Whether a protein can transmit disease in mammals has been an open question for some time. The latest test of this idea provides some strong evidence in favour, but is unlikely to end the debate.
Studies of the infectious agent responsible for the transmissible spongiform encephalopathies have spanned more than six decades. But the slow course of these diseases, and experimental difficulties in their study, has left the nature of this agent in doubt. Now, a report in Science by Legname et al.1 represents a major step towards proving that a protein is the only essential element of the infectious agent.
Transmissible spongiform encephalopathies (TSEs) are uniformly fatal neuro-degenerative diseases, and include scrapie in sheep, Creutzfeldt–Jakob disease (CJD) in humans, bovine spongiform encephalopathy (BSE) and the latter condition transmitted to humans (variant CJD). The discovery that the infectious agent that causes scrapie is far more radiation-resistant than known viruses or bacteria — which rely on the very radiation-sensitive DNA or RNA — led to the suggestion2 that this agent is a protein (without a required nucleic acid). Proteins are far less sensitive to radiation damage than are DNA and RNA. With the genetic3 and biochemical4,5,6,7 identification of PrP, a protein essential for the propagation and infectivity of TSEs8, the ‘protein-only’ hypothesis2,9 took concrete form (‘prion’ means ‘infectious protein’).
The protein PrP is a cell-surface protein that is normally sensitive to protein-degrading enzymes (proteases) and has a turnover time of a few hours. In TSE-infected cells, by contrast, PrP is partly protease-resistant and quite stable (reviewed in ref. 10). Because it is the main component of purified infectious material4, this protein was itself proposed to be the infectious agent in TSEs9. Theoretically, the normal form of the protein, PrPC (where ‘C’ denotes ‘cellular’), converts rarely into an abnormal form (PrPSc for ‘scrapie’) that then catalyses the conversion of other normal molecules into the same abnormal form. The accumulated PrPSc then somehow causes neurodegeneration.
Proving the protein-only theory has been difficult, however. For example, because the infectious material is highly aggregated and heterogeneous in size, even the best preparations need 105 PrP molecules to infect one mouse. So it has not yet been possible to prove that PrP is sufficient for infectivity by purifying the protein from infected brains.
Another approach has taken its cue from the ‘amyloid’ nature of PrPSc. This abnormal form — unlike PrPC — contains a high percentage of β-sheets, one of the two main types of structural feature in proteins. Furthermore, PrPSc is filamentous and protease-resistant and shows birefringence on staining with the dye Congo Red. These four properties define amyloid, and gross amyloid deposits of PrP are seen in the brains of many CJD patients. Full-length, mature, soluble PrPC has been converted to β-sheet-rich forms in vitro — but this material was not infectious.
There has been success on one front: the basic mechanism postulated for the TSE protein-only model has been demonstrated biochemically. PrPSc purified from infected brains can induce the conversion in vitro of small amounts of PrPC to the PrPSc form, in a reaction that shows all the specificity of the TSE disease process11. However, the limited extent of this reaction in vitro has so far made it impossible to demonstrate the generation of new infectivity, over that present in the input PrPSc.
How can one discover whether a protein is infectious? In budding yeast, clear evidence was provided by the genetic properties of the non-chromosomal genes [URE3] and [PSI+], which represent the prion forms of the Ure2 and Sup35 proteins, respectively12. The same genetic approaches served to demonstrate that [PIN+] (also in budding yeast13) and [Het-s] (in the filamentous fungus Podospora anserina)14 represent the prion forms of the Rnq1 and HET-s proteins, respectively. But these methods are not feasible in mammals. However, [Het-s] can be efficiently transmitted to uninfected P. anserina cells by introducing amyloid fibrils of recombinant HET-s protein (not by introducing the soluble form or other aggregates of this protein). These results confirm that [Het-s] is a prion, and show that self-propagating amyloid is the infectious material15. Recently, similar results have been achieved in the [PSI+] system16,17.
In the latest work, Legname et al.1 engineered mice so that they lacked the normal PrP protein, instead producing a fragment of PrP (amino acids 89–230) at 16 times the usual levels. The idea was that this fragment might be more easily triggered than the full-length protein to convert to the abnormal form. The authors then prepared amyloid filaments of PrP89–230 in vitro and inoculated the filaments into the animals' brains. They found that these mice died with a scrapie-like syndrome after about 550 days, whereas mice inoculated with a control buffer survived at least 620 days. Crucially, brain tissue from the amyloid-inoculated mice was then highly infectious when inoculated into normal mice (Fig. 1).
Unfortunately, the authors do not report whether the brains of the buffer-inoculated mice are also infectious. If they are not, then the results show that amyloid filaments composed of this PrP fragment can initiate an infectious process — apparently putting to rest the ‘protein-only’ dispute. If the brains of buffer-inoculated animals are infectious, however, then one can conclude that overproduction of the PrP fragment results in prion formation de novo (perhaps accelerated by the injection of amyloid). The latter conclusion would also be a strong argument in favour of the protein-only model, as this is one of the genetic criteria for prions that were established by studies in yeast12.
One other control would have been useful. The authors used amyloid filaments of the PrP fragment for infection, but they argue, paradoxically, that it is not amyloid that is the infectious form. For this reason, among others, injecting the soluble PrP fragment, or a non-specific aggregate, would have been helpful.
Nonetheless, this work adds to the already considerable evidence for the protein-only model. It is essential that the transgenic mice be made widely available, as this infectivity assay will be very useful in defining precisely the structure of the infectious material.
About this article
Nature Reviews Genetics (2005)