Prion Strain Problem: overview
Encipherment and propagation of strains
The prion hypothesis--the famously heretical notion that naked protein particles, without a stitch of nucleic acid, can cause transmissible disorders of the nervous system such as mad cow disease and the similar Creutzfeldt-Jakob disease (CJD) of humans--has gained increasing acceptance in recent years.
But even true believers acknowledge a major stumbling block: the existence of multiple strains of prions with different, and apparently inheritable, characteristics. "This was the fly in the ointment for the prion hypothesis," says Glenn Telling, a molecular biologist who works at the University of California, San Francisco, with Stanley Prusiner, the leading proponent of the hypothesis. "How can you have stably inherited information--strain differences--without a genetic component?"
Within the past year or two, a possible solution to that quandary has begun to emerge. Prions may consist of a cellular protein that has misfolded into an abnormal three-dimensional structure, so researchers have speculated that different prion strains consist of the same protein misfolded in different ways. The idea is that each one can impose its brand of misfolding on the normal protein by simple contact, thereby transmitting disease with unique strain characteristics. In this issue, Telling, Prusiner, Pierluigi Gambetti and their colleagues--working with prions from two human diseases--provide compelling new evidence for that idea.
They show that a single type of normal prion protein in the brains of living mice can be converted into two different forms, depending on the type of abnormal human prion that initiates the conversion. The result is two different patterns of pathological changes, as would be expected for different prion strains. Dennis Choi, a neurologist at Washington University in St. Louis, calls the findings "a very powerful piece of support for the idea that an abnormal protein can confer its abnormal configuration onto the host prion protein."
Prion strains may consist of differently folded abnormal versions of the protein that transmit their characteristics to the normal prion protein by forcing it to fold the same way.
The puzzle the UCSF workers are addressing dates from 1985, when Prusiner's lab and others found that prions, the protein particles found in the brain in neurodegenerative disorders like CJD and kuru in humans, mad cow disease, and scrapie in sheep, are misfolded or "flipped" versions of cellular prion protein (PrP), a normal component of brain cells. That flipping, or conformational change, may occur in the brain spontaneously for unknown reasons or because of mutations in cellular PrP, and cause buildup of abnormal prions and nerve cell damage. But because some of the diseases are transmissible, Prusiner also proposed that infection with one of the misfolded versions can induce disease by forcing healthy PrP molecules to refold themselves into abnormal prions.
But at the time, few other researchers would consider the idea of an infectious agent composed only of protein, a problem compounded by the existence of prion strains. They insisted that the disease-causing agent had to have some kind of genetic material, perhaps a virus lurking in the prion preparations. Then, in the December 1994 Journal of Virology, Richard Bessen and Richard Marsh of the University of Wisconsin, Madison, reported some of the first evidence that different misfolded conformations of the protein might explain the strains.
Studying mink transmissible encephalopathy, a prion disease, they found two strains when the disease was transmitted to hamsters, each one associated with different incubation times, symptoms, and distributions within the brain--as well as with a distinct conformation of the hamsters' PrP. In a follow-up study in Nature in June 1995, Bessen, who had moved to the National Institutes of Health Rocky Mountain Laboratories in Hamilton, Montana, Byron Caughey, and their Rocky Mountain colleagues showed that the two prion conformations could be spread to one type of cellular PrP when the proteins were mixed together in a cell-free system.
Now, Telling, Prusiner, and their colleagues have extended the work to human prions in a mouse model. They used a strain of transgenic mice carrying a chimeric human/mouse PrP gene that was developed in their lab for susceptibility to human prions. They injected the animals' brains with extracts from the brains of patients who died either from CJD or from fatal familial insomnia (FFI), an inherited disorder caused by a mutation in the PrP gene.
Prusiner and his colleagues knew that the prion proteins causing the two disorders differ in the way they fold because when they are cut with a protein-splitting enzyme, the CJD material gives a 21-kilodalton fragment, while that from the FFI patients yields a 19-kilodalton fragment. The question then was, would the prions produced in mice infected with the human prions yield the same protein fragments as the parent molecules?
And indeed, the animals produced PrP fragments matching those of the original inoculants. "Clearly, you can impart two different conformations to the same primary structure," Prusiner says. "We don't know at this point whether we can keep propagating them in these mice. We suspect that we can, and that's the ongoing experiment." He also suspects that many more conformations are possible.
No study involving prions goes undebated, however, and this one is no exception. At the heart of the criticism lies a problem that has dogged prion research since it first began: No one has yet been able to take the definitive step of generating purified prions in a system that would rule out the possible presence of viruses and show that they transmit scrapielike diseases. "The group has taken a relatively crude [brain] extract," says Michael Harrington, who studies nervous system diseases at the California Institute of Technology in Pasadena. "Therefore, it's not defined." The paper, he adds, is evidence that conformational differences yield strains, "but it doesn't confirm it."
To Prusiner and his colleagues, the failure of years of research to turn up a virus is powerful evidence that none is involved in these disorders. But Telling says, "there are people who will go to their graves believing these diseases are caused by viruses." Prusiner concurs: "Some will say that forever, and there's nothing to say to them except that the evidence is overwhelming. ... They can think what they want. I can't help them."
The fundamental event in prion diseases seems to be a conformational change in cellular prion protein (PrP-C) whereby it is converted into the pathologic isoform PrP-Sc. In fatal familial insomnia (FFI), the protease-resistant fragment of PrP-Sc after deglycosylation has a size of 19 kilodaltons, whereas that from other inherited and sporadic prion diseases is 21 kilodaltons. Extracts from the brains of FFI patients transmitted disease to transgenic mice expressing a chimeric human-mouse PrP gene about 200 days after inoculation and induced formation of the 19-kilodalton PrP-Sc fragment, whereas extracts from the brains of familial and sporadic Creutzfeldt-Jakob disease patients produced the 21-kilodalton PrP-Sc fragment in these mice. The results presented indicate that the conformation of PrP-Sc functions as a template in directing the formation of nascent PrP-Sc and suggest a mechanism to explain strains of prions where diversity is encrypted in the conformation of PrP-Sc.
For many years the prion diseases, also called transmissible spongiform encephalopathies, were thought to be caused by slow-acting viruses (1), but it is now clear that prions are not viruses and that they are devoid of nucleic acid (2, 3). Prions seem to be composed only of PrP-Sc molecules, which are abnormal conformers of a normal, host-encoded protein designated PrP-C (3, 4). PrP-C has a high alpha-helical content and is virtually devoid of beta-sheets, whereas PrP-Sc has a high beta-sheet content (4, 5); thus, the conversion of PrP-C into PrP-Sc involves a profound conformational change. Formation of PrP-Sc is a post-translational process that does not appear to involve a covalent modification of the protein (6).
The prion diseases are unique in that they may present as inherited and infectious disorders (3, 7). More than 20 different mutations of the human (Hu) PrP gene segregate with dominantly inherited disease; five of these have been genetically linked to familial Creutzfeldt-Jakob disease (fCJD), Gerstmann-Sträussler-Scheinker disease, and fatal familial insomnia (FFI) (8). The most common prion diseases of animals are scrapie of sheep and bovine spongiform encephalopathy; the latter may have been transmitted to people through foods (9).
To extend studies on the transmission of wild-type and mutant prions from sporadic Creutzfeldt-Jakob disease (sCJD) and fCJD patients, respectively, to transgenic mice expressing a chimeric mouse-human PrP gene [Tg(MHu2M) mice] (10, 11), we inoculated these mice with mutant prions from the brains of patients who died of FFI. Transmission of human prions to Tg(MHu2M) mice involves the conversion of chimeric MHu2M PrP-C into MHu2M PrP-Sc through a process that is thought to involve the binding of PrP-Sc to PrP-C as PrP-C undergoes a structural transition (12, 13). A point mutation of the PrP gene at codon 178 [in which an Asp residue at position 178 is mutated to Asn (D178N)] is the cause of FFI, but a Met residue must be encoded at position 129 on the mutant allele for the FFI phenotype to be manifest (14). The same D178N mutation segregates with a subtype of fCJD, but in this case, Val is encoded on the mutant allele at position 129. The D178N mutation is thought to destabilize the structure of PrP-C, resulting in its transformation into PrP-Sc (13, 15). Some investigators have reported transmission of FFI prions to non-Tg and Tg(HuPrP) mice; the incubation times exceeded 400 days, and only a minority of the inoculated Tg(HuPrP) mice expressing both human and mouse PrP-C developed disease (16). These findings with Tg(HuPrP) mice are in accord with earlier studies showing that transmission of human prions to Tg(HuPrP) mice is inhibited by mouse PrP-C, and this inhibition can be abolished by ablation of the mouse PrP gene (Prnp-O/O) (10, 11)Tg(MHu2M)Prnp-O/O mice (17) were inoculated intracerebrally with extracts prepared from brain tissue obtained after the death of individuals who died of FFI, fCJD(E200K) (with a mutation in which Glu at position 200 has mutated to Lys), or sCJD. The mice developed signs of experimental prion disease about 200 days after inoculation . At the time of writing, inoculation of Tg(MHu2M)Prnp-O/O mice has resulted in primary passage of prions from at least one brain region from three of four FFI patients. As previously reported, Tg(MHu2M)Prnp-O/O mice are susceptible to prions from patients who carried the E200K mutation (11). Extracts from patients who died with fCJD(E200K) or sCJD(M/M129) (homozygous for Met at position 129) caused neurologic dysfunction in Tg(MHu2M)Prnp-O/O mice between 170 and 190 days after inoculation. All samples were 10% (w/v) brain homogenates that were diluted 1:10 before inoculation. If the PrP gene of the patient carried11).
The failure to transmit disease with brain homogenates of frontal and insula cortices from FFI patient V-58 is apparently not related to heterozygosity at codon 129, because homogenate from patient IV-16, who has the same haplotype, transmitted the disease. Patients V-58 and IV-16 are phenotypically similar and exemplify the FFI phenotype of the codon 129 heterozygotes with especially long duration (18). The sleep disorder was comparable in patients V-58 and IV-16, and spongiosis was actually more severe in patient IV-16 than in patient V-58 (18). It is noteworthy that a homogenate prepared from the parietal cortex of a patient with fCJD(D178N, V/V129) has so far failed to transmit disease. whether the titer of prions in this particular sample is low or the Val (V) residue at position 129 in combination with the D178N mutation prevents transmission to Tg(MHu2M)Prnp-O/O mice remains to be established.
Prion proteins in extracts from the brains of Tg(MHu2M) mice inoculated with FFI(D178N, M129) were compared with those inoculated with fCJD(E200K) or sCJD. Mouse brain homogenates were digested with proteinase K (100 µg/ml) for 1 hour at 37°C followed by denaturation by boiling in 3% SDS. The denatured PrP-Sc was then digested with glycopeptide N- glycosidase (PNGase F) to remove Asn-linked oligosaccharides. As previously described, the human brain extracts prepared from FFI patients yielded a 19-kD protein, whereas extracts from brains of patients with fCJD(E200K) or typical sCJD contained a 21-kD protein (19). Because the amino acid sequences of HuPrP-Sc molecules from FFI and fCJD(E200K) patients differ at two residues, it was not surprising that the conformations of PrP-Sc as reflected by the size of the protease-resistant PrP fragments are different. In contrast, it was unexpected that PrP-Sc found in Tg(MHu2M) mice inoculated with FFI prions would be 19kD, whereas that in Tg(MHu2M) mice injected with fCJD(E200K) was 21 kD (Fig. 1A). These findings demonstrate that the conformation of HuPrP-Sc in the inoculum is replicated in the brains of the Tg(MHu2M)Prnp-O/O mice by conversion of MHu2M PrP-C into MHu2M PrP-Sc. Wild-type PrP-Sc in the brain extract from a patient who died of sCJD was found To be 21 kD. Transmission of sCJD to Tg(MHu2M)Prnp-O/O mice produced MHu2M PrP-Sc, also of size 21 kD, again demonstrating the fidelity of the
We emphasize that whereas the primary structures of the PrP-Sc molecules in the three different human brain inocula are distinct, The amino acid sequences of the PrP-Sc molecules in the brains of inoculated Tg(MHu2M) mice are invariant. The MHu2M PrP transgene was sequenced and found to be the same as the construct used for microinjections during production of the mice. Protein immunoblot analysis of PrP-Sc from brains of Tg(MHu2M) mice inoculated with FFI, fCJD(E200K), and sCJD.
Homogenates of human or mouse brain were prepared as described (30). (A) Comparison of fCJD(E200K) and FFI. Samples analyzed are from the In FFI-inoculated mice, PrP-Sc deposition was most intense in the thalamus and the rostral part of the corpus callosum. In FFI patients PrP-Sc deposition and neuropathologic changes are marked in the antero-ventral and mediodorsal nuclei of the thalamus (18, 22). Intermediate intensities of immunostaining were found in the deeper layers of the frontal cortex and in the lateral portions of the caudate nuclei. Little or no immunostaining was found in the habenula or the hypothalamus. Staining in the hippocampus was also negative except for the stratum lacunosum molecularae where most of the spongiform degeneration and reactive astrocytic gliosis occurred. The absence of PrP-Sc deposition in the habenula is unique to FFI because deposition invariably occurs in this region in response to CJD and scrapie In contrast to FFI-inoculated mice, inoculation of Tg(MHu2M) mice with fCJD(E200K) and sCJD prions induced PrP-Sc accumulation in many areas of the central nervous system (Fig.<2, B and C). Although inoculation with fCJD(E200K) and sCJD as well as with iatrogenic CJD prions (24) resulted in accumulations of PrP-Sc in the brainstem, that was not the case for FFI. These differences in PrP-Sc deposition reflect earlier studies on prion strains where the The neuropathologic changes in the brains of five Tg(MHu2M)Prnp-O/O mice inoculated with prions from three different FFI patients were examined. They were characterized by moderate to severe spongiform degeneration and astrocytic gliosis in the deeper layers of the frontal cortex and rostral part of the cingulate gyrus, the thalamus, the lateral portions of the caudate nucleus, and in the white matter tracts of the cerebral hemispheres. Immunohistochemical examination of FFI-inoculated brains showed that regions with the largest amount of PrP staining corresponded to the regions with the most severe neuropathological changes in both gray and white matter. The accentuated immunostaining resulted from multiple Comparison of FFI-inoculated Tg- (MHu2M)Prnp-O/O mice with the same type mice inoculated with prions from fCJD- (E200K) and sCJD patients revealed two main histopathologic differences. First, FFI produced no vacuolation in the hypothalamus, whereas a mild To moderate degree of vacuolation was found with both sCJD and fCJD(E200K) prions. Secondly, FFI produced moderate to severe vacuolation of the corpus callosum. In contrast, there was only mild vacuolation of the corpus callosum with sCJD and none with fCJD(E200K). Our studies show that human prions inoculated into Tg(MHu2M)Prnp-O/O mice instruct with substantial fidelity the formation of distinct If such properties are propagated, then this will suggest that different mutant human PrPs will have generated distinct strains of prions. The existence of prion strains has posed a conundrum as to the mechanism by which strain-specific characteristics are encrypted (23, 26). Although differences in the size of protease-resistant fragments of PrP-Sc have not been a general characteristic of prion strains (27), The hyper and drowsy strains of prions isolated from mink by serial Our results provide a plausible mechanism for explaining diversity in a pathogen that lacks nucleic acid; the biological properties of prion strains seem to be encrypted in the conformation of PrP-Sc. Because prion strains produce different disease phenotypes, such findings raise the possibility that deviations in the phenotypes of other degenerative disorders may also reflect conformational variants in pathologic proteins. Variations in the conformation of PrP-Sc are reproduced through templating of the PrP-Sc in the inoculum onto the substrate PrP-C. Deciphering the molecular events by which the conformation of one protein is imparted to another and the mechanism responsible for the apparently high degree of fidelity associated with this process should be of considerable interest. Indeed, the foregoing data violate the widely and long-held idea that amino acid sequences are the sole determinants of the tertiary structures of biologically active proteins (29).
REFERENCES AND NOTES
23 August 1996; accepted 28 October 1996