Self-propagating amyloid as a critical regulator for diverse cellular functions

Roles of Amyloid-Like Functional Fungal Prions

Many mammalian and fungal prions have structural and physical features of amyloids. Prions are cytoplasmic genetic elements and self-propagate in a non-Mendelian manner. Prion inheritance is achieved by the conformational conversion of natively folded or disordered proteins into amyloid forms. The amyloid form of prion proteins acts as a self-template that converts and sequesters native prion proteins into the prion amyloids. Due to the consistency of amyloid-based propagation of prion conformations, prions as cytoplasmic genetic elements are stably inherited by daughter cells over many generations. The result is the continued loss and/or gain of function of the prion proteins once a particular prion state has been established. Thus, formation of amyloid-like prions could impart cells with new cellular phenotypes which are maintained as long as that specific prion conformation is propagated.

In budding yeast Saccharomyces cerevisiae, a wide range of proteins are believed to have the potential to function as yeast prions. Surprisingly, over one-third of ∼700 wild strains were suggested to contain prion-like elements (3). Furthermore, over 100 proteins containing a glutamine/asparagine (Gln/Asn)-rich domain, which tends to induce protein aggregation, may be candidate yeast prion proteins (4). These results suggest that prions are more common than expected and put forward the intriguing hypothesis that cells may utilize prions (amyloids) for physiological processes.

Sup35, a translation termination factor, is the most studied yeast prion protein and its prion formation results in a cytoplasmically heritable phenotype called [PSI⁺] (5–7). In [PSI⁺] cells, Sup35 forms insoluble amyloid-like aggregates (8), which inactivate the function of Sup35 as a translation terminator. This results in a dramatic increase in the frequency of read-throughs at nonsense mutations, which alter gene expression in [PSI⁺] yeast. Interestingly, however, it was recently identified that such read-throughs can occur at specific stop codons even in non-prion [psi^-] cells (9).

Consequently, both advantages and disadvantages of the [PSI⁺] state have been reported. More than half of the artificially induced [PSI⁺] variants are lethal or sick, and thus this prion was proposed to be a disease state (10–12). In contrast, [PSI⁺] cells exhibit growth advantage under certain conditions including ethanol or lithium, depending on their genetic background (3, 13). The presence of [PSI⁺] had previously been observed only in laboratory strains, but [PSI⁺] has recently been found in wild strains (3), suggesting that [PSI⁺] states are beneficial at least to these strains.

Nonsense mutation of ade1 or ade2 causes red colony formation by accumulating a red adenine precursor while [PSI⁺] cells form white or pink colony by read-through of the nonsense mutation. The variation in colour between white and red indicates the total intracellular Sup35 activity which is dictated by amyloid polymorphism (14). Fragile amyloids fragment easily and thus yield more seeds necessary for further propagation of the specific prion conformation, resulting in a strong and stable read-through phenotype (15, 16). There are also phenotypically non-specific [PSI⁺] cells which give rise to colonies with various colours. This phenomenon is commonly explained by either maturation or multiple variant hypotheses, or both (17). Under the maturation hypothesis, immature Sup35 prions may undergo different conformational changes to form a diverse set of structurally stable variants during the maturing process. Conversely, a single cell with multiple prion variants could also explain this phenomenon because daughter cells inheriting one of the many variants may inherit a specific phenotype associated with that particular prion variant. Thus, amyloids such as [PSI⁺] prions may have the potential to fine tune its phenotypic effects by maturating prion conformation or by competition of various prion conformations in a cell.

In addition to autonomous self-propagating properties, the conformation and function of amyloids could also be modulated by interactions between amyloid-forming proteins. For instance, the structural conversion of a soluble yeast prion protein to prion conformation is strongly affected by the presence of other amyloids. This is highlighted by the observation that the formation and maintenance of [PSI⁺] depend on other amyloids such as Rnq1 prions (4, 18–20). Rnq1 amyloids with distinct conformations affect the appearance of different [PSI⁺] variants (21), indicating that pre-existing Rnq1 amyloids could regulate conformations of Sup35 amyloids in cross-seeding reactions and resulting phenotypes of [PSI⁺] yeast. Furthermore, overexpression of Gln/Asn-rich proteins such as Lsm4 and Pin4 leads to elimination of prion states (22, 23), which provides a potential means for phenotypic changes. The prion loss was suggested to be caused by the binding of Gln/Asn-rich amyloids to pre-existing prions, leading to the formation of enlarged prion aggregates with decreased efficiency of prion fragmentation and propagation. Alternatively, Gln/Asn-rich proteins sequester Hsp104 chaperone away from the diffusible cytoplasmic pool, resulting in the elimination of the prion state, as these chaperones are believed to be critical for prion propagation (24, 25).

Similar to Sup35 prions, the prion form of the transcriptional regulator Sfp1 in [ISP⁺] prion-state yeast was reported to regulate translation termination (26). The [ISP⁺] state shows enhancement of SUP35 expression, which correlates with antisupression of nonsense suppressor mutations in [ISP⁺] yeast, while Δsfp1 strains show the reduction of SUP35 expression and nonsense mutations are suppressed (27). The phenotypic differences between [ISP⁺] and Δsfp1 strains revealed that prion states are also capable of acquiring novel functions.

It has been demonstrated that multiple yeast prions can control the transcription of FLO11, a regulator of multicellularity and adhesion (28–30). Conversely, transcription regulators of FLO11 have been predicted to have higher propensity to be prions compared with randomly selected transcription factors (30). These findings suggest that some prions could act as a molecular switch of multicellularity. Indeed, prion formation of Mot3 results in the acquisition of an adhesive phenotype and formation of more elaborated biofilm (30). Remarkably, the conversion of soluble Mot3 to the prion form is regulated by environmental conditions. High ethanol levels in culture media could induce the prion-state yeast [MOT3⁺], proposed to be due to a perturbation of protein homeostasis. In addition, hypoxia eliminates [MOT3⁺] by repressing the expression of MOT3 and thus decreases the number of heritable Mot3 prion seeds. Similar phenotypes were also observed in other prion strains. Overexpression of a transcriptional repressor Cyc8 induces the prion state [OCT⁺] and causes higher flocculation (31). A wild wine yeast was reported to show a [PSI⁺]-dependent adhesive phenotype to solid media (3).

The drug-resistant phenotypes are also widely acquired by other yeast prions. For instance, the prion form of a chromatin remodeller, Swi1 (33), leads to formation of the prion state [SWI⁺], which is resistant to microtubule disruption (4). Furthermore, several wild yeast strains are resistant to DNA-damaging agents or antifungal drugs, and these phenotypes are eliminated by the ectopic expression of Sup35 or Mot3 lacking a prion-forming domain. These results suggest that the drug resistance is caused by Sup35 or Mot3 prions (3).

The HET-s protein from the filamentous fungus Podospora anserina is involved in a genetically controlled programmed cell death phenomenon termed heterokaryon incompatibility. Filamentous fungi have developed a self/non-self recognition system that prevents the vegetative fusion of individuals that differ in specific loci termed het-loci. The het-s locus exists as two incompatible allelic variants termed het-s and het-S. While a soluble HET-S protein is expressed from het-S leading to [Het-S] strains, HET-s protein derived from het-s gene can form two distinct conformations of soluble and amyloid (prion) forms, leading to neutral (prion-free) [Het-s*] and prionized [Het-s] phenotypes, respectively (34–36). The crosses between [Het-s*] and [Het-S] strains lead to the formation of viable mixed cells (heterokaryon). Conversely, when [Het-s] strains containing Het-s prions fuse with [Het-S] strains, the hyphal fusion triggers a lethal reaction at the contact site, thereby preventing mixing of the cytoplasmic components of the two hyphal types. This phenomenon is also observed in meiotic drive by sexual crosses between [Het-s] and [Het-S] (37). These results establish that HET-s prions function as a trigger for cell death.

Transcriptional and Translational Regulation by Amyloid-Like Aggregation of RNA-Binding Proteins

In some types of cancerous cells, the LC domains of FUS, Ewing's sarcoma (EWS) and TAF15 (collectively called as FET proteins) are translocated onto various DNA-binding proteins, which in turn activate transcription of the bound DNA (40). The C-terminal domain (CTD) of RNA polymerase II can bind to the polymerized fibres of LC domains and this binding allows initiation of transcription. Subsequent CTD phosphorylation can then release the bound RNA polymerase from the LC domains to facilitate transcriptional elongation. Taken together, amyloid-like fibres of LC domains regulate transcription by recruiting and releasing RNA polymerase II.

A yeast RNA-binding protein Whi3 that negatively regulates a G1 cyclin has recently been reported to be a mnemon. Mnemon acts as protein-based memory through super-assembly of the protein of interest (Whi3, in this case) which can be inherited asymmetrically at mitosis or maintained in non-dividing cells (41). In response to pheromone of an opposite mating type, Saccharomyces cerevisiae haploid cells show cell cycle arrest in the G1 phase and form mating projection towards the pheromone source. However, long-term pheromone exposure releases the cell from cell cycle arrest and cell division is restarted. As a ‘molecular memory’, this escape phenotype is inherited only by mother cells during cell division and is not affected by further cell division even in the absence of pheromone. The prolonged exposure to pheromone induces the formation of super-assembly of Gln/Asn-rich Whi3, which is resistant to detergent and proteinase as observed for amyloid. The formation of super-assembly was proposed to release mRNA of G1 cyclin CLN3 from translational inhibition due to Whi3 binding, thereby promoting escape from the G1 phase arrest. Ectopic overexpression of GFP-tagged Whi3 formed a super-assembly which is asymmetrically segregated into mother cells during cell division, consistent with the behaviour of the haploid cells exposed to mating pheromone. Furthermore, Ashbya Whi3, an orthologue of Saccharomyces Whi3, also forms a large protein assembly and the aggregation was suggested to asynchronize nuclear division cycles in multinucleate cells by heterogeneously clustering CLN3 mRNA (42).

Cytoplasmic polyadenylation element-binding protein (CPEB) of the sea slug Aplysia regulates translation of its target mRNAs and is required for persistent synaptic facilitation in sensory neurons. Both the formation of prion-like CPEB multimers and sustained CPEB-dependent local protein synthesis for synaptic growth play critical roles in long-term facilitation (43, 44). The amyloid-like aggregates of Drosophila Orb2A, an orthologue of CPEB, were also reported to function in the maintenance of long-term memory (45). An F5Y point mutation in Orb2A dramatically reduces amyloid-like oligomerization and causes defects of long-term memory, indicating that Orb2A aggregation is crucial for the persistence of memory.

Expanding Roles of Functional Amyloid in Eukaryotic and Prokaryotic Cells

Recent studies have expanded the list of functional amyloids in both eukaryotic and prokaryotic systems. For example, human RIP1/RIP3 amyloid was shown to be necessary for tumour necrosis factor (TNF)-induced programmed necrosis (46). Three TNF-induced pathways, activation of NF-κB, apoptosis and programmed necrosis, are regulated by polyubiquitination and caspase-mediated digestion of RIP1. The inhibition of polyubiquitination and caspase activity by inhibitor of apoptosis (IAP) antagonists and viral infection was demonstrated to result in necroptosis and the heterodimeric amyloid formation of RIP1 and RIP3 is believed to play a critical role in programmed necrosis by activating their kinase activities.

In immune systems, functional amyloid-like aggregates act to switch cellular signalling. In human cells, one of the antiviral immune response is initiated by RIG-I receptor that recognizes viral RNA (47). The binding of viral RNA induces the conformational change of RIG-I, leading to amyloid-like fibril formation of the mitochondrial antiviral signalling protein (MAVS). In this aggregate state, MAVS activates transcription factors such as IRF3, which in turn triggers the immune response.

A variety of peptides and protein hormones can form amyloids and they are stored in pituitary secretory granules as amyloids (48). The rigid amyloid structure and reversibility of amyloid formation of the peptides and protein hormones allow long-term storage and controlled release of monomeric hormones, respectively. Pmel17, a melanocyte protein necessary for eumelanin deposition in mammals and found in melanosomes, was shown to form an amyloid structure. Pmel17 amyloids accelerate the polymerization of reactive small molecules into melanin (49). Pmel17 amyloids also appear to play a role in reducing the toxicity associated with melanin formation by sequestering and minimizing diffusion of highly reactive, toxic melanin precursors out of the melanosome. Enteric bacteria such as Escherichia coli have a fibrillar component of the extracellular matrix called curli (50), which contains amyloid structure and is involved in adhesion, flocculation and biofilm formation. Saccharomyces cerevisiae and dimorphic fungus, Candida albicans, also have amyloid-forming proteins whose functions are related to cell adhesion (51, 52).

Conclusions and Perspectives

A growing body of evidence indicates that diverse amyloids act as functional units in various biological processes. These results imply that functional amyloid is pervasive in a variety of organisms and systems. Importantly, some of these functions are directly linked to a fitness advantage in response to environmental stress, which is essential to cell survival (53). The recent development of methods to search for novel amyloids includes computational analysis (4), proteomics (54) and cross-seeding assays (19, 20). These methods as well as unprecedented technologies will facilitate both identification of novel functional amyloids and their unexpected physiological roles in the cell.

Acknowledgements

We thank Kelvin Hui for his comments on the article.

Conflict of Interest

None declared.

Abbreviations

Abbreviations
 
• Asn

  asparagine

 
• b-isox

  biotinylated isoxazole

 
• CPEB

  cytoplasmic polyadenylation element-binding protein

 
• CTD

  C-terminal domain

 
• EWS

  Ewing's sarcoma

 
• FET

  FUS/EWS/TAF15

 
• FUS

  fused in sarcoma

 
• Gln

  glutamine

 
• IAP

  inhibitor of apoptosis

 
• LC

  low complexity

 
• MAVS

  mitochondrial antiviral signalling protein

 
• TNF

  tumour necrosis factor

References