A genetic platform for the study of protein

perturbation and prion-based inheritance

by Gregory A. Newby

B.S., Biological Sciences Carnegie Mellon University (2009)

SUBMITTED TO THE DEPARTMENT OF BIOLOGY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY AT THE MASSACHUSETTS INSTITUTE OF TECHNOLOGY

June 2017

Massachusetts Institute of Technology 2017. All rights reserved.

Signature redacted Signature of Author...... Department of Biology redacted C e rtifie d B y .. Signature ...... I.R..... F...... / Gerald R. Fink Professor of Biology Acting Thesis Supervisor in Place of Susan Lipdquist (deceased)

Accepted By...... Signature redacted. Amy E. Keating MASSACHUSETTS INSTITUTE Professor of Biology OF TECHNOLOGY Chair, Graduate Student Committee

MAY 2 3 2017 LIBRARIES I 1 2 A genetic platform for the study of protein perturbation and prion-based inheritance

By Gregory Arthur Newby Submitted to the Department of Biology in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Biology

SUMMARY Proteins mediate every cellular process. In order for life to exist, each protein must be finely tuned to carry out its function at the proper time and place. Because the environment is dynamic and often unpredictable, the regulation of proteins must be responsive to environmental stimuli. Mutations, age, and severe or repeated insults can decrease the quality of protein regulation, leading to disease. The study of protein regulation and its dysfunction in disease are of vital importance. Regulatory and disease phenomena involving protein assembly or aggregation are common but currently understudied on account of their intractability with existing techniques. In order to equip scientists with better tools to tackle these difficult phenomena, my collaborators Ahmad Khalil and Szilvia Kiriakov (Boston University) and I developed the yTRAP platform (standing for: yeast transcriptional reporters of aggregating proteins). yTRAP couples the activity of a synthetic transcriptional activator to a protein's solubility. It enables sensitive measurement of a protein's state within a eukaryotic cellular context, preserving complex interactions that may be lost using in vitro techniques. yTRAP can be measured in high throughput to enable large studies, screens, and selections on aggregation phenomena. The reporter output is modular and can be customized to desired purposes and measurement modalities. Using a fluorescent output, the signal from yTRAP is readily quantifiable. The combination of these desirable properties enables many kinds of previously-impossible studies. Furthermore, because of its exquisite sensitivity, yTRAP can be used to broadly screen for protein perturbation beyond the context of aggregation. For example, it can report on alterations in protein localization, binding partners, or degradation. I applied yTRAP to track yeast prions, which have previously been difficult to study due to a lack of simple and reliable reporters. Prions are protein-based elements of inheritance that have profound implications for the evolution of single-celled organisms. I first utilized yTRAP to identify factors that faithfully switch prion states on and off, thus proving that prion states can be deterministically regulated. I used these factors to create new cellular tools out of prions: 1) a memory device that records elevated temperatures experienced by a yeast population, and 2) an anti-prion drive that eliminates prions from mating partners and progeny. Separately, I conducted an ecological study into the yeast prion [SW*]. I found that [SWI*] confers a 'pioneering' cellular program that encourages migration and diverse mating partners. Loss of the prion confers a protective 'settled' cellular program with growth and survivability advantages. yTRAP greatly facilitated this study through reliable tracking of the prion state. Prion-like phenomena are now ripe for study with yTRAP.

Thesis Supervisor: Susan Lindquist, Professor of Biology

3 4 Acknowledgments

I will be forever grateful for the time I spent in Susan Lindquist's laboratory. Susan was an amazing scientist, mentor, and person who gave so much of herself to everyone she worked with. In our many meetings throughout the years, I never failed to leave revitalized and excited about the work to come. She had enormous vision for the future and a never-ending stream of ideas. She was also a highly skilled and thoughtful manager who assembled a top-notch team to help run the lab - together they brought in and supported an amazing set of people with whom it was a pleasure to come to work every day. In large part that pleasure was due to the friendly and collaborative environment fostered directly by Sue. Her death was felt deeply by all of us; we will always miss the days we shared together when our laboratory was still whole. I hope that in this thesis and in my future work, I can live up to Susan's high expectations.

I would like to thank all of the members of the Lindquist lab, past and present, for their friendship and scientific contribution to my experience and training. In countless ways each person added to the unique and happy environment, where unlimited help was never more than a question away. I would particularly like to thank members of the prion subgroup, Peter Tsvetkov, Can Kayatekin, Erinc Hallacli, Kendra Frederick, Sohini Chakrabortee, Lauren Pepper, Bill Hesse, Dan Jarosz, Manoshi Datta, my co-leader of chocolate subgroup Georgios Karras, my rotation mentor Mikko Taipale, the administrative staff Brooke Bevis, Linda Clayton, Bob Burger, Audrey Madden, Rosemary Benson, and the materials assistants Ndubuisi Azubuine and Tsering Yangchen. I could fill the page with the names of my many labmates, who all contributed to the wonderful experience I had has a graduate student.

I would like to thank my collaborator Ahmad "Mo" Khalil, his graduate student Szilvia Kiriakov, and the other members of his lab. Mo and Szilvia have been my closest collaborators since very early on in my thesis work. It has been a pleasure working closely with them, and I have come to think of Mo as a second mentor to me. Together, the three of us co-authored the manuscript that I have included in this thesis as chapter 2. A great deal of credit for the ideas, text, and figures is owed to them. I hope that our collaborative work together is only just beginning.

I would like to thank the members of my thesis committee, Gerry Fink, Dane Wittrup, Tim Lu, Graham Walker, and lain Cheeseman, for their advice and encouragement provided at each meeting, and for their input as I consider future steps. I have valued their guidance greatly.

My graduate work was made possible by the many wonderful scientists who trained me before. I would like to thank my mentors and from Carnegie Mellon, Peter Berget, Beth Jones, and Jim Burnette, and also my many labmates and teachers. I thank Andreas PlOckthun who hosted and mentored me during my Fulbright project, and his laboratory.

I would also like to thank the wonderful friends I have made in graduate school and those I met long before. We have had a lot of fun through all manner of good times, and

5 stuck together even through much sadness. I thank The Ten Groomsmen, Rohit Ramnath, Fowler Brown, Jon Chastek, Andy Echenique, Kevin Knockenhauer, Rob Mathis, Isaac Oderberg, Ethan Sokol, Bradely Yates, and Timmy Zhu. I would also like to thank the Ramnath and Zhu families, who on several occasions hosted me in their homes and treated me as a son. I thank all of my classmates for making the transition to graduate school a fun and easy one. I especially thank all members of the Biomansion community, who lived together with me, hosted, and/or attended many fun events every single week.

Finally, I would like to thank my dear family - the many cousins and relatives scattered throughout the globe, my sisters Kristin, Laura, Colette, and Alexandra, my parents Richard and Hilary, and my wife Hee Yeon. Over the years they have all shown me so much love and support that I am overwhelmed to think of it. I would especially like to thank my parents for the incredible amount of effort it took on their parts to get me to where I am today, and for inspiring me through their strong work ethic, inquisitive spirits, and commitment to reason. And to Hee Yeon I owe a very special debt of gratitude - she willingly signed up for this partnership despite not knowing what trials lay ahead! Each shared day with her is a new adventure, and her steadfastness is a constant comfort and source of strength for the journey.

6 Table of Contents

Chapter 1: Introduction...... 9 S u m m a ry ...... 1 0 References ...... 29

Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives...... 37 S u m m a ry ...... 3 8 Introduction...... 38 R e s u lts ...... 4 2 D is c u s s io n ...... 6 3 M e th o d s ...... 6 8 R e fe re n c e s ...... 8 1 Supplem ental figures ...... 88 Supplemental tables ...... 98

Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual e x p lo ra tio n ...... 1 1 3 S u m m a ry ...... 1 1 4 Results and discussion ...... 114 M e th o d s ...... 1 2 4 R e fe re n c e s ...... 1 2 9 Supplem ental figures ...... 132 Supplem ental Table...... 138

Chapter 4: Conclusion and future directions ...... 139 C o n c lu s io n ...... 1 4 0 F u tu re d ire c tio n s ...... 14 2 1. New studies facilitated by yTRAP technology...... 142 11. Potential extensions to the yTRAP system ...... 147 l1l. The application of prions to synthetic biology ...... 150 IV. The ecological ram ifications of prions ...... 151

Appendix A: Selection for the [psi-] phenotype with canavanine ...... 153

Appendix B: RIP1/RIP3 prion propagation in yeast ...... 157

7 8 Chapter 1

Introduction

*This chapter has been published previously: Newby GA, Lindquist S. Blessings in disguise: biological benefits of prion-like mechanisms. Trends Cell Biol. 2013; 23(6):251-9.

9 Chapter 1: Introduction

Blessings in disguise: biological benefits of prion-like mechanisms

Gregory Arthur Newby' 2 , Susan Lindquist'

1 Department of Biology, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, US 2 Whitehead Institute for Biomedical Research, 9 Cambridge Center, Cambridge MA 02142, USA 3 Howard Hughes Medical Institute, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA

Prions and amyloids are often associated with disease, but related mechanisms provide beneficial functions in nature. Prion-like mechanisms (PriLiMs) are found from bacteria to humans, where they alter the biological and physical properties of prion-like proteins. We have proposed that prions can serve as heritable bet-hedging devices for diversifying microbial phenotypes. Other, more dynamic proteinaceous complexes may be governed by similar self-templating conformational switches.

Additional PriLiMs continue to be identified and many share features of self-templating protein structure (including amyloids) and dependence on chaperone proteins.

Here, we discuss several PriLiMs and their functions, intending to spur discussion and collaboration on the subject of beneficial prion-like behaviors.

Defining prions, amyloids, and similar phenomena

Prions have been defined as 'infectious proteins' that can assume a profoundly altered conformation and propagate that conformation in a self-templating process. The mammalian prion protein PrP is the founding example of such self-propagating conformations and is the only established prion that is infectious to humans. The best

10 characterized prion proteins are found in fungi, where their self-propagating states are transmitted to mating partners and progeny as epigenetic elements of inheritance. A highly sophisticated system of remodeling factors ensures that the prion template is divided into oligomeric prion seeds that are inherited with very high fidelity1. Most of these prions form an unusually stable aggregate structure known as an amyloid fiber, which is typically defined by three characteristics: (i) a structure consisting of beta strands running perpendicular to the axis of the fiber, resulting in a stereotypical cross- beta diffraction pattern; (ii) high stability characterized by resistance to denaturation by heat and sodium dodecyl sulfate (SDS); and (iii) binding to hydrophobic dyes such as thioflavin T and Congo Red. Owing to their unique physical properties, nature has also made extensive and diverse use of amyloids ranging from bacterial biofilm components to melanin scaffolding in humans 2 5 .

However, not all related phenomena fit squarely into the categories of prions and amyloids. Several mechanisms have been described as 'prion-like', meaning that an initial conformational change of a protein can template the conversion of other proteins to a similar or identical conformation6 -. Unlike bona fide prions, these need not be transmissible between individuals. Some mechanisms have been called 'amyloid-like'

(see Glossary), meaning that they have some but not all of the properties of amyloids9-

11. Amyloid and amyloid-like aggregates are subsets of prion-like phenomena because they template soluble proteins to adopt their fold as proteins are added to the aggregate. In this opinion, we illustrate the breadth of beneficial PriLiMs and their cognate prion-like proteins (PriLiPs) by discussing several examples of the functions

11 Chapter 1: Introduction they provide: building stable structures, signal propagation, dynamic scaffolding of ribonucleoprotein (RNP) granules, and bet-hedging in microorganisms.

Bet-hedging prions enhance phenotypic diversity and adaptation in microorganisms

Bet-hedging mechanisms are used to diversify microbial phenotypes. In fluctuating environments this allows some fraction of the population to 'win' and thrive in conditions when most would 'lose', or perish 12,13 . For example, bacterial persister cells can survive antibiotic treatment, potentially saving the population of bacteria from extinction. The cost of this mechanism is that, until they switch out of their persistence phenotype, such cells grow much slower than normal cells in the absence of antibiotics. Although antibiotics may be encountered rarely and persister cells have a severe growth defect, it is advantageous for the species to conserve this bet-hedging mechanism to survive occasional exposure to such strenuous environments

Similarly, we have found that fungal prions produce various new phenotypes that are often disadvantageous but can provide great advantages in particular

15 circumstances -19. We have proposed that such prions act as bet-hedging mechanisms: at a low frequency in a population of yeast cells, prion conformations are nucleated, resulting in a heritable, altered activity that underlies a phenotypic change.

Due to the self-propagating nature of prions and to the mechanisms that ensure their orderly distribution to progeny, prion phenotypes are heritable. Rare cells, however, switch back to the non-prion state when they lose the prion template. A recent example of a prion that confers antibiotic resistance is the yeast prion [MOD+] 2 0,2 1 (for the

12 nomenclature of yeast prions, see Box 1).

[MOD'] cells are resistant to azole-based Box 1. Yeast prions confer non-Mendelian traits and depend on chaperones to antifungals, but in rich media they have a propagate growth disadvantage. To date, bet-hedging In 1994, prion propagation was proposed to explain some perplexing, non-Mendelian phenotypes identified in yeast85. A yeast prion segregates in a non-Mendelian functions for prions have been described fashion because it is not based on a mutation in DNA inherited through chromosomes, but rather on a self- only in Saccharomyces cerevisiae. However, propagating protein conformation inherited through the cytosol. If a cell containing the prion state of a protein (a [PRION'] cell) mates with a cell containing that protein in many findings suggest that they are a nonprion state (a [prion-] cell), the non-prion proteins are rapidly templated and take on the self-propagating through the microbial world (see prion conformation. Because all meiotic progeny inherit widespread part of the parental cytosol, the vast majorify will display the prion phenotype, rather than 50% as one might have below and Box 1) and we expect that more expected if the phenotypes were based on two different alleles of a gene. We refer the interested reader to these excellent reviews on yeast prion biology31'86 87. will soon be discovered elsewhere. The [PRION*]/[prion-] nomenclature is used for all yeast prions - square brackets indicate the non- Mendelian segregation of the prion phenotype and capital Three independent studies predicted letters indicate the dominant phenotype in mating (the self-propagating conformational change); lower-case prion-like sequences in the S. cerevisiae letters designate the recessive phenotype usually associated with soluble, untemplated protein. Chaperones are intimately involved in prion 22 24 genome computationally - , one following propagation - perturbing chaperone function often results in an increased rate of prion appearance or loss (or 31 5 88 up with experimental evidence of prion-like both) ,3, . Most fungal prions rely on Hspl 0420,23, a protein disaggregase that can sever amyloid fibers and generate new ends for growth". By inhibiting this enzyme behavior23. Each study identified sets of over several generations, [prion] cells can be reliably 8 generated from a [PRION'] population 1. Hsp104 proteins that were significantly enriched for cooperates with Hsp70 and Hsp40 to exert this prion- propagating activity in a delicately balanced process that seems to have been fine-tuned to allow prion regulatory functions - transcription factors propagation8 . One prion, [GAR+], does not appear to result from an amyloid conformation and is not dependent on Hsp104, but still requires Hsp70 to propagate into and RNA-binding 52 proteins. Importantly, daughter cells . Homologs of all of these chaperones are found because these proteins regulate many broadly throughout many branches of life, perhaps indicating a conserved ability to propagate prions. Bacterial homologs were recently found to be capable of genes that often act cooperatively, bet- replacing yeast chaperones to propagate a prion in yeast90 and yeast prions have been successfully hedging prions involving such factors could nucleated in the bacterial cytoplasm91. Flies, worms, and plants also have Hspl04 homologs - it will be interesting to see whether these are also capable of propagating allow cells to immediately acquire complex, yeast prions or their own, endogenous PriLiPs. Mammals have no Hspl 04 homolog and had been thought to lack heritable phenotypes 15- 17. Some prion states disaggregase machinery, but recently Hspl 10 has been shown to cooperate with Hsp70 and Hsp40 to this effect 92. Although the mammalian machinery was not able to remodel the yeast prion Sup35, it may yet have similar 13 activity for PriLiMs in its native cellular context. Chapter 1: Introduction

Figure 1. Hypothesis: bet-hedging prions (00 "mm w) are adaptive and can respond to stress. Yeast prion states provide advantages in various environments 1 7,18 and prion switching increases in response to environmental stress 35 . Two types of prion-switching induction are proposed - stochastic and specific. The 'blue environment' signifies unpredictable environmental stresses in which prions are induced stochastically. This might be observed for any stress that significantly perturbs protein homeostasis and stresses the chaperone machinery involved in maintaining prion states. Note that each different prion causes a different phenotype, indicated by the color of the cell. After competition, a prion state that proved advantageous dominates the population of cells. The 'green environment' signifies an environmental stress that induces a specific prion preadapted to enhance survival in that condition. This is more likely to occur for stresses that are encountered regularly throughout the evolution of the organism. Specific prion induction has been observed for [MOT3*] in ethanol 36 and for [GAR+] in the presence of bacterial competitors52 . Note that there will generally be a low frequency of appearance and disappearance of each prion state (not depicted). may confer 'preadapted' complex phenotypes to enhance survival in environments that are encountered rarely but repeatedly, for which bet-hedging strategies are favored12 ,2 .

Other prions may act as evolutionary capacitors allowing random variation to accumulate cryptically for many generations before being tested by a small proportion of the population 26 . An example of this is the prion [PSI+], formed by a translation termination factor. The [PSI'] prion allows ribosomes to read through stop codons, uncovering previously silent genetic variation on a genome-wide level2 7. Phenotypes that provide a consistent advantage can become 'fixed' in the genome - that is, independent of the prion - by the accumulation of new mutations or by the genetic reassortment of pre-existing variation18 19.

The phenotypes produced by conformational changes in a prion protein can be compared with the phenotypes that are created by genetic mutations 28 ,29. In many cases, the prion conformations are self-propagating amyloid states that are inactive, similar to loss-of-function or null mutations in genes. Furthermore, most prions can adopt multiple amyloid conformations with different fragmentation and elongation rates.

These create prion 'strains' that have unique ratios of soluble:amyloid protein and thus different activity levels 30. These prion strains are akin to an allelic series of a gene,

14 I1

Pian siWtcng bmodlv icreased COj~eoD

&unnKoftta.t

Nomena~nfff Paon Ww"n at 60500 AewJ

zohag 4QS COMPeftm

Switc"~ of oe prwne ON

5keamlot pe JA" n"Jj',

tuning the level of a protein's activity and thus the phenotypic consequences of the prion state31 .

Depending on the genetic background and the particular prion protein involved,

S. cerevisiae prion proteins switch between prion and non-prion states at frequencies of between 10-2 and 10-7 31-34. Thus, prion-based phenotypes can be sampled much more frequently on average than loss-of-function mutations, which occur at frequencies of between 10-6 and 10-8. Furthermore, because prion inheritance depends on protein homeostasis machinery (Box 1), they might naturally switch more frequently under conditions that stress protein homeostasis; that is, when cells are not well adapted to their environment (Figure 1). This has been observed for [PS/*] 32-35 . Increased

15 Chapter 1: Introduction switching under stress could be of great advantage to a population of yeast, allowing individuals to sample multiple, potentially life-saving phenotypes when they most need them. This, in effect, changes the bets that the population of yeast has on the table. If the stress persists, the few cells that survive pass on to their progeny the protein states that saved them.

There are also cases where specific stresses induce specific prions. This is likely to occur for more predictable conditions that cells encounter regularly, allowing them to evolve a prion response that increases viability in the new environment (Figure 1).

Ethanol was observed to increase the appearance of the yeast prion [MO T3*] 36 , which derepresses anaerobic genes, whereas certain bacterial competitors could induce the yeast prion [GAR+] (Jarosz and Lindquist, unpublished), which overcomes glucose repression. In both cases, [PSI+] is not induced, so prion induction appears specific; however, the mechanisms of induction remain unknown.

Environmental adaptation via bet-hedging prions has two major advantages over adaptation through genetic mutation: (i) it allows a microbial population to have diverse, heritable, and complex responses to environmental conditions, even when the population is not large enough for substantial genetic diversity; and (ii) bet-hedging prions allow for fast reversion from a loss-of-function or 'null' prion state of a protein, when reversion from a loss-of-function mutation at the DNA level is quite rare.

To exemplify the first point, a small yeast colony growing on a plant may benefit from having some members stay attached while others detach to follow the flow of rainwater and spread the population. Prions that regulate surface adhesion may be ideal to promote colony diversity. Indeed, a wild strain of yeast was recently found to

16 adhere to agar growth medium after washing only when the translation termination factor Sup35 was in its prion conformation, the [PSI+] state 19. Additionally, the FLOI1 gene in yeast, which is a central regulator of colony morphology and adhesion, is regulated by multiple well-characterized yeast prions - URE2, CYC8/SSN6, MOT3,

SFP1, and SW1 all affect its transcription 37- 4 3 . Besides adhesion, prions confer numerous different phenotypes that vary from strain to strain and could be used to diversify small populations. Consistent with this, the growth of [PSI'] and [psi] yeast have been compared across many conditions, and often one state or the other confers a marked benefit to growth1 7,18,35.

The second advantage of bet-hedging prions, the relatively fast rate of reversion from a hypomorphic change in activity due to the prion state, derives from the frequency at which loss-of-function mutations are beneficial to organisms. By far the most common genetic mutations sampled are loss-of-function, and often these are adaptive. It may be beneficial to lose the function of a gene because of the energy cost associated with it or because new environmental conditions disfavor the original gene 44 - 4 6. However, microbial populations cannot adapt exclusively to their current environment at the expense of all others, because conditions in nature are always in flux (Figure 2).

Summer and winter, dry and wet, and nutrient-rich and nutrient-poor conditions are just a few examples of the cycles to which many organisms must have adapted to have survived to the present. At the same time, new environments with different intrinsic physical properties and changing microbial competitors are also being sampled. S. cerevisiae was recently shown to undergo such drastic environment changes as to live

17 Chapter 1: Introduction

Figure 2. Hypothesis: bet-hedging prions allow rapid phenotypic diversification, acquisition of complex traits, and facile reversion to previous phenotypes. (a) Different combinations of prion/non-prion conformations among many available prion proteins allow shuffling of heritable phenotypes. The red and blue cells indicate two possible combinations of prions, and thus heritable phenotypes, between genetically identical cells in a population. A cell will switch to a new prion state at a rather low frequency. Thus, it is possible to generate new combinations of prion states that are not present or may have previously died out. (b) Cells experience slowly oscillating environments and may benefit from resampling phenotypes that were advantageous in the past. Adaptations made through bet- hedging prions are reversed more frequently than are mutations. This could allow cells to adapt to previously encountered environments more quickly. (c) Cells frequently sample new, complex environments; for example, as different microbial competitors and surfaces are encountered. Shuffling the states of multiple prion proteins (indicated by different yeast cell colors) allows rapid phenotypic diversification enhancing the likelihood that some members of the population will adapt to and survive in each new environment. Here, the yeast sample environments progressing from leaf, to fruit, to insect, to liquid culture, each with its own set of microfauna, and different prion states dominate the population in each environment. In the next, unknown environment another combination of prion states may be advantageous. Many prion combinations may be present at a low frequency in the population before entering the environment and the stresses of a new environment may induce additional prion switching to enhance adaptation. on grapes in the summer and to survive the winter in the gut of wasps 47. A null mutation that is favorable in one environment could easily be deleterious in the next set of environments, which will consist of both familiar and novel elements, but genetic changes revert at a rather low frequency. Bet-hedging prions allow organisms rapidly to acquire and revert from loss-of-function phenotypes and other sampled traits, testing new phenotypes and resampling expression programs that were advantageous in the past (Figure 2).

Although bet-hedging prions have so far been observed only in fungi, we expect that more will soon be discovered in other microbes. The first yeast prions identified in

S. cerevisiae, Sup35 and Ure2, have domains rich in glutamine and asparagine residues (also called Q/N-rich or prion-forming domains). This unusual feature was successfully used to identify other S. cerevisiae proteins that could behave as prions and modulate the activity of a fused reporter23 (22 of the 90 tested Q/N-rich domains could do this, or 24%). To our knowledge, no screen has been conducted to search for prion-forming domains in the abundance of protozoan genomes that have recently

18 (bi Frequently-escuuntered. (a) (bj Frequently-encuuntered, predkiable envronments SumWer

Change prion states A Change phen~otypes

Env. 2 Env 3 En.w4 Epnv Env I cOMk a" uruwedctable e~nwents WIth diverta compwrt%*oe j-ro0 {- I Advanteota yeast pflof s*$ TRW I&A

become available. In 2000, Michelitsch and Weissman surveyed the 28 prokaryotic genomes that were available at the time, but found few Q/N-rich sequences compared with the content of S. cerevisiae22. However, an enormous 24% of proteins in

Plasmodium falciparum, the protozoan parasite that causes malaria, are Q/N-rich48, compared with 1.5% of S. cerevisiae proteins and 0.3% of human proteins 22 ,4 9.

Furthermore, a computational analysis found that the propensity to form amyloids increases as organism complexity decreases 50, but the only single celled organisms screened were S. cerevisiae and Paramecium tetraurelia, both eukaryotes. Clearly, a high-throughput analysis of the thousands of microbial genomes available could provide a wealth of information regarding potential bet-hedging prions.

It is important to note, however, that not all yeast prions contain Q/N-rich sequences. The Het-s prion of the fungus Podospora anserina5 l and the S. cerevisiae prion Mod520 are both able to form amyloids and propagate heritably though they lack

19 Chapter 1: Introduction any Q/N-rich domain. Furthermore, some yeast prions do not form amyloids at all - the prion [GAR-] appears to consist of a self-propagating, non-amyloid interaction between two proteins, the proton pump Pmal and the glucose signaling protein Std152. Another prion, [#], consists of a self-activating vacuolar protease 3

The evolutionary benefits of bet-hedging prions are just beginning to be explored and remain controversial. An alternative hypothesis is that the ability of many prions to form amyloids is an undesirable disease state5 4. Indeed, for essential yeast prion proteins like Sup35, some amyloid strains that have been generated by overexpression are so strong that they deplete cells of its essential activity, which kills them 55. However, even if this lethality occurs at natural expression levels, it could be an acceptable cost for the benefit of adaptability that bet-hedging prions provide to the population 15"1 6.

Throughout evolution, detrimental mutations are experienced much more frequently than beneficial ones, yet mutations remain the dominant force in evolution. It is difficult to assess the impact of prion switching over the course of evolutionary history because no direct trace is left behind. However, comparative genomics may be one method of determining how some prions have been utilized in the past5 6. Others include determining the conservation of prion-forming domains and examining snapshots of adapting cells recently taken from their natural habitat. A recent study surveying 700 wild S. cerevisiae isolates found that prions were present in at least one-third of the strains 19 . Prion loss was induced by transiently inhibiting a chaperone involved in maintaining prions. When assayed under 12 different growth conditions, prion loss frequently conferred a growth disadvantage. Thus, these prions had adaptive value. It is

20 likely that these results underestimate the number of cells that are utilizing prions in natural populations, because only a small number of conditions were tested.

Further supporting the usefulness of prions in fungi, Medina and colleagues observed broad conservation of many prion-like domains 57. The authors computationally searched through the 103 sequenced fungal genomes for homologs of 29 Q/N-rich proteins that can function as prions in S. cerevisiae23 . Strikingly, more than 99% of the fungi have at least a few homologous proteins containing Q/N-rich domains - only one distant relative lacked any such homolog. It remains to be shown whether these fungal prion-like domains function as bet-hedging prions or as another kind of prion, or whether their behavior is not prion-like at all. However, several of the Sup35 homologs were able to propagate the [PSI+] prion in S. cerevisiae58-60. It seems likely that prions are widely used as bet-hedging devices throughout fungi and in other branches of life.

Bet-hedging strategies like this may or may not be employed by more complex, multicellular organisms. These provide a specialized and more stable environment (or niche) for most cells and typically produce fewer progeny. Nevertheless, many other uses for PriLiMs have been identified, several of which we discuss below.

Amyloid-based PriLiMs have useful physical properties

Some PriLiMs composed of self-templating amyloids are highly regulated and are activated reliably in response to particular signals. These functional protein complexes do not act as genetic elements. Some are used for the physical properties that an amyloid fiber provides: scaffolding meshworks, coating surfaces, or binding to pigments.

21 Chapter 1: Introduction

These phenomena have been well reviewed elsewhere as types of functional amyloid 2 -5 and we only briefly mention their functions here.

In microorganisms, the physical properties of extracellular amyloids have been used to alter cellular interactions with surfaces. Diverse bacteria use amyloid fibers as a component of biofilms, which help to accumulate nutrients and protect bacteria from harsh conditions6 1,6 2 . It was recently proposed that cell-surface proteins in yeast also mediate biofilm attachment and function as amyloids 6 3. Both bacteria and fungi are able to coat themselves with amyloid fibers made of proteins called chaplins and class I hydrophobins, respectively64. These proteins can enhance attachment of the microbe to a host or allow it to escape an aqueous environment and spread spores through the air.

PriLiMs used for their physical properties are also found in metazoa. Insects and fish use amyloid fibers as eggshell components 2. In humans, Pmel17 forms amyloid fibers that bind toxic melanin precursors and scaffold their polymerization in melanosomes, which are subsequently transferred to surrounding cells 65 . Recently, various hormone peptides were found to be stored in an amyloid state in mammalian pituitary secretory granules 66. The widespread use of these PriLiMs establishes amyloid formation as a common structural state that, when adopted, alters the physical properties of proteins.

Figure 3. Prion-like mechanisms (PriLiMs) can alter the biological properties of a protein. (a) Prion-like assemblies may alter protein-protein interactions. Mitochondrial antiviral signaling (MAVS) protein, on the surface of mitochondria, interacts with tumor necrosis factor (TNF) receptor-associated factors (TRAFs) after prion-like aggregation . (b) Other proteins gain catalytic function when they assemble into amyloid. Here, RIP1 and RIP3 are depicted as inactive kinases that are activated on assembly. This activity is thought to be in part due to enhanced auto- and cross-phosphorylation in the 6 69 assembled form, which is prevented by other factors before assembly , . The kinase image was adapted from PDB entry 2J21 for purely illustrative purposes.

22 "11

Stable PriLiMs as a part of biological signaling cascades

Prion-like aggregation can also alter biological activity, changing interactions with other macromolecules. Several phenomena have recently been described in which prion-like aggregation is used to propagate a biological signal, providing a gain of function for the constituent protein or proteins (Figure 3).

Two such PriLiMs are involved in antiviral signaling. The first mechanism involves a templated conformational change to a fibrous state of the human mitochondrial antiviral signaling (MAVS) protein on the surface of mitochondria6.

The initial conformational switch appears to be templated by the RIG-1 protein when it binds to double stranded viral RNA in the cytoplasm. In its assembled form, MAVS interacts with tumor necrosis factor (TNF) receptor-associated factors (TRAFs) and

23AsmmbI MAW

23 Chapter 1: Introduction propagates a signal that results in the induction of type I interferons and other antiviral molecules 6. The second mechanism can be triggered by Vaccinia virus, which inhibits caspases to prevent the host cell from undergoing apoptosis 67-69. When this happens, another cellular death mechanism is deployed. The cellular kinases RIP1 and RIP3 interact and rapidly form amyloid fibers 68. In the amyloid state, the kinase domains of

RIP1/3 are activated and phosphorylate downstream targets to cause programmed necrosis of the cell and an inflammatory response in the surrounding tissue68 70 . Such signaling PriLiMs may be used at key steps in antiviral responses, because viruses might have more difficulty evolving mechanisms to interfere with self-templating amyloid assembly than with signaling cascades, which are inherently reversible. Such mechanisms are not likely to be restricted to mammals

Another signaling PriLiM is the self-perpetuating conformation of cytoplasmic polyadenylation element-binding protein (CPEB) from the neurons of the sea slug

Aplysia. In its non-prion state, CPEB binds and inhibits the translation of mRNAs that are involved in building stable synapses 71. The repeated stimulation of neurons with the learning-associated neurotransmitter serotonin causes the assembly of CPEB into an amyloid state. CPEB gains activity in this form, enhancing the translation of target mRNAs. This plays a major role in strengthening and stabilizing synaptic boutons for

7 73 long-term potentiation 2, . The Drosophila homolog Orb2A also forms oligomers in neurons that are required for the stabilization of long-term but not short-term memory.

Removal of the prion-like domain in Orb2A abolishes long-term memory. Mammals also express several CPEB proteins that contain Q-rich domains in neurons, but whether prion conversion contributes to memory in mammals is not yet established74. Certainly,

24 a self-perpetuating PriLiM such as CPEB seems an ideal way to perpetuate the memory of stimulation for long periods of time, with the large size of the complex keeping it local and synapse specific.

Astonishingly, when neuronal Aplysia CPEB was expressed in yeast, it readily assembled into a heritable, prion-like state7,7 5. The activity of the CPEB increased in this prion-like state, as it does in neurons, activating the translation of target mRNAs containing its recognition sequence - a cytoplasmic polyadenylation element. This demonstrates that stable PriLiPs from other organisms, even ones that are present only in differentiated, non-dividing cells, can be propagated indefinitely as prions in yeast.

Using yeast as a model for these mechanisms could be of great advantage for studying phenomena from less genetically tractable organisms.

Like Aplysia CPEB, some endogenous yeast prions may have altered function rather than simply decreased function in the prion state. The [ISP'] prion does not confer the same phenotypes found in Asfpl strains, but rather the additional phenotype of nonsense antisuppression76 .

It is unlikely that stable PriLiMs are used exclusively for either their physical properties or signaling, but rather for a combination of both. An interesting avenue for future research is to determine how the physical structure of amyloids may help to scaffold the interactions of signaling PriLiPs and how amyloids that are used for their physical properties, such as CsgA in biofilms, may alter their interactions with binding partners upon assembly.

25 Chapter 1: Introduction

Dynamic PriLiMs help to form reversible RNP granules

Prion-like domains are also involved in the assembly of dynamic RNP granules that process and modify RNA. Although it has been known for some time that Q/N-rich, Q- rich, or other low-complexity domains are essential for forming some RNP granules 8,77,78, how these large assemblies are regulated and structured remains elusive. Unlike amyloids, stress granules are composed of many different proteins that can undergo rapid exchange with the cytoplasm 77,79. Recently, a clue to this puzzle was found by Kato, Han, and colleagues. Even with no RNA present, many RNPs could be precipitated together from mammalian cell extracts using a crystalline compound that is thought to mimic the surface of a cross-beta sheet9,80 . The retention of GFP-tagged protein in a hydrogel composed of the RNA-binding protein FUS provided an in vitro assay for interactions between these low-complexity sequences. The FUS fibers comprising the hydrogels were amyloid like as assessed by their stereotypical diffraction pattern and appearance by electron microscopy. However, unlike amyloids these assemblies could incorporate different proteins, were rapidly reversible, and were not

SDS resistant. Thus, concerted, templated conformational changes among different low-complexity domains could be the basis of RNP granule formation.

Such a mechanism is prion-like in that one protein templates another to fold into the same basic structure, but differs from other PriLiMs because it is much more dynamic, perhaps allowing the segregation of interacting domains into a 'liquid' or gel- like phase separated from the rest of the cytosol8 1,8 2 . Phosphorylated FUS monomers no longer interact with the assembled FUS hydrogel, suggesting that assembly could be regulated by post-translational modification 0 .

26 A screen of Q/N-rich domains in yeast identified several RNP granule components with domains that could act as yeast prions and perhaps have bet-hedging functions 2 3 ,8 3 . Nrpl, Pub1, and Hrpl, which associate with yeast stress granules, and

Lsm4, which contributes to P body formation, could all form amyloid-like fibers and propagate the activity state of a fused reporter 23. Notably, like FUS fibers, Hrpl fibers were not SDS resistant. The physical state of these yeast proteins in such RNP granules remains to be determined, but they may assemble in a dynamic fashion. If, instead of forming such reversible assemblies, a small fraction of the cellular population inactivates these RNA-binding proteins by nucleating an amyloid, it might serve as a bet-hedging mechanism to diversify cellular phenotypes.

RNP granules are found broadly throughout eukaryotes - some regulate RNAs spatiotemporally in gametes and embryos, whereas others are used to transport RNA down neuronal dendrites 79. How these dynamic complexes are assembled and regulated in vivo at a molecular level remains largely unknown and will be a fascinating avenue of future research.

Concluding remarks

We have discussed several biological functions that PriLiMs have in nature. It is likely that many more PriLiMs await discovery in diverse cellular pathways. In Caenorhabditis elegans, 1% of proteins have Q/N-rich, prion-like domains and in Drosophila the fraction is even greater at 3.5%22. Some might function as stable or dynamic PriLiMs and a few may even have bet-hedging functions. The yeast prions [GAR+], [Het-s], and [MOD+] demonstrate that even proteins without canonical prion-like domains can function as

27 Chapter 1: Introduction prions. The real number of self-templating PriLiMs functioning in nature may be much greater than we can currently predict by sequence.

Despite the diversity of PriLiMs, some basic principles are likely to be shared. For example, they may all take advantage of the cell's core protein homeostasis machinery.

The S. cerevisiae prion proteins investigated to date all depend on Hspl04 and/or

20 2 3 4 95 84 Hsp70 , , , 2, . MAVS aggregation in extracts from human cells appears to be dependent on Hsp90 6 and mammalian stress granule regulation involves Hsp70 and

perhaps other chaperones 8. Aplysia CPEB is readily propagated in yeast, where it forms a yeast prion and is also subject to chaperone activity7. These connections to protein

homeostasis may make them intrinsically responsive to diverse internal and extracellular conditions. This, however, is clear: prion-like mechanisms are not restricted to disease, but are broadly used for the benefit of life.

Acknowledgements

The authors thank Daniel Jarosz, Randal Halfmann, Isaac Oderberg, Kevin

Knockenhauer, and members of the Lindquist laboratory for helpful discussion and

critical reading of the manuscript. They thank Tom DiCesare for helping to produce figures and for training on graphical software. S.L. is an investigator of the Howard

Hughes Medical Institute. G.N. is supported by a fellowship from the National

Science Foundation.

28 References

1. Byrne, L. J. etal. The number and transmission of [PSI] prion seeds (Propagons) in the yeast Saccharomyces cerevisiae. PLoS One 4, e4670 (2009). 2. Fowler, D. M., Koulov, A. V, Balch, W. E. & Kelly, J. W. Functional amyloid--from bacteria to humans. Trends Biochem. Sci. 32, 217-24 (2007). 3. Maury, C. P. J. The emerging concept of functional amyloid. J. Intern. Med. 265, 329-34 (2009). 4. Otzen, D. Functional amyloid: Turning swords into plowshares. Prion 4, 256-264 (2010). 5. Shewmaker, F., McGlinchey, R. P. & Wickner, R. B. Structural insights into functional and pathological amyloid. J. Biol. Chem. 286, 16533-40 (2011). 6. Hou, F. et al. MAVS forms functional prion-like aggregates to activate and propagate antiviral innate immune response. Cell 146, 448-61 (2011). 7. Si, K., Lindquist, S. & Kandel, E. R. A Neuronal Isoform of the Aplysia CPEB Has Prion-Like Properties. Cell 115, 879-891 (2003). 8. Gilks, N. et al. Stress Granule Assembly Is Mediated by Prion-like Aggregation of TIA-1. Mol. Biol. Cell 15, 5383-5398 (2004). 9. Kato, M. et al. Cell-free formation of RNA granules: low complexity sequence domains form dynamic fibers within hydrogels. Cell 149, 753-67 (2012). 10. Majumdar, A. et al. Critical role of amyloid-like oligomers of Drosophila Orb2 in the persistence of memory. Cell 148, 515-529 (2012). 11. Adda, C. G. et al. Plasmodium falciparum merozoite surface protein 2 is unstructured and forms amyloid-like fibrils. Mol. Biochem. Parasitol. 166, 159-71 (2009). 12. King, 0. & Masel, J. The evolution of bet-hedging adaptations to rare scenarios. Theor. Popul. Biol. 72, 560-575 (2007). 13. Kussell, E. & Leibler, S. Phenotypic diversity, population growth, and information in fluctuating environments. Science 309, 2075-8 (2005). 14. Kussell, E., Kishony, R., Balaban, N. Q. & Leibler, S. Bacterial persistence: a model of survival in changing environments. Genetics 169, 1807-14 (2005). 15. Halfmann, R. & Lindquist, S. Epigenetics in the extreme: prions and the inheritance of environmentally acquired traits. Science (80-.). 330, 629-632 (2010). 16. Halfmann, R., Alberti, S. & Lindquist, S. Prions, protein homeostasis, and phenotypic diversity. Trends Cell Biol. 20, 125-133 (2010).

29 Chapter 1: Introduction

17. True, H. & Lindquist, S. A yeast prion provides a mechanism for genetic variation and phenotypic diversity. Nature 407, 477-483 (2000). 18. True, H., Berlin, I. & Lindquist, S. Epigenetic regulation of translation reveals hidden genetic variation to produce complex traits. Nature 431, 184-187 (2004). 19. Halfmann, R. et al. Prions are a common mechanism for phenotypic inheritance in wild yeasts. Nature 482, 363-368 (2012). 20. Suzuki, G., Shimazu, N. & Tanaka, M. A yeast prion, Mod5, promotes acquired drug resistance and cell survival under environmental stress. Science 336, 355-9 (2012). 21. Tanaka, G. S. and M. Expanding the yeast prion world: Active prion conversion of non-glutamine/asparagine-rich Mod5 for cell survival. Prion 7, (2013). 22. Michelitsch, M. D. & Weissman, J. S. A census of glutamine/asparagine-rich regions: implications for their conserved function and the prediction of novel prions. PNAS 97, 11910-5 (2000). 23. Alberti, S., Halfmann, R., King, 0., Kapila, A. & Lindquist, S. A systematic survey identifies prions and illuminates sequence features of prionogenic proteins. Cell 137,146-58 (2009). 24. Harrison, P. M. & Gerstein, M. A method to assess compositional bias in biological sequences and its application to prion-like glutamine/asparagine-rich domains in eukaryotic proteomes. Genome Biol. 4, R40 (2003). 25. Levy, S. F. & Siegal, M. L. in Evolutionary Systems Biology (ed. Soyer, 0. S.) 751, 431-452 (Springer New York, 2012). 26. Masel, J. Evolutionary capacitance may be favored by natural selection. Genetics 170,1359-71(2005). 27. Griswold, C. K. & Masel, J. Complex adaptations can drive the evolution of the capacitor [PSI], even with realistic rates of yeast sex. PLoS Genet. 5, e1000517 (2009). 28. Chernoff, Y. 0. Mutation processes at the protein level: Is Lamark back? Mutat. Res. - Rev. Mutat. Res. 488, 36-64 (2001). 29. Uptain, S. M. & Lindquist, S. Prions as protein-based genetic elements. Annu. Rev. Microbiol 56, 703-41 (2002). 30. Tanaka, M., Collins, S. R., Toyama, B. H. & Weissman, J. S. The physical basis of how prion conformations determine strain phenotypes. Nature 442, 585-9 (2006). 31. Liebman, S. W. & Chernoff, Y. 0. Prions in yeast. Genetics 191,1041-1072 (2012).

30 32. Sideri, T. C., Koloteva-Levine, N., Tuite, M. F. & Grant, C. M. Methionine oxidation of Sup35 protein induces formation of the [PSI+] prion in a yeast peroxiredoxin mutant. J. Biol. Chem. 286, 38924-31 (2011). 33. Allen, K. D., Chernova, T. a, Tennant, E. P., Wilkinson, K. D. & Chernoff, Y. 0. Effects of ubiquitin system alterations on the formation and loss of a yeast prion. J. Biol. Chem. 282, 3004-13 (2007). 34. Lancaster, A. K., Bardill, J. P., True, H. L. & Masel, J. The Spontaneous Appearance Rate of the Yeast Prion [PSI+] and Its Implications for the Evolution of the Evolvability Properties of the [PSI+] System. 184, 393-400 (2010). 35. Tyedmers, J., Madariaga, M. L. & Lindquist, S. Prion switching in response to environmental stress. PLoS Biol. 6, 2605-2613 (2008). 36. Holmes, D. L., Lancaster, A. K., Lindquist, S. & Halfmann, R. Heritable remodeling of yeast multicellularity by an environmentally responsive prion. Cell 153, 153-165 (2013). 37. Brtckner, S. & M6sch, H.-U. Choosing the right lifestyle: adhesion and development in Saccharomyces cerevisiae. FEMS Microbiol. Rev. 36, 25-58 (2012). 38. St'ovifek, V., Vechov, L. & Palkov, Z. Yeast biofilm colony as an orchestrated multicellular organism. Commun. lntegr. Biol. 5, 203-5 (2012). 39. Teixeira, M. C. et al. The YEASTRACT database: a tool for the analysis of transcription regulatory associations in Saccharomyces cerevisiae. Nucleic Acids Res. 34, D446-51 (2006). 40. Abdulrehman, D. et al. YEASTRACT: providing a programmatic access to curated transcriptional regulatory associations in Saccharomyces cerevisiae through a web services interface. Nucleic Acids Res. 39, D136-40 (2011). 41. Monteiro, P. T. et al. YEASTRACT-DISCOVERER: new tools to improve the analysis of transcriptional regulatory associations in Saccharomyces cerevisiae. Nucleic Acids Res. 36, D132-6 (2008). 42. Barrales, R. R., Jimenez, J. & lbeas, J. I. Identification of novel activation mechanisms for FLO1 1 regulation in Saccharomyces cerevisiae. Genetics 178, 145-56 (2008). 43. Mao, X., Nie, X., Cao, F. & Chen, J. Functional analysis of ScSwil and CaSwil in invasive and pseudohyphal growth of Saccharomyces cerevisiae. Acta Biochim Biophys Sin 41, 594-602 (2009). 44. Olson, M. V. When Less Is More: Gene Loss as an Engine of Evolutionary Change. Am. J. Hum. Genet. 64, 18-23 (1999). 45. Morris, J. J., Lenski, R. E. & Zinser, E. R. The Black Queen Hypothesis: Evolution of dependencies through adaptive gene loss. MBio 3, (2012).

31 Chapter 1: Introduction

46. Gore, J., Youk, H. & van Oudenaarden, A. Snowdrift game dynamics and facultative cheating in yeast. Nature 459, 253-6 (2009). 47. Stefanini, I. et al. Role of social wasps in Saccharomyces cerevisiae ecology and evolution. PNAS 109, 13398-403 (2012). 48. Singh, G. P. et al. Hyper-expansion of asparagines correlates with an abundance of proteins with prion-like domains in Plasmodium falciparum. Mol. Biochem. Parasitol. 137, 307-19 (2004). 49. Osherovich, L. Z. & Weissman, J. S. The Utility of Prions. Dev. Cell 2, 143-151 (2002). 50. Tartaglia, G. G., Pellarin, R. & Cavalli, A. Organism complexity anti-correlates with proteomic b-aggregation propensity. Protein Sci. 14, 2735-2740 (2005). 51. Saupe, S. J. A Short History of Small s. Prion 1, 110-115 (2007). 52. Brown, J. C. S. & Lindquist, S. A heritable switch in carbon source utilization driven by an unusual yeast prion. Genes Dev. 23, 2320-32 (2009). 53. Roberts, B. T. & Wickner, R. B. Heritable activity: a prion that propagates by covalent autoactivation. Genes Dev. 17, 2083-7 (2003). 54. Kelly, A. C., Shewmaker, F. P., Kryndushkin, D. & Wickner, R. B. Sex, prions, and plasmids in yeast. PNAS (2012). doi:10.1 073/pnas. 1213449109 55. McGlinchey, R. P., Kryndushkin, D. & Wickner, R. B. Suicidal [PSI+] is a lethal yeast prion. PNAS 108, 5337-41 (2011). 56. Giacomelli, M. G., Hancock, A. S. & Masel, J. The conversion of 3' UTRs into coding regions. Mol. Biol. Evol. 24, 457-64 (2007). 57. Medina, E. M., Jones, G. W. & Fitzpatrick, D. A. Reconstructing the fungal tree of life using phylogenomics and a preliminary investigation of the distribution of yeast prion-like proteins in the fungal kingdom. J Mol Evol 73, 116-133 (2011). 58. Santoso, A., Chien, P., Osherovich, L. Z. & Weissman, J. S. Molecular basis of a yeast prion species barrier. Cell 100, 277-88 (2000). 59. Nakayashiki, T., Ebihara, K., Bannai, H. & Nakamura, Y. Yeast [PSI+] 'prions' that are crosstransmissible and susceptible beyond a species barrier through a quasi- prion state. Mol. Cell 7, 1121-30 (2001). 60. Afanasieva, E. G., Kushnirov, V. V, Tuite, M. F. & Ter-Avanesyan, M. D. Molecular basis for transmission barrier and interference between closely related prion proteins in yeast. J. Biol. Chem. 286, 15773-80 (2011). 61. Barnhart, M. M. & Chapman, M. R. Curli biogenesis and function. Annu. Rev. Microbiol. 60, 131-47 (2006).

32 62. Blanco, L. P., Evans, M. L., Smith, D. R., Badtke, M. P. & Chapman, M. R. Diversity, biogenesis and function of microbial amyloids. Trends Microbiol. 20, 66-73 (2012). 63. Ramsook, C. B. et al. Yeast cell adhesion molecules have functional amyloid- forming sequences. Eukaryot. Cell 9, 393-404 (2010). 64. Gebbink, M. F. B. G., Claessen, D., Bouma, B., Dijkhuizen, L. & W6sten, H. a B. Amyloids--a functional coat for microorganisms. Nat. Rev. Microbiol. 3, 333-41 (2005). 65. Fowler, D. M. et al. Functional amyloid formation within mammalian tissue. PLoS Biol. 4, e6 (2006). 66. Maji, S. K. et al. Functional amyloids as natural storage of peptide hormones in pituitary secretory granules. Science 325, 328-32 (2009). 67. Chan, F. K.-M. et al. A role for tumor necrosis factor receptor-2 and receptor- interacting protein in programmed necrosis and antiviral responses. J. Biol. Chem. 278, 51613-21 (2003). 68. Li, J. et al. The RIP1/RIP3 Necrosome Forms a Functional Amyloid Signaling Complex Required for Programmed Necrosis. Cell 150, 339-350 (2012). 69. Cho, Y. S. et al. Phosphorylation-driven assembly of the RIP1-RIP3 complex regulates programmed necrosis and virus-induced inflammation. Cell 137, 1112- 23 (2009). 70. Moquin, D. & Chan, F. K.-M. The molecular regulation of programmed necrotic cell injury. Trends Biochem. Sci. 35, 434-41 (2010). 71. Fernbndez-Miranda, G. & M6ndez, R. The CPEB-family of proteins, translational control in senescence and cancer. Ageing Res. Rev. 11, 460-472 (2012). 72. Miniaci, M. C. et al. Sustained CPEB-dependent local protein synthesis is required to stabilize synaptic growth for persistence of long-term facilitation in Aplysia. Neuron 59, 1024-36 (2008). 73. Si, K., Choi, Y. B., White-Grindley, E., Majumdar, A. & Kandel, E. R. Aplysia CPEB Can Form Prion-like Multimers in Sensory Neurons that Contribute to Long-Term Facilitation. Cell 140, 421-435 (2010). 74. Theis, M., Si, K. & Kandel, E. R. Two previously undescribed members of the mouse CPEB family of genes and their inducible expression in the principal cell layers of the hippocampus. Proc. Nat/. Acad. Sci. U. S. A. 100, 9602-7 (2003). 75. Heinrich, S. U. & Lindquist, S. Protein-only mechanism induces self-perpetuating changes in the activity of neuronal Aplysia cytoplasmic polyadenylation element binding protein (CPEB). PNAS 108, 2999-3004 (2011).

33 Chapter 1: Introduction

76. Rogoza, T. et al. Non-Mendelian determinant [ISP+] in yeast is a nuclear-residing prion form of the global transcriptional regulator Sfpl. PNAS 107, 10573-7 (2010). 77. Thomas, M. G., Loschi, M., Desbats, M. A. & Boccaccio, G. L. RNA granules: the good, the bad and the ugly. Cell. Signal. 23, 324-34 (2011). 78. Decker, C. J., Teixeira, D. & Parker, R. Edc3p and a glutamine/asparagine-rich domain of Lsm4p function in processing body assembly in Saccharomyces cerevisiae. J. Cell Biol. 179, 437-49 (2007). 79. Buchan, J. R. & Parker, R. Eukaryotic stress granules: the ins and outs of translation. Mol. Cell 36, 932-41 (2009). 80. Han, T. W. et al. Cell-free formation of RNA granules: bound RNAs identify features and components of cellular assemblies. Cell 149, 768-79 (2012). 81. Weber, S. C. & Brangwynne, C. P. Getting RNA and protein in phase. Cell 149, 1188-91 (2012). 82. Brangwynne, C. P. et al. Germline P granules are liquid droplets that localize by controlled dissolution/condensation. Science 324, 1729-32 (2009). 83. Buchan, J. R., Muhlrad, D. & Parker, R. P bodies promote stress granule assembly in Saccharomyces cerevisiae. J. Cell Biol. 183, 441-55 (2008). 84. Osherovich, L. Z. & Weissman, J. S. Multiple Gln/Asn-rich prion domains confer susceptibility to induction of the yeast [PSI+] prion. Cell 106, 183-194 (2001). 85. Wickner, R. B. [URE3] as an altered URE2 protein: evidence for a prion analog in Saccharomyces cerevisiae. Science (80-.). 264, 566-9 (1994). 86. Crow, E. T. & Li, L. Newly identified prions in budding yeast, and their possible functions. Semin Cell Dev Biol 22, 452-459 (2011). 87. Tuite, M. F. & Serio, T. R. The prion hypothesis: from biological anomaly to basic regulatory mechanism. Nat. Rev. Mol. Cell Biol. 11, 823-33 (2010). 88. Winkler, J., Tyedmers, J., Bukau, B. & Mogk, A. Chaperone networks in protein disaggregation and prion propagation. J. Struct. Biol 179, 152-60 (2012). 89. Park, Y.-N. et al. Differences in the curing of [PSI+] prion by various methods of Hsp104 inactivation. PLoS One 7, e37692 (2012). 90. Reidy, M., Miot, M. & Masison, D. C. Prokaryotic Chaperones Support Yeast Prions and Thermotolerance and Define Disaggregation Machinery Interactions. Genetics 192, 185-193 (2012). 91. Garrity, S. J., Sivanathan, V., Dong, J., Lindquist, S. & Hochschild, A. Conversion of a yeast prion protein to an infectious form in bacteria. PNAS 107, 10596-10601 (2010).

34 92. Shorter, J. The mammalian disaggregase machinery: Hspl 10 synergizes with Hsp70 and Hsp40 to catalyze protein disaggregation and reactivation in a cell-free system. PLoS One 6, e26319 (2011).

35 Chapter 1: Introduction

36 Chapter 2

A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

This chapter previously submitted for publication. Authors: Gregory A. Newby, Szilvia Kiriakov, Can Kayatekin, Erinc Hallacli, Peter Tsvetkov, Christopher P. Mancuso, J. Maeve Bonner, William R. Hesse, Sohini Chakrabortee, Anita L. Manogaran, Susan W. Liebman, Susan Lindquist, Ahmad S. Khalil

AUTHOR CONTRIBUTIONS G.A.N., S.K., S.L., and A.S.K. conceived of the study. G.A.N. and S.K. performed and analyzed all experiments with the exception of the following. C.K., E.H., P.T., C.P.M., and S.C. worked with G.A.N. and S.K. to develop and optimize yTRAP methodology. C.P.M. assisted with the development of reverse-yTRAP. J.M.B. aided in the development of expression systems. W.R.H. and C.K. developed the Htt-induction system. A.L.M. and S.W.L. developed the [PSI+] prion variant strains. S.L. and A.S.K. oversaw the study. G.A.N., S.K., and A.S.K. wrote the paper with input from all authors.

37 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

SUMMARY

Protein aggregation is a hallmark of many diseases, but also underlies a wide range of positive cellular functions. Despite its importance, tools to probe protein aggregation quantitatively in living cells have been elusive. Here we develop a versatile genetic sensor that transmits the aggregation state of proteins to synthetic transcriptional outputs. We use the sensor to quantify protein aggregation associated with human disease and to identify modulators of the self-templating, heritable states of yeast prions. We design a hyper-inducing prion fusion that can be used to encode synthetic memories in yeast, which we demonstrate by engineering cells to measure and remember heat. Using high-throughput screens enabled by our sensor, we identify mutant prion alleles that cure endogenous prions. We use these prion-curing alleles to create "anti-prion drive" systems that reverse the dominant, non-Mendelian inheritance of prions and, in some cases, eliminate them from yeast populations. Our work establishes a quantitative, high-throughput, and generalizable toolkit to study and control diverse protein aggregation processes in cells.

INTRODUCTION

Protein aggregation is a frequent occurrence, and we are only beginning to understand its diverse functions and dysfunctions. Cellular aggregates are associated with a staggering number of human diseases, including neurodegeneration', type 11 diabetes 2, systemic amyloidosis 3, and aging4. It is also becoming increasingly clear that protein aggregates function in a wide range of indispensable biological processes. These

38 include signal transduction in the human immune system 5-7, synaptic regulation and memory 8, RNA regulation in the cellular stress response9- 11, melanin production 12, and microbial biofilm formation 1 .

Broadly, aggregation is the assembly of proteins into large complexes that can grow indefinitely. Aggregate structures can range from amorphous, those that have little or no regular structure, to those that are ordered, such as the characteristic cross-beta- sheet structure of amyloids. This conformational diversity, coupled with the diversity of components that can be contained in a single aggregate as well as the variation in aggregate size, renders typical structural techniques inapplicable. Correspondingly, available methods to describe protein aggregation in vivo are largely qualitative and ad hoc. Our current understanding of this emerging biology is limited due to a lack of tools to interrogate protein aggregation phenomena. Such technology is urgently required to enable quantitative and high-throughput studies to elucidate the causes and consequences of aggregation, including cell-based screens for regulators of functional aggregation or therapies to treat diseases associated with aggregates.

Since their discovery as the causative agents of transmissible neurodegenerative disorders, prions have served as model protein aggregation systems. Prions are self- propagating protein conformations found across biological organisms, from bacteria to mammals. Much of what we know about these elements comes from studies of yeast prions. Most known prion proteins form amyloid aggregates that template the amyloid conformation onto newly synthesized proteins. "Switching" between the soluble, non- prion conformation and the prion aggregate in yeast is rare15-17, but can result in a dramatic change in activity that leads to new cellular phenotypes that are faithfully

39 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives transmitted to progeny18-23. Because the aggregated prion conformations are transmitted through the cytoplasm, they and their associated phenotypes are inherited in a non-Mendelian fashion and do not show the 2:2 segregation pattern associated with nucleic acid based inheritance24 2 .

The role of prion aggregation in cells remains an active area of investigation. The canonical example of a yeast prion is [PS/+], which is formed by the translation termination factor Sup35 (brackets denote non-Mendelian inheritance of the prion phenotype, capital letters denote its dominant inheritance)26 . The N-terminal domain of

Sup35 is highly aggregation prone and causes the protein to switch from a soluble, functional form to a self-templating amyloid at a low frequency (usually about one in ten million cells) 16. As Sup35 enters the amyloid aggregate, the accompanying loss of soluble protein causes ribosomes to read through stop codons, uncovering previously silent genetic variation on a genome-wide level. [PS/I]-associated phenotypes are diverse and often disadvantageous, but in some environments can provide advantages1 8,19. This has led to the compelling suggestion that fungal prions may have adaptive value1 8,2 1 ,27,2 8. Indeed, the [Het-s] prion in Podospora anserina clearly has a positive function in the widespread fungal phenomenon of heterokaryon incompatibility, which may prevent the spread of infectious cytoplasmic pathogens 29,30. Tools to detect and manipulate prions in a wide variety of contexts are necessary to comprehensively explore the adaptive hypothesis, map the full nature of biological benefits they may encode, and even exploit them to engineer new cellular functionalities.

Studies with [PSI+] and other yeast prions have revealed the general molecular properties that allow prions to adopt and propagate amyloid conformations. First, a prion

40 protein must harbor a sequence that can readily adopt an amyloid fold. Most known prions contain prion-forming domains (PrDs) rich in glutamine (Q) and/or asparagine (N) residues. In silico screens for Q/N-richness have been used to identify aggregation- prone PrDs 31, including human proteins important for disease 3 2 . However, not all prions possess this sequence bias and thus could not be identified based on this requirement 2' 33. Second, to propagate the amyloid, a prion protein must be able to participate in interactions with chaperone proteins, for example, relying on Hsp104 to fragment yeast prion fibers into smaller prion "seeds" that can template new protein34 36

Based on these properties, at least ten different proteins that can act as prions have been found in S. cerevisiae2 1,3 7. However, recent explorative studies suggest that S. cerevisiae may harbor many more yet uncharacterized prion-like elements that do not necessarily adhere to these properties 20,33. Thus, determining the pervasiveness of these elements and their aggregation mechanisms will require direct, high-throughput testing.

Here, we use synthetic biology to develop a versatile genetic toolkit for sensing and controlling protein aggregation states in cells. yTRAP (yeast Transcriptional

Reporting of Aggregating Proteins), is based on coupling the solubility state of a protein- of-interest to the regulation of synthetic transcriptional outputs, producing sensitive and quantifiable aggregation-dependent signals. We first construct a panel of yTRAP sensors and apply them to quantitatively detect prion states in individual yeast as well as populations of yeast cells. Exploiting the modular synthetic design of yTRAP, systems can be developed that allow high throughput measurement of multiple prions simultaneously. We utilize yTRAP sensors to identify and characterize factors that

41 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives influence the aggregation of disease proteins and yeast prions. We construct a synthetic protein fusion that acts, to our knowledge, as the most potent genetic inducer of the

[PSI'] prion. Additionally, by developing a yTRAP-based high-throughput screen of mutant libraries, we select for mutant prion alleles that cure endogenous [PRON+] states after transient expression. Finally, we show that this genetic toolkit provides a basis for programming two new cellular functionalities using prion elements. First, we employ the hyper-inducing prion fusion to engineer synthetic memory in yeast cells. By linking heat to prion induction, a prion-encoded memory of transient temperature changes is maintained by the yeast population. Second, we construct "anti-prion drive" systems to cure prions and reverse the dominant inheritance of prion states in mating partners and progeny. Our results establish a generalizable cell-based technology to observe, screen, and manipulate protein aggregation phenomena.

RESULTS

Development of a genetic tool that sensitively detects protein aggregation states

To detect protein aggregation in living cells, we developed a genetic tool termed yTRAP

(yeast Transcriptional Reporters of Aggregating Proteins). The yTRAP cassette integrates into the Saccharomyces cerevisiae genome as a single copy in an efficient, one-step process. The cassette is composed of: (1) a 'yTRAP fusion' of the protein-of- interest with a synthetic transcriptional activator (synTA), constitutively expressed at low levels; (2) a reporter gene regulated by a synTA-responsive promoter (Figures 1A and

S1A). The system functions by linking transcriptional activation of the reporter to the

42 solubility of the yTRAP fusion. In the soluble state, the synTA binds to its cognate promoter and induces strong expression of the reporter gene. Aggregation of the fusion protein limits the availability of the synTA to regulate its promoter, thus reducing the transcriptional output of the system. The synthetic fusion can either join existing aggregates endogenous to the cell, or be the sole source of a heterologous aggregating protein. Moreover, the synTAs feature a synthetic biology design utilizing engineered zinc finger proteins that target synthetic binding sequences (BS); the modular design of these regulators enables yTRAP fusions to be readily constructed for any protein and customization of transcriptional outputs 38,39.

We first calibrated the yTRAP reporter output for different levels of soluble synTA. Using an estradiol-inducible system 40, we titrated synTA protein level in the cell and measured the corresponding reporter output using flow cytometry. This revealed a quantifiable linear relationship between soluble synTA protein levels and reporter outputs across a -50-fold dynamic range, with only 1.5% basal expression of the reporter gene in the absence of synTA (Figure S1 B).

To validate the tool, we applied yTRAP to detect the self-templating states of the well-established yeast prion, [PSI+]. We designed a sensor for the [PS-] prion by encoding a yTRAP fusion for the Sup35 NM domain, where NM refers to the N-terminal and Middle regions of Sup35 ("PSI sensor", Figure 1 B). The N-terminal domain is Q/N-

rich and forms the amyloid core of the prion, while the M region is highly charged and

important for chaperone binding41 .

Sup35, like several other prion proteins, can adopt multiple self-templating amyloid conformations. Each conformation incorporates new protein at different rates.

43 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

These conformations, or 'prion variants,' result in varying ratios of soluble:aggregated protein in the cell. Stronger variants of [PSI+] efficiently template more protein into the aggregate, and thus the cellular activity of that protein is more severely affected. We introduced the PSI sensor into three yeast strains representing variants of this prion: strong [PSI+], weak [PSI+], and [psi]. Each variant was clearly distinguishable by flow cytometry, with stronger aggregation yielding less yTRAP signal (Figure 1 B). We separately confirmed that the yTRAP fusion protein was incorporated into amyloid aggregates in cells harboring the [PSI+] prion but not in cells harboring the non-prion

[psi-] state using semi-denaturing detergent agarose gel electrophoresis (SDD-AGE), a traditional biochemical approach of separating detergent-insoluble aggregates from other cellular proteins (Figure 1 B). Because of the sensor's high degree of sensitivity and the resulting large separation between [prion-] and [PRION+] states, we found that

Figure 1. Development of a genetic tool to detect protein aggregation states in cells. (A) Schematic of the yTRAP system, which functions by transmitting the aggregation state of proteins to synthetic transcriptional outputs. The system consists of a yTRAP fusion of a gene-of-interest (GOI) to a synthetic transcriptional activator (synTA), and a reporter gene controlled by a synTA-responsive promoter. The fusion is expressed from a weak, constitutive promoter and functions as a cellular probe of protein solubility. It strongly activates the reporter in the soluble state (left), but not in an aggregated state (right). BS: 'binding site' of the synTA. (B) Top: Genetic design of the [PSI'] prion sensor. Bottom left: Histograms of the sensor output for yeast strains harboring different variants of the prion: strong [PSI+], weak [PSI+], and [psi]. Bottom right: SDD-AGE analysis confirming that the yTRAP fusion protein was converted to amyloid states in [PSI+] cells (anti-HA to detect the yTRAP fusion). (C) Fluorescence image of an agar plate showing the sensor output for yeast colonies harboring different variants of the prion: strong [PSI+] (top left), weak [PSI+] (top right), and [psi-] (bottom). (D) Top: Genetic design of the [RNQ+] prion sensor. Bottom left: Histograms showing the sensor output for [RNQ+] and [mq-] cells. Bottom right: SDD-AGE analysis confirming that the yTRAP fusion protein was converted to an amyloid state in [RNQ+] cells (anti-HA to detect the yTRAP fusion). (E) Left: Yeast strains were constructed to harbor yTRAP sensors for a panel of asparagine (N) or glutamine (Q) rich protein domains, and a construct for galactose-inducible expression of polyQ- expanded Huntingtin Exon1 (Htt-Q103). Right: Relative aggregation of the Q-rich and N-rich domains was measured in the presence of Htt-Q103 induction. Relative aggregation was obtained using yTRAP signal measured by flow cytometry. The ratio of uninduced / galactose-induced signal for each strain was determined, and normalized to control strains that lack the Htt-Q103 construct (n = 3, error bars, SD).

44 I

A SOLUBLE STATE AGGREGATED STATE [prion- ] [PRION+] aggregate formation

native or R

pro,teind REPORTER REPORTE R r OFF yRPfsion I yTRAP fusion G01 SYnTA 8x VnTA G01 synTA 8x VnTA

B [PSI+] PRION SENSOR D [RNQ+] PRION SENSOR

TRP sion RAP sion S35- synTAm s yTA BS BS

strong weak strong I weak [PSIl I[PSI+] I[psi-I [RNQJ+ Irnq-] [RNQ+I [rnq] 100- 100- 4. 80- 4 80- E 60- E 60- 0I. 0 40- 0 40- 20- 20- A

1. 2 3 A5 2 0- I 102 103 104 105 102 103 104 105 mNeonGreen (a.u.) mNeonGreen (a.u.)

C E SENSING WITH DISEASE AGGREGATES strong weak Sensors of 0-/N-rich proteins c 2.5 [PSII [Psl+] 0 * 0-rich TRAP fsion W 2 - N-rich Q-rich synTA 01) 1.5 1 N-rich synTA CD 0m 0.5 Huntingtin overexpression 0 0

pGAL d'0 [psi]

45 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives prion states could also be readily evaluated by photographing yeast colonies on agar plates (Figure 1C), fluorescence microplate readings on yeast cultures (Figure S1C), and fluorescence microscopy of single cells (Figure S1 D). Previous methods to interrogate prions have relied on the presence of detectable cellular phenotypes that are affected by aggregation of that specific protein. Generally, these prion phenotypes are

18 20 highly dependent on the genetic background of the yeast strain , . Uniquely, yTRAP allows facile, rapid, and single-cell measurement of aggregation state with no particular requirements for the genetic background. To demonstrate this, we measured the [PSI+] prion in three distinct genetic backgrounds (Figure S1 E). Each yeast strain showed a strong shift in yTRAP signal dependent on the prion state. Taken together, these results demonstrate that we have established a highly-sensitive genetic sensor that can quantitatively distinguish between prion states in diverse strain backgrounds.

yTRAP enables interrogation of diverse prion and aggregation-prone elements

The wealth of published data on [PSI+] is in large part a consequence of having a color phenotype to aid in its interrogation 42 . Finding reliable, natural phenotypes for other yeast prions has been challenging, and great effort has gone into engineering them for each individual prion. Our platform overcomes these challenges because it is based on the general principle of coupling solubility state to synthetic transcriptional outputs. This enables the rapid construction of sensors for a variety of different proteins with a single platform. To demonstrate this, we assembled a sensor for a second yeast prion,

[RNQ+], by encoding a yTRAP fusion using Rnql, the amyloid-forming protein

46 determinant of this prion ("RNQ sensor", Figure 1 D). Many wild and laboratory strains of yeast harbor the [RNQ+] prion, also known as [PIN+]. However, it has been a particularly difficult prion to track due to the complete lack of growth phenotypes conferred by the prion. The only established [RNQ+] phenotype is that its presence enhances the frequency at which other yeast prions appear. When we introduced the RNQ sensor into a [RNQ+] strain, we observed a low-fluorescence (sensor OFF) state (Figure 1 D). After curing the strain of the [RNQ+] prion by transient growth in guanidine hydrochloride, which eliminates all Hsp104-dependent prions 36 ,4 3 , the sensor's output switched to a high-fluorescence (sensor ON) state, easily distinguished from its [RNQ+] parent population. Incorporation of the yTRAP fusion protein into the amyloid state in [RNQ+] cells was confirmed by SDD-AGE analysis (Figure 1 D).

Extending this general approach, we built sensors for a panel of other yeast prion proteins: Swil, Mot3-PrD, and Newl-PrD3 1 . Importantly, with these sensors, we were able to not only detect existing prion states, but also to actively isolate newly formed prion states and subsequently observe their curing by treatment with guanidine hydrochloride (Figures S1C and S1 F). Thus, yTRAP enabled the tracking of each of these prions states as aggregation changed in both directions in live cells. These results establish a synthetic, "plug-and-play" toolkit for detecting and isolating prion states, applicable for different prions and across genetic backgrounds.

Next, we explored whether yTRAP could be used to interrogate aggregation of human disease proteins (a subject of intense research but without reliable toolkits) using the huntingtin protein associated with Huntington's disease as a model system.

Individuals with longer polyglutamine (polyQ) tracts at this locus, which results in a more

47 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives aggregation-prone protein, contract the disease with higher penetrance and at an earlier age of onset44 . We compared the solubility of yTRAP sensors for a highly polyQ- expanded huntingtin (Htt-Q103, containing 103 glutamine residues) to a shorter expansion (Htt-Q25, containing only 25 glutamine residues). Both yTRAP fusion proteins were expressed at the same level, yet yTRAP sensor outputs confirmed that

Htt-Q103 is more aggregated than Htt-Q25 (Figure S1G).

PolyQ aggregates have been shown to sequester other cellular proteins, a property that likely contributes to their toxicity45-4 8. We therefore tested the ability of yTRAP to track the aggregation of matched Q- and N-rich protein domains in the context of Htt-Q103 aggregation. We generated yTRAP sensor strains for a panel of previously-characterized Q- and N-rich yeast proteins, pairs of which differ only in Q/N content49, and measured the change in yTRAP signal in response to Htt-Q103 overexpression. Note that both Q- and N-rich proteins are aggregation prone, with N- rich proteins having a greater propensity to form amyloid 49; however, Q-rich proteins were previously observed to co-localize with Htt-Q103 aggregates by fluorescence microscopy, while N-rich aggregates did not50. As assessed by yTRAP, Q-rich proteins showed a marked shift to an aggregated state when Htt-103Q was expressed, while matched N-rich control proteins were unaffected (Figure 1 E). Our results support the model that Q-rich proteins are particularly susceptible to sequestration by polyQ- expanded huntingtin, and illustrate the potential of yTRAP to screen for factors that influence disease-relevant aggregation.

48 Designing custom yTRAP sensor systems: positive detection of aggregation

The yTRAP tool is built on programmable synthetic components. As a result, yTRAP- based genetic circuits can be constructed to produce custom sensor programs. For example, we designed a sensor system with reverse output logic, which turns ON upon aggregate formation. In the 'reverse-yTRAP' scheme, the synTA-responsive promoter controls expression of a TetR repressor (fused to the red fluorescent protein mKate2), which in turn regulates a downstream, Tet-repressible reporter (Figures S2A and S2B).

When the yTRAP fusion protein is soluble, TetR-mKate2 is expressed and represses the downstream reporter; repression is relieved by aggregation of the yTRAP fusion.

We built and introduced a RNQ reverse-yTRAP sensor into [RNQ+] and [rnq-] cells, revealing strongly separated high green / low red fluorescence states and low green / high red states, respectively (Figure S2A). This additional fluorescent output that gains signal upon aggregation will be advantageous when screening for aggregation- promoting factors.

Designing custom yTRAP sensor systems: multiplex tracking and isolation of prion states

Thus far, the study of prions in high throughput has been limited to interrogating the behavior of individual prions. This is, in part, because each yeast prion has required a particular genetic background for detection. By unlocking additional synthetic transcriptional channels, the yTRAP toolkit enables the tracking of multiple prion proteins at the single-cell level in a population of yeast cells. To do this, we designed a

49 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

'dual prion sensor' by utilizing two orthogonal pairs of synTAs / responsive promoters 38 .

The dual prion sensor consists of two cassettes, each encoding (1) a yTRAP fusion comprising a protein-of-interest fused to a distinct synTA and (2) a corresponding synTA-controlled unique reporter (Figures 2A and S2C). We used this framework to develop a dual PSI / RNQ sensor (Figure 2A). We introduced this sensor into a set of strains harboring each of the four combinations of prion and non-prion states. Each strain could be clearly differentiated by flow cytometry using two fluorescent channels

(Figure 2B). With the availability of a suite of orthogonal synTA-promoter pairs 38, we envision these yTRAP combinations can be expanded to allow greater numbers of proteins to be simultaneously monitored.

We next sought to apply the dual PSI / RNQ sensor to simultaneously track the behavior of these two prions in response to a stimulus. Specifically, we subjected an initial population of [PSI+] [RNQ+] cells harboring the sensors to guanidine hydrochloride treatment. We sampled cells at different durations of treatment and allowed them to recover in growth media lacking guanidine for a minimum of six generations to establish heritable prion states.

Figure 2. A dual prion sensor enables multiplex detection and isolation of [PSI*] and [RNQ+] states (A) Genetic design of the dual prion sensor for [PSI'] and [RNQ+]. Each prion protein is coupled to a distinct transcriptional output channel through orthogonal synTA / promoter pairs (synTA1 and synTA2, and their associated BS (binding site)) (B) Flow cytometry measurements of the dual PSI / RNQ sensor for four yeast strains harboring all four combinations of the [PSI+] and [RNQ+] prions. (C) Left: Time course measurements of dual prion sensor cells tracking the conversion from a [PSI+] [RNQ+] population to a fully cured [psi-] [mq-] population by guanidine hydrochloride treatment. Samples were collected at specified times during growth in guanidine-containing media and diluted to halt the effect of guanidine. Cells were then grown to saturation in the absence of guanidine and measured the next day by flow cytometry to quantify the population composition (n = 3, error bars, SD). Right: Overlaid fluorescence image of an agar plate showing the heterogeneous population of prion states following 16 hr of treatment with guanidine. Single colonies of any prion combination can be isolated (red colonies, [psi] [RNQ+]; green, [PSI+] [mq-]; yellow, [psi] [mq-]; no fluorescence, [PSI+] [RNQ+]).

50 We then assessed prion states using flow cytometry (Figure 2C, left). Based on this, we could readily distinguish combinations of prion / non-prion states that were sampled by the population as it evolved from entirely [PSI+] [RNQ+] to [psi] [mq-]. 16

hours of guanidine treatment yielded the maximum number of cells cured of only one prion, after which cells cured of both prions became the prevalent sub-population

(Figure 2C, left). The heterogeneous population could also be directly observed on agar

A DUAL PRION SENSOR B 15s [psi-][RNQO+ [psi-][rnc( PSI sensor a) 104

0 Sp35 synTA 1 synTA 1 NM BS a)C 10 .- - RNO sensor

E 2 [PSI+ . RNQO+ [PSIt J[rnq] RAqfusion 10 2 . .4 . . .. 102 10, 10 105 Rnq1 synTA 2 synTA 2 BS mNeonGreen fluorescence (a.u.)

C MULTIPLEX TRACKING OF PRION STATE COMBINATIONS

16 hr treatment 1001 80- 0 60- + [PSI+J[RNQ+l C- -m- [psi- [RNQ+]- 40- +-PS1+]rnq-] - #4-0 -*[psi-][rnq- OR 20- f. 012 16 20 24 36 time in guanidine treatment (hrs)

51 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

plates (Figure 2C, right). Notably, at all time points, the number of cells cured of [RNQ+]

was always greater than the number cured of [PSI+], indicating that [RNQ+] is more

readily cured by guanidine hydrochloride, as has been previously reported 51. These

results demonstrate the utility of yTRAP for detecting and selecting yeast prion

combinations.

Prion switching has been hypothesized to be a mechanism for promoting

phenotypic diversity in microbial populations 33 52 . These tools could, in principle, be

used to multiplex the observation of prion switching in screens of diverse treatments

and environments.

Synthetic cellular memory of thermal inputs using a hyper-inducing prion fusion

yTRAP transmits the heritable prion state of the cell into distinct transcriptional outputs.

We have utilized this aspect of yTRAP to enable detection and isolation of prions within

cellular populations. However, this principle can also be used to convert and store

intrinsic or extrinsic signals into stable prion states, forming the basis of a cellular

memory. To enable this, it is important to have a means by which to induce the formation of [PRION+] cells within a population, so that these induced prion state

memories can be "wired" to the signal of interest.

There currently exists no method to fully induce an amyloid prion in a yeast

population, though there are methods to increase the frequency of their appearance. In

general, prions can be induced by transiently overexpressing the proteins that encode

5 3 them ,54 . The increased concentration of prion protein enhances the frequency of

52 nucleating the prion conformation. However, this typically induces the prion in only a small percentage of the population.

A second method for increasing the frequency of prion induction is to stimulate cross-seeding. Cross-seeding occurs when a protein nucleates its prion conformation off of a different prion that already exists in the cell 55,56. Because prion aggregates are highly specific for proteins of identical sequences, cross-seeding is typically a rare event. By creating a fusion between Sup35 and Rnq1 (called "NM-Rnql"), the frequency of [PSI+] induction was shown to be increased, presumably by the stimulation of cross-seeding 57. According to the cross-seeding model, as the fusion protein is incorporated into Rnq1 amyloid, the forced proximity of the Sup35 PrD increases the likelihood of its conversion to the amyloid state. Endogenous Sup35 will then template off of the nucleated Sup35 PrD to propagate [PS/].

We reasoned that by bringing the Rnq1 sequence closer to the amyloid-forming

N domain of Sup35, we might further enhance the cross-seeding effect (Figure 3A). To test this, we engineered an "N-Rnql-M" fusion. We transiently expressed this fusion in cells harboring the dual PSI / RNQ sensor, and compared its efficiency in inducing

[PSI+] with the previously-used factors Sup35NM-Rnql (NM-Rnql) and Sup35NM (NM).

Strikingly, N-Rnql-M was able to induce [PSI] in >99% of the population, as compared with 2% of cells for NM and 10% for NM-Rnql (Figure 3B). [PSI+] induction by N-Rnql-

M was confirmed by conventional red/white assay (Figure S3A). Moreover, in a [mq-] strain with no aggregate off which to template, expression of the fusion protein did not lead to [PSI+] induction (Figure S3B).

53 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Figure 3: Designing a hyper-inducing prion fusion to encode synthetic cellular memory (A) Architecture of the hyper-inducing prion fusion, which is a fusion of the Sup35 N domain, the full Rnql protein, and the Sup35 M domain. Placing Rnql in close proximity to the N domain is thought to enable efficient cross-seeding of the [RNQ+] amyloid onto Sup35. (B) Left: Yeast strains were constructed to harbor the dual PSI / RNQ sensor and a construct for estradiol-inducible expression of [PS/l]-inducing proteins N-Rnql-M, NM-Rnql, and NM. Right: Histograms of the PSI sensor output before (uninduced) and after (induced) transient expression of prion-inducing fusions. N-Rnql-M led to the conversion of >99% of cells to the [PSI] state. In contrast, NM-Rnql and NM alone induced only -10% and 2% of cells, respectively. (C) Left: Schematic of a synthetic memory circuit triggered by thermal inputs. By placing expression of N-Rnql-M under the control of a temperature-responsive promoter (pSSA4, SSA4 promoter), the hyper-inducing fusion converts cells to the [PSI+] state at an efficiency dictated by the temperature experienced. Middle: Schematic of the thermal memory experiment. Right: Percentage of the yeast population converted to the [PSI+] state after transient exposure to different temperatures, measured by yTRAP. A control strain that expressed NM (gray bars) instead of the hyper-inducing prion fusion showed no prion induction at any temperature.

Because of the dual prion sensor, we could additionally observe the effects of the prion fusions on the [RNQ+] prion. Interestingly, transient expression of NM-Rnql led to curing of the [RNQ+] prion in a majority of cells, while N-Rnql-M fusion showed very little curing (Figure S3C). This unexpected curing of [RNQ+] by NM-Rnql, but not N-

Rnql-M, may explain their large difference in [PSI+] induction potential, and highlights the importance of tracking multiple prions at once, as prions may interact and switch together in complex ways. These results demonstrate the usefulness of our genetic platform for rapidly building and studying prion-inducing factors.

Equipped with this hyper-inducing prion fusion, we next sought to build a cellular memory device based on [PSI+] induction. We hypothesized that, by linking expression of N-Rnql-M to the presence of an environmental signal, we could transmit and heritably store signals in the prion state of cells. Using temperature as a proof-of- principle, we constructed strains that express N-Rnq1-M or NM under the control of the heat-responsive SSA4 promoter in [ps-] [RNQ+] dual prion sensor cells (Figure 3C). We subjected cultures of these sensor cells to a two-hour stimulus of elevated temperature between a range of 300C - 380C. We then recovered the cells for over 10 generations 54 A HYPER-INDUCING B [PS1i psi PRION FACTOR FUSION

PSI sensor NM Designing a hyper-inducing prion fusion

Sup35 C ST35 synTA x E Rnqi NM-Rnql ~0- Expression of fusion estradiol uninduced hyper- N-Rnql-M Induced N-Rnql-M Inducing fusion - enhanced 3 5 cross-seeding 101 10 10 PSI sensor mKate2 fluorescence (a.u.)

C PRION-ENCODED SYNTHETIC THERMAL MEMORY

PSi sensor TRANSIENT EXPOSURE TO * N-Rnq1-M ELEVATED TEMPERATURE 60- * NM control

W -5 &.- NMST5synTA 240- CL Expression of I- ~20- hyper-inducing fusion MEMORY -10 generations thermal 0 Input TIME 30 335 35 36.5 38 pSSA4 temperature (deg C) at 250C and measured the percentage of the population that switched to [PS/I] using flow cytometry. We observed that, as the temperature increased, so did the proportion of the population that had induced and maintained the [PSI'] prion (Figure 3C). In contrast, the NM-expressing control strain showed no [PSI'] induction for any temperature.

Thus, the strain harboring this cellular thermal memory device retained a memory of the temperature it had experienced long after the stimulus was removed. The

55 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives memory of that temperature can theoretically be propagated indefinitely to future progeny. By replacing the promoter controlling expression of N-Rnq1-M, other stimuli could also be encoded by the [PSI+] prion and remembered by a population of yeast cells.

A high-throughput genetic screen identifies prion-curing alleles

Development of a means to prevent aggregation or remove aggregates is of paramount importance because of the many disease-associated protein aggregates1 . yTRAP coupled with fluorescence-activated cell sorting (FACS) enables rapid high-throughput screening of pooled libraries of mutants for rare alleles that prevent aggregation in trans. To demonstrate this, we used yTRAP to screen for mutant prion alleles that cure

[PSI'] and [RNQ+] while it is being maintained by the endogenous wild type protein

(Figure 4A). As a source of mutants, we utilized a yeast variomics library, which contains a theoretical diversity of -200,000 mutants per gene 58. We mated cells containing libraries of Sup35 and Rnq1 mutants to [PSI+] and [RNQ+] yTRAP sensor strains, respectively, and performed two successive rounds of FACS enrichment for highly fluorescent cells cured of their prion state. Following the second round of selection, we isolated and sequenced the enriched plasmids (Table Si). We then tested the prion-curing mutant alleles by introducing them on inducible plasmids into a [PS/+]

[RNQ+] strain harboring the dual prion sensor. The putative prion-curing mutants were transiently expressed and the cells then recovered for more than ten generations in non-inducing media to assay for heritable prion curing.

56 We recovered many of the previously reported prion-curing alleles of Sup35 59, including Q1OR, Q15R, and Q24R (Figure 4B). We also recovered several new [PSI+]- curing alleles. Notably these new alleles are the first prion-curing alleles that do not involve the introduction of a charged residue: Q61 L and the double mutant Y46C, Q95L.

Notably, all [PS/*]-curing mutations were located in the Sup35 N domain comprising the first 123 amino acids, even though the random mutations were made throughout the entire sequence spanning 685 amino acids.

The screen for [RNQ+]-curing alleles yielded two effective mutations: A288-298, and a nonsense mutation at amino acid 313 (stop313; Figure 4C). The A288-298 deletion in Rnq1 spans an oligopeptide repeat, and was previously identified as a naturally occurring allele in some S. cerevisiae strains 60. The Yeast Genome Database contains sequences for 45 different genetic backgrounds of S. cerevisiae. Of these 45 strains, 4 of them harbored this A288-298 mutation. None of the strains harbored the stop313 mutation, nor any of the selected Sup35 mutations. Interestingly, following transient expression, both of the RNQ1 mutations caused an increase in yTRAP signal that did not overlap with the [rnq-] control (Figure 4C). These mutants heritably remodeled the prion to a more soluble state. A small percentage of the population that had overexpressed Rnq1 stop313 were cured of [RNQ'], indicated by a small high- fluorescence peak. The A288-298 mutation caused a greater overall shift in the population, with a very broad profile indicating a diversity of inherited weaker [RNQ+] or

[rnq-] states.

Because these experiments were conducted in the PSI / RNQ dual prion sensor, we could evaluate the effect of each mutant allele on the other prion protein as well.

57 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Figure 4. Prion-curing mutations identified using a high-throughput genetic screen. (A) Schematic of the yTRAP FACS screen for prion-curing alleles. [PSI] and [RNQ+] strains with prion sensors were mated to cells harboring a library of random mutations in the respective prion gene. Two successive rounds of sorting for high-fluorescent cells were performed to enrich for mutant prion alleles that cure the endogenous prions. Plasmids were isolated from high-fluorescence cells and sequenced. (B) Characterization of selected [PS/+]-curing mutants. Curing efficiencies were assessed by transiently overexpressing the mutants in a [PSI+] strain, followed by >10 generations of recovery in non-inducing media, and then measuring yTRAP outputs by flow cytometry (n = 2, error bars, range between duplicates). (C) Characterization of selected [RNQ+]-curing mutants. Histograms showing the sensor output after mutant proteins were transiently overexpressed in a [RNQ+] strain followed by >10 generations of recovery in non-inducing media. Yellow, Rnql nonsense mutation at residue 313 (stop 313); light green, A288-298 mutation; gray, [RNQ+] control strain; dark green, [mq-] control strain.

Mutant and wild type Rnq1 expression had no measurable effect on the [PSI+] prion in this strain (data not shown). In contrast, the wild type Sup35 and the majority of the

[PS/*]-curing Sup35 mutants also cured [RNQ+] (Figure S4). The two exceptions were the double mutant S17R, Q56R and the double mutant Q1OR, Q24R. These selectively cured [PSI+], but not [RNQ+]. Our results illustrate the potential of yTRAP to enable high-throughput screens for aggregation-modifying mutants and factors without the laborious and limiting process of establishing downstream or indirect cellular phenotypes.

Anti-prion drive systems engineered using prion-curing alleles

Gene drives are genetic systems that bias the standard Mendelian inheritance of a particular gene to increase its prevalence in a population 61 . These can be naturally- occurring or synthetic systems, such as the recent examples of CRISPR/Cas9-based endonuclease gene drives62. Fungal prions can also act as "drives" due to their self- perpetuating nature; but instead of propagating nucleic acid sequence changes in cells, prions propagate changes in phenotype6 3. Specifically, self-templating prion aggregates

58 A B [PSI+]-CURING MUTANTS

GENETIC SCREEN FOR percentage [PSI+] cells cured PRION-CURING ALLELES 0 20 40 60 8C Y46C, Q95L 015R, K1391, 0242R Q15R, P65S [PRION+] :;AMUTANT 01OR, 024R STRAIN X(@LLIBRARY N8D, Q70R, A237T (@ C yTRAP mutant library CI Q15R sensor of yeast gene *3 Q15R, 1152T E 061L Ulg S17R, 056R select for cr) 022R diploid 17 S17R (I) G20D Q10P, N229D Q1OR I Q18R, Y32H I G25D I sort for N81, V246A W. SENSOR ON Wr W0 population

...... , ...... , ...W5%1 C [RNO+]-CURING MUTANTS mNeonGreen (a.u.) isolate plasmids 100- Co, from sorted cells 80- [RNQ+] control Rnql stop313 E 60- Rnql A288-298 [rnq-] control transform enriched 0 plasmids into a1 40- sensor W strain.andprntal sort 20- 0- 10 101 10 101 mNeonGreen (a.u.) mNeonGreen (a.u.) are transferred to mating partners during cytoplasmic fusion. Upon sporulation, the yeast prion is inherited by all meiotic progeny, and all progeny continue to propagate the prion state and associated phenotypes. These properties enable prions to function as protein-based genetic elements with non-Mendelian patterns of inheritance.

Prion-curing mutant alleles offer a unique opportunity to reverse the typical 4:0 phenotypic inheritance of the [PSI+] and [RNQ*] prions. If a strain is engineered to express a prion-curing mutant, it will cure the prion that is present in its mating partner instead of acquiring a prion state, and future progeny will not inherit the prion (Figure

59 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Figure 5. Engineered anti-prion drive systems eliminate prions from mating partners and progeny (A) Diagram of typical, dominant prion inheritance (left) and anti-prion drive inheritance (right) through mating and sporulation. Rather than 4:0 inheritance typical for [PSI+] and [RNQ+], an anti-prion drive will cause reversal of this inheritance pattern, resulting in 0:4 segregation ("full drive") or curing of the prion in spores that do not inherit the drive expression cassette ("semi drive"). (B) Left: Genetic design of the anti-PSI drive system. The drive strain harbored a PSI sensor and a cassette encoding strong, constitutive expression of a [PS/I]-curing allele (Sup35NMAPD = S17R, Q56R). pGPD, yeast GAPDH promoter. Right: The drive eliminates [PSI+] from all progeny spores ("full drive"). Three replicate sets of tetrad dissections resulting from mating the anti-[PS/] drive strain to a [PSI+] strain. Cells spotted onto (-Leu) selective media for the drive cassette showed the expected 2:2 segregation. Fluorescence images of spores on non-selective media showed the prion state via yTRAP. "WT Sup35" is a control in which the wild type Sup35 NM domain was expressed from the drive cassette instead of the [PS/I]-curing allele (Sup35APD). "Cured" are the WT control spores cured of the prion state by guanidine treatment. (C) Left: Genetic design of the anti-RNQ drive system. The drive strain harbored a RNQ sensor and a cassette for the strong, constitutive expression of a [RNQ+]-curing allele (Rnq1APD = stop313). Right: The drive eliminated [RNQ+] from spores that do not inherit the drive cassette ("semi drive"). As above for Sup35NM, wild type Rnql and its guanidine-cured derivatives were used as controls.

5A). We term this mode of inheritance an "anti-prion drive" (APD), as it biases the inheritance of prion phenotypes away from 4:0 and toward 0:4.

We used the Sup35 S17R, Q56R double mutant to construct an "anti-PSI drive" due to its efficient and highly selective [PSil]-curing potential (Figures 4B and S4), and the stop313 Rnq1 mutant to construct an "anti-RNQ drive". We cloned each prion-curing allele or a wild type control into a plasmid with strong constitutive expression, and integrated these into the genome of a strain harboring the respective yTRAP sensor

(Figures 5B and 5C). We then mated these strains to the corresponding [PRION+] strain also harboring the yTRAP sensor. We selected for diploids, sporulated them, and dissected tetrads to determine the segregation frequency of the phenotype (Figures 5B and 5C).

The anti-PSI drive cured every tetrad of the prion state ("full drive"): spores had the same fluorescence as guanidine-cured controls (Figures 5B and S5B) and were not further curable by growing on guanidine hydrochloride-containing media (Figure S5B).

The anti-RNQ drive also caused curing of the [RNQ+] prion, but acted as a "semi drive".

60 A NORMAL PRION ANTI-PRION DRIVE INHERITANCE INHERITANCE

[PRION+] [PRION+] ANTI-PRION STRAIN DRIVE (APD) STRAIN Q ) y gx STRAINT[prioA]

select for select for diploid diploid I I te dissectsprua sdissect

FULL 3 DRIVE ALL PROGENY [PRION'] 3 SEMI S3 3 DRIVE progeny with progeny with APO cassette APO cassette

B ANTI-PSI DRIVE Sup35APD FULL DRIVE WT Sup35 Cured (S17R, Q56R) tetrad 1 PSI sensor z-m tetrad 2 yTRAusIon|. tetrad 3 S;T35 synTA V

Drive overexpression (D 0

pGPD

Rnq1AP ANTI-RNQ DRIVE WT Rnql Cured (stop3l 3) C SEMI DRIVE tetrad 1 RNO sensor tetrad2

T A fsIon tetrad 3 Rnql synTA -a 0 0 Drive overexpression (D C 0 0 pGPD

61 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

The spores that segregated with the APD cassette exhibited intermediate yTRAP fluorescence (Figure 5C), indicative of acquiring a variant of the [RNQ+] prion; these spores could be further cured with guanidine hydrochloride to a bright [mq-] state

(Figure S5D). In contrast, spores that no longer had the anti-RNQ construct were completely cured of the [RNQ+] prion, and were not further cured on guanidine hydrochloride-containing media (Figure S5D). For both the anti-PSI and anti-RNQ drives, all dissected tetrads showed the expected 2:2 segregation of the APD cassette, which allows growth on medium lacking leucine (Figure 5B, -Leu). These results suggest a mechanism of [RNQ+] curing by the stop313 mutant, whereby the mutant alters the prion variant to one that requires the stop313 Rnq1 protein to propagate.

Once that mutant allele is removed and only the wild type Rnq1 protein remains, the new prion variant can no longer be propagated.

One of the RNQ1 mutant alleles identified in our screen was also found in nature

(Figure 4C). To explore the effect of this natural allele when mated with [RNQ+] partners, we tested it in the same context as the drives above. Interestingly, we found that the natural RNQ1 mutant did not cure any prions in progeny cells, but rather functioned as a "prion-remodeling drive" (RD), capable of altering the solubility of Rnq1 in all progeny (Figures S5E and S5F). Specifically, spores that did not inherit the A288-

298 RNQ1 allele were heritably changed to a more soluble state of [RNQ+], indicated by higher fluorescence (Figure S5F). Spores that did inherit the drive cassette showed an even greater degree of solubility. This suggests that variant alleles of prions can unlock new arrays of prion conformational states through mating and undergoing meiosis. Akin to the diversification of a chromosome during genetic recombination, the effect of prion-

62 remodeling alleles on heritable protein conformations during mating may act as a form of protein-based recombination.

In addition to discovering the remodeling effect of a natural prion allele, we are able to engineer anti-prion drives to reverse and finely tune the propagation of yeast prions in a population.

DISCUSSION

We have developed a powerful genetic toolkit that enables quantitative monitoring and control of cellular aggregates. yTRAP offers a number of advantages over existing biochemical or microscopy-based methods of assaying protein aggregates. By coupling protein solubility to synthetic transcriptional activation of fluorescent reporters, the toolkit allows for quantitative and sensitive measurement of aggregation states in living cells.

Importantly, the system is also amenable to high-throughput methods that enable screening by genetic, chemical, or other means. yTRAP features a context- independent, orthogonal, and modular design. As a result, yTRAP sensors for desired proteins can be readily constructed as genetic fusions in which functionality does not rely on specific growth media / conditions or genetic backgrounds. Additionally, more complex sensor systems can be engineered using additional orthogonal synTA- promoter pairs to enable multiplexed observation and control of aggregation-prone proteins.

It is becoming increasingly clear that protein aggregation plays key roles in cellular processes, as wide ranging as disease initiation and progression, signaling, and

63 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives evolution. Prions are model aggregation systems that illustrate this principle. Prions were first discovered based on their ability to transmit lethal neurodegenerative diseases in mammals 64. However, prion-like mechanisms can also serve diverse cellular functions in many organisms 65. Herein, we applied yTRAP to address a number of challenging problems concerning yeast prions, using it to (1) encode sensors that detect prion states in yeast and (2) program new, prion-mediated cellular functions. The sensors can be deployed across different genetic backgrounds (including, in principle, wild isolates of yeast), enabled us to detect a variety of prions and prion combinations using a common platform. This included [RNQ*], which lacks a growth phenotype and has previously been laborious to track. We then applied the sensor systems in a variety of studies, including the measurement of a hyper-inducing prion fusion and high- throughput screens to identify mutant prion alleles that cure prions.

Our studies uncovered a number of noteworthy findings regarding yeast prions.

First is a fusion of two prion genes that allowed for virtually 100% induction of the [PSI+] prion (Figure 3B). Prions generally switch at very low spontaneous rates 54 , but this fusion reveals that prion switching can be regulated and deterministic. Given the simplicity of the design, we expect that similar hyper-inducing genes could be found in nature and designed for other prions. Second, we isolated the first mutations that cure

[PSI+] without the introduction of charged amino acids, as well as the first, to our knowledge, that cure [RNQ+] in trans (which were also uncharged substitutions). This knowledge expands the classes of mutations that can impede prion propagation, previously thought to be limited to substitutions generating charged residues 59. Finally, we discovered that overexpression of Sup35NM can cure the [RNQ*] prion, as well as

64 two mutant alleles of Sup35NM that do not show this effect (Figure S4). Though it was known that the overexpression of Q/N-rich proteins can destabilize prions by interfering with chaperone proteins 66, our demonstration is the first to our knowledge that revealed an effect of Sup35 on the [RNQ*] prion. The two mutant Sup35 alleles we discovered that escape this effect could help to explore the basis of this curing.

In addition to enabling the detection of prions and screens for prion modulators, yTRAP is a versatile platform for encoding new functionalities using yeast prions. yTRAP circuits can be engineered such that a prion state controls expression of genes other than fluorescent reporters, such as genes that encode biosynthetic enzymes or that affect cell viability. Moreover, because prions act as stable, heritable elements, they can encode cellular memories when prion switching is wired to a stimulus 67. We engineered a thermal memory device based on this principle of prion induction, wherein transient temperature upshifts drive the induction of a heritable prion state in a specific, measurable fraction of the population. This could be applied to engineer cellular systems, where new metabolic programs and other processes can be strongly regulated by a small, transient change in temperature. Such a tool may be a more efficient option compared to expensive chemical means for driving expression of desired products in bioreactors. Prion-encoded memory devices will be useful in building more complex functions as synthetic biological systems become more advanced.

A second prion functionality enabled by our yTRAP technology was the creation of anti-prion drives, capable of regulating and curing prion propagation in wild type populations. Just as gene drives bias the inheritance of DNA so that the drive itself is inherited at an increased frequency 68 69, phenotypic drives increase the rate at which a

65 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives phenotype is inherited by progeny. Yeast prions inherently act as phenotypic drives, converting other proteins of the same kind to their own self-templating conformation and being inherited by all meiotic progeny. Prion-associated phenotypes are far reaching and can affect growth rate 8 22 33 , cell adhesion and biofilm formation22 , ethanol production 27, and antibiotic resistance21 . Anti-prion or prion-remodeling drives, such as the ones demonstrated here, could be used to regulate prion phenotypes in the contexts of ecology, industry, and medicine. An important feature of a phenotypic drive is that it can affect wild or unmodified populations in trans through mating with laboratory- designed strains. For example, if a disease-causing fungus resisted treatment due to prion-dependent antibiotic resistance, cells harboring an engineered anti-prion drive could be applied to decrease the number of resistant pathogens. Recent evidence shows that prions or prion-like mechanisms are much more common in yeast than were predicted by computational analysis of protein sequences 33. Furthermore, both human cells and bacteria are capable of propagating prion conformations 5-7, 70. Finding new means to control aggregation in these diverse contexts will be useful for research purposes and could also uncover novel therapeutic strategies. yTRAP represents a powerful tool to screen for these regulators in high throughput.

The molecular components making up the yTRAP toolkit are synthetic, do not appear to interfere with normal biology, and are broadly functional across eukaryotic systems 38. Thus, yTRAP could be moved into human cell lines or even live animal models to study and track aggregation in other relevant contexts. This capability would be useful for discovering aggregation-prone elements in the human proteome and characterizing their biological functions. Indeed, mounting evidence suggests that

66 aggregation mechanisms are found in human cell biology. Several examples exist that are part of normal mammalian physiology and aggregate in prion-like manners: 1)

Pmel17 in melanin synthesis 12; 2) CPEBs involved in synapse regulation8 ; and 3) CARD domains and the RIP1/3 necrosome in innate cellular immunity 5-7,71 . As none of these domains/proteins share common sequence determinants, high-throughput tools like yTRAP will be necessary to uncover more of them and study their functions.

The establishment of yTRAP technology in mammalian systems could also aid the study and treatment of human diseases. Aggregation is a common pathological feature of many incurable neurodegenerative diseases, including Alzheimer's Disease and Parkinson's Disease7 2. The development of tractable cellular models for these aggregation-associated diseases in order to investigate their molecular underpinnings and discover therapeutics remains a significant challenge73. The yTRAP principle could be applied to develop aggregation sensors for disease-associated proteins directly in neuronal cells to establish drug screening capabilities. yTRAP-based high throughput detection of aggregates may be an attractive alternative to the extreme and unnatural overexpression of disease proteins that are commonly used in model systems to generate toxicity from aggregating proteins.

Meanwhile and in lieu of tractable human cell models, yeast has emerged as a useful model system for studying disease-associated aggregation74-77. In this work, we used yTRAP to quantify the sequestration of individual proteins in response to expression of polyglutamine-expanded huntingtin. Moreover, the principles of templated cross-seeding and high throughput screening of random mutants gave us tools to control the [PSI'] prion state. The same principles could also be used to template toxic

67 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives oligomers to non-toxic conformations or screen for drugs or mutant disease alleles that impact the aggregation of disease proteins. Surprisingly, one [RNQ+]-modulating mutant that we identified in a screen is already present in nature. This variant allele can affect the prion state of mating partners in trans, acting as a form of protein-based recombination. It causes an altered prion state, not present in either parent, to be inherited by progeny, thus increasing the overall diversity of the population. Natural, prion-modulating mutants are not limited to yeast - one such natural allele of the human prion disease protein, PRNP, appears to be protective against prions and grants immunity to prion disease when expressed in mice78. Just as we identified alleles that cure yeast prions, we envision that yTRAP can be used to find drugs or mutant alleles that inhibit the aggregation of human disease proteins. yTRAP establishes a technological framework for studying, understanding and controlling the diverse roles of protein aggregation phenomena across cellular regulation and disease.

METHODS

Cloning and vector construction

Plasmids used in this study are listed in Table S3 (appendix D). The basic yTRAP plasmid pGAN147 was constructed by PCR and subsequent Gibson assembly of components into pDML1 1279 (Figure S1A). Gateway@ cloning was used to insert genes-of-interest into the ccdB site in pGAN147 to form yTRAP fusions. Gateway@ cloning was also used to insert genes, such as mutant prion-curing alleles, into pAG- series vectors8 O and their derivatives.

68 Reverse-yTRAP and dual yTRAP plasmids were constructed with standard cloning techniques and Gibson assembly. Reverse-yTRAP required a TetR-repressible promoter (Figure S2A). To obtain a suitable promoter, we generated and screened a small panel of nine variant promoters. Each variant was created by inserting tetO2 sites at the vicinity of promoter control elements such as TATA-box, transcription factor binding sites and the transcriptional start site of the S. cerevisiae ADHI promoter.

[RNQ+] and [rnq-] yeast strains carrying the tetR-mKate2 yTRAP circuit (pGAN230) were used to screen this panel, and the promoter with the greatest dynamic range was selected for use (pSK221).

To develop dual yTRAP, we constructed a second yTRAP plasmid using a second engineered zinc finger and its paired DNA binding site38 (Figure S2C). This second plasmid was generated from pGAN147, exchanging sites of homology from the

HO locus to the LEU2 locus to re-target the integration, and NATr resistance for KanMX.

Prion-inducing fusions were generated by PCR and subsequent Gibson assembly into the estradiol-inducible plasmid pHES83540 . pJMB101 harbors the estradiol-responsive transcription factor composed of the chimeric ZEM regulator

(Zif268 - Estradiol Receptor ligand binding domain - Msn2 activation domain) 40. For heat-inducible expression, the estradiol-responsive promoter was swapped for the

SSA4 promoter by standard cloning techniques.

Yeast strains and growth conditions

Yeast strains used in this study were primarily derived from YJW508 (MA Ta, leu2-3,112; his3-11,-15; trpl-1; ura3-1; adel-14; can1-100; [PSI']; [RNQ']), YJW584

69 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

(MA Ta, leu2-3,112; his3-11,-15; trpl-1; ura3-1; adel-14; can1-100; [psi]; [RNQ*]) 81 , and standard laboratory W303 (MA Ta, leu2-3,112; his3-11,-15; trpl-1; ura3-1; ade2-1; can1-100; [psi]; [RNQ*]). A complete list of strains generated in this study is shown in

Table S2 (appendix C). Growth media was complete standard synthetic media or media lacking certain amino acids with either 2% glucose, 2% galactose, 2% raffinose, or 2% glycerol supplemented as the carbon source (glucose was used unless otherwise specified below). YPD plates used for prion curing contained 5mM guanidine hydrochloride. Media was supplemented with 50 pg/mL adenine hemisulfate (Sigma-

Aldrich, A9126) to eliminate the color of ade- cells, which otherwise interferes with fluorescence readings. Yeast transformations were conducted using a lithium acetate competent cell protocol as previously described8 2.

The construction of variant [PSI'] strains in W303, [PSI+] in 74D-694 (MA Ta ade1-14 leu2-3, 112 ura3-52 trp1-289 his3-200)35, and [PSI'] in 1 OB-H49a (MA Ta ade2-

1 SUQ5 Iysl-1 his3-11,15 leul karl-I ura3::KANMX)83, which were used to characterize the sensor, was done as follows: Strains were cured of all prions by passing cells on rich media containing 5mM guanidine hydrochloride8 4. Prion loss was verified by mating strains to [prion-] tester strains carrying a copper-inducible Sup35NM-

GFP plasmid. Diploids containing diffuse fluorescence after a four-hour incubation with

50 pM CUSO4 verified that the original strains did not contain [PSI+]. Prion variants were introduced into recipient strains through cytoduction. To generate cytoductants, the above recipient strains were made rhoO by streaking cells on rich media containing 10 pg/ml ethidium bromide and verified by no growth on glycerol-containing media.

Recipient strains were mated in excess to the donor strain (C1OB-H49a: Mata SUQ5

70 ade2-1 lys1-1 his3-11,15 leul karl-I cyhR)8 3 that contained the appropriate prion variant and a defective karl allele, which inhibits nuclear fusion8 5. Cytoductants were distinguished from donor or diploids by the presence of auxotrophic markers, ability to mate to MATa strains, and growth on media containing glycerol as the sole carbon source. The presence of [PSl] or [RNQ*] were confirmed by the presence of cytoplasmic aggregates using tester strains. The [PSl] strains were also confirmed to display a change in color on rich media and grow on SD-Ade media.

Flow cytometry

Cells were diluted from overnight cultures in synthetic complete media 100-fold into fresh media, and incubated 6 hours at 300C to reach exponential growth before measurement. 10,000 - 50,000 events were acquired using a MACSQuant VYB cytometer with a 96-well plate platform (Miltenyi Biotech), and data was processed using FlowJo. Events were gated by forward and side scatter, and median fluorescence values were calculated. In order to exclude dead cells from analysis, 1 Oug/mL propidium iodide (Sigma Aldrich Cat. No. P4864) or 1 ug/mL DAPI (Roche Diagnostics

Cat. No. 10236276001) stain was used. The B1 channel (525/50 filter) was used to measure green fluorescence. The Y3 channel (661/20 filter) was used to measure red fluorescence or propidium iodide stain. The V1 channel (450/50 filter) was used to measure DAPI.

Western blotting and densitometry

1 OD unit of log phase cells were collected by centrifugation and protein was extracted with the following TCA extraction method. Cells were resuspended in 1mL

71 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives dH20, followed by the addition of NaOH to a concentration of 0.25M and 2- mercaptoethanol to a concentration of 1 % (v/v), and were incubated on ice 15 minutes.

TCA was added to a concentration of 6.5% (w/v) and samples were incubated a further

10 minutes on ice before centrifugation at 12,000 rcf to collect precipitated protein.

Precipitate was dissolved in HU buffer (200mM Tris HCI pH 6.8, 8M urea, 5% SDS,

1.5% DTT, bromophenol blue) and incubated at 650C for at least one hour.

Samples were separated by SDS-PAGE and transferred to nitrocellulose membranes using the Invitrogen iBlot2 (7-minute transfer). For primary antibodies,

Santa Cruz Mouse monoclonal anti-HA (sc-7392) was used to blot against the 6xHA epitope on the synTA. Rabbit polyclonal anti-PGK1 (ABIN568371; antibodies- online.com) was used as a loading control. For fluorescent secondary antibodies, LI-

COR IRDye 800CW donkey anti-mouse (product 926-32212) and Li-cor IRDye 680RD donkey anti-rabbit (product 926-68073) were used. Blots were imaged with the LI-COR

Odyssey system. Band quantification / densitometry was performed using the Image

Studio Lite software.

SDD-AGE

SDD-AGE was performed as previously described 86. Briefly, saturated cultures of yeast were spheroplasted and lysed, then loaded onto SDD-AGE gels. Gels were run for 4-6 hours at 40 volts, and transferred to nitrocellulose membranes by overnight liquid transfer. Membranes were treated as western blots. Santa Cruz Mouse monoclonal anti-HA (sc-7392) was used as the primary antibody, and Sigma Aldrich HRP- conjugated rabbit anti-mouse IgG (A-9044) was used as the secondary antibody.

72 Agar plate fluorescence photography

Photographs were collected using a Bio-Rad Chemi-Doc MP. Colonies were grown for 2 days at 300C on agar yeast plates and imaged on the second day. For green fluorescence, the blue LED was used for excitation, and the 530nm/28nm filter for emission. For red fluorescence, the green LED was used for excitation, and the

605nm/50nm filter for emission. The Bio-Rad Image Lab software was used for false- coloring. Overlays were assembled using the transparency parameter in Adobe

Photoshop.

Epifluorescence microscopy

Epifluorescence images were taken at 100x magnification (Plan Apo 100x oil objective, NA 1.4) using an Eclipse Ti-E inverted microscope (Nikon Instruments, Inc.).

Images were acquired in DIC and in fluorescent (GFP) channels. Filters and light sources were automatically controlled by the supplier's software (NIS-Elements

Advanced Research).

Fluorescence plate reader measurements

Measurements were performed using an Infinite M1000 PRO microplate reader

(Tecan Group Ltd.). Cells were grown to saturation in synthetic complete media. 200uL of triplicate saturated cultures were added to clear-bottom, flat, black microtiter plates

(Corning Product #3631). Absorbance at 600nm was collected to measure cell density, raw fluorescence (ex. 488nm/5nm, em. 520nm/5nm) was collected to measure yTRAP signal, and fluorescence was calculated by normalizing raw fluorescence by absorbance.

73 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Isolation of Swil, Mot3-PrD, and New1-PrD prion states

Galactose-inducible plasmids for Swil, Mot3-PrD, and Newl-PrD were introduced into sensor strains for the same protein. Cells were grown overnight in ura drop-out media supplemented with galactose to induce expression. Subsequently, cultures were plated on glucose agar plates. Colonies on agar plates were photographed in the green fluorescent channel to assess prion state. Low-fluorescence colonies were picked and streaked for further testing. Colonies were confirmed to contain a prion by streaking on guanidine hydrochloride plates - those that were cured and returned to a bright fluorescent state were [PRION+] switched cells.

Relative aggregation measurement

To measure relative aggregation of Htt-Q25 and Htt-Q103 (Figure S1G), triplicate sensor strains for each protein (yGAN01 6-017) were grown for 6 hours at 300C and yTRAP signal was measured by flow cytometry. Relative aggregation was calculated as the average of the least-aggregated samples divided by each sample: Relative aggregation = (average Htt-Q25 yTRAP signal) / (yTRAP signal).

To measure relative aggregation of Q-/N-rich proteins in response to Htt overexpression (Figure 1 E), yTRAP sensor plasmids for these protein domains were transformed into cells containing integrated galactose-inducible Huntingtin exon 1 harboring 103 glutamine residues (Htt-Q1 03) or into an identical strain lacking the Htt-

Q103 construct (producing yGAN018-029). Triplicate cultures were grown overnight to saturation in synthetic complete media containing raffinose as a carbon source. The following day, each strain was diluted 50-fold into raffinose media and, separately,

74 media containing galactose to induce Htt. Cells were grown for 6.5 hours at 300C and then fluorescence was measured by flow cytometry. Aggregation in Htt-induced samples was calculated by dividing median fluorescence of each strain in raffinose by median fluorescence in galactose. To normalize for the effect of carbon source, the raf/gal ratio of Htt-expressing cells was divided by the raf/gal ratio of control cells, yielding the final relative aggregation value: Relative aggregation = (yTRAP signal of

Htt-expressing cells in raffinose / yTRAP signal of Htt-expressing cells in galactose) /

(yTRAP signal of non-expressing cells in raffinose / yTRAP signal of non-expressing cells in galactose).

Guanidine-curing time course

The dual PSI / RNQ sensor strain (ySK293) was diluted to OD 0.001 into triplicate 50mL cultures of synthetic complete media with 5mM guanidine hydrochloride.

The zero time point negative control was diluted to OD 0.001 without guanidine hydrochloride. After 12 hours of shaking at 30 oC in 500mL baffled flasks, 50uL samples were collected from each of the curing flasks and diluted 100-fold into 5mL of fresh medium lacking guanidine hydrochloride to halt prion curing. Additional samples were collected every two hours until 22 hours after guanidine treatment started. In addition to collecting final, 22hr samples by dilution into medium lacking guanidine hydrochloride, samples were also diluted 100-fold into fresh media containing 5mM guanidine hydrochloride to continue curing. The following day, after a total of 36 hours of growth, all samples had grown to stationary phase and were diluted another 100-fold in fresh medium lacking guanidine hydrochloride. After 6 hours of growth, the prion status was assessed by flow cytometry as described above, gating to distinguish populations of

75 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives prions. One recovered culture that had been treated with guanidine hydrochloride for 16 hours was spread on an agar plate for photography.

Inducing [PSI+] with transient expression of prion fusions

Duplicate colonies harboring estradiol-inducible prion domains and fusions were picked into 500pL of non-inducing synthetic complete media and grown overnight at 30

0C with shaking in a deep-well 96 well plate. Saturated cultures were diluted 100x into

500pL synthetic complete media with or without 100nM estradiol. After 24 hours of growth, cultures were centrifuged 5 min at 2000rcf, and media was beat off. Cultures were resuspended in 500pL of fresh media, then diluted 100x into non-inducing media and grown 24 hours further. Cultures were then diluted 100-fold and grown 6 hours to reach log phase before measurement by flow cytometry to assess prion state. This allows for 8 to 12 generations of growth in non-inducing conditions, so that only heritable changes in prion state were assessed.

Heat induction and memory experiments

Cultures were grown at 250C in complete synthetic media. Cultures were pre- screened with flow cytometry for background switching because of leakiness of the

SSA4 promoter. Only [psi-] cultures were used in the experiment. Saturated cultures were diluted to OD600 of 0.02 in 96-well PCR plates. Cultures were grown for 4 hours at

250C, then transferred to an Eppendorf PCR instrument (Mastercycler Pro) incubating a gradient of temperatures for 2 hours. After heat treatment, samples were returned to

250C for overnight growth. The next day, cultures were diluted 1000x into fresh media.

Selection of prion-curing alleles

76 We used a previously described variomics library 58. The mutant libraries for

Sup35 and Rnq1 were grown in liquid media and sporulated as described 87. MATa spores containing the mutant plasmid and chromosomal deletion were selected by growing in liquid media (SD-arg-his-leu-ura + 60ug/mL canavanine + 200ug/mL G418) for 5 generations. For each step, at least 2 million viable cells were used to maintain 10- fold coverage of the theoretical diversity of 200,000 mutants. The appropriate library of spores was mated with either PSI or RNQ sensor strains harboring the [PSI+] and

[RNQ+] prions. 6 million mutant spores were mixed with 2-fold excess of yTRAP sensor cells. These were plated on YPD agar plates and incubated at 300C for 16 hours to mate. Cells were then scraped from the plate and collected in dH20. Aliquots were plated on diploid- and haploid-selective media to determine the mating efficiency.

Diploid selective media was SD-ura+100ug/mL nourseothricin (NAT) + 200ug/mL G418

(NAT selects for the yTRAP sensor, G418 selects for the chromosomal deletion, media lacking uracil selects for the mutant plasmid covering the chromosomal deletion).

Diploid cells containing both the mutant plasmid and the yTRAP sensor were present in larger numbers than haploids containing only the mutant plasmid, so we estimated that mating was greater than 50% efficient. The pool of scraped cells was diluted in diploid- selective media and grown 6 hours to OD 1.0. At this point, cells were diluted 100-fold into diploid-selective media and grown a further 13 hours until FACS sorting (OD - 1.0).

FACS sorting was performed using a FACS Aria II (BD Biosciences). The top

0.005% of most fluorescent cells (indicating a non-prion state) were collected (300-400 cells). Over 5 million cells were examined in each pool. After sorting, cells were recovered for 2.5 hours in SD-ura media. NAT and G418 were then added to maintain

77 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives selection for diploid cells. Cultures were shaken 2 days at 301C until reaching saturation. Plasmids were purified from each sample as previously described 8

Plasmids were amplified by electroporation into bacteria followed by subsequent miniprep.

Mutant plasmids were retransformed into haploid [PSI'] or [RNQ+] yTRAP sensor strains for a second round of selection. Empty plasmids were also transformed separately to serve as negative controls. Transformants were grown in selective liquid media for 40 hours with intermittent dilution to keep cells from reaching stationary phase, and then subjected to a second FACS sort. Collection gates were set so that all cells from control strains containing empty plasmids would be excluded, and only cells with greater fluorescence were collected. 20,000 cells were collected for both libraries and recovered as before. Plasmids were purified from yeast and transformed into bacteria. 24 clones of each library were prepped and sequenced (Table Si).

Prion-curing with transient expression of mutant alleles

Estradiol-inducible mutant plasmids were transformed into the PSI / RNQ dual- sensor strain yGANO36. The strain also contains a pAG303GPD-Sup35C plasmid, which eliminates toxicity due to Sup35NM expression. Estradiol induction was performed as follows: overnight cultures in glucose, ura drop-out media were diluted

1000x into ura drop-out media supplemented with 5nM estradiol. After 24 hours of induction, cultures were measured by flow cytometry to determine effects on solubility during overexpression (Table SI). Cultures were also diluted 1000x into fresh media lacking estradiol to recover and examine heritable effects. After 24 hours of recovery,

78 cultures were diluted 100x into fresh media and grown 6 hours before measurement by flow cytometry (Figure 4B and 4C, Table Si).

The Saccharomyces Genome Database was used to search for known strains that harbor our selected mutant prion-curing alleles in SUP35 and RNQ1. The Rnq1

A288-298 mutation was harbored in 4 of the 45 available yeast strains: 217_3, PW5,

DBVPG6044, and YJM789.

Anti-prion drive mating, sporulation, and tetrad dissection

Plasmids harboring the 'drive cassettes' (pGAN257-261) or wild type controls were transformed into MATa sensor strains for the respective prion (forming yGAN060-

064). The anti-PSI drive was derived from a clone with high expression of the drive cassette. Strains used to test the anti-PSI drive strain harbored a constitutive Sup35C construct to prevent toxicity from overexpression of Sup35NM. To test drives, anti-prion drive strains were mated to 'tester strains' (yGAN065 for PSI or yGAN066 for RNQ).

Mating was conducted by spreading a water-suspended colony of the tester strain onto agar YPD plates and letting it dry. A separate plate with large, spotted colonies of the drive-containing strains was replica plated onto the same YPD plate, and incubated for

6 hours at 30 oC to allow for mating. The YPD plate was then replica plated onto his and leu drop-out agar plates, allowing only diploids to grow. Diploid colonies were streaked once onto the same media, then picked into 5mL pre-sporulation media (20g/L bacto- peptone, 1Og/L yeast extract, 40g/L glucose, 100mg/L adenine hemisulfate) and grown at 300C overnight.

79 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Overnight cultures in pre-sporulation media were centrifuged and resuspended in

5mL dH20 to wash. This process was repeated for a total of 3 washes before cells were resuspended in 200pL of dH20. 50pL was spotted onto agar SPO plates (1g/L potassium acetate, 50pM uracil, 100pM tryptophan, 50mg/L leucine, 25mg/L of every other natural amino acid, 25mg/L adenine, 2.5mg/L p-aminobenzoic acid, and 20g/L agar). Plates were incubated 2 nights at 300C to sporulate.

Following this incubation, a few mg of cells were picked with a pipet tip and resuspended into enzyme mix (1M sorbitol, 0.1M EDTA, 1Omg/mL zymolyase). This was quickly vortexed to mix and left at room temperature 5-10 minutes. A sterile loop was used to streak a line of cells onto an agar YPD plate. Tetrads were dissected using a Singer MSM dissection microscope. Dissected spores were grown 2 days at 301C and then arrayed into cultures in 96 well plates. Spores were spotted onto synthetic complete media and leu dropout media to image fluorescence (prion state) and plasmid content, respectively (Figure 5B and 5C). Spores were also streaked onto guanidine hydrochloride plates to cure their prions and compared by flow cytometry to uncured cultures.

80 REFERENCES

1. Aguzzi, A. & Lakkaraju, A. K. K. Cell Biology of Prions and Prionoids: A Status Report. Trends Cell Biol. 26, 40-51 (2016). 2. Mukherjee, A., Morales-Scheihing, D., Butler, P. C. & Soto, C. Type 2 diabetes as a protein misfolding disease. Trends Mol. Med. 21, 439-449 (2015). 3. Blancas-Mejia, L. M. & Ramirez-Alvarado, M. Systemic Amyloidoses. Annu Rev Biochem. 745-774 (2013). doi:10.1146/annurev-biochem-07261 1- 130030.Systemic 4. L6pez-Otin, C., Blasco, M. A., Partridge, L., Serrano, M. & Kroemer, G. The hallmarks of aging. Cell 1194-217 (2013). doi:10.1016/j.cell.2013.05.039 5. Cai, X. et al. Prion-like Polymerization Underlies Signal Transduction in Antiviral Immune Defense and Inflammasome Activtion. Cell 156, 1207-1222 (2014). 6. Franklin, B. S. et al. The adaptor ASC has extracellular and 'prionoid' activities that propagate inflammation. Nat. Immunol. 15, 727-737 (2014). 7. Li, J. et al. The RIP1/RIP3 Necrosome Forms a Functional Amyloid Signaling Complex Required for Programmed Necrosis. Cell 150, 339-350 (2012). 8. Majumdar, A. et al. Critical role of amyloid-like oligomers of Drosophila Orb2 in the persistence of memory. Cell 148, 515-529 (2012). 9. Han, T. W. et al. Cell-free formation of RNA granules: bound RNAs identify features and components of cellular assemblies. Cell 149, 768-79 (2012). 10. Kato, M. et al. Cell-free formation of RNA granules: low complexity sequence domains form dynamic fibers within hydrogels. Cell 149, 753-67 (2012). 11. Protter, D. S. W. & Parker, R. Principles and Properties of Stress Granules. Trends Cell Biol. 26, 668-679 (2016). 12. Fowler, D. M. et al. Functional amyloid formation within mammalian tissue. PLoS Biol. 4, e6 (2006). 13. Austin, J. W., Sanders, G., Y, W. W. K. & Collinson, S. K. Thin aggregative fimbriae enhance Salmonella enteritidis biofilm formation. 162, 295-301 (1998). 14. Taglialegna, A., Lasa, 1. & Valle, J. Amyloid structures as biofilm matrix scaffolds. J. Bacteriol. 198, JB.00122-16 (2016). 15. Aigle, M. & Lacroute, F. Genetical aspects of [URE3], a non-mitochondrial, cytoplasmically inherited mutation in yeast. MGG Mol. Gen. Genet. 136, 327-335 (1975). 16. Lancaster, A. K., Bardill, J. P., True, H. L. & Masel, J. The Spontaneous Appearance Rate of the Yeast Prion [PSI+] and Its Implications for the Evolution of the Evolvability Properties of the [PSI+] System. 184, 393-400 (2010).

81 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

17. Allen, K. D., Chernova, T. a, Tennant, E. P., Wilkinson, K. D. & Chernoff, Y. 0. Effects of ubiquitin system alterations on the formation and loss of a yeast prion. J. Biol. Chem. 282, 3004-13 (2007). 18. True, H. & Lindquist, S. A yeast prion provides a mechanism for genetic variation and phenotypic diversity. Nature 407, 477-483 (2000). 19. True, H., Berlin, I. & Lindquist, S. Epigenetic regulation of translation reveals hidden genetic variation to produce complex traits. Nature 431, 184-187 (2004). 20. Halfmann, R. et al. Prions are a common mechanism for phenotypic inheritance in wild yeasts. Nature 482, 363-368 (2012). 21. Suzuki, G., Shimazu, N. & Tanaka, M. A yeast prion, Mod5, promotes acquired drug resistance and cell survival under environmental stress. Science 336, 355-9 (2012). 22. Holmes, D. L., Lancaster, A. K., Lindquist, S. & Halfmann, R. Heritable remodeling of yeast multicellularity by an environmentally responsive prion. Cell 153, 153-165 (2013). 23. Du, Z., Zhang, Y. & Li, L. The Yeast Prion [SWI+] Abolishes Multicellular Growth by Triggering Conformational Changes of Multiple Regulators Required for Flocculin Gene Expression. Cell Rep. 13, 2865-2878 (2015). 24. Wickner, R. B. [URE3] as an altered URE2 protein: evidence for a prion analog in Saccharomyces cerevisiae. Science (80-. ). 264, 566-9 (1994). 25. Osherovich, L. Z., Cox, B. S., Tuite, M. F. & Weissman, J. S. Dissection and design of yeast prions. PLoS Biol. 2, (2004). 26. Wickner, R. B., Masison, D. C. & Edskes, H. K. [PSI] and [URE3] as yeast prions. Yeast 11, 1671-1685 (1995). 27. Jarosz, D. et al. Cross-kingdom chemical communication drives a heritable, mutually beneficial prion-based transformation of metabolism. Cell 158, 1083- 1093 (2014). 28. Wickner, R. B., Edskes, H. K., Bateman, D., Kelly, A. C. & Gorkovskiy, A. The yeast prions [PSI+] and [URE3] are molecular degenerative diseases. Prion 5, 258-262 (2011). 29. Saupe, S. J. & Daskalov, A. The [Het-s] prion, an amyloid fold as a cell death activation trigger. PLoS Pathog. 8, 8-10 (2012). 30. Maddelein, M.-L., Dos Reis, S., Duvezin-Caubet, S., Coulary-Salin, B. & Saupe, S. J. Amyloid aggregates of the HET-s prion protein are infectious. Proc. Nat/. Acad. Sci. U. S. A. 99, 7402-7407 (2002). 31. Alberti, S., Halfmann, R., King, 0., Kapila, A. & Lindquist, S. A systematic survey identifies prions and illuminates sequence features of prionogenic proteins. Cell 137, 146-58 (2009).

82 32. March, Z. M., King, 0. D. & Shorter, J. Prion-like domains as epigenetic regulators, scaffolds for subcellular organization, and drivers of neurodegenerative disease. Brain Res. 1647, 9-18 (2016). 33. Chakrabortee, S. et al. Intrinsically Disordered Proteins Drive Emergence and Inheritance of Biological Traits. Cell 167, 1-13 (2016). 34. Byrne, L. J. et al. The number and transmission of [PSI] prion seeds (Propagons) in the yeast Saccharomyces cerevisiae. PLoS One 4, e4670 (2009). 35. Chernoff, Y. 0., Lindquist, S. L., Ono, B., Inge-Vechtomov, S. G. & Liebman, S. W. Role of the chaperone protein Hsp1 04 in propagation of the yeast prion-like factor [psi+]. Science 268, 880-884 (1995). 36. Jung, G. & Masison, D. C. Guanidine hydrochloride inhibits Hsp104 activity in vivo: A possible explanation for its effect in curing yeast prions. Curr. Microbiol. 43, 7-10 (2001). 37. Crow, E. T. & Li, L. Newly identified prions in budding yeast, and their possible functions. Semin Cell Dev Biol 22, 452-459 (2011). 38. Khalil, A. S. et al. A synthetic biology framework for programming eukaryotic transcription functions. Cell 150, 647-58 (2012). 39. Keung, A. J., Bashor, C. J., Kiriakov, S., Collins, J. J. & Khalil, A. S. Using targeted chromatin regulators to engineer combinatorial and spatial transcriptional regulation. Cell 158, 110-120 (2014). 40. Aranda-Diaz, A., Mace, K., Zuleta, I., Harrigan, P. & EI-Samad, H. Robust Synthetic Circuits for Two-Dimensional Control of Gene Expression in Yeast. ACS Synth. Biol. acssynbio.6b00251 (2016). doi:10.1021/acssynbio.6b00251 41. Liu, J. J., Sondheimer, N. & Lindquist, S. L. Changes in the middle region of Sup35 profoundly alter the nature of epigenetic inheritance for the yeast prion [PSI+]. Proc. Natl. Acad. Sci. U. S. A. 99 Suppl 4,16446-53 (2002). 42. Cox, B. S. PSI, a cytoplasmic suppressor of super-suppressor in yeast. Heredity (Edinb). 20, 505-21 (1965). 43. Ferreira, P. C., Ness, F., Edwards, S. R., Cox, B. S. & Tuite, M. F. The elimination of the yeast [PSI+] prion by guanidine hydrochloride is the result of Hsp104 inactivation. Mol. Microbiol. 40, 1357-1369 (2001). 44. Lee, J. M. et al. CAG repeat expansion in Huntington disease determines age at onset in a fully dominant fashion. Neurology 78, 690-695 (2012). 45. Wear, M. P. et al. Proteins with intrinsically disordered domains are preferentially recruited to polyglutamine aggregates. PLoS One 10, 1-27 (2015). 46. Preisinger, E., Jordan, B. M., Kazantsev, a & Housman, D. Evidence for a recruitment and sequestration mechanism in Huntington's disease. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 354, 1029-1034 (1999). 47. Schaffar, G. et al. Cellular toxicity of polyglutamine expansion proteins: Mechanism of transcription factor deactivation. Mol. Cell 15, 95-105 (2004).

83 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

48. Park, S. H. et al. PolyQ proteins interfere with nuclear degradation of cytosolic proteins by sequestering the Sisip chaperone. Cell 154, 134-145 (2013). 49. Halfmann, R. et al. Opposing Effects of Glutamine and Asparagine Govern Prion Formation by Intrinsically Disordered Proteins. Mol. Cell 43, 72-84 (2011). 50. Kayatekin, C. et al. Prion-like proteins sequester and suppress the toxicity of huntingtin exon 1 TL - 111. Proc. Natl. Acad. Sci. 111 VN-, 12085-12090 (2014). 51. Derkatch, I. L. et al. Dependence and independence of [PSI(+)] and [PIN(+)]: a two-prion system in yeast? EMBO J. 19, 1942-52 (2000). 52. Halfmann, R. & Lindquist, S. Epigenetics in the extreme: prions and the inheritance of environmentally acquired traits. Science 330, 629-632 (2010). 53. Chernoff, Y. 0., Derkach, I. L. & Inge-Vechtomov, S. G. Multicopy SUP35 gene induces de-novo appearance of psi-like factors in the yeast Saccharomyces cerevisiae. Curr. Genet. 24, 268-270 (1993). 54. Liebman, S. W. & Chernoff, Y. 0. Prions in yeast. Genetics 191, 1041-1072 (2012). 55. Derkatch, I. L., Bradley, M. E., Zhou, P., Chernoff, Y. 0. & Liebman, S. W. Genetic and Environmental Factors Affecting the de novo Appearance of the [PSI+] Prion in Saccharomyces cerevisiae. Genetics 147, 507-19 (1997). 56. Derkatch, I. L. et al. Effects of Q/N-rich, polyQ, and non-polyQ amyloids on the de novo formation of the [PSI+] prion in yeast and aggregation of Sup35 in vitro. Proc. Natl. Acad. Sci. U. S. A. 101, 12934-9 (2004). 57. Choe, Y. J., Ryu, Y., Kim, H. J. & Seok, Y. J. Increased [PSI+] appearance by fusion of rnq1 with the prion domain of sup35 in saccharomyces cerevisiae. Eukaryot. Cell 8, 968-976 (2009). 58. Huang, Z. et al. A functional variomics tool for discovering drug resistance genes and drug targets. 2Cel Reports 3, 577-585 (2013). 59. DePace, A. H., Santoso, A., Hillner, P. & Weissman, J. S. A critical role for amino- terminal glutamine/asparagine repeats in the formation and propagation of a yeast prion. Cell 93, 1241-1252 (1998). 60. Kelly, A. C., Shewmaker, F. P., Kryndushkin, D. & Wickner, R. B. Sex, prions, and plasmids in yeast. PNAS (2012). doi:10.1073/pnas.1213449109 61. Burt, A. Site-specific selfish genes as tools for the control and genetic engineering of natural populations. Proc. Biol. Sci. 270, 921-8 (2003). 62. Esvelt, K. M., Smidler, A. L., Catteruccia, F. & Church, G. M. Concerning RNA- guided gene drives for the alteration of wild populations. Elife 3, e03401 (2014). 63. Dalstra, H. J. P., Swart, K., Debets, A. J. M., Saupe, S. J. & Hoekstra, R. F. Sexual transmission of the [Het-S] prion leads to meiotic drive in Podospora anserina. PNAS 100, 6616-21 (2003).

84 64. Prusiner, S. B. Novel proteinaceous infectious particles cause scrapie. Science 216, 136-144 (1982). 65. Newby, G. & Lindquist, S. Blessings in disguise: biological benefits of prion-like mechanisms. Trends Cell Biol 23, 251-259 (2013). 66. Yang, Z., Hong, J. Y., Derkatch, I. L. & Liebman, S. W. Heterologous gln/asn-rich proteins impede the propagation of yeast prions by altering chaperone availability. PLoS Genet. 9, e1003236 (2013). 67. Lu, T. K., Khalil, A. S. & Collins, J. J. Next-generation synthetic gene networks. Nat. Biotechnol. 27, 1139-1150 (2009). 68. DiCarlo, J. E., Chavez, A., Dietz, S. L., Esvelt, K. M. & Church, G. M. Safeguarding CRISPR-Cas9 gene drives in yeast. Nat Biotechnol 33, 1250-1255 (2015). 69. Sinkins, S. P. & Gould, F. Gene drive systems for insect disease vectors. Nat. Rev. Genet. 7, 427-435 (2006). 70. Yuan, A. H. & Hochschild, A. A bacterial global regulator forms a prion. 201, 198- 201 (2017). 71. Hou, F. et al. MAVS forms functional prion-like aggregates to activate and propagate antiviral innate immune response. Cell 146, 448-61 (2011). 72. Knowles, T. P. J., Vendruscolo, M. & Dobson, C. M. The amyloid state and its association with protein misfolding diseases. Nat. Rev. Mol. Cell Biol. 15, 384-96 (2014). 73. Khurana, V., Tardiff, D. F., Chung, C. Y. & Lindquist, S. Toward stem cell-based phenotypic screens for neurodegenerative diseases. Nat. Rev. Neurol. 11, 339- 50 (2015). 74. Khurana, V. & Lindquist, S. Modelling neurodegeneration in Saccharomyces cerevisiae: why cook with baker's yeast? Nat. Rev. Neurosci. 11, 436-449 (2010). 75. Tardiff, D. F., Khurana, V., Chung, C. Y. & Lindquist, S. From yeast to patient neurons and back again: A powerful new discovery platform. Mov. Disord. 29, 1231-1240 (2014). 76. Kaganovich, D., Kopito, R. & Frydman, J. Misfolded proteins partition between two distinct quality control compartments. Nature 454, 1088-1095 (2008). 77. Choe, Y.-J. et al. Failure of RQC machinery causes protein aggregation and proteotoxic stress. Nature 531, 191-5 (2016). 78. Asante, E. A. et al. A naturally occurring variant of the human prion protein completely prevents prion disease. Nature 522, 478-81 (2015). 79. Landgraf, D., Huh, D., Hallacli, E. & Lindquist, S. Scarless gene tagging with one- step transformation and two-step selection in Saccharomyces cerevisiae and Schizosaccharomyces pombe. PLoS One 11, 1-24 (2016).

85 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

80. Alberti, S., Gitler, A. & Lindquist, S. A suite of Gateway@ cloning vectors for high- throughput genetic analysis in Saccharomyces cerevisiae Simon. Yeast2 24, 913-919 (2007). 81. Osherovich, L. Z. & Weissman, J. S. Multiple Gin/Asn-rich prion domains confer susceptibility to induction of the yeast [PSI+] prion. Cell 106, 183-194 (2001). 82. Knop, M. et al. Epitope tagging of yeast genes using a PCR-based strategy: more tags and improved practical routines. Yeast 15, 963-972 (1999). 83. Kochneva-Pervukhova, N. V., Poznyakovski, A. I., Smirnov, V. N. & Ter- Avanesyan, M. D. C-terminal truncation of the Sup35 protein increases the frequency of de novo generation of a prion-based [PSI+] determinant in Saccharomyces cerevisiae. Curr. Genet. 34, 146-151 (1998). 84. Tuite, M. F., Mundy, C. R. & Cox, B. S. Agents that cause a high frequency of genetic change from [psi+] to [psi-] in Saccharomyces cerevisiae. Genetics 98, 691-711 (1981). 85. Conde, J. & Fink, G. R. A mutant of Saccharomyces cerevisiae defective for nuclear fusion. Genetics 73, 3651-3655 (1976). 86. Halfmann, R. & Lindquist, S. Screening for amyloid aggregation by Semi- Denaturing Detergent-Agarose Gel Electrophoresis. J. Vis. Exp. 20-22 (2008). doi:10.3791/838 87. Pan, X. et al. A robust toolkit for functional profiling of the yeast genome. Mol. Cell 16, 487-496 (2004). 88. Chen, T. F., De Picciotto, S., Hackel, B. J. & Wittrup, K. D. Engineering fibronectin-based binding proteins by yeast surface display. Methods in Enzymology 523, (Elsevier Inc., 2013).

86 87 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

SUPPLEMENTAL FIGURES

Figure S1. Related to Figure 1. yTRAP design, characterization, and sensors for detecting yeast prions and disease aggregates. (A) Schematic of the yTRAP sensor genetic construct. This consists of (1) a yTRAP fusion between a gene-of-interest (GOI) and a synthetic zinc finger (ZF)-based transcriptional activator (ZF-43-8) (Khalil et al., 2012), and (2) a reporter harboring 8x ZF binding sites upstream of a minimal promoter driving the expression of a mNeonGreen reporter gene. This plasmid is integrated as a single copy into the HO locus of the yeast genome. SV40 NLS, SV40 nuclear localization sequence; VP16, Herpes simplex virus VP16 transcriptional activation domain; PKI NES, protein kinase inhibitor nuclear export sequence; SUP35 term, SUP35 terminator; minCYC1, minimal CYCI promoter; ADH1 term, ADHI terminator. (B) yTRAP reporter output as a function of soluble synTA (no fusion), titrated using an estradiol induction system (Aranda-Diaz et al., 2016). Top: Western blot against synTA (green) and a loading control (red). RNQ sensor strains (synTA with fusion to Rnql) in the [RNQ+] and [mq-] states were also included for comparison of synTA levels (right lanes). Bottom: The relationship between soluble synTA (no fusion) protein level and yTRAP output, fit by a linear regression (R 2 >0.95). (C) Diverse prions measured by fluorescence plate reader. Top: yTRAP output for prion and non- prion strains containing the Rnql, Swil, Mot3-PrD, and Newl-PrD yTRAP sensors. Green: Initial non- prion state of the strain after transformation with the yTRAP sensor (W303 is already [RNQ+], no green bar). Gray: [PRION+] strains isolated after overexpression of the prion. Orange: [PRION+] strains cured of the prion by transient guanidine treatment. Bottom: yTRAP outputs of the variant PSI strains (from Figure 1 B), which were also cured with guanidine treatment ('cured'). (n=3, error bars, SD). (D) Fluorescence micrographs of PSI sensor cells. (E) Flow cytometry histograms of the PSI sensor in [PSI'] and [psi] strains of yeast with diverse genetic backgrounds as indicated. (F) Top: Flow cytometry of the New1 -PrD, Swi1, and Mot3-PrD sensors in [PRION+] and [prion-] states. Right: SDD-AGE analysis confirming that the Mot3-PrD yTRAP fusion was converted to an amyloid state in the low-fluorescence (yTRAP OFF) population. (G) Aggregation of polyQ-expanded Htt quantified by yTRAP. Left: Schematic of the Htt-PolyQ yTRAP sensors. Middle: Relative aggregation of Htt-Q25 vs. Htt-Q103 quantified by flow cytometry. Relative aggregation was calculated as the inverse of median fluorescence, normalized to the Htt- Q25 sample (see Materials and Methods). Right: Western blot against the yTRAP fusions (synTA, green; loading control, red).

88 A yTRAP fusion yTRAP reporter

pS 5stUR;5 HO nCIi ADHI] Gateway Synthetic transcriptional cassette activator (synTA) (MCS) B D weak [P1+1 strona P1 1 estradiol [nM]

anti-PGK1 E DIFFERENT STRAINS control anti-HA W303 74-D694 1OB-H49A (synTA) 100 f 7 1-- strong so- [PSI+] E weak ai 00 0R 40 1 . I [PSI+] Ii 20 [psi] 40000 010 1 10 lo10102 10 1 0 10 10ic 10, 10 Ill mNeonGreen fluorescence (a.u.) i DIFFERENT PRIONS 0 2D 40 8 80 10 12D 14D F synTA protein band intensity (a.u.) ,qN , '*N Newl-PrD SENSOR Swil-PrD SENSOR Mot3-PrD SENSOR Rio 100- 8 E 6 Icured U C FLUOR. PLATE READER [PRION*I 40

-..50000- 20 (prion-] 4O.40000 -PRION-1 w1 30000 -cured [PRIONi] 6 50K Co 2000 Cn 0LLDOi 10 ie 10' 10' 1... lit 10 10 10 10000 0A G mNeonGreen fluorescence (a.u.) [(W .l (SWI-I [MOT3- [NEW1- PRD] PD]1 ZEE 0n G C0) TRAP fusIon| 12 2 40000 synTA i.E .2Cu ,20000 (D 1 - I Sensors of -, Htt-Q25 polyQ proteins e & wsith different Htt-Q103 tract lengths 0 Pep 49

89 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Figure S2. Related to Figure 2. Reverse-yTRAP and dual yTRAP sensor systems. (A) The reverse-yTRAP sensor turns ON in the aggregated state. Top: Schematic of reverse-yTRAP. In the soluble state, the yTRAP fusion drives expression of TetR (fused to mKate2), which represses expression of the mNeonGreen reporter from an engineered, Tet-controlled ADHI promoter (pADH1). In the aggregated state, TetR is not produced and repression of mNeonGreen is relieved. Bottom: Histograms of red (left) and green (right) fluorescence of [RNQ+] and [mq-] cells harboring a reverse- yTRAP RNQ sensor. (B) Schematic of the reverse-yTRAP sensor genetic constructs. (C) Schematic of the dual sensor genetic constructs. This system has two yTRAP cassettes, each using a distinct, orthogonal synTA / promoter pair (ZF-43-8 and ZF-37-12) (Khalil et al., 2012).

90 A REVERSE-yTRAP

[prion-J STATE [PRION+] STATE

RED RED ~ON C=&MEOFF

GREEN GREEN t H OFF ON toto pDHI tetO 'pADH1

100 [mq] 100- [rnq-] 80 [RNQO+ 80- [RNQ+] E 60- E 60- 0 0 40- 40- 20- 20-

0 0- 02 - 10 102 103 104 102 10 10 mKate2 (a.u.) mNeonGreen (a.u.)

B Reverse-yTRAP fusion TetR cassette

HO p NLSIVP16ZF-PVJNESJMA[N 8ZFBmInCYCi TeR Gateway Synthetic transcriptional cassette activator (synTA) (MCS)

yTRAP reporter

LEU2 EpADH 2x tetO sites

C yTRAP fusion 1 yTRAP reporter 1

- mEnI~W~1 - ~D!~ L HO m ~IIJ.trm.~ui Gateway Synthetic transcriptional cassette activator (synTA 1) (MCS) yTRAP fusion 2 yTRAP reporter 2

LEU2 pSUP35 SV40 NIS VPI6 PKI NES OxHA t ' Gateway Synthetic transcriptional cassette activator (synTA 2) (MCS)

91 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Figure S3. Related to Figure 3. A hyper-inducing prion fusion induces [PSI'] specifically in [RNQ*] cells but does not affect the [RNQ*] prion. (A) Induction of [PSI+] by the hyper-inducing prion fusion, N-Rnq1-M, confirmed by the red/white assay, in which [psi] colonies are red and [PSI+] are pink or white (Liebman and Chernoff, 2012). Agar plate of yeast cells harboring estradiol-inducible N-Rnql-M grown without estradiol (uninduced, left) and with estradiol (induced, right). (B) Flow cytometry histograms of the PSI sensor output with and without expression of the hyper- inducing fusion N-Rnql-M in a [ps-] [rnq-] strain. (C) Flow cytometry histograms of the RNQ sensor output with and without expression of [PSI+]- inducing factors.

92 A [PS+] PHENOTYPE

N-Rnql-M N-Rnql-M uninduced induced

B C [RNQ+l (mq] FUSION N-Rnql-M in [rnq-] cells NM

100- uninduced Induced 80- E E 60- NM-Rnql '4- 06 0 40- 20- uninduced induced 0- 1 8 5 N-Rnql-M 101 10 10, PSI sensor mKate2 fluorescence (a.u.) 10 10 105 RNQ sensor mNeonGreen fluorescence (a.u.)

93 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Figure S4. Related to Figure 4. Sup35NM induction cures [RNQ*], but some mutants do not. Percentage of cells cured of [RNQ+] after overexpression and recovery of the indicated Sup35NM mutant alleles in dual PSI / RNQ sensor strains. Red denotes the allele used in the anti-PSI drive.

94 J1

percentage [RNQ+] cells cured 0 10 20 30 40 510 40SI 50 G25D 022R Q15R Q18R, Y32H Q10P N229D Y46C 095L C N81, V246A Q15R, 1152T WT E G20D L. Co) 015R QiOR Co) Q15R, P65S 061 L Y46C, Q95L S17R Q15R, K1391, Q242R N8D, Q70R, A237T S17R, Q56R Q10R, Q24R

95 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Figure S5. Related to Figure 5. Anti-prion and prion remodeling drive systems. (A) Table of dissected tetrads enumerating spores that exhibited the anti-PSI drive phenotype, conferred by either Sup35APD (S17R, Q56R mutant) or wild type Sup35NM control. (APD: anti-prion drive). (B) Flow cytometry histograms measuring the [PSI+] state of spores produced by the anti-PSI drive and wild type control. Top Left: One set of spores from a single tetrad harboring wild type Sup35NM. Bottom Left: Those same spores after treatment with guanidine to test for further curing. Top Right: One set of spores from a single tetrad harboring the anti-PSI drive. Bottom Right: Those same spores after treatment with guanidine to test for further curing. * denotes spores that inherited the drive cassette (LEU+). (C) Table of dissected tetrads enumerating spores that exhibited the anti-RNQ drive phenotype, conferred by either Rnq1APD (stop313) or wild type Rnq1 control. (D) Flow cytometry histograms measuring the [RNQ+] state of spores produced by the anti-RNQ drive and wild type control. Top Left: One set of spores from a single tetrad harboring wild type Rnql. Bottom Left: Those same spores after treatment with guanidine to test for further curing. Top Right: One set of spores from a single tetrad harboring the anti-RNQ drive. Bottom Right: Those same spores after treatment with guanidine to test for further curing. * denotes spores that inherited the drive cassette (LEU+). (E) Table of dissected tetrads enumerating spores that exhibited the RNQ remodeling drive phenotype, conferred by Rnq1RD (A288-298). (RD: remodeling drive) (F) Flow cytometry histograms measuring the [RNQ+] state of spores produced by the RNQ remodeling drive, compared to control [RNQ+] (light gray) and [mq-] (dark gray) strains. * denotes spores that inherited the drive cassette (LEU+).

96 -I

A C

TETRADS DRIVE TETRADS DRIVE CASSETTE DISSECTED PHENOTYPE CASSETTE DISSECTED PHENOTYPE

WT Sup35 9 0 / 36 spores WT Rnq1 7 1 / 28 spores Sup35wD 11 42 / 44 spores Rnq1 AD 11 43 / 44 spores (SI7R 066R) (9top3l3)I

Sup35APD Rnq1 B WT Sup35 FULL DRIVE D WT Rnq1 SEMI DRIVE 100- spore 1 spore 1* 100 - spore 1 spore 1 spore 2 spore 2* spore 2* spore 2* 80- spore 3' spore 3 80- spore 3' spore 4* spore 4 spore 4* spore 4 E 60 E 60- 0 0 40- 40- 0- 20- 20- 4 0- 0 1 3 5 1 0' 10" 10 5 10' 10" 10 10 103 10 10 103 10 GUANID INE I GUANIDINE cured cured cured cured 100- spore I spore 1* 100- spore 1 spore 1 spore 2 spore 2* spore 2* spore 2* 80- spore 3* spore 3 80- spore 3 spore 3' spore 4* spore 4 E spore 4* spore 4 E 60- %b-60- 0 40- 0 40- 20- 20- 0- 0- 10 103 10 10 103 101 101 103 L10 10 103 101 mNeonGreen (a.u.) mNeonGreen (a.u.) mNeonGreen (a.u.) mNeonGreen (a.u.)

100 A288-298 cured spore E F [RNQ*] control RN REMODELING DRIVE I 80- il spore 1* spore 2 E 60- spore 3 TETRADS DRIVE 0 spore 4* CASSETTE DISSECTED PHENOTYPE 40- Rnq1R 12 36 /48 spores 20- -51 10 10 10 RNQ sensor mNeonGreen fluorescence (a.u.)

97 Chapter 2: A genetic sensor for protein aggregation enables synthetic memory and anti-prion drives

Supplemental Table 1. Related to Figure 4. Sup35 and Rnql mutations isolated from genetic screen and their activities

Frequency Increase sensor Sup35 mutations out of 24 Heritable [PSI+] curing? solubility during seqs. overexpression? Q15-R, K139-1, Q242-R 4 Yes Yes Q18-R, Y32-H 2 Some (1-10%) Yes Q61-L 2 Yes Yes Q10-R, Q24-R, D500-V 2 Some (1-10%) Yes Q10-P, N229-D, H613-R 1 Some (1-10%) Yes N8-D, Q70-R, A237-T I Yes Yes N8-l, V246-A I Some (1-10%) Yes S17-R, Q56-R 1 Yes Yes Q15-R, P65-S 1 Yes Yes Y46-C, Q95-L 1 Yes Yes Q15-R, 1152-T, 1675-M 1 Yes Yes S17-R, L521-Q, L569-V 1 Yes Yes Q22-R 1 Yes Yes Q15-R 1 Yes Yes G20-D, N321-S 1 Some (1-10%) Yes Q1 5-R, N437-Y 1 Yes Yes G25-D 1 Some (1-10%) Yes Q10-R 1 Yes Yes

Frequency Increase sensor Rnql mutations out of 24 Heritable [RNQ+] curing? solubility during seqs. overexpression? L91-P 6 No No L138-P, stop 313 5 Yes Yes V100-G, A69-T 4 No Yes L97-F, R112-G, G149-S 4 No Yes F214-L, A288-298 2 Yes Yes S8-P, Y256-C 1 No Yes S241-G 1 No No S12-P, S241-G 1 No No

During selection, these mutants were expressed from endogenous promoters. Solubilizing activity may exist at that level of expression.

98 Table S2: Yeast strains used in Chapter 2

Strain Plasmids transformed Prion content Parent Phenotype Purposev Reference Base strain for transformation of Rnql, Mot3-PrD, Newl-PrD, and Landgraf et al., 4437 [RNQ+] Htt-related sensors PLoS ONE 2016 Osherovich and For PSI+][RNQ+] base Weissman, Cell YJW508 [PSI+][RNQ+] strain 2001 Osherovich et al., YJW584 [psi-][RNQ+] For [psi-][RNQ+] base strain PLoS Biol. 2004 Byrne et al.,PLoS YJW679 [PSI+][rng-] For [PSI+][rng-] base strain ONE 2009 (YJW508 cured by Osherovich and guanidine) For [psi-][rnq-] Weissman, Cell YJW509 [psi-][rng-] base strain 2001 L2776 strong [PSI+] YJW584 Detecing [PSI+] prion This study L2775 weak [PSI+] YJW584 Detecing [PSI+] prion This study Detecing [PSI+] prion in L2910 [psi-] alternate background This study Detecing [PSI+] prion in L2770 strong [PSI+] L2910 alternate background This study True and Detecing [PSI+] prion in Lindquist, Nature y3103 [psi-] alternate background 2000 Detecing [PSI+] prion in L2778 strong [PSI+] y3103 alternate background This study yGAN001 pGAN200 strong [PSI+] L2776 NATr Detecing [PSI+] prion This study yGAN002 pGAN200 weak [PSI+] L2775 NAT' Detecing [PSI+] prion This study yGAN003 pGAN200 [psi-] YJW584 NATr Detecing [PSI+] prion This study Detecing [PSI+] prion in yGAN004 pGAN200 strong [PSI+] L2770 NATr alternate background This study Detecing [PSI+] prion in yGAN005 pGAN200 [psi-] L291 0 NATr alternate background This study

99 Table S2: Yeast strains used in Chapter 2

Detecing [PSI+] prion in yGAN006 pGAN200 strong [PSI+] L2778 NATr alternate background This study Detecing [PSI+] prion in yGAN007 pGAN200 [psi-] y3103 NATr alternate background This study yGAN008 pGAN201 [RNQ+] 4437 NATr Detecting [RNQ+] prion This study yGAN009 pGAN201 [rng-] 4437 NATr Detecting [RNQ+] prion This study yGAN010 pGAN202 [SWI+][RNQ+] 4437 NAT Detecting [SWI+] prion This study yGAN01 1 pGAN202 [swi-][RNQ+] 4437 NAT Detecting [SWI+] prion This study Detecting [MOT3-PrD+] yGAN012 pGAN203 [MOT3-PrD+][RNQ+] 4437 NATr prion This study Detecting [MOT3-PrD+] yGAN013 pGAN203 [mot3-prd-][RNQ+] 4437 NAT' prion This study Detecting [NEW1-PrD+] yGAN014 pGAN204 [NEW1-PrD+][RNQ+] 4437 NATr prion This study Detecting [NEW1-PrD+] yGAN015 pGAN204 [newl-prd-][RNQ+] 4437 NATr prion This study Detecting solubility of Htt- yGAN016 pGAN205 [RNQ+] 4437 NATr Q25 This study Detecting solubility of Htt- yGAN017 pGAN206 [RNQ+] 4437 NAT Q103 This study Parent strain for Huntingtin yWRH148 pWRH74, pWRH75 [RNQ+] 4437 HIS+ LEU+ overexpression This study Detecting solubility of NAT, HIS+ Gall1-PrD Q->N with Htt- yGAN018 pGAN2O7,pWRH74, pWRH75 [RNQ+] yWRH148 LEU+ 103Q overexpression This study Detecting solubility of NATr HIS+ Gall 1 -PrD with Htt-1 03Q yGANO19 pGAN2O8,pWRH74, pWRH75 [RNQ+] yWRH148 LEU+ overexpression This study

Detecting solubility of Lsm4- NATr HIS+ PrD N->Q with Htt-1 03Q yGANO20 pGAN209,pWRH74, pWRH75 [RNQ+] yWRH148 LEU+ overexpression This study

Detecting solubility of Lsm4- NATr HIS+ PrD with Htt-103Q yGANO21 pGAN21O,pWRH74, pWRH75 [RNQ+] yWRH148 LEU+ overexpression This study

Detecting solbuility of Ure2- NATr HIS+ PrD N->Q with Htt-103Q yGANO22 pGAN211,pWRH74, pWRH75 [RNQ+] yWRH148 LEU+ overexpression This study

100 Detecting solubility of Ure2- NATE HIS+ PrD with Htt-1 03Q yGANO23 pGAN212,pWRH74, pWRH75 [RNQ+] yWRH148 LEU+ overexpression This study Detecting solubility of yGANO24 pGAN207 [RNQ+] 4437 NAT' Ga11-PrD This study Detecting solubility of yGANO25 pGAN208 [RNQ+] 4437 NAT' Gall 1-PrD Q->N This study Detecting solubility of Lsm4- yGAN026 pGAN209 [RNQ+] 4437 NAT' PrD N->Q This study Detecting solubility of Lsm4- yGANO27 pGAN210 [RNQ+] 4437 NATr PrD This study Detecting solbuility of Ure2- yGAN028 pGAN211 [RNQ+] 4437 NATr PrD N->Q This study Detecting solubility of Ure2- yGANO29 pGAN212 [RNQ+] 4437 NATr PrD This study NAT' ySK293 pSK226, pSK260 [PSI+][RNQ+] YJW508 KanMXr Dual PSI / RNQ sensing This study NATr ySK301 pSK226, pSK260 [psi-][RNQ+] YJW584 KanMXr Dual PSI / RNQ sensing This study NATE ySK305 pSK226, pSK260 [PSI+][rng-] YJW679 KanMXr Dual PSI / RNQ sensing This study NAT' ySK352 pSK226, pSK260 [psi-][rng-] YJW509 KanMXr Dual PSI / RNQ sensing This study Reverse-yTRAP sensing of ySK425 pSK221, pGAN230 [RNQ+] YJW584 NATrLEU+ [RNQ+] This study Reverse-yTRAP sensing of [rnq-] (pSK425 cured by ySK426 pSK221, pGAN230 [rnq-] YJW584 NATr LEU+ guanidine) This study

pGAN213, pJMB101, NATr HIS+ Estradiol induction of sTA to yGANO30 pGAN251 [RNQ+] 4437 TRP+ test signal response This study

NAT' KanMXr yGANO31 pGAN252, pJMB101 [psi-][RNQ+] ySK301 HIS+ TRP+ Estradiol induction of NM This study

101 Table S2: Yeast strains used in Chapter 2

NAT' KanMXr Estradiol induction of N- yGANO32 pGAN253, pJMB101 [psi-][RNQ+] ySK301 HIS+ TRP+ RNQ-M This study

NAT' KanMXr Estradiol induction of NM- yGANO33 pGAN254, pJMB101 [psi-][RNQ+] ySK301 HIS+ TRP+ RNQ This study NAT KanMXr yGANO34 pGAN256 [psi-][RNQ+] ySK301 HIS+ Synthetic memory of heat This study

NATr KanMXr Control for synthetic yGANO35 pGAN255 [psi-][RNQ+I ySK301 HIS+ memory of heat This study [PSI+][RNQ+] strain with NATr Sup35C overexpression to KanMX prevent toxicity from yGANO36 pGAN263, pJMB101 [PSI+][RNQ+] ySK293 HIS+ Sup35NM This study

NAT pGAN271, pGAN263, KanMXr For estradiol induction of yGANO37 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Rnq1 wt This study

NATr pGAN272, pGAN263, KanMXr For estradiol induction of yGANO38 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Rnql stop313 This study

NATr pGAN273, pGAN263, KanMXr For estradiol induction of yGANO39 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Rnq1 del-288-298 This study

NAT pGAN274, pGAN263, KanMXr For estradiol induction of yGAN040 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM wt This study

102 NAT' pGAN275, pGAN263, KanMX For estradiol induction of yGAN041 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM Y46C, Q95L This study

NAT' For estradiol induction of pGAN276, pGAN263, KanMXr Sup35NM Q15R, K1391, yGANO42 pJMB101 [PSI+][RNQ+] GANO36 HIS+ URA+ Q242R This study

NAT' pGAN277, pGAN263, KanMXr For estradiol induction of yGANO43 pJMB101 [PSI+][RNQ+] GANO36 HIS+ URA+ Sup35NM Q15R, P65S This study

NATr pGAN278, pGAN263, KanMX For estradiol induction of yGANO44 pJMB101 [PSI+][RNQ+] GANO36 HIS+ URA+ Sup35NM Q1OR, Q24R This study

NAT' For estradiol induction of pGAN279, pGAN263, KanMX Sup35NM N8D, Q70R, yGANO45 pJMB101 [PSI+][RNQ+] GANO36 HIS+ URA+ A237T This study

NATr pGAN280, pGAN263, KanMXr For estradiol induction of yGAN046 pJMB101 [PSI+][RNQ+] GANO36 HIS+ URA+ Sup35NM Q15R This study

NATr pGAN281, pGAN263, KanMXr For estradiol induction of yGANO47 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM Q15R, 1152T This study

NATr pGAN282, pGAN263, KanMXr For estradiol induction of yGANO48 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM Q61L This study

103 Table S2: Yeast strains used in Chapter 2

NAT' pGAN283, pGAN263, KanMXr For estradiol induction of yGANO49 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM S17R, Q56R This study

NAPr pGAN284, pGAN263, KanMXr For estradiol induction of yGAN050 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM Q22R This study

NATr pGAN285, pGAN263, KanMXr For estradiol induction of yGAN051 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM S17R This study

NAT pGAN286, pGAN263, KanMXr For estradiol induction of yGAN052 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM G20D This study

NATr pGAN287, pGAN263, KanMXr For estradiol induction of yGAN053 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM Q10P, N229D This study

NAT pGAN288, pGAN263, KanMXr For estradiol induction of yGAN054 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM Q1OR This study

NATr pGAN289, pGAN263, KanMXr For estradiol induction of yGAN055 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM Q18R, Y32H This study

NATr pGAN290, pGAN263, KanMXr For estradiol induction of yGAN056 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM G25D This study

104 NAT' pGAN291, pGAN263, KanMX For estradiol induction of yGAN057 pJMB101 [PSI+][RNQ+] yGANO36 HIS+ URA+ Sup35NM N81, V246A This study

pGAN200, pGAN262, NATr TRP+ Wt control of anti-PSI drive yGAN060 pGAN257 [psi-][RNQ+] YJW584 LEU+ haploid This study

pGAN200, pGAN262, NATrTRP+ Anti-PSI drive haploid, yGAN061 pGAN258 [psi-] YJW584 LEU+ GPDp-S17R, Q56R This study

NATr TRP+ Wt control of anti-RNQ drive yGAN062 pGAN2O1, pGAN259 [psi-][RNQ+] YJW584 LEU+ haploid This study

[psi-] weakened NATr TRP+ Anti-RNQ drive haploid, yGAN063 pGAN201, pGAN260 [RNQ+] YJW584 LEU+ GPDp-RNQ1 stop313 This study RNQ-remodeling drive [psi-] weakened NATr TRP+ haploid, GPDp-RNQ1 yGAN064 pGAN201, pGAN261 [RNQ+] YJW584 LEU+ del288-298 This study [PSI+][RNQ+] strain to mate NATr TRP+ with drives and test anti-PSI yGAN065 pGAN200, pGAN262, pAG413 [PSI+][RNQ+] YJW584 HIS+ drive phenotype This study [PSI+][RNQ+] strain to mate with drives and test anti- yGAN066 pGAN201, pAG413 [PSI+][RNQ+] YJW584 NATr HIS+ RNQ drive phenotype This study

pGAN200, pGAN262, NATr TRP+ Wt control of anti-PSI drive yGAN067 pGAN257 [PSI+] YJW584/yJW508 HIS+ LEU+ diploid This study

pGAN200, pGAN262, NAT' TRP+ Anti-PSI drive diploid, yGAN068 pGAN258 [psi-] YJW584/yJW508 HIS+ LEU+ GPDp-S17R, Q56R This study

NATr HIS+ Wt control of anti-RNQ drive yGAN069 pGAN201, pGAN259 [RNQ+] YJW584/yJW508 LEU+ diploid This study

105 Table S2: Yeast strains used in Chapter 2

NAT, HIS+ Anti-RNQ drive diploid, yGAN070 pGAN201, pGAN260 weakened [RNQ+] YJW584/yJW508 LEU+ GPDp-RNQ1 stop313 This study RNQ-remodeling drive NATr HIS+ diploid, GPDp-RNQ1 yGAN071 pGAN201, pGAN261 weakened [RNQ+] YJW584/yJW508 LEU+ del288-298 This study Spore 1 from tetrad 1 of wt yGAN072 pGAN200, pGAN262 [PSI+] YJW584/yJW508 NATrTRP+ NM control drive This study Spore 2 from tetrad 1 of wt yGAN073 pGAN200, pGAN262 [PSI+] YJW584/yJW508 NATr TRP+ NM control drive This study

pGAN200, pGAN262, NATr TRP+ Spore 3 from tetrad 1 of wt yGAN074 pGAN257 [PSI+] YJW584/yJW508 LEU+ NM control drive This study

pGAN200, pGAN262, NAT' TRP+ Spore 4 from tetrad 1 of wt yGAN075 pGAN257 [PSI+] YJW584/yJW508 LEU+ NM control drive This study

pGAN200, pGAN262, NATr TRP+ Spore 1 from tetrad 1 of yGAN076 pGAN258 [psi-] YJW584/yJW508 LEU+ anti-NM drive This study

pGAN200, pGAN262, NATrTRP+ Spore 2 from tetrad 1 of yGAN077 pGAN258 [psi-] YJW584/yJW508 LEU+ anti-NM drive This study Spore 3 from tetrad 1 of yGAN078 pGAN200, pGAN262 [psi-] YJW584/yJW508 NATrTRP+ anti-NM drive This study Spore 4 from tetrad 1 of yGAN079 pGAN200, pGAN262 [psi-] YJW584/yJW508 NATr TRP+ anti-NM drive This study Spore 1 from tetrad 1 of wt yGAN080 pGAN201 [RNQ+] YJW584/yJW508 NAT RNQ control drive This study Spore 2 from tetrad 1 of wt yGAN081 pGAN201, pGAN259 [RNQ+] YJW584/yJW508 NATr LEU+ RNQ control drive This study Spore 3 from tetrad 1 of wt yGAN082 pGAN201 [RNQ+] YJW584/yJW508 NATr RNQ control drive This study Spore 4 from tetrad 1 of wt yGAN083 pGAN201, pGAN259 [RNQ+] YJW584/yJW508 NATr LEU+ RNQ control drive This study

106 Spore 1 from tetrad 1 of yGAN084 pGAN201 [rng-] YJW584/yJW508 NATr anti-RNQ drive This study Spore 2 from tetrad 1 of yGAN085 pGAN201, pGAN260 weakened [RNQ+] YJW584/yJW508 NATr LEU+ anti-RNQ drive This study Spore 3 from tetrad 1 of yGAN086 pGAN201, pGAN260 weakened [RNQ+] YJW584/yJW508 NAT LEU+ anti-RNQ drive This study Spore 4 from tetrad 1 of yGAN087 pGAN201 [rnq-] YJW584/yJW508 NATr anti-RNQ drive This study Spore 1 from tetrad 1 of yGAN088 pGAN201, pGAN261 weakened [RNQ+] YJW584/yJW508 NATr LEU+ RNQ-remodeling drive This study Spore 2 from tetrad 1 of yGAN089 pGAN201 weakened [RNQ+] YJW584/yJW508 NAT, RNQ-remodeling drive This study Spore 3 from tetrad 1 of yGAN090 pGAN201 weakened [RNQ+] YJW584/yJW508 NATr RNQ-remodeling drive This study Spore 4 from tetrad 1 of yGAN091 pGAN201, pGAN261 weakened [RNQ+] YJW584/yJW508 NATr LEU+ RNQ-remodeling drive This study

107 Table S3: Plasmids used in Chapter 2

Yeast Integration Plasmid Content selection locus Purpose Reference Youk and Lim, Science pNH603 empty -his HIS3 Parent plasmid for single-integration at HIS3 2014 Youk and Lim, Science pNH604 empty -trp TRP1 Parent plasmid for single-integration at TRP1 2014 Youk and Lim, Science pNH605 empty -leu LEU2 Parent plasmid for single-integration at LEU2 2014 Landgraf et pFA6-HO-GPDp-mCherry-ADH 1 term- Parent of pGAN147. Promoter and MCS cut al., PLoS pDML1 12 NAT-HO nourseothricin HO out to insert yTRAP cassettes ONE 2016 Alberti et al., pAG303 pAG303-GPD-ccdB -his HIS3 Parent plasmid for Sup35C integration Yeast 2007 Alberti et al., pAG304 pAG304-GPD-ccdB -trp TRP1 Parent plasmid for Sup35C integration Yeast 2007 Alberti et al., pAG305 pAG305-GPD-ccdB-HA -leu LEU2 Parent plasmid for PSI-inducing alleles Yeast 2007 Alberti et al., pAG416 pAG416-GPD-ccdB-GFP -ura N/A Parent plasmid for pZ-inducible mutant alleles Yeast 2007 Empty plasmid as a marker to allow selection Alberti et al., pAG413 pAG413-GAL-ccdB -his N/A of diploids Yeast 2007 Alberti et al., pAG426 pAG426-GAL-ccdB-GFP -ura N/A Parent plasmid for pGAN101 Yeast 2007 Plasmid for overexpression of prion domains to pGAN101 pAG426-GAL-ccdB-mKate2 -ura N/A induce prions This study SUP35p-ccdB-sTA(43-8)-SUP35term 8xop-crCYC1 p-NeonGreen- This is the empty basic yTRAP sensor plasmid, pGAN147 ADHiterm nourseothricin HO constructed from pDML112 This study pGAN101 with Swil inserted into Plasmid for overexpression of Swil domains to pGAN160 ccdB site with LR reaction -ura N/A induce [SWI+] This study pGAN101 with Mot3-PrD inserted into Plasmid for overexpression of Mot3-PrD to pGAN161 ccdB site with LR reaction -ura N/A induce [MOT3-PRD+] This study

108 pGAN101 with New1-PrD inserted Plasmid for overexpression of New1 -PrD to pGAN 162 into ccdB site with LR reaction -ura N/A induce [NEW1-PRD+] This study pGAN147 with Sup35NM inserted into pGAN200 ccdB site with LR reaction nourseothricin HO For sensing [PSI+] This study pGAN147 with Rnql inserted into pGAN201 ccdB site with LR reaction nourseothricin HO For sensing [RNQ+] This study pGAN147 with Swi1 inserted into pGAN202 ccdB site with LR reaction nourseothricin HO For sensing [SWI+] This study pGAN147 with Mot3-PrD inserted into pGAN203 ccdB site with LR reaction nourseothricin HO For sensing [MOT3-PRD+] This study pGAN147 with New1-PrD inserted pGAN204 into ccdB site with LR reaction nourseothricin HO For sensing [NEW1-PRD+] This study pGAN147 with Htt-Q25 inserted into pGAN205 ccdB site with LR reaction nourseothricin HO For sensing Htt-Q25 This study pGAN147 with Htt-Ql03 inserted into pGAN206 ccdB site with LR reaction nourseothricin HO For sensing Htt-Ql 03 This study pGAN 147 with Gall 1-PrD inserted pGAN207 into ccdB site with LR reaction nourseothricin HO For sensing Gall 1-PrD This study pGAN 147 with Gal1-PrD Q->N inserted into ccdB site with LR pGAN208 reaction _nourseothricin HO For sensing Gall11-PrD Q->N This study pGAN147 with Lsm4-PrD N->Q inserted into ccdB site with LR pGAN209 reaction nourseothricin HO For sensing Lsm4-PrD N->Q This study pGAN147 with Lsm4-PrD inserted into pGAN210 ccdB site with LR reaction nourseothricin HO For sensing Lsm4-PrD This study pGAN147 with Ure2-PrD N->Q inserted into ccdB site with LR pGAN211 reaction nourseothricin HO For sensing Ure2-PrD N->Q This study pGAN147 with Ure2-PrD inserted into pGAN212 ccdB site with LR reaction nourseothricin HO For sensing Ure2-PrD This study pGAN147 With reporter only (no yTRAP fusion). 8xop-crCYC 1p- for measuring background levels of reporter pGAN213 NeonGreen-ADHl term nourseothricin HO protein This study SUP35p-ccdB-sTA(37-12)- SUP35term 8x37-12op-crCYClp- For sensing 2nd cassette with orthogonal ZF pSK247 NeonGreen-ADH1term KanMX LEU2 (dual yTRAP) This study

109 Table S3: Plasmids used in Chapter 2

pSK247 With Rnq1 inserted into ccdB For sensing [RNQ+] alongside another protein pSK226 site with LR reaction KanMX LEU2 (dual yTRAP) This study pSUP35-SUP35NM-43-8-VP16, 8x43- pSK260 8 op-pCYC1-mKate2 nourseothricin HO Sensing [PSI+] with mKate2 This study pSK221 pNH6O5-ADH1p(tetO2)-EGFP -leu LEU2 Tet-repressible construct for reverse yTRAP This study SUP35p-ccdB-sTA(43-8)-SUP35term 8xop-crCYC1 p-TetR-mKate2- pGAN230 ADH1term nourseothricin HO pGAN147 with TetR-mKate2 as reporter This study pNH603GALp-FLAG-HttExl -Q1 03- Htt-Q103 overexpression (dead GFP tag for pWRH74 ymsfGFP (Y66L) -his HIS3 consistency with other cell lines) This study pNH605GALp-FLAG-HttExl -Q1 03- Htt-Q103 overexpression (dead GFP tag for pWRH75 ymsfGFP (Y66L) -leu LEU2 consistency with other cell lines) This study pNH604 crADH 1 p-ZEM-ADH 1 term(C. pJMB101 albicans) -trp TRP1 ZEM estradiol-inducible transcription factor This study Aranda-DIaz et al., ACS Parent plasmid for estradiol induction from an Synth. Biol. pHES835 pNH6O3 pZ-YFP -his HIS3 integrated cassette 2016 pNH603 pZ-(43-8)sTF-ADH1term(C. pGAN251 albicans) -his HIS3 For estradiol induction of the sTA This study pNH603 pZ-Sup35NM-ADHlterm(C. pGAN252 albicans) -his HIS3 For estradiol induction of NM This study pNH603 pZ-N-Rnql-M-ADHlterm(C. pGAN253 albicans) -his HIS3 For estradiol induction of N-Rnq1-M This study pNH603 pZ-NM-Rnql-ADHlterm(C. pGAN254 albicans) -his HIS3 For estradiol induction of NM-Rnql This study pNH603 pSsa4-Sup35NM- pGAN255 ADHlterm(C. albicans) -his HIS4 For heat-sensitive expression of NM (control) This study pNH603 pSsa4-N-Rnql-M- pGAN256 ADHlterm(C. albicans) -his HIS5 For heat-sensitive expression of N-Rnq1-M This study pAG305-GPDp-Sup35NM wt-HA- pGAN257 CYCiterm -leu LEU2 Anti-prion drive control plasmid expressing NM This study pAG305-GPDp-Sup35NM S17R, pGAN258 Q56R-HA-CYC1term -leu LEU2 Anti-prion drive for PSI This study pAG305-GPDp-Rnq1 wt-HA- Anti-prion drive control plasmid expressing pGAN259 CYCiterm -leu LEU2 Rnq1 This study

110 pAG305-GPDp-Rnql stop313-HA- pGAN260 CYC1term -leu LEU2 Anti prion drive for RNQ This study pAG305-GPDp-Rnql del288-298-HA- pGAN261 CYCIterm -leu LEU2 Prion-remodeling drive for RNQ This study Sup35C overexpression to eliminate toxicity pGAN262 pAG304-GPDp-Sup35C-CYCiterm -trp TRP1 from Sup35NM expression This study Sup35C overexpression to eliminate toxicity pGAN263 pAG303-GPDp-Sup35C-CYClterm -his HIS3 from Sup35NM expression This study Parent plasmid (gateway compatible) for pGAN270 pAG416-pZ-ccdB-mKate2 -ura N/A cloning estradiol-inducible mutants This study pGAN271 pAG416-pZ-RNQ1wt-mKate2 -ura N/A For estradiol induction of-Rnql wt This study pGAN272 pAG416-pZ-RNQ stop313-mKate2 -ura N/A For estradiol induction of Rnql mutant This study pAG416-pZ-RNQ1 del288-298- pGAN273 mKate2 -ura N/A For estradiol induction of Rnq1 mutant This study pGAN274 pAG416-pZ-Sup35NM wt-mKate2 -ura N/A For estradiol induction of Sup35NM wt This study pAG416-pZ-Sup35NM Y46C, Q95L- pGAN275 mKate2 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM Q15R, K1391, pGAN276 Q242R-mKate2 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM Q15R, P65S- pGAN277 mKate2 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM Q1OR, Q24R- pGAN278 mKate2 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM N8D, Q70R, pGAN279 A237T-mKate2 -ura N/A For estradiol induction of Sup35NM mutant This study pGAN280 pAG416-pZ-Sup35NM Q15R-mKate2 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM Q15R, 1152T- pGAN281 mKate2 -ura N/A For estradiol induction of Sup35NM mutant This study pGAN282 pAG416-pZ-Sup35NM Q61 L-mKate3 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM S17R, Q56R- pGAN283 mKate4 -ura N/A For estradiol induction of Sup35NM mutant This study pGAN284 pAG416-pZ-Sup35NM Q22R-mKate5 -ura N/A For estradiol induction of Sup35NM mutant This study pGAN285 pAG416-pZ-Sup35NM S17R-mKate6 -ura N/A For estradiol induction of Sup35NM mutant This study pGAN286 pAG416-pZ-Sup35NM G20D-mKate7 -ura N/A For estradiol induction of Sup35NM mutant This study

111 Table S3: Plasmids used in Chapter 2

pAG416-pZ-Sup35NM Q10P, N229D- pGAN287 mKate8 -ura N/A For estradiol induction of Sup35NM mutant This study pGAN288 pAG416-pZ-Sup35NM Q1OR-mKate9 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM Q18R, Y32H- pGAN289 mKatel0 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM G25D- pGAN290 mKatel 1 -ura N/A For estradiol induction of Sup35NM mutant This study pAG416-pZ-Sup35NM N8, V246A- pGAN291 mKate12 -ura N/A For estradiol induction of Sup35NM mutant This study

112 Chapter 3

SWI Pioneers: A fungal prion promotes geographic and sexual exploration

113 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

SUMMARY

To thrive in an ever-changing environment, microbes must distribute their progeny widely to colonize new and fertile grounds. Furthermore, they must continually evolve and adapt to the stresses of their surroundings. In both of these regards, diversity is key

- if an entire population moved together or responded to the environment in the same way, it could easily go extinct. Here we show that the epigenetic prion switch [SWI*] 1 3 establishes specialized subpopulations with "pioneering" and "settled" phenotypic programs in Saccharomyces cerevisiae. In the pioneering state, cells readily expand geographically, carried by the flow of water. Pioneers are also more likely to find and mate with genetically diverse partners as mother cells shun their own daughters. In the settled state, cells form protective flocs and tend to remain in their current position.

Settled cells are better able to withstand harsh conditions like drought and alkaline pH.

The [SWI'] prion is a concerted epigenetic switch that specializes yeast subpopulations to the ecological advantage of the population as a whole. Our findings, together with the recent identification of the first bacterial prion switch4 , suggest that microbial specialization may be a powerful and common adaptation strategy.

RESULTS AND DISCUSSION

Microbes face a constant struggle for resources and survival in an environment that changes by the hour. Wide temperature fluctuations, severe drought, and other environmental factors can cause mass extinction events, leaving vast regions for surviving cells to recolonize. Likewise, events such as floods or fruit falling from trees may fill new areas with nutrients, enabling expansion and growth of microbial

114 populations. However, exploration of a new area with unknown resources carries a

significant risk of death. How have microbes evolved to spread efficiently without

excessive risk?

Prions are powerful evolutionary devices by which yeast cells switch between

phenotypic programs, 'hedging their bets' to survive in fluctuating and unpredictable

environments5-10. Prion proteins can exist in at least two stable conformations, each

having an altered function - typically a soluble, high-activity "[prion] state", and an

aggregated, low-activity "[PRION*] state." At least ten different proteins in S. cerevisiae

have the capacity to propagate in alternate prion states'1 13 . Thus, the combination of

potential prion states can lead to a plethora of switchable phenotypes.

An advantage of prion-based phenotypes is that they can be totally reversed by a return

to the original protein conformation. In contrast, a new phenotype gained by genetic

mutation is very difficult perfectly reverse, particularly in the case of loss-of-function

mutations and deletions. Prion-based adaptation is thus ideal for rapidly changing

between different phenotypic programs that are each advantageous in different

situations14 . The [SWI*] prion has been shown to cause slow growth in non-fermentable

carbon sources 2 and a loss of invasiveness and flocculation', but has not previously

been demonstrated to provide any advantage to yeast.

The protein that underlies the [SWI/] prion state, Swil, is a DNA-binding protein and

subunit of the SWI/SNF chromatin remodeling complex. The [SW*] prion stochastically

arises at a low rate in the population 3, and is readily lost upon stress 15. The [SWI/] prion

was recently shown to regulate adhesion, invasion, and flocculation through a

prominent effect on the expression level of flocculin genes. In the [SWI/] state, FLOl

115 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

Figure 1. The [SWI*] prion enhances the migration of cells with the flow of water. A) A diagram of the experimental procedures used to test the migration of cells on solid media. Pre- grown spots of yeast colonies experience 'rainfall' as drops of water are pipetted over them. The plates are tilted to allow the water to flow across the surface. After drying and incubating for yeast growth, colonies established by the migrated cells appear. B) Photograph showing the migration of [swi-] and [SWI+] cells on an agar plate. C) Quantification of total cell number after migration experiment. Error bars indicate standard deviation from triplicate experiments. D) Left: Photograph showing the migration of cells on an agar plate after [SWI+] and [swi-] cells were mixed in the indicated proportions. Arrows indicate large colonies where small flocs of [swi-] cells have migrated. Center: The same image collected in a green fluorescence channel. Due to the yTRAP sensor (Newby, et. al., in review) in both strains, [swi-] colonies are brightly fluorescent. Right: Migration experiment from Panel B photographed in the green fluorescence channel for comparison. and FLO11, and thus flocculation and invasion, are greatly repressed. Could this loss of multicellularity confer advantages in microbial migration?

One likely means of microbial migration is the flow and agitation of water, especially during rainfall. We designed experiments to test the effect of [SWI+]-dependent flocculation on migration under rain-like conditions. [SWI/] and [sw-] strains were spotted side-by-side on agar plates. After one day of growth on solid substrate, small droplets of water were added onto each spot of yeast and the plate was tipped to one side to allow the water droplet to flow down its surface (Fig 1A). A multichannel pipette was used to ensure equal 'rainfall' on each spot. Plates were dried and photographed after one more day of growth (Fig 1 B). The [SWI*] prion conferred a clear advantage for the spreading of yeast away from the original spot, allowing cells to colonize the entire area touched by the flowing water. Scraping each lane separately into solution and subsequent measurement by flow cytometry revealed that there were about four times more [SWI/] cells than [swi] cells at the conclusion of this experiment (Fig 1 C), demonstrating the potential growth advantage of a migratory state. This migratory enhancement is specifically due to the prion's effect mediated through the flocculin

116 -I

I I A B C [swi] [SWIi] Total cells after migration Pre-grown yeast spots Raindrol flow Agar plate 1X109 1. 'Raindrops' of water added to yeast spots J

2. Plate tilted to let raindrops flow 6x1 0" Raindrop flow 0

2x108 r 0 3. Plate dried and incubated to let N X migrating cells grow Migrated cells I D

[SwrI:[swtJ 1:20 1:10 1:4 1:1 1:20 1:10 1:4 1:1 O fsSlINwN

genes, as [SWI/] does not affect migration in cells that do not express flocculins (Fig

S1A).

Could a small population of [SWI*] cells confer a migratory benefit to their [sw-] neighbors? It is logical to expect that [SWI/] cells might disrupt larger flocs that would otherwise be formed by adjacent [swt] partners, thus encouraging the escape and migration of [sw-] cells with flowing water. We tested for such a cooperative migratory

117 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

Figure 2. The [SWIP] prion encourages diverse mating partners. A) A diagram showing the mating tendencies of [SWI+] and [swi-] haploids (budding yellow cell). Left: A [swi-] cell will readily switch its mating type after dividing and mate with its own daughter. The resulting diploid cannot out-cross with diverse mating partners (orange cell). Not depicted: an additional generation may occur during mating type switching, at which point all four haploids from the lineage will mate pairwise. Right: In the [SWI+] pioneering state, mating type switching is inhibited and thus mother cells cannot mate with their own daughters. This increases the likelihood that they will mate with genetically diverse partners. B) Relative out-crossing efficiencies of [swi-] cells (blue) and [SWI+] cells (red). Elimination of the prion from the [SWI+] strain ('cured') returns its out-cross ratio to the low state. The ratio was calculated by normalizing to the mating efficiency of ho- control cells (Fig S2A). Error bars indicate standard deviation of triplicate mating tests. effect by seeding different ratios of [SWI/] : [sw-] cells into spots on agar plates. The same simulated rainfall procedure was applied and migrated cells were grown for one day and imaged (Fig 1 D). Only when [SWI/] and [sw-] populations were mixed, larger colonies appeared within the migrants; these were likely formed by the migration of clusters of cells. To confirm the prion state of these colonies, we utilized the yTRAP sensor for Swil (Newby, et. al. in review), which yields bright fluorescence in [sw] cells and dim fluorescence in [SWI*] cells. The large migrated colonies from mixed spots were indeed brightly fluorescent, and thus [sw-] (Fig 1 D, left and center, arrows). In contrast, when [SWI*] and [sw-] were spotted separately, no bright, [sw-] clustered cells migrated (Fig 1 D, right). We conclude that [SWI/] not only migrates more efficiently itself, but also can enhance the migration of otherwise-immobile, [swr] neighbors.

In liquid growth media, the [SWI/] prion has a noticeable effect on the partitioning of cells into clumped flocs (dominated by [sw-] cells) versus the soluble 'supernatant'

(dominated by [SWI] cells)1. We quantified this effect using yTRAP - [SWI/] and [sw-] cells were co-inoculated in equal measure and grown for 16 hours. We then used flowcytometry to measure the prion state of cells in the supernatant vs the total culture after dispersion of flocs (Fig S1 B). The [SWI/] cells were generally less fit and comprised less than 40% of the total culture. However, in the supernatant fraction 118 U. I A [swi- [SW/+] Tends to in-breed Tends to out-cross

MAT a MAT A MAT a MAT A 4%i Diverse strain [swi-] yn Diverse strain [SWI+]

MAT A MAT a MAT a MAT A MAT a MAT A Diverse strain SWITCHED [swi] Diverse strai. SWITCHING [swI- INHIBITED [SWIl]

MAT A MAT a [SWg] Diverse strain Homozygous Heterozygous Diploid Diploid [swi [SWI']

B 0.8- m [swi~ ]

*0 0.6- [SW]

. 0.4- T 0 0.2-

0.0-"

Np

[SWI] cells comprised more than 70%. When cultures were grown in the presence of ethanol, a by-product of yeast growth that is known to enhance flocculation16 , the fraction of [SWI*] cells in the supernatant increased to over 90%. We hypothesized that 119 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

Figure 3. [SWIr] cells have lower fitness and do not survive dry conditions. A) The survival of cells from overnight cultures (no stress), 6-day starved cultures (starvation) and antibiotic-treated cultures (caspofungin) were measured using a viability stain and flow cytometry. Blue indicates [swi-] cells and red, [SWI+] cells. B) Growth comparison of [swi-] cells (blue) and [SWI+] cells (red) in media buffered at pH 7.5. Cell density was measured by absorbance every 15 minutes. C) Survival of [SWI+] and [swi-] strains after drying under blowing air for 24 hours. The survival of [SWI+] was below the detection limit by viability stain and flow cytometry (1 in 10,000 cells). Error bars indicate standard deviation (n=3) D) Recovery of dried [SWI+] and [swi-] strains in liquid culture in microtiter plates. After rehydration of dried cells into the original volume of the overnight culture, a 25-fold dilution into fresh media was made in microtiter wells. Growth after 4 days is depicted. these supernatant cells would be more capable of migration, as they would overflow during rainfall. We tested the overflow of liquid cultures in a laboratory setup (Fig S1C).

Indeed, [SWI/] cells had a strong advantage in escaping to colonize new grounds (Fig

S1 D). These results establish a clear benefit to [SWI*] cells in migration in response to the common environmental stimulus of flowing or falling water. We therefore term

[SWI'] cells "pioneers," and the non-migratory [swi] cells "settled."

Swil's effect on transcription is not limited to flocculins. In fact, the SW/1 gene is so named because its mutation leads to a defect in transcription at the HO locus, which is required for mating type switching. Mating type switching occurs after a diploid cell has sporulated to produce four haploid spores. Haploid cells can mate with one another to return to the more robust diploid state, but only if the two mating partners are of opposite mating types. Spores containing a functional HO gene will commonly divide once, after which the mother cell switches its mating type and mates with its own daughter cell, generating a homozygous diploid 17. While this does not increase genetic diversity, it does return both cells to the more robust diploid state without the risks of out-crossing (for example, the risks of acquiring transposons, viral infections, 2p DNA, or non-adaptive prion states). Because mating type switching occurs readily in the lab, it

120 -I A B Growth at pH 7.5 100- +.0 (swI- M [SW/I+ _0 Ism]l M [swIl 0.6 C- 50. 10 1 M .0& -i[S 0 0.4 0. imss.

0.2

0.0 0 20 40 60 Time (hours) ~01 0

C D

m SWrj] 4- Recovery after drying m[swIl _T_ 3- [SWI] C 2- [swi] <1 in a1. 0.01%

0- I

L _ I is a mystery why so many natural yeast isolates are heterozygous and appear to have mated most recently with diverse partners (out-crossing) rather than daughter cells 18.

We hypothesized that the [SWI*] prion, like SWII mutants, may confer a defect in HO expression and thereby increase out-crossing frequency. To test this, we measured the out-crossing efficiency of [SWI] and [sw-] HO' spores. Our measurement was normalized to identical ho- control spores to specifically determine the prion effects mediated through HO (experimental procedure: Fig S2). Strikingly, the [SWI/] prion

121 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

greatly increased the out-crossing efficiency of HO' spores (Fig 2). Elimination of the

prion by passaging on guanidine2 (curing) returned the out-crossing frequency to that of the original [sw-] strain, confirming that the effect was indeed mediated through the

prion state. Thus, the [SWI*] prion acts as a deterrent for in-breeding and encourages

the diversification of genomes by out-crossing. We propose that a combined effect of

enhanced migration, which would increase the chance of encountering diverse strains, with enhanced out-crossing, vastly increases the likelihood of generating new hybrid

diploids.

[SW*] is acutely sensitive to perturbations in the Hsp70 chaperone system. Proteotoxic

stress such as elevated temperature will readily eliminate the prion from cells15. How

might cells benefit from eliminating the prion in response to stress? To test this, we

compared the survival and growth of [SWI/] and [sw-] strains across a number of

conditions (Fig 3). In every condition tested, the 'settled' [sw-] state was advantageous.

After 6 days of glucose starvation or one day of antimicrobial (caspofungin) treatment,

the flocculated [sw-] cells showed increased survival (Fig 3A). The growth defect of

[SWI/] cells was particularly dramatic when media was buffered to pH 7.5 (Fig 3B), but was also present when cells were grown at favorable pH supplemented with glucose or

raffinose (Fig S3). [swf] cultures were also uniquely able to withstand dry conditions -

after 24 hours under a draft of sterile air, about 3% of [swr] cells survived, while the

survival of [SWI] cells was below the detection limit (Fig 3C). Upon rehydration and

subsequent incubation, [swI] cells recovered and grew in microtiter wells, while [SW*]

cells did not (Fig 3D). This demonstrates the advantage of the settled cellular program

and provides an evolutionary rationale for switching to the settled state under stress.

122 Notably, the human homolog of Swil, ARID1A, is commonly mutated in multiple types of cancer 19. Loss of ARID1A confers increased cellular migration and metastases, and

2 02 1 is associated with poor prognosis , . This is reminiscent of the prion state of Swil, which also mimics a loss-of-function'. Our discovery that Swil regulates cellular migratory states may point to a general role of Swil family members in controlling cell migration.

5 0 Prions are thought to act as bet-hedging elements in yeast ,6,8-1 . That is, they confer environment-specific growth advantages so that a diverse population with many prion states - and thus many hedged bets - is more likely to have subsets of cells that are already adapted to unpredictable future environments. The [SW*] prion has not previously been shown to confer a growth advantage in any environment, so its function as a bet-hedging element was mysterious. We have demonstrated that [SWI/] can indeed benefit the growth of the total population when the opportunity to migrate arises

(Fig 1 C). We propose that, in addition to simple bet-hedging of growth rates, prions can generate specialized subpopulations of cells with the ability to take advantage of unique ecological niches. For example, [SW*] cells are specialized "pioneers" that travel with the flow of water and mate with diverse partners. With the recent discovery of heritable prion states in bacteria4 , it is plausible that similar specialization strategies are widely used in single-celled organisms to switch between risky and potentially beneficial phenotypic programs.

123 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

METHODS

Cloning and vector construction:

The yTRAP sensor plasmid pGAN202 (Newby, et. al. in review) was used to detect the [SW*] prion. This plasmid integrates into the HO locus and generates a large deletion. It produces bright green fluorescence in the [sw-] state and dim fluorescence in the [SWI/] state, allowing the detection of prion in a cell or colony.

The multicopy, episomal plasmid pGAN 160 (Newby, et. al. in review) was used to induce the [SWI/] prion in cells by overexpression of Swil from the GALl promoter in galactose.

The FLO8-correction plasmid pGAN300 was made using pAG305GAL20 pAG305GAL was cut with the restriction enzymes Xbal and Xhol, followed by ligation of the wild-type FLO8 coding sequence. The restriction enzyme BstAPI was used to cut the plasmid between the start codon of FLO8 and the site of its mutation in laboratory strains to favor correction of the endogenous copy.

Yeast strains and growth media:

Yeast strains used in this study were BY4741 trpl:KanMX from the Invitrogen deletion library (Cat. no. 95401.H2), SK1 21 (HO+ a/a), and derivatives BY4741 22 (his3dl, leu2A, met15A, ura3A, ho-). For a complete list, please see Supplemental Table 1.

Growth media was YPD or complete standard synthetic media (CSM) or media lacking certain amino acids with either 2% glucose, 2% galactose, or 2 % raffinose supplemented as the carbon source. [SW'] was cured by streaking on YPD agar plates

124 supplemented with 5mM guanidine hydrochloride. Yeast transformations were

conducted using a lithium acetate competent cell protocol as previously described23 . To

calculate the OD600 of flocculating cells, flocs were fully dispersed by the addition of

25mM EDTA followed by vigorous vortexting.

To induce the [SW'] prion, the Swil overexpression plasmid pGAN160 was

transformed into a strain containing the yTRAP sensor (pGAN202). The Swil protein

was transiently overexpressed by growth for 16 hours in galactose followed by plating

on glucose. Potential [SW*] colonies had heritable, low-fluorescence. The presence of

the prion was confirmed if streaking on guanidine media returned cells to the bright,

[swi-] state.

Migration from liquid culture:

To test for the ability of cells to emerge from an overflowing liquid pool, two-tiered

holes were produced in agar YPD medium. Overnight 50pL cultures of CSM +2%

glucose +5% ethanol (strains yGAN102 and yGAN103) were pipetted into the bottom

hole and allowed to settle for 1 hour. A multichannel pipette was used to transfer 800pL

of sterile water into each hole simultaneously, so that the bottom tier overflows onto the

top. Then, the water was aspirated away and the plates were dried under flowing air in a

hood for one hour, then transferred to 300C to allow colonies to grow. Migrated colonies

were photographed on a Bio-Rad ChemiDoc.

In order to make the two-tiered holes, a custom mold was ordered from the MIT

machine shop made from PDMS. 6 pins were inserted into the mold, each of which

formed a two-tiered hole. The bottom tier hole had a radius of 0.15 in and a height of

125 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

0.47 in. The top hole had a radius of 0.9 in and a height of 0.47 in. Each pin was placed

1.4 in apart. The sides of each tier were slanted at 60 to ease removal from solidified agar.

Supernatant/total culture quantification:

Quadruplicate, 4mL YPD or YPD +5% ethanol cultures were inoculated with 50%

[SW*] (yGAN103) and 50% [swi-] (yGAN102) to a final OD600 of 0.02. Cultures were grown for 16 hours at 300C with shaking. Tubes were let sit for 15 minutes on the bench without agitation. Samples were collected from the top of the culture (-0.1 OD600 units), then cultures were fully resuspended with EDTA to disperse flocs. Additional samples were collected. All samples were centrifuged, resuspended in 500uL fresh

YPD and incubated 4 hours at 300C in deep well 96-well plates to reach log phase before measurement. EDTA was added to disperse flocs and samples were measured by flow cytometry using the MACSquant VYB (Miltenyi Biotec). The percentage of

[SW*] cells was calculated by gating in the GFP/SSC channels.

Migration on solid agar plates:

Overnight cultures of yGAN102 and yGAN103 (OD600 - 8) were spotted on rectangular YPD plates. After 16 hours of growth at 300C, 14uL of sterile water was pipetted on the spots using a multichannel pipet to ensure even flow. Plates were immediately tipped to the side until vertical, allowing the drops of water to flow down the surface of the plate. Plates were incubated 16 hours more at 300C, then photographed.

The BioRad ChemiDoc was used to collect images in the green fluorescent channel. An

Epson document scanner was used to take normal photographs.

126 To calculate the total number of cells after migration, the original yeast spot and all yeast that grew on its trail of migration were scraped into 500uL CSM. This was diluted 100-fold and measured by flow cytometry as above. The number of cells per pL was used to calculate the total number of cells that were present on the plate.

For mixtures of [SWI*] and [swi-] cultures, EDTA was added to all cultures to disperse flocs (and for consistency in [SWI] cultures, which did not flocculate). ODs were normalized, and cultures were mixed in the given ratios, then centrifuged and resuspended in media lacking EDTA. These were spotted onto agar plates as above.

To calculate the number of migrated [swi-] cells, all cells halfway down the plate or more were collected by scraping into 500pL of CSM. These were measured by flow cytometry and gated in the green/SSC channels to determine the total number of [swi-] cells.

Out-cross ratio measurement:

The procedure to calculate the out-cross ratio is summarized in Fig S2. The procedure is complicated because HO+ haploids cannot be propagated for experiments.

The homozygous, HO diploid SKi was sporulated by overnight incubation in 2% potassium acetate. SKi spores were mated to flo8- [SWI*] and [swi-] strains (yGAN 100 and yGAN101) by co-inoculation into YPD and overnight growth. Hybrid diploids were selected by growth on media lacking histidine, supplemented with 100ug/mL nourseothricin. The [SW*] diploid was cured by passaging on guanidine to generate an additional control for the effect of the prion. This generated yeast strains yGAN104-106

(see Supplemental Table 1).

127 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

These hybrid diploids contained one functional copy of HO (from SK1) and one mutant copy (from the [SWI] sensor strain). Due to the HO-integrated pGAN202, the mutant ho- allele is marked with the expression of a green fluorescent protein. Upon sporulation, the green, ho- spores will be preferentially able to mate with an exogenous strain - HO+, non-fluorescent spores can instead switch mating types and mate with their daughters. To test whether the [SWI*] prion affects this tendency, we sporulated the diploids for 7 days in 2% potassium acetate, followed by vigorous vortexing. We then mixed triplicate mating tubes of YPD containing these spores and BY4741 trpl:KanMX, at an OD600 of 0.01 each. Mating tubes were incubated at 300C for 16 hours.

Diploids formed between the exogenous strain (BY4741 trpl:KanMX) and the spores were selected by 20-fold dilution into synthetic media lacking tryptophan and supplemented with 1 OOpg/mL geneticin (Sigma Aldrich Al 720). After 24 hours of growth, selection was continued with a further 100-fold dilution into the same selective media. After another 24 hours of growth, cultures were diluted 100-fold and grown 6 hours to log phase before flow cytometry measurement as above. The outcross ratio was calculated as the number of non-fluorescent diploids (formed from HO+ spores) divided by the number of fluorescent diploids (formed from ho- spores). This allowed us to normalize specifically to the effect of the [SWI/] prion mediated through the HO locus.

Survival measurement:

Small colonies of FLO8+ [swi-] or [SWI/] strains (yGAN102 and yGAN103) were picked into triplicate cultures of synthetic complete media. 100ng/mL caspofungin

(Abcam ab145180) was used for caspofungin treatment. After overnight growth (or one

128 week of incubation in the case of glucose-starved cultures), cultures were centrifuged and resuspended in fresh CSM. 10ug/mL propidium iodide (Sigma Aldrich Cat. no.

P4864) and EDTA were added to disrupt flocs and stain dead cells. Cultures were measured by flow cytometry and dead/live cells were gated in the red fluorescence channel.

Growth curve determination:

Growth curves were collected using the ThermoFisher MultiSkanTM GO at 300C.

Cells were diluted to OD600 of 0.01 in the indicated media. Absorbance at 600nm was collected every 15 minutes after 15 seconds of shaking. For pH 7.5 media, pH was buffered with 100mM sodium phosphate.

REFERENCES

1. Halfmann, R. et al. Prions are a common mechanism for phenotypic inheritance in wild yeasts. Nature 482, 363-368 (2012). 2. Holmes, D. L., Lancaster, A. K., Lindquist, S. & Halfmann, R. Heritable remodeling of yeast multicellularity by an environmentally responsive prion. Cell 153, 153-165 (2013). 3. Chakrabortee, S. et al. Intrinsically Disordered Proteins Drive Emergence and Inheritance of Biological Traits. Cell 167, 1-13 (2016). 4. True, H. & Lindquist, S. A yeast prion provides a mechanism for genetic variation and phenotypic diversity. Nature 407, 477-483 (2000). 5. Yuan, A. H. & Hochschild, A. A bacterial global regulator forms a prion. 201, 198- 201 (2017). 6. Du, Z., Zhang, Y. & Li, L. The Yeast Prion [SWI+] Abolishes Multicellular Growth by Triggering Conformational Changes of Multiple Regulators Required for Flocculin Gene Expression. Cell Rep. 13, 2865-2878 (2015). 7. Wilson, B. G. & Roberts, C. W. SWI/SNF nucleosome remodellers and cancer. Nat. Rev Cancer 11, 481-492 (2011).

129 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

8. He, F. et al. Decreased expression of ARID1A associates with poor prognosis and promotes metastases of hepatocellular carcinoma. J. Exp. Clin. Cancer Res. 34, 47 (2015). 9. Yan, H. B. et al. Reduced expression of the chromatin remodeling gene ARID1A enhances gastric cancer cell migration and invasion via downregulation of E- cadherin transcription. Carcinogenesis 35, 867-876 (2014). 10. Newby, G. & Lindquist, S. Blessings in disguise: biological benefits of prion-like mechanisms. Trends Cell Biol 23, 251-259 (2013). 11. Liebman, S. W. & Chernoff, Y. 0. Prions in yeast. Genetics 191, 1041-1072 (2012). 12. Crow, E. T. & Li, L. Newly identified prions in budding yeast, and their possible functions. Semin Cell Dev Biol 22, 452-459 (2011). 13. Suzuki, G., Shimazu, N. & Tanaka, M. A yeast prion, Mod5, promotes acquired drug resistance and cell survival under environmental stress. Science 336, 355-9 (2012). 14. Du, Z., Park, K.-W., Yu, H., Fan, Q. & Li, L. Newly identified prion linked to the chromatin-remodeling factor Swil in Saccharomyces cerevisiae. Nat. Genet. 40, 460-465 (2008).

15. Du, Z., Goncharoff, D. K., Cheng, X. & Li, L. Analysis of [ SWI + ] formation and propagation events. Mol. Microbiol. 104, 105-124 (2017). 16. Hines, J. K. et al. [SWI+], the prion formed by the chromatin remodeling factor Swil, is highly sensitive to alterations in hsp70 chaperone system activity. PLoS Genet. 7, 27-29 (2011). 17. Smukalla, S. et al. FLO1 Is a Variable Green Beard Gene that Drives Biofilm-like Cooperation in Budding Yeast. Cell 135, 726-737 (2008). 18. Lee, C.-S. & Haber, J. E. Mating-type switching in Saccharomyces cerevisiae. Microbiol Spectr 3, (2015). 19. Magwene, P. M. in Ecological Genomics 781, 37-48 (2014). 20. Alberti, S., Gitler, A. & Lindquist, S. A suite of Gateway@ cloning vectors for high- throughput genetic analysis in Saccharomyces cerevisiae Simon. Yeast2 24, 913-919 (2007). 21. Kane, S. M. & Roth, R. Carbohydrate metabolism during ascopore development in yeast. J. Bacteriol. 118, 8-14 (1974). 22. Brachmann, C. B. et al. Designer deletion strains derived from Saccharomyces cerevisiae S288C: A useful set of strains and plasmids for PCR-mediated gene disruption and other applications. Yeast 14, 115-132 (1998). 23. Knop, M. et al. Epitope tagging of yeast genes using a PCR-based strategy: more tags and improved practical routines. Yeast 15, 963-972 (1999).

130 131 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

SUPPLEMENTAL FIGURES

Supplemental Figure 1. FLO8-dependent migratory benefits of [SWl cells. A) Comparison of migration of [SWI+] and [swi-] cells lacking a functional FLO8 gene, which is required for the expression of Flo1 and Flo 11. B) Fraction of the total liquid culture (lined bars) or supernatant only (plain bars) that is [SWI+] after growth in YPD or YPD + 5% ethanol initiated at equal inoculum. Measurements were made using flow cytometry on the yTRAP sensor that reports on prion status by fluorescence (Newby, et. al. in review). Flocs were disrupted chemically to solubilize cells for measurement. Error bars indicate standard deviation from triplicate cultures. C) A diagram of experimental procedures to test the ability of cells to migrate in liquid culture. Yeast cultures are inoculated at the bottom of a two-tiered agar well. Water is added until it overflows onto the second tier, followed by aspiration of the water. After incubation to allow cell growth, colonies established by migrated cells appear on the second tier. D) Photograph comparing the migration of [SWI*] and [sw-] cells in liquid media. [SW*] cells migrate so efficiently that they form a lawn of colonies on the second tier.

132 A B Fraction [SWI1 in liquid culture

[swri [SWIJ [swrt [swr m TW flo& flo8' flog eo M Supematant 1.0-

070.5.

.O--

C ncate cultureinolded.,.,rplate

LID [swi- [swr] 1. Allowed to settle, deep pa overflowed with water ml

2. Plate dried and incubated for growth I

133 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

Supplemental Figure 2. Experimental procedure for out-cross ratio measurement. [SWI*] and [swi-] diploid strains were sporulated. Each had the genotype HO+/ho-, where the ho- locus was marked with a cassette expressing the green fluorescent protein NeonGreen. A large pool of these spores was diluted into mixed culture with an ho-, selectable, haploid tester strain. After allowing ample time for mating to occur, we selected for hybrids formed by mating events between the spores and the tester strain. We then determined the ratio of non-fluorescent HO+ spores that out- crossed with the tester strain to fluorescent ho- spores that out-crossed.

134 I [swi] [Sw l/ I

Wt HO Diploid w"t H (repressed) Diploid

ho: GFP ho: GFP

I Sporulation ISporulation

Can switch mating type and [SWI-repressed wt HO mate with daughter cells acts like ho mutant wtHO

1W ho. P WWQ o.GFP ho: GP

annot switch mating type - Tester more likely to mate with tester strain strain hstrainn ~hoJU Mate & select Mate & select \ho l

Tester strain ho 'Tester strain hor Tester strain ho Tester strain ho,

wt HOho GFP wt HO ho: GFP

Out-cross ratio Out-cross ratio (non-fluorescent over fluorescent cells) (non-fluorescent over fluorescent cells) low INCREASES

135 Chapter 3: SWI Pioneers: A fungal prion promotes geographic and sexual exploration

Supplemental Figure 3. [SWI*] pioneers are less fit than [swil settled cells. A) Growth comparison of [sw-] cells (blue) and [SWI/] cells (red) in standard growth media supplemented with glucose, or B) raffinose. Cell density was measured by absorbance every 15 minutes.

136 A Glucose 1.0.

0.8.

Q0.6-

0 0.4.

0.2-

0.0;0 0 20 40 Time (hours)

B Raffinose 1.0.

0.8-

CD 0.6-

0 0.4-

0.2-

0.01.. 0 20 40 60 Time (hours)

137 Table SI. Yeast strains used in Chapter 3

Plasmids Swil Prion Strain transformed State Parent Genotype Purpose Reference Brachmann, CB, MA Ta, his3d1, leu2A, met15A, ura3A, Base strain for comparison of et al. Yeast. BY4741 ho-, flo8- Swil Drion states 1998. Non-flocculant, [swi-] strain. Control for FLO8 involvement. Used for growth curves and MA Ta, his3dl, leu2A, met15A, ura3A, for out-cross ratio yGAN100 pGAN202 - BY4741 ho-:yTRAP Swil sensor, flo8- measurements This study Non-flocculant [SWI+] strain. Control for FLO8 involvement. Used for growth curves and MATa, his3dl, leu2A, met15A, ura3A, for out-cross ratio yGAN101 pGAN202 + yGAN100 ho-:yTRAP Swil sensor, flo8- measurements This study Flocculant [swi-] strain. Used pGAN202, MA Ta, his3dl, leu2A, met15A, ura3A, for migration experiments and yGAN102 pGAN300 - yGAN100 ho-:yTRAP Swil sensor, FLO8:LEU2+ survivability experiments. This study Flocculant [SWI+] strain. Used pGAN202, MA Ta, his3dl, leu2A, met15A, ura3A, for migration experiments and yGAN103 pGAN300 + yGAN102 ho-:yTRAP Swil sensor, FLO8+:LEU2+ survivability experiments. This study Kane, SM and Used for out-cross ratio Roth R. J. SK1 - MA Ta/MA Ta, HO+ measurements. Bacteriol. 1974. MA Ta, his3dl, leu2A, met15A, ura3A, Used for out-cross ratio Invitrogen Cat. BY4741 trp1:KanMX ho-, flo8- measurements. no. 95401.H2 MA Ta/MA Ta,HO+/ho-:yTRAP Swil yGAN100, sensor, HIS3/his3dl, LEU2/leu2A, (YJW508 cured by guanidine) yGAN104 pGAN202 - SK1 MET15/metl5A, URA3/ura3A For [psi-][rng-] base strain This study MA Ta/MA Ta, HO+/ho-:yTRAP Swil yGAN 101, sensor, HIS3/his3dl, LEU2/Aeu2A, (YJW508 cured by guanidine) yGAN105 pGAN202 SKI MET15/met15A, URA3/ura3A For [psi-][mg-] base strain This study MA Ta/MA Ta, HO+/ho-:yTRAP Swil yGAN 101, sensor, HIS3/his3dl, LEU2/leu2A, (YJW508 cured by guanidine) yGAN106 pGAN202 - SK1 MET15/metl5A, URA3/ura3A For [psi-][mg-] base strain This study

138 Chapter 4

Conclusion and future directions

139 Chapter 4: Conclusion and future directions

CONCLUSION yTRAP is a versatile technology that can sensitively detect prion like propagation in yeast. With it, one can quantitatively detect a given protein's solubility in vivo and in high throughput. These advantageous properties enable previously-impossible studies.

Using yTRAP, I identified mutant prion alleles that can cure the [PSI'] and [RNQ*] prions after transient expression. I also found a fusion architecture that enables 100% induction of the [PSI'] prion - the only fully efficient inducer known to date. With these reliable prion-switching factors, I constructed two synthetic devices out of prions: 1) a memory circuit that records elevated temperatures in the prion state of the population, with a simple readout using yTRAP, and 2) 'anti-prion drive' cells that cure prions in mating partners and progeny, reversing the typical inheritance pattern of prions (see

Chapter 2).

yTRAP also assisted ecological studies on the yeast prion [SWI*] - using it, I could rapidly find prion-switched cells and keep track of prion status throughout experiments (see Chapter 3). I observed that the [SWI*] prion greatly enhances the migration of cells in flowing water and increases the tendency of cells to out-cross, potentially increasing geographic and genetic diversity in nature. [swi] cells, on the other hand, tended to stay in place and were more robust to stressful conditions. In a

mixed population, the presence of sub-stoichiometric numbers of [SWI/] cells increased the migratory ability of their [swi] neighbors. Thus, the epigenetic prion switch establishes two specialized subpopulations of yeast - "pioneer" [SWI*] cells that explore and colonize new territory, and "settled" [sw-] cells that tend to remain clustered in place

140 with increased survivability. This ability to generate specialized subpopulations likely brings ecological advantages to the population as a whole.

Before constructing the yTRAP system, I explored other ways to screen and select for protein aggregation. The Sup35C-tag, conferring [PSI]/[ps-] phenotypes based on the aggregation state of the tagged protein, has previously been used to study diverse prion-like phenomena. I constructed a nonsense CANi allele that enables conditional selection for the soluble ([psi]) state (see Appendix A). Such a system could be useful for the selection or screening of solubilizing factors without the use of a flow cytometry system.

In general, the versatility of yTRAP makes it preferable to other techniques for studying aggregation. The [PS/+]-associated phenotypes (red-to-white color change and growth in the absence of adenine) are conferred by a plethora of other mutations that cause nonsense suppression, leading to high background. [PSI+] phenotypes also can have strong effects on the biology of the cell, leading to altered growth in a variety of conditions. yTRAP utilizes a synthetic and orthogonal transcriptional activator, which has minimal effect on the cell except for the programmable reporter. yTRAP can be multiplexed to allow for the quantitative readout of the solubility of multiple proteins at once. Unlike [PSI+] phenotypes, which can take up to 7 days to appear after plating, yTRAP can be measured at any point, even in individual cells.

Currently, the most common methods to track aggregation quantitatively involve fractionation of cell lysate into soluble material and pelleted material, followed by immunoblotting. However, such techniques are costly both in time and money.

Furthermore, extraction of proteins into lysate disrupts their endogenous localization

141 Chapter 4: Conclusion and future directions and other cellular machinery (such as chaperones) that may be involved in the

aggregate in question. yTRAP allows for simple photographs or quantitative flow

cytometry to measure the solubility in the native context, without disruption of the

cellular system.

I utilized yTRAP to construct a model of prion propagation of the human proteins

RIP1 and RIP3, involved in innate immunity (see Appendix B). I found that these proteins could propagate as prions in yeast. Exploring their chaperone sensitivity, I

found that Hsp70 disruption severely increases the aggregation of RIP1, causing it to

enter into the prion state. Whether this Hsp70-dependence is also true in human cells

remains an interesting avenue of future exploration. This yeast model could be useful in

the future for screening for inhibitors or genetic interactors with the RIP1/RIP3

necrosome. In the following future directions section, I will discuss other potential

applications of the yTRAP technology.

FUTURE DIRECTIONS

New studies facilitated by yTRAP technology

yTRAP facilitates many kinds of experiments that were previously difficult or impossible:

1. yTRAP can provide insight into the mechanism-of-action of disease modifiers.

The Lindquist lab and others have specialized in using drug or genetic screens in

yeast to identify modifiers of toxicity for human disease-relevant proteins such as

alpha synuclein, a-beta, and TDP-43. These screens involve measuring changes

in yeast growth due to altered toxicity of the disease protein. However, it is a

difficult process to tease out the mechanism-of-action for any hit from a yeast

142 screen. Reproducible enhancement or suppression of growth rate could be due

to a perturbation on the disease protein itself (perhaps causing it to be degraded,

solubilized, or aggregated in an alternative manner), or on any number of other

pathways related to cellular toxicity. Using yTRAP, one can look directly for

perturbations on the disease protein of interest in the presence of the modifier.

2. yTRAP itself can be used as a platform for the discovery of new genes and drugs

that affect disease proteins. Currently, diseases caused by aggregating proteins

(for example, Alzheimer's disease and Parkinson's Disease) have no cure.

Treatments can ameliorate the symptoms, but do not fix the underlying problem.

Using yTRAP screens, akin to screens conducted in Chapter 2 for prion-curing

alleles, one could screen for genes or drugs that directly affect aggregation

events that underlie disease. A yTRAP-based screen has three advantages over

traditional toxicity-based screens:

a. Screens can be done at low (more physiologically-relevant) levels of

protein. Previous screens relied on immense and unnatural

overexpression of toxic proteins, but in human disease progression, this is

typically not the context. Often disease is thought to result from a slow

build-up of aggregated species and related events that happen over time.

By examining aggregation at low protein concentrations, yTRAP can shed

light on early events in disease progression and be used to screen for

genes and drugs that may be relevant during these early time points. This

could be the key to finding a therapy or preventative measure for

aggregation-associated diseases.

143 Chapter 4: Conclusion and future directions

b. Screens can be done on aggregates with very low or no toxicity. Some

aggregates may only be toxic in the context of specific cell types, for

example SOD1 aggregates interfering with neuronal function in the

disease ALS. These specific cell types may not be suitable for high

throughput screening, and cell types suitable for screening do not

experience toxicity no matter how extremely SOD1 expression is driven.

Thus, researchers were previously unable to screen for modifiers. Using

yTRAP, one can find inhibitors of aggregation without the requirement that

this is linked to a cell growth phenotype.

c. Validated hits from the screen will directly affect the conformational state

of the protein of interest, as opposed to a downstream pathway involved in

toxicity. There is currently no therapy for a neurodegenerative disease that

accomplishes this, quite likely because it has not been possible to screen

for them. For many disease-related proteins (for example, SOD1 or

Huntingtin) the disease-causing mutation is in the aggregating protein.

Any treatment that affects toxicity without altering the protein's

aggregation is like treating a symptom rather than the root of the disease.

The ideal treatment would be to push the protein back to its normal

conformation so that the root of the disease is eliminated. Though screens

aimed towards these efforts were impossible before, they are now

possible with yTRAP.

3. yTRAP could be used to identify important residues mediating protein

aggregation, and thus potential drug target sites, for example in the disease

144 proteins TDP-43 or a-synuclein (associated with ALS and Parkinson's disease,

respectively). One could produce a library of yTRAP sensors through random

mutagenesis of the protein of interest. Mutations that cause either aggregation or

solubilization will yield a corresponding increase or decrease in fluorescent

signal. These cells can be collected by FACS and their plasmids sequenced to

determine the relevant mutations (and thus, the sites on the protein important for

aggregation/solubility). Alternatively, comprehensive alanine scanning could be

conducted and rapidly measured with yTRAP.

4. yTRAP could be used to profile the effects of toxic proteins and conditions to

understand their cellular consequences. This first requires the creation of a

library of yTRAP sensors, for example the set of all yeast proteins or the set of all

human transcription factors. One could then apply a stimulus such as a drug or

toxic protein before scanning the array of yTRAP sensors. This would illuminate

the particular proteins or pathways perturbed by this stimulus.

5. yTRAP could be used to identify chaperone clients. By chemically or genetically

perturbing a chaperone, client proteins will aggregate or be degraded. By

challenging a panel of yTRAP sensors with chaperone inhibitors, clients could be

quickly identified without complicated reagents or biochemical purification.

6. yTRAP could be used to profile the effects of a library of ORFs on a single

protein of interest. For example, a yTRAP sensor strain could be mated to the

arrayed yeast ORFeome overexpression library. Upon overexpression, the effect

of each ORF on the protein-of-interest can be quantified.

145 Chapter 4: Conclusion and future directions

7. yTRAP could be used to identify new prions. Proteins from S. cerevisiae or from

other organisms can be cloned into the yTRAP system in high throughput. The

identical protein can be transiently overexpressed in its own sensor strain, and

then repressed. After a few generations of recovery, sensor strains can be

screened for heritable maintenance of a low-fluorescence (aggregated) state.

8. yTRAP could be used to measure the aggregation propensity of a set of proteins.

For example, one could make yTRAP sensor strains out of the human proteome.

During overexpression of each protein in its own sensor strain, yTRAP signal can

be measured to determine the aggregation state of each protein.

9. Multiplexed yTRAP sensors could be used to elucidate networks of protein

interactions and assemblies. For example, there are a number of described RNP

granules like stress granules, P-bodies, spliceosomes, and the nucleolus. But is

every granule made the same way, or do different stimuli lead to different RNP

granule components? What are the organizers that nucleate granules and the

subsequent order of assembly? How does the perturbation of one gene or

component affect the others? How is this altered in disease? These are

examples of the questions that can now be accessed using high throughput

yTRAP measurement and quantification. Doing a time-course in living cells, one

can study all of these phenomena in real time.

146 Potential extensions to the yTRAP system yTRAP is limited in several aspects. Further development of this technology will allow researchers to tackle even more challenging problems. I will briefly outline current limitations and present thoughts on how to address them.

1. The default yTRAP signal is variable between different proteins. This could be

due to a number of properties of the protein-of-interest such as the number and

localization of its binding partners, its natural segregation between the nucleus

and the cytoplasm, or its stability. In the process of optimizing the yTRAP sensor

for Sup35 and Rnq1, it was necessary to include both a nuclear localization

signal and a nuclear export signal on the yTRAP tag. This presumably keeps the

yTRAP fusion protein shuttling between the nucleus and cytoplasm so it has the

opportunity both to activate transcription and be sensitive to cytoplasmic

aggregates. However, the strength of nuclear import and export signals may

have to be tuned for some proteins where the dynamic range of signal is poor.

For example, some sensor strains are very dark when the protein of interest is

likely not aggregated to begin with. To allow for tracking relevant perturbations in

such a protein, it may need to be tagged with additional nuclear localization

sequences. Alternately, some proteins may be so aggregation prone or so

soluble that finer control is necessary over the expression level of the yTRAP

fusion protein. I chose the SUP35 promoter to drive the yTRAP fusion in the

current system because it is extremely stable in expression and is known to

generate levels of protein that are in the right range to exist either in an

aggregated or soluble state. It may be useful to replace the constitutive SUP35

147 Chapter 4: Conclusion and future directions

promoter with an inducible promoter in order to wield more control over this

important factor of the system.

2. Alterations in yTRAP signal can be indicative of many different perturbations.

Relevant perturbations could include a shift in binding partner, localization,

degradation or synthesis rate, or aggregation/solubilization. My studies have

focused on proteins that are known to be aggregation-prone, in which case the

sensor works marvelously. However, some potential future experiments I

described above would involve profiling proteins with unknown properties. In

these cases, alterations in signal might be due to diverse perturbations. For

some purposes, this may be sufficient to draw links between interacting

pathways. However, it would be useful to have a built in system that could be

used to more easily narrow down the possibilities. For example, the yTRAP tag

could itself include a fluorescent protein so that changes in the synthesis or

degradation of the protein could be measured alongside yTRAP signal. yTRAP

could also interface with existing panels of yeast profiling strains, so that

alterations in localization could be more readily determined by microscopy. The

yTRAP tag currently has 6 tandem HA tags, facilitating blotting and pull downs.

Immunoprecipitation followed by mass spectrometry could be used to determine

whether a change in yTRAP signal is accompanied by a change in binding

partners.

3. Aggregates and other protein perturbations are not limited to the nucleus and

cytoplasm. Such events can happen in any cellular compartment or even in

extracellular space. yTRAP relies on tracking the solubility of the protein-of-

148 interest with a transcription factor tag, so it is not suitable for proteins that exist in

other membrane-bound compartments. One possibility is to re-target such

proteins to the cytoplasm for measurement, but this eliminates one of the key

benefits of yTRAP, namely that proteins can be studied directly in their native

environment. Other creative systems must be developed to track aggregation

and other perturbations in contexts outside of the nucleus and cytoplasm. Some

inspiration may be taken from existing systems - for example, the RIP1/RIP3

kinases that increase in activity after assembly into the amyloid state. If an in situ

assay could be developed using enzymes such as these that are activated upon

aggregation, it may be possible to track aggregation in cellular compartments or

extracellular space.

4. We have so far only shown that yTRAP can be multiplexed to examine two

proteins at the same time. However, using more pairs of fused transcriptional

activators and their paired reporter cassettes, there is no limit to the number of

sensors that could be multiplexed. The first limitation one may encounter is that

any more than 3 fluorescent proteins are difficult to distinguish simultaneously.

However, flow cytometry systems have been made that are able to

simultaneously detect as many as 8 different fluorescent dye-conjugated

antibodies, compensating for spectral overlap. The modular yTRAP reporter

cassettes could be reprogrammed to express surface proteins against which dye-

conjugated antibodies already exist. In this manner, the sensor could be readily

multiplexed to track 8 proteins simultaneously with flow cytometry. Beyond that,

149 Chapter 4: Conclusion and future directions

high-throughput sequencing of expressed mRNA barcodes could extend

multiplexing capacity much further.

5. So far, yTRAP has only been implemented in S. cerevisiae. However, the basic

concept of the technology should be functional in any cell. Synthetic

transcriptional activator tags and their paired promoters could be used in

bacteria, tissue culture, or even in live animals to track aggregation or other

protein perturbations quantitatively. The expansion of yTRAP to other cell types

of interest will be a useful for studying protein perturbations in their native

context, for example the progression of disease-associated aggregation in living

brains or the propagation of inflammosomes in macrophages. Earlier this year, a

bacterial protein was observed to propagate as a prion for the first time.

Searching for bacterial prions could also be facilitated by yTRAP.

The application of prions to synthetic biology

In chapter 2, I demonstrated two simple prion-based cellular devices - a heat memory device and an anti-prion drive. Prions have unique properties that could be useful for synthetic biology: 1) prions are protein-conformation based switches, allowing for post- translational regulation of fused domains such as the yTRAP transcriptional activator, 2) prions are natural toggle switches, some of which have switching frequencies under one

in ten million, 3) switching factors can be readily selected with yTRAP, 4) new orthogonal prions can be generated readily by scrambling Q/N-rich domains, increasing the available number of switches. yTRAP could facilitate the identification of new, orthogonal prions as well as screens to find factors that specifically switch them.

150 Equipped with a sufficient number of high-fidelity switches, more complex systems could be engineered. For example, one could program a yeast cell to release a peptide drug only if a certain combination of environmental conditions is met. The more switches available, the more complex and fine-tuned that response may be.

The hyper-inducing prion fusion identified in chapter 2 can already be used as a memory element in yeast. When expressed from an environment-sensitive promoter, the cell population will record the environmental stimulus for later read-out. This could be used, for example, to record conditions such as low oxygen or low glucose for quality control in bioreactors. Cells could alternately be programmed to switch in the presence of virulent microbes, for use as biosensors in the gut or environment.

Prion-curing alleles can be used to heritably activate transcription of a pre-

programmed yTRAP reporter gene. For example, in a bioreactor, a temporary signal could be added which causes cells to express the curing allele, or anti-prion drive cells could be added to eliminate prions through mating. This would avoid the continual use of inducing agents that might otherwise be used to regulate expression. I am eager to see what other prion-based tools may be constructed as our methods become more sophisticated.

The ecological ramifications of prions

In chapter 3, I described a prion switch that regulates specialized subpopulations of yeast - "pioneer" [SW*] cells colonize new territory and mate with diverse partners, while "settled" [swi-] cells tend to remain clustered and genetically homogenous. Each

151 Chapter 4: Conclusion and future directions

S. cerevisiae prion has the capacity to similarly form a distinct subpopulation with unique advantages and disadvantages to each. Approximately ten such prion proteins are known to exist currently, but it is likely that many more remain yet undiscovered.

With the discovery this year of a bacterial prion - Rho in C. botulinum, which can propagate as a prion in E. coli - it seems likely that prion switches are common in the microbial world. Specializing subpopulations of cells to be suited to particular tasks and conditions is an elegant strategy to adapt to harsh environments. Because prion states switch at a low basal frequency, large populations will constantly be sampling new epigenetic states at only a minor cost. If one prion state confers an advantage, cells harboring it will produce more progeny, which also inherit the prion. Thus, as long as that state is advantageous, it will be largely maintained by the subpopulation. Prion- based microbial specialization is similar to the toxin-antitoxin-mediated microbial persister phenomenon. However, prion-based epigenetic regulation has the capacity to control much more than growth rate. A particularly interesting avenue for future research is the identification of new specialized microbial subpopulations, the mechanisms that govern them, and their ecological ramifications. It may someday become possible to push disease-causing microbes to non-virulent specialized states. It is my hope that this body of work and the yTRAP technology will facilitate yet more interesting work in the future.

152 Appendix A

Selection for the [psi-] phenotype with canavanine

153 Appendix A: Selection of the [psi] phenotype with canavanine

SUMMARY

The [PSI'] prion has been the subject of much investigation. The phenomenon first gained interested due to the unusual inheritance pattern of its red/white phenotype1 .

Later, it was used a model for prion propagation. Many studies were made possible due to the selective growth advantage of white [PSI'] on media lacking adenine. By growing

red ade- cells on media lacking adenine, spontaneous [PSI+] white colonies would

appear. A similar system was used to test whether other proteins could propagate as

prions after tagging with Sup35C 2. However, it may sometimes be useful to select for soluble, [psi] phenotypes, for example when searching for factors which solubilize

proteins. Therefore, I constructed a selection system that allows only [psi] colonies, which faithfully terminate translation at stop codons, to grow. Though the yTRAP technology is more versatile and easily measured, there may be some instances where this strong canavanine selection system would be preferred.

RESULTS

I cloned the CANI protein coding sequence into pAG323GPD 3, driving it with strong,

constitutive expression. When expressed, this produces a membrane channel that imports arginine and its toxic mimetic, canavanine. By introducing a nonsense mutation

into CAN1, functional protein will only be expressed in suppressor strains, such as

strains with the [PSI+] phenotype. When canavanine is present in the medium, [PSI+]

cells will be selectively killed.

154 I used site-directed mutagenesis to alter CANI codon 370 to each of the three stop codons. This codon was selected because it was predicted to encode for a disordered loop in the protein, before the last transmembrane domain. It is not known which amino acid, if any, is placed at the site of [PS/*]-induced read-through. I did not want to place the mutation in a transmembrane domain because it may be sensitive to mutation. Preliminary experiments indicated that TAA and TAG each had approximately equal effects, while TGA seemed to allow for some read-through even in the [ps-] state

(data not shown). The TAA stop codon was used for future experiments. Available prion states were compared in 74D-694 and W303 strains of yeast after transformation with the 2pm plasmid expressing CAN1-370-TAA. Cells were spotted on synthetic media lacking arginine and histidine, supplemented with 60pg/mL canavanine, or on synthetic complete media (Fig 1). On canavanine media, [psr] strains had a strong growth advantage.

SD-arg-his + 60ug/mL canayanine CSM

74D-694

W303 [sI+[PIN+i 7PS+;PIN+I

-sI1PIN+ A &s!-[PIN+)

Figure 1. Canavanine selection for [ps-]. Top: 74D-694 genetic background. Bottom: W303 genetic background. Left: Selective media lacking histidine, arginine, and supplemented with 60pg/mL canavanine. Right: Non-selective media control. Prion status of strains indicated. Overnight cultures were spotted at OD 0.4 with serial 5- fold dilutions proceeding to the right.

155 Appendix A: Selection of the [psi-] phenotype with canavanine

REFERENCES

1. Cox, B. S. PSI, a cytoplasmic suppressor of super-suppressor in yeast. Heredity (Edinb). 20, 505-21 (1965). 2. Alberti, S., Halfmann, R., King, 0., Kapila, A. & Lindquist, S. A systematic survey identifies prions and illuminates sequence features of prionogenic proteins. Cel 137, 146-58 (2009). 3. Alberti, S., Gitler, A. & Lindquist, S. A suite of Gateway@ cloning vectors for high- throughput genetic analysis in Saccharomyces cerevisiae Simon. Yeast2 24, 913-919 (2007).

156 Appendix B

RIP1/RIP3 prion propagation in yeast

157 Appendix B: RIP1/RIP3 prion propagation in yeast

SUMMARY

RIP1 and RIP3 are human innate immunity proteins that assemble to an amyloid conformation in response to viral infection1 . Unusually, this amyloid state is composed of both proteins in equal ratio - typically amyloid fibers are highly specific for identical protein constituents. I reconstituted the RIP1I/RIP3 prion in a yeast model. This model can be used to identify regulators and study its biology with the tools of yeast genetics.

We identified that RlP1/RIP3 aggregation is increased by Hsp70 inhibition.

RESULTS

RIP3 is more aggregation prone than RIPi yTRAP sensors were constructed for the full length RIP1 and RIP3 proteins. Sensor plasmids were integrated into W303. A western blot confirmed that both proteins were expressed at roughly equal levels (Fig 1A). However, SDD-AGE indicated that RIP3 formed SDS-resistant aggregates even at this low expression level, while RIP1 did not

(Fig 1 B). Based on the yTRAP signal, RIP3 was far less soluble in the cell than RIP1, by a factor of approximately 20 (Fig 1 C).

Figure 1. RIP1 and RIP3 aggregate in yeast. A) Western blot against the HA epitope of the yTRAP fusion. RIP1 and RIP3 sensor strain lysates were loaded. RIPI-synTA expected molecular weight: 108.5kDa. RIP3-synTA expected molecular weight: 89.5kDa. B) SDD-AGE blot against RIP1 and RIP3 sensor strain lysates. RIP3 alone shows a strong smear of aggregated, SDS-resistant protein. C) Aggregation of RIP1 and RIP3 in yTRAP sensor strains (inverse yTRAP signal). Normalized to RIP1. D) Aggregation of RIP1 in its yTRAP sensor strain in response to overexpression of the indicated proteins. E) RIP1 intrinsically disordered domain overexpressed in the RIP1 sensor strain. Green channel: yTRAP signal for full length RIPI solubility. Red channel: overexpressed RIP1 intrinsically disordered domain tagged with the red fluorescent protein mKate2. Microscopy channels of DIC, green fluorescence, and red fluorescence overlaid.

158 A B

75kDa

Anti-HA western Anti-HA SDD-AGE

am monomer 6# Anti-PGK1 - n (loading control) \qN4~ q~ q-

C D Fold aggregaton propensity (normalized to RIPI) Fold RIPI aggregation after 161w overoxpresslon Is- C t6 a CL * 15. 0 2

C45 32 S .- OM 0. / 0 4e Sb I, E

159 Appendix B: RIP1/RIP3 prion propagation in yeast

RIPI aggregates when either itself or RIP3 is overexpressed

Episomal 2pm plasmids encoding galactose-inducible RIP1 and RIP3 fragments tagged with the red fluorescent protein mKate2 were transformed into the RIP1 sensor strain.

The inducible fragments were the intrinsically disordered domains of RIP1/RIP3. For

RIP1, this is amino acids 295-583. For RIP3, this is amino acids 295-518 (its C- terminus). RIP1 aggregates upon overexpression of either RIP domain, but not the aggregation prone prion domain of Ure2, as measured by yTRAP (Fig 1 D). Using microscopy, red foci of aggregated protein are visible and correlate with decreased green yTRAP fluorescence (Fig 1 E).

RIPI and RIP3 can form a 2-component prion in yeast

The RIP1 and RIP3 sensors were integrated into W303 yeast of opposite mating types.

As a control, the RIP1 sensor was also integrated into both mating types. Selectable plasmids were added so that each individual haploid could be selected for by growth on media lacking histidine or media lacking tryptophan. The complementary haploid strains were mated overnight in YPD, and diploids were selected by streaking on media lacking both histidine and tryptophan. When RIP1 sensor strains were mated with RIP3 sensor strains, some diploid colonies were dark in the green fluorescent channel (Fig 2A).

When RIP1 sensor strains were mated to each other, only bright colonies were generated (Fig 2B). Dark colonies must have both RIP1 and RIP3 aggregated, since soluble RIP1 alone would be enough to induce green fluorescence. These dark colonies

160 could be propagated, but the prion state was unstable and bright colonies arise with every passage (data not shown).

RIP1 is highly sensitive to Hsp70 perturbation

Uracil-selectable plasmids expressing dominant negative alleles of Hsp70 (K69M) or

Hspl04 (K218T, K620T) from the constitutive TDH3 promoter were transformed into

bright RIP1-RIP3 sensing diploids. Inhibition of Hsp70 caused a particularly large effect,

inducing aggregation of RIP1 as measured by yTRAP signal (Fig 2C). After streaking on

5-FOA to eliminate the Hsp70-inhibiting plasmid, some cells returned to a bright fluorescent state (Fig 2D). However, most remained dark, indicating that the transient

Hsp70 inhibition had induced a heritable prion state. These results suggest that Hsp70

may be a strong regulator of RIP1/RIP3 assembly. This remains a promising avenue of

research in the context of human cells. This yeast model of RIP1/RIP3 assembly may

be useful in the future to screen for modulators of this pathway in high throughput.

161 Appendix B: RIP1/RIP3 prion propagation in yeast

B A RIP1-3 diploids RIP1-1 diploids

C D Elimination of 8. - DN Hsp70 on 5-FOA 0 C. ,5 6* C 0 4. o

6 0. U. I I

Figure 2. RIPI/RIP3 forms a bipartite prion in yeast regulated by Hsp70. A) Mating tubes were streaked on dual-selection media for the diploids of a RIP1 yTRAP sensor strain with a RIP3 yTRAP sensor strain or B) a RIP1 sensor strain with another RIP1 sensor strain. Red arrows indicate darker colonies, indicating the aggregation of both proteins sensed by these diploids. C) Bright RIP1-RIP3 sensor diploids were transformed with a dominant negative (DN) Hsp70 plasmid, a dominant negative Hsp104 plasmid, or an empty plasmid control. D) Dark RIPI-RIP3 sensor diploids containing the DN Hsp70 plasmid were streaked on 5-FOA so that only cells which lose the plasmid may grow. Some cells return to a bright state (indicated by red arrows). Others remain dark, indicating they inherited a stable prion state through transient inhibition of Hsp70.

162 METHODS

For a detailed description of methods, see Chapter 2. Briefly, the Bio-Rad ChemiDoc was used to visualize western blots, SDD-AGE, and fluorescent yeast colonies. The

MACSquant VYB (Miltenyi Biotec) was used to measure yTRAP signal quantitatively for fold-aggregation measurements. A Nikon Eclipse epifluorescence microscope was used to collect fluorescence micrographs. For overexpression, samples were measured after

12 hours of growth in media supplemented with 2% galactose.

REFERENCE

1. Li, J. et al. The RIP1/RIP3 Necrosome Forms a Functional Amyloid Signaling Complex Required for Programmed Necrosis. Cell 150, 339-350 (2012).

163