial formation of PrP^Sc or whether it represents different selection patterns is currently unknown. Ghaemmaghami et al¹¹ observed co-occurrence of types 1 and 2 at P1, but the two types were segregated in subsequent passages (Figure 1). Studies now in progress with serial passages of the human 1/2 PrP^Sc to human PrP^C-expressing transgenic mice may show whether a similar outcome also applies to human prion 1/2 co-occurrence (I. Cali and Q. Kong, personal communication). This kind of experiment may also help clarify whether PrP^Sc species associated with sCJD are authentic strains or represent intermediate states.

It has recently been observed that the prion principle (ie, the recruitment of a normal protein into a conformationally pathogenic isoform) also applies to several other neurodegenerative diseases.¹⁸ In fact, Alzheimer disease and tau protein–related diseases (or some of the basic histopathological features of these diseases) have been transmitted to receptive hosts. Similar findings, albeit preliminary, have also been obtained in other neurodegenerative diseases, including Parkinson disease and amyotrophic lateral sclerosis. Strain presence is one of the issues raised by these findings.¹⁰ More than 10 years ago, it was reported that familial forms of Alzheimer disease with distinct disease phenotypes are associated with distinct electrophoretic banding patterns of amyloid β, the primary pathogenic peptide in Alzheimer disease.¹⁹ Pathogenic presenilin-1 mutations, which give rise to a malignant phenotype, were associated with significantly more amyloid β fragments truncated at the N-terminus than full-length amyloid β; the opposite ratio was observed in other mutations with typical phenotypes and in sporadic Alzheimer disease. This finding suggests that the cleavage of amyloid β at different sites, which generates distinct banding patterns, is associated with distinct phenotypes, a scenario similar to that leading to the formation of PrP^Sc types 1 and 2 in prion diseases. Therefore, it would not be surprising if strains are shown to exist and play a role in Alzheimer disease and other nonprion neurodegenerative diseases.

The use of human synthetic strains in experiments such as those conducted by Ghaemmaghami et al¹¹ may help clarify some of these issues. In addition, animal models of spontaneous human prion disease and cells capable of propagating human prion strains must be developed, and the application of prion technologies to other neurodegenerative diseases must be aggressively pursued.

Conclusions

The study by Ghaemmaghami et al¹¹ focuses on the evolutionary process that prions undergo after serial passages in syngeneic hosts to reach a conformer that is stable and most pathogenic and that fulfills the characteristics of the fully adapted prion strain. A major merit of their study is the use of several synthetic prion preparations that are purposely heterogeneous but are all capable of causing a prion disease in the same host. Theoretically, synthetic prions are powerful tools to dissect conformational and other features that are related to banding pattern, infectivity and pathogenicity. For example, at P1, Ghaemmaghami et al¹¹ generated each of the three PrP^Sc banding patterns that seemingly are similar to those observed in sCJD. Detailed conformational analyses of the synthetic prion developed in their study (and of future, more efficient ones) may reveal the features that lead to evolution into PrP^Sc type 1, type 2, or both types at the same time. More importantly, such analyses may reveal the conditions that make a prion highly pathogenic or harmless, and whether these conditions have therapeutic applications.