The results of Bruce et al.2 and Hill et al.3 underline the need to gain a better understanding of the nature of the agents responsible for transmissible spongiform encephalopathies (TSEs). The most favoured view is that the agents are composed solely of an altered form of a host-encoded protein known as PrP and lack a foreign nucleic acid; this is the ‘protein only’, or prion, hypothesis4.

There are well-documented instances of TSE agents changing their characteristics, or phenotype, upon passage (repeated transmission through experimental animals). Indeed, Hill et al. report a change in fragment size upon transmission of BSE and vCJD to their transgenic mice. However, strain phenotypes can remain stable12; thus the Edinburgh group of Bruce and colleagues have reported that different strains adapted to, and repeatedly passaged in, a single line of inbred mice retain their distinguishing features even though they must be composed of PrP protein with the same amino-acid sequence6. In addition, a single strain, as illustrated by BSE, can retain its phenotype even when passaged through different hosts. In these circumstances, then, strain phenotype is maintained even though the agent is composed of PrPs of different amino-acid sequences, raising the question of what determines strain phenotype — clearly, in these cases, it is not the primary structure of the PrP protein.

Supporters of the ‘protein only’ hypothesis argue that the small number of different strains that have been convincingly demonstrated so far can be explained by different conformations of the PrP protein, perhaps in combination with modifications such as glycosylation. Such differences of course need to ‘breed true’ upon passage.

On the other hand, opponents of the ‘protein only’ hypothesis point to the potentially large number of TSE strains (the Edinburgh group claim to have evidence for around 20) and argue that an unimaginable plasticity would be required for these to be accommodated by PrP conformation alone. They suggest that a more likely explanation is that, in addition to PrP, an informational molecular component is present in the infectious agent. Whatever the nature of the agent, our understanding of TSE biology is evidently incomplete.