|Home | About | Journals | Submit | Contact Us | Français|
Amyloid formation is a hallmark of several systemic and neurodegenerative diseases. Extracellular amyloid deposits or intracellular inclusions arise from the conformational transition of normally soluble proteins into highly ordered fibrillar aggregates. Amyloid fibrils are formed by nucleated polymerization, a process also shared by prions, proteinaceous infectious agents identified in mammals and fungi. Unlike so called non-infectious amyloids, the aggregation phenotype of prion proteins can be efficiently transmitted between cells and organisms. Recent discoveries in vivo now implicate that even disease-associated intracellular protein aggregates consisting of α-synuclein or Tau have the capacity to seed aggregation of homotypic native proteins and might propagate their amyloid states in a prion-like manner. Studies in tissue culture demonstrate that aggregation of diverse intracellular amyloidogenic proteins can be induced by exogenous fibrillar seeds. Still, a prerequisite for prion-like propagation is the fragmentation of proteinaceous aggregates into smaller seeds that can be transmitted to daughter cells. So far efficient propagation of the aggregation phenotype in the absence of exogenous seeds was only observed for a yeast prion domain expressed in tissue culture. Intrinsic properties of amyloidogenic protein aggregates and a suitable host environment likely determine if a protein polymer can propagate in a prion-like manner in the mammalian cytosol.
Amyloidoses are degenerative diseases that are associated with fibrillar aggregates of diverse host-encoded proteins. More than 30 human diseases are associated with amyloid deposition, highly ordered fibrillar aggregates with cross β-sheet structure that are formed by proteins with little sequence similarity.1 Amyloid diseases comprise both systemic and neurologic diseases such as Alzheimer disease, Parkinson disease and Chorea Huntington.2 The fundamental event underlying amyloidoses is the aggregation of aberrantly folded proteins or peptides into amyloid polymers. According to the nucleated polymerization model, seeds composed of abnormally folded proteins template the conformational transition of soluble cognate proteins (Fig. 1A).3 The mechanism of nucleated polymerization is also shared by prions, proteinaceous infectious proteins4 (Fig. 1B) so far identified in mammals and lower eukaryotes. Amyloid diseases and prion diseases can be sporadic or heritable. The infectious nature ultimately separates prion diseases from other amyloidoses. Prion diseases in mammals are fatal neurodegenerative disorders that affect both humans and animals.5–7 Similar to other amyloidoses, human prion diseases can arise spontaneously with an incidence of approx. 1:106 or are inherited due to germline mutations in the prion protein coding sequence.8 Importantly, prions are also infectious and transmit experimentally or naturally within or even between species. It is now widely accepted that an abnormal isoform of the host-encoded prion protein PrP,4,6 constitutes the causative agent of prion diseases. Abnormal PrP (PrPSc) acts as a template that recruits and converts normal prion protein (PrPC) into a structured PrPSc polymer. Strong support for the protein-only hypothesis comes from several recent studies on in vitro generated prion infectivity.9–11 PrPC is a glycosylated protein attached to the outer leaflet of the cell membrane by a glycosyl-phosphatidylinositol-anchor. Conversion of PrPC to PrPSc likely occurs on the cell surface or within early endosomal compartments.12–14
Remarkably, self-perpetuating prion states have also been demonstrated for a growing class of diverse proteins in lower eukaryotes.15 Rather than causing disease, fungal prions constitute epigenetic elements of inheritance that can “encode” and transmit conformational information. In most cases, a conformational switch of the normally soluble and functional prion protein inactivates the protein and leads to a change in phenotype. The induction of the prion phenotype can be favorable under some growth conditions.16 Thus, fungal prions “act as genes” that, although their information content is lower than nucleic acid, may allow rapid adaption to environmental changes.15,17 However, the fact that the [PSI+] and [URE3] prion phenotypes are not prevalent in wild strain isolates suggests that they might not be beneficial under all circumstances.18 Reversion between the non-prion state and the prion state can occur sporadically at very low rate.19 Overexpression of soluble homotypic protein increases rates with which the prion phenotypes arise de novo, likely due to misfolding of overproduced protein.15,20 Spontaneous induction is strongly enhanced by the prion state of the yeast protein Rnq1 and prion aggregates thereof appear to cross-seed aggregation of several other prion proteins.21 Naturally, yeast prions transmit to daughter cells during cell division and the prion phenotype is heritably propagated. New rounds of prion replication are initiated by segregation of prion polymers into smaller seeds, a process that crucially involves the disaggregase Hsp104 and other members of the chaperone machinery.22 Yeast prions are considered infectious because the prion phenotype can be experimentally transmitted to naïve yeast cells by a process called cytoduction.23
The discrimination between non-infectious aggregation-prone proteins and prions has been challenged by recent findings that both systemic and tissue-specific amyloidoses can be induced by seeding-like mechanisms that share similarities with prion propagation.24–26 Degeneration of interconnected brain regions has been reported for several amyloid-related neurodegenerative disorders, including progressive supranuclear palsy and progressive tauopathy in Alzheimer disease. The observed network degeneration27 implicates spreading of protein misfolding rather than cell autonomous intracellular protein aggregate formation. Indeed, tauopathy was shown to be transmittable experimentally by injecting transgenic mouse brain extracts that contain mutant P301S Tau fibrils into mice expressing wild-type human Tau.28 Wild-type Tau aggregation was triggered at the site of injection and at more distant areas, strongly suggesting that protein-misfolding was capable of spreading within the brain.28 Moreover, inclusions of aggregated α-synuclein, a major component of intracellular Lewy bodies associated with several neurodegenerative diseases, were found in neurons that were grafted into the brain of Parkinson disease patients, implicating transmission of α-synuclein misfolding from host to mesencephalic transplants.29,30 These findings provoke the question whether the concept of infectious protein conformers may also be applicable to a broader spectrum of amyloids.
Clearly, the findings that Tau and α-synuclein aggregation spreads in connected brain areas raise the important question how intracellular protein aggregates can trigger aggregation of cognate soluble proteins in neighboring cells. Cytosolic aggregate formation of Tau, α-synuclein and Huntingtin can also be recapitulated in cell culture by overexpression of wild-type proteins, mutant proteins or disease-associated fragments thereof. Expression of these proteins frequently leads to the formation of so-called aggresomes, pericentriolar inclusions that arise when the ubiquitin-proteasome system that normally degrades misfolded proteins is impaired.31 Aggresome formation is an active process of the cell to sequester otherwise detrimental protein aggregates for storage or subsequent clearance by autophagosomes.32
Recent data now provide evidence that exogenous proteinaceous aggregates can trigger cytosolic assembly of aggregation-prone proteins that are normally soluble. Amyloidogenic proteins in their monomeric or polymeric states have been shown to be internalized by cells.33–41 Recombinant fibrils derived from polyglutamine peptides, the microtubule binding region of Tau (MTBR), Aβ and the prion domain of the Saccharomyces cerevisiae Sup35 were readily internalized by tissue culture cells and translocated to the cytosol.42–45 But how exactly do apparently high molecular protein aggregates penetrate the cell? PrPSc has been repeatedly shown to be taken up by various cell types in vitro46,47 and co-localization with endosomal markers or uptake inhibition at low temperatures implicates involvement of the endosomal pathway.48,49 Likewise, MTBR in its fibrillar state was internalized by fluid phase endocytosis.45 The transmission of α-synuclein polymers was reduced considerably in different cell types expressing a dominant negative mutant of dynamin, suggesting that internalization was dependent on endocytic processes.39,41 By contrast, Ren et al. suggested direct penetration of the lipid bilayer as a mechanism for polyglutamine fibril translocation to the cytoplasm. Importantly, membrane permeabilization was also observed for some oligomers from α-synuclein, polyglutamine peptides and prion protein peptide 106–126 but not for fibrils.40,50,51 Internalization mechanisms may therefore differ depending on the size and structure of the polymer, the protein aggregate or the cell-type.
Exogenous oligomers or amyloid-like fibrils can also initiate aggregation of cytosolic homotypic proteins in tissue culture models. Early evidence that extracellular protein polymers can induce intracellular aggregation of homotypic proteins came from studies with different α-synuclein oligomers. Intriguingly, some but not all types of α-synuclein oligomers derived from recombinant protein were capable of inducing intracellular wild-type and mutant α-synuclein inclusions.40 We have previously shown that the yeast Sup35NM prion domain tagged with the antibody epitope of hemagglutinin (NM-HA) remains soluble when expressed cytosolically in murine N2a neuroblastoma cells.52 Addition of exogenous recombinant fibrils but not monomeric Sup35NM to the cell culture medium efficiently triggered aggregation of the cognate soluble protein (Fig. 2A).43 Induction of NM-HA aggregates was also observed in a PrP deficient hippocampal cell line stably expressing NM-HA, demonstrating that both Sup35NM fibril uptake and propagation of the NM-HA aggregate phenotype occurred in another cell line and were independent of the expression of the endogenous prion protein PrP (Fig. 2B and C). Seeding of intracellular aggregates by external amyloid fibrils was also recently shown for two other disease-related amyloidogenic proteins. Self-assembly into cytosolic or nucleoplasmic inclusions is drastically increased with proteins exhibiting polyglutamine (polyQ) tracts above a pathogenic threshold of approx. 37 repeats. Pathogenic polyQ proteins have been shown to recruit non-pathogenic soluble polyQ proteins into juxtanuclear inclusions when proteins are co-expressed in mammalian cells.53,54 Interestingly, exogenous recombinant polyQ (Q44) fibrils were taken up by the cell and sequestered ectopically expressed soluble cyan fluorescent protein tagged Htt exon 1 (CFP-HttQ25) into aggresomes44 (Fig. 3A, left). Seeding of intracellular protein aggregates by external amyloid fibrils was also demonstrated in a cell culture model for Tau aggregation.45 Transiently expressed wild-type Tau exhibits a low tendency to spontaneously self-assemble into aggregates, while the microtubule binding region MTBR readily forms inclusions in cell culture. However, exposure of cells to recombinant MTBR fibrils induced aggregation of endogenous wild-type Tau (Fig. 3A, left).
Importantly, spontaneously formed MTBR aggregates were also naturally transferred between cells, suggesting that transmission of protein aggregates could also initiate aggregation of Tau in neighboring cells (Fig. 3A, right). Natural transmission of Htt polyglutamine polymers in cell culture was rather inefficient but could be drastically increased by disruption of co-cultured donor cells44 (Fig. 3A, right). Transmission of donor-cell derived α-synuclein and subsequent inclusion body formation was also observed in co-culturing experiments with cells overexpressing α-synuclein and control cells, suggesting that aggregates were transmitted between cells.41 If α-synuclein transmission in cell culture can also initiate aggregation of soluble homotypic protein, it has not been elucidated so far. Cell-to-cell transmission of NM-HA aggregates in cell culture is currently under investigation. However, extracts from cells displaying NM-HA aggregates can induce NM-HA aggregation in recipient mammalian cells, providing experimental evidence that the cell-derived aggregates are taken up by cells (Fig. 3B, right). But how are protein aggregates spread between cells? Exocytosis of α-synuclein in its monomeric and aggregated state was recently demonstrated.55 Efficient spread of mammalian prions at least in tissue culture is achieved by direct cell-to-cell contact, for example by nanotubes,56,57 or by release of prion infectivity into the medium, likely by exosomes.58–61 Similar mechanisms might also account for the transfer of other protein aggregates within cell populations.
Induction of cytosolic protein aggregation by external amyloid seeds and transfer of proteinaceous aggregates between cells are characteristic features of the prion life cycle. However, mitotic stability of protein aggregates is ultimately required for stable prion maintenance in tissue culture and yeast. Once infected, cells can produce infectious prions over many cell generations.19,62 Remarkably, a transient exposure of N2a cells expressing NM-HA to recombinant Sup35NM fibrils was sufficient to initiate self-seeded propagation of NM-HA aggregates that was maintained over continuous cell divisions (Fig. 3B, left). Intriguingly, NM-HA aggregates induced by recombinant fibrils varied from frequent small punctate aggregates to few long, unbranched polymers. Single cell cloning revealed that different types of aggregates were faithfully propagated by progeny cells without any obvious loss of the prion phenotype. The fact that extracts from N2a cells containing NM-HA aggregates were also capable of inducing the prion phenotype in yeast (albeit at a low rate) argues that aggregates formed in mammalian cells share features with Sup35 yeast prions. In conclusion, NM-HA aggregates induced in mammalian cells fulfill several criteria for prions: (1) spontaneous prion induction is an extremely rare event, (2) the prion phenotype can be induced by external seeds, (3) prions are stably maintained during cell division and (4) prions are transmissible to naïve cells where they initiate continuous aggregation of normally soluble homotypic proteins.
If amyloidogenic proteins such as α-synuclein or Tau have the capacity to trigger a self-perpetuating aggregation mechanisms that progresses independent of exogenous seeds remains to be determined. The high penetrance of the NM-HA prion phenotype is in strong contrast to the low persistence of CFP-HttQ25 aggregates induced by recombinant polyglutamine fibrils (approx., 3–8% of cells with persistent inclusions in induced cells versus 0–4% of inclusions in non-induced control cells).44 Asymmetrical distribution of polyglutamine inclusions between daughter cells was also observed in tissue culture and a drosophila model of Huntington disease,63 suggesting that except for a small fraction, induced CFP-HttQ25 aggregates might have been diluted out in subsequent cell passages.
The inefficient propagation of induced polyglutamine aggregates and the formation of polyglutamine aggresomes led to the proposal that mammalian cells might have evolved efficient mechanisms to prevent prion-like aggregate propagation in the cytosol.44 Our studies on the aggregation behavior of the yeast Sup35NM prion domain clearly demonstrate that at least some cytosolic proteinaceous aggregates can escape the quality control system of the cell and can be stably maintained upon continuous passage once initiated. If NM-HA aggregates display any characteristics of aggresomes needs further investigation, but clearly single juxtanuclear NM-HA structures were hardly ever detected. Remarkably, transient overexpression of huntingtin exon I with an extended polyglutamine stretch (HD72Q-GFP) in cells stably expressing NM-HA also led to co-aggregation with NM-HA.64 However, aggregation of NM-HA depended on the continuous presence of HD72Q-GFP and was lost once the polyglutamine polypeptide was no longer expressed.64 A high content of glutamine might thus initiate self-assembly of aggregation-prone proteins with similar aggregation-prone stretches, but might be insufficient to allow stable propagation of the aggregated state in a prion-like manner. The fact that NM-HA aggregates induced by recombinant Sup35NM fibrils were not diluted during cell division but efficiently transmitted to daughter cells in subsequent passages (Fig. 2) implicates that aggregates display an intrinsic fragility and/or polymers are efficiently segregated into smaller seeds that can be transmitted during cell division, as has been shown for fungal prions. How this is achieved and if the cellular chaperone machinery is involved remains to be elucidated. Fragmentation of yeast prions crucially depends on Hsp104,22,65–68 a chaperone with disaggregating activity with no known orthologs in the mammalian cytosol. Several other chaperones with homologs in mammals are involved in yeast prion propagation.69,70 Although it was shown that the aggregation domain of Sup35 can to some extend be replaced by a polyQ tract,71 amino acid composition differences in polyQ stretches and the prion domain of Sup35NM account for the difference in initiation of aggregation and mitotic stability of the respective protein aggregates in yeast.71–73 A recent systematic approach revealed a strong bias towards asparagines versus glutamines in identified prion domains in yeast.74 Which sequences allow efficient prion replication in the mammalian cytosol remains to be shown. However, the interplay between intrinsic properties of amyloidogenic proteins and cellular factors in the host cell environment likely accounts for the efficient propagation of the prion state in the mammalian cytosol. Importantly, the prion-like replication of NM-HA aggregates in N2a cells argues that transmission of structural information might also play a physiological role for certain so-far unidentified mammalian proteins, as has been demonstrated for proteins in lower eukaryotes.75,76 It will be one of the major challenges of the future to identify and experimentally validate mammalian proteins with prion-forming capacities and to dissect the cellular machinery that enables their replication as prions.
We are thankful to M. Suhre and T. Scheibel (Universität Bayreuth, Germany) for providing a Sup35 plasmid and the production of rec. Sup35NM and S. Kremmer (Helmholtz Zentrum München, Germany) for generating NM antibodies. We thank the members of the Vorberg and Schätzl laboratories for constructive comments on this work. Financial support for this work was provided by the Deutsche Forschungsgemeinschaft (SFB 596 B14, VO 1277/1-2) and by the EU NoE Neuroprion.
Previously published online: www.landesbioscience.com/journals/prion/article/10013