Prions show their metal

Transcription

1 Prions show their metal Ian Jones describes the evidence that increasingly links prion proteins and copper ions. Could a defect in the metabolism of this simple metal be at the heart of 'mad cow' disease? The preoccupation with prion proteins in recent years has related almost entirely to their role as key to the infectious agents that cause the transmissible spongiform encephalopathies (TSEs) such as BSE and nvcjd. Questions surrounding the role of prion protein in these diseases are many and varied: how does the infectious agent arise, how is it transmitted, how does it propagate and how does it survive the processes that would ordinarily inactivate any other infectious agent? Addressing these areas is essential if we are to understand and control the occurrence and spread of spongiform diseases. However, the focus on the pathogenic form of the agent has obscured, to some degree, scientific effort aimed at addressing a fundamental yet still unanswered question: what is the function of the normal or wild type non-pathogenic form of prion protein? Recently, and largely as a result of being able to produce large amounts of wild type prion protein in vitro through recombinant expression systems, evidence has been accumulating about what that function might be. Causes and effects The background to prion disease is well known. Records of scrapie, an endemic disease of sheep, date back more than 200 years, while many other animals are known to have experienced similar neurodegenerative diseases. Diseases with a similar pathology arise sporadically within the human population and include Creutzfelt-Jakob disease (CJD) and Gerstmann-Stäussler-Scheinker syndrome (GSS). In some cases such as Fatal Familial Insomnia (FFI), the TSE tracks through families, clearly indicating a strong association with a host gene. The introduction of infected material into a susceptible mammal, usually through consumption, leads to further infection, making the TSEs unique in being both genetic and transmissible diseases. The finding that the scrapie agent could cause a similar disease in rodents led to the biochemical characterisation of the infectious agent and the introduction of the term 'prion' by Stanley Prusiner in This name recognised the fact that the characteristics of the agent did not allow it to be classified with any known infectious agent at the time. Purifying the infectious scrapie agent led to the prion amino acid sequence and, subsequently, to the identity of the encoding gene, Prnp. Interestingly, Prnp was found to be a normal host gene and produces the wild type prion protein (PrP c ), as well as the pathogenic form (PrP Sc ) in infected animals. Both prion forms are now known to share the same amino acid sequence. Subsequent analysis has confirmed that the sequence of Prnp is linked to the characteristics of the disease such as predisposition (in the case of the hereditary forms), incubation time and the ability to transmit to other species. Importantly, work by neuroscientist Charles Weissmann in Zurich and others on 'knockout' mice lacking Prnp shows that PrP c, in addition to a PrP Sc, is necessary for disease progression. The current prion hypothesis suggests that a change in prion protein conformation causes pathogenesis. The wild type prion protein is thought to be benign, but the aberrant form is capable of direct or indirect neurotoxicity, and capable of stimulating more wild type prion to undergo conformational change. The prion protein

2 Researchers described the nucleotide sequence of the prion gene in the mid-1980s - as a protein of just over 250 amino acids that had no homology with any sequence known at the time. The prion sequence is still considered unique, although researchers have recently described a related molecule, Doppel, which may indicate an ancient gene duplication event during the evolution of prion proteins. The prion sequence indicates that this protein is destined for the cell surface where it is retained by a glycosylinositol phospholipid (GPI) anchor. The wild type molecule at the cell surface cycles from the membrane to sacs within the cell interior called endocytic vesicles where it may be degraded (see Box). Experiments suggest that the wild type molecule transforms to the pathogenic form in the vesicles, either sporadically or after interaction with ingested PrP Sc. Jekyll and Hyde prions The wild type prion (PrP c ) is found in the secretory pathway of cells expressing the protein (1) and moves to the plasma membrane where it is anchored by its GPI tail (2). There it may bind to an extracellular ligand (possibly copper) (3) before being cycled from the membrane into endocytic vesicles (4). At some point its cargo is released and the protein either passes to the lysozome for degradation or back to surface for another round of ligand binding. In this respect it resembles many other membrane-resident proteins. The pathogenic form (PrP Sc ) also finds its way to endocytic vesicles where it co-opts some of the wild type form to become pathogenic (5). PrP Sc is resistant to degradation, a hallmark of the infectious form, so accumulates. Neurotoxicity is probably linked to the conversion event itself, perhaps through its interference with normal PrP c turnover, because there is considerable evidence to show that the accrued PrP Sc is not inherently toxic. Examining the prion protein sequence for structural clues to its function indicates an unusually 'split' molecule (see below). The carboxyl terminal half of the protein contains amino acid sequences of mixed character, with early secondary structure predications indicating at least two long stretches of a-helix. The presence of these a-helices was subsequently proved correct; a protein fragment encoding the C-terminal half of the protein provided the first three-dimensional structure of any prion related sequence by NMR spectroscopy in The domain is a tightly folded globular structure made up of three a-

3 helices, two of which are held together by a single disulphide bond, and two short sections of b-sheet. Anatomy of the curious prion protein Subsequent biochemical analysis has shown that this domain folds extremely rapidly and is relatively stable and unaffected by its environment. Experiments to determine the structure of the amino terminal half of the prion protein have not been as successful, reflecting the highly unusual amino acid sequence and secondary structure for this region. Following an initial 'signal peptide' region, there is a short series of hydrophobic amino acids followed by an eight amino acid motif, the octarepeat. As its name implies this motif is repeated perfectly four times and includes residues typical of those that interact with metal ions. Following the octarepeat region a second hydrophobic stretch of amino acids occurs before the beginning of the well characterised C-terminal domain. Clearly, the prion protein is a molecule that is front end loaded and much of the research effort has focused on trying to discover the function of such an unusual sequence combination. Clues to function Efforts to produce the amino terminal half of the protein in isolation and to study it biochemically have proven difficult. However, researchers have ascribed biochemical properties that may reflect those of the complete protein to peptides with sequences derived from this region. In the mid-1990s M. Hornshaw and colleagues at the MRC Neurochemical Unit in Newcastle showed that peptides encoding the octarepeat sequence bind to copper ions in vitro. Peptides with sequences corresponding to the central hydrophobic sequence of the prion protein, between the octarepeats and the C-terminal domain, are cytotoxic to cultured cells in vitro. They also have a propensity to aggregate into fibril structures reminiscent of those found in diseased TSE brains, suggesting that this region could become exposed during the conformational change when the prion protein becomes pathogenic. However, research on isolated peptides cannot necessarily be extrapolated to a full length folded protein. The observations of copper binding by prion peptides therefore remained of speculative interest until the late 1990s, when David Brown, then at the University of Cambridge, showed that this is also a property of the complete recombinant prion protein; roughly four copper ions are bound to each prion molecule, one for each octarepeat. Binding is dependent on ph, suggesting that the protein can be loaded with copper at neutral ph, for example on the cell surface, and then release the ion under acidic conditions, such as in the endocytic vesicles.

4 Additionally, Brown and coworkers showed that copper is present in subcellular fractions containing prion protein isolated from normal mouse brain, but not in equivalent fractions isolated from the brains of prion knockout mice. Recently, and in collaboration with Brown, my group at the University of Reading showed that copper binding to the prion protein in vitro leads to a novel protein conformation that depends on the presence of the octarepeats. The same protein conformation could not be achieved by adding copper after protein folding. These findings lend credibility to the idea that prion might be a cell surface copper binding protein, and that this could relate to the mechanism of pathogenesis. This idea of the prion protein as a cuproprotein is enticing because some known neuropathological disorders, notably Menkes syndrome and Wilson's disease, are the result of aberrant copper metabolism. Could a change in copper binding associated with the change to pathogenic prion be directly linked with the pathology of TSEs? Until very recently, that link has always been speculative and has suffered from two major drawbacks. It has been known since the earliest transgenic prion knockout mice that resident wild type protein is required for the propagation of the infectious form. Subsequent work using mouse varieties with shorter and shorter forms of the prion protein have mapped the minimum amino acids essential for the disease propagation. These experiments have shown that the presence of the octarepeats, the sequences thought to be mainly responsible for copper binding, are not required for pathogenesis. Moreover, Weissmann's group, now in London, has shown recently that introducing a prion gene devoid of the octarepeats to a prion knockout mouse results in renewed susceptibility to disease. The octarepeats clearly have a role to play because aberrant numbers of the repeat are associated with some human TSEs, but at least in experimental systems they are not necessary for prion infectivity. In addition, the affinity of full length prion protein for copper is relatively low (µm) and is insufficient to capture and retain the low levels of copper present in biological tissue. Both of these observations led to the view that copper binding by prion protein was interesting, but lacked obvious physiological relevance. A clearer view From the earliest observations of copper binding by octarepeat peptides, the stoichiometry of binding has been the subject of some dispute. Most studies agree on at least four free copper binding sites per prion molecule, but slightly higher values often recorded imply additional binding sites outside of the octarepeats. Their identity has been obscure, however, until recent studies in which researchers determined the copper content of prion protein after loading with a copper-amino acid chelate rather than free copper. The affinity of free amino acids for copper approximates to that observed for full length prion protein binding. The only way that copper can be stripped from the chelate by prion protein is therefore via sites within the molecule that are higher affinity, probably outside of the octarepeat region. Using this approach, researchers have revealed a new binding site for a single copper ion. It can bind to much lower levels of Cu (fm not µm) and lies within the central hydrophobic region known to be essential for disease propagation and peptide toxicity in vitro. Moreover, the same region appears to be involved in the earliest conformational changes that are associated with conversion to the pathogenic form. The two main objections to copper binding by prion protein being a physiologically relevant process are now redundant and it seems clear that defects in copper metabolism are, directly or indirectly, involved in the observed pathogenesis of the TSEs. If copper is involved in prion function and dysfunction, what roles could it fulfil? Free copper, being redox active, is toxic and is quickly scavenged so copper is always found as a complex

5 in biological systems. In proteins, copper is generally acting as a metal centre for enzyme function, or is in the process of being transported. Prion protein could be involved in either of these functions. Adding radioactive copper to cells expressing prion protein shows that prion can bind to the metal at the cell surface and internalise it, a role consistent with a scavenging mechanism. Additionally, the copper ion could act as a redox centre once bound within the prion protein. Cells expressing prion protein are more resistant to oxidative stress than cells that do not express prion protein, and may play an active role in mopping up free radicals. Indeed, the copper bound prion protein that we have produced in vitro dissipates added superoxide and protects cells from oxidative damage. One factor that might support a more active role for prion protein in copper metabolism is the unusual degree of homology between prion proteins from different vertebrates. Strikingly, for example, the primary sequence of prion proteins in man and mouse is essentially the same. The degree of amino acid conservation is astonishingly high, with the regions of strictest homology lying in the amino terminal region where copper binds. A simple metal ion transporter would probably not require such strict sequence conservation, but a combined metal ion transporter protein and enzyme would explain the evident restriction on sequence diversity. Since prion concentrations are highest at nerve junctions, it is not hard to envisage how a defective protein would lead to neurotoxicity. What now? Armed with a plausible idea of how pathogenesis could arise, the way in which copper is bound to the surrounding sequence becomes highly significant. A simple question posed in light of the new high affinity copper site is: what is the role of copper binding by the octarepeats? What are the repeats doing and why are they conserved? More intriguing still is the idea that copper (or other metal ion) binding itself may, in part, modulate the conversion event. The high affinity site is in the region of the protein that appears to determine some of the disease characteristics; will mutations known to affect these disease traits influence the affinity or range of metal ions that can be bound? Equally, if perturbation of copper metabolism is at the heart of the pathology, then intervention strategies that target copper directly may offer prospects for disease management. The search for sensitive and robust diagnostics should also take account of the fact that metal ions may modulate the conformation of the protein; antibodies that can distinguish the metal dependent conformation of prion protein may be useful diagnostic reagents. Lastly, a role for a redox active metal in the TSEs may draw them into a larger group of neurological disorders, including Alzheimer's disease, that may have a shared basis of action. In that case there is every hope that advances in any one field will be beneficial to a wide range of related diseases. Ian Jones is professor of virology at the school of animal and microbial sciences, University of Reading, Reading RG6 6AJ